• Open Access

Oxidative stress and aging: beyond correlation


Simon Melov, Buck Institute for Age Research, 8001 Redwood Blvd, Novato, CA 94945, USA. Tel.: 415 209 2068; fax: 415 209 2232; e-mail: smelov@buckinstitute.org


The oxidative stress theory of aging has become increasingly accepted as playing a role in the aging process, based primarily on a substantial accumulation of circumstantial evidence. In recent years, the hypothesis that mitochondrially generated reactive oxygen species play a role in organismal aging has been directly tested in both invertebrate and mammalian model systems. Initial results imply that oxidative damage, specifically the level of superoxide, does play a role in limiting the lifespans of invertebrates such as Drosophila melanogaster and Caenorhabditis elegans. In mammalian model systems, the effect of oxidative stress on lifespan is less clear, but there is evidence that antioxidant treatment protects against age-related dysfunction, including cognitive decline.

Oxidative stress and aging

It has become increasingly accepted that damage resulting from reactive oxygen species (ROS), or oxidative stress, plays a role in the aging process. The level to which oxidative stress contributes to aging may vary between organisms, tissues and distinct cell types. The oxidative stress theory of aging proposes that ROS, which primarily result from normal mitochondrial metabolism, cause progressive damage resulting in the functional decline that defines aging (Harman, 1972). Between 0.4 and 4% of the oxygen consumed by the mitochondria is converted into ROS including superoxide (Chance et al., 1979; Turrens & Boveris, 1980; Boveris, 1984; Turrens et al., 1985; Hansford et al., 1997). This superoxide can react with iron–sulphur clusters, causing the release of dangerous free iron. In addition, superoxide is converted by superoxide dismutase (SOD) to hydrogen peroxide, which can react with free iron to produce the highly reactive hydroxyl radical, which in turn will readily damage biomolecules including DNA, protein and membrane lipids (Fridovich, 1978; Halliwell & Gutteridge, 1999).

Organisms have evolved defences against these potentially dangerous molecular species, including three forms of SOD, which detoxifies superoxide. Mammals have two isoforms of copper–zinc SOD; extracellular (SOD3) and cytosolic (SOD1), and a manganese SOD which is localized to the mitochondrial matrix (SOD2) (Halliwell & Gutteridge, 1999). In addition, recent studies have demonstrated that SOD1 is also localized to the mitochondrial intermembrane space (Okado-Matsumoto & Fridovich, 2001; Sturtz et al., 2001; Higgins et al., 2002). SOD converts superoxide to hydrogen peroxide, which is then converted to water by either catalase or glutathione peroxidase. There are a range of other protective mechanisms against ROS, including chemical antioxidants such as vitamins C and E (Halliwell & Gutteridge, 1999), and systems to remove molecules once damaged.

For many years, evidence for the oxidative stress theory of aging was limited to correlative studies demonstrating that oxidative damage accumulates with age, or that the rate of accumulation of oxidative damage scales with lifespan (Beckman & Ames, 1998). Damage to DNA, protein and lipids has been shown to increase with age in many organisms including humans, mice, Drosophila melanogaster and the nematode Caenorhabditis elegans (Sohal et al., 1993; Yan & Sohal, 1998; Yasuda et al., 1999; Hamilton et al., 2001; Stadtman, 2001). Reducing antioxidant defences or increasing oxidative stress has resulted in a shortening of lifespan (Melov et al., 1999; Ishii et al., 1998; Moskovitz et al., 2001) which can be interpreted as support for the theory, though there are many ways to reduce lifespan which have nothing to do with normal aging. In addition, a correlation between metabolic rate and lifespan was described, which has been interpreted as revealing an underlying relationship between ROS production and lifespan (Pearl, 1928). Longer-lived organisms also have an increased resistance to oxidative stress relative to shorter-lived species, consistent with the free radical theory of aging (Kapahi et al., 1999).

Taken together, these correlative studies present a strong circumstantial case for the universal involvement of oxidative stress in the aging process. Recently, the oxidative stress theory of aging has been tested directly; this work is discussed below and summarized in Tables 1 and 2. Much of this critical work has been carried out in invertebrate model systems.

Table 1.  Summary of genetic interventions testing the oxidative stress hypothesis of aging
ManipulationSystemEffect on lifespanEffect on oxidative stressReference
  • *

    ND: not determined.

