key: cord-0006735-dx5svokn
authors: Halliwell, Barry
title: Drug Antioxidant Effects: A Basis for Drug Selection?
date: 2012-10-30
journal: Drugs
DOI: 10.2165/00003495-199142040-00003
sha: d54e2a8d34b11ab4c925b429270bd840dee8a8bc
doc_id: 6735
cord_uid: dx5svokn

A free radical is any species capable of independent existence that contains one or more unpaired electrons. Free radical reactions have been implicated in the pathology of more than 50 human diseases. Radicals and other reactive oxygen species are formed constantly in the human body, both by deliberate synthesis (e.g. by activated phagocytes) and by chemical side- reactions. They are removed by enzymic and nonenzymic antioxidant defence systems. Oxidative stress, occurring when antioxidant defences are inadequate, can damage lipids, proteins, carbohydrates and DNA. A few clinical conditions are caused by oxidative stress, but more often the stress results from the disease. Sometimes it then makes a significant contribution to the disease pathology, and sometimes it does not. Several antioxidants are available for therapeutic use. They include molecules naturally present in the body [superoxide dismutase (SOD), α-tocopherol, glutathione and its precursors, ascorbic acid, adenosine, lactoferrin and carotenoids] as well as synthetic antioxidants [such as thiols, ebselen (PZ51), xanthine oxidase inhibitors, inhibitors of phagocyte function, iron ion chelators and probucol]. The therapeutic efficacy of SOD, α- tocopherol and ascorbic acid in the treatment of human disease is generally unimpressive to date although dietary deficiencies of the last two molecules should certainly be avoided. Xanthine oxidase inhibitors may be of limited relevance as antioxidants for human use. Exciting preliminary results with probucol (antiatherosclerosis), ebselen (anti-inflammatory), and iron ion chelators (in thalassaemia, leukaemia, malaria, stroke, traumatic brain injury and haemorrhagic shock) need to be confirmed by controlled clinical trials. Clinical testing of N-acetylcysteine in HIV-1-positive subjects may also be merited. A few drugs already in clinical use may have some antioxidant properties, but this ability is not widespread and drug- derived radicals may occasionally cause significant damage.

A free radical is any species capable of independent existence that contains one or more unpaired electrons. Free radical reactions have been implicated in the pathology of more than 50 human diseases. Radicals and otherreactive oxygen species areformed constantly in the human body, both by deliberate synthesis (e.g. by activated phagocytes) and by chemical side-reactions. They are removed by enzymic and nonenzymic antioxidant defence systems. Oxidative stress, occurring when antioxidant defences are inadequate, can damage lipids, proteins, carbohydrates and DNA. A few clinical conditions arecaused by oxidative stress, but moreoften the stress results from the disease. Sometimes it then makes a significant contribution to the disease pathology, and sometimes it does not. Several antioxidants are available for therapeutic use. They include molecules naturally present in the body[superoxide dismutase (SOD), a-tocopherol, glutathione and its precursors, ascorbic acid, adenosine, lactoferrin and carotenoidsJ as well as synthetic antioxi-dants{such as thiols, ebselen (PZ51), xanthine oxidaseinhibitors, inhibitors ofphagocyte function, iron ion chelators and probucol]. The therapeutic efficacy of SOD, a-tocopherol and ascorbic acid in the treatment ofhuman disease is generally unimpressive to date although dietary deficiencies ofthe lasttwomolecules shouldcertainly beavoided. Xanthine oxidase inhibitors may beof limited relevance as antioxidants for human use. Exciting preliminary results with probucol (antiatherosclerosis), ebse1en (anti-inflammatory) , and iron ion chelators (in thalassaemia, leukaemia, malaria, stroke, traumatic brain injuryand haemorrhagic shock) needto be confirmed by controlled clinical trials. Clinical testing ofN-acetylcysteine in HIV-1-positive subjects may also be merited. A few drugs already in clinical use may havesome antioxidant properties, but this ability is not widespread and drug-derived radicals may occasionally cause significant damage.

A radical might donate its unpaired electron to another molecule (a reducing radical) or it might take an electron from another molecule in order to pair (an oxidising radical). However, if a radical gives one electron to, takes one electron from or possible spin states. Indeed, the technique of electron spin resonance detects radicals because it measures the energy changes that occur as electrons change their spin state. Because electrons are more stable when paired together in orbitals, radicals are in general more reactive than nonradical species, although there is a considerable variation in their individual reactivity (reviewed by Halliwell & Gutteridge 1989) .

Radicals can react with other molecules in a number of ways. Thus, if 2 radicals meet, they can combine their unpaired electrons (symbolised by -) and join to form a covalent bond (a shared pair of electrons):

Free radicals and their relationship to human disease is an important research area at the moment, especially since it has been reported that free radical reactions are implicated in the pathology of over 50 human diseases (listed in Halliwell & Gutteridge 1989) .

1.1 Definition Electrons in atoms occupy regions of space known as orbitals. Each orbital can hold a maximum of 2 electrons, spinning in opposite directions. A free radical can be simply defined as any species capable of independent existence (for however brief a time) that contains one or more unpaired electrons, an unpaired electron being one that is alone in an orbital. Most biological molecules are nonradicals, containing only paired electrons (Halliwell & Gutteridge 1989 ) .

An electron occupying an orbital by itself has 2 A· + A· -+ A-A (Eq. I) simply joins on by addition to a nonradical, that nonradical becomes a radical. Thus, a feature of the reactions of free radicals with nonradicals (which includes most biological molecules) is that they tend to proceed as chain reactions: one radical begets another and so on. Only when 2 free radicals meet can these chain reactions be terminated (reaction 1).

The most reactive free radical known is the hydroxyl radical (·OH), whose formation plays a key role in the damage suffered by tissues exposed to high-energy radiation (Von Sonntag 1987) . When tissues are exposed to, for example, 'Y-radiation, most of the energy they take up is absorbed by water, largely because there is more water in tissues than any other molecule. The radiation causes one Reaction 2 removes the 'OH, but leaves behind a carbon-centered radical (Co) in the membrane. Carbon-centered radicals formed from polyunsaturated fatty acid side-chains usually undergo molecular rearrangement to give conjugated diene structures, which can have various fates. Thus , if 2 such radicals collided in the membrane, crosslinking of fatty acid side-chains could occur as the 2 electrons join to form a covalent bond ( fig. 1 ). Reaction with membrane proteins is also a possi-of the oxygen-hydrogen covalent bonds in water to split, leaving a single electron on the hydrogen atom and one on the oxygen atom and creating two radicals, H· and ·OH. Since it is so reactive, 'OH does not persist for even a microsecond before combining with a molecule in its immediate vicinity . Because it is a radical, however, its reactions leave behind a legacy in the cell in the form of propagating chain reactions . Thus, if 'OH attacks DNA, free radical chain reactions occur and cause strand breakage, deoxyribose fragmentation and extensive chemical alteration of the purine and pyrimidine bases (Von Sonntag 1987) .The latter pattern of base changes in DNA is apparently diagnostic for attack by 'OH (Aruoma et al. 1989a; Halliwell & Aruoma 1991) . Imperfect repair of DNA damage caused by 'OH can result in proto-oncogene activation and carcinogenesis. Hence, high-energy radiation can lead to cancer, which implies that 'OH is a complete carcinogen (reviewed by Breimer 1988) .

Perhaps the best characterised biological damage caused by 'OH (although not necessarily the most important: see Halliwell 1987 ) is its ability to initiate the free radical chain reaction known as lipid peroxidation ( fig. 1 ). This occurs when the 'OR is generated close to or within membranes and attacks the fatty acid side-chains of the membrane phospholipids. It preferentially attacks side-chains derived from fatty acids with several double bonds, such as arachidonic acid.The 'OH abstracts an atom of hydrogen from one of the carbon atoms in the side-chain and combines with it to form water: bility. However, under physiologicalconditions, the most likely fate of carbon-centered radicals is to combine with oxygen, creating yet another type of radical, the peroxyl radical (sometimes abbreviated to 'peroxy radical'):

I I -C· + Oz -+ -C-Oz ' I I Peroxyl radicals can combine with each other or attack membrane proteins ( fig. 1 ), but they are also reactive enough to attack adjacent fatty acid side-chains, abstracting hydrogen:

Another carbon-centered radical is generated, and so the chain reaction (reactions 3 and 4) can continue. Hence, one 'OH can result in the conversion of several hundred fatty acid side-chains into lipid hydro peroxides (Halliwell & Gutteridge 1989 ). Accumulation oflipid hydroperoxides in a membrane disrupts its function and can cause it to collapse. In addition, lipid hydroperoxides can decompose to yield a wide range of highly cytotoxic products, the most studied'of which are aldehydes (reviewed by Esterbauer et al. 1988) . Most attention has been paid in the literature to malonaldehyde (sometimes called malondialdehyde) but this is relatively unreactive compared with such noxious products as 4-hydroxynonenal (Curzio 1988; Esterbauer et al. 1988) . Peroxyl radicals and reactive aldehydes can also cause severe damage to membrane proteins, inactivating receptors and membrane-bound enzymes (e.g. Dean et al. 1986; Roubal & Tappel 1966 ).

Interest in the role of free radicals in human metabolism was raised by the discovery in 1968 of an enzyme specific for a free radical substrate, superoxide dismutase (SOD). This enzyme activity is present in all human cells and functions to remove superoxide radical, the I-electron reduction product of oxygen:

Superoxide dismutase removes Oiby catalysing a dismutation reaction : one Oiis oxidised to 02 and the other reduced to H202: (Eq. 6) In the absence of SOD, dismutation of Oioccurs nonenzymically but at a rate approximately 4 orders of magnitude less at physiological pH.

The discovery of SOD led to the realisation that

Oiis formed in vivo in living organisms, and SOD functions to remove it. Formation of some Oiin vivo appears to be a chemical accident (Fridovich 1983; Imlay & Fridovich 1991) . For example, when mitochondria are functioning, some of the electrons passing through the respiratory chain leak from the electron carriers and pass directly onto oxygen, reducing it to Oi- (Fridovich 1983 ). Many molecules oxidise on contact with oxygen, such as when an adrenalin solution left to stand on a laboratory bench 'goes off and eventually forms a pink product. The first stage in this oxidation is transfer of an electron from the adrenalin onto 02, giving Oi-. Such oxidations undoubtedly proceed in vivo as well (reviewed in Halliwell & Gutteridge 1989) . For example, several sugars, including glucose, react with proteins to produce oxygen radicals. Thus, it has been suggested that years of exposure of body tissues to elevated blood glucose in diabetic patients can result in them suffering an 'oxidative stress' that may contribute to the side effects of hyperglycaemia (discussed by Wolff & Dean 1987) . Elevated blood glucose leads to a nonenzymatic glycosylation of proteins, to which free radical generation might contribute (Wolff & Dean 1987) .

Although some of the Oiproduction in vivo may be accidental, much is functional. Thus, activated phagocytic cells generate Oias has been shown for monocytes, neutrophils, eosinophils and most types of macrophage (Curnutte & Babior 1987) . Radical production is important in allowing 573 phagocytes to kill some of the bacterial strains that they can engulf. This can be illustrated by examining patients with chronic granulomatous disease, a collective name given to several conditions in which the membrane-bound NADPH oxidase system in phagocytes that makes the Oidoes not work (Curnutte & Babior 1987) . Such patients have phagocytes that engulf and process bacteria perfectly normally, but several bacterial strains are not killed and are released in viable form when the phagocytes die. Thus, patients suffer severe, persistent, and multiple infections with several organisms, especially Staphylococcus aureus.

Another killing mechanism used by neutrophils (but not by macrophages) is the enzyme myeloperoxidase (Weiss 1989) . It uses H202 produced by dismutation of Oito oxidise chloride ions into hypochlorous acid, HOCI, a powerful antibacterial agent (many household bleach products contain the sodium salt of hypochlorous acid):

(Eq. 7)

More recently discovered examples of physiological roles for 0iinclude its possible involvement in fibroblast proliferation (Meier et al. 1990; Murrell et al. 1990 ) and in vasoregulation (Halliwell 1989a) . It is now widely thought (reviewed by Marietta 1989) , but not yet completely certain (Myers et al. 1990 ) that endothelium-derived relaxing factor (EDRF), a humoral agent, produced by vascular endothelium, which is an important mediator of vasodilator responses induced by several pharmacological agents (including acetylcholine and bradykinin), is identical with nitric oxide. Nitric oxide (NO) has one unpaired electron and thus qualifies as a free radical. Vascular endothelium may also produce small amounts of Oiwhich can react with NO. Both NO itself (if generated in excess) and some of the products of its reaction with Oimay be highly cytotoxic (Beckman 1990; Beckman et al. 1990; Saran et al. 1990 ), although others have disputed this and argued that the interaction of NO and Oimay have a beneficial role in regulating vascular tone (Halliwell 1989a; Saran et al. 1990 ).

Superoxide dismutases remove Oiby converting it into 0iand H202. They work in conjunction with two enzymes that remove H202 in human cells: catalase and glutathione peroxidase (reviewed by Halliwell & Gutteridge 1989) . The study of inborn errors of metabolism strongly suggests that glutathione peroxidase is the more important in removing H202 in human tissues, probably because it is located in the same subcellular compartments (mostly cytosol and some in mitochondria) as SOD. Glutathione peroxidase is the only human enzyme known so far that requires the element selenium for its activity: a selonocysteine residue (side-chain -SeH instead of -SH, as in normal cysteine) is present at the enzyme's active site. It is unlikely , however, that the sole function of selenium in humans is to act as a cofactor for glutathione peroxidase (reviewed by Levander 1987) . Glutathione peroxidase removes H202 by using it to oxidise reduced glutathione (GSH) into oxidised glutathione (GSSG):

H202 is a fairly unreactive molecule that can cross cell membranes easily. Like Oi-, it may perform some physiological roles, e.g. control of platelet adherence (Salvemini & Botting 1990) . Since the H202 molecule has no unpaired electrons, H202 cannot be described as a radical. Hence , the term 'reactive oxygen species' has been introduced in the biomedical literature to describe collectively not only Oiand ·OH (radicals) but also H202 (nonradical). Hypochlorous acid (HOCl) produced by myeloperoxidase can also be included under this heading: it is a powerful oxidising agent but not a radical, having no unpaired electrons. H202, Oi-, ·OH and HOCI are sometimes collectively called 'oxidants'. This is a valid description for H202, ·OH and HOCl , which are oxidising agents. However, Oihas both oxidising and reducing properties. Indeed, the latter property is made use of in a popular assay for 0i-, the SOD-inhibitable reduction of cytochrome c, often applied to measure Drugs 42 (4) 1991

Oiproduction by phagocytes (Curnutte & Babior 1987) : cyt c (Fe+++) + Oi 7 -+ 02 + cyt c (Fer") (Eq, 9) Another reactive oxygen species (but not a radical) is singlet oxygen l~g, a form of oxygen that has undergone an electronic rearrangement that allows it to react much faster With biological molecules than does 'normal' (ground-state) oxygen. For example, singlet 02 reacts directly with polyunsaturated fatty acid side-chains in membrane lipids to give lipid peroxides. Singlet 02 is produced when several pigments are illuminated: light is absorbed by the pigment, which enters a higher electronic excitation state. As this returns to the ground state, energy is transferred onto oxygen, to make singlet 02. Several compounds found in plants, some cosmetics and some drugs (e.g. tetracyclines) can cause singlet 02 formation in the presence of light (reviewed by Epstein 1982; Kanofsky 1989) .

