DNA AND CELL BIOLOGYVolume 21, Number 4, 2002© Mary Ann Liebert, Inc.Pp. 251–257

Review Articles

HO2?: The Forgotten Radical

AUBREY D.N.J. DE GREY

ABSTRACT

HO2?, usually termed either hydroperoxyl radical or perhydroxyl radical, is the protonated form of superox-ide; the protonation/deprotonation equilibrium exhibits a pKa of around 4.8. Consequently, about 0.3% ofany superoxide present in the cytosol of a typical cell is in the protonated form. This ratio is rather accuratelyreflected by the published literature on the two species, as identified by a PubMed search; at the time of writ-ing only 28 articles mention “HO2,” “hydroperoxyl” or “perhydroxyl” in their titles, as against 9228 men-tioning superoxide. Here it is argued that this correlation is not justifiable: that HO2?’s biological and bio-medical importance far exceeds the attention it has received. Several key observations of recent years arereviewed that can be explained much more economically when the participation of HO2? is postulated. It issuggested that a more widespread appreciation of the possible role of HO2? in biological systems would be ofconsiderable benefit to biomedical research.

251

INTRODUCTION

THE “BIBLE” OF FREE RADICAL BIOLOGY is Halliwell and Gut-teridge’s Free Radicals in Biology and Medicine, whose

third edition was published in 1999 (Halliwell and Gutteridge,1999). As the originating reactive oxygen species formed bynumerous metabolic processes, in particular by mitochondrialelectron transport, superoxide is naturally discussed at greatlength there. By contrast, its protonated form, HO2? or hy-droperoxyl radical, is only mentioned briefly, and does not evenrate an entry in the index. The most prominent reference toHO2? is the following sentence (p. 61):

Although at the pH of most body tissues the ratio of[O2?2]/[HO2?] will be large (100/1 at pH 6.8, 1000/1 atpH 7.8), the high reactivity of HO2? and its uncharged na-ture, which might allow it to cross membranes more read-ily than the charged O2?

2, have combined to maintain in-terest in this species.

It is difficult to agree with either the beginning or the end ofthis statement. First, a ratio of between two and three orders ofmagnitude is small when compared to the ratio in reactivity of

HO2? and O2?2 with many biomolecules, so the chemistry ofsuperoxide in living systems is in large part dominated by thereactions of HO2?. Second, interest in HO2? has conspicuouslynot been maintained: at the time of writing, just 28 articles listedin PubMed mention “HO2,” “hydroperoxyl,” or “perhydroxyl”in their titles, of which the last study of the radical’s biologi-cal activity is from 1996 (Dix et al., 1996).

In this article we highlight a selection of significant findingsfrom the literature of the past decade that can, it will be argued,be best understood by a consideration of the possible involve-ment of HO2?, a scenario that has generally not been adequatelydiscussed in relation to those observations. The examples cho-sen focus on the mitochondrion, but HO2?’s potential relevanceelsewhere is also implied. It is hoped that these examples willinspire a restoration of interest in HO2?’s possible biologicalroles.

A CHALLENGE TO DELOCALIZEDCHEMIOSMOSIS

Mitchell’s chemiosmotic theory was first published in 1961(Mitchell, 1961). It was not until 1966, however, that in a much

Department of Genetics, University of Cambridge, Cambridge CB2 3EH, UK.

more detailed exposition of his model (Mitchell, 1966) Mitchellmade explicit a component of it, which is now widely consid-ered indispensable: its spatially delocalized nature. The protoncircuit that transfers energy from the electron transport chain tothe ATP synthase involves the travel of protons through theaqueous medium on either side of the mitochondrial inner mem-brane; this has been robustly demonstrated by experiments in-volving external manipulation of the transmembrane proton gra-dient (Thayer and Hinkle, 1975; Jagendorf and Uribe, 1996).What is less clear, however, is how constrained is the path thatthose protons take within the two aqueous compartments. Thedelocalized chemiosmotic theory states that the path is con-strained only by the boundary of each aqueous compartment—in other words, the broadly proton-impermeant plasma mem-brane and inner mitochondrial membrane for the cytosoliccompartment and the inner mitochondrial membrane alone forthe matrix compartment. By this model, any two points withinthe same compartment are at equal electrochemical potential.(The speed of proton movement by the Grotthuss mechanismis so great that effects due to protons entering and exiting acompartment at distinct sites can be neglected.)

