Biological Free Radicals

BY GRAHAM S. TIMMINS AND MICHAEL J. DAVIES

1 Introduction and Scope of Review

This review covers recent literature on the use of EPR techniques to investigate the formation and reactions of radicals in biochemical, biological, and medical systems during the period 1994 (when this area was last reviewed) to early 1998. It covers both direct EPR spectroscopy and spin trapping studies as well as related techniques; it does not cover recent developments in the synthesis and chemistry of spin traps, the formation and reactions of radicals in enzymes and metalloproteins, DNA damage, or spin labelling studies; these topics are covered elsewhere in this volume. Food irradiation, and its detection by EPR spectroscopy, has recently been extensively and thoroughly reviewed, and is also therefore not covered. Owing to the increasing interest in, and use of, EPR in the biomedical field, this review cannot be all encompassing and complete due to space limitations; we have, however, endeavoured to cover major advances that have occurred during this time period, and apologise for any omissions. Emphasis has been placed on novel discoveries and processes, and hence we have deliberately omitted the majority of (the very large number of) studies where EPR spin trapping has been employed in the assessment of putative antioxidant / scavenging compounds, in which (typically) the trapping of HO· or O2·- by DMPO (to give the well-characterised DMPO-OH or DMPO-OOH adducts) has been employed purely as a competitive 'clock' reaction. The very large volume of literature that has developed over the last few years on the trapping on nitric oxide (NO·), which has a wide variety of important biological functions, is reviewed briefly, with particular emphasis on EPR methods.

The literature covered in this review has been subdivided in a manner similar to that in previous reviews of this area with the sub-sectioning dictated by the exogenous / endogenous compounds or stimuli which result in radical formation, rather than the identity of the radicals so formed.

2 Metal Ions

2.1 Iron – The origin of the oxygen atom in HO· formed during the Fenton reaction (Fe2+ / H2O2) has been investigated using both H2O2 and H2O labelled with 17O and the spin trap DMPO. The DMPO-OH adduct was observed to give 17O couplings when labelled peroxide was employed, but not when the H2O contained 17O, establishing unequivocally that the oxygen atom in the trapped HO· arises solely from the peroxide. The role and selectivity of HO· in the degradation of hyaluronic acid, other polysaccharides and mono-saccharides has been examined using both direct rapid-flow and spin trapping EPR. In these studies little selectivity was observed with low-molecular-weight substrates, but only a few of the possible radicals were observed with some of the polymers; this may be due to the increased stability of some of these species and their slower rates of rearrangement. The selectivity of HO· attack on collagen, model peptides and free amino acids has also been investigated by EPR spin trapping using nitroso spin traps.

HO· generation by a series of iron-containing minerals (magnetite and haematite), an iron-exchanged zeolite, and asbestos fibres (chrysolite and crocidolite) has been investigated using DMPO as a spin trap. Zeolite and asbestos fibres proved to be effective catalysts for HO· generation, whereas the oxides were mostly inert. The total surface concentration of iron in these materials has been reported to be unrelated to the yield of HO·, as only a few of the iron ions on the surface appear to be active; this may be related to their redox state and coordination. A somewhat similar study has been carried out with coal mine dust, with all dust (from various mines in the USA) and standard samples giving rise to HO· generation in the presence of DMPO. In this case however there was a positive correlation between the rate of radical formation and surface iron content, and this extended to the ability of these dusts to induce lipid peroxidation. This catalytic activity has been suggested to be important in the development of pneumoconiosis and other pulmonary diseases in coal workers.

The ability of Fe2+/ascorbic acid mixtures to induce lipid oxidation on liposomes has been investigated using a number of different spin traps. In this system no evidence was obtained for HO· generation, with only carbon-centred radicals detected; these presumably arise from the lipid moieties. This conclusion is supported by the observation that the signal intensity of these radicals was dependent on the lipid concentration.

The role of iron ions in catalysing radical formation in vivo in chronic iron-loaded rats or cultured hepatocyte cells has been investigated. In the former study a secondary radical trapping technique was employed where the initially generated HO· reacts with added DMSO to give methyl radicals which are then trapped by PBN to give a stable radical adduct. The methyl radical adduct to PBN was detected in the bile of animals 10 weeks after being fed on an iron-loading diet and 40 minutes after i.p. injection of the spin trap. Desferal (desferrioxamine) completely inhibited this radical formation, and great care was taken to exclude artifactual radical formation ex vivo. In a second study 11 cultured hepatocytes were treated with various levels of iron and the formation of lipid-derived radicals detected using the spin trap POBN. Radical formation was found to be both time-, and iron concentration-, dependent over a 24 hour period. Two iron chelators - desferal and a hydroxypyrid-4-one (CP20) inhibited radical formation when these materials were introduced either before or simultaneously with the iron. A further chelator, pyoverdin, was not protective.

2.2 Copper – The role of two metal ion complexes - HgCl2 and [Fe(CN)6]3 - in releasing metal ions from Cu(l)-metallothioneins thereby promoting radical formation, has been investigated. Two DMPO adducts, the HO· adduct and a carbon-centred species, were detected when HgCl2 was used, but no oxygen-derived radicals were detected with [Fe(CN)6]3-. These results, in conjunction with inhibitor studies (using SOD and catalase), suggest that this mercury salt can displace copper ions from metallothioneins, and that the released ions subsequently autoxidise to give oxygen-centred radicals. The role of chelating ligands on the ability of copper ions to generate alkoxyl and peroxyl radicals from alkyl hydroperoxides has been studied, with the rate and extent of radical generation reported to depend crucially on the nature of the ligand. Some of the species which did not permit radical formation with the Cu2+ complex, catalysed spin adduct formation in the presence of the reductants cysteine and glutathione. Cu2+ binding to substrates has been shown to have dramatic effects on the positional selectivity of radical attack on collagen, a number of small peptides, and model compounds. EPR has also been employed to study the formation of radicals during the reaction of Cu2+ with NADH and the role of such reactions in site-specific DNA damage. This redox couple generates a carbon-centred radical, probably NAD+, from NADH, which is postulated to undergo further oxidation to NAD+ with generation of O2·-. Dismutation of the latter to H2O2 and subsequent reaction of this material with Cu+ is suggested to give rise to the DNA-damaging species.

The effect of both copper and iron ions on the apoptotic cell death induced in human promyelocytic HL-60 cells by four antioxidants (ascorbate, gallic acid, n-propyl gallate and caffeic acid) has been studied and it has been reported that such cell death is enhanced by Cu2+, but reduced by Fe3+, despite the fact that both metal ions enhanced the intensity of the ascorbyl radical signal, but reduced those from gallate and caffeic acid. The authors conclude that the ability of these metal ions to modulate radical concentrations is not the sole determinant of cytotoxic activity.

2.3 Chromium – The role of chromium ions in radical formation in biological systems has been the subject of continuing widespread interest. The reduction of Cr6+ by thiols and ascorbate has been examined. In the former case reduction gives rise to thiyl radicals; in the added presence of H2O2 or organic hydroper- oxides, these systems give HO· and hydroperoxide-derived radicals via a Cr-mediated pseudo-Fenton reaction. Reduction by ascorbate has been reported to give both Cr4+ and Cr5+ as well as carbon-centred radical adducts to DMPO. Reaction of the former ion with tBuOOH and cumene hydroperoxides gave rise to carbon-centred radical adducts to DMPO and enhanced yield of Cr5+, suggesting that Cr4+-mediated formation of radicals from lipid hydroperoxides might play a role in Cr6+-mediated carcinogenesis. Further studies by the same group have demonstrated that tetraperoxochromate(V) complexes do not play a significant role in the formation of radicals from H2O2 ; these results support the Cr5+-complexation / Fenton reaction model of carcinogenesis. Cr5+ binding to DNA has been examined and shown to be dependent on the nature of the buffer present; in this system radicals proved more effective than Cr5+ in producing strand breaks. Spin trapping studies using MNP as the spin trap, have detected base and nucleoside-derived radicals via direct formation of HO· (or possibly as a result of the degradation of Cr5+-peroxo complexes) in Cr6+ / NAD(P)H / H2O2 systems. The nucleoside-derived radicals are formed predominantly as a result of addition of HO· to the base, rather than via hydrogen-atom abstraction from the sugars; similar behaviour has been observed with homo-polymers. In contrast, studies with Cr6+ and glutathione have suggested that the formation of DNA single-strand breaks does not involve free HO·, or require added H2O2, but does involve molecular oxygen. The formation of such strand breaks was accompanied by the generation of chromium-DNA complexes.

