INTRODUCTION

TOP ABSTRACT INTRODUCTION CHEMISTRY OF ROS CELLULAR SOURCES AND REGULATION... MEASUREMENT OF ROS IN... LIGAND-INDUCED ROS PRODUCTION... SIGNALING MOLECULES "TARGETED"... MECHANISMS OF ROS ACTION CONCLUSION REFERENCES

ALL MULTICELLULAR ORGANISMS depend on highly complex networks of both extracellular and intracellular signals to orchestrate cell-cell communication in diverse physiological processes such as developmental organogenesis, maintenance of normal tissue homeostasis, and repair responses to tissue injury. Typically, extracellular signals are composed of growth factors, cytokines, hormones, and neurotransmitters that bind to specific cell surface receptors. These receptor-ligand interactions then generate various types of intracellular signals that may involve changes in ion concentrations (ion channel-linked receptors), activation of trimeric GTP-binding regulatory proteins (G protein-coupled receptors), and activation of receptor kinases (enzyme-linked receptors). Downstream signaling is then relayed by second messengers (such as cAMP, Ca²⁺, and phospholipid metabolites) and by protein phosphorylation cascades. Ultimately, these intracellular signaling pathways lead to the activation of transcription factors that regulate the expression of specific sets of genes essential for diverse cellular functions.

Molecular oxygen (dioxygen; O₂) is essential for the survival of all aerobic organisms. Aerobic energy metabolism is dependent on oxidative phosphorylation, a process by which the oxidoreduction energy of mitochondrial electron transport (via a multicomponent NADH dehydrogenase enzymatic complex) is converted to the high-energy phosphate bond of ATP. O₂ serves as the final electron acceptor for cytochrome-c oxidase, the terminal enzymatic component of this mitochondrial enzymatic complex, that catalyzes the four-electron reduction of O₂ to H₂O. Partially reduced and highly reactive metabolites of O₂ may be formed during these (and other) electron transfer reactions. These O₂ metabolites include superoxide anion (O₂·) and hydrogen peroxide (H₂O₂), formed by one- and two-electron reductions of O₂, respectively. In the presence of transition metal ions, the even more reactive hydroxyl radical (OH·) can be formed. These partially reduced metabolites of O₂ are often referred to as "reactive oxygen species" (ROS) due to their higher reactivities relative to molecular O₂.

ROS from mitochondria and other cellular sources have been traditionally regarded as toxic by-products of metabolism with the potential to cause damage to lipids, proteins, and DNA (78). To protect against the potentially damaging effects of ROS, cells possess several antioxidant enzymes such as superoxide dismutase (which reduces O₂· to H₂O₂), catalase, and glutathione peroxidase (which reduces H₂O₂ to H₂O). Thus oxidative stress may be broadly defined as an imbalance between oxidant production and the antioxidant capacity of the cell to prevent oxidative injury. Oxidative stress has been implicated in a large number of human diseases including atherosclerosis, pulmonary fibrosis, cancer, neurodegenerative diseases, and aging (56, 103). Yet the relationship between oxidative stress and the pathobiology of these diseases is not clear, largely due to a lack of understanding of the mechanisms by which ROS function in both normal physiological and disease states.

Accumulating evidence suggests that ROS are not only injurious by-products of cellular metabolism but also essential participants in cell signaling and regulation (76, 237). Although this role for ROS is a relatively novel concept in vertebrates, there is strong evidence of a physiological role for ROS in several nonmammalian systems. In bacteria, the OxyR protein functions as a transcriptional regulator of H₂O₂-inducible genes and has been shown to be directly activated by oxidation (53, 278). A recent study (324) has shown that H₂O₂ oxidizes two conserved cysteines in OxyR to form intramolecular disulfide linkages that trigger the activation of this transcription factor, presumably by changing its conformation. The Escherichia coli SoxR transcription factor is activated specifically in response to O₂·-generating redox-cycling agents such as paraquat and menadione (95). Activated SoxR mediates SoxS gene transcription, resulting in an increase in SoxS protein that then activates the transcription of several other genes including superoxide dismutase (SOD) (63). SoxR is a homodimer that contains two redox-active iron-sulfur [2Fe-2S] centers that are sensitive to oxidation by O₂· (109, 312). The iron-sulfur centers of SoxR must be in their oxidized state for them to be transcriptionally active (85), thus providing a plausible mechanism by which O₂· transmits its gene regulatory signal. In plant cells, generation of H₂O₂ in response to various pathogens elicits localized cell death to limit spread of the pathogen (165) and a more systemic response involving the induction of defense genes regulating plant immunity (10). In sea urchins, fertilization triggers extracellular production of H₂O₂ by a plasma membrane oxidase with the simultaneous release of ovoperoxidase (263). Peroxidase- or H₂O₂-catalyzed cross-linking of extracellular proteins then forms a protective envelope around the freshly fertilized oocyte.

