Correspondence to: Dr L. Kruidenier, Barts and The London, Queen Mary's School of Medicine and Dentistry, Department of Adult and Paediatric Gastroenterology, Suite 31, Dominion House, 59 Bartholomew Close, London EC1A 7BE, UK. E-mail: email@example.com
Virtually all inflammatory mediators investigated to date seem to be dysregulated in the inflamed intestinal mucosa of patients with inflammatory bowel disease. However, which of these are actually involved in the initiation and perpetuation of intestinal tissue damage is still not fully understood. Amongst these mediators are the reactive oxygen metabolites, produced in large amounts by the massively infiltrating leucocytes. These reactive oxygen metabolites are believed to constitute a major tissue-destructive force and may contribute significantly to the pathogenesis of inflammatory bowel disease.
This paper provides a concise overview of reactive oxygen metabolite biochemistry, the types of cell and tissue damage potentially inflicted by them, and the endogenous antioxidants which should prevent these harmful effects. An up-to-date summary of the available human experimental data suggests that reactive oxygen metabolite-mediated injury is important in both the primary and downstream secondary pathophysiological mechanisms underlying intestinal inflammation. Nonetheless, how the individual components of the mucosal antioxidant enzymatic cascade respond to inflammatory conditions is a neglected area of research. This particular aspect of intestinal mucosal oxidative stress therefore merits further study, in order to provide a sound, scientific basis for the design of antioxidant-directed treatment strategies for inflammatory bowel disease patients.
One of the major fundamental tissue-destructive mechanisms is oxidative stress through an excessive release of reactive oxygen metabolites (ROM).1–3 Numerous in vitro and in vivo studies have shown convincingly that ROM, a term used for all metabolites of molecular oxygen, including oxygen free radicals, are capable of causing, directly, reversible and irreversible damage to any oxidizable biomolecule. Consequently, they have been implicated in cell or tissue damage of practically every disease.4–12
In this review, we discuss why ROM are notoriously suspect in inflammatory diseases and, in particular, evaluate the evidence implicating oxidative stress in the pathogenesis of intestinal inflammation.
Reactive oxygen metabolites
Biochemistry and characteristics
Superoxide anion (O2–•). The primary ROM is superoxide anion (O2–•). O2–• is formed from the single-electron reduction of molecular oxygen, and is an oxygen free radical because it contains an unpaired electron. O2–• is generated through a variety of sources in both physiological and pathophysiological conditions. Probably most pertinent to the pathogenesis of intestinal inflammation, however, is its generation by neutrophils and macrophages.13–16 These cells, on interaction with pro-inflammatory agents, such as cytokines, immune complexes or bacterial products, undergo a so-called respiratory burst. This process involves a sudden stimulus-induced activation of the membrane-bound enzyme nicotinamide adenine dinucleotide phosphate (NADPH) oxidase, which in turn evokes the release of large amounts of ROM. Although it is beyond doubt that O2–• generation by phagocytes (and to a lesser extent by eosinophils, lymphocytes and fibroblasts) is essential for an effective host defence against bacterial infection, its continuous overproduction during inflammatory processes may also cause extensive tissue destruction.17
In addition to certain exogenous O2–• sources, such as cigarette smoke and ionizing radiation,1 that are particularly relevant in lung diseases8,11 and radiation pathologies,18 O2–• is generated intracellularly. In the mitochondrion, electrons continuously leak from their carriers within the respiratory chain, passing directly on to molecular oxygen. As much as 1–5% of the total oxygen consumption by normal tissues might thus be transformed into O2–•,19,20 which makes the mitochondrion the major endogenous intracellular O2–• production site under non-inflammatory conditions.21 Significant amounts of O2–• can also be generated by a variety of endogenous enzyme systems, such as the peroxisomal enzyme xanthine oxidase, which is activated on re-introduction of oxygen after periods of hypoxia (reviewed in Kooij22).
Despite the enormous potential for its production, O2–• itself is not considered as a particularly reactive intermediate.2,23–25 It does not rapidly cross lipid membrane bilayers and it dismutates spontaneously at physiological pH [reaction (1)]:
Paradoxically, however, the danger in O2–• lies in its neutralization. Reaction (1), accelerated by the superoxide dismutases (SOD; discussed below), is the first of a cascade of other oxidant reactions (illustrated in Figure 1), yielding much more powerful ROM, such as hydrogen peroxide (H2O2), hypochlorous acid (HOCl) and the hydroxyl radical (OH•).