It may initially seem hard to imagine how this could bewrong, but some remarkably direct evidence suggests that it is.In a series of studies between 1969 and 1984 (see Tedeschi,1980, for a review of the main work), the group of Tedeschifound that mitochondria could synthesize ATP despite gener-ating no proton gradient whatever. Their technique for mea-suring the membrane potential was with microelectrodes ingiant mitochondria, a method that can be argued to be less dis-

constrained to a region of the cytosol, the proton-pumping ac-tion of the electron transport chain will acidify that region rel-ative to the rest of the cytosol. The smaller the volume of theregion that contains the chemiosmosis-engaged protons, thegreater this acidification will be. The models mentioned above(Kell, 1979; de Grey, 1999) propose that it is a layer only ,1nm thick on the outside of the inner membrane, allowing theacidification to be considerable.

Guidot et al. (1995) obtained a result that appears to implythat this acidification indeed occurs as a result of mitochondrialrespiration. They isolated mitochondria from a strain of yeast thatlacked the mitochondrial isoform of superoxide dismutase(SOD), and exposed them to an extramitochondrial source of su-peroxide. They then measured the rate of dismutation of that su-peroxide, which must be nonenzymatic unless there is SOD inthe intermembrane space (a topic that will be discussed below).They found that the rate of dismutation was markedly greaterwhen the mitochondria were actively respiring than when, by anyof a variety of means, the formation of a proton gradient was pre-vented. The effect was also found in wild-type yeast (i.e., withmitochondrial SOD). Guidot et al. attributed this to a respiration-dependent acidification of the intermembrane space (into whichextramitochondrially generated superoxide would readily diffuse,because of the pores in the outer membrane). This would pro-duce the effect observed, because the lower pH would increasethe proportion of superoxide that was protonated to HO2?, andwould thus increase the rate of nonenzymatic dismutation (Fig.1), according to the rate constants derived by Bielski (1978):

DE GREY252

FIG. 1. According to the surface-restricted (Kell, 1979; de Grey, 1999), but not the textbook delocalized (Mitchell, 1961), ver-sion of the chemiosmotic theory, mitochondria with a substantial proton gradient (left) acidify the outer surface of the inner mem-brane, leading to greater protonation of superoxide and thus faster dismutation than when the proton gradient is weak (right).This is in accordance with observed rates (Guidot et al., 1995).

ruptive to the electrical properties of the organelle than the morecommon approach, introduction of lipophilic cationic dyes thatequilibrate according to the Nernst equation. A model was pro-posed in 1979 (Kell, 1979) that went some way towards ex-plaining the phenomenon; it was soon seen to possess an im-portant flaw (Mitchell, 1981), but a refinement that “repairs” itwas proposed recently (de Grey, 1999).

A study involving superoxide dismutation (Guidot et al.,1995) appears to shed new light on this issue and strongly chal-lenge delocalized chemiosmosis. Whatever the details of themechanism, if the cytosolic component of the proton circuit is

[These values, together with a pKa of 4.69 for the equilib-rium between O2?2 1 H1 and HO2?, were derived (Bielski,1978) by measuring the second-order rate constant for the de-cay of O2?2 as a function of pH. This also requires knowledgeof the molar extinction coefficients of O2?2 and HO2?, whoseless accurate measurements had previously implied slightly dif-ferent values for the pKa.]

However, Guidot et al. did not draw attention to the fact thatthis model flatly contradicts the generally accepted version of thechemiosmotic theory (namely, the delocalized model), and sup-ports the heretical alternative (Kell, 1979; de Grey, 1999) men-

O2?2 1 O2?

2 1 2H2O R O2 1 H2O2 1 2OH2 k , 0.35 M21s21

HO2? 1 O2?2 1 H2O R O2 1 H2O2 1 OH2 k 5 (1.02 6 0.49) 3 108 M21s21

HO2? 1 HO2? R O2 1 H2O2 k 5 (8.60 6 0.62) 3 105 M21s21

tioned above. In consequence, attention has not yet been paid tothis potentially pivotal result by bioenergeticists at large.

ENDOGENOUS SUPEROXIDE IN THEMITOCHONDRIAL INTERMEMBRANE SPACE

The previous section discussed an experiment involving ex-ternally generated superoxide diffusing into the mitochondrialintermembrane space; other work bears on the relevance of pen-etration into the intermembrane space of superoxide generatedby the electron transport chain (ETC). It has been argued (Tur-rens, 1997) that all the superoxide made by electron leak fromthe ETC emerges on the matrix side of the membrane, but theevidence cited in support of this conclusion only demonstratesthat some does. Additionally, however, it has been noted (But-ler et al., 1982) that superoxide reacts very rapidly with ferri-cytochrome c, whose millimolar concentration in the inter-membrane space would seem sufficient to eliminate anysuperoxide there before it can undergo any potentially toxic re-actions. (The reaction with ferricytochrome c restores the su-peroxide to molecular oxygen, and the electron is simply trans-ported to Complex IV, so any potential for toxicity is averted.)