The apparent paradox in the roles of ROS as essential biomolecules in the regulation of cellular functions and as toxic by-products of metabolism may be, at least in part, related to differences in the concentrations of ROS produced. This is analogous to the effects of nitric oxide (NO·), which has both regulatory functions and cytotoxic effects depending on the enzymatic source and relative amount of NO· generated (210). NO· functions as a signaling molecule mediating vasodilation when produced in low concentrations by the constitutive isoform of nitric oxide synthase (NOS) in vascular endothelial cells (309) and as a source of highly toxic oxidants utilized for microbicidal killing when produced in high concentrations by inducible NOS in macrophages (178). Indeed, all phagocytic cells have a well-characterized O₂·-generating plasma membrane oxidase capable of producing the large amounts of ROS required for its function in host defense. In this review, we examine the evidence for the presence of similar plasma membrane oxidases that generate much lower amounts of ROS in nonphagocytic cells for purposes of cell signaling and regulation. We summarize the cellular effects of a wide range of ligand-receptor interactions that have been shown to generate intracellular or extracellular ROS. Emerging concepts on the mechanisms by which ROS may function as signaling molecules are discussed. For purposes of this review, we focus primarily on the cellular effects of ligand-mediated (endogenous) production of ROS rather than on the effects of exogenous oxidative stress. Also, the role of NO· as a signaling molecule is not discussed here except in the context of its ability to interact with and modulate O₂· signaling.

	CHEMISTRY OF ROS

TOP ABSTRACT INTRODUCTION CHEMISTRY OF ROS CELLULAR SOURCES AND REGULATION... MEASUREMENT OF ROS IN... LIGAND-INDUCED ROS PRODUCTION... SIGNALING MOLECULES "TARGETED"... MECHANISMS OF ROS ACTION CONCLUSION REFERENCES

The biological chemistries of ROS primarily determine the ability of different species to react with specific cellular substrates within the microenvironment in which they are produced. These ROS-substrate reactions are likely to form the basis for our understanding of ROS specificity and their mechanisms of action. ROS have often been "loosely" categorized as free radicals, but this is incorrect because not all ROS are free radicals. A free radical is defined as any atomic or molecular species capable of independent existence that contains one or more unpaired electrons in one of its molecular orbitals (102). Molecular O₂ itself qualifies as a free radical because it has two unpaired electrons with parallel spin in different -antibonding orbitals. This spin restriction accounts for its relative stability and paramagnetic properties. O₂ is capable of accepting electrons to its antibonding orbitals, becoming "reduced" in the process, and, therefore, functioning as a strong oxidizing agent.

A one-electron reduction of O₂ results in the formation of O₂· either by enzymatic catalysis or by "electron leaks" from various electron transfer reactions. O₂· chemistry in aqueous solution differs greatly from that in organic solvents. In contrast to its remarkable stability in many organic solvents, O₂· in aqueous solution is short-lived. This "instability" in aqueous solutions is based on the rapid dismutation of O₂· to H₂O₂, a reaction facilitated by higher concentrations of the protonated form of O₂· (HO₂·) in more acidic pH conditions. Thus the dismutation reaction has an overall rate constant of 5 × 10⁵ M¹ · s¹ at pH 7.0. SOD speeds up this reaction almost 10⁴-fold (rate constant = 1.6 × 10⁹ M¹ · s¹) (79). This implies that any reaction of O₂· in aqueous solution will be in competition with SOD or, in its absence, with the spontaneous dismutation reaction itself. NO· reacts with O₂· at near diffusion-limited rates and is, therefore, one of the few biomolecules that is able to "outcompete" SOD for O₂· (119). Thus in most biological systems, unless sufficiently high concentrations of NO· or other similarly reactive molecules are present, generation of O₂· usually results in the formation of H₂O₂.

Although dismutation of O₂· probably accounts for much of the H₂O₂ produced by eukaryotic cells, H₂O₂ can also be formed by direct two-electron reduction of O₂, a reaction mechanism shared by a number of flavoprotein oxidases (186). Unlike O₂·, H₂O₂ is not a free radical and is a much more stable molecule. H₂O₂ is able to diffuse across biological membranes, whereas O₂· does not. H₂O₂ is a weaker oxidizing agent than O₂·. However, in the presence of transition metals such as iron or copper, H₂O₂ can give rise to the indiscriminately reactive and toxic hydroxyl radical (OH·) by Fenton chemistry.

	CELLULAR SOURCES AND REGULATION OF ROS

TOP ABSTRACT INTRODUCTION CHEMISTRY OF ROS CELLULAR SOURCES AND REGULATION... MEASUREMENT OF ROS IN... LIGAND-INDUCED ROS PRODUCTION... SIGNALING MOLECULES "TARGETED"... MECHANISMS OF ROS ACTION CONCLUSION REFERENCES

Cellular production of ROS occurs from both enzymatic and nonenzymatic sources. As stated earlier, any electron-transferring protein or enzymatic system can result in the formation of ROS as "by-products" of electron transfer reactions. This "unintended" generation of ROS in mitochondria accounts for ~1-2% of total O₂ consumption under reducing conditions (78). Due to high concentrations of mitochondrial SOD, the intramitochondrial concentrations of O₂· are maintained at very low steady-state levels (298). Thus unlike H₂O₂, which is capable of diffusing across the mitochondrial membrane into the cytoplasm (44), mitochondria-generated O₂· is unlikely to escape into the cytoplasm. The potential for mitochondrial ROS to mediate cell signaling has gained significant attention in recent years, particularly with regard to the regulation of apoptosis (25, 40, 46, 154, 166, 302). There is evidence to suggest that tumor necrosis factor (TNF)-- and interleukin (IL)-1-induced apoptosis may involve mitochondria-derived ROS (269, 273, 311). It has also been suggested that the mitochondria may function as an "O₂ sensor" to mediate hypoxia-induced gene transcription (45, 68).