Hydrogen peroxide (H2O2). Any system producing O2–• will also produce H2O2 and, consequently, inflammatory phagocytes also generate and release remarkable amounts of H2O2.13–16 Although H2O2 has been shown to directly exert non-specific irreversible damage to epithelial cells,26 it is generally considered as a relatively weak ROM. Its high in vivo reactivity, therefore, is not only attributed to its stability and diffusibility, but particularly to its ability to react with partially reduced metal ions, such as Fe2+ or Cu+, to form OH• in the so-called Fenton reaction [reaction (2)]:27
The formation of OH• from H2O2 can be bypassed through the two-electron reduction of H2O2 to water, catalysed by catalase (CAT) [reaction (3)] or glutathione peroxidase (GPO) [reaction (4)]:
Hypochlorous acid (HOCl). Instead of being neutralized to water, H2O2 can also be metabolized by the enzyme myeloperoxidase to form the potent chlorinating as well as oxidizing agent HOCl (commonly known as bleach) [reaction (5)]:
This reaction is specifically considered to be relevant in inflammatory processes, as the haemoprotein myeloperoxidase is one of the most abundant proteins in phagocytes. It is estimated to represent 5% of neutrophil protein and 1% of monocyte protein, and myeloperoxidase is believed to be present in human macrophages as well.28,29 When activated, neutrophils can secrete myeloperoxidase extracellularly.30
HOCl is estimated to be 100–1000 times more toxic than O2–• or H2O2, and seems to have distinct biochemical targets.31 For example, it is capable of inactivating essential enzymes,31 of oxidizing plasma membrane thiol (SH) groups,32 of disrupting certain protein and plasma membrane functions33 and of decreasing the adhesive properties of some extracellular matrix components.34 In addition, exposure to HOCl seems to increase endothelial permeability through the mobilization of cellular Zn2+ molecules.35 No specific HOCl neutralizing enzyme has been found as yet, but HOCl can be removed by reaction with albumin and ascorbic acid.36
Hydroxyl radical (OH•). OH• is considered to be the most reactive ROM.24,37 In contrast with H2O2, for instance, OH• inactivates the pivotal mitochondrial enzyme pyruvate dehydrogenase,38 depolymerizes gastrointestinal mucin and directly inflicts DNA damage.39,40 OH• is formed from H2O2 through the Fenton reaction [reaction (2)] or from O2–• through another transition metal-dependent reaction, called the iron-catalysed Haber–Weiss reaction [reaction (6)]:41
Recently, OH• has been shown to be produced via certain alternative, but inflammation-relevant, pathways. They include the generation of OH• during the inactivation of Cu/Zn-SOD by H2O2,42 and through the interactions between O2–• and HOCl [reaction (7)],43 HOCl and reduced iron ions [reaction (8)]44 and H2O2 and nitric oxide (NO) [reaction (9)]:45
Again, no known enzyme exists to facilitate the detoxification of OH•. OH•-induced tissue damage may be prevented, however, by the binding (‘sequestration’) of transition metal ions by, for instance, albumin, caeruloplasmin, ferritin, transferrin and metallothionein.1
Reactive nitrogen metabolites (RNM). In analogy with ROM, reactive nitrogen metabolites (RNM) are derived from nitrogen, and include NO and peroxynitrite. In recent years, considerable interest has been shown in the role of RNM in cellular redox reactions, and an overwhelming amount of papers have been published to describe their role in inflammation. Although NO is a highly lipid-soluble molecule with a considerably long half-life, capable of diffusing several cell diameters from its site of synthesis, there is still much debate as to whether NO production is actually harmful to inflamed tissues (reviewed in Grisham et al.46 and Kubes and McCafferty47). It is also unclear whether the NO-producing enzyme, nitric oxide synthase, is expressed by inflammatory cells.
NO itself probably is not particularly noxious and may even have some beneficial, inflammation-reducing, effects (reviewed in Lefer and Lefer48). For example, NO has been shown to protect epithelial cells against H2O2-mediated toxicity,49,50 to protect macrophages from cytokine-induced cytotoxicity51 and to diminish leucocyte binding to endothelial cells.52 In addition, through its very rapid interaction with O2–•[reaction (10)], NO may form a sink for O2–• and its toxic downstream ROM derivatives.53–55
However, the product of reaction (10), the peroxynitrite anion, is generally believed to be considerably more reactive and damaging than its precursors.56 Moreover, it has a relatively long lifetime and passes easily through lipid bilayers.57 Peroxynitrite oxidizes mitochondrial membrane lipids,58 is believed to damage sodium channels in the colon59 and increases microvascular permeability and oedema formation.60 Peroxynitrite formation, which depends on the existence of sufficient levels of available O2–• and NO, may be self-limiting through the capability of NO to inactivate NADPH oxidase, the enzyme responsible for phagocytic O2–• production.61 On the other hand, nitric oxide synthase seems to be capable of producing O2–• itself.62 In the absence of a specific peroxynitrite detoxifying mechanism, proteins such as haemoglobin and neutrophilic myeloperoxidase, which are abundantly present during inflammation, are thought to play a major defensive role against peroxynitrite.63
Excessive generation of ROM and RNM may, in principle, result in the attack and damage of all cellular and extracellular components (see Figure 1). It is important to realize, however, that the overall high reactivity and short half-life of ROM implies that the tissue damage they inflict is generally close to the site of ROM generation. In addition, as stated by McCord,3 reactivity and toxicity are not synonymous. A particular ROM may not be particularly bioreactive, but may still be extremely toxic when it strikes a crucial target effectively.