Catalase reducedDrosophilaReducedND*Mackay & Bewley (1989)
Catalase increasedDrosophilaNo effectReduced sensitivityGriswold et al. (1993), Orr & Sohal (1992), Sun & Tower (1999)
SOD1 reducedDrosophilaReducedIncreased sensitivityPhillips et al. (1989)
SOD1 increasedDrosophilaNo effectNo effectSeto et al. (1990)
SOD1 increasedDrosophilaIncreased in some strainsReduced sensitivity in some strainsOrr & Sohal (1993), Reveillaud et al. (1991)
SOD1 increasedDrosophilaIncreasedReduced sensitivityParkes et al. (1998), Sun & Tower (1999), Staveley et al. (1990)
SOD1 increasedMouseNDReduced sensitivityEpstein et al. (1987), Cardozo-Pelaez et al. (1998)
SOD1 and catalase increasedDrosophilaIncreasedReduced sensitivityOrr & Sohal (1994), Sohal et al. (1995)
SOD1 and catalase increasedDrosophilaNo added benefit over SOD1 aloneNo added benefit over SOD1 aloneSun & Tower (1999)
SOD2 increasedDrosophilaNo effectNo effectMockett et al. (1999a)
SOD2 increasedDrosophilaIncreasedNDSun et al. (2002)
Glutathione reductase increasedDrosophilaIncreased lifespan under hyperoxiaReduced protein carbonylation under hyperoxiaMockett et al. (1999b)
Glutathione peroxidase increasedMouseNDReduced peroxide levelsMirochnitchenko et al. (1995)
Peptide methionine sulphoxide reductase increasedDrosophilaIncreasedIncreased resistanceRuan et al. (2002)
Peptide methionine sulphoxide reductase decreasedMouseReducedIncreased sensitivityMoskovitz et al. (2001)
Table 2.  Summary of pharmacological or dietary interventions testing the oxidative stress hypothesis of aging
ManipulationSystemEffect on lifespanReduction of oxidative stressCognitive benefitsReference
  • *

    ND: not determined.

Euk-8, Euk-134C. elegansIncreasedND*NDMelov et al. (2000)
N-acetylcysteineDrosophilaIncreasedNDNDBrack et al. (1997)
MelatoninDrosophilaIncreasedYesNDBonilla et al. (2002)
Vitamin EMouseNoneNDNDLipman et al. (1998), Morley & Trainor (2001)
Vitamin ERatNDYesYesJoseph et al. (1998)
Glutathione (GSH)MouseNoneNDNDLipman et al. (1998)
LA and/or acetyl-L-carnitine (ALCAR)RatNDYesYesLiu et al. (2002b)
Strawberry extractRatNDYesYesJoseph et al. (1998, 1999)
Strawberry extractMouseNoneNDNDLipman et al. (1998)
Spinach extractRatNDNoYesJoseph et al. (1998, 1999)
Blueberry extractRatNDYesYesJoseph et al. (1999)
N-tert-butyl-α-phenylnitrone (PBN)GerbilNDYesYesCarney et al. (1991), Dubey et al. (1995)
N-tert-butyl-α-phenylnitrone (PBN)HouseflyNoneNDNDDubey et al. (1995)

Tests of the oxidative stress theory of aging in invertebrate model systems: genetic interventions

There are now several examples of genetic manipulation of the steady state level of ROS, or of the ability to repair oxidative damage, in Drosophila. Levels of enzymes that form the organism's front line of defence against reactive oxygen such as superoxide dismutase and catalase have been modified. Other studies have increased levels of enzymes that repair damaged molecules. Increasing levels of some antioxidant defences and repair systems has been demonstrated to increase lifespan concomitant with a decrease in oxidative damage, a true test and the best support to date for the hypothesis that this damage limits lifespan under normal conditions.