Usually such 'photosensitisation' reactions are a nuisance. For example, many of the side effects of the nonsteroidal anti-inflammatory drug benoxaprofen could have been related to its ability to act as a photosensitiser (Artuso et al. 1990; Sik et al. 1983) , and patients with certain forms of porphyria become light-sensitive because photosensitising porphyrins accumulate in their skin (Mathews-Roth 1987) . However, photosensitisation can be made use of in medicine. Components of a porphyrin derivative called haematoporphyrin derivative (HPD) are taken up by tumour tissues . Irradiation of the tumour with light at a wavelength absorbed by HPD can then damage or kill the tumour: this method has been applied in the treatment of some types of lung cancer (reviewed by Ash & Brown 1989) , although its real clinical benefit remains to be fully evaluated.

Small amounts of singlet 02 can be generated in peroxidising lipid systems ( fig. 1 ; Wefers 1987) but , apart from photosensitisation reactions, singlet 02 does not seem to be a major damaging species in the human body. However, it can be produced by acti vated eosinophils under certain circumstances (Kanofsky 1989 Superoxide can re-reduce Fe+++ to Fe?": 575 ticipate in the formation of species able to abstract hydrogen. There is considerable debate (e.g. see Aruoma et al. 1989b; Minotti & Aust 1987) about which species initiate peroxidation in biological lipid systems (such as lipoproteins and membranes); candidates include ferryl, perferryl, Fe++-Fe+++-02 complexes and ·OH. However, there is general agreement that metal ions are needed for formation of initiating species, except in organisms exposed to high doses of ionising radiation. Second, transition metal ions decompose lipid hydroperoxides into peroxyl and alkoxyl (lipid-O) radicals that can abstract hydrogen and perpetuate the chain reaction of lipid peroxidation. This may be represented, in a simplified form, by the equations below, in which L· symbolises a carbon-centered radical ( The word 'antioxidant' can be defined in various ways. Often the term is implicitly restricted to chain-breaking antioxidant inhibitors of lipid per-Reducing agents, such as ascorbic acid or Oi-, can accelerate these metal ion-dependent peroxidation reactions because Cut and Fe?" ions seem to decompose peroxides faster than do Cur" and Fe+++, respectively. The end-products of these complex metal ion-catalysed breakdowns of lipid hydro peroxides include the cytotoxic aldehydes mentioned previously, as well as hydrocarbon gases such as ethane and pentane (reviewed in Halliwell & Gutteridge 1989) . Copper ions can also convert H202 into ·OH (Aruoma et al. 199Ia) . Generation of ·OH and other reactive species appears to account for some or all of the toxicity observed when Oiand H202 are generated in excess in vivo (reviewed in Halliwell & Aruoma 1991; Halliwell & Gutteridge I990a; Imlay & Linn 1988) . There are very few significant cellular targets in mammals that can be attacked by Oior H202 generated at physiological rates.

It has also been suggested that the peroxynitrite species generated by reaction of Oiwith NO may decompose to produce ·OH under physiological conditions (Beckman 1990 ) but doubt has been expressed about this (Saran et al. 1990 ).

Iron and copper ions also accelerate lipid peroxidation, for at least 2 reasons. First , they can par-LO· + LH --. LOH + L· LOO· + LH --. LOOH + L· L· + 02 --. LOO· (Eq. 16) (Eq. 17) (Eq. 18) oxidation, such as a-tocopherol (see below). However, the author prefers a broader definition -an antioxidant is any substance that, when present at low concentrations compared to those of an oxidisable substrate, significantly delays or prevents oxidation of that substrate (Halliwell & Gutteridge 1989 ). The term 'oxidisable substrate' includes almost everything found in living cells, including proteins, lipids, carbohydrates and DNA.

The major intracellular antioxidants in the human body are probably the enzymes superoxide dismutase, catalase and glutathione peroxidase. Two isoenzymes of SOD exist in human cells: an enzyme containing copper and zinc ions (largely located in the cytosol) and a mitochondrial SOD containing manganese ions. Both catalyse the same reaction (reaction 6; Fridovich 1983 ). In addition, membranes contain the chain-breaking antioxidant a-tocopherol, a lipid-soluble molecule that is located in the interior of biological membranes. a-Tocopherol is the most important chain-breaking antioxidant present in human membranes (Ingold et al. 1986) but not the only one (Esterbauer et al. 1989 ). Attached to the hydrophobic structure of atocopherol is an -OH group whose hydrogen atom is very easy to remove. Thus, when peroxyl and alkoxyl radicals are generated during lipid peroxidation, they combine preferentially with the antioxidant, e.g.:

LOO + tocopherol-OH --+ LOOH + tocopherol-Or (Eq. 19) instead of with an adjacent fatty acid. This terminates the chain reaction, hence the name chainbreaking antioxidant. It also converts the a-tocopherol into a new radical, tocopherol-O. This radical is poorly reactive, being unable to attack adjacent fatty acid side-chains , so the chain reaction is stopped. Evidence exists (Esterbauer et al. 1989; McCay 1985) that the tocopherol radical can migrate to the surface of membranes and lipoproteins and be converted back to a-tocopherol by reactions with ascorbic acid (vitamin C) and possibly with GSH, although it is not yet absolutely certain that this happens in vivo ). However, it is widely believed that both vitamin C and a-Drugs 42 (4) 1991 tocopherol help to minimise the consequences of lipid peroxidation in membranes, should this process begin. The terms 'a-tocopherol' and ' vitamin E' are often used synonymously. This is not quite correct: vitamin E is defined nutritionally as a factor needed in the diet of rats for normal reproduction (Diplock 1985) , and compounds other than a-tocopherol (e.g. (3-, 'Y-and a-tocopherols) have some effect in this assay. However, a-tocopherol is the most effective lipid-soluble antioxidant in animals .

Antioxidant defence in the extracellular compartments of the human body relies largely on different strategies than does intracellular defence. a-Tocopherol is still important. Thus , the content of a-tocopherol in plasma lipoproteins helps to determine their resistance to lipid peroxidation, but it is not the only antioxidant present. This is relevant to the development of atherosclerosis, since peroxidation oflow-density lipoproteins appears to contribute to the progression of atherosclerosis (Esterbauer et al. 1989 ). Indeed, low plasma concentrations of a-tocopherol and vitamin C seem to be associated with an increased incidence of myocardial infarction and of some types of cancer (Gey et al. 1987; Gey 1990 ). However, it could be argued that these low levels are merely indicative of a poor diet, which predisposes to most diseases. Thus, although one could assume that taking commercial vitamin pills would protect against atherosclerosis , improving one's diet is a more logical recommendation. If low density lipoproteins isolated from human plasma are placed under oxidative stress, peroxidation does not begin until tocopherols , {3carotene, Iycopene and phytofluene have been oxidised (Esterbauer et al. 1989) .

By contrast with the intracellular environment, enzymic antioxidant defence enzymes are much less prominent in extracellular fluids. Blood plasma, tissue fluid, cerebrospinal fluid, synovial fluid and seminal plasma contain little or no catalase activity, and only low activities of superoxide dismutase and selenium-containing glutathione peroxidase can be measured. There is also very little reduced glutathione (GSH) in most extracellular fluids -about 2 JLmoljL in human plasma (Svardal et al. 1990 ). The selenium-containing glutathione peroxidase enzyme has been purified from human plasma, and shown to be different from the intracellular enzyme. The functional role of this enzyme is something of an anomaly , since its K m (substrate concentration for half-maximal velocity) for GSH is in the millimolar range and so it would be unlikely to function effectively at the micromolar levels of GSH found in plasma (reviewed in Halliwell & Gutteridge 1990b) . Although much higher (~100 JLmoljL) concentrations of GSH are present in human lung lining fluid, its content of glutathione peroxidase is low (Cantin et ai, 1987) .

'Extracellular' superoxide dismutase (EC-SOD) enzymes have also been described. Although EC-SODs contain copper and zinc, they are very different from intracellular CuZnSOD in that EC-SODs have a much higher relative molecular mass (about 135000), and possess attached carbohydrate. The biological role of EC-SODs is unclear; such low activities are present in plasma and in other extracellular fluids that bulk-scavenging of Oiin extracellular fluids has always seemed (to this author) to be an unlikely physiological function for them. Marklund has shown that some EC-SODs bind to heparin and has proposed that , in vivo, they may be associated with endothelial cell surfaces as a protective antioxidant 'layer' over the cells (Marklund, 1990) . Trace amounts of EC-SODs are found attached to cells in various tissues, so the term 'extracellular' does not really mean a location within extracellular fluids.

Thus, enzymic removal of Oiand H202 contributes little to the antioxidant activit y of extracellular fluids. Indeed, Oiand/or H202 generated extracellularly in small amounts by endothelium (Arroyo et al. 1990 ), lymphocytes (Maly 1990 ), platelets (Salvemini & Botting, 1990) , fibroblasts (Murrell et al. 1990 ) and other cells (Burdon & Rice-Evans 1989 ) may play important physiological roles and thus should not be scavenged with 100% efficiency. Halliwell and Gutteridge (l990a,b) have argued that a major antioxi-577 dant defence of human plasma is to prevent Oiand H202 from reacting to form dangerous species such as ·OH, by binding transition metal ions in forms that will not stimulate free radical reactions, or by otherwise preventing the metal ions from participating in such reactions. Thus, safe sequestration of iron and copper ions into forms that will not catalyse free radical reactions is an important antioxidant strategy in the human body (Halliwell & Gutteridge I990a, b) .

How is this sequestration achieved? The irontransport protein transferrin in plasma from healthy humans is only 20 to 30% loaded with iron, so that the content of free ionic iron in plasma is effectively nil. Iron, bound to transferrins will not participate in ·OH radical formation or lipid peroxidation (Aruoma & Halliwell 1987; . The physiological importance of this sequestration of iron ions into 'safe' forms is clearly illustrated by an inspection of the multiorgan damage seen in patients with iron-overload disease (McLaren et al. 1983 ), in whom low-molecular-mass iron ion complexes capable of stimulating lipid peroxidation and ·OH generation (Gutteridge et al. 1985) circulate in the plasma.

Haemoglobin can be liberated into plasma from disrupted erythrocytes, and myoglobin can be released from muscles after injury (e.g, myocardial infarction, or skeletal muscle damage during strenuous exercise). Both of these haem-containing proteins are potentially dangerous in that they can accelerate peroxidation of lipids (Galaris et al. 1988; Kanner & Harel 1985) , including plasma lipoproteins (Bruckdorfer et al. 1990) , in the presence of H202 . They do this by two mechanisms. First, reaction of the protein with equimolar concentrations of H202 produces an oxo-haem species, probably ferryl, whose formation is associated with production of a reactive oxygen species [possibly a tyrosine peroxyl radical (Davies 1990) ] that can stimulate lipid peroxidation and damage other biomolecules (Davies 1990; Galaris et al. 1988; Puppo & Halliwell, 1988a) . Second, excess H202 can cause degradation of the haem rings of myoglobin and haemoglobin, releasing from the proteins iron ions that are capable of stimulating ·OH pro-duction and lipid peroxidation (Gutteridge 1986; Puppo & Halliwell 1988a,b) . Haem, released from damaged haem proteins, is also well known to be a powerful stimulator of lipid peroxidation (e.g. Gutteridge & Smith 1988) .

Because myoglobin and haemoglobin can be liberated into body fluids after muscle injury (e.g, in the 'crush' syndrome) plasma contains the haemoglobin-binding proteins known as haptoglobins, as well as a haem-binding protein (haemopexin). Binding of haemoglobin to haptoglobins, or of haem to haemopexin, diminishes their effectiveness in stimulating lipid peroxidation (Gutteridge & Smith 1988 ). The haemoglobin-haptoglobin or haem-haemopexin complexes are rapidly cleared from the circulation.

The plasma copper-containing protein caeruloplasmin is thought to play an essential role in iron metabolism, but it also has antioxidant properties (Gutteridge & Stocks 1981) . First , caeruloplasmin has a ferroxidase activity -it oxidises Fe++ to Fe+++ whilst reducing oxygen to water. However, unlike the nonenzymatic oxidation of Fe":" ions by 02, this caeruloplasmin-catalysed oxidation of Fe":" does not release any damaging oxygen radicals: they are kept on the active site of the protein. The ferroxidase activity of caeruloplasmin allows it to inhibit iron ion-dependent lipid peroxidation and ·OH formation from H202 under most conditions, and this is probably the major antioxidant activity of the protein (Halliwell & Gutteridge 1990b) .

Albumin can also bind copper ions, and it usually inhibits copper ion-dependent lipid peroxidation and ·OH radical formation (reviewed in Halliwell 1988) . In fact, the reactions often continue at the sites of metal ion binding and damage the protein, but the high concentration of albumin in plasma and the rapid turnover of this protein mean that such damage is probably biologically insignificant. Thus, binding to albumin of any copper ions released into plasma may lead to damage to the protein if Oiand H202 are generated in plasma. However, albumin may well prevent the copper ions from attaching to more important targets, such as -SH groups essential to membrane function, e.g. on the membrane proteins of endo-Drugs 42 (4) 1991 thelial cells or of erythrocytes, where binding of Cut" ions could lead to oxidative damage. Binding to albumin might also help to stop copper ions from accelerating the peroxidation of low-density lipoproteins and promoting atherosclerosis (Esterbauer et al. 1989 ). The copper ion-albumin complex might be a safe 'transit form' of copper, that can be removed from the circulation by the liver (Halliwell 1988) .

Albumin transports fatty acids in the blood, and the bile pigment bilirubin is bound to it. It has been claimed that bilirubin acts as an antioxidant inhibitor of lipid peroxidation in vitro (Stocker 1990 ).

Perhaps bilirubin protects albumin-bound fatty acids against peroxidation in vivo. Indeed, Stocker (1990) has argued that induction of the enzyme haem oxygenase (shown to occur in cells exposed to oxidative stress) serves not only to remove a pro-oxidant (haem) but also to increase antioxidants (e.g. bilirubin). However, the prevention of copper ion-mediated damage by albumin is probably a property of the protein itself. Albumin is also a powerful scavenger of HOCI in plasma .