However, Kerver et al. (1997) demonstrated in an eleganthistochemical study that singlet oxygen is formed in the inter-membrane space. The relevance of this result is that, while thereaction of superoxide with ferricytochrome c (or with SOD inthe oxygen-regenerating step of its cycle) produces ground-stateoxygen, nonenzymatic dismutation is reported (Corey et al.,1987) to produce singlet oxygen. Although this is, of course,not proof that nonenzymatic dismutation of superoxide is tak-ing place in the intermembrane space, it constitutes an argu-ment by elimination, in that no alternative mechanism for theformation of singlet oxygen in that compartment has ever beensuggested. Thus, whether or not some mitochondrially gener-ated superoxide emerges in the matrix, evidently some alsoemerges in the intermembrane space; additionally, the presenceof ferricytochrome c is not sufficient to eliminate it. The es-sential role of HO2? in nonenzymatic dismutation is evidentfrom the rate constants mentioned in the previous section. How-ever, Kerver et al. did not mention this, instead stating that thedismutation of superoxide anions was responsible.

Additionally, Balzan et al. (1999) engineered a transgenicsuperoxide dismutase in yeast, with a leader sequence thatcaused it to be localized to the intermembrane space. This wasshown to improve the resistance to hyperoxia of yeast thatlacked the mitochondrial SOD, again suggesting that superox-ide is present in the intermembrane space. However, it must be

siders the protonated form, HO2?, because, as mentioned in theIntroduction, HO2? is uncharged and thus more membrane-per-meant than superoxide.

A recent report (Okado-Matsumoto and Fridovich, 2001)suggests that mitochondria possess a superoxide dismutase iso-form in the intermembrane space. If this is correct, it is diffi-cult to understand why addition of transgenic SOD should bebeneficial (Balzan et al., 1999), or why the presence of a pro-ton gradient should considerably accelerate dismutation (Guidotet al., 1995). A possible explanation is that the intermembranespace SOD does not exist in yeast, the organism used in boththose studies; this is hinted at by the observation (Guidot et al.,1995) that the respiration-dependent component of nonenzy-matic dismutation was much greater in yeast mitochondria thatin mitochondria isolated from two rat tissues. Kerver et al.(1997) studied rat tissues, however.

Finally, it must be noted that the occurrence of nonenzymaticdismutation requires the collision of two superoxide molecules(one or both protonated), and thus their presence in the inter-membrane space of the same mitochondrion at the same time.This is not in accordance with the classical estimate (Chanceet al., 1979) of a concentration of superoxide in the intermem-brane space of the order of 10211 M. Thus, that estimate mayneed to be reconsidered.

AGE-RELATED MITOCHONDRIALDYSFUNCTION

A number of groups, working with various tissues of vari-ous species, have reported that mitochondrial superoxide (orhydrogen peroxide) production rises with age (Sawada andCarlson, 1987; Sohal et al., 1990, 1994) (although the robust-ness of this finding has also been questioned; Barja, 1999). Ithas also been reported that the mitochondrial proton gradientdeclines with age (Sastre et al., 1996; Hagen et al., 1997). Thiscombination of changes is highly paradoxical at first sight, be-cause superoxide production is found to rise with increasingmembrane potential (Hansford et al., 1997). [A mechanism ex-plaining this last result in the case of Complex III concerns theease of movement of electrons between the hemes of cy-tochrome b (Demin et al., 1998); a corresponding model forComplex I must await more detailed understanding of its en-zymatic mechanism than is yet available.] An explanation isproposed here that revolves around the role of HO2? in initi-ating lipid peroxidation, something that it (but not superoxide)can do even in the absence of pre-existing lipid hydroperox-ides (Bielski et al., 1983):

HO2?: THE FORGOTTEN RADICAL 253

conceded that the absence of mitochondrial SOD might itselfcause a rise in the level of intermembrane space superoxide,because matrix superoxide might have a longer lifetime, andthus have greater potential to diffuse through the inner mem-brane. This possibility is made more plausible when one con-

[Note that, as shown by the above results and others in thesame study (Bielski et al., 1983), and since, HO2? abstracts onlybisallylic H atoms of fatty acids.]