The endoplasmic reticulum (ER) is another membrane-bound intracellular organelle that, unlike mitochondria, is primarily involved in lipid and protein biosynthesis. Smooth ER (lacking bound ribosomes) contains enzymes that catalyze a series of reactions to detoxify lipid-soluble drugs and other harmful metabolic products. The most extensively studied of these are the cytochrome P-450 and b₅ families of enzymes that can oxidize unsaturated fatty acids and xenobiotics and reduce molecular O₂ to produce O₂· and/or H₂O₂ (18, 43, 78). Although there does not appear to be a direct link between ER-derived oxidants and growth factor signaling, there is evidence for redox regulation of ER-related functions such as protein folding and secretion (22, 32, 120, 219). Bauskin et al. (32) showed that Ltk, a nonreceptor tyrosine kinase (non-RTK) expressed mainly in lymphocytes, leukemia cells, and neurons, is activated by forming disulfide-linked multimers in response to thiol-oxidizing agents. It has also been suggested that an O₂·-generating microsomal NADH oxidoreductase may function as a potential pulmonary artery O₂ sensor in pulmonary artery smooth muscle cells (198, 199).

Nuclear membranes contain cytochrome oxidases and electron transport systems that resemble those of the ER but the function of which is unknown (78, 102). It has been postulated that electron "leaks" from these enzymatic systems may give rise to ROS that can damage cellular DNA in vivo (102).

Peroxisomes are an important source of total cellular H₂O₂ production (37). They contain a number of H₂O₂-generating enzymes including glycolate oxidase, D-amino acid oxidase, urate oxidase, L--hydroxyacid oxidase, and fatty acyl-CoA oxidase. Peroxisomal catalase utilizes H₂O₂ produced by these oxidases to oxidize a variety of other substrates in "peroxidative" reactions (296). These types of oxidative reactions are particularly important in liver and kidney cells in which peroxisomes detoxify a variety of toxic molecules (including ethanol) that enter the circulation. Another major function of the oxidative reactions carried out in peroxisomes is -oxidation of fatty acids, which in mammalian cells occurs in mitochondria and peroxisomes (9). Specific signaling roles have not been ascribed to peroxisome-derived oxidants, and only a small fraction of H₂O₂ generated in these intracellular organelles appears to escape peroxisomal catalase (37, 230).

In addition to intracellular membrane-associated oxidases, soluble enzymes such as xanthine oxidase, aldehyde oxidase, dihydroorotate dehydrogenase, flavoprotein dehydrogenase and tryptophan dioxygenase can generate ROS during catalytic cycling (78). The most extensively studied of these is the O₂·-generating xanthine oxidase, which can be formed from xanthine dehydrogenase after tissue exposure to hypoxia (188, 222). Xanthine oxidase is widely used to generate O₂· in vitro to study the effect of ROS on diverse cellular processes; however, no studies have implicated a direct physiological role for endogenous xanthine oxidase in cell signaling.

Autooxidation of small molecules such as dopamine, epinephrine, flavins, and hydroquinones can be an important source of intracellular ROS production (78). In most cases, the direct product of such autooxidation reactions is O₂·. Although there is no known role for autooxidation of small molecules in growth factor and/or cytokine signaling, such reactions may induce oxidative stress and alter the overall cellular redox state. There is the suggestion that the prooxidant effects of dopamine autooxidation may be involved in the dopamine-induced apoptosis that is implicated in the pathogenesis of neurodegenerative diseases such as Parkinson's disease (214, 215, 320).

Plasma membrane-associated oxidases have been implicated as the sources of most growth factor- and/or cytokine-stimulated oxidant production (97, 143, 191, 253, 280, 293), although the precise enzymatic sources have yet to be fully characterized. The best characterized of the plasma membrane oxidases in general is the phagocytic NADPH oxidase, which serves a specialized function in host defense against invading microorganisms (reviewed in Refs. 20, 258). This multicomponent enzyme catalyzes the one-electron reduction of O₂ to O₂·, with NADPH as the electron donor through the transmembrane protein cytochrome b₅₅₈ (a heterodimeric complex of gp91^phox and p22^phox protein subunits). The transfer of electrons occurs from NADPH on the inner aspect of the plasma membrane to O₂ on the outside. During phagocytosis, the plasma membrane is internalized as the wall of the phagocytic vesicle, with what was once the outer membrane surface now facing the interior of the vesicle. This targets the delivery of O₂· and its reactive metabolites internally for localized microbicidal activity (20).

Recent studies (33, 81, 110, 127, 192, 200) have suggested that functional components of the phagocytic NADPH are present in nonphagocytic cells. Expression of p22^phox has been demonstrated in the adventitial smooth muscle cells of coronary arteries (19) and the aorta (183). Moreover, polymorphism of the p22^phox gene has been associated with an increased risk of coronary artery disease in young Caucasian men (84). Increased aortic adventitial O₂· production contributes to hypertension by blocking the vasodilatory effects of NO· (305). 3-Hydroxy-3-methylglutaryl-CoA reductase inhibitors appear to have a "cardioprotective" role not only by their effect on lowering cholesterol but also by inhibiting the activity of an O₂·-generating oxidase in vascular endothelial cells and, thereby, improving NO·-dependent vasodilation (303). Angiotensin II (ANG II)-induced hypertension is mediated, at least in part, by direct interactions between O₂·, generated by a NAD(P)H oxidase, and NO· in vascular smooth muscle cells (80). p22^phox has been shown to be a functional component of this ANG II-stimulated oxidase (183, 322). TNF- also stimulates O₂· production in vascular smooth muscle cells by a p22^phox-based NADH oxidase and appears to upregulate p22^phox gene expression in these cells (60). Both gp91^phox and p22^phox are expressed in pulmonary neuroepithelial cells where a NADPH oxidase-like enzyme may function as an O₂ sensor by activating O₂- and H₂O₂-sensitive K⁺ channels (304). However, it was later demonstrated that mice with chronic granulomatous disease (lacking a functional gp91^phox subunit) have preserved hypoxic vasoconstrictive responses and O₂ sensing, suggesting that this oxidase is not involved (14). More recently, Suh et al. (279) demonstrated that Mox-1, a gene encoding a homologue of the catalytic subunit of the phagocytic gp91^phox, is expressed in a number of tissues including vascular smooth muscle. In this study, Mox-1 expression in NIH/3T3 cells was associated with increased O₂· production, serum-stimulated cell growth, and a transformed phenotype. This study suggests that an O₂·-generating oxidase similar (but not identical) to the phagocytic NADPH oxidase is present in some nonphagocytic cells where it functions primarily as a regulator of cellular growth responses.

