Drs Elena Maklashina and Brian A. C. Ackrell, Molecular Biology Division (151S), VA Medical Center, 4150 Clement Street, CA 94121, USA. Tel.: +1 415 752 9676; fax: +1 415 750 6959; e-mail: addresses: email@example.com, firstname.lastname@example.org
The bulwark of the mitochondrial theory of aging is that a defective respiratory chain initiates the death cascade. The increased production of superoxide is suggested to result in progressive oxidant damage to cellular components and particularly to mtDNA that encodes subunits assembled in respiratory complexes. Earlier studies of respiration in muscle mitochondria obtained from large cohorts of patients supported this notion by showing that either singly or in combinations, the respiratory complexes exhibited decreased activity in the elderly. The following critique of the most cited publications over the past decade points out the systematic errors that put earlier work at odds with recent findings. These later investigations indicate that aging has no overt effect on either the electron transport system or oxidative phosphorylation.
Comparative studies with animals suggest that longer lifespans correlate best with low oxygen radical (ROS) production specifically by mitochondria (Perez-Campo et al., 1998). Thus oxidant damage to mitochondria or other cell components provides the most widely believed mechanism for the aging process. Debate still centres on whether electron transport dysfunction is instrumental in the process or, instead, the cumulative damage to protein, lipid and DNA (Rustin et al., 2000; Szibor & Holtz, 2003). A wealth of genetic and biochemical information exists to indicate that ROS are not only responsible for oxidative stress, but also act as secondary messengers by influencing the interplay of various signalling pathways involved in the decision by cells to undergo proliferation, senescence or apoptosis (Finkel & Holbrook, 2000; Hekimi & Guarente, 2003).
The electron transport chain
A major source of cellular superoxide (), which yields relatively stable peroxide (H2O2) by dismutation and the highly reactive hydroxyl (•OH) species by Fenton chemistry, is the respiratory chain located in the inner membrane of eukaryotic mitochondria and shown schematically in Fig. 1. In this energy transducing apparatus, electrons from NADH–ubiquinone oxidoreductase (complex I) are passed downhill via the ubiquinone (Q) pool and complex III (ubiquinone–cytochrome c oxidoreductase) to cytochrome oxidase (complex IV), where oxygen is reduced to water. Other major entry points for reducing equivalents into the chain are the three membrane-bound dehydrogenases succinate-Q oxidoreductase (complex II), ETF-Q oxidoreductase, and α-glycerolphosphate-Q oxidoreductase. These enzymes provide direct links to TCA cycle activity, fatty acid oxidation and glycolysis, respectively, and help gear activity to electron flow in the chain through competition for oxidized Q. An adequate supply of reducing equivalents is as critical to energy conservation as a properly functioning electron transport chain. Owing to pumping of protons into the intermembrane space by complexes I, III and IV during redox cycling, electron flow produces an electrochemical proton gradient (ΔµH+) across the inner membrane, which controls mitochondrial respiration and ATP synthesis by ATP synthase (complex V), ionic equilibria essential to mitochondrial stability and cell viability, and the production of . For simplicity the subsets of reactions constituting each of these processes can be grouped into the modules shown in Fig. 2.
Metabolic control analysis applied to mitochondria has shown that control of respiration in state 3 (high ATP demand) is distributed between the respiratory chain and phosphorylation system; control in heart and muscle is mainly by the respiratory chain, whereas in brain, liver and kidney it is by the phosphorylation system. Hence, within limits, heart and muscle might be expected to be more sensitive to respiratory chain deficiencies than the other tissues (Rossignol et al., 2000). In state 4 conditions (minimum ATP demand) respiration is controlled mainly by proton leak with some by substrate oxidation. Under physiological conditions mitochondrial function and thus production is altered by a variety of transient and long-term effectors, e.g. Ca2+ stimulates NADH delivery to the electron transport chain (Denton & McCormack, 1985); nitric oxide (Cooper, 2002) or phosphorylation (Lee et al., 2002a) modulates complex IV; oxidized glutathione inhibits reversibly complex I (Taylor et al., 2003) and α-ketglutarate dehydrogenase (Nulton-Persson et al., 2003) to slow TCA cycle activity; glucagon stimulates complex II (Brand et al., 1990), thyroxine modulates proton leak (Harper & Brand, 1993); and part of the anti-aging effect of calorie-restricted (CR) diets is to cause lowering of ΔµH+, and thus production, by increasing proton leak and decreasing substrate oxidation. Insulin reverses these effects (Lambert & Merry, 2003). Chronic ROS spillage can also influence expression of metabolic enzymes as part of the stress response mediated by signalling pathways through redox-sensitive transcription factors, kinases and phosphatases, as can thyroxin (Winder et al., 1980), exercise (Hood, 2001) and growth on CR diets (Kayo et al., 2001).