Ascorbic acid, present in plasma from healthy humans at concentrations of 50 to 200 jlmolfL, has multiple antioxidant properties (listed in Halliwell 1990 ), probably including the ability to regenerate a-tocopherol by reducing o-tocopheryl radicals at the surface of lipoproteins and membranes. High concentrations of ascorbic acid are also present intracellularly, especially in brain and lung cells. Mixtures of ascorbic acid and H202 with iron or copper ions have powerful pro-oxidant properties, however, because they can form ·OH and other reactive species. This emphasises the need for careful sequestration of 'free' transition metal ions in the human body in order for ascorbic acid to exert its antioxidant action. The fact that large amounts of ascorbic acid can safely be given to healthy humans illustrates how effective this sequestration of iron and copper ions is, and means that ascorbic acid is probably an important antioxidant in healthy humans. The administration of ascorbic acid to patients with iron overload can lead to serious con-sequences, however, unless desferrioxamine is given simultaneously (Nienhuis 1981) .

It thus seems that a major antioxidant action of blood plasma is the prevention of transition metal ions from accelerating damaging free-radical reactions such as lipid peroxidation. By contrast, the metal ion-binding antioxidant defences of some other extracellular fluids may be much weaker. Human cerebrospinal fluid (reviewed by Halliwell 1989b) contains little transferrin , albumin or caeruloplasmin, but has high concentrations of ascorbic acid (about 10 times those in plasma) and also contains uric acid, which has some antioxidant properties (Ames et al. 1981) . Synovial fluid has lower concentrations of albumin, transferrin and caeruloplasmin than does plasma (Gutteridge 1987) , whereas the fluid that lines the alveoli of the lung also has a low protein content but a very high concentration of ascorbic acid (reviewed by Heffner & Repine 1989). It has also been shown that human alveolar lining fluid, unlike other human extracellular fluids, contains significant concentrations of glutathione (Buhl et al. 1989) . Seminal plasma may have poor antioxidant capacity, since spermatozoa seem to peroxidise readily (Aitken 1989), but more studies on this fluid are needed.

There is still a general tendency in the literature to equate 'oxidative damage' with 'lipid peroxidation', so that antioxidants are often characterised only by their ability to inhibit lipid peroxidation. However, membrane lipids are only one possible target of oxidative damage ( fig. 2) . Indeed, very few toxins that injure cells by generating excess reactive oxygen species seem to cause damage by accelerating the bulk peroxidation of membrane lipids (discussed by Kappus 1987 and by Halliwell & Gutteridge 1984 , 1990a .

Oxidati ve stress in cells and tissues results from increased generation of reactive oxygen species and/ or from decreases in antioxidant defences. In-579 creased production of 0iand HzOz can occur as a result of:

I. Raised Oz concentrations (sometimes inadvertantl y; culture of most mammalian cells under 95% Oz or even under air exposes them to higher Oz concentrations than are present in vivo).

2. Adding certain toxins (such as alloxan, paraquat or doxorubicin (adriamycin)] that increase intracellular formation of reactive oxygen species.

3. Activating a large number of phagocytes at a localised site (Oiand HzOz are produced by activated phagocytes and are essential for the killing of many bacterial strains , but they 'can do tissue damage when generated in excess).

One early event frequently observed in mammalian cells subjected to oxidative stress is DNA strand breakage, which leads to activation of poly(ADP-ribose) synthetase, an enzyme that cleaves NAD+ to give ADP-ribose, which is then polymerised. Excess activation of poly(ADP-ribose) synthetase can lead to large falls in cellular NAD+ concentrations, so interfering with ATP synthesis (Schraufstatter et al. 1987) . Oxidative stress can also lead to rises in intracellular 'free' Cat" within the cell by inactivating enzyme systems in the endoplasmic reticulum and mitochondria that sequester free Car", and possibly also by damaging transport systems in the plasma membrane that eject Ca?" from the cell (discussed by Albano et al. 1991 ; Orrenius et al. 1989 ). If free intracellular Car" becomes too high, proteases can be activated that cleave the cytoskeleton. Calcium could also activate endonucleases that produce DNA strand breaks (Orrenius et al. 1989 ).

In addition, cells contain a small low-molecular-mass iron pool, which supplies iron for the synthesis of ferroproteins. The exact intracellular location of this pool is not clear, but it may be largely compartmentalised into organelles such as mitochondria. Its existence may explain why superoxide dismutase and HzOz-removing enzymes are such important intracellular antioxidants; it is vital to remove as much Oiand HzOz as possible before they come into contact with this low molecular mass iron pool and generate damaging species such as ·OH (Halliwell & Gutteridge 1990b Fig. 2 . Schematic illustrating the multiple derangements of cell metabolism that can be brought about by oxidative stress. Direct damage to DNA, proteins and/or lipids is possible. Secondary damage can arise when oxidative stress produces rises in 'free' intracellular metal ions, such as Ca++ and possibly Fer" , Ca++can stimulate proteases and nucleases, damaging both DNA and the cytoskeleton. Membrane 'blebs' are 'blow-outs' of the plasma membrane characteristic of cells under stress; the rupture of large blebs may lead to cell death (Orrenius et al. 1989). damage to proteins and to DNA can occur not only as a result of activation of hydrolytic enzymes, but also by generation of·OH . The relative significance of these 2 mechanisms is an area of current debate (Halliwell 1987; Halliwell & Aruoma 1991 ; Imlay & Linn 1988; Orrenius et al. 1989; Schraufstatter et al. 1987; Starke & Farber 1985) . In addition, oxidative stress can create more metal ion promoters of free radical reactions . Thus , Oican mobilise a small percentage of the iron ions stored within the protein ferritin (Biemond et al. 1984; Bolann & Ulvik 1990) and may also be able to release iron ions from the active sites of some iron-sulphur proteins (Flint & Emptage 1991) . Similarly, HzOz in excess can 'degrade haem proteins to liberate iron (Gutteridge 1986 ; Pup po & Halliwell I988a,b).

In evaluating the likelihood of a proposal that a compound already present in vivo or administered as a drug is acting as an antioxidant in vivo, it is important to ask the right questions; as summarised in table I (Halliwell 1990 ). Antioxidants can act by scavenging reactive oxygen species (e.g. SOD removing Or), by inhibiting their formation (e.g. by blocking activation of phagocytes), by binding transition metal ions and preventing formation of ·OH and/or decomposition of lipid hydroperoxides, by repairing damage (e.g. a-tocopherol repairing peroxyl radicals and so terminating the chain reaction oflipid peroxidation) or by any combination of the above.

In testing putative antioxidant activity , it is important to use biologically relevant reactive oxygen species. Fairly simple methods exist for measuring the rate of reaction ofa compound with Oi-, HzOz, HOCI, ·OH, peroxyl radicals, singlet Oz and oxidants generated when haem proteins react with HzOz. Methods also exist to measure the ability of compounds to suppress metal ion-dependent ·OH generation, production of reactive oxygen species by phagocytes or inhibition of the peroxidation of fatty acids or of membranes (reviewed by Halliwell 1990) . In all cases, it is essential to test the compound at the concentrations that are actually present in vivo. which are often too low for direct scavenging of reactive oxygen species to be feasible. For example, although many drugs react with ·OH in vitro, it is unlikely that this is often a significant mechanism of antioxidant action in vivo simply because the drugs are usually present at concentrations that are too low to compete with biological targets for any ·OH generated (explained in table I).

It must also be borne in mind that antioxidants Table I . Questions to ask when evaluating the proposed role of an 'antioxidant' in vivo 1. What biomolecule is the compound supposed to protect? For example. an inhibitor of lipid peroxidation is unlikely to be useful if the oxidative damage is mediated by an attack upon proteins or DNA.

2. Is the compound present in vivo at or near that biomolecule at sufficient concentration? For example, many compounds have been suggested to act as 'OH scavengers in vivo. In order to compete with biological molecules for 'OH, a scavenger has to be present at at least millimolar concentrations in vivo. Most drugs never achieve this sort of concentration. 3. How will the compound protect -by scavenging ROS, by preventing their formation or by repairing damage? 4. For naturally occurring antioxidants, is antioxidant protection the primary biological role of the molecule or a secondary one? For example. SOD has probably evolved as an antioxidant enzyme. By contrast, transferrin has probably evolved as an iron transport protein, although the binding of iron ions to transferrin stops them accelerating radical reactions, giving this protein an important secondary role in extracellular antioxidant defence. 5. If the antioxidant acts by scavenging a ROS, can the antioxidant-derived radicals themselves do biological damage? 6. Can the antioxidant cause damage in biological systems different from those in which it exerts protection?

Abbreviation: ROS = reactive oxygen species; SOD = superoxide dismutase.

can sometimes do damage in biological systems. This may occasionally be because reactive oxygen species have useful physiologicalfunctions and thus too much antioxidant activity can be counter-productive. Hence, intravenous administration of excess a-tocopherol to premature babies has been claimed to lead to an increase in serious infections, presumably by suppression of bacterial killing by phagocytes (discussed by Ehrenkranz 1989) . In addition , many lipid-soluble chain-breaking antioxidant inhibitors of lipid peroxidation can have prooxidant properties under certain circumstances, often because they can bind Fe+++ ions and reduce them to Fe++ (Laughton et al. 1989) . Even a-tocopherol can be made to exert pro-oxidant effects in vitro, although these almost certainly have no physiological relevance (discussed by Halliwell 1990) . Propyl gallate, an antioxidant sometimes used in the food industry, has limited solubility in water, but this is enough to allow it to accelerate both 'OH formation from H202 by Fenton chemistry and DNA damage by the antibiotic bleomycin, in both cases by its ability to reduce Fe+++ to Fer" (Aruoma et al. 1990; Gutteridge & Xaio-Chang 1981) . Many phenolic compounds found in plants (especially flavonoids) have been styled as 'antioxidants' because they inhibit lipid peroxidation -hence the appearance of 'bioflavonoids' (plus or minus vitamin C) on the shelves of health-food stores. However, several plant phenolics can accelerate oxidative damage to nonlipid biomolecules such as DNA (e.g, Laughton et al. 1989) . They can do this by reducing Fe+++ ions to Fe?" and/ or by oxidising to produce Oiand H202. Thus, an antioxidant in one system is not an antioxidant in all systems, and this must be borne in mind. A similar principle might be applicable to ascorbic acid. Tissue injury can result in release of intracellular metal ions and a consequent stimulation offree radical reactions (section 3.2). Whereas ascorbic acid is a good antioxidant in the absence of trans ition metal ions, in the presence of such ions it can accelerate oxidative damage. It thus follows that the large doses (often > 10 g/day) of ascorbic acid sometimes recommended for consumption might be harmless to healthy subjects, but not necessarily harmless in certain human diseases. In sickle-cell disease (Jain & Williams 1985) , rheumatoid arthritis , adult respiratory distress syndrome (ARDS) ] and iron overload disease (Nienhuis 1981) , ascorbic acid concentrations in body fluids are markedly subnormal. In rheumatoid arthritis and ARDS, this may be because of the rapid uptake and oxidation of ascorbate by activated neutrophils as well as by the reaction of ascorbate with 0iand HOCl. However, replacement of this ascorbate might not necessarily be totally beneficial. It might even be argued that this loss of ascorbic acid is beneficial, if tissue injury is making metal ions available (Halliwell & Gutteridge 1990b) .

Drugs 42 (4) 1991

Does increased formation of free radicals and other reactive oxygen species cause any human disease? Radiation-induced cancer may be initiated by radicals, as discussed previously. The signs produced by chronic dietary deficiencies of selenium (Keshan disease) or of vitamin E (neurological disorders seen in patients with inborn errors in the mechanism of intestinal fat absorption) might also be mediated by reactive oxygen species (reviewed by Levander 1987; Howard 1990 ). In the premature infant, exposure of the incompletely vascularised retina to elevated concentrations of oxygencan lead to retinopathy of prematurity, which in its most severe forms leads to blindness. Several controlled clinical trials have documented the efficiency of a-tocopherol in minimising the severity of the retinopathy, suggesting (but by no means proving) a role for lipid peroxidation (Howard 1990) . Vitamin E supplementation may also diminish the incidence of intraventricular haemorrhage (Chiswick et al. 1991 ).

For most human diseases, the excess formation of reactive oxygen species is not a cause of the disease, but is secondary to the primary disease process. For example, activated neutrophils produce Oi-, H202 and HOCI as part of the mechanism of killing bacteria. However, if large numbers of phagocytes become activated in a localised area, they can produce tissue damage. For example, the synovial fluid in the swollen.joints of rheumatoid patients contains many activated neutrophils and there is reasonable evidence that reactive oxygen species and other products derived from neutrophils and macrophages are contributing to the joint injury (reviewed by Halliwell et al. 1988a) . In some forms of ARDS the lung damage is thought to be mediated by an influx of neutrophils into the lung, where they become activated to produce prostaglandins, leukotrienes, proteolytic enzymes, such as elastase, and reactive oxygen species (Cross et al. 1990 ). Among other effects, the reactive oxygen species can inactivate the proteins (such as aqantiproteinase) within the lung that normally inhibit the action of serine proteinases, such as elastase, and prevent them from attacking lung elastic fibres. The precise contribution that generation of reactive oxygen species makes to lung damage in AROS is unknown, but deserves more investigation in view of the high mortality rate of AROS .

Another example of secondary oxidative damage is provided by the work of Akaike et al. (1990) . They found that superoxide dismutase protects mice against the lethality of an influenza virus infection, apparently by protecting the lungs against excess reactive oxygen species generated by activated phagocytes and by the enzyme xanthine oxidase. Similar oxidative stress may occur in some other viral infections (Peterhans et al. 1988 ) and during malarial infections (Clark et al. 1986; Hunt & Stocker 1990) .

Reactive oxygen species seem also to play a role in the formation of foam cells in atherosclerotic plaques (reviewed by Steinberg et al. 1989 ). Interest in the role of lipid peroxidation in atherogenesis was raised when it was found that the compound probucol has a marked antiatherogenic effect in hyperlipidaemic rabbits (Carew et al. 1987) . Probucol, like a-tocopherol, has easily donatable hydrogen atoms and is a powerful chain-breaking inhibitor of lipid peroxidation. That does not mean, of course, that any protective action of probucol in vivo is due to inhibition of lipid peroxidation.

Macrophages possess receptors for low-density lipoprotein (LOL) but if LOL has become peroxidised it is recognised by a separate class of macrophage receptors known as the scavenger receptors. There may be more than one type of scavenger receptor. LOL bound to these receptors is taken up with enhanced efficiency, so that cholesterol rapidly accumulates within the macrophage and may convert it into a foam cell . Arterial endothelial cells, smooth muscle 583 cells and macrophages have been shown to be capable of oxidising LOL in vitro so that macrophages will internalise it faster. The modification process that allows recognition by the scavenger receptors involves the derivatisation of lysine residues of the apoprotein B moiety of LOL by lipid peroxidation products , such as the cytotoxic aldehyde 4-hydroxynonenal (Steinberg et al. 1989 ).Human atherosclerotic lesions accumulate the pigment ceroid, which is widely thought to be an endproduct of peroxidation ). More rigorous evidence for the presence of peroxidised lipid in atherosclerotic plaques is provided by the use of antibodies. For example, antibodies directed against LOL that has undergone peroxidation or has been treated with 4-hydroxynonenal, bind to rabbit aortic atherosclerotic lesions. In addition, LOL eluted from such lesions can bind to antibody specific for malondialdehyde-treated LOL. Plasma samples from hypercholesterolaemic rabbits and from several human subjects have been found to contain antibodies that will react with peroxidised LOL, suggesting that peroxidation does indeed take place in vivo (Steinberg et al. 1989 ).

