Medical Biology 62:71-77, 1984 Oxygen Radicals: A Commonsense Look at Their Nature and Medical Importance B. Halliwell From the Department of Biochemistry, University of London King's College, London, U.K. Introduction "Oxygen radicals" are now popular subjects for research papers; several hundred are published each year. Many of these pass rapidly into oblivion, joining the great mass of unread scientific literature that clogs library shelves and dilutes important research findings to an increasingly great extent. The basic chemistry of oxygen-derived species was established years ago by radiation chemists (1,6), but "superoxide" is still endowed with miraculous properties by the uninitiated. Demonstration that the action of a disease or toxin in vivo produces increased lipid peroxidation (a currently-popular scientific activity) means nothing more than the fact that its action produces increased lipid peroxidation: it does not automatically follow that the lipid peroxidation causes the damaging effects of the drug or disease. The purpose of this paper is to explain: i) what oxygen radicals are ii) the evidence that oxygen radicals are important in vivo iii) what needs to be done to establish a role for oxygen radicals and lipid peroxidation in human disease. What are the oxygen radicals and how are they produced? Electrons within atoms and molecules occupy regions of space known as "orbitals". Each orbital can hold a maximum of two electrons. A single electron alone in an orbital is said to be "unpaired" and a radical is defined as any species that contains one or more unpaired electrons. Such a definition embraces the atom of hydrogen (one unpaired electron) and the ions of such transition metals as iron, copper and manganese (cf. Holmberg, this volume). The diatomic oxygen molecule, O2, has two unpaired electrons and thus qualifies as a radical. Most of the oxygen taken up by human cells is reduced to water by the action of the cytochrome oxidase complex in mitochondria. This requires the addition of four electrons to each oxygen molecule, O2 + 4H+ + 4e- ---> 2H2O (1) For chemical reasons (reviewed in ref. 21 and 28), O2 likes to receive its electrons one at a time, producing a series of partially reduced intermediates O2 add le- O2- add le- H2O2 add le- ---> ---> ---> 2H (two unpaired superoxide hydrogen peroxide electrons) (one unpaired (no unpaired electron) electron) OH OH- hydroxyl radical hydroxyl ion (2) (one unpaired electron) (no unpaired electron) | | | | | add le- H+ | add H+ H2O H2O Cytochrome oxidase keeps the partially reduced intermediates on the pathway to water tightly bound to its active site (21); they do not escape into free solution. Superoxide Superoxide ion is the one-electron reduction product of oxygen. Dissolved in organic solvents, it is an extremely reactive species, e.g. it can displace chlorine from such unreactive chlorinated hydrocarbons as carbon tetrachloride (CCl4) (40). In aqueous solution O2- is poorly reactive, acting as a reducing agent (e.g. it will reduce cytochrome c or nitro-blue tetrazolium) and slowly undergoing the dismutation reaction, in which one molecule of superoxide reduces another one to form hydrogen peroxide (H2O2 ). The dismutation reaction occurs in stages; O2- must first combine with a proton to yield the hydroperoxyl radical, HO2, O2- + H+ ---> HO2 (3) HO2 + O2 + H+ ---> H2O2 + O2 (4) -------------------------------------------------------- overall O2- + O2- + 2H+ ---> H2O2 + O2 (5) At physiological pH the low concentration of H+ ions slows the rate of dismutation. Despite the low reactivity of O2- in aqueous solution, systems producing it do a great deal of damage in vitro (e.g. they fragment DNA and polysaccharides, kill bacteria and animal cells in culture) and in vivo (e.g. when O2- generating systems are injected into the footpads of rats inflammation is produced, their instillation into the lungs of rats and rabbits produces oedema and cell death, and infusion of them into vascular beds produces endothelial cell damage and extensive leakage from the blood vessels) (21,26,28). Depending on the circumstances, damage caused by O2- generating systems might be attributed to (i) O2- itself, e.g. exposure of tissue fluids to O2- causes formation of a factor chemotactic for neutrophils that brings more of them into the area and hence can potentiate inflammation (ii) HO2 radical, which is more reactive than O2- (6). Formation of HO2 is favoured at pH values lower than "physiological", but the phagocytic vacuole operates