Production of ROS
Production of increases with reduction of the respiratory chain, and thus is higher in state 4 than state 3 conditions when there is greater restriction imposed on electron flow by a higher ΔµH+. The amount is low and dependent on mitochondrial phenotype and substrate utilized. Recent experiments with rat mitochondria have shown that production, measured as H2O2 diffusing into the external medium, is barely detectable with complex I or complex II substrates. It attains significance in heart and skeletal muscle but not in liver mitochondria oxidizing palmitoyl carnitine, which yields more reducing equivalents. The proportion of electron flux diverted from energy conservation to formation is estimated to be < 0.15% (St-Pierre et al., 2002). The sites of electron leakage are the FMN co-factor (Liu et al., 2002) and [Fe–S] cluster N-2 (Genova et al., 2001) of complex I and the Q·– intermediate formed at centre Qo of complex III. Other potential sites are ETF-Q oxidoreductase and its FAD-containing substrate, the electron transferring flavoprotein (ETF2), functioning in the fatty acid oxidation pathway (β-oxidation) (St-Pierre et al., 2002). Production of would also increase should electron pressure in the chain rise because a downstream component (Fig. 1) is incapacitated. The question arises as to whether this happens in the absence of disease. Inherited base changes and deletions in mtDNA that result in a dearth of, or mutated subunits for, respiratory chain complexes are at the root of numerous human degenerative diseases (Wallace et al., 1995). Mitochondrial DNA encodes the cytochrome b subunit of complex III, seven subunits of complex I, three subunits of complex IV, and two subunits of complex V. Other possible explanations for losses of activity by a respiratory chain component are direct oxidant damage, repressed synthesis or an inadequate supply of a prosthetic group that prevents assembly. Iron homeostasis is particularly radical-sensitive (Soum & Drapier, 2003). Of the respiratory components shown in Fig. 1, only complexes IV and V do not contain [Fe–S] clusters.
The importance of the electron transport chain as a determinant in longevity is evident in the extended lifespan exhibited by Caenorhabditis elegans mutants with a diminished Q pool or when carrying clk-1 mutations inhibiting biosynthesis of endogenous Q9 (Larsen & Clarke, 2002). Lifespan is further increased (to five-fold) in clk-1daf-2 double mutants, where antioxidant defences are also upgraded. Inhibition of daf-2, a mediator in the insulin-like signalling pathway, changes mitochondrial activity in addition to up-regulating catalase and superoxide dismutase (Hekimi & Guarente, 2003, and references therein). Increased lifespan also results from mutations in the Fe–S (Reiske) subunit of complex III (isp-1 mutants), which slow the initial step of the Q cycle and, hence, the availability of Q· for interaction with oxygen (Feng et al., 2001). Extra longevity resulting from inactivation of complex I, III, IV or V by short interference RNA (Dillin et al., 2002) probably reflects the side reduction of fumarate relieving electron pressure in the main chain. By contrast, by interfering with Q reduction, the mev-1 mutation in complex II causes over-production of and exaggerated protein carbonyl formation, and leads to ced-3- and ced-4-dependent apoptosis and aging (Senoo-Matsuda et al., 2003).