Peroxidation ofLOL within the vessel wall could have a number of deleterious effects. It has been suggested that products formed in peroxidised LOL, such as lysophosphatidyl choline, might act as chemotactic factors for blood monocytes, encouraging their recruitment into an atherosclerotic lesion and development into macrophages (Steinberg et al. 1989 ). In addition, low (submicromolar) concentrations of lipid hydroperoxides might accelerate cyclooxygenase-and lipoxygenase-catalysed reactions in endothelium and in any platelets present, leading to enhanced formation of prostaglandins and leukotrienes (Warso & Lands 1983) . Oxidised LOL might also stimulate the production of cytokines and growth factors by macrophages, which are complex multifunctional cells (Yokode et al. 1988 ). There have been several speculations that oxidation products of cholesterol might also be involved in atherogenesis; cholesterol is oxidised into a wide variety of products in peroxidising lipid systems (Sevanian & Peterson 1986; Smith 1987) and oxidised cholesterol has been reported to be toxic to many cell types, including arterial smooth muscle cells and endothelial cells. It seems likely that cell breakdown in the advanced atherosclerotic plaque can lead not only to activation of lipoxygenases, but also to release of transition metal ions that can stimulate free radical reactions such as lipid peroxidation (Halliwell I989c) .

The clinical consequences of severe atherosclerosis (stroke and myocardial infarction) may also involve reactive oxygen species. One of the most exciting areas of basic research in recent years has been that of ischaemia/reperfusion injury (Me-Cord, 1985) , although it has yet to yield clinically useful developments. When tissues are deprived of 02 they are injured and after a period the injury becomes irreversible and the tissue will die. The length of this critical period depends on the extent of 02 deprivation (whether it is total ischaemia or merely hypoxia) and the tissue in question: skeletal muscle can be rendered bloodless for hours without much injury , whereas in the brain the period is much shorter. For the heart , up to about 60 minutes seems to be a tolerable period. Thus , ischaemia injures cells and will eventually kill them, and the aim should be to restore blood flow as soon as possible: hence, the use of thrombolytic agents in the treatment of myocardial infarction (reviewed by Bolli 1988) .

However, studies on isolated organs and in animals have shown that, provided the period of ischaemia does not itself do irreversible damage, tissue function is better preserved upon reoxygenation if certain antioxidants are included in the reperfusion medium (Bolli 1988; McCord 1985; Simpson & Lucchesi 1987) . Hence, restoration of the oxygen supply, although obviously beneficial overall, causes increased formation of reactive oxygen species in the damaged tissue and temporarily worsens the injury. For example, depending on the model system studied, the heart can respond to reperfusion by showing several dysfunctions , such as arrhythmias, delayed return of contractile function ('stunning') and , it has been claimed , an increase in infarct size. Thus, inclusion of antioxidants when blood flow is restored offers protective effects in many animal experiments: antioxidants that have Drugs 42 (4) 1991 been used include thiol compounds (e.g. mercaptopropionylglycine), catalase, recombinant human superoxide dismutase and desferrioxamine, a chelator that binds iron ions and usually stops them from accelerating free radical reactions (Gutteridge et aI. 1979) . The use of antioxidants in the treatment of shock and in the preservation of organs for tranplantation is also receiving attention (Bolli 1988; Currin et aI. 1990; Fuller et aI. 1988; Sanan et aI. 1989) , as discussed in section 4.

Superoxide dismutase and catalase have been extensively explored as protective agents when added to reperfusates in many types of experiment. In some models significant protection is observed whilst in others the enzymes are without effect. The author's general impression is that SOD is often very protective against myocardial arrhythmias and sometimes protective against stunning, whereas its ability to diminish infarct size is less well established (Bolli 1991; Cohen 1989; Downey et aI. 1991; Engler & Gilpin 1989) . Protection by some antioxidants is often only observed when the antioxidant is added before ischaemia or at the start of reperfusion: add ition after reperfusion has started (even after only 60 seconds) gives no protective effect. This phenomenon has been demonstrated with desferrioxamine and with mercaptopropionylglycine (Bolli et aI. 1989) in an open-chest dog model of myocardial stunning after reperfusion .

Why should ischaemia lead to increased generation of reactive oxygen species when tissues are reoxygenated? The answer is not yet clear: possibilities include release of iron ions, catalytic for free radical reactions , from their normal storage sites within the cell (Gower et aI. 1989a; Halliwell 1989a; Sanan et aI. 1989) ,disruption of mitochondrial respiratory chains so that more electrons leak to oxygen to form Oiand increases in the activities of certain enzymes, such as xanthine oxidase, that generate Oi- (McCord 1985) . However, there is considerable doubt about the role of xanthine oxidase in human myocardial injury (discussed by Kehrer 1989 ).

Mechanical (e.g. crushing), chemical or ischaemic injury to tissues may cause cells to rupture and release their contents into the surrounding area ( fig. 3) . These contents will include transition metal ions, which are potentially capable of increasing free radical reactions in the surrounding area. For example, admin istration of cytotoxic drugs to patients with acute myeloid leukaemia has been shown to create a temporary 'iron overload' state, probably due to extensive drug-induced lysis of the leukaemic cells. This increased iron availability could contribute to the side effects of cytotoxic chemotherapy (Halliwell et al. 1988b ). Thus, as figure 3 shows, any disease that increases cell death could conceivably result in an increased rate of free radical reactions. This may be completely trivial, i.e. the increased rates of free radical reactions may make no significant contribution to the disease pathology. An example of this may be the increased amounts of lipid hydroperoxides measured in dystroph ic muscle, even though there is no evidence that inhibitors ofIipid peroxidation or other antioxidants are therapeutically useful in muscular dystrophy (Davison et al. 1988; Stern et al. 1982) . Sometimes, however, this injury-related damage may be important: perhaps the greatest interest in this area lies in the sequelae of traumatic or ischaemic injury to the brain . Some areas of the human brain are rich in iron but cerebrospinal fluid has no significant iron-binding capacity, since its content of transferrin is low. Thus, it has been proposed (Halliwell & Gutteridge 1985 ) that injury to the brain by mechanical means (trauma) or by oxygen deprivation (stroke) can result in release of iron 585 ions into the surrounding area. These ions can facilitate further damage to the surrounding areas by accelerating free radical reactions ( fig. 3 ). This proposal has been given some support from animal studies, using agents (such as amino steroids) that interfere with iron-dependent radical reactions. One such aminosteroid, U74006F, has been observed, in animal experiments, to decrease reperfusion injury, post-traumatic spinal cord degeneration and neurological damage after head injury (reviewed by Hall & Braughler 1989) .

Tissue destruction and degeneration can result in increased oxidative damage, by such processes as metal ion release, lipoxygenase activation ( fig.  3) , phagocyte stimulation and disruption of mitochondrial electron transport chains (so that more electrons 'escape' to oxygen to form 02"-). Hence, almost any disease is likely to be accompanied by increased formation of reactive oxygen species, and it is not surprising that the list of diseases in which their formation has been implicated is long and grows longer (Halliwell 1987) . For atherosclerosis, rheumatoid arthritis, some forms of ARDS, reoxygenation injury , and traumatic or ischaemic damage to the central nervous system, there is reasonable evidence to suggset that free radical reactions make a significant contribution to the disease path- ology. As has been stressed previously, however (Halliwell & Gutteridge 1984) , it is equally likely that in some (perhaps most) diseases, the increased formation of reactive oxygen species is an epiphenomenon that is of little clinical relevance and antioxidants are unlikely to be of therapeutic benefit. This is by no means a new ' situation. Tissue injury usually increases synthesis of prostaglandins, yet cyclooxygenase inhibitors are of little benefit in many human diseases. Thus, in rheumatoid arthritis, they decrease pain and swelling of the joints, but fail to halt the joint destruction processes. . These general principles can help us to interpret the results of antioxidant therapy in various human diseases, and to make proposals for the future .

Attempts to use antioxidants in the treatment of human disease can be divided into 3 main areas:

1. Administration ofantioxidants that occur naturally in the human body, such as a-tocopherol, glutathione or superoxide dismutase (SOD).

2. Administration of synthetic antioxidants, such as probuco1 or che1ating agents that suppress iron ion-dependent free radical reactions .

3. The possibility that drugs developed to protect against-other mechanisms of tissue injury might have additional physiological action because they have antioxidant properties; for example, the angiotens in converting enzyme inhibitor captopril has been suggested to function additionally as an antioxidant in vivo (Bagchi et al. 1989 ). However, a detailed evaluation of these proposals has shown that the antioxidant activity of captopril at the concentrations actually achieved in vivo is likely to be very limited (Aruoma et al. 1991b ).

Copper-zinc containing SOD has a well-established anti-inflammatory effect in animal models of acute inflammation, in part because it can de-Drugs 42 (4) 1991 crease the number of neutrophils entering sites of inflammation. It also diminishes their adherence to endothelium (Suzuki et al. 1989 ). However, SOD enzymes of equal catalytic activity do not suppress acute inflammation to the same extent in rats, and it has been suggested that SOD must bind to specific sites to exert its anti-inflammatory activity. Hence, the surface charge distribution on the SOD protein is an important factor (Jadot et al. 1986; Schalkwijk et al. 1985) . Similar results have been obtained in studies of the ability of SOD to suppress adjuvant-induced polyarthritis in rats (Vaille et al. 1990 ).

There is a considerable amount of older literature claiming that injection of SOD into the knee joints of patients with osteoarthritis or rheumatoid arthritis produces a beneficial effect, but these data have been severely criticised (reviewed by Greenwald 1990) . Even the early proponents of the use of SOD in humans now admit that SOD is not, and will probably never become, a drug of choice in the-treatment of rheumatoid arthritis (Flohe 1988) . The author has been unable to find any report of a well-controlled clinical trial of the effect of SOD in inflammatory joint disease or autoimmune disease in a high-quality journal, although brief reports of uncontrolled studies continue to be published. Now that recombinant human copperzinc-containing (Ambrosio et al. 1986 ) and manganese-containing SODs (Beck et al. 1988 ) are available, such well-controlled studies should be . performed to settle the question as to whether SOD will be of any use as an anti-inflammatory agent. A wide variety of SOD conjugates are available , including polyethyleneglycol (PEG)-SOD, cationised SOD, genetically engineered SOD polymers, pyran SOD arid SOD-albumin complexes (reviewed by Greenwald 1990) . These have longer circulating half-lives than the unconjugated SOD molecules.

There has also been intense interest in the use of SOD in protection against reoxygenation injury. The reported effects of SOD in decreasing infarct size after prolonged myocardial ischaemia seem increasingly questionable (Bolli et al. 1991; Downey 1991 ; section 3.2) and so hope is being concen-trated in 2 areas: does a combination of SOD and a thrombolytic agent produce increased functional activit y and decreased mortality compared with a thrombolytic agent alone in cases of recent myocardial infarction , and would SOD be helpful in extending the preservation time of organs for transplantation? An answer to the first question must await the results of controlled clinical trials currently in progress, although Petch (1990) and Editorial (1989a) have reported that reperfusion injury does not seem to be a significant problem in the clinical use of thrombolytic agents. One reason for this could be that reperfusion in most animal model systems is sudden , whereas thrombolytic reperfusion in humans is slow, and slow reoxygenationmay minimise the injury (e.g. Perry & Wadhwa 1988) . Although markers of oxidative reperfusion injury can be detected during postischaemic reperfusion of the hearts of patients after open-heart surgery (del Nido et al. 1987; Ferrari et al. 1990) , the clinical significanceof this injury may be questioned (Bell et al. 1990; Hermens et al. 1988) . Even if human myocardial reoxygenation injury were clinically significant, SOD need not be the ideal protective agent. For example, SOD failed to protect against myocardial stunning in open-chest dogs, although the combination of SOD and catalase was protective ., Further comments about this are made in section 5.

Treatment of human kidney transplant recipients with bovine SOD produced no significant decrease in rates of acute renal failure, but there was a suggestion of benefit that merits further study (Schneeburger et al. 1989) . The reported ability of SOD in some animal experiments to offer.protection against endotoxic or hypovolaemic shock (Carden et al. 1989; Schneider 1988) , pancreatitis (Sanfey et al. 1984 ), gastrointestinal ischaemia (McCord 1985 Vohra et al. 1989 )and preservation of organs other than kidney (Fuller et al. 1988; Huang et al. 1987; Weiss et al. 1989) does lnot yet appear to have been assessed critically in humans. Even in animal experiments, there is considerable variabilit y in results between different laboratories (e.g. McCord 1985) .

The gene encoding human extracellular SOD 587 (EC-SOD; section 1.7) has been cloned and expressed (Tibell et al. 1987) . Since reoxygenation injury to tissues probably involves endothelial damage (Forman et al. 1989; Halliwell 1989a; Tsao & Lefer 1990 )and some EC-SODs bind to endothelia (Marklund 1990) , it has been proposed by Marklund's group that EC-SODs might be especially efficient as protective agents. To the author's knowledge, this proposal has not yet been tested in humans by controlled clinical trial. Inoue et al. (1990) constructed and expressed in yeast a fusion gene consisting of cDNA encoding human Cu-ZnSOD and heparin-binding peptide. The resulting enzyme retained activity and was able to bind to and protect endothelial cells.

Several low-molecular-mass compounds that react with 0 2'have been described. They usually contain transition metal ions. Examples include iron porphyrins (Nagano et al. 1989 ), a complex of manganese ions with the chelating agent desferrioxamine (Beyer & Fridovich 1989) or copper ions chelated to amino acids or to anti-inflammatory drugs (Sorenson 1984; Weser et al. 1980) . Superoxide scavengers that do not involve metal ions, such as nitroxides, have also been described (Mitchell et al. 1990 ). Apart from the various copper complexes (Sorenson 1984) little information is available on the toxicity of SOD mimics (Steinkuhler et al. 1990 ). Indeed, the potential use of these compounds in treatment of human disease is uncertain in view of the availability of recombinant human SOD enzymes of very low toxicity.