at an acid pH and the pericellular pH of macrophages has been reported to be 6 or less (15) (iii) H2O2 (see below) (iv) hydroxyl radical (see below) (v) singlet oxygen. Singlet O2 is an especially reactive form of oxygen capable of rapidly oxidising many molecules, including membrane lipids. Its formation in O2-- generating systems has often been proposed but clear-cut evidence for a damaging role of singlet O2 in such systems has not been obtained. One of the problems is that the "scavengers" of singlet O2 frequently used react with other radical species as well (for reviews see ref. 26 and 28). What is the evidence that O2- is formed in vivo in human cells? Any electron transport chain operating in the presence of O2 "leaks" some of the electrons, passing them directly onto O2. Since O2 prefers to take electrons one at a time, O2- is produced. Such O2- production can be demonstrated in vitro using mitochondria and microsomes from a range of animal tissues. The rate at which O2- is produced rises as the concentration of O2 in the system is raised (e.g. see ref. 20). A number of compounds slowly become oxidised on exposure to O2 and O2- is generated; these include adrenalin, tetrahydrofolate, reduced FMN and oxyhaemoglobin (21,24). Since human cells contain mitochondria, endoplasmic reticulum, oxidisable compounds and oxygen, it is likely that O2- is formed within them in vivo. Backing up this evidence, for those who do not like extrapolating from in vitro experiments, is the fact that human cells contain high levels of superoxide dismutase (SOD) activity (45). This enzyme, for which O2- is the specific substrate (35), is known to be a very important anti-oxidant in bacteria and small mammals (26) and its presence in human cells is good evidence that O2- is formed in vivo. During the maturation of erythrocytes most enzymes are lost, but SOD remains. It is not a great stretch of the imagination to associate this with the ability of oxyhaemoglobin to release O2- radical and methaemoglobin. Another source of O2- in vivo is the respiratory burst of phagocytic cells such as neutrophils, monocytes, eosinophils and macrophages (3, 16, 25). The amount of O2- produced might sometimes be controlled by the O2 tension of body fluids (14). Host defence against invading bacteria is dependent on the circulating neutrophils, which respond to contact with particles they recognise as foreign by producing a "burst" of O2 radical. The particle is engulfed (the piece of membrane surrounding it being the segment that produces O2- on contact; cf. Segal, this volume), and other vesicles then fuse with the phagocytic vesicle. This exposes the engulfed particle to other anti- bacterial mechanisms, including cationic proteins, lysosomal enzymes and myeloperoxidase (3, 16, 25). Which of these processes is the most important in bacterial killing? Human and other animal neutrophils can kill some strains of bacteria under anaerobic conditions, when O2- cannot form. Obviously, the other mechanisms are important here. Many other bacterial strains are not killed in the absence of O2, however, even though engulfment and vesicle fusion proceed normally. In chronic granulomatous disease (CGD), an inborn error of metabolism, the respiratory burst does not occur but other aspects of phagocytic action proceed normally. CGD was first described in humans because it is accompanied by severe and recurrent infections affecting lymph nodes, skin, lungs and liver (43). The symptoms of CGD provide direct evidence for the production of O2- by human phagocytic cells in vivo and for its role in bacterial killing. It follows therefore that if neutrophils become activated in the wrong place, or to excessive extents (as in the autoimmune diseases, 25) then the oxygen radicals they release could do a lot of damage. It must be remembered, however, that phagocytic cells also produce hydrolytic enzymes (elastase, neutral proteases etc.), chemotactic factors, prostaglandins, leukotrienes and other chemicals, so that damage by activated phagocytes could be due to any one of these factors or to any combination of them. It cannot be attributed a priori to oxygen radicals. Hydrogen Peroxide O2- generating systems produce H2O2 by the dismutation reaction (eqn. 5) and a number of oxidase enzymes produce H2O2 directly, examples being glycollate oxidase and amino acid oxidases. SOD enzymes remove O2- by greatly accelerating the dismutation reaction, so if we accept that O2- is formed in vivo in humans then we must accept that H2O2 vapour is present in expired human breath (48), a likely source being H2O2 released from alveolar macrophages (3, 25) although a contribution from peroxide- producing oral bacteria (10) cannot be ruled out. That H2O2 is formed in vivo in humans is further supported by the presence of enzymes specific for its removal, such as catalase and glutathione peroxidase. The latter enzyme requires selenium for its activity (13; cf. Diplock, this volume). H2O2 is probably more damaging than is O2- in in vitro experiments in aqueous solution, but many cells seem to tolerate its presence and bacteria often produce H2O2 (e.g. ref. 10). On the other hand, the toxicity of O2- generating systems to several animal cells in culture has been attributed to formation of H2O2 (e.g. ref. 44). Why this should be so is discussed in the next section. Hydroxyl radical Hydroxyl radical is produced when water is exposed to high-energy ionising radiation and hence its properties have been well documented by radiation chemists (6, 49). Unlike the hydroxyl ion, the hydroxyl radical is fearsomely reactive, combining with most molecules found in vivo at near diffusion-controlled rates. Hence any OH produced in vivo will react at or close to its site of formation. The extent of the damage done would therefore depend on what the site of formation was (e.g. production of OH close to DNA could lead to strand breakage whereas production close to an enzyme molecule already present in excess in the cell, such as lactate dehydrogenase, might have no biological consequences). Hydroxyl radical is produced whenever H2O2 comes into contact with copper (I) ions (Cu+) or iron (II) ions (Fe2+). Dr. Gutteridge has reviewed in this volume the substantial evidence that metal complexes capable of causing hydroxyl radical formation are present in vivo in human cells (also see ref. 28). Particularly important in vivo are complexes of iron salts with phosphate esters such as ATP and GTP (17, 19) or with DNA (18). Organisms take great care to ensure that as much iron or copper as possible is bound to transport proteins or functional proteins such as transferrin, caeruloplasmin or haemoglobin. Metals bound to these proteins are inactive or only weakly active in catalysing OH production (28, 50). Since both H2O2 and metal complexes are present in vivo in humans, it is logical to assume that OH radicals can form. Direct evidence for this is difficult to obtain. Many methods exist for demonstrating the existence of OH in vitro (see ref. 24 and 28 for reviews) but in vivo any OH formed is likely to react so close to its site of formation that the use of these methods is impractical, although some new techniques (such as the ability of OH to convert dimethylsulphoxide into methane (36) or its ability to hydroxylate aromatic rings in characteristic ways (37) show promise for in vivo use. One can also attempt to infer the formation of OH radical in vivo by observing the damage done (as in rheumatoid arthritis, see below). In vitro, phagocytic cells have been shown to produce OH radical (11-13) and the killing of bacteria can sometimes be prevented by reagents that react with this species (3, 16, 25). It was mentioned in the previous section that the killing of animal cells in culture by O2- generating systems can sometimes be attributed to H2O2. It could, of course, be achieved by H2O2 itself; some enzymes are known to be inactivated by H2O2 although the best examples come from plant rather than animal systems (11). There is another possibility, however, H2O2 generated externally crosses cell membranes easily and could penetrate inside the cell and cause OH to be formed. Externally added scavengers of OH would not prevent this since they could not reach the correct place. By contrast, O2- crosses cell membranes only slowly (42) unless there is a specific channel for it (the only known example of this being the erythrocyte membrane, which has an "anion channel" through which O2- can move(3). Hydroxyl radical will never cross a membrane: it will react with whatever membrane component if meets first. What is lipid peroxidation and is it of medical importance? Lipid peroxidation has been broadly defined by A. L. Tappel in the USA as "oxidative deterioration of polyunsaturated fatty acids", i.e. fatty acids that contain more than two carbon-carbon double bonds. Oxygen-dependent deterioration, leading to rancidity, has been long recognised as a problem in the storage of fats and oils and is even more relevant today with the popularity of "polyunsaturated" food