Protection against ROS
The protection afforded the respiratory chain against oxyradicals is both formidable and multifaceted. Scavengers in the inner membrane are ubiquinol (QH2), α-tocopherol and ascorbate. Ubiquinol eliminates lipid peroxyls and recycles any tocopherol or ascorbate radicals formed, and then is itself replenished by the respiratory chain. The benefit of attacking radicals at their source is evident in the vastly superior protection against apoptosis afforded by mitochondria-targeted compared to untargeted quinone and vitamin E derivatives in fibroblast cultures (Jauslin et al., 2003). Another defence is to uncouple respiration through uncoupling proteins (Brand, 2000). Main countermeasures in the matrix compartment are millimolar concentrations of glutathione for buffering redox changes, and the enzymes Mn-superoxide dismutatase (SOD2) and glutathione peroxidase (GPx1) acting in concert to convert to H2O. GPx1 (Cao et al., 2003a) and cytoplasmic catalase (Cao et al., 2003b) are both activated by phosphorylation by the c-Abl and Arg tyrosine kinases acting in response to oxidant levels. The dire consequences associated with acute oxidative stress are emphasized in nullizygous (SOD2−/–) mice. The deletion is neolethal with widespread organ damage and cell loss. Mitochondria are extremely debilitated, suffering major losses of respiratory function and a marked sensitization of the permeability transition pore that leads to premature apoptosis and cell death. The affront is more benign for the heterozygous mutant. Respiratory capacity is at a higher level, the membrane shows only a partial proton leak, and there is evidence of rapidly accumulating oxidative damage. Significantly, control mice undergo similar declines later in life (Melov et al., 1999; Kokoszka et al., 2001).
Does aging affect electron transport?
Aging is a complex mix of morphological and biochemical changes as single cells and organs fall gradually into physiological decline. Obvious traits in humans are loss of mass and strength by skeletal muscle (sarcopenia) and onset of such diseases as Alzheimer's disease, cataract, atherosclerosis and Parkinsonism. At the cellular level the accepted hallmarks are increasing mutations/deletions in mtDNA, lipid peroxidation, accumulation of nondigested debris (lipofuscin) and higher levels of ROS. These processes are more likely to be observed in post-mitotic than in proliferating tissues, where damaged material not repaired or turned over is diluted by cell division. The picture is further confounded by the fact that mitochondria are constantly renewed and adapting in response to metabolic/physical stimuli.
Inspection of publications from the past decade dealing with age-related loss of electron transport activity in various tissues (discussed in detail below) shows that almost as many investigations provide supportive evidence as those that do not.
(i) Skeletal muscle
In their benchmark publication over a decade ago, Trounce et al. (1989) reported a statistically significant decline in the activities of complexes I, II and IV of mitochondria isolated from the vastus lateralis muscle of 29 orthopaedic patients aged from 20 to 90 years. Complexes III and V were not assayed. The data were interpreted as reflecting a disproportionate decline in oxidative capacity with age, because activities were related to mitochondrial protein and because the oxidation of tyramine by monoamine oxidase present in the outer membrane remained constant. The choice of monoamine oxidase activity as an invariant marker for mitochondria is suspect. The outer membrane of skeletal muscle mitochondria was already known to contain two enzymes oxidizing tyramine, MAOA and MAOB (Kalaria & Haric, 1987), each one potentially able to compensate for any decline of activity by the other. The activity of each form is known to increase many-fold with age in other tissues (Saura et al., 1997; Maurel et al., 2003) and lower MAOB activity in platelets is associated with smoking, gender and life-long alcohol consumption (Snell et al., 2002). That complexes I and IV seemed to undergo disproportionate losses of activity with age was later corroborated with matrix citrate synthase activity as reference (Cooper et al., 1992). However, the statistical analysis was conducted on too few muscle samples (n = 17) to cover the 20–90 years age range and was heavily skewed by the inclusion of seven patients over 50 years old with Parkinson's disease.