In exploring how this combination of changes can comeabout, one must examine in more detail what factors determine

linoleic acid (18:2) 1 HO2? R linoleoyl? 1 H2O2 k 5 (1.18 6 0.20) 3 103 M21s21

linolenic acid (18:3) 1 HO2? R linolenoyl? 1 H2O2 k 5 (1.70 6 0.35) 3 103 M21s21

arachidonic acid (20:4) 1 HO2? R arachidonoyl? 1 H2O2 k 5 (3.05 6 0.29) 3 103 M21s21

the proportion of electrons that escape from the respiratory chainto form superoxide. In addition to the membrane potential, thisproportion is affected by the proportion of the time that the“leaky” sites in the respiratory chain possess an electron—thedegree of reduction of those sites. This is, in turn, determinedby the efficiency of electron flow both upstream and downstreamof those sites. The classic example is that antimycin A blockselectron transfer on the matrix-side ubiquinone-binding site ofcytochrome b (Stater, 1973), and thus increases the reduction ofthe cytosolic-side ubiquinol-binding site, which is where super-oxide production by Complex III seems mainly to occur (Hanet al., 2001), whereas myxothiazol blocks electron flow up-stream of this site (von Jagow et al., 1984); correspondingly, an-timycin A increases and myxothiazol decreases the rate of su-peroxide production at Complex III (Turrens et al., 1985).

The efficiency of the respiratory chain enzymes can be hy-pothesized to alter as a function of age due to at least two fac-tors: oxidative damage to its constituent proteins, and changesto their environment. Because superoxide production rises withage, oxidative damage occurs at a higher rate; moreover, thehalf-life of mitochondrial turnover, and thus the lifetime of theaverage mitochondrial membrane protein, is increased in olderindividuals (Rooyackers et al., 1996). However, some enzymesare much more sensitive than others to this damage: those con-taining iron–sulphur clusters are particularly susceptible (Gard-ner and Fridovich, 1992; Andersson et al., 1998). The majormicroenvironmental change with age may be the reduced levelof cardiolipin in the inner membrane (Paradies et al., 1997).Again, this affects some proteins more than others; ComplexIV is particularly dependent on cardiolipin as a cofactor (Fryand Green, 1980; Paradies et al., 1997).

The sensitivity of Complex IV to cardiolipin levels provides

a plausible explanation for increased superoxide production: be-cause it is the terminal enzyme of the respiratory chain, its lossof efficiency can be predicted to increase the degree of reduc-tion of the earlier enzymes that are responsible for superoxideproduction. This effect may outweigh the inhibitory effect ofthe lowered membrane potential on superoxide production. Butwhat remains to be explained is why the cardiolipin level fallsin the first place, because the enzymes necessary for its syn-thesis are present and the required rate of synthesis is presum-ably very low (being set by the rate of mitochondrial turnover).

Again, a possible answer may be found in the role of HO2?(Fig. 2). Cardiolipin is the only negatively charged phospho-lipid present in significant quantity in the inner mitochondrialmembrane: the other phospholipids present there, phos-phatidylcholine and phosphatidylethanolamine, are zwitterionic(Hovius et al., 1990). Negative fixed charges on the surface ofa membrane cause a surface potential (an attraction of cationsand a repulsion of anions), which in this case has been esti-mated to be about 60 mV; this results, in turn, in a reductionof pH at the membrane of about one unit (McLaughlin, 1977).Thus, lower cardiolipin levels will result in a less acid aqueousphase at the membrane surface, and hence, in a reduction in theproportion of superoxide released that is converted to HO2?.This is likely to reduce the rate of peroxidation of mitochon-drial lipids and proteins, even if there is a higher rate of pro-duction of superoxide, because it has been shown that HO2? isresponsible for the vast majority of initiation of lipid peroxi-dation reactions in the inner membrane (Antunes et al., 1996;Kowald, 1999). The effect is independent of the transmembraneproton gradients, although the lowering of that gradient withage (Sastre et al., 1996; Hagen et al., 1997) also inhibits HO2?

formation as explained previously (Fig. 1).

DE GREY254

FIG. 2. Cardiolipin in the mitochondrial inner membrane (left) causes a substantial acidification of the surface aqueous phase(independent of transmembrane proton gradient), due to its fixed negative charge. This increases HO2? levels and consequentlylipid peroxidation rates, relative to a mitochondrion with a depleted cardiolipin concentration (right). CL, cardiolipin.