products. Some of the best studies on peroxidation chemistry have been carried out by food chemists. Initiation of peroxidation in a membrane or polyunsaturated fatty acid is due to the attack of any species that can "pull off" a hydrogen atom from one of the - CH2 - groups in the carbon chain. Hydroxyl radical and possibly HO2 can do this, but H2O2 and O2- cannot. Hence O2- does not initiate lipid peroxidation. Since a hydrogen atom has only one electron, removing it leaves behind an unpaired electron on the carbon. The resulting carbon radical - CH -, undergoes molecular rearrangement to form a conjugated diene, which then combines rapidly with O2 to give a O2 | peroxy radical, - CH -. Peroxy radicals are capable of abstracting a hydrogen atom from other fatty acids and so setting off a chain reaction that can continue until the membrane fatty acids are completely oxidised to hydroperoxides (eqn. 6) O2 | - CH - + - CH2 - ---> peroxy adjacent fatty acid radical carbon chain O2H | - CH - + - CH - (6) carbon radical, lipid forms another hydroperoxide peroxy radical Lipid hydroperoxides are stable under physiological conditions until they come into contact with transition metals such as iron or copper salts. Cu2+, Fe2+ or Fe3+ salts as well as haem and haem proteins (e.g. cytochromes, haemoglobin) can interact with lipid peroxides. These metals or their complexes cause lipid hydroperoxides to decompose in very complicated ways, producing radicals that can continue the chain reaction of lipid peroxidation (as in eqn. 6), as well as cytotoxic aldehydes and hydrocarbon gases. Most attention is paid in the literature to malonaldehyde, but this is a very minor endproduct of lipid peroxidation (for reviews see ref. 4, 26, 32). Does lipid peroxidation occur normally in vivo in humans? This question is surprisingly difficult to answer: little evidence for lipid peroxides or their decomposition products can be found in healthy human tissues (28). Expired human breath contains gaseous hydrocarbons that might have originated from decomposition of lipid hydroperoxides, but they might also have been produced by bacteria in the gut or even on the skin. Animal cell membranes contain tocopherol (vitamin E), which is a powerful inhibitor of lipid peroxidation, and proteins such as caeruloplasmin and glutathione peroxidase probably help to protect against this process in vivo (27). Diseased tissues, or tissues isolated after exposure of animals to such toxins as ethanol, phenylhydrazine and paraquat often show evidence of increased peroxidation. Simple in vitro experiments demonstrate quite clearly that dead or damaged tissues peroxidise more rapidly than living ones, presumably because of membrane disruption by enzymes released from lysosomes, release of metal ions from their storage sites and failure of antioxidant mechanisms. Thus evidence that a toxin increases lipid peroxidation in vivo does not prove the sequence of events toxin ---> lipid peroxidation ---> damage (7) but is equally explained by the sequence toxin ---> cell damage or death ---> lipid peroxidation (8) Of course, toxins released by dead or dying cells undergoing peroxidation might cause further damage to healthy cells, although there is little evidence for this in vivo. Among the many claims I have seen in the literature for lipid peroxidation as an agent of the damage induced by a toxin, I have seen clear evidence for sequence 7 only in the case of the hepatotoxic effects of carbon tetrachloride (32). Sequence 8 is a much better explanation of the in vivo effects on membrane lipids of, for example, paraquat. An often quoted illustration of the importance of lipid peroxidation in vivo is the accumulation of "age pigment" in various human tissues. Chemical analysis of age pigment shows convincingly that it is an endproduct of oxidative damage to lipids (41). However, the lipids in question seem to be taken into lysosomes before they are degraded; they are not "normal cell lipids". The exposure of lipids to hydrolytic enzymes and metal ions within lysosomes no doubt facilitates their peroxidation, and so more peroxidised material accumulates within cells as lysosomes get older and have engulfed more lipid material. The TBA test The TBA (thiobarbituric acid) test is one of the most widely used (and abused!) tests for measuring lipid peroxidation. The simplicity of performing the test (the material under study is merely heated with acid and TBA and the formation of a pink colour measured at 523 nm) conceals its essential