Experiments with primary cultures of rat hepatocytes exposed to Cr6+ have shown that melatonin does not attenuate Cr5+ formation, but does limit HO· formation, and protects against DNA single-strand breaks, cytotoxicity and lipid peroxidation. Related studies have shown that reduction of Cr6+ to Cr5+ can be detected in cultured lung cells, and in intact animals injected i.p. with Cr6+. In the latter studies Cr5+ was found predominantly in the liver, with small amounts in the blood. No Cr5+ signal was detectable in heart, spleen, kidney, and lung. Pretreatment of the animals with metal ion chelators reduced the Cr5+ concentration. Cr6+ reduction has also been detected, using a surface coil resonator, in skin of living rats topically exposed to Cr6+; removal of the stratum comeum enhanced the formation and decay of Cr5+ suggesting that the skin can be an entry route for chromium into animals and humans. Other in vivo experiments have shown that radical formation can be detected in chromium poisoned rats (via gastric dosage) by spin trapping using 4-POBN. Carboncentred radical adducts, believed to be derived from endogenous lipids in the liver, were detected in bile.

2.5 Other Metal Ions – The catalytic effect of manganese on the autoxidation of dopamine has been investigated by monitoring both semi-quinone radical formation and (the six-line spectrum of) Mn2+. The catalysis is complex and reported to occur via formation of a transient complex. A vanadium-1, 10-phenanthroline complex has been shown to cleave DNA in the presence of H2O2 via binding of the complex to DNA and the pH-dependent formation of HO·; the latter has been detected by EPR spin trapping. Palladium and platinum ions have been reported to enhance strand breakage induced in super-coiled DNA by Fenton systems, with this enhancement ascribed, on the basis of both EPR spin trapping using DMPO and product analysis, to an enhanced yield of HO·. The exact mechanism of this process remains to be established.

3 Hydroperoxides

3.1 Alkyl / Aryl Hydroperoxides – Detailed mechanistic studies on the reaction between alkyl hydroperoxides (and related materials) and heme-containing proteins have been carried out by Mason and co-workers. The reaction of hematin with tBuOOH gives rise to peroxyl and alkoxyl radicals (trapped with DMPO), and methyl radicals (trapped with MNP). Alteration of the spin trap concentration has shown that the alkoxyl radical is the initial species, generated by homolytic scission of the O-O bond. The majority of the peroxyl radicals observed are methyl peroxyl radicals formed by reaction of O2 with methyl radicals arising from β-scission of the initial alkoxyl species. Some direct formation of tBuOO· was also detected, though this is a minor pathway. A similar overall conclusion (i.e. the major initial reaction is homolysis of the O-O bond to give alkoxyl radicals) has been reported in studies on the reaction of cytochrome c with both tBuOOH and cumene hydroperoxide, and rabbit cytochrome P450 1A2 with cumene hydroperoxide. In the latter case, the only peroxyl radicals detected were methyl peroxyl species, with the formation of these radicals being oxygen dependent. These studies have been extended to the reaction of P450 with linoleic acid hydroperoxide, where a similar overall mechanism has been shown to operate. At high P450 concentrations a protein-derived radical was also detected. In more recent work peroxyl, alkoxyl, methyl and protein-derived radicals have been detected in the reactions of methemoglobin and metmyoglobin with 1BuOOH. Again the major initial reaction appears to be peroxide homolysis, though the detection of protein- (globin-) derived radicals (suggested to be from valine residues in the case of methemoglobin, and both valine and tyrosine with metmyoglobin) was suggested to arise from heterolytic cleavage of the peroxide with formation of a high-oxidation-state species at the heme centre, which subsequently abstracts a hydrogen atom from neighbouring amino acid residues (see also below). These two pathways (homolysis and heterolysis) have been suggested to occur concurrently. The role of chelation of heme by hemopexin in limiting oxidative damage induced by free heme in the presence of hydroperoxides has been examined and it has been concluded that the reduced yield of oxidising species (radicals and high-oxidation-state iron complexes) formed on complexation by hemopexin arises from steric hindrance of the access of the hydroperoxide to the bound heme. The radicals involved in benzoylperoxide-mediated damage to DNA have been investigated and it has been shown that both phenyl (Ph·) and benzoyloxyl (PhCO2·) radicals are formed on decomposition induced by Cu+. Both of these species react with DNA bases, nucleosides, sugars, RNA and DNA, with the major reaction being addition to the base. Comparison of these data with those obtained with Ph· alone (generated from the diazonium salt) suggests that PhCO2· is the major damaging species in reactions with DNA.