Cellular targets of oxidative attack
Membrane lipids. The polyunsaturated fatty acids located within the cell membrane lipid bilayer are major targets for ROM attack.1,64 They are particularly effectively attacked by OH•, thereby initiating the process of lipid peroxidation. Once initiated, lipid peroxidation continues as a chain reaction to generate lipid hydroperoxides and aldehydes, and a single oxidative event can thus affect many lipid molecules. The accumulation of hydroperoxides in the cell membrane has a profound effect on its fluidity and, as such, on the activity of transmembrane enzymes, transporters, receptors and other membrane proteins.65,66 As a result, lipid peroxidation causes changes in membrane permeability and selectivity, and ultimately leads to alterations in cell volume homeostasis and cellular metabolism.67 Moreover, hydroperoxides and aldehydes are directly toxic to cells and organelles,68 have neutrophil chemotactic properties69 and may regulate cytokine production.70
Lipid peroxidation accelerates only when cellular detoxification systems have failed to remove the precursors of OH•, in particular H2O2, effectively. Once initiated, lipid peroxidation is most successfully combated by lipid-soluble antioxidants such as α-tocopherol (vitamin E),2 although it has also been shown that NO can act as a chain-breaking antioxidant against lipid peroxidation.71
Proteins. Proteins are the most abundant cell constituents, which make them important ROM targets.72 Moreover, a relatively minor structural (oxidative) modification of a single protein can lead to a marked change (in most cases lowering) in its biological activity. Similar to lipid peroxidation, OH• seems to be most effective in inducing oxidative protein damage.72 The process of protein oxidation frequently introduces new functional groups, such as hydroxyls and carbonyls, which contribute to altered function, turnover and degradation.73 Secondary effects include protein fragmentation, cross-linking and unfolding.73
Another permanent modification that can adversely affect protein function is the nitration of protein-bound tyrosines by peroxynitrite.74,75 Tyrosine is an important amino acid, involved in (de)phosphorylation reactions and signal transduction pathways,72 and its nitration may not only compromise protein function, but may also have serious consequences in cellular regulation. It is of particular interest that the enzymes (SOD) which catalyse reaction (1) have been identified as specific protein targets of nitration. Nitration of cytoplasmic Cu/Zn-SOD has been reported to occur in vitro without loss of enzymatic activity,76 whereas tyrosine nitration of mitochondrial Mn-SOD is associated with the loss of enzyme function, and has been detected in rejected human kidney allografts.77 It should be mentioned, however, that protein nitration and inactivation might also occur through a peroxynitrite-independent, myeloperoxidase-catalysed, pathway.78,79
DNA. Both nuclear and mitochondrial DNA are known targets of ROM attack,80,81 which can result in many types of DNA modification. The most common are base hydroxylations and strand cleavage, leading to adenosine triphosphate depletion and gene mutations, which can in turn result in malignant transformation or cell death. O2–• is relatively unreactive with DNA,82 and other ROM implicated in DNA damage are peroxynitrite83,84 and NO, which may directly damage chromatin.85 The most substantial portion of DNA modifications, however, is thought to involve in situ produced OH• as the attacking species.39,40 Once H2O2 escapes from its cytosolic neutralizing enzymes and reaches the nucleus, it will react with chromatin-bound iron (or copper released from oxidatively damaged Cu/Zn-SOD),42,86 producing OH•in situ, which in turn will attack nearby DNA residues.87 Substantial protection against OH•-mediated DNA damage may come from the thiol-rich protein metallothionein (see below). Metallothionein is not only a powerful OH• scavenger,88 it also accumulates in the cell nucleus during certain phases of the cell cycle.89
Apoptotic cell death. Cell death can follow two distinct pathways: necrosis or apoptosis.90 Apoptosis (or controlled cell death) differs from necrosis (chaotic cell death due to overt injury) by distinct morphological and biochemical features, such as chromatin condensation, membrane surface blebbing, DNA fragmentation and, finally, the breakdown and autodigestion of the cell into a series of smaller units (apoptotic bodies). Although the early biochemical events that dictate the mode of cell death are still unclear, several lines of evidence implicate ROM as modulators of apoptosis.91In vitro exposure to low doses of ROM, or a depletion of cellular antioxidants, has been shown to result in apoptosis,92–97 and, conversely, apoptosis can be blocked by the addition of antioxidant compounds.98–101