complexity. Consider a typical experiment. A lipid system, perhaps with added metal ions, chelating agents or other reagents, in incubated in the presence of air. Then TBA plus acid are added and the mixture heated at 100 degrees Celcius. The air, metals and other reagents are still present, so as much or even more oxidative damage to the lipid can be done during the TBA test itself as happened during the initial incubation. The pink colour is due to the formation of an adduct between TBA and malonaldehyde (MDA) under acidic conditions. Indeed, the TBA assay is often calibrated with MDA and the results of peroxidation assays are often expressed as "amounts of MDA formed". Some papers in the literature give the mistaken impression that TBA reacts only with free MDA and so measures the production, but it was shown as long ago at 1958 in studies with peroxidising fish oil that 98 % of the MDA that reacts in the TBA test was not present in the original sample assayed but forms from lipid peroxides that decomposed during the acid-heating stage of the TBA assay. More recent studies confirm this and show that the apparent "TBA reactivity" of say, serum, varies with the exact concentration of acid, type of acid and period of heating used in the TBA assay (23). The amount of MDA formed during the initial incubation of the system as opposed to during the assay depends on such factors as the iron salt concentration (4, 23, 32). An apparent "inhibitor" of lipid peroxidation as detected by the TBA test might actually inhibit the peroxidation process, but could equally well interfere with decomposition of the peroxides during the acid-heating stage of the assay. Similarly, absolute values for the "TBA reactivity" of body fluids or tissue extracts are meaningless, although changes in these values may be significant provided that the same assay in employed in the same way each time. Of course, many scientists are aware of these problems with the TBA assay and there are ways around them (2, 41), including the use of other assay systems in conjunction with the TBA test (4, 27). I have included these cautions to encourage a more critical attitude to some of the published literature. Oxygen Radicals and Disease Free radicals have been suggested to be involved in the pathology of a number of diseases. In several cases the evidence consists only of observations of increased lipid peroxidation in diseased tissues, which is ambiguous (see above). I have chose to look in detail at two cases where the evidence at first sight is more convincing, cancer and inflammatory joint disease. Cancer Any substance that reacts with DNA is potentially carcinogenic. Exposure of DNA to O2- generating systems causes extensive strand breakage and degradation of deoxyribose (9, 39), an effect shown in vitro to be due to formation of OH. Both bacteria and animal cells in culture suffer DNA damage on exposure to O2- generating systems, which can be shown to be mutagenic (46, 47). It is therefore tempting to attribute the increased risk of development of cancer in chronically inflamed tissues to generation of oxygen radicals by phagocytic cells, although there is no direct evidence for this. Great excitement was generated by reports that cancer cells in culture and from some transplantable tumours in animals are deficient in SOD activity, especially in their mitochondria (for a review see ref. 34). The relevance of these studies to human cancer is not at all clear, however, since human tumours biopsied during surgery show no defects in any SOD activity (31, 45). Rheumatoid arthritis I have already speculated on the role of oxygen radicals in the autoimmune diseases. Rheumatoid arthritis has some of the features of an autoimmune disease but its exact cause is unknown. The synovial fluid of the inflamed joint swarms with neutrophils. Since the fluid contains increased concentrations of products that activated neutrophils release (including lactoferrin, 5) and end-products of arachidonic acid metabolism), then at least some of these neutrophils must be activated and thus producing superoxide, and hence H2O2 in vivo. Human synovial fluid is poor in SOD, catalase and glutathione peroxidase activities (8) but does contain iron complexes capable of catalyzing a reaction between O2- and H2O2 to form OH (38). There is as yet no direct proof that OH is formed in vivo, but evidence consistent with its formation includes the observation that the hyaluronic acid in synovial fluid is degraded in rheumatoid joints, and the type of degradation