Cytochrome b₅₅₈, along with the cytosolic components p67^phox, p47^phox, and p40^phox, makes up the core protein subunits of the phagocytic NADPH oxidase complex. Activation of the oxidase, however, requires the additional participation of Rac2 (or Rac1 in mouse macrophages) and Rap1A, members of the Ras superfamily of small GTP-binding proteins. During activation, Rac2 binds GTP and migrates to the plasma membrane along with the cytosolic components to form the active oxidase complex. A requirement for Rac1 in the activation of the mitogenic oxidase has also been demonstrated in nonphagocytic cells (55, 128). The insert region in Rac1 (residues 124-135) appears to be essential for O₂· production and stimulation of mitogenesis in quiescent fibroblasts but not for Rac1-induced cytoskeletal changes or activation of Jun kinase (128).

Another GTP-binding protein, p21^Ras, appears to function upstream from Rac1 in oxidant-dependent mitogenic signaling (121, 281). Sundaresan et al. (281) showed that dominant negative expression of Rac1 inhibits not only the growth factor- and/or cytokine-generated rise in intracellular ROS in NIH/3T3 cells but also the ROS production in cells overexpressing a constitutively active isoform of Ras (H-Ras^V12). Stable transfection of the same Ras plasmid (H-Ras^V12) in fibroblasts induces cellular transformation and constitutive production of large amounts of O₂· (121). In this study, mitogenic signaling in Ras-transformed fibroblasts was demonstrated to be redox sensitive but independent of the mitogen-activated protein (MAP) kinase (MAPK) or c-Jun NH₂-terminal kinase (JNK) pathways. Results from our laboratory (292) demonstrate the requirement for p21^Ras in the generation of intracellular O₂· by mitogens such as platelet-derived growth factor (PDGF)-BB, fibroblast growth factor (FGF)-2, and transforming growth factor (TGF)- in nontransformed human lung fibroblasts. However, studies from our laboratory (292, 293) also suggest the presence of a distinctly different ROS-generating enzymatic system in these cells that is independent of p21^Ras regulation and that primarily generates extracellular H₂O₂ in response to TGF-1.

In contrast to these relatively recent reports of pyridine- and flavin-linked oxidase activities in nonphagocytic cells, enzymes involved in phospholipid metabolism have been known to exist for several decades. Membrane phospholipids, in addition to their structural role in providing membrane integrity, are substrates for the action of the phospholipases (PLs) PLA₂, PLC, and PLD. Although these enzymes are important for the generation of lipid second messengers, they have generally not been associated with ROS production in nonphagocytic cells. A recent report by Touyz and Schiffrin (297), however, suggests that ANG II-induced O₂· production in smooth muscle cells is dependent on the PLD pathway.

PLA₂ hydrolyzes phospholipids to generate arachidonic acid. Arachidonic acid then forms the substrate for cyclooxygenase- and lipoxygenase (LOX)-dependent synthesis of the four major classes of eicosanoids: prostaglandins, prostacyclins, thromboxanes, and leukotrienes. These synthetic pathways involve a series of oxidation steps that involve a number of free radical intermediates (78). Arachidonic acid metabolism, particularly involving the LOX pathway, which leads to leukotriene synthesis, has been reported to generate ROS (30, 170, 207, 272). LOX activity has also been implicated in redox-regulated signaling by ANG II (306), epidermal growth factor (EGF) (196), and IL-1 (36). There is the suggestion that LOX-derived lipid peroxidation products may be involved in the oxidative stress response to asbestos (74). TNF--induced apoptosis appears to be mediated by a LOX-dependent but ROS-independent mechanism (213). A lipid-metabolizing enzyme in fibroblasts similar to 15-LOX has been shown to generate large amounts of extracellular O₂· that appears to be independent of flavoenzyme activity (212).

	MEASUREMENT OF ROS IN BIOLOGICAL SYSTEMS

TOP ABSTRACT INTRODUCTION CHEMISTRY OF ROS CELLULAR SOURCES AND REGULATION... MEASUREMENT OF ROS IN... LIGAND-INDUCED ROS PRODUCTION... SIGNALING MOLECULES "TARGETED"... MECHANISMS OF ROS ACTION CONCLUSION REFERENCES

The high reactivity and relative instability of ROS make them extremely difficult to detect or measure in biological systems. Thus assessments of ROS and free radical generation have largely been made by indirect measurement of various end products resulting from the interaction of ROS with cellular components such as lipids, protein, or DNA (78, 231). Most methods for identification of specific ROS are based on reactions with various "detector" molecules that are oxidatively modified to elicit luminescent or fluorescent signals. Although detailed descriptions of these methods are beyond the scope of this review, we briefly discuss the limitations of the more commonly used assays.