Another investigation often cited in support of the mitochondrial theory of aging was by Boffoli et al. (1994), who found a statistically significant age-related decline in the activities of complexes I, II and IV, but not of complex III, among a cohort of orthopaedic patients (n = 59) between the ages of 17 and 91 years. No mitochondrial marker was used for reference. Loss of activity by complex IV was attributed to oxidant damage because there was no accompanying loss of enzyme judged from cytochrome aa3 difference spectra. Attempts to establish that the levels of complex I and III also remained constant with age, a critical control, were inconclusive. A similar selective decline of complexes I and IV had been observed by Sugiyama et al. (1993) to occur late in the lifespans of dogs and rats in diaphragm, skeletal muscle and heart but not liver mitochondria. These findings corroborated previous studies on rats (Torii et al., 1992), but were counter to those of Lenaz et al. (1997), who reported that only complex I declined with age in liver, heart and muscle. In mouse, the activities of complexes changed over time without any perceived pattern other than post-mitotic tissues such as brain, heart and muscle were more adversely affected than proliferating tissues such as liver and kidney (Kwong & Sohal, 2000).
A major flaw in most of these investigations (5/7) was the assumption that preparations isolated from young and aged tissues contained like amounts of impurities and loss of activity correlated with damage to the respiratory chain (but see Bowling et al., 1993; Rasmussen et al., 2003). The sporadic sensitivity of complexes, particularly that of complex II, which is nuclear encoded, is also somewhat puzzling considering that there are well-established methods for assaying each complex. However, most of the investigations (5/7) can be criticized for the use of imprecise assay procedures for membrane preparations, specifically the omission of inhibitors needed to preclude non-specific activity, and for not activating complex II prior to assay. Thus, with impure preparations of complex I, it is important to measure only rotenone-sensitive activity, particularly when ubiquinone analogs are used as electron acceptors. Antimycin A distinguishes complex III activity from the non-specific interaction between ubiquinol and cyt. c. Variable fractions of complex II in membrane preparations (40–60%) are inevitably in a deactivated state due to the tight (KD ∼ 0.01 µm) binding of inhibitory oxaloacetate picked up during isolation. It is resistant to centrifugation or column passage. Activation is usually achieved by incubation with chaotropes or succinate at elevated temperatures (30–38 °C). Repeat assays in the presence of high concentrations of malonate, a competitive inhibitor, allow adjustment for non-specific reduction by endogenous substrates.
Barrientos et al. (1996) re-investigated muscle mitochondria from 132 orthopaedic patients between the ages of 15–95 years and confirmed the inverse relationship between age and substrate oxidation intimated by Trounce et al. (1989). Importantly, the effect of aging became much less significant once confounding variables such as exercise (Berthon et al., 1995; Brierly et al., 1996) and smoking (Smith et al., 1993) were taken into account in a multivariate statistical analysis. Exercise increases the volume density of mitochondria in young skeletal muscle, less so in old muscle, and shifts substrate metabolism to a higher reliance on fatty acids, which yield more ATP/mole. These responses are mediated via a number of nuclear transcription factors (reviewed in Hood, 2001). Increasing levels of the peroxisome proliferator-activated receptor γ (PPARα/γ) co-activator PGC-1α and factors NFR-1 and NFR-2 promote expression of the genes of the TCA cycle, respiratory chain and heme biosynthesis. PGC-1α in conjunction with PPARα allows the genes of the β-oxidation pathway to be regulated separately. Factors NRF-1 and Tfam increase the transcription/replication of mtDNA and co-ordinate the supply of nuclear and mitochondria-encoded subunits destined for assembly in the respiratory chain. Benefits from exercising are lost by bed rest or sedentary lifestyle. In several species, aging is associated with reduced levels of protein transcripts and mitochondrial volume in muscle and other tissues (Welle et al., 2003, and references therein), and with lower tissue rates of protein synthesis (Rooyackers et al., 1996) and oxidative capacity (Kent-Braun & Ng, 2000).
With improved isolation methods, Chretien et al. (1998) obtained similar yields of functionally intact mitochondria (90%, respiratory control values with succinate > 4.0) from both young and senescent deltoid muscle biopsy samples taken from orthopaedic patients aged 0–60 years. In agreement with other recent investigations (Barrientos et al., 1996; Rasmussen et al., 2003), little correlation between the activities of the individual respiratory complexes and age was observed. The fact that the marked scattering of data points is lessened considerably if activity ratios between respiratory complexes are plotted is a further indication that physiological variables as well as methodological inaccuracies are contributory factors. Rasmussen et al. (2003), also using highly purified preparations, extended these observations by showing the lack of age-related effects on P/O ratios, state 4 respiration, the respiratory chain, pyruvate dehydrogenase or the TCA cycle. Significantly, β-oxidation did decline with age consistent with the expected change in mitochondrial plasticity with age (see above). The noted lack of effect on phosphorylation efficiency would suggest that phospholipid peroxidation has little affect on bioenergetic function.