observed can be reproduced by exposing pure hyaluronic acid in vitro to OH radical (22). TBA-reactive material is also present in serum and synovial fluid of rheumatoid patients. There are significant correlations (38) between the content of TBA-reactive material in synovial fluid, its content of catalytic iron complexes and both clinical ("knee score") and laboratory ("white cell count" and "fluid content of C-reactive protein") assessments of disease activity. Thus there is certainly evidence for oxygen radicals being produced in the rheumatoid joint and having some deleterious effects. The question to be answered in how important are oxygen radicals in relation to other agents of damage. The pathology of rheumatoid arthritis is very complex and the number of potentially damaging agents, including hydrolytic enzymes, prostaglandins and leukotrienes, is enormous (29). Some scientist have tried to assess the importance of oxygen radicals by examining the effects of injecting SOD directly into inflamed joints (33; see Marklund, this volume), whereas our group, reasoning that iron complexes are required for O2- dependent formation of highly reactive OH radical, is examining the effect of iron-chelating drugs that can prevent OH formation (such as desferrioxamine, 12) on animal models of acute and chronic inflammation (7). REFERENCES 1. Anbar M, Neta P: A compilation of specific bimolecular rate constants for the reactions of hydrated electrons, hydrogen atoms and hydroxyl radicals with inorganic and organic compounds in aqueous solution. Int J Appl Radiat Isot 18: 495-523, 1967 2. Aust SD, Svingen BA: The role of iron in enzymatic lipid peroxidation. In: Free radicals in biology, vol. 5. ED. W. A. Pryor. Academic Press, New York 1982 3. Babior BM: Oxygen-dependent microbial killing by phagocytes. N Engl J Med 298: 721-725, 1978 4. Barber AA, Berheim F: Lipid peroxidation: its measurement, occurrence and significance. Adv Gerontol Res 2: 355-403, 1967 5. Bennett RM, Eddie-Quartey AC, Holt PJL: Lactoferrin: an iron binding protein in synovial fluid. Arthritis Rheum 16: 186- 190, 1973 6. Bielski BHJ, Shive GG: Reaction rates of superoxide radicals with the essential amino acids. In: Oxygen free radicals and tissue damage. Ciba Found Symp 65 (new series). Excerpta Medica, Amsterdam 1979 7. Blake DR, Hall ND, Bacon PA, Dieppe PA, Halliwell B, Gutteridge JMC: Effect of a specific iron chelating agent on animal models of inflammation. Ann Rheum Dis 42: 89-93, 1983 8. Blake DR, Hall ND, Treby DA, Halliwell B, Gutteridge JMC: Protection against superoxide and hydrogen peroxide in synovial fluid from rheumatoid patients. Clin Sci 61: 483- 486, 1981 9. Brawn K, Fridovich I: DNA strand scission by enzymically generated oxygen radicals. Arch Biochem Biophys 206: 414- 419, 1981 10. Carlsson J, Iwami Y, Yamada T: H2O2 excretion by oral streptococci and effect of lactoperoxidase-thiocyanate-H2O2. Infect Immun 40: 70-80, 1983 11. Charles SA, Halliwell B: Effect of H2O2 on spinach chloroplast fructose bis phosphatase. Biochem J 189: 373- 376, 1980 12. Gutteridge JMC, Richmond R, Halliwell B: Inhibition of the iron-catalysed formation of hydroxyl radicals from superoxide and of lipid peroxidation by desferrioxamine. Biochem J 184: 469-472, 1979 13. Diplock AT: Metabolic and functional defects in selenium deficiency. Phil Trans R Soc Lond B 294: 105-117, 1981 14. Edwards SW, Hallet MD, Lloyd D, Campbell AK: Decrease in apparent Km for oxygen after stimulation of respiration of rat polymorphonuclear leucocytes. FEBS Lett 161: 60-64, 1983 15. Etherington DJ, Silver IA, Gibbons R: An in vitro model for the study of collagen degradation during acute inflammation. Life Sci 25: 1885-1892, 1979 16. Fantone JC, Ward PA: Role of oxygen-derived free radicals and metabolites in leukocyte-dependent inflammatory reactions. Am J Pathol 107: 397-413, 1982 17. Flitter W, Rowley DA, Halliwell B: Superoxide-dependent formation of hydroxyl radicals in the presence of iron salts. What is the physiological iron chelator? FEBS Lett 158: 310- 312, 1983 18. Floyd RA: DNA-ferrous iron catalysed hydroxyl free radical formation from hydrogen peroxide. Biochem Biophys Res Commun 99: 1209-1215, 1981 19. Floyd RA: Direct demonstration that ferrous ion complexes of di- and triphosphate nucleotides catalyse hydroxyl free radical formation from hydrogen peroxide. Arch Biochem Biophys 225: 263-270, 1983 20. Freeman, BA, Crapo JD: Hyperoxia increases oxygen radical production in rat lungs and lung mitochond-ria. J Biol Chem 256: 