Prolonged deficiency of a-tocopherol, as in fat malabsorption syndromes, produces severe neurological damage (reviewed by Bieri et al. 1983 ). The antioxidant action of a-tocopherol in plasma lipoproteins might help to protect against atherosclerosis, as suggested by epidemiological studies (Gey et al. 1987; Gey 1990) . Hence, there are good reasons for maintaining an adequate vitamin E intake by eating a healthy diet and for giving supplementary vitamin E to patients with diseases af-fecting fat absorption. Therapeutic administration of vitamin E to premature babies has been reported to diminish the severity of retrolental fibroplasia (although there is some controversy about this: discussed in Ehrenkranz 1989) and the incidence of both haemolytic syndrome of prematurity and intraventricular haemorrhage (Chiswick et al. 1991; Sinha et al. 1987) . Vitamin E supplementation has also been recommended as beneficial in patients suffering from haemolytic syndromes caused by an inborn lack of glutathione synthetase or glucose-6-phosphate dehydrogenase in erythrocytes (e.g. Corash et al. 1980 ). However, it has been reported that high doses of intravenous vitamin E can produce an increase in the infection rate in premature babies, probably by depressing phagocyte function (Engle et al. 1988 ). Depression of phagocyte function can also be demonstrated in cells isolated from adults taking oral vitamin E but this effect does not seem to be clinically significant (Bendich & Machlin 1988) .

However, attempts to use vitamin E to treat human diseases in which oxidative stress is thought to be important, such as anthracycline-induced cardiac damage (reviewed by Powis 1989) , have been largely disappointing. One reason for this could be that it takes a considerable time to raise the vitamin E content of many tissues, especially that of brain, whereas oxidative damage can often be very fast (e.g. within minutes in the reoxygenation injury experiments described by Bolli et al. 1989 ). In addition, oxidative damage frequently occurs by mechanisms other than lipid peroxidation (Powis 1989 ; section 2.1).

Pharmacological doses of vitamin E have been claimed to be beneficial in thrombotic vascular disease (reviewed by Bieri et al. 1983 ), but even if true this might be due to effects other than antioxidant action, such as partial inhibition of lipoxygenase (Reddanna et al. 1985) or a weak suppression of platelet aggregation (Jandak et al. 1989) . It is also interesting to note that the reported protective effects of o-tocopheryl succinate against oxidative damage in isolatedhepatocytes could not be explained simply by its hydrolysis to free tocopherol and may have involved 'cytoprotective Drugs 42 (4) 1991 effects' of the succinate ester itself (Carini et al. 1990; Fariss 1990; Fariss et al. 1989) .

Several structural analogues of a-tocopherol with improved antioxidant activity (as assayed in vitro and/or tested in animals) have been described (Ingold et al. 1986 Mukai et al. 1989) . A watersoluble analogue of a-tocopherol, 'Trolox C', also exists (Thomas & Bielski 1987) but caution should be employed in using it upon humans because it has been shown to be capable of accelerating oxidative damage to substrates other than lipids in vitro (Aruoma et al. 1990 ). In general, the author feels that antioxidant inhibitors of lipid peroxidation are not likely to be of general use in treating human disease, because lipid peroxidation is often a late stage in the oxidative damage mechanism (section 2.1). However, they may have great benefit as protective agents in some cases, e.g. in protecting against the development of atherosclerosis, provided that they can enter and remain in LDL (Esterbauer et al. 1989) . The results of a clinical trial (Deprenyl and Tocopherol Antioxidative Therapy of Parkinsonism, or DATATOP) in which the combination of vitamin E with an inhibitor of monoamine oxidase B (deprenyl) is being evaluated in the treatment of Parkinson's disease are awaited with interest (Parkinson Study Group 1989) . So far, however, the beneficial effect of deprenyl has been much more obvious than that of a-tocopherol.

Vitamin C is essential in the human diet as a cofactor for several hydroxylase enzymes (reviewed by Levine 1986 ) and deficiency of it should be avoided, especially as deficiency might also predispose to atherosclerosis, as is suggested by .epidemiological studies (Gey et al. 1987) .It is possible that smokers need a higher vitamin C intake than nonsmokers , since smoking appears to lead to increased vitamin C consumption (e.g. Chow et al. 1986 )and vitamin C is thought to be an important antioxidant in the lung (Bendich et al. 1986; Cross et al. 1990) . It may also protect against the carcinogenicity of nitrosamines (Shamberger 1984; Tannenbaum et al. 1991 ). In the absence of metal ions, ascorbate is a powerful antioxidant in vitro (Frei et al. 1989 ), but in their presence, it can be dangerous, as discussed in section 1.7.

Administration of very high doses (up to 20g daily in some cases) of vitamin C to healthy humans appears to do no significant harm (Bendich et al. 1986 ) but , to the author's knowledge, no clinical benefit has yet been demonstrated in wellcontrolled clinical trials in any disease: indeed, highdose vitamin C appeared to lead to lethal haemorrhagic tumour necrosis in 4 patients with advanced cancer (Cameron & Campbell 1974) . Cessation of megadose therapy may lead to rebound low levels of vitamin C in body fluids (e.g. Tsao & Salimi 1984) .

A number oflipophilic ascorbate esters (e.g. ascorbyl palmitate) have also been described (e.g. Baader et al. 1988 ), but reports of properly controlled trials of their use in the treatment of human disease have not yet appeared.

Under certain circumstances, neutrophils can potentiate injury to tissues (section 3.2). For example, antibodies preventing neutrophil adherence decreased infarct size in a dog model of cardiac reoxygenation injury after ischaemia (Werns & Lucchesi 1990 ). The first step in neutrophil accumulation in several tissues involves adherence to the endothelium, a process that is regulated by many substances produced by endothelial cells. For example, prostacyclin decreases adherence, whereas Oiincreases it (Suzuki et al. 1989 ).

The nucleoside adenosine (adenine-ribose) affects the function of both neutrophils (e.g. it decreases Oiproduction) and endothelium and it has been shown to have protective effects in some animal models of reoxygenation injury (Forman et al. 1989; Grisham et al. 1989; Pitarys et al. 1991) . Indeed, it has been argued that adenosine is a natural anti-inflammatory agent during transient injury , helping to prevent premature activation of the inflammatory response (Cronstein et al. 1990; Engler 1987) . For example, 5-amino-4-imidazole carboxamide-riboside accelerates adenosine release from ischaemic cells, and has been reported to be 589 protective against ischaemic injury (Engler 1987) . Compounds acting in a comparable way to adenosine might be very useful as anti-inflammatory agents and in protection against reoxygenation injury (e.g. Kaminski & Proctor 1990) , to the extent that this phenomenon is clinically important (section 4.1.1).

Lactoferrin is an iron ion-binding protein , similar to transferrin but with a higher affinity for iron ions (especially at acidic pH values). It is present in human secretions (milk, saliva, tear fluid, nasal secretions) and released by activated neutrophils. Iron ions bound to lactoferrin are unable to catalyse free radical reactions (Aruoma & Halliwell 1987; . Indeed, it has been argued that the release of lactoferrin by activated neutrophils is a protective mechanism that helps to diminish iron-dependent free radical reactions in the surrounding area (Halliwell et al. 1981) . For example, Ward et al. (1983) found that lactoferrin was able to protect rats against lung injury mediated by reactive oxygen species. Lactoferrin has many other effects on phagocyte function , however (Birgens 1984) . Several groups are cloning and expressing the gene for human lactoferrin (e.g. Stowell et al. 1991 ) and it will be interesting to see if the human protein has any therapeutic uses.

Subjecting cells to oxidative stress frequently leads to depletion of intracellular nicotinamide nucleotides by activation of poly(ADP-ribose) synthetase (section 2.1). Several authors have reported protection against cell killing under these circumstances by inhibitors of poly(ADP-ribose) synthetase, such as aminobenzamide, or by supplementation of the cell culture medium with nicotinamide or nicotinic acid (niacin) [e.g. Schraufstatter et al. 1987; Weitberg 1989 ]. Weitberg (1989) found that the prolonged ingestion of supplementary nicotinic acid by human volunteers led to an increase in the NAD+ concentrations in their circulating lymphocytes, and that lymphocytes isolated from these volunteers were more resistant to oxidative dam-age. Wang et al. (1990) found that pretreatment of hamsters with nicotinamide decreased the lung fibrosis observed after subsequent exposure to the antitumour antibiotic bleomycin.

Whether these various observations will tum out to have therapeutic relevance to humans is unclear. A potential problem with the use of ADPribose synthetase inhibitors is that blocking the action of this enzyme may decrease DNA repair, so increasing the risk of carcinogenesis: this effect has been demonstrated in mice whose pancreatic ,B-cells were damaged by treatment with streptozotocin, leading to diabetes. The animals could be protected against the diabetogenic action of streptozotocin by administration of ADP-ribose synthetase inhibitors, but a significant number of animals then developed insulinomas (Okamoto 1986 ). The results with nicotinamide again illustrate the point that a healthy diet may be a good protection against tissue injury due to oxidative damage .

Glutathione has multiple functions in human metabolism, including the detoxification of xenobiotics. It is also important in antioxidant defence, e.g. by helping to protect DNA against damage by ionising radiation and by acting as a substrate for glutathione peroxidases and transferases (Held 1988; Ross 1988) . There have therefore been many suggestions that reduced glutathione (GSH) may be therapeutically useful as an antioxidant and as a general cytoprotective agent, e.g. in the preservation of organs for transplantation and in protecting against tissue damage by cytotoxic drugs such as cyclophosphamide (Currin et al. 1990; Friedman et al. 1990; Menasche et al. 1986 ). Use of aerosolised GSH solutions has been suggested as a means of diminishing lung damage by reactive oxygen species generated extracellularly from activated phagocytes in such diseases as cystic fibrosis (Buhl et al. 1990 ). There is also evidence for abnormalities ofGSH metabolism in HIV-infected cells and the use of GSH or its precursors in asymptomatic HIV-positive and AIDS patients has been suggested (reviewed by Halliwell & Cross 1991) .

Drugs 42 (4) 1991 GSH itself is not rapidly taken up by cells, but several approaches can be used to overcome this . Methyl and ethyl esters of glutathione have been described, which can apparently cross cell membranes and be hydrolysed to GSH within the cell (Anderson & Meister 1989a) . However, there have been some reports of toxicity of GSH esters, possibly due to contaminating compounds arising during their preparation (Thomalley 1991). The compound L-2-oxathiazolidine-4-carboxylate has been reported to be hydrolysed to cysteine in vivo,which may lead to an increased synthesis of GSH (Meister 1988; Porta et al. 1991) . N-Acetylcysteine has been used as an antioxidant in a wide variety of experiments (Moldeus et al. 1986 ) and is very effective in the treatment of paracetamol (acetaminophen) overdosage in humans (Smilkstein et al. 1988) . Paracetamol is metabolised by the hepatic cytochrome P450 system to a reactive quinoneimine that can combine with GSH , so causing severe GSH depletion in the liver. N-Acetylcysteine may protect by entering cells and being hydrolysed to cysteine, which stimulates GSH synthesis . Additionally, N-acetylcysteine has also been widely used in humans for treatment of various respiratory disorders; it has very limited toxicity but its therapeutic benefits in these conditions have been questioned.

Mixtures of thiols with transition metal ions can be cytotoxic because of reactions that produce both oxygen radicals (02'and 'OH) and sulphur-containing radicals such as RS·, RSO· and RSOz· (Rowley & Halliwell 1982; Sevilla et al. 1989) . GSH is unstable in solution, particularly if iron or copper ions are present, and the oxidation products can cause damage. This must be borne in mind if solutions containing GSH or other thiols are to be stored for any length of time, e.g. in cell culture media or in fluids for organ preservation (e.g. Boudjema et al. 1990; Currin et al. 1990 ). Cysteine seems to be more toxic than GSH in such systems (Saez et al. 1982) , perhaps because it oxidises faster and because the cysteine-So radical can abstract hydrogen atoms from fatty acids and initiate lipid peroxidation (Schoneich et al. 1989 ). Thus, cysteine accelerates iron ion-dependent peroxidation of lipids, whereas GSH and N-acetylcysteine do not (Haenen et al. 1989) . Some thiols can also mobilise iron ions from ferritin (e.g. Bonomi et al. 1989) . Indeed, cysteine is much more toxic when administered to animals that are N-acetylcysteine or GSH (Anderson & Meister 1989b) . Similarly, high levels of plasma homocysteine in homocystinuric patients predispose to atherosclerosis (Ueland & Refsum 1989) .

Sulphur-containing radicals can also be formed when thiols are oxidised by ·OH and possibly by other reactive oxygen species: 591 sitising drugs (Mathews-Roth 1987) . They may also have an anticancer effect (Bendich & Olson 1989; Stahelin et al. 1991) . It thus seems important to maintain an adequate dietary intake of vegetables. Carotenoids can also act as inhibitors of lipid peroxidation under certain circumstances, but their actions are complex (discussed by Krinsky 1989). Whether they will ever have a major therapeutic use remains unclear as yet.

Carotenoid pigments, such as {3-carotene, are powerful quenchers of singlet 02 and have found therapeutic use in protecting porphyria patients against photosensitisation damage to the skin, and in protecting animals against damage by photosen-As has been stressed previously (Halliwell 1990 ), when proposing the use of scavengers to remove reactive oxygen species, it is necessary to ask what happens to the radicals derived from the scavenger.

N-Acetylcysteine has been reported to have protective effects in many animal models of lung injury. By contrast, it potentiated the lung damage produced by the anticancer drug bleomycin in hamsters (Giri et al. 1988 ). Bleomycin-dependent lung damage is thought to involve formation of an iron ion-bleomycin complex, which produces strand breakage in DNA in the presence of certain reducing agents (reviewed by Halliwell & Gutteridge 1989) . In vitro, N-acetylcysteine did not potentiate iron ion-dependent damage to DNA by bleomycin, whereas cysteine did (Evans & Halliwell, unpublished data) and so it is possible that the N-acetylcysteine acted as a precursor of intracellular cysteine, which could reduce the iron-bleomycin complex upon the nuclear DNA. Thus, indiscriminate use ofthiol compounds as therapeutic agents cannot be recommended.

Many thiols have been tested for their ability to protect cells and small animals against ionizing radiation. They include GSH and its precursors (cysteine, cysteamine), dimercaprol (British Anti-Lewisite) and mercaptopropionylglycine (von Sonntag 1987) . In general, the author's impression is that cysteamine (e.g. Isseis et al. 1987 ) and cysteine (section 4.1.8) are more toxic than the others, perhaps because they oxidise faster if transition metal ions are present and/or produce sulphur-containing radicals that are more damaging to biological molecules.

The thiol compound mercaptopropionylglycine (MPG) has been shown to be protective against reperfusion injury after ischaemia in animal models of myocardial stunning and infarction (discussed by Bolli et al. 1989; Puppo et al. 1990 ). It may act by its ability to scavenge reactive oxygen species, but which species it acts upon is as yet uncertain (Puppo et . al. 1990) . Several clinical studies have claimed that MPG is beneficial in the treatment of various diseases (listed in Bolli et al. 1989 ) but wellcontrolledtrials of its effects against reperfusion injury do not appear to have been published.