Thus, cardiolipin decline may be a regulated, compensatoryresponse to age-related deterioration of other cellular compo-nents, such as the lysosomal apparatus that is responsible formitochondrial turnover (Grisolia et al., 1979). If accumulatingundegradable material such as lipofuscin impairs lysosomalfunction, as has been postulated by Brunk (Brunk et al., 1992;Terman and Brunk, 1998), that could explain the slowing ofmitochondrial turnover with age (Rooyackers et al., 1996). Alonger lived mitochondrion must be adjusted to inflict free rad-ical-mediated membrane damage upon itself more slowly, orelse it will become unable to support a proton gradient and willbe a drain on cellular resources (de Grey, 1997). Thus, pro-duction of the free radicals that are responsible for this damagemust be inhibited. In support of this hypothesis, restoration ofcardiolipin levels in old rats by dietary supplementation withacetyl-L-carnitine causes a rise in the rate of oxidative damage(Hagen et al., 1998).

SUPEROXIDE PRODUCTION BY COMPLEX III

Another intriguing observation with relevance to the role ofoxidative damage in aging is the absence of a correlation, acrosshomeotherm species, between maximum lifespan and antioxi-dant enzyme levels (adjusted for specific metabolic rate). Earlywork by Cutler suggested that such a correlation did exist forsuperoxide dismutase but not for other antioxidant enzymes(Tolmasoff et al., 1980; Cutler, 1991); even SOD is now known,however, to exhibit, if anything, a lower level in longer livedspecies of comparable metabolic rate if birds are included inthe calculation (Ku and Sohal, 1993; Perez-Campo et al., 1998).Evolution of slower aging is accompanied by the expectedchanges in two other parameters related to free radical damage,namely the rate of production of superoxide as a proportion ofoxygen consumption (Perez-Campo et al., 1998) and the de-gree of unsaturation of membrane (especially mitochondrial in-ner membrane) phospholipids (Pamplona et al., 1996), whichdetermines their susceptibility to free radical-mediated peroxi-dation (Cosgrove et al., 1987). But the natural expectation isthat the level of antioxidant enzymes, which is just as much un-der genetic control as are the two parameters just mentioned,should also be adjusted in the appropriate direction when evo-lutionary pressure for increased longevity is present.

It was recently proposed (de Grey, 2000) that this has failedto occur because the location in which the lifespan-limiting ox-idative damage occurs is one in which such enzymes are inef-fective, because they are absent; namely, the mitochondrial in-termembrane space. Whereas superoxide generated in themitochondrial matrix may be easily and completely detoxifiedby mitochondrial SOD, any that is generated on the outside ofthe inner membrane will have a longer lifetime and—in viewof the more acidic environment there than in the matrix—willbe more likely to become protonated to HO2? and react with aphospholipid of the membrane. This may initiate a chain reac-tion that propagates through the membrane and eventually dam-ages the mitochondrial DNA, which is attached to the mem-brane (Albring et al., 1977).

This particular model (de Grey, 2000) is seriously challengedby the recent report (Okado-Matsumoto and Fridovich, 2001) ofan intermembrane space SOD, and might in any case be consid-

ered fragile in view of its reliance on the assumption that evolu-tion has, for some reason, been unable to evolve a SOD isoformdirected to the intermembrane space. But recently a variation onthe hypothesis has been put forward (Muller, 2000) that escapesboth of these objections. The cytosolic-side quinone-binding siteof Complex III is not at the membrane surface but slightly buriedwithin the membrane, and a glutamic acid residue lies very closeto it. Muller (2000) proposed that, on those occasions when anelectron passing from ubiquinol to the heme group goes astrayand becomes attached to a molecule of oxygen, forming super-oxide, a proton from the glutamic acid residue is likely also tobe transferred to the oxygen. In other words, HO2? may be formedwithin the inner membrane itself. Such a scenario allows the pos-sibility of initiation of lipid peroxidation before any antioxidantenzymes, even including a putative intermembrane space SOD,have had access to the originating free radical. It may thus, es-pecially if increasing knowledge of the mechanism of superox-ide production by Complex I leads to a similar model, be re-garded as a highly plausible explanation for the ostensiblyparadoxical failure of homeotherm evolution to use regulation ofantioxidant enzyme levels as one of its strategies to select forlong life.

CONCLUSION

The evidence surveyed in this article demonstrates that HO2?

probably plays a central role in mediating the toxic side effectsof aerobic respiration, due to its high reactivity with biomole-cules and its membrane permeance. Additionally, inclusion ofHO2? in the analysis of the chemiosmotic proton circuit is seento offer a valuable insight into a controversy that has simmeredfor decades due to the absence of any conclusive experimentaldata. Because HO2? is present in any compartment where su-peroxide is present and the pH is physiological, biochemical re-actions and pathways that produce extramitochondrial super-oxide—such as plasma membrane electron transport(O’Donnell and Azzi, 1996; de Grey, 1998; McArdle et al.,2001)—should also be carefully analyzed in the context of theeffects that HO2? may have.