10986-10992, 1981 21. Fridovich I: Superoxide radical, an endogeneous toxicant. Ann Rev Pharmacol Toxicol 23: 239-257, 1983 22. Greenwald AA, Moy WW: Effect of oxygen-derived free radicals on hyaluronic acid. Arthritis Rheum 23: 455-463, 1980 23. Gutteridge JMC: Free radical damage to lipids, amino acids, carbohydrates and nucleic acids as determined by thiobarbituric acid reactivity. Int J Biochem 14: 649-653, 1982 24. Halliwell B: Free radicals, oxygen toxicity and aging. In: Age pigments. Ed. R. S. Sohal. Elsevier, Amsterdam 1981 25. Halliwell B: Production of superoxide, hydrogen peroxide and hydroxyl radicals by phagocytic cells: a cause of chronic inflammatory disease? Cell Biol Int Rep 6: 529-542, 1982 26. Halliwell B: Superoxide and the superoxide theory of oxygen toxicity: a critical appraisal. In: Copper proteins, vol. 2. Ed. R. Lontie. CRC Press, Florida 1984 27. Halliwell B, Gutteridge JMC: Free radicals in biology and medicine. Oxford University Press, Oxford 1984 28. Halliwell B, Gutteridge JMC: Oxygen toxicity, oxygen radicals, transition metals and disease. Biochem J 219: 1- 14, 1984 29. Henderson B: The biochemistry of the human synovial lining with special reference to alterations in metabolism in rheumatoid arthritis. Path Res Pract 172: 1-24, 1981 30. Lynch RE, Fridovich I: Permeation of the erythrocyte stroma by superoxide radical. J Biol Chem 253: 4697-4699, 1978 31. Marklund SL, Westman NG, Lundgren E, Roos G: Copper and zinc- containing SOD, manganese-containing SOD, catalase and glutathione peroxidase in normal and neoplastic human cell lines and normal human tissues. Cancer Res 42: 1955-1961, 1982 32. McBrien DCH, Slater TF (eds): Free radicals, lipid peroxidation and cancer. Academic Press, London 1982 33. Michelson AM, Puget K: Aspects medicaux des superoxyde dismutases. CR Seances Soc Biol 173: 380-394, 1979 34. Oberley LW: Superoxide dismutase and cancer. In: Superoxide dismutase, vol. 2. Ed. L. W. Oberley. CRC Press, Florida 1982 35. Pain RH: Dressing the SOD. Nature 306: 228, 1983 36. Repine JE, Eaton JW, Anders MW, Hoidal JR, Fox RB: Generation of hydroxyl radical by enzymes, chemicals and human phagocytes in vitro. J Clin Invest 64: 1642-1651, 1979 37. Richmond R, Halliwell B, Chauhan J, Darbre A: Superoxide- dependent formation of hydroxyl radicals: detection of hydroxyl radicals by the hydroxylation of aromatic compounds. Anal Biochem 118: 328-335, 1981 38. Rowley D, Gutteridge JMC, Blake D, Farr M, Halliwell B: Lipid peroxidation in rheumatoid arthritis. Thiobarbituric-acid- reactive material and catalytic iron salts in synovial fluid from rheumatoid patients. Slin Sci (in press) 1984 39. Rowley DA, Halliwell B: DNA damage by superoxide-generating systems in relation to the mechanism of action of the anti- tumour antibiotic adriamycin. Biochim Biophys Acta 761: 86- 93, 1983 40. Sawyer DT, Gibian MT: The redox chemistry of superoxide ion. Tetrahedron 35: 1471-1481, 1979 41. Sohal RS (ed): Age pigments. Elsevier, Amsterdam 1981 42. Takahashi MA, Asada K: Superoxide anion permeability of phospholipid membranes and chloroplast thylakoids. Arch Biochem Biophys 226: 558-566, 1983 43. Tauber AI, Borregaard N, Simons E, Wright J: Chronic granulomatous disease: a syndrome of phagocyte oxidase deficiencies. Medicine 62: 286-309, 1983 44. Weiss SJ, Young J, LoBuglio AF, Slivka A, Nimeh NF: Role of hydrogen peroxide in neutrophil-mediated destruction of cultured endothelial cells. J Clin Invest 68: 714-721, 1981 45. Westman NG, Marklund SL: Copper-and zinc-containing superoxide dismutase and manganese-containing superoxide dismutase in human tissues and human malignant tumours. Cancer Res 41: 2962-2966, 1981 46. Weitburg AB, Weitzman SA, Destrempes M, Latt SA, Stossel TP: Stimulated human phagocytes produce cytogenetic changes in cultured mammalian cells. N Engl J Med 308: 26-30, 1983 47. Weitzman SA, Stossel TP: Mutation caused by human phagocytes. Science 212: 546-547, 1981 48. Williams MD, Leight JS, Chance B: Hydrogen peroxide in human breath and its probable role in spontaneous breath luminescence. Ann NY Acad Sci 45: 478-483, 1983 49. Willson RL: Free radicals and tissue damage: mechanistic evidence from radiation studies. In: Biochemical mechanism of liver injury. Ed. T. F. Slater. Academic Press, London 1978 50. Winterbourn CC: Lactoferrin-catalysed hydroxyl radical formation. Additional requirement for a chelating agent. Biochem J 210: 15-19, 1983 Address: B. Halliwell Department of Biochemistry, University of London King's College Strand, London UK