Ebselen (PZ51; 2-phenyl-I ,2-benzisoselenazolin-3(2H)-one) is an organoselenium compound that catalyses removal of peroxides by reaction with GSH , and was developed as a glutathione peroxidase 'mimic'. It has been shown to have anti-in-flammatory effects in a number of small mammals (Parnham et al. 1991) . However, it is uncertain whether these effects can be attributed to removal of H202, since ebselen also inhibits cyclooxygenase, lipoxygenase, the production of reactive oxygen species by activated neutrophils and lipid peroxidation in membranes (Cotgreave et al. 1989; Hayaishi & Slater 1986; Leurs et al. 1989; Parnham et al. 1991) . Ebselen does not appear to have been effectively tested in controlled clinical trials for beneficial effects in humans to date, although its metabolism in the human body has been described (Fischer et al. 1988) . One potentially worrying observation is that ebselen also inhibits elongation and desaturation of long-chain fatty acids in rat liver microsomes (Laguna et al. 1989) , although the toxicity of the drug to animals has been reported as very low (Parnham et al. 1991) .

Allopurinol inhibits the production of Oi-, H202 and uric acid by the enzyme xanthine oxidase. It inhibits by first acting as a substrate for the enzyme, so producing the true inhibitor, oxypurinol (Spector 1988) . Oxypurinol is also the major metabolite formed when allopurinol is adminstered to humans (Reiter et al. 1983) . When it was proposed that xanthine oxidase is the major source of the reactive oxygen species generated when ischaemic tissues are reoxygenated (McCord 1985) , interest naturally arose in the use of allopurinol or oxypurinol as protectors against tissue damage. Indeed, some animal experiments found that allopurinol has protective effects against cardiac, cerebral or gastrointestinal reoxygenation injury (Itoh et al. 1986; McCord 1985; Parks 1989) as well as in some animal models of haemorrhagic shock (Allan et al. 1986 ). It might be thought more logical to use oxypurinol as a protective agent, since it is the true inhibitor of xanthine oxidase (Spector 1988) . However, allopurinol was reported to be more effective than oxypurinol in reducing urinary uric acid excretion in humans (Chalmers et al. 1968) . Similarly B103U, an analogue of oxypurinol in which the oxygen at position 6 is replaced by Drugs 42 (4) 1991 sulphur, did not inhibit the enzyme any better in vivo than did oxypurinol (Spector et al. 1989) , although it was a more powerful inhibitor in vitro.

Since allopurinol is widely used to treat hyperuricaemia in humans (Chalmers et al. 1968 ), its toxicological profile is well known and it might therefore seem to be the agent of choice in testing for beneficial effects in diseases to which ischaemia/reperfusion is thought to contribute: it has always seemed to the author that preventing the formation of excess reactive oxygen species is more likely to be effective than attempting to scavenge them once they have been formed. Unfortunately the evidence that xanthine oxidase makes any significant contribution to the generation of reactive oxygen species in human heart and brain is very weak (section 3.2; Wajner & Harkness 1989; Werns & Lucchesi 1990 ). In addition, many studies in which protection has been shown have required pretreatment of animals with the drug, which is not usually possible in clinical cases of human ischaemia/reperfusion. However, xanthine oxidase activity has been detected in human gut, in the palmar fascia of patients with Dupuytren's contracture (Murrell et al. 1987) , in the circulation of patients after reperfusion of a tourniqued limb (Friedl et al. 1990 ) and in the synovia of patients with rheumatoid arthritis (Merry et al. 1989) . The therapeutic possibilities of treatment with allopurinol in these diseases have been discussed, but not yet evaluated clinically to date. Allopurinol and oxypurinol could also be beneficial in enhanced preservation of organs for transplantation (Chambers et al. 1987) . Salim (1990) has recently reported that allopurinol was as effective as cimetidine in preventing the recurrence of duodenal ulcer in humans.

As well as inhibiting xanthine oxidase, oxypurinol can also scavenge HOCI (Grootveld et al. 1987) , which might contribute to its protective effects under certain circumstances. In vitro, both allopurinol and oxypurinol can react with 'OH, but the concentrations achieved in vivo are probably too low for this to be a significant mechanism of protection (discussed by Halliwell 1990 ). An additional mechanism by which allopurinol or oxy-purinol could protect tissues is by preventing oxidation of hypoxanthine, so enhancing its salvage for reincorporation into adenine nucleotides when the tissue is reoxygenated (Lasley et aI. 1988) .

Other inhibitors of xanthine oxidase that have been described in the literature include lodoxamide (White 1981) , pterin-6-aldehyde (Nishino & Tsushima 1986 ) and amflutizole (Wems & Lucceshi 1990) .

Generation of reactive oxygen species by activated phagocytes may contribute to inflammatory and reoxygenation injury (section 3.2). Thus, it is possible that inhibitors of phagocyte recruitment, adherence to endothelium, or activation of the respiratory burst oxidase complex could be therapeutically useful, as has already been discussed in the case of adenosine (section 4.1.5). Several anti-inflammatory drugs have been suggested to suppress Oiformation by phagocytes, but convincing evidence for this mechanism of action in vivo has been obtained only in a few cases (discussed by Halliwell et aI. 1988a) . Cross (1990) has reviewed the compounds that have been shown to inhibit phagocyte Oigeneration in vitro: none of them appears suitable for therapeutic use as yet. Hart et aI. (1990) reported that apocynin, a phenolic compound of plant origin, is oxidised to a toxic product by phagocytes containing peroxidase (such as neutrophils, which possess myeloperoxidase) and thus exerted an anti-inflammatory effect in rats. Only activated phagocytes (producing the H202 necessary for peroxidase action) would be affected.

Chelating agents have been successfully used in the treatment of metal ion poisoning for many years, e.g. dimercaprol to treat arsenic poisoning, penicillamine and trientine to treat Wilson's disease and diethylenetriaminepentaacetic acid to remove lead or plutonium (Jones 1991; Jones & Pratt 1976) . Graf and Eaton (1990) have pointed out that phytic acid (inositol hexaphosphate) , an insoluble constituent of the human diet, is also a powerful 593 chelator of iron ions that suppresses iron-dependent free radical reactions. Based on this observation, the potential therapeutic use of inositol 1,2,6triphosphate in human disease is being evaluated (Claxson et aI. 1990 ).

The only reasonably specific iron ion chelator approved for routine clinical use in humans is deferoxamine (sometimes called desferrioxamine: commercial preparations of the drug are usually deferoxamine B methanesulphonate). Deferoxamine, which is produced by Streptomyces pilosus, is widely and effectively used for the prevention and treatment of iron overload in patients who have ingested toxic oral doses of iron salts, or who require multiple blood transfusions, e.g. for the treatment of thalassaemia (Modell et aI. 1982) . Deferoxamine also binds Al+++ ions, although with a stability constant 6 orders of magnitude less than for its binding of Fe"?" ions, and has been used to decrease body aluminium loads in renal dialysis patients (McCarthy et aI. 1990 ). It has recently been claimed that deferoxamine has beneficial effects in the treatment of patients with Alzheimer's disease (McLachlan et aI. 1991) .

Deferoxamine is a powerful chelator of Fe"?". In the presence of physiological buffer systems, deferoxamine inhibits iron ion-dependent lipid peroxidation and the generation of highly-reactive oxidising species such as 'OH from Oiand H202

in the presence of iron ions (reviewed by Halliwell I989d). Large doses of deferoxamine can be injected into animals or humans: at least 50 mg/kg bodyweight per day appears fairly safe in thalassaemic patients (Davis et aI. 1983 ) provided that they do not become iron-deficient (Porter et aI. 1989a) . Hence, deferoxamine has been proposed for use as a 'probe' to study the importance of irondependent free radical reactions as mediators of tissue injury (Gutteridge et aI. 1979 ). Indeed, deferoxamine has been used in this way in many animal model systems (reviewed by Halliwell 1989d) . It is generally observed to be protective, probably by its iron-chelating actions in most cases, although it has also been shown to interfere with the uptake of paraquat by isolated lung cells (van der Wal et aI. (990), which partially explains its pro-tective effect against cytotoxicity of this herbicide. Deferoxamine can also react directly with several reactive oxygen species, including R02", Oi-, "OH, HOC1, oxidants produced by mixing haem proteins with H202, and peroxynitrite formed by reaction of NO with Oi-. Most of these reactions are of little or no significance in vivo, simply because the concentration of deferoxamine achieved during normal therapeutic use is too low for scavenging of reactive oxygen species to be a feasible mechanism of action (Halliwell 1989d) . However, the ability of deferoxamine to scavenge peroxynitrite has been suggested to account for its ability to protect against reoxygenation injury in some myocardial systems (Beckman et a1. 1990 ), a proposal which merits further investigation. It has also been shown (at high concentrations) to block the formation of xanthine oxidase activity in cultured bovine pulmonary endothelial cells ( Rinaldo & Gorry 1990) . Deferoxamine is very poor at removing iron ions bound to transferrin or lactoferrin, which may be advantageous since iron bound to these proteins does not stimulate free radical reactions (Aruoma & Halliwell 1987; Gutteridge et a1. 1981) .

Deferoxamine, at therapeutically relevant concentrations of 20 to 100 ,umoljL, can inhibit cell proliferation in vitro, and in vivo (Estrov et al. 1988) . It may do so by depriving the cells of iron and inhibiting the enzyme ribonucleoside diphosphate reductase, but there is considerable debate as to whether or not this is the only cytostatic mechanism. Thus, when deferoxamine is administered by continuous infusion over several days, as in some studies of its effect upon chronic inflammation or autoimmune disease in animals, it is possible (but remains to be proved) that inhibition of the proliferation of inflammatory cells such as lymphocytes could contribute to the protective effects observed. The cytostatic action of deferoxamine or other iron ion-chelating agents might conceivably be of therapeutic use in suppressing Iympocyte-mediated reactions (Bierer & Nathan 1990; Bradley et al. 1986 ), e.g. graft rejection or the replication of HIV (Baruchel et al. 1991 ; Tabor et a1. 1991) .

Drugs 42 (4) 1991 Because iron is essential for normal cell function, prolonged administration of any powerful iron-chelating agent to humans (other than patients suffering from iron overload) will probably cause serious side effects. Hence, proposals to use long term deferoxamine administration in the treatment of chronic disease . in humans, such as Alzheimer's disease or rheumatoid arthritis, must be approached with caution , as must prolonged treatment of aluminium-overloaded patients with this drug (Blake et a1. 1985; McCarthy et a1. · 1990) . However, the side effects of chronic deferoxamine administration should not preclude its acute use, e.g. in comb ination with thrombolytic agents in the treatment of myocardial reperfusion injury, in haemorrhagic shock (Sanan et a1. 1989) or in ameliorating cell damage by toxins such as the anthracycline anti tumour antibiotics (Halliwell et a1. 1988b ). Deferoxamine might be especially useful in myocardial reperfusion systems (Menasche et a1. 1990 ). For example, it can sometimes protect against myocardial stunning much more effectively than does SOD (Bolli et a1. 1990 ). Thus , if clinical trials (section 4.1.1) show that recombinant human SOD has a beneficial effect when combined with thrombolytic agents in the treatment of recent myocardial infarction, then the possibility of using deferoxamine as a much cheaper alternative is well worth exploring. Deferoxamine might also have applications in the preservation of organs for transplantation (Gower et al. 1989b; Yoon et al. 1989) . It is interest ing to note that, despite the putative iron ion dependency of the lung damage induced by the anticancer drug bleomycin in animals (section 4.1.8) neither deferoxamine nor other iron chelators so far tested appear to offer any protective effects (Tryka 1989; Ward et a1. 1988) . Hallaway et a1. (1989) have described high-molecular-mass derivatives of deferoxamine, made by attaching it to polymers such as dextran or hydroxyethyl starch. Conjugation to these polymers was reported not to diminish Fe+++-binding activity, but to markedly raise the LDso and the circulating plasma lifetime in animals. Whether these high-molecular-mass forms of deferoxamine will be therapeutically useful remains to be seen.

A major problem with deferoxamine is that it is not absorbed through the gut and has to be administered to thalassaemic patients by intravenous or subcutaneous infusion, leading to poor patient compliance. The drug is also expensive. There has therefore been a search for orally active chelating agents that are, hopefully, cheaper (Jones 1982) . Some work has been reported with pyridoxal isonicotinoyl hydrazone (discussed by Hershko 1988), deferrithiocin (Biener & Nathan 1990; Wolfe 1990) and 2,3-dihydroxybenzoate (Grady et al. 1978) , but most attention is currently concentrated on the 3hydroxypyrid-4-ones. For example, 1,2-dimethyl-3hydroxypyridin-4-one (LI) has undergone preliminary tests in humans. It appears to be more toxic than deferoxamine (Editorial I989b) , and other hydroxypyridones (such as the 1,2-diethyl compound) are now undergoing evaluation (Porter et al. 1989b (Porter et al. ,c, 1990 Dobbin & Hider 1990; Hershko et al. 1991) .

The increased toxicity of hydroxypyridones, when compared with deferoxamine, might be due to featu res of their chemistry (Halliwell & Grootveld 1987) . Three moles of chelator are needed to complex I Fe+++ ion completely, whereas deferoxamine chelates Fe+++ at a I : I molar ratio: the deferoxamine molecule 'wraps around' the iron ion, encasing it in an envelope of organic material. Upon dilution, 3 : I hydroxypyridone: iron ion complexes tend to dissociate into 2: I and I : I complexes, leaving available coordination sites on the iron that could allow the complexes to catalyse free radical reactions. Indeed, Kontoghiorghes et al. (1986) found that 3 : I chelator-iron complexes did not stimulate iron-dependent free radical reactions, whereas 2 : I or I : I complexes seemed able to do so. In addition, hydroxypyridones can mobilise iron ions from transferrin (Kontoghiorghes 1986a) or lactoferrin (Kontoghiorghes I986b) . Thus, one can envisage a scenario in which iron ions safely protein-bound (i.e. unable to stimulate free radical reactions) are mobilised and transfemed to other sites in the body at which damage can be caused (Halliwell & Grootveld 1987) . Caution therefore should be employed in proposing long term administra-595 tion of these compounds to humans, especially those not suffering from iron overload, without close medical supervision (Hershko 1988; Nathan & Piomelli 1990) .

One aspect of the use of iron-chelating drugs that deserves further attention is their antimalarial action . Deferoxamine (Hershko & Peto 1988; Pollack et al. 1987) inhibits the growth of malarial parasites in culture and in infected animals. Hydroxypyridones are also effective against parasites in vitro (Heppner et al. 1988; Hershko et al. 1991) . Similar suggestions have been made for use of chelators against trypanosomal infections (Shapiro et al. 1982) . Clark et al. (1986) have proposed that reactive oxygen species are involved in the pathology of malarial infections, and various antioxidants have been found to be beneficial in malariainfected animals (e.g. Thumwood et al. 1989) . By contrast, several antiparasitic drugs have been claimed to exert their action by increased generation of reactive oxygen species in vivo (reviewed by Docampo 1990) .