Lucigenin has been used extensively as a chemiluminescent substrate for the detection of O₂· in many biological systems (97, 101, 121, 168, 235, 274). This is based on the ability of lucigenin (Luc²⁺) to be reduced to LucH·⁺, which can then react with O₂· to yield an unstable dioxetane compound that yields light (emission of photons) in the process of spontaneously decomposing to its ground-state electronic configuration (73). It has recently been demonstrated (171) that several enzymatic systems in vitro are capable of univalently reducing Luc²⁺ even in the absence of O₂· generation, resulting in the formation of LucH·⁺ that can autooxidize to generate O₂·. The ability of lucigenin to "redox cycle" in this manner, much like the effects of paraquat and nitro blue tetrazolium, is thought to be a major limitation in the use of this compound in the detection of O₂· (171). However, the importance of this redox cycling effect of lucigenin when applied to biological systems has been questioned by other investigators (6, 168, 274). The use of an alternative chemiluminescent probe, coelenterazine, may present less of a concern with regard to the potential redox-cycling effects of these compounds (287).

Another technique that is widely used for the detection of intracellular H₂O₂ production in response to growth factor stimulation is based on oxidation of the nonfluorescent substrate 2',7'-dichlorofluorescin (DCFH) to the fluorescent product 2',7'-dichlorofluorescein (DCF) (24, 196, 280). The esterified form of DCFH, dichlorofluorescin diacetate (DCFH-DA), is able to cross cell membranes and then, as a result of deacetylation by intracellular esterases, forms DCFH, which becomes "trapped" intracellularly. This property allows for assessment of intracellular oxidant production. It is becoming increasingly clear, however, that this assay lacks specificity and may be a better indicator of overall changes in the intracellular redox state of the cell than a direct measure of intracellular H₂O₂ production (122, 247). Recent work (244, 245) also suggests that similar to the problem with lucigenin, a series of free radical chain reactions with DCFH in the presence of peroxidase (and even in the absence of H₂O₂) may give rise to "artificial" DCF-dependent fluorescence and O₂ consumption.

Assessments of extracellular ROS production appear to be relatively more reliable and may allow for quantitative measurements. The method for detection of O₂· release with the method of SOD-inhibitable reduction of ferricytochrome c, initially described for neutrophils by Babior et al. (21), has been successfully used in other cell systems (187, 268). However, this method, although quite specific for O₂·, lacks sensitivity and may underestimate net O₂· production due to the potential reoxidation of ferricytochrome c by other oxidants (16).

Cellular production of extracellular H₂O₂ release can be measured by a fluorometric method that is based on the ability of heme peroxidases to catalyze H₂O₂-dependent dimerization of substituted phenolic compounds (221, 248). The major advantages of this method are its high specificity and sensitivity for H₂O₂ as determined with the use of horseradish peroxidase (used in most assays). The high sensitivity of such assays is related, at least in part, to the high reactivity of horseradish peroxidase (severalfold higher than catalase) with H₂O₂ (221). We (293) have been able to detect absolute concentrations of H₂O₂ in the 10⁸ M range and a steady-state release of extracellular H₂O₂ of ~22 pmol · min¹ · 10⁶ cells¹ from TGF-1-treated fibroblasts in cell culture using this method. Because the fluorescence of the dimerized product is dependent on pH, careful consideration must be made for possible pH changes and overoxidation of the substrate (221).

The only analytic method that, at least theoretically, measures free radicals directly is electron spin resonance spectroscopy. However, it is relatively insensitive and requires steady-state concentrations in the micromolar range for detection and is, therefore, of limited value in most biological systems. This problem may be circumvented by the technique of spin trapping, which involves the addition of a "spin trap" (usually a nitrone or a nitroso compound) that reacts rapidly with free radicals to form more stable radical adducts that accumulate to concentrations in the detectable range (115, 240). Thus all of the currently available methods for the detection of ROS or free radicals in biological systems, including electron spin resonance spectroscopy, are, by definition, indirect (because detector compounds are required) and semiquantitative (because specific ROS are in competition with other biomolecules that can react with these detector molecules). However, if the redox chemistries of the various detector molecules are carefully considered and the potential limitations of these assays are understood, they can provide useful information on the production and regulation of ROS in biological systems.

	LIGAND-INDUCED ROS PRODUCTION IN NONPHAGOCYTIC CELLS

TOP ABSTRACT INTRODUCTION CHEMISTRY OF ROS CELLULAR SOURCES AND REGULATION... MEASUREMENT OF ROS IN... LIGAND-INDUCED ROS PRODUCTION... SIGNALING MOLECULES "TARGETED"... MECHANISMS OF ROS ACTION CONCLUSION REFERENCES

A variety of cytokines and growth factors that bind receptors of different classes have been reported to generate ROS in nonphagocytic cells (Table 1). Currently available information on the enzymatic source(s) and physiological actions of ROS generated by these ligand-receptor interactions are summarized here (see Fig. 1 for 3 such pathways).