The incidence of deletions in the mtDNA of aged muscle homogenates is low (< 0.1%). Upon examination of single fibres from different muscles and animals it is clear that mutated mtDNA occurs focally and is coincident with localized loss of complex IV (COX) activity (Kopsidas et al., 2002; McKenzie et al., 2002). Accumulation of the same deletion product throughout the abnormal region is taken as evidence for the replicative advantage of deleted over wild-type mtDNA (clonal expansion). The affected segments retain SDH activity (COX−/SDHnormal regions) or have increased (COX−/SDH++) levels (ragged red fibres) as the nucleus attempts to overcome bioenergetic insufficiency. The COX−/SDH++ regions in the latter phenotype are longer, lead to atrophy and breakage, and loss of the fibre (sarcopaenia). Mitochondria in COX− regions do not contain detectable full-length DNA, but heterogeneous mixtures of rearranged mtDNA species, leading to the suggestion that only full-length mtDNA is functional (Kopsidas et al., 2002). More recent evidence suggests that many of the earlier reported deletions represent partial duplications (aborted) or damaged mtDNA in the process of degradation. A similar pattern of cytochrome oxidase deficiency occurs among cardiomyocytes (< 1%) (Müller-Höcker, 1989).
Although aged mammalian hearts suffer some minor age-related losses of cardiomyocytes and are less tolerant to ischaemia/reperfusion damage, loss of electron transport function is not consistently observed. Some laboratories report no attenuation of respiratory complexes with age (Torii et al., 1992; Barazzoni et al., 2000; Kwong & Sohal, 2000), whereas others have observed decreased complex I and IV activities (Sugiyama et al., 1993; Andreu et al., 1998), transcript levels (Andreu et al., 1998; Hudson et al., 1998) and protein synthesis (Hudson et al., 1998). In rat models of congestive heart failure, declines in transcripts for subunits of complex IV and enzyme activity are associated with decreased levels of the nuclear factors PGC-1α, Tfam and NRF-2 (Garnier et al., 2003). Cardiac interfibrillar mitochondria located between the myofibrils exhibit lower protein yield and oxidative capacity in elderly rats. The damage has been located to complex III and is manifested as an extra leakage of electrons from the Qo site. Subsarcolemmal mitochondria (SSM) lying beneath the plasma membrane are not affected (Moghaddas et al., 2003).
No such confusion exists with regard to respiratory capacity in the human heart. No age-related loss of activity by complexes I, II, III or IV could be detected in heart homogenates by Miróet al. (2000) among patients (n = 59) 8–86 years old, of whom 16 were smokers. As observed with human skeletal muscle mitochondria, the data points exhibited a high degree of scattering but no statistically significant decline with age whether activities were plotted directly, as ratios of activities or relative to the mitochondrial marker citrate synthase. The findings were consistent with those of Marin-Garcia et al. (1998), who showed in addition that complex V activity and transcripts for complex IV were also unaffected. No up-regulation of SOD or glutathione peroxidase occurred despite the observed increase in lipid peroxidation in homogenates.