Reactive oxygen metabolites as immunomodulators
The traditional view of ROM as directly causing non-specific injury to cells through a series of local chemical reactions has, in recent years, been complemented by numerous in vitro reports that show more subtle, immunomodulatory effects of ROM. O2–•, for example, has been shown to mediate the infiltration and accumulation of neutrophils at sites of inflammation,102,103 and to be involved in the mobilization of arachidonic acid.104 Furthermore, H2O2 appears to act as a neutrophil chemoattractant,105 to elicit leucocyte rolling,106 to activate T lymphocytes107 and to induce angiogenesis.107 H2O2, like O2–•, is also capable of mobilizing arachidonic acid.104
Most of these oxidant effects are thought to be mediated by shared, redox-sensitive, regulatory pathways, which are discussed in detail elsewhere.108 Apparently, ROM are involved in the (in)activation of a variety of kinases and transcription factors, their up- or down-regulation depending on the magnitude of the redox change. In particular, the transcription factors nuclear factor-κB and activator protein-1 have received much attention in this respect. They are activated during intestinal inflammation in epithelial and inflammatory cells,109 where they lead to the up-regulation of a number of inflammatory genes, including those encoding tumour necrosis factor-α, interleukin-1, -6 and -8, inducible nitric oxide synthase, major histocompatibility complex class I antigens and the adhesion molecules E-selectin and vascular cell adhesion molecule-1.108,110 Redox-mediated, nuclear factor-κB/ activator protein-1-induced gene expression may be of particular relevance in the context of chronic inflammatory processes. Some of the cytokines whose genes are switched on by nuclear factor-κB, such as tumour necrosis factor-α and interleukin-1, are themselves activators of nuclear factor-κB.111 Moreover, interleukin-1 and tumour necrosis factor-α are also known to induce cellular ROM production,112–114 and ROM, in particular H2O2, have been shown to activate nuclear factor-κB.111,115,116 Evidence exists that ROM and pro-inflammatory cytokines work synergistically to further intensify transcription factor activation.111,117 In this context, it is also of interest to note that another major tissue-destructive force, an extensive group of proteases called the matrix metalloproteinases,118,119 also appears to be under ROM control, probably through the same regulatory pathways. In fact, the genes that code for matrix metalloproteinases and their inhibitors are modulated by activator protein-1,120,121 and ROM at concentrations achieved at sites of inflammation can activate multiple matrix metalloproteinase family members122,123 and at the same time inactivate their inhibitors,124 thereby allowing uncontrolled protease and collagenase activity to damage tissues. Such ROM-/cytokine-/transcription factor-regulated self-sustaining regulatory loops may contribute to the perpetuation and exacerbation of chronic inflammation and tissue damage, particularly when the local immune response is ‘hyper-responsive’ and fails to successfully down-regulate the immune reaction. The inflammatory bowel diseases show signs suggestive of such mechanisms.125
Inflammatory bowel disease
The expression of the inflammatory bowel diseases, Crohn's disease and ulcerative colitis, is generally thought to depend on the interplay of environmental, genetic and immunological factors (reviewed in Fiocchi125). Yet, an integrated concept explaining the initiating event(s) and/or fundamental abnormalities in inflammatory bowel disease in relation to the pathophysiological changes has not yet emerged. It does seem certain that inflammatory bowel disease is amplified and propagated by an uncontrolled and sustained host immune response (reviewed in Brandtzaeg et al.126), as the disease is paralleled by an extensive inflammatory infiltrate in the lamina propria, consisting of polymorphonuclear neutrophils, eosinophils and plasma cells. However, the final steps leading from such an excessive and enduring mucosal immune activation to tissue injury are still not fully understood. Ultimately, only a limited number of effector mechanisms, including ROM, might be responsible for the excessive cellular/tissue damage, chronic inflammation and destruction of normal tissue that is observed in inflammatory bowel disease.