REFERENCES

ALBRING, M., GRIFFITH, J., and ATTARDI, G. (1977). Associationof a protein structure of probable membrane derivation with HeLacell mitochondrial DNA near its origin of replication. Proc. Natl.Acad. Sci. USA 74, 1348–1352.

ANDERSSON, U., LEIGHTON, B., YOUNG, M.E., BLOMSTRAND,E., and NEWSHOLME, E.A. (1998). Inactivation of aconitase andoxoglutarate dehydrogenase in skeletal muscle in vitro by superox-ide anions and/or nitric oxide. Biochem. Biophys. Res. Commun.249, 512–516.

ANTUNES, F., SALVADOR, A., MARINHO, H.S., ALVES, R., andPINTO, R.E. (1996). Lipid peroxidation in mitochondrial inner mem-branes. I. An integrative kinetic model. Free Radic. Biol. Med. 21,917–943.

BALZAN, R., AGIUS, D.R., and BANNISTER, W.H. (1999). Clonedprokaryotic iron superoxide dismutase protects yeast cells against ox-idative stress depending on mitochondrial location. Biochem. Bio-phys. Res. Commun. 256, 63–67.

HO2?: THE FORGOTTEN RADICAL 255

BARJA, G. (1999). Mitochondrial oxygen radical generation and leak:Sites of production in states 4 and 3, organ specificity, and relationto aging and longevity. J. Bioenerg. Biomembr. 31, 347–366.

BIELSKI, B.H.J. (1978). Reevaluation of the spectral and kinetic prop-erties of HO2 and O2

2 free radicals. Photochem. Photobiol. 28,645–649.

BIELSKI, B.H.J., ARUDI, R.L., and SUTHERLAND, M.W. (1983).A study of the reactivity of HO2/O2

2 with unsaturated fatty acids. J.Biol. Chem. 258, 4759–4761.

BRUNK, U.T., JONES, C.B., and SOHAL, R.S. (1992). A novel hy-pothesis of lipofuscinogenesis and cellular aging based on interac-tions between oxidative stress and autophagocytosis. Mutat. Res. 275,395–403.

BUTLER, J., KOPPENOL, W.H., and MARGOLIASH, E. (1982). Ki-netics and mechanism of the reduction of ferricytochrome c by thesuperoxide anion. J. Biol. Chem. 257, 10747–10750.

CHANCE, B., SIES, H., and BOVERIS, A. (1979). Hydroperoxide me-tabolism in mammalian organs. Physiol. Rev. 59, 527–605.

COREY, E.J., MEHROTRA, M.M., and KHAN, A.U. (1987). Waterinduced dismutation of superoxide anion generates singlet molecu-lar oxygen. Biochem. Biophys. Res. Commun. 145, 842–846.

COSGROVE, J.P., CHURCH, D.F., and PRYOR, W.A. (1987). Thekinetics of the autoxidation of polyunsaturated fatty acids. Lipids 22,299–304.

CUTLER, R.G. (1991). Antioxidants and aging. Am. J. Clin. Nutr. 53,373S–379S.

DE GREY A.D.N.J. (1997). A proposed refinement of the mitochon-drial free radical theory of aging. BioEssays 19, 161–166.

DE GREY, A.D.N.J. (1998). A mechanism proposed to explain the risein oxidative stress during aging. J. Anti-Aging Med. 1, 53–66.

DE GREY A.D.N.J. (1999). Incorporation of transmembrane hydrox-ide transport into the chemiosmotic theory. Bioelectrochem. Bioen-erg. 49, 43–50.

DE GREY, A.D.N.J. (2000). The non-correlation between maximumlongevity and enzymatic antioxidant levels among homeotherms; im-plications for retarding human aging. J. Anti-aging Med. 3, 25–36.

DEMIN, O.V., KHOLODENKO, B.N., and SKULACHEV, V.P.(1998). A model of O2?-generation in the complex III of the electrontransport chain. Mol. Cell Biochem. 184, 21–33.

DIX, T.A., HESS, K.M., MEDINA, M.A., SULLIVAN, R.W., TILLY,S.L., and WEBB, T.L. (1996). Mechanism of site-selective DNAnicking by the hydrodioxyl (perhydroxyl) radical. Biochemistry 35,4578–4583.

FRY, M., and GREEN, D.E. (1980). Cardiolipin requirement by cy-tochrome oxidase and the catalytic role of phospholipid. Biochem.Biophys. Res. Commun. 93, 1238–1246.

GARDNER, P.R., and FRIDOVICH, I. (1992). Inactivation-reactiva-tion of aconitase in Escherichia coli. A sensitive measure of super-oxide radical. J. Biol. Chem. 267, 8757–8763.

GRISOLIA, S., KNECHT, E., HERNANDEZ-YAGO, J., and WAL-LACE, R. (1979). Turnover and degradation of mitochondrial andtheir proteins. Ciba Found. Symp. 75, 167–188.

GUIDOT, D.M., REPINE, J.E., KITLOWSKI, A.D., FLORES, S.C.,NELSON, S.K., WRIGHT, R.M., and MCCORD J.M. (1995). Mi-tochondrial respiration scavenges extramitochondrial superoxide an-ion via a nonenzymatic mechanism. J. Clin. Invest. 96, 1131–1136.

HAGEN, T.M., INGERSOLL, R.T., WEHR, C.M., LYKKESFELDT,J., VINARSKY, V., BARTHOLOMEW, J.C., SONG, M.H., andAMES, B.N. (1998). Acetyl-L-carnitine fed to old rats partially re-stores mitochondrial function and ambulatory activity. Proc. Natl.Acad. Sci. USA 95, 9562–9566.

HAGEN, T.M., YOWE, D.L., BARTHOLOMEW, J.C., WEHR, C.M.,DO, K.L., PARK, J.Y., and AMES, B.N. (1997). Mitochondrial de-cay in hepatocytes from old rats: Membrane potential declines, het-erogeneity and oxidants increase. Proc. Natl. Acad. Sci. USA 94,3064–3069.

HALLIWELL, B., and GUTTERIDGE, J.M.C. (1999) Free Radicalsin Biology and Medicine (Oxford University Press, Oxford, UK).

HAN, D., WILLIAMS, E., and CADENAS, E. (2001). Mitochondrialrespiratory chain-dependent generation of superoxide anion and itsrelease into the intermembrane space. Biochem. J. 353, 411–416.

HANSFORD, R.G., HOGUE, B.A., and MILDAZIENE, F. (1997). De-pendence of H2O2 formation by rat heart mitochondrial on substrateavailability and donor age. J. Bioenerg. Biomembr. 29, 89–95.

HOVIUS, R., LAMBRECHTS, H., NICOLAY, K., and DE KRUIJFF,B. (1990). Improved methods to isolate and subfractionate rat livermitochondria. Lipid compensation of the inner and outer membrane.Biochim. Biophys. Acta 1021, 217–226.

JAGENDORF, A.T., and URIBE, E. (1996). ATP formation caused byacid-base transition of spinach chloroplasts. Proc. Natl. Acad. Sci.USA 55, 170–177.

KELL, D.B. (1979). On the functional proton current pathway of elec-tron transport phosphorylation. An electrodic view. Biochim. Bio-phys. Acta 549, 55–99.

KERVER, E.D., VOGELS, I.M., BOSCH, K.S., VREELING-SINDE-LAROVA, H., VAN DEN MUNCKHOF, R.J., and FREDERIKS,W.M. (1999). In situ detection of spontaneous superoxide anion andsinglet oxygen production by mitochondrial in rat liver and small in-testine. Histochem. J. 29, 229–237.

KOWALD, A. (1999). The mitochondrial theory of aging: Do damagedmitochondria accumulate by delayed degradation? Exp. Gerontol. 34,605–612.

KU, H.H., and SOHAL, R.S. (1993). Comparison of mitochondrial pro-oxidant generation and anti-oxidant defenses between rat and pigeon:Possible basis of variation in longevity and metabolic potential.Mech. Ageing Dev. 72, 67–76.

MCARDLE, A., PATTWELL, D., VASILAKI, A., GRIFFITHS, R.D.,and JACKSON, M.J. (2001). Contractile activity-induced oxidativestress: Cellular origin and adaptive responses. Am. J. Physiol. 280,C621–C627.

MCLAUGHLIN, S. (1977). Electrostatic potentials at membrane-solu-tion interfaces. Curr. Top. Membr. Transport 9, 71–143.

MITCHELL, P. (1961). Coupling of phosphorylation to electron andhydrogen transfer by a chemiosmotic type of mechanism. Nature 191,144–148.

MITCHELL, P. (1966). Chemiosmotic coupling in oxidative and photo-synthetic phosphorylation. Biol. Rev. Camb. Philos. Soc. 41, 445–502.

MITCHELL, P. (1981). Proton conduction and chemi-osmosis. Chem.Britain 17, 166.

MULLER, F. (2000). The nature and mechanism of superoxide pro-duction by the electron transport chain: Its relevance to aging. J. Am.Aging Assoc. 23, 227–253.