Another compound whose action in vivo may relate to iron ion chelation is ICRF-187, which was originally synthesised as an anti-tumour agent and extensively tested in animals and humans, but with disappointing results. ICRF-187 is hydrolysed in vivo to produce a metal ion chelator structurally similar to edetic acid (EDTA). ICRF-187 has been reported to ameliorate the cardiotoxicity of anthracyclines in human breast cancer patients, while not affecting the therapeutic efficiency of these drugs (Speyer et al. 1988 ). It has therefore been suggested (Halliwell et al. 1988b; Halliwell & Bomford 1989 ) that careful use of low concentrations of chelating agents may help to minimise anthracycline-dependent cardiotoxicity in other cancers, particularly the leukaemias, in which chemotherapy can cause transient iron overload (section 3.3). ICRF-187 is the (+)-isomer of the racemic mixture first described as ICRF-159 [I ,2-bis-(3,5-dioxopiperazine-I -yl)propane]. ICRF-187 is preferred for use because of its increased solubility in water. It has also been reported to protect against the diabetogenic action of alloxan in mice, which is thought to be due to increased free radical formation in the pancreatic islets (El-Hage et al. 1981) .

Since iron-dependent free radical reactions appear to make significant contributions to injury to the CNS after stroke or trauma (section 3.3), several antioxidants (e.g. U74006F) have been developed that can cross the blood-brain barrier (Hall & Braughler 1989) . These compounds are 21-aminosteroids ('lazaroids') without glucocorticoid activity and they are powerful inhibitors of iron iondependent lipid peroxidation in vitro. although whether they act in vivo as iron ion chelators and/ or as chain-breaking antioxidants is unclear as yet. They have been reported to diminish brain or spinal cord injury resulting from stroke or trauma in animal experiments (Hall & Braughler 1989; Jacobsen et al. 1990; Steinke et al. 1989) . Results of clinical trials in humans are awaited with interest.

Several phenolic chain-breaking antioxidants have been proposed for use in humans. They include do phenol), and R-830 (2,6-di-tert-butyl-4-[2'-thenoyl]phenol). All have shown anti-inflammatory effects in animal models of acute inflammation (Swingle etal. 1987) . R-830 is reported to be in early clinical trials as a topical preparation for skin inflammation, but little information on human use of the other compounds has appeared. It must be remembered that chain-breaking antioxidants can often show pro-oxidant properties towards nonlipid biomolecules (section 2.2), that any anti-inflammatory action they have could be related to interference with the synthesis of prostaglandins and/or leukotrienes (e.g. by inhibiting lipoxygenase and/or cyclooxygenase) rather than by inhibiting lipid peroxidation, and that lipid peroxidation is often not a major mechanism of cell injury by oxidative stress (section 2.1). These reasons may account for the observation that the antiinflammatory activity of commercially available Drugs 42 (4) 1991 phenolic antioxidants has been described as 'not impressive' (Swingle et al. 1987) .

Many flavonoids and other phenols of plant origin have been reported to have anti-inflammatory effects in some animal model systems (Lewis 1989; Villar et al. 1984) and to inhibit lipid peroxidation in vitro (reviewed by Larson 1988) , but similar cautions apply (e.g. Donatus et al. 1990; Hodnick et al. 1986; Laughton et al. 1989) . Flavonoids are often powerful inhibitors of lipoxygenase and cyclooxygenase (e.g. Moroney et al. 1988 ) and they exert a wide range of other biochemical effects, so that it is usually difficult to attribute any biological actions that they have to antioxidant activity. Several plant phenolics can be oxidised by myeloperoxidase to give toxic products (Simons et al. 1990 ), which may explain some of their reported effects in suppressing Oigeneration by activated neutrophils (Hart et al. 1990 ). Some phenols can also exert pro-oxidant effects in vitro (Laughton et al. 1989 ).

Interest in the role of lipid peroxidation in atherosclerosis (section 3.2) was raised when it was found that the phenolic compound probucol (4,4'- [isopropylidenedithio] bis [2,6-di-tert-butylphenol]) has a marked antiatherogenic effect in hyperlipidaemic rabbits (Carew et al. 1987 ). Probucol was introduced as an agent that lowers blood cholesterol concentrations, but it is also a powerful chainbreaking inhibitor of lipid peroxidation (Parthasarathy et al. 1986 ) that does not appear to have pro-oxidant activity to other biomolecules (Aruoma et al. 1990 ). The potential therapeutic use of probucol to diminish the development of atherosclerotic lesions in coronary arteries after bypass surgery is currently being evaluated. There is no direct proof as yet that any antiatherogenic effect of probucol is actually related to an inhibition of lipid peroxidation and not to other effects on monocytes and macrophages within atherosclerotic lesions.

Many drugs already used in humans have been discovered to be capable of acting as antioxidants in vitro in some assay systems . They include non-steroidal anti-inflammatory drugs (discussed by Halliwell et al. 1988a) , the angiotensin converting enzyme inhibitor captopril (Bagchi et al. 1989) , the sulphasalazine metabolite 5-aminosalicylate (Aruoma et al. 1987 ) and several compounds acting as Car" antagonists (e.g. Janero et al. 1988 ). However, very few drugs are present in vivo at concentrations that would allow them to compete with biological molecules for reaction with 'OH or HOCI (table I): possible exceptions include salicylate, penicillamine and gold thiols as used in the treatment of rheumatoid arthritis (reviewed by Halliwell et al. 1988a ) and sulfasalazine or aminosalicylates as used in the treatment of inflammatory bowel disease (Ahnfelt-Ronne et al. 1990 ). For drugs that can act as radical scavengers in vivo, it is important to ensure that toxic drug-derived radicals are not produced during the reaction. For example, sulphur-containing radicals derived from penicillamine are capable of inactivating clq-antiproteinase, the major inhibitor of serine proteases in human extracellular fluids (Aruoma et al. 1989c ).

Tissue injury mechanisms in human disease are very complex. Taking ischaemia/reperfusion injury as an example, the extent of protection by antioxidants in different animal model systems can vary enormously. Just by slight changes in experimental conditions, a mechanism of tissue injury that is largely free radical-dependent can be easily changed to something that is scarcely affected by antioxidants. For example, patients with myocardial infarction are all different. They have had different periods of ischaemia. Some of them may have been treated with thrombolytic agents, some not, and so on. Thus , calcium blockers or free radical scavengers alone may not afford satisfactory protection against ischaemia/reperfusion after thrombolysis. To overcome this we might attempt to protect against all potential mechanisms of cell injury simultaneously. Polypharmacy is not encouraged, but in fact quite often a single drug may act by multiple mechanisms. Many of the drugs in 597 clinical use may well be doing several different things.

Drugs with multiple mechanisms of protective action, including antioxidant properties, may be one way forward to minimising tissue injury in human disease. Certainly, the potential therapeutic effects of N-acetylcysteine in HIV-I infection (Halliwell & Cross 1991) , probucol in atherosclerosis, ebselen in chronic inflammation and iron chelators in Alzheimer's disease, leukaemia, organ preservation and malaria merit urgent evaluation. Probably less attention should be paid to SOD and vitamin E, which have been therapeutically disappointing (except for vitamin E in ameliorating the consequences of prematurity and fat malabsorption syndromes).

Clinical evidence supporting the radical scavenger mechanism of 5-aminosalicylicacid

The role of free oxygen radicals and sperm function

Dependence on 02-generation by xanthine oxidase of pathogenesis of influenza virus infection in mice

Stimulation of lipid peroxidation increases the intracellular calcium content of isolated hepatocytes

The protective action of allopurinol in an experimental model of haemorrhagic shock and reperfusion

Reduction in experimental infarct size by recombinant human superox ide dismutase: insights into the pathophysiology ofreperfusion injury

Uric acid provides an antioxidant defense in humans against oxidant-and radical-causing aging and cancer: a hypothesis

Glutathione monoesters

Marked increase of cysteine levels in man y regions of the brain after administration of 2-oxothiazolidine-4-carbox ylate

Endothelial cells as a source of oxygen free radicals: an ESR study

DNA strand breaks photosensitized by benoxaprofen and other non-steroidal antiinflammatory agents

Evaluation of the ability of the angiotensin-converting enzyme inhibitor captopril to scavenge reactive oxygen species

Superoxide-dependent and ascorbatedependent format ion of hydroxyl radicals from hydrogen peroxide in the presence of iron

The scavenging of oxidants by sulphasalazine and its metabolites

Iron ion-dependent modification of bases in DNA by the superoxide radical-generating system hypoxanthine/xanthine oxidase

The mechanism of initiation of lipid peroxidation

Apparent inact ivation of a l-antiproteinase by sulphur-containing radicals derived from penicillamine

An evaluat ion of the antiox idant and potential pro-oxidant propert ies of food additives and of Trolox C, vitamin E and probucol

Copper-iondependent damage to the bases in DNA in the presence of hydrogen peroxide

Photod ynamic therap y-achievements and prospects

The suppression of gran-uIOC)1e functions by lipophili c antioxidants

Enhanced prostaglandin production in the ischemic-reperfused myocardium by captopril .Iinked with its free radical scavenging action . Prostaglandins

What is the significance of ceroid in human atherosclerosis? Archives of Pathology and Laboratory Medicine III

Efficient production of active human manganese superoxide dismutase in Escherichia coli

Ischaemic injury mediator

Apparent hydroxyl radical product ion by peroxynitrite : implications for endothelial injury from nitric oxide and superoxide

Inflammatory response, neutrophil activation, and free radical production after acute myocardial infarction : effect of thrombolytic treatm ent

Biological actions of carotenoids

The ant ioxidant role of vitamin C

Safety of oral intake of vitamin E

Frido vich I. Chara cterization of a supero xide dismutase mimic prepared from desferrioxamine and Mn02

Iron mobilization from ferritin by superoxide derived from stimulated polymorphonuclear leukocytes: possible mechanism in inflammation diseases

Medical uses of vitam in E

The effect of desferrithiocin, an oral iron chelator, on T-cell function

The biological significance of lactoferrin in haematology

Cerebral and ocular toxicity induced by desferrioxamine

On the limited ability of superoxide to release iron from ferritin

Oxygen-derived free radicals and postischemic myocardial dysfunction ('stunned myocardium')

Superoxide dismutase ten years later: a drug in search of a use

Marked reduction of free radical generation and contractile dysfunction by antioxidant therap y begun at the time of reperfusion

lronmediated radical reactions upon reperfusion cont ribute to myocardial stunning

Molecular aspects of the removal of ferritin-bound iron by DL-dihydrolipoate

Effect of oxidized and reduced glutathione in liver preservation

Desferrioxamine treatment prevents chronic islet allograft damage

Ionizing radiation-induced mutagenesis

Endothelium-derived relaxing factor (nitric oxide) lipoprotein oxidation and atherosclerosis

Systemic glutathione deficiency in symptom-free HIV-seropositive patients

Augmentation of glutathione in the fluid lining the epithelium of the lower respiratory tract by directly administering glutathione aerosol

Free radicals and the regulation of mammalian cell proliferation

Biokinetics of dietary RRR-a-tocopherol in the male guinea pig at three dietary levels of vitamin C and two levels of vitamin E: evidence that vitamin C does not " spare" vitamin E in vivo

The orthomolecular treatment of cancer

Normal alveolar epithelial lining fluid contains high levels of glutathione

Reperfusion injury following circulatory collapse: the role of reactive oxygen metabolites

Antiatherogenic effect of probucol unrelated to its hypocholesterolaemic effect:evidence that antio xidants in vivo can selectively inhibit LDL degradation in macrophage-rich fatty streaks and slow the progression of atherosclerosis in the Watanabe heritable hyperlipidemic rabbit

Comparative evaluat ion of the antio xidant properties of a-tocopherol, a-tocopherol polyethylene glycol 1000 succinate and a-tocopherol succinate in isolated hepatocytes and liver microsomal suspensions

A comparative study of the xanthine oxidase inhibitors allopurinol and oxypurinol in man

Free radicals and cardioplegia: allopurinol and oxypurinol reduce myocardial injury following ischemic arrest

Vitamin E supplement ation and periventricular hemorrhage in the newborn

Lower levels of vitamin C and carotenes in plasma of cigarette smokers

Oxygen-derived free radicals in the pathogenesis of parasitic disease

The anti-inflammatory effects of D-myo-inositol-I ,2,6-triphosphate (PP56) on animal models of inflammation

Free radicals in ischemic and reperfusion myocardial injury: is this the time for clinical trials?

Reduced chronic hemolysis during high-dose vitamin E administration in Mediterranean-type glucose-S-phosphate dehydrogenase deficiency

Studies on the anti-inflammatory activity of Ebselen

The adenosine /neutrophil paradox resolved: human neutrophils possess both AI and A2 receptors that promote chemotaxis and inhibit 02 -generation, respectively

Inhibitors of the leukocyte superoxide generating oxidase

Oxidative stress and abnormal cholesterol metabolism in patients with adult respiratory distress syndrome

Chronic granulomatous disease. Advances in

Lemasters 11. Evidence that Carolina rinse solution protects sinusoidal endothelial cells against reperfusion injury after cold ischemic storage of rat liver

Interaction between neutrophils and 4-hydroxyalkenals and consequences on neutrophil mobility. Free Radical Research

Detection of myoglobin-derived radicals on reaction of metmyoglobin with hydrogen peroxide and other peroxidic compounds

Ocular toxicity of high-dose intravenous desferrioxamine

Active oxygen in neuromuscular disorders

Free radical-mediated fragmentation of monoamine oxidase in the mitochondrial membrane: role of lipid radicals

Evidence of myocardial free radical injury during elective repair of tetralogy of Fallot

Fat-soluble vitamins, their biochemistry and applications

Iron chelation therapy

Sensitivity of parasites to free radical damage by antiparasitic drugs

Cytotoxic and cytoprotective activities of curcumin

Superoxide dismutase therapy for myocardial ischaemia

Reperfusion injury after thrombolyti c therapy for acute myocardial infarction

Vitamin E and retinopathy of prematurity: still controversial

Reduction in the diabetogenic effect of alloxan in mice by treatment with the antineoplastic agent ICRF-187

Vitamin E decreases superoxide anion production by polymorphonuclear leukocytes

Consequences of activation and adenosine -mediated inhibition of granulocytes during myocardial ischemia

Can superoxide dismutase alter myocardial infarct size?

Chemical phototoxicit y in humans

cytotoxic products oflipid peroxidation . lSI Atlas of Science

Continuous monito ring of in vitro oxidation of human low density lipoprotein

In vitro cytotoxicity of deferoxamine on human bone marrow haematopoietic progenitors

Oxygen toxicity: unique cytoprotecti ve properties of vitamin E succinate in hepatocytes

Hara TM. a-Tocopheryl succinate protects hepatocytes from chemical -induced toxicity under physiological calcium conditions

A novel biologically active selenoorganic compound VIII: biotransformation ofEbselen

Occurrence of oxidative stress during reperfusion of the human heart

A new mode of actio n for superoxide: inactivation of enzymes with [Fe-S] clusters. Free Rad ical Research Comm unicatio ns

Superoxide dism utase for therapeutic use: clinical experience, dead ends and hopes

Endothelial and myocard ial injury d uring ischemia and reperfusion: pathogenesis and therapeutic implications

Ascorbate is an outstanding antioxidant in human blood plasma

Superoxide radical, an endogenous toxica nt

Ischemia-reperfusion in humans. Appearance of xanthine oxidase activit

Glutathione protects cardiac and skeletal muscle from cyclophosphamide-ind uced injury

Free radical damage and organ preservation: fact or fiction?