Evidence of increased reactive oxygen metabolite levels in inflammatory bowel disease
The chronic presence of numerous, activated, myeloperoxidase-containing phagocytes in the inflamed intestine of inflammatory bowel disease patients implies a prolonged and intense mucosal exposure to an arsenal of toxic and damaging agents, including ROM and RNM. Yet, attempts to directly quantify ROM/RNM levels in the inflammatory bowel disease intestinal mucosa have been limited. The main reason for this lies undoubtedly in the technical difficulty of measuring ROM and RNM directly in cells and tissues, due to their short biological half-lives.127 Over the years, several easy-to-use techniques have been developed, including histochemistry128–132 and colorimetric assays,133–136 but they all lack sufficient sensitivity and specificity. More reliable techniques, such as electron spin resonance spectroscopy136,137 and chemiluminescence,138 involve highly specialized and expensive laboratory equipment. Moreover, most of such methods are conducted ex vivo, and their extrapolation to the in vivo situation has been questioned.139
Nevertheless, several laboratories have assessed ROM levels directly in inflammatory bowel disease by applying chemiluminescence techniques in colonic biopsy specimens of ulcerative colitis and Crohn's disease patients.140–143 Their results were highly consistent: when compared with normal control mucosa, ROM production was considerably increased in the inflammatory bowel disease samples, was positively correlated with inflammatory bowel disease activity and appeared to be neutrophil-derived. In analogy, direct NO measurements in ulcerative colitis and Crohn's disease patients revealed greatly increased colonic concentrations.144–146 Again, these could be correlated with clinical and endoscopic indices of disease activity. Additional evidence of increased NO production in inflammatory bowel disease has been obtained indirectly, through the analysis of the mucosal levels of the NO-producing enzyme nitric oxide synthase. In numerous studies, nitric oxide synthase activity and protein levels were reported to be up-regulated in the inflamed inflammatory bowel disease mucosa.144,147–157 Thus, both direct and indirect measurements of ROM/RNM levels in inflammatory bowel disease strongly suggest an increased mucosal production of and exposure to O2–•, NO or their downstream metabolites.
Evidence of oxidative damage in inflammatory bowel disease
Excessive inflammatory ROM production in the gut has been held responsible for the enhanced electrolyte and water secretion, culminating in diarrhoea, in inflammatory bowel disease patients.158 However, experimental evidence of in vivo oxidant injury in inflammatory bowel disease is scarce, non-circumstantial and generally lacks causative importance. Most has been obtained from drug-induced intervention studies aimed at reducing the generation or effects of ROM in various animal models of intestinal inflammation and in inflammatory bowel disease patients (reviewed in Kruidenier and Verspaget,159 and see below). In fact, direct evidence has been provided in one study only, in which McKenzie et al. showed that the loss of glyceraldehyde-3-phosphate dehydrogenase enzyme activity in colon epithelial crypt cells, harvested directly from inflamed lesions of Crohn's disease and ulcerative colitis patients, resulted from oxidation by HOCl.160
Instead, measurement of oxidative injury in vivo, and hence in human inflammatory bowel disease, has commonly relied on the assessment of oxidatively modified marker molecules.40, 127, 132, 161, 162 The increased levels of malondialdehyde and 4-hydroxynonenal, found in colonic biopsies from inflammatory bowel disease patients,163,164 for instance, provide evidence for excess lipid peroxidation reactions. This is also true of the increased breath ethane and pentane excretion in these patients,165–167 which are non-invasive markers of lipid peroxidation, and have been correlated with disease activity in most cases. With regard to protein damage, the carbonyl content of cells/tissues has been widely used as a convenient marker of oxidative protein damage,168 whereas the nitration of tyrosine residues in proteins to 3-nitrotyrosine is an indication of the presence of peroxynitrite-modified proteins.56,74 In colonic biopsies from Crohn's disease and ulcerative colitis patients, the protein carbonyl content has been reported to be increased,143 as has the immunohistochemical expression of 3-nitrotyrosine in mainly the lamina propria mononuclear cells of both Crohn's disease and ulcerative colitis mucosa.149,152,154 Obvious correlations with disease activity, however, have never been established. Mucosal DNA oxidation in human inflammatory bowel disease has been evaluated in one study only, in which the DNA oxidation product, 8-hydroxy-2′-deoxyguanosine, was found to be increased in Crohn's disease biopsies.143 More information is available regarding the role of apoptosis in inflammatory bowel disease.169 Markers of this mode of cell death include Fas and Fas-ligand, and DNA strand breaks detected by electrophoresis or immunohistochemistry. From a number of studies, the pattern emerges that, in the inflammatory bowel disease mucosa, apoptosis is increased in the epithelium,170–172 but decreased in T lymphocytes and neutrophils.173–176 These disturbances may have important pathogenetic implications, because they may lead to an increase of epithelial turnover and the accumulation of inflammatory cells,169 thereby hampering immune down-regulation. To date, however, no evidence exists to indicate that the defective apoptosis in inflammatory bowel disease is directly mediated through excessive ROM production.
All of these findings strengthen the concept that an uncontrolled, excessive ROM/RNM production and/or disturbances of the redox status in inflammatory diseases, such as inflammatory bowel disease, may have a considerable impact on the course of the disease. On the one hand, ROM and RNM pose a serious threat of deleterious effects by oxidizing and damaging important cellular structures and macromolecules, and, on the other, they can have a profound effect on the expression of a variety of immune and inflammatory molecules, thereby influencing the recruitment and activation of inflammatory cells and tissue-destructive mechanisms. Obviously, it is an absolute necessity for cells and tissues to carefully regulate ROM/RNM levels, even more so under inflammatory conditions. To do so, cells are equipped with an elaborate antioxidant defence system.2 Roughly, the components of this system can be categorized into non-enzymatic and enzymatic antioxidants.