View larger version (31K): [in this window] [in a new window]   Fig. 1.   Proposed signaling pathways activated by specific ligand-mediated reactive oxygen species (ROS) production in nonphagocytic cells. A: mitogenic signaling by ROS in response to platelet-derived growth factor (PDGF) has been primarily studied in NIH/3T3 cells (121, 281), although a similar pathway may also be activated in vascular smooth muscle cells (280) and primary lung fibroblasts (292). One of the proposed mechanisms of this effect is ROS-mediated inactivation of protein tyrosine phosphatase (PTP)-1B (28, 163). EGF, epidermal growth factor; RTK, receptor tyrosine kinase. B: in primary lung fibroblasts, transforming growth factor (TGF)-1 induces Src kinase(s)-dependent myofibroblast differentiation and activation of a cell surface-associated H2O2-generating NADH oxidase (292, 293, 295). * In mouse osteoblastic cells, TGF-1-induced extracellular H2O2 has been shown to activate egr-1 in an autocrine manner (216). RS/TK, receptor serine/threonine kinase. C: in vascular smooth muscle cells, both angiotensin (ANG) II and serotonin [5-hydroxytryptamine (5-HT)] activate a NAD(P)H oxidase-like enzyme that generates intracellular ROS (97, 158). ROS-mediated activation of the p38 mitogen-activated protein kinase (MAPK) pathway appears to mediate ANG II-induced cell hypertrophy (299), whereas the extracellular signal-regulated kinase (ERK) 1/2 MAPK pathway mediates serotonin-induced cell proliferation (159). GPCR, G protein-coupled receptor; 5-HT TRP, 5-HT transporter. See text for details.

Cytokine receptors. Cytokine receptors fall into a large and heterogeneous group of receptors that lack intrinsic kinase activity and are not directly linked to ion channels or G proteins. Cytokines such as TNF-, IL-1, and interferon (IFN)- were among the first reported to generate ROS in nonphagocytic cells (187, 190, 191). There is no consensus on the specific species produced, the enzymatic source, or the site of generation of ROS for these cytokines. In the case of TNF-, recent reports (49, 169, 257, 270, 319) suggest that a mitochondrial source of ROS is required for activation of the transcription factor nuclear factor (NF)-B and NF-B-dependent gene transcription. There is also the suggestion that TNF- may activate NF-B by an ROS-independent mechanism and that TNF- and oxidants may produce a synergistic effect in this activation (126). TNF--induced generation of mitochondrial ROS has been implicated in apoptotic cell death (269). Overexpression of Mn SOD inhibits TNF--induced apoptosis, supporting a role for mitochondrial O₂· production in mediating this effect (181). The mechanism of TNF--induced apoptosis appears to be related, at least partially, to depletion of GSH, leading to redox-dependent formation of ceramide from sphingomyelin (172, 273). Gotoh and Cooper (94) showed that TNF- activates apoptosis signal-regulating kinase-1 (ASK-1), a member of the MAPK kinase kinase superfamily, by inducing oxidant-dependent dimerization of ASK-1. The resistance to apoptosis in some cancer cells may be related to the constitutive activation of NF-B by autocrine production of TNF- (86). ROS-dependent mechanisms also appear to be involved in TNF--induced expression of cell adhesion molecules (12, 147, 233, 318), IL-8 expression (147, 266), production of chemokines (27, 62), and induction of cardiac myocyte hypertrophy (206).

RTKs. A number of growth factors that bind RTKs have been shown to generate intracellular ROS essential for mitogenic signaling. A decade ago, PDGF was shown to stimulate cellular production of H₂O₂, which was thought to function as a "competence factor" in BALB/3T3 cell growth (264). Sundaresan et al. (280) demonstrated the ability of PDGF to transiently increase intracellular concentrations of H₂O₂, which was required for PDGF-induced tyrosine phosphorylation, MAPK activation, DNA synthesis, and chemotaxis. PDGF also activates the signal transducer and activator of transcription (STAT) family of transcription factors by a H₂O₂-dependent mechanism (271). Interestingly, H₂O₂ appears to antagonize the PDGF effect on disruption of gap junction communication by abrogating the phosphorylation of a gap junctional protein, connexin 43, by a tyrosine kinase- and redox-dependent mechanism (116). PDGF has also been shown to regulate gene expression by O₂·-dependent pathways. PDGF stimulates flavoenzyme-dependent O₂· generation in human aortic smooth muscle cells by PKC- and wortmannin-sensitive pathways (184). In this study, PDGF-induced O₂· production appears to participate in vascular lesion formation by activating NF-B and inducing monocyte chemoattractant protein-1 expression. PDGF-induced O₂· (but not H₂O₂) appears to be involved in its upregulation of inducible NOS- and NO·-dependent release of PGE₂ in fibroblasts (134).

Receptor serine/threonine kinases. All receptor serine/threonine kinases described to date in mammalian cells are members of the TGF- superfamily. TGF-1 is the prototype of this large family of polypeptide growth factors that bind receptor serine/threonine kinases and include the TGF-s, activins, inhibins, bone morphogenetic proteins, and Mullerian-inhibiting substance (185, 227). Each member of the TGF- superfamily activates a heteromeric complex of type I-type II receptors in a combinatorial manner, leading to the phosphorylation and/or activation of the recently described SMAD proteins that translocate to the nucleus to mediate gene transcription (185). Unlike RTK(s)-linked growth factors, TGF-1 typically inhibits growth of most target cells. Proliferative responses to TGF-1 appear to be primarily related to the indirect effects on the autocrine production of mitogenic growth factors (29, 164) or their receptors (241, 291). These multifunctional effects of TGF- on cellular growth regulation as well its diverse effects on cell differentiation and extracellular matrix production are critical in complex biological processes such as embryogenesis, fibrosis, and carcinogenesis.