Cooxidation of salicylate and choleste rol during the oxidatio n of metmyog lobin by H202

Th e antioxidant hypothesis of cardiovascular disease: epidemiology and mechanisms

Plasma levels of antioxidant vitamins in relatio n to ischaemic heart disease and cancer

Effects of repeated adminis tration of N-acetyl-L-cysteine on sulfhydryl levels of differen t tissues and bleomycin-ind uced lung fibrosis in hamsters

Protection against oxidative damage in cold-stored rabbi t kidneys by desferrioxarnine and indomethacin

Measurement by HPLC of desferrioxa rnine-available iron in rabbit kidneys to assess the effect of ischaemia on the distribution of iron within the total pool

The development of new iron-chelating drugs II

Antiox idant functions of phytic acid

Superoxide dismu tase and catalase as therapeutic

agents for human diseases: a critical review

Therape utic usages of oxygen radica l scavengers in human diseases: myths and realities

Adenosine inh ibits ischernia-reperfusion -induced leukocyte adherence and extravasation

Moorhouse CPo Action of uric acid, allopurino l and oxypurinol on the myeloperoxidase-derived oxidant hypochlorous acid

Iron promoters of the Fento n reaction and lipid peroxidation can be released from haemoglobin by peroxides

Bleomycin-detectable iron in knee-joint synovial fluid from art hritic patie nts and its relationship to the extracellular antioxidant activi ties of caeru loplasmin, transferrin and lactoferrin

Antioxida nt protection by haemopexin of haem-stimulated lipid peroxidation

Ceruloplasmin: physiological and pathological perspectives

Enhancement of bleomycin-iro n free radical damage to DNA by antiox idants and their inhibition of lipid peroxidat ion

Inhibition of the ironcatalyzed forma tion of hydroxyl radicals from superoxide and oflipid peroxidat ion by desferrioxarnine

Inhibition of lipid peroxidat ion by the iron-binding protein lactoferr in

low-molecular-weight iron comple xes and oxygen radical reactions in idiopathic haemochromatosis

Effect ofthiols on lipid peroxidation in rat liver microsomes

Central nervous system trauma and stroke

Modu lation of deferoxamine toxicity and clearance by covalent attach ment to biocompatible polymers

Oxidants and human disease: some new concepts

Albumin, an important extracellular antioxidant?

Superoxide , iron, vascular endot helium and reperfusion injury

Oxidants and the central nervous system: some funda mental questions

Free radicals, reactive oxygen species and human disease: a critical evaluat ion with special reference to atherosclerosis

Protect ion against tissue damage in vivo by desferrioxam ine: what is its mechanism of action ?

1990 laboratory and electrophysiological assessment

Increased survival of skin flaps by scavengers of superoxide radical

Oxidative stress and the redox status of malaria-infected erythrocytes

Assays of metabolic superoxide production in Escheri chia coli

A new vitamin E analogue more active than a-tocopherol in the rat curative myopathy bioassay

Further studies of a new vitamin E analogue more active than a-tocopherol in the rat curative myopathy bioassay

Inhibition of oxygen toxicity by targeting superoxide dismutase to endothelial cell surface

Influence of oxidative stress induced by cysteamine upon the induction and development of thermotolerance in Chinese Hamster ovary cells

Effect of allopurinol on ischemia and reperfusion-induced cerebral injury in spontaneously hypertensive rats

Novel21-aminosteroids that inhibit iron-dependent lipid peroxidation and protect against central nervous system trauma

Antiinflammatory activity of superoxide dismutases : inhibition of carrageenaninduced edema in rats. Free Radical Research Communications I

Reduced levels of plasma ascorbic acid (vitam in C) in sickle cell disease patients: its possible role in the oxidant damage to sickle cells in vivo

a-Tocopherol, an effective inhibitor of platelet adhesion

Protection of cardiac memo brane phospholipid against oxidative injury by calcium antagonists

Newer trends in therapeutic chelating agents

New developments in therapeutic chelating agents as antidotes for metal poisoning

Therapeutic chelating agents

Actions of adenosine on nitroblue tetrazolium deposition and surface pH during intestinal reperfusion injury

Initiation of membranal lipid peroxidation by activated metmyoglobin and methaemoglobin

Singlet oxygen production by biological systems

A survey of chemicals inducing lipid peroxidation in biological systems

Concepts related to the study of reactive oxygen and cardiac reperfusion injury

The study of iron mobilisat ion from transferrin using a-ketohydroxy heteroaromatic chelators

Iron mobilisation from lactoferrin by chelators at physiological pH

In vitro screening of iron chelators using models of free radical damage

Action of Ebselen on rat hepatic microsomal enzyme-catalyzed fatty acid chain elongation, desaturation and drug biotransformation

The antioxidants of higher plants

Allopurinol enhanced adenine nucleotide repletion after myocardial ischemia in the isolated rat heart

Antioxidant and pro-oxidant actions of the plant phenolics quercetin, gossypol and myricetin

Inhibition of superoxide anion radical production by Ebselen (PZ5I) and its sulfur analogue (PZ25) in guinea pig alveolar macrophages

A global view of human selenium nutrition

New concepts in the biology and biochemistry of ascorbic acid

Anti-inflammatory drugs from plant and marine sources

The determination of dehydroascorbic acid and ascorbic acid in the serum and synovial fluid of patients with rheumatoid arthritis

The B lymphocyte: a newly recognized source of reactive-oxygen species with immunoregulatory potential

Expression of extracellular superoxide dismutase by human cell lines

Nitric oxide: biosynthesis and biological significance

Photo protection by carotenoids

Clinical experience with desferrioxamine in dialysis patients with aluminium toxicity

Vitamin E: interactions with free radicals and ascorbate

Oxygen-derived free radicals in post ischemic tissue injury

Intramuscular desferrioxamine in patients with Alzheimer's disease

Iron overload disorders: natural history, pathogenesis, diagnosis and therapy

Human fibroblasts release reactive oxygen species in response to treatment with synovial fluids from patients suffering from arthritis

Glutathione metabolism and its selective modification

A comparative study of free radical scavengers in cardioplegic solutions

Deferoxamine reduces neutrophil-mediated free radical production during cardiopulmonary bypass in man

Oxygen free radicals, inflammation and synovitis: the current status

Mechanism of action of paracetamol protective agents in mice in vivo

The role of iron in the initiation of lipid peroxidation

Biologically active metal-independent superoxide dismutase mimics

Macrophages and atherogenesis

Survival and desferrioxamine in thalassaemia major

Lung protection by a thiolcontaining antioxidant: N-acetylcysteine

Selectivity of neutrophil 5-lipoxygenase and cyclooxygenase inhibition by an anti-inflammatory flavonoid glycoside and related aglycone l1avonoids

Synthesis and stopped-flow investigation of antioxidant activity of tocopherols. Finding of new tocopherol derivatives having the highest antioxidant activity among phenolic antioxidants

Free radicals and Dupuytren's contracture

Modulation of fibroblast proliferation by oxygen free radicals

Vasorelaxant properties of the endothelium-derived relaxing factor more closely resemble 5-nitrosocysteine than nitric oxide

Superoxide dismutase mimics based on iron in vivo

Introduction: oral iron chelators

Vitamin C and iron

Interaction of milk xanthine oxidase with folic acid

Molecular basis of experimental diabetes: degeneration, oncogenesisand regeneration of pancreatic ,B-eells of islets of Langerhans

Role of Ca 2 + in toxic cell killing

Effect of deprenyl on the progression of disability in early Parkinson's disease

Oxygen radicals: mediators of gastrointestinal pathophysiology

The pharmacology of ebselen

Probucol inhibits oxidati ve modification of low densit y lipoprotein

Effect of human recombinant superoxide dismutase on canine myocardial infarction

Gradual reintroduction of oxygen reduces reperfusion injury in cat stomach

Dangers of thrombolysis

Autotoxicity and reactive oxygen in viral disease

Reduction of myocardial reperfusion injury by intravenous adenosine administered during the early reperfusion period

Desferrioxamine suppresses Plasmodium falciparum in Astus monkeys

L-2-oxothiazolidine-4-carboxylic acid, a cysteine prodrug: pharmacokinetics and effects on thiols in plasma and lymphocytes in human

Animal toxicology of iron chelator LI

Desferrioxamine ototoxicity : evaluation of risk factors in thalassaemic patients and guidelines for safe dosage

Relative oral efficacy and acute toxicity of hydroxypyrid-4-one iron chelators in mice

Free radical formation by antitumor quinones

Formation of hydroxyl radicals from hydrogen peroxide in the presence of iron: is haemoglobin a biological Fenton reagent?

Formation of hydroxyl radical in biological systems. Does myoglobin stimulate hydroxyl radical formation from hydrogen peroxide?

Scavenging of hypochlorous acid and of myoglobin-derived oxidants by the cardioprotective agent mercaptopropionylglycine

Inhibition of 5-lipoxygenase by vitamin E

On the metabolism of allopurinol

Protection by deferoxamine from endothelial injury: a possible link with inhibition of intracellular xanthine oxidase

Glutathione, free radicals and chemotherapeutic agents

Damage to proteins, enzymes and amino 603 acids by peroxidizing lipids

Superoxide-dependent formation of hydroxyl radicals in the presence ofthiol compounds

The production offree radicals during the autoxidation of cysteine and their effect on isolated rat hepatocytes

Oxygen-derived free radicals and the prevention of duodenal ulcer relapse: a new approach

The effect of free radical scavengers on platelet adhesion and aggregation

Protection by desferrioxamine against histopathological changes of the liver in the post-oligaemic phase of clinical haemorrhagic shock in dogs: correlation with improved survival rate and recovery

The role of oxygen-derived free radicals in the pathogenesis of acute pancreatitis

Reaction of NO with superoxide

Joosten LAB, van den Bersselaar L. Cationization of catalase, peroxidase and superoxide dismutase

First clinical experiences with superoxide dismutase in kidney transplantation -results of a double-blind randomized study

Prophylactic and curative effects of recombinant human superoxide dismutase in lethal rat endotoxemia

Thiyl radical attack on polyunsaturated fatty acids: a possible route to lipid peroxidation

Oxidant injury of cells

The cytotoxic and mutagenic properties of cholesterol oxidation products

ESR investigations of the reactions of radiation-produced thiyl and DNA peroxyl radicals: formation of sulfoxyl radicals

Genetic toxicology of ascorbic acid

In vivo and in vitro activity by diverse chelators against Trypanosoma brucei brucei

The phototoxic effect of benoxaprofen and its analogs on human erythrocytes and rat peritoneal mast cells

Metabolic activation of natural phenols into selective oxidative burst agonists by activated human neutrophils

Free radicals and myocardial ischaemia and reperfusion injury

Vitamin E supplementation reduces frequency of periventricular haemorrhage in very preterm babies

Efficacy of oral N-acetylcysteine in the treatment of acetaminophen overdose

Cholesterol autoxidation 1981-1986

Bis(3,5-diisopropylsalicylato)copper(II), a potent radioprotectant with superoxide dismutase mimetic activity

Oxypurinol as an inhibitor of xanthine oxidase-cata-Iyzed production of superoxide radical

Inhibition of xanthine oxidase by 4-hydroxy-6-mercaptopyra-zoI0

Protective effect of the bispiperazinedione ICRF-187 against doxorubicin-induced card iac toxicity in women with advanced breast cancer

Ii-carotene and cancer prevent ion: the Basel study

Ferric iron and superoxide ions are required for the killing of cultured hepatocytes by hydrogen peroxide

Beyond cholesterol: modifications of low-density lipoprotein that increase its atherogenicity

A trial of the 21-aminosteroid U74006F in a primate model of chronic cerebral vasospasm

Cytotoxicity of a low molecular weight CuzZnz superoxide dismutase active center analog in human erythroleukemia cells

Drug trial of superoxide dismutase in Duchenne's muscular dystrophy

Induction of haem oxygenase as a defence against oxidative stress

Expression of eloned human lactoferrin in baby-hamster kidney cells

Superoxide med iates reperfusion-induced leukocyteendothelial cell interactions

Determination of reduced, oxid ized, and protein-bound glutathione in human plasma with precolumn derivatization with monobromobimane and liquid chromatography

Anti-inflammatory activity of antioxidants

Inhibition by desferrioxamine of in-vitro replication of HIV-I

Inhibition of nitrosamine formation by ascorbic acid

Oxidation and reaction of Trolox C

tocopherol analogue, in aqueous solution: a pulse-radiolysis stud

Esterification of reduced glutathione

Antioxidants can prevent cerebral malaria in Plasmodium berghei-infected mice

Expression of human extracellular superoxide dismutase in chinese hamster ovary cells and characterization of the product

ICRF-187 and polyhydroxyphenyl derivatives fail to protect against bleomycin induced lung injury

Evidence of rebound effect with ascorbic acid

Time course and mechanism of endothelial dysfunction in isolated ischemic-and hypoxic-perfused rat hearts

Plasma homocysteine, a risk factor for vascular disease : plasma levels in health , disease and drug therapy

Anti-inflammatory activity of various superox ide dismutases on polyarthritis in the Lewis rat

Mechanism of protection of alveolar type II cells against paraquat-induced cytotoxicity by deferoxam ine

Anti-inflammatory and antiulcer properties of hypolactin-8-g1ucoside, a novel plant flavonoid

Ischemic inj ury to newborn rabbit ileum : protective role of human superoxide dismutase

The chemical basis of radiation biology

Distribution of xanthine dehydrogenase and oxidase activities in human and rabbit tissues

Niacin attenuates bleomycin-induced lung fibrosis in the hamster

Deferoxamine infusion does not inhibit bleomycin-induced lung damage in the rat

Evidence of role of hydroxyl radical in complement and neutrophil-mediated tissue injury

Lipid peroxidation in relation to prostacyclin and thromboxane physiology and pathophysiology

The antioxidant action of extracellular fluids

Singlet oxygen in biological systems

Oxygen radical scavengers improve vascular potency and bonemuscle cell survival in an ischemic extremity replant model

Tissue destru ction by neutrophils

Effect of nicotin ic acid supplementat ion in vivo on oxygen radical-induced genetic damage in human lymphocytes

Free radicals and ischemic tissue injury

Structure of CU2 (indomethacin), and the reaction with superox ide in aprotic systems

Inhibit ion of oxidation enzymes by anti -allergy drugs

Seminars in Hematolog y

Glucose autox idation and protein modification

Stimulated arach ido nate metabolism during foam cell transformation of mouse peritoneal macro phages with oxidized low densit y lipoprot ein

The role of metal ions in ischemia jreperfusion injury in skin flaps

For further information

The author thanks Dr JMC Gutteridge for his helpful comments. Research funding from the British Heart Foundation and the Arthritis and Rheumatism Council (UK) is gratefully acknowledged.