This group includes several dietary compounds with antioxidant properties, which normally originate from natural sources, such as fruits, vegetables and plant extracts (reviewed in Aruoma177). In particular, certain minerals (e.g. zinc), vitamins (C and E) and the flavonoids found in these extracts are considered to be of prime interest in this context, and have also been exploited commercially.2,177
In addition, the human gut naturally contains and/or produces a variety of non-enzymatic antioxidant defences.2 These include water-soluble agents, such as glutathione, metallothionein, ascorbic acid (vitamin C), uric acid and some plasma proteins, as well as lipid-soluble defences, such as α-tocopherol (vitamin E), bilirubin and ubiquinol (reduced coenzyme Q10).
Glutathione. Reduced glutathione, a tripeptide with a reactive sulphydryl group, can act on multiple levels of antioxidant defence. Apart from being a substrate for the antioxidant enzyme GPO (see below), glutathione serves as a scavenger of several ROM, including O2–•, OH•, peroxynitrite and lipid hydroperoxides.2,178 In addition, it has been shown to be involved in the direct repair of oxidative DNA lesions, and in the protection against ROM- or cytokine-induced apoptosis.94,179 During its antioxidant function, reduced glutathione is converted to its oxidized state, upon which it is reduced back again by glutathione reductase.
Metallothionein. The ‘sequestration’ of free metal ions is an important antioxidant defence mechanism, preventing OH• formation via the Fenton and Haber–Weiss reactions [reactions (2) and (6)]. Metallothionein is a small thiol-rich protein that effectively binds potentially harmful metals, such as copper and zinc (reviewed in Davis and Cousins180). As a result of its high thiol content, however, an additional asset of metallothionein function is its capability to directly scavenge OH•.181 These characteristics make metallothionein an effective inhibitor of O2–•-, H2O2- and peroxynitrite-dependent lipid peroxidation,182,183 and its nuclear localization enables metallothionein to protect against oxidative DNA damage and apoptosis.88,183–185
Localization studies of metallothionein in the normal gut revealed it to be expressed in the cytoplasm and nucleus of luminal and crypt enterocytes.186–188 Metallothionein synthesis is related to cellular zinc homeostasis,180 but also seems to be under inflammatory cytokine control. The metallothionein gene contains activator protein-1 binding sequences, and inflammatory cytokines, such as interleukin-1, interleukin-6, tumour necrosis factor-α and interferon-γ, have been shown to induce metallothionein.180,189 Metallothionein is consumed during OH• scavenging,181 and can be damaged on exposure to O2–• and HOCl through the mobilization of zinc.190
Probably the most important component of the endogenous defence system against ROM-mediated damage constitutes an intricate and highly conserved enzymatic machinery that exists in all mammalian cells. The enzymes SOD, CAT and GPO form the backbone of this enzymatic antioxidant cascade (see Figure 1).
Superoxide dismutases. Soon after the discovery of SOD activity by McCord and Fridovich in 1969,191 it became clear that this enzyme is absolutely necessary to maintain life in aerobic organisms.19 As mentioned before, SOD detoxify O2–• by converting it to H2O2[reaction (1)] in what appears to be the fastest enzyme-catalysed reaction known.192 In consequence, loss of SOD function might induce cellular oxidative toxicity through an increase in O2–• levels or, more importantly, in its downstream metabolites, such as OH• or peroxynitrite. In humans, three forms of SOD have been identified, each with distinct distribution and metal components.19
Cu/Zn-SOD is a cyanide-sensitive homodimer of approximately 32 kDa, which is diffusely located throughout the cytoplasm and, to a lesser extent, in the nucleus, but is absent in mitochondria.193,194 It is by far the most abundant SOD isoform, constituting approximately 70% of the total SOD activity,195 and can be found in the epithelium and all types of phagocytes in most organs.196 Apart from O2–•, Cu/Zn-SOD can also accept H2O2 as a substrate to interact with reduced copper (Cu+). This reaction reduces H2O2, forming OH• adducts of nearby target molecules (such as Cu/Zn-SOD itself).42 In addition, Cu/Zn-SOD can accept peroxynitrite as a substrate,76 forming nitronium ions that can subsequently transfer nitrate groups to tyrosines on various proteins. The dual role of SOD as an O2–• scavenger and an H2O2 producer is increasingly being recognized.197 In humans, overexpression of Cu/Zn-SOD is thought to be associated with Down's syndrome pathogenesis,198 and the Cu/Zn-SOD gene has been linked to the motor neuron degenerative disorder amyotrophic lateral sclerosis.199