Age-related effects on electron transport in brain tissue seem restricted primarily to complex I. Studies by Bowling et al. (1993) of crude mitochondrial preparations from the fronto-parietal cortex of rhesus monkeys showed a statistically significant decline in complex I activity relative to that of citrate synthase and in complex IV, but to a lesser degree. Complex II–III activity was apparently not affected with age. The scattergrams presented are the result in the main of imprecise assay protocols and complex II not being activated prior to assay (see above). A similar disparity between the age-related declines of complex I and IV activities was evident in mouse brain (Navarro et al., 2002), where the losses in 18-month-old animals amounted to 40% and 15%, respectively, relative to complex II activity, which remained constant. As an indication of increasing oxidative stress, thiobarbituric acid reactive substances (50%), cytoplasmic SOD (52%) and matrix SOD (108%) had all increased. Other workers found no significant age-related declines for any of the complexes in rat brain (Filburn et al., 1996; Sandhu & Kaur, 2003). A report by Ferrándiz et al. (1994) of loss of activities in non-synaptic but not in synaptic mouse mitochondria must be accepted with caution. The specific activities determined for complex I are some 30-fold too high with respect to those of the other complexes, complex II was not activated prior to assay and the possibility that preparations from older tissue contained more impurity was not excluded. However, the distinction between the two preparations would seem to be real. Lenaz et al. (1997) observed decreases in the turnover numbers (kcat) for complex I of 39% and 18%, respectively, in non-synaptic and synaptic mitochondria isolated from rat cortex. The impact on respiration, however, is questionable. It has been determined that rat complex I activity could be inhibited by 72% in non-synaptic mitochondria (Davey & Clark, 1996) and 25% in synaptic mitochondria (Davey et al., 1998) before respiration and ATP synthesis become impaired. What is important is that the gradual pro-oxidizing shift in the glutathione redox state with age (Kamzalov & Sohal, 2003) reduces complex I activity and abolishes the threshold.
(iv) Proliferating tissue
Respiratory chain complexes in mitotic tissues such as liver and kidney from a variety of species seem not to decline with age (Sugiyama et al., 1993; Barazzoni et al., 2000; Kwong & Sohal, 2000; Navarro et al., 2002). In cell culture both human fibroblasts (Martinez et al., 1991) and rat hepatocytes (Hagen et al., 1997) have been shown to develop distinct populations judged on the basis of membrane potential and replicative ability. Mild oxidant challenge to human lung fibroblast cultures hastens replicative senescence and stimulates mitochondrial biogenesis (mtDNA copy number, mRNA, transcription factors) to meet the energy needs for survival. At high oxidant levels, however, genes are down-regulated as a prelude to apoptosis and cell death (Lee et al., 2002b).
Earlier investigations that provided evidence for age-related inactivation of respiratory chain components and, thus, for the mitochondrial theory of aging can be challenged on several counts. Chief among these was the decision in most investigations not to use reference markers for the mitochondrial content of their preparations. As it stands, the possibility that the observed declines in activity were due to higher levels of impurity in preparations from older tissue cannot be excluded and is, in fact, likely. No age-related loss of activity by respiratory complexes is evident from later experiments utilizing mitochondrial preparations of consistently high quality and purity. There is also the question of why groups studying the same type of biopsy samples should detect activity losses by different complexes unless there were selective preparative damage or assay problems. The use of inadequate assay procedures in the early investigations was discussed previously.
There is agreement as to the vulnerability of complex I in brain to age-related inactivation. The modifying agent(s) is unknown. A complex I deficiency is often found in post mortem brain tissue from patients with Parkinsonism, a late onset neurological disease (Schapira et al., 1990). It is of interest therefore that Parkinsonian-like symptoms are induced in rat models following administration of specific complex I inhibitors rotenone (Betarbet et al., 2000) and MPTP (Heikkila et al., 1984), both of which are environmental toxins. Another agent may be the peroxynitrite (ONOO−) radical present at low level in the cell (Murray et al., 2003).
The pervading view that the electron transport chain in post-mitotic tissues becomes dysfunctional with aging, and thereby becomes a force in the aging process, is not supported by the recent data. Also, Rasmussen et al. (2003) showed that neither state 4 respiration rates nor oxidative phosphorylation were affected by aging. This would suggest that lipid peroxidation does not influence greatly the permeability of the inner membrane to protons, as was previously suspected, or bioenergetic sufficiency. Moreover, the retention of full activity by the respiratory chain is evidence that radical-induced mutations in mtDNA have little effect on the normal supply of pristine subunits for assembly in the respiratory chain. All of these data run counter to the mitochondrial theory of aging.
This work was supported in part by the Department of Veterans Affairs and by NIH grant GM61606.