O’DONNELL, V.B., and AZZI, A. (1996). High rates of extracellularsuperoxide generation by cultured human fibroblasts: Involvementof a lipid-metabolizing enzyme. Biochem. J. 318, 805–812.

OKADO-MATSUMOTO, A., and FRIDOVICH, I. (2001). Subcellu-lar distribution of superoxide dismutases (SOD) in rat liver: Cu, Zn-SOD in mitochondria. J. Biol. Chem. 276, 38388–38393.

PAMPLONA, R., PRAT, J., CADENAS, S., ROJAS, C., PEREZ-CAMPO, R., LOPEZ-TORRES, M., and BARJA, G. (1996). Lowfatty acid unsaturation protects against lipid peroxidation in liver mi-tochondria from long-lived species: The pigeon and human case.Mech. Ageing Dev. 86, 53–66.

PARADIES, G., RUGGIERO, F.M., PETROSILLO, G., andQUAGLIARIELLO, E. (1997). Age-dependent decline in the cy-tochrome c oxidase activity in rat heart mitochondria: Role of car-diolipin. FEBS Lett. 406, 136–138.

PEREZ-CAMPO, R., LOPEZ-TORRES, M., CADENAS, S., ROJAS,C., and BARJA, G. (1998). The rate of free radical production as adeterminant of the rate of aging: Evidence from the comparative ap-proach. J. Comp. Physiol. B 168, 149–158.

ROOYACKERS, O.E., ADEY, D.B., ADES, P.A., and NAIR, K.S.

DE GREY256

(1996). Effect of age on in vivo rates of mitochondrial protein syn-thesis in human skeletal muscle. Proc. Natl. Acad. Sci. USA 93,15364–15369.

SASTRE, J., PALLARDO, F.V., PLA, R., PELLIN, A., JUAN, G., O’-CONNOR, J.E., ESTRELA, J.M., MIQUEL, J., and VINA, J. (1996).Aging of the liver: Age-associated mitochondrial damage in intacthepatocytes. Hepatology 24, 1199–1205.

SAWADA, M., and CARLSON, J.C. (1987). Changes in superoxideradical and lipid peroxide formation in the brain, heart and liver dur-ing the lifetime of the rat. Mech. Ageing Dev. 41, 125–137.

SLATER, E.C. (1973). The mechanism of action of the respiratory in-hibitor, antimycin. Biochim. Biophys. Acta 301, 129–154.

SOHAL, R.S., ARNOLD, L.A., and SOHAL, B.H. (1990). Age-relatedchanges in antioxidant enzymes and prooxidant generation in tissuesof the rat with special reference to parameters in two insect species.Free Radic. Biol. Med. 9, 495–500.

SOHAL, R.S., KU, H.H., AGARWAL, S., FORSTER, M.J., and LAL,H. (1994). Oxidative damage, mitochondrial oxidant generation andantioxidant defenses during aging and in response to food restrictionin the mouse. Mech. Ageing Dev. 74, 121–133.

TEDESCHI, H. (1980). The mitochondrial membrane potential. Biol.Rev. Camb. Philos. Soc. 55, 171–206.

TERMAN, A., and BRUNK, U.T. (1998). Lipofuscin: mechanisms offormation and increase with age. APMIS 106, 265–276.

THAYER, W.S., and HINKLE, P.C. (1975). Kinetics of adenosine

triphosphate synthesis in bovine heart submitochondrial particles. J.Biol. Chem. 250, 5336–5342.

TOLMASOFF, J.F., ONO, T., and CUTLER, R.G. (1980). Superoxidedismutase: Correlation with life-span and specific metabolic rate inprimate species. Proc. Natl. Acad. Sci. USA 77, 2777–2781.

TURRENS, J.F. (1997). Superoxide production by the mitochondrialrespiratory chain. Biosci. Rep. 17, 3–8.

TURRENS, J.F., ALEXANDRE, A., and LEHNINGER, A.L. (1985).Ubisemiquinone is the electron donor for superoxide formation bycomplex III of heart mitochondria. Arch. Biochem. Biophys. 237,408–414.

VON JAGOW, G., LJUNGDAHL, P.O., GRAF, P., OHNISHI, T., andTRUMPOWER, B.L. (1984). An inhibitor of mitochondrial respira-tion which binds to cytochrome b and displaces quinone from theiron–sulfur protein of the cytochrome bc1 complex. J. Biol. Chem.259, 6318–6326.

Address reprint requests to:Aubrey D.N.J. de Grey

Department of GeneticsUniversity of Cambridge

Downing StreetCambridge CB2 3EH, UK

E-mail: ag24@gen.cam.ac.uk

HO2?: THE FORGOTTEN RADICAL 257