The second SOD isoform Mn-SOD is a homotetramer of 96 kDa containing one manganese atom per subunit;200 it is exclusively located in the mitochondria.194 Mn-SOD constitutes approximately 15% of the total SOD activity in most tissues,195 where it has been detected in epithelial cells as well as in phagocytes.200 The importance of this particular SOD isozyme and the potential toxicity of mitochondrially produced O2–• have been effectively illustrated in Mn-SOD knockout mice, which die within several days after birth.201 Cu/Zn-SOD knockouts and extracellular-SOD knockouts, on the contrary, survive quite well until they are stressed.202,203
Extracellular-SOD is the third, and most recently described, SOD family member.204 It is the dominant SOD isoform in the plasma and the interstitium,194 and is the only known extracellular enzyme to scavenge O2–•. It is a secretory, tetrameric, copper- and zinc-containing glycoprotein of 135 kDa with a high affinity for glycosaminoglycans, such as heparin.204 Due to this heparin affinity, it may exist at very high concentrations in unique extracellular compartments, whereas it is estimated to make up only 0.5–17% of the total SOD activity in tissues.195 In the human lung, extracellular-SOD has a specific distribution in the connective tissue matrix and smooth muscle cells around larger vessels and airways.205 Based on these findings, it has been speculated to contribute to the protection of collagen matrix elements against ROM, or to be involved in the modulation of vascular tone by the prevention of NO conversion to peroxynitrite.206
Catalase and glutathione peroxidase. Although CAT and GPO share their substrate H2O2[reactions (3) and (4)], both enzymes have certain distinct features. CAT is one of the most efficient enzymes known: it cannot be saturated by H2O2 at any concentration. In mammalian cells, CAT is largely contained in peroxisomes207 and some of it seems to be secreted. GPO performs the role of CAT in the extracellular environment. It is a largely selenium-dependent enzyme that is predominantly found in the cytoplasm, but also in mitochondria and peroxisomes.207 At least five GPO isozymes are present in mammals, some of which have been localized to mature intestinal absorptive epithelial cells.208
Functionally, GPO has a much higher affinity for H2O2 than CAT,209 and only GPO can react effectively with lipid hydroperoxides210 and prevent peroxynitrite-mediated oxidations.211 In human inflammatory cells, GPO and its substrate glutathione are preferably found in monocytes compared to neutrophils, whereas CAT is found in much higher levels in neutrophils than in monocytes.196 All of these observations concur with various in vitro data,212,213 which indicate that the glutathione/GPO redox cycle acts as the primary defence against a low, continuous exposure to H2O2, whereas CAT becomes more significant in conditions of acute, severe oxidative stress.
Regulation of antioxidant enzyme expression
Expression regulation. Only a balanced, co-ordinate action of these three antioxidant enzymes secures a low steady state concentration of ROM in the cell, and hence their activities need to be very precisely regulated. The regulation of SOD isozyme expression has been investigated in numerous in vitro and in vivo studies. All indicate that Cu/Zn-SOD is expressed constitutively, whereas Mn-SOD is highly inducible. Cu/Zn-SOD gene transcription seems to be under the regulation of Sp1-related factors,214 and is marginally affected by exposure to cytokines,188,215–219 oxidants220–222 or other forms of stress, such as ultraviolet irradiation.218 In macrophages, Cu/Zn-SOD activity can be induced by thyroid hormones and insulin.223 The Mn-SOD gene contains consensus sites for nuclear factor-κB and activator protein-1 transcription factor binding,224–226 which makes it easily inducible under inflammatory conditions. Profound up-regulation of Mn-SOD expression has been demonstrated following exposure to various oxidants,220–222,227,228 tumour necrosis factor-α,188, 215–219, 226, 229, 230 interleukin-1,215–217,226 interleukin-6,218 interferon-γ,216,231 irradiation218 or endotoxins.215,226,232 Insulin and dexamethasone have been shown to reduce Mn-SOD activity in both macrophages and intestinal epithelial cells.223,233 The expression of extracellular-SOD, like that of Cu/Zn-SOD, is not influenced by its substrate or other ROM.221 It can, however, be elevated by interferon-γ and is depressed by interleukin-1, tumour necrosis factor-α and particularly transforming growth factor-β.216
Much less information is available concerning the expression regulation of the H2O2-metabolizing enzymes CAT and GPO. Their expression levels were induced after exposure to H2O2,220,222 but treatment with O2–• or tumour necrosis factor-α had no effect.188,219,220 Insulin increased the activities of macrophage CAT and GPO, whereas thyroid and glucocorticoid hormones reduced GPO activity.223