G protein-coupled receptors. G protein-coupled receptors are the largest family of cell surface receptors, with >100 members characterized in mammals (9, 239). A number of ligands that bind to these receptors have been shown to generate ROS in different cell systems. Examples of these ligands include ANG II, serotonin [5-hydroxytryptamine (5-HT)], bradykinin, thrombin, and endothelin (ET).

Ion channel-linked receptors. Ion channel-linked receptors mediate rapid synaptic signaling between electrically excitable cells by transiently opening and closing an ion channel formed by the ligand-bound receptor complex. Such ligands typically are neurotransmitters (e.g., acetylcholine, 5-HT, glutamate, glycine, and -aminobutyric acid). Compared with the other classes of receptors discussed above, relatively little is known about ROS signaling by ion channel-linked receptors. However, there is growing interest in the role of ROS in mediating neuronal cell death by some excitatory neurotransmitters. Glutamate induces ROS production in neuronal cells by mechanisms that have been shown to be both dependent and independent of the changes in intracellular ion concentrations (35, 260). Work by Sensi and colleagues (260, 261) suggests that ROS generation in these cells is a result of intracellular Zn²⁺-mediated mitochondrial depolarization. Acetylcholine has been shown to activate ATP-sensitive K⁺ channels and increase mitochondrial ROS that may function as an intracellular signal for acetylcholine-induced preconditioning in cardiomyocytes (317).

	SIGNALING MOLECULES "TARGETED" BY ROS

TOP ABSTRACT INTRODUCTION CHEMISTRY OF ROS CELLULAR SOURCES AND REGULATION... MEASUREMENT OF ROS IN... LIGAND-INDUCED ROS PRODUCTION... SIGNALING MOLECULES "TARGETED"... MECHANISMS OF ROS ACTION CONCLUSION REFERENCES

Although a large number of signaling pathways appear to be regulated by ROS, the signaling molecules targeted by ROS are less clear. There is growing evidence, however, that redox regulation might occur at multiple levels in the signaling pathways from receptor to nucleus.

Receptor kinases and phosphatases themselves may be targets of oxidative stress. Growth factor receptors are most commonly activated by ligand-induced dimerization or oligomerization that autophosphorylates its cytoplasmic kinase domain (107). Ligand-independent clustering and activation of receptors in response to ultraviolet light have also been well demonstrated (243, 249), and this effect appears to be mediated by ROS (118, 226). Exogenous H₂O₂ (usually in the millimolar range) has been shown to induce tyrosine phosphorylation and activation of the PDGF-, PDGF-, and EGF receptors (82, 87, 91). Lysophatidic acid-induced activation of the EGF receptor appears to be mediated by the intermediate formation of ROS (57). A study by Knebel et al. (138) suggests that the mechanism of these effects may be related to ROS-mediated inhibition of the dephosphorylation of RTKs by inactivation of membrane-bound protein tyrosine phosphatase(s). A similar action of either O₂·- or NO·-generating stimuli on the phosphorylation-dephosphorylation balance leading to PDGF receptor activation has also been reported (41). However, to our knowledge, no studies have demonstrated the ability of endogenous, growth factor-stimulated ROS to regulate its own (or other) receptor(s) activation. On the other hand, H₂O₂ production in response to EGF has been shown to require a functional EGF receptor kinase domain (24), suggesting that growth factor-induced ROS may function downstream from EGF receptor activation. This does not exclude the possibility that, under certain pathological conditions associated with oxidative stress, ROS may directly activate cell surface receptors.

Because most growth factors and cytokines appear to generate ROS at or near the plasma membrane, phospholipid metabolites are potentially important targets for redox signaling. Takekoshi et al. (286) showed that the oxidized forms of diacylglycerol were more effective in activating PKC than its nonoxidized forms. PKC activation and protein tyrosine phosphorylation appear to be required for H₂O₂-induced PLD activation in endothelial cells and fibroblasts (197, 209). This effect may be mediated by the upstream activation of growth factor receptors because suramin (an inhibitor of receptor activation) blocks H₂O₂-induced PLD activation (208). EGF signaling of PLA₂ activation and arachidonic acid release is sensitive to antioxidants suggesting that PLA₂ may be a target of ROS (88).

Non-RTKs belonging to the Src family (Src kinases) and Janus kinase (JAK) family have been reported to be regulated by various forms of oxidative stress (2, 7, 271). As in the case of RTKs, these reports relate primarily to the effects of exogenously added oxidants, and the significance of such actions in specific growth factor- or cytokine-mediated signaling is less clear. Simon et al. (271) have demonstrated, however, that PDGF-induced activation of the JAK-STAT pathway is redox sensitive, suggesting a role for endogenous oxidant production in mediating this effect. Src kinases and JAKs appear to be involved in H₂O₂-induced activation of p21^Ras in fibroblasts (2). In another study, Lander et al. (150) have shown that p21^Ras may be a common signaling target of NO· and ROS. This effect may involve the recruitment of phosphatidylinositol 3'-kinase to the plasma membrane where it associates directly with the effector domain of p21^Ras before being activated (64).