Reducing the activity of catalase or SOD1 results in a greater sensitivity to oxidative stress and reduced longevity (Mackay & Bewley, 1989; Phillips et al., 1989). Based on these data, many groups set out to determine whether the converse was true, whether increased catalase or SOD activity would increase lifespan. The results indicated that elevated catalase expression increased resistance to oxidative stress, but without extending lifespan (Orr & Sohal, 1992; Griswold et al., 1993; Sun & Tower, 1999). In contrast, overexpression of SOD1 increased lifespan, but only under specific genetic paradigms (Orr & Sohal, 1993; Seto et al., 1990; Staveley et al., 1990; Reveillaud et al., 1991; Orr & Sohal, 1994; Sohal et al., 1995; Parkes et al., 1998). The discrepancy in the results of these various genetic systems may be due, in part, to experimental complications including the need to separate any confounding effects of transgene expression on development from its effects on adult lifespan, and the necessity for identical genetic backgrounds for the control and transgenic strains. One study (Sun & Tower, 1999) addressed both of these potential difficulties by using an inducible system to overexpress SOD1 only after the completion of development. The authors found that increasing SOD1 activity 1.2–1.5 fold in adult flies resulted in an increase in mean and maximum lifespan in a dose-dependent manner in two different genetic backgrounds. The experimental design was strengthened by the fact that the control lines and the overexpressing lines were identical except for the heat pulse required to initially turn on the expression of the extra SOD1. The heat pulse itself was demonstrated to have no effect or a negative effect on lifespan. Therefore, this study supports the hypothesis that oxidative damage plays a role in limiting the lifespan of Drosophila.

A second study that found a positive effect of SOD1 overexpression on lifespan is notable in that the expression was limited to motor neurones in adult flies (Parkes et al., 1998), implicating the significance of oxidative stress in the nervous system as limiting lifespan in the fly. This study also controlled for the genetic backgrounds of the control and transgenic flies.

Overexpression of SOD2 has also resulted in conflicting results with respect to fly lifespan (Mockett et al., 1999a; Sun et al., 2002). The recent study by Sun et al. (2002), in which SOD2 was overexpressed, experimentally used the same paradigm as used previously with SOD1 (Sun & Tower, 1999). An inducible expression system allowed the expression of SOD2 to be activated after development was complete, and assured that the control strain had a genetic background identical to that of the overexpressing population. These experiments revealed a 12–29% increase in lifespan. A second study, in which an extra copy of SOD2 was present throughout life, found no effect on lifespan (Mockett et al., 1999a). Possible explanations for the discrepancy between these two studies include a potential negative effect of excess SOD2 activity during development, or perhaps differences in the localization of the extra SOD2 activity in the two genetic systems (Sun et al., 2002).

Another enzyme involved in the removal of ROS, glutathione reductase (GR), was overexpressed in Drosophila with the expectation that the amount of reduced glutathione (GSH) would be increased, resulting in increased resistance to oxidative stress and an increased lifespan (Mockett et al., 1999b). Under normoxic conditions, there was no effect of GR overexpression on GSH content, resistance to paraquat or lifespan. In contrast, transgenic flies exhibited reduced protein carbonylation and extended lifespan only under hyperoxic conditions. This study indicates that, in contrast to detoxification of superoxide, the rate of conversion of GSSG to GSH may be limiting to lifespan in the fly under hyperoxic but not normoxic conditions. Taken with the lack of lifespan extension seen in catalase overexpression experiments, this study also implicates that controlling levels of superoxide is more important to lifespan than levels of hydrogen peroxide.

Another genetic manipulation that extended lifespan of Drosophila was the overexpression of peptide methionine sulphoxide reductase (MSRA) (Ruan et al., 2002), an enzyme which repairs oxidized methionines in proteins. Expression of MRSA has been demonstrated to decrease with age in rats (Petropoulos et al., 2001), and reducing levels of MSRA reduces lifespan in mice (Moskovitz et al., 2001). In the fly experiments, overexpression was located predominantly to the nervous system, which in combination with the SOD1 experiments mentioned above again implicates oxidative stress in the nervous system as important in limiting the lifespan of Drosophila. In addition to having an extended lifespan, the animals were resistant to oxidative stress, and their reproductive period and movement levels were maintained longer compared to non-transgenic flies. This study indicates that not just the levels of ROS themselves, but also damage resulting from ROS, can be limiting to Drosophila lifespan.

Tests of the oxidative stress theory of aging in invertebrate model systems: pharmacological interventions

Pharmacological interventions expected to reduce ROS levels have also been demonstrated to increase lifespan in two systems, the nematode C. elegans (Melov et al., 2000) and Drosophila (Brack et al., 1997; Bonilla et al., 2002). The use of drugs to reduce levels of ROS eliminates some of the complexities that can result from genetic manipulations. Controls and treated groups are identical, with no complications resulting from different genetic backgrounds, or from altered physiology in response to a redirection of resources to allow the expression of an unnaturally high level of one protein. It is important, however, that the pharmacological agents are demonstrated to have no unexpected effects on aspects of physiology such as reproduction or metabolic rate.

Two catalytic antioxidants, Euk-8 and Euk-134, were demonstrated to extend the lifespan of C. elegans (Melov et al., 2000). These compounds mimic the catalytic activities of both superoxide dismutase and catalase, and thus are more effective than non-catalytic antioxidants such as vitamin C or vitamin E. Treatment of C. elegans with the Euk compounds extended lifespan without affects on growth or reproduction. Levels of ROS or forms of oxidative damage have not yet been assayed in treated animals to demonstrate that treatment is reducing oxidative stress.

In Drosophila, feeding of two compounds, melatonin (Bonilla et al., 2002) and N-acetylcysteine (NAC), has been demonstrated to extend lifespan (Brack et al., 1997). The hormone melatonin is an antioxidant and free radical scavenger (Reiter et al., 2002), while NAC is a precursor to glutathione synthesis and feeding is expected to increase steady-state levels of the antioxidant GSH. Feeding melatonin to flies increased their resistance to paraquat in addition to lengthening lifespan, indicating that the melatonin may function as an antioxidant in this system. The NAC experiments did not test for changing levels of oxidative damage or resistance to oxidative stress in the treated flies, so it is possible that the lifespan extension observed in these experiments is due to some other effect of NAC supplementation. These studies did, however, report that there was no effect of NAC on the body weight or motility of the flies.

Tests of the oxidative stress theory of aging in mammalian model systems

As discussed above, there is convincing evidence that oxidative stress limits lifespan in invertebrate model systems, particularly oxidative stress in the nervous system. However, to date, there have been no reported experiments in which the lifespan of mammals has been lengthened by increasing either endogenous antioxidants such as SOD1 and glutathione peroxidase (Epstein et al., 1987; Mirochnitchenko et al., 1995), or exogenous antioxidant levels including vitamin E and GSH (Lipman et al., 1998; Morley & Trainor, 2001). Decreasing MRSA in the mouse results in a shortened lifespan and increased sensitivity to oxidative stress (Moskovitz et al., 2001), but a study investigating whether this type of oxidative damage is limiting to the lifespan of a wild-type animal has not yet been conducted. There are a number of examples that have been reported in which the age-related decline in cognitive function in mammals has been slowed or reversed by reducing the level of ROS (Carney et al., 1991; Joseph et al., 1998, 1999; Liu et al., 2002a). Age-related declines in cognitive function in humans are of increasing concern as improved health care allows greater numbers of people to reach old age, providing an impetus for studying the role of oxidative stress in such declines and the ability of antioxidants to help maintain cognitive functions.

Supplementing the regular diet of rats with extracts of fruits and vegetables containing high levels of antioxidants slows or reverses age-related cognitive defects (Joseph et al., 1998). Chronic long-term dietary supplementation with vitamin E, extracts of strawberry or extracts of spinach protect against oxidative stress in the brain as measured by dichlorofluorescien diacetate (DCF) fluorescence (Ueda et al., 1997), which measures the level of oxidants produced by cells harvested directly after death. The diet also retarded age-associated decreases in cognitive function as determined by the Morris water maze, which measures hippocampal function (Morris, 1984).

Extracts of blueberry, spinach or strawberry added to the diet of old animals for 8 weeks reversed the age-related functional decline concomitant with reduced oxidative stress (Joseph et al., 1999). The strawberry and blueberry extracts but not the spinach extract resulted in significantly reduced levels of oxidative stress, while all extracts significantly restored cognitive function as measured in the Morris water maze.

The results of these experiments suggest that by decreasing the level of oxidative stress, cognitive function is maintained or restored in aged rats. Alternatively, it is possible that the reduced oxidative stress and restored cognitive function are due to independent effects of the extracts. This may be why a cognitive function in the spinach-fed group increased in the absence of an observed reduction in oxidative stress. However, only one measure of oxidative stress (DCF-fluorescence) was tested. Other measures of oxidative stress that measure the end product of ROS damage, such as isoprostane content or protein carbonyl content, would have been valuable in these studies.