Autocatalytic inactivation. Any antioxidant enzyme introduced to a site of inflammation will be susceptible to attack by the ROM and RNM present. Indeed, in vitro experiments have shown that practically all ROM are capable of inactivating one or several antioxidant enzymes. Exposure to O2–•, for instance, led to a rapid inhibition of CAT,234,235 but not of GPO activity,235 and H2O2 depressed the activities of Cu/Zn-SOD42,236 and GPO.235 Low concentrations of neutrophil-derived HOCl very quickly inactivated GPO, whereas CAT was inactivated at higher HOCl levels and Cu/Zn-SOD was not easily affected by HOCl at all.237 Similarly, OH• has been reported to inhibit CAT and GPO activities, but not Cu/Zn-SOD.237 As far as RNM are concerned, NO has been shown to inactivate CAT and GPO.238 On the other hand, it was capable of rescuing Cu/Zn-SOD from H2O2-mediated inactivation.236 Exposure to peroxynitrite effectively inactivated both Mn-SOD239 and GPO.240
Antioxidant enzyme levels in inflammatory bowel disease
The differences in the regulation of expression between SOD, CAT and GPO may not only reflect their different roles in normal physiology, but may also endanger the efficient removal of ROM under inflammatory conditions such as inflammatory bowel disease. A decrease in antioxidant enzyme activity, or an unbalanced overexpression of one of these enzymes, may increase the vulnerability of cells to ROM.241 For instance, an increase in SOD would deplete the cell of O2–•, but would increase H2O2 production, which might be deleterious unless sufficient CAT or GPO was available. Likewise, excess GPO could unnecessarily deplete glutathione and/or NADPH reserves, even though CAT was present. This concept of an imbalanced antioxidant enzyme response has been appreciated in several (inflammatory) diseases.242–244
In inflammatory bowel disease, the endogenous intestinal expression of SOD isozymes has been investigated in a limited manner (see Table 1). Focusing on Cu/Zn-SOD only, decreased SOD protein and activity levels have been reported in resected mucosa from inflammatory bowel disease patients with active disease,251 and in inflamed mucosal biopsies from Crohn's disease and ulcerative colitis patients.143,254 No information is available regarding the other SOD isoforms, Mn-SOD and extracellular-SOD. Intestinal CAT and GPO activity levels in inflammatory bowel disease have been evaluated in only a handful of studies. These enzymes do not seem to be affected by the inflammatory process,255–258 despite the fact that some smaller studies have reported an increased GPO activity in inflammatory bowel disease mucosa.245,252
Table 1. Reported changes in the levels of endogenous tissue antioxidants in human inflammatory bowel disease
In summary, the hypothesis that oxygen radicals are pathogenic factors in inflammatory bowel disease is certainly not ridiculous. All the available experimental evidence suggests that ROM/RNM-mediated events are important in both the primary and downstream secondary pathophysiological mechanisms underlying intestinal inflammation. If so, do antioxidant treatment regimens deserve a place in the inflammatory bowel disease clinic? In fact, attenuating oxidative stress in inflammatory bowel disease patients has already been a therapeutic strategy for 50 years. Commonly used drugs, in particular sulfasalazine and its active moiety 5-aminosalicylic acid, are potent ROM scavengers (reviewed in Miles and Grisham259). In addition, there have been attempts to specifically prevent or attenuate intestinal oxidative stress through either the inhibition of ROM-producing enzymes or the direct scavenging of ROM. Most investigations have been carried out in animal models of colitis and have been extensively reviewed elsewhere.159 Specific antioxidant trials in inflammatory bowel disease patient groups are rare, and all are uncontrolled. Some of them have been withdrawn.260,261 In fact, only two studies remain. Published in the mid-1980s, high positive response rates (> 80% remission) were observed when patients with severe Crohn's disease (n=30) or ulcerative colitis (n=4) were treated with free or liposomal-encapsulated bovine Cu/Zn-SOD.262,263 Since then, no further SOD-based clinical trials in inflammatory bowel disease patients have been reported. Obviously, the therapeutic applicability of natural SOD has its limitations in terms of its limited cell permeability, short circulating half-life, immunogenicity and cost of production. Several innovative antioxidant agents have recently been developed that have overcome these limitations, and these await further clinical evaluation.264
Before we start introducing any drug with antioxidant activity into the gut, however, it is imperative that we learn more about the status of the endogenous antioxidant defences in the normal and inflamed intestinal mucosa. To date, data on the mucosal concentration, activity and localization of the most important (anti) oxidant enzymes in Crohn's disease and ulcerative colitis are scarce or, at most, fragmentary. We also need to understand their association with parameters of oxidative damage. In combination with animal experiments designed to evaluate the functional relationship between the (transgenic) expression of antioxidant enzymes and the development and course of intestinal inflammation, these studies might give a fresh impulse to the application and development of antioxidant therapy for inflammatory bowel disease.