ROS signaling by the induction of changes in intracellular Ca²⁺ ([Ca²⁺]_i) has been studied in a number of cell types including smooth muscle cells (reviewed in Ref. 285). Roveri et al. (246) showed that H₂O₂ induces a rapid increase in [Ca²⁺]_i that appears to be related to inositol 1,4,5-triphosphate-sensitive Ca²⁺ stores in the sarcoplasmic reticulum (SR) followed by a slower increase in [Ca²⁺]_i that is most likely derived from the extracellular space (246). Both O₂· (98, 283, 284) and H₂O₂ (99) have been shown to inhibit the activity of ATP-dependent Ca²⁺ of the SR, which would result in passive diffusion of SR Ca²⁺ into the cytosol. Such effects may also be more important in oxidative stress responses (179) than in receptor-mediated signaling by growth factors and/or cytokines.

The MAPKs comprise a large family of PKs that include ERK1 (p44^MAPK)/ERK2 (p42^MAPK), JNKs (also known as the stress-activated PKs), and p38 MAPKs. Because the MAPK pathways mediate both mitogen- and stress-activated signals, there has been significant interest in the redox regulation of these pathways. A number of groups (4, 23, 100, 195, 201, 277) have demonstrated the ability of exogenous oxidants to activate the ERK MAPK pathway. The mechanism(s) for this effect is unclear, and the precise molecular target(s) is unknown. Some studies suggest that ROS-mediated ERK activation may be an upstream event at the level of growth factor receptors (100), Src kinases (7), and/or p21^Ras (201). Another potential mechanism for this effect may be oxidant-induced inactivation of protein tyrosine phosphatases (PTPs) and/or protein phosphatase A (307). There are relatively fewer reports of endogenous ROS regulating the ERK MAPK pathway. However, Wilmer et al. (310) showed that IL-1-induced activation of ERK2 and JNK in human mesangial cells was inhibited by antioxidants, suggesting that ligand-stimulated ROS may be involved in mediating this effect. Also, serum- and 12-O-tetradecanoylphorbol-13-acetate-induced p21 induction appears to be mediated by redox-regulated ERK activation in HeLa cells (70). Results from our laboratory (159) suggest that redox regulation of the ERK pathway may be ligand and cell specific. 5-HT-induced ERK1/ERK2 activation in smooth muscle cells is redox sensitive, whereas FGF-2-stimulated activation of ERK1/ERK2 in lung fibroblasts is not (Thannickal et al., unpublished observations). Recent work also suggests that cyclin D1 expression, independent of the ERK MAPK pathway, may be a downstream target of redox-regulated mitogenic signaling (218). In vascular endothelial cells, cyclic strain-induced ROS can modulate egr-1 expression, at least partially, via the ERK signaling pathway (314).

Other members of the MAPK family have also been implicated as potential targets of ROS. Big MAPK-1 (BMK-1) appears to be much more sensitive than ERK1/ERK2 to H₂O₂ in several cell lines tested and suggests a potentially important role for BMK-1 as a redox-sensitive kinase (3). A study by Lo et al. (177) suggests that TNF-- and IL-1-induced activation of JNK may be mediated by intracellular H₂O₂. ANG II (306) and EGF (17) have also been reported to have similar effects. Generation of ROS in response to hypoxia-reoxygenation and ischemia-reperfusion mediates JNK activation in cardiac myocytes (144) and perfused rat heart (54). Although not directly related to the MAPK family, p66shc, a splice variant of the p52shc/p46shc protein involved in mitogenic signaling from activated receptors to p21^Ras, was found to be a target of H₂O₂ and responsible for mediating stress apoptotic responses (194).

NF-B, a transcription factor that regulates the expression of a number of genes involved in immune and inflammatory responses, has long been considered oxidant responsive (193, 256). Significant progress has been made on the mechanisms of activation of this transcription factor, and the regulation of upstream IB kinases that phosphorylate IB, a critical step in NF-B activation, has been recently characterized (130). However, the mechanism(s) by which ROS regulate this activity has remained elusive. Recent reviews (38, 167) on this subject have emphasized the importance of the recognition that a redox-dependent activation of NF-B is cell and stimulus specific as opposed to the concept that oxidative stress is a common mediator of diverse NF-B activators. Li and Karin (167) found that when a redox-regulated effect is observed, it appears to occur downstream from the IB kinases, at the level of ubiquitination and/or degradation of IB.

Activator protein-1 (AP-1), a transcriptional complex formed by the dimerization of Fos-Jun or Jun-Jun proteins, has also been demonstrated to be regulated by redox mechanisms. Both exogenous oxidants (11, 211) and ligand-induced ROS (176, 232) have been implicated in AP-1 activation. Unlike NF-B, however, AP-1 appears also to be activated by some antioxidants (52, 321). In vitro experiments suggest that a single cysteine residue in the highly conserved tri-amino acid sequence Lys-Cys-Arg of the DNA binding domain of Fos and Jun proteins confers redox sensitivity to AP-1 (1). It was suggested from this study that this single cysteine residue was unlikely to account for the formation of inter- or intramolecular disulfide bonds but rather for the possible conversion of this cysteine to a reversible oxidation product such as sulfenic acid. Further evidence suggests that the redox regulation of AP-1 DNA binding is facilitated by the reducing activity of redox factor-1 (Ref-1) protein that may act directly on this critical cysteine residue (315). Other transcription factors in which DNA binding is regulated by similar redox mechanisms include Sp-1 (313), c-Myb (203), p53 (234) and egr-1 (117).

	MECHANISMS OF ROS ACTION

TOP ABSTRACT INTRODUCTION CHEMISTRY OF ROS CELLULAR SOURCES AND REGULATION... MEASUREMENT OF ROS IN... LIGAND-INDUCED ROS PRODUCTION... SIGNALING MOLECULES "TARGETED"... MECHANISMS O