Administration of the free radical spin-trapping compound N-tert-butyl-α-phenylnitrone (PBN) to gerbils reduces oxidative damage and reverses age-associated decreases in short-term memory (Carney et al., 1991). Old gerbils were injected with PBN for 14 consecutive days prior to behavioural analysis and death. Oxidative damage was measured by determining the carbonyl content of brain proteins. Protein carbonyl content increased by 85% in the old animals compared to young controls and was significantly reduced with PBN treatment. Short-term memory, as measured using an eight-arm radial maze, decreased with age and was restored to the level observed in the young gerbils with PBN treatment. What is remarkable about these studies is that 14 days of treatment had such a dramatic effect on short-term memory. The effect of PBN on protein carbonyls appears to be very species and tissue specific. Later studies also showed a reduction in protein carbonyls in the brains of gerbils with PBN treatment but found no reduction in protein carbonyls in gerbil heart or old mouse brain (Dubey et al., 1995). Additionally, it was determined that PBN treatment does not increase the lifespan of the housefly. It is not clear as to whether the effect of PBN on protein carbonyls and short-term memory is a product of its antioxidant properties. Later studies have concentrated on the effects of PBN as a transcriptional modulator (Miyajima & Kotake, 1997; Kotake et al., 1998; Sang et al., 1999). Therefore, it is possible that PBN has other unknown effects on gerbil physiology that would explain the benefits derived from such a short period of treatment.

Mitochondria are a primary source of ROS, which are produced as a by-product of the generation of ATP by oxidative phosphorylation. Age-associated mitochondrial dysfunction can be reversed by feeding rats mitochondrial metabolites (Liu et al., 2002b). The hypothesis that improvement of mitochondrial function would result in reduced oxidative damage and reductions in age-related cognitive dysfunction was tested in rats (Liu et al., 2002a). Lipid peroxidation levels were significantly higher in old rats than in young rats. Feeding old rats R-α-lipoic acid (LA), an antioxidant and a co-enzyme for pyruvate dehydrogenase and α-ketoglutarate dehydrogenase, for 8 weeks decreased the level of lipid peroxidation to that seen in young rats. Feeding old rats both LA and acetyl-l-carnitine (ALCAR), which is required for transporting long-chain fatty acids into the mitochondria, resulted in significant reductions in oxidized nucleic acid in the hippocampus, cerebral cortex and in white matter as well as a reversal of age-associated deficits in spatial memory using the Morris water maze. Both ALCAR and LA on their own are able to reduce the level of oxidized nucleic acids in many regions of the brain and improve performance in the water maze. The mechanism through which this occurs may be due to the known antioxidant activity of LA.

The most encouraging view of the studies described above is that the age-related loss of neurological function in mammals can be partially reversed through the supplementation of antioxidants. However, there is no definitive proof that the mechanism through which these various compounds improve function is via their antioxidant properties. It is likely that there will be further reports demonstrating the ability of compounds containing antioxidant activities to reverse age-related dysfunction. With additional reports such as those described above, it may become clear that age-related neuronal dysfunction can be caused by oxidative stress and thus be reversed through antioxidant supplementation. The results in these studies suggest that age-related decline in cognitive function in humans could be reduced by consuming diets rich in antioxidants or through supplementing their diet with antioxidants.


The studies discussed above demonstrate the first direct tests of the oxidative stress hypothesis of aging. As in many areas of molecular medicine, model systems have been used in aging research to identify basic phenomenon and interventions that may prove to be valid and beneficial in human aging. Invertebrate models such as Drosophila and C. elegans have proved useful for the study of aging primarily because of their short lifespans, their powerful genetics and because of the ease of producing large, synchronized populations for aging studies. They allow for the direct testing of theories, carried out through either genetic or pharmacological intervention, much more quickly and economically than can be done in mammals. However, conclusions drawn from these systems may not always be applicable to mammals. None of the genetic modifications that have been successful in lifespan extension in flies has yet been successful when tested in mice, though future experiments may find that aspects of transgene expression, whether tissue or temporal specificity, impact the results of overexpression studies. Inconsistencies such as these demonstrate the importance of moving such studies from invertebrate into mammalian model systems to identify universal mechanisms of aging.

Mammalian models such as rats, gerbils and mice should prove useful in studies testing the role of oxidative stress in lifespan, as they have in the caloric restriction paradigm (Weindruch et al., 1982). Genetic and pharmacological studies beyond those discussed above are currently underway in many laboratories. In addition to being more relevant models for human aging, mammalian models are excellent models for the dissection of the role of oxidative stress in age-related dysfunction such as cognition. The goal of much of aging research is not to extend the lifespan of humans per se, but to improve and extend the functional lifespan of aging individuals. To this extent, the prevention of age-associated cognitive decline is of major interest and importance. Mammalian model systems are already providing promising results in studies testing the efficacy of antioxidants in the treatment of age-related dysfunction, and therefore the immediate future looks to be an exciting and interesting one with regard to the oxidative stress theory of aging.