Cytochrome oxidase deficiency generated by RNAi of sft-1 resulted in lifespan extension accompanied by decreased germ cell production but no obvious slowdown of developmental rate, whereas the lifespan extension caused by RNAi of oxa-1 was associated with a pronounced slow growth phenotype and lethal defects in embryonic development. The combination of dead embryos and long-lived progeny in the broods of oxa-1(RNAi) animals could reflect some mitochondrial “threshold effect” in which the threshold for mitochondrial function is surpassed in some embryos, leading to catastrophic mitochondrial damage and the inability to invoke compensatory pathways.
The embryonic lethality and slow growth of viable progeny observed with RNAi of oxa-1 correlates with phenotypic observations of oxa-1 homologues in other organisms. Inactivation of both OXA1 genes in S. pombe is lethal , whereas a knockout mutant of S. cerevisiae is viable, but is unable to respire and can only grow on fermentable carbon sources . The function of these various OXA1 proteins is conserved, as N. crassa, human and S. pombe homologues can rescue S. cerevisiae oxa1 null mutants . In yeast, defects in Oxa1 ablate the function of the cytochrome oxidase complex and greatly reduce levels of Complex III and F0F1ATPase . Overall, the data suggest that OXA1 is important for the biogenesis of several different respiratory chain complexes (reviewed in ) and this may account for the severity of the oxa-1 RNAi phenotype seen in C. elegans. Consistent with this, no OXA1 deficient patients have been identified and this condition may thus be embryonic lethal in humans.
SURF1 deficient patients, on the other hand, usually survive into early childhood with a progressive neurodegenerative disease. In contrast to the human SURF1 deficient phenotype, no abnormal neuromuscular function was observed in sft-1(RNAi) worms, despite detailed observations of movement and pharyngeal pumping rates (data not shown). A possible caveat to this conclusion, however, is that RNAi in C. elegans is known to be relatively ineffective in the nervous system, thus neuronal phenotypes may have been masked in our experiments. It is also possible that the lack of a neuromuscular phenotype in the sft-1 deficient worms may simply reflect overall energy requirements and the level of energy generation in different tissues. Previous studies of metabolic activity in C. elegans, as monitored by oxygen consumption, indicate that only a small proportion of total energy expenditure is on movement and the greatest single energetic demand is for reproduction . There is a significant increase in oxygen consumption at the L3 and L4 larval stages, the time of extensive proliferation of germ cells (reviewed in . Reflecting this dependence of reproduction on energy generating capacity, mutations leading to deficiency of two other components of the mitochondrial oxidative phosphorylation system, Complex I (nuo-1 gene) and the F1F0 ATP synthase (atp-2 gene) also show a marked abnormality in gonadal development .
Decreased fecundity of sft-1(RNAi) worms was attributed to a failure of both oogenesis and spermatogenesis, as unfertilized oocytes were not observed at a significant level and WT sperm was only able to partially restore the brood size. Consistent with the hypothesis that decreased energy generating capacity in sft-1(RNAi) worms due to COX deficiency may account for the reproductive failure, we observed high levels of COX activity in the germ line of WT animals. sft-1::gfp was not observed in the germline, but a likely explanation for this is the phenomenon of gene silencing that is often triggered by transgene expression in the germline .
Both sft-1(RNAi) and oxa-1(RNAi) worms can be described as having Mit phenotypes (that is, those associated with mitochondrial dysfunction). Knockdown of either gene resulted in a significantly extended lifespan (particularly in the case of oxa-1) in at least two independent experiments, while only the oxa-1(RNAi) worms were resistant to oxidative stress induced by treatment with paraquat. The relationship between mitochondrial function and lifespan in C. elegans is complex, although most observations support two contributory mechanisms, nutritional restriction and damage due to ROS. Many different observations link nutritional/energetic status and lifespan in a wide range of different species. Food deprivation, either environmental or genetically determined (for example, eat-2 gene mutations ) and mutations in genes involved directly (for example, nuo-1) or indirectly (for example, clk-1) in mitochondrial energy metabolism have all been shown to prolong life in C. elegans. These observations are incorporated in the “rate of living model” for longevity in which increased life expectancy is attributed to a generalized slowing down of metabolic function.
In nearly all cases examined thus far, lifespan extension in Mit mutants has been shown to act independently of the insulin/IGF signaling pathway . Therefore, the dependence of sft-1(RNAi) animals on daf-16 for lifespan extension, suggesting that sft-1 may act upstream of daf-16 to regulate longevity, is noteworthy, putting sft-1 knockdown in a different phenotypic category from most other Mit phenotypes. The FoxO-like forkhead/winged helix transcription factor DAF-16 is thought to be a master regulator of aging, integrating metabolic signals, stress signals as well as reproductive signals from the germline to modulate longevity . Perhaps the dependence of lifespan extension in sft-1(RNAi) animals on daf-16 reflects the relative importance of reduced fecundity in promoting longevity when sft-1 expression is reduced.
The second model for the determination of lifespan in C. elegans is based on the accumulation of damage due to ROS, although recent analyses suggest a rather complex relationship between oxidative stress and aging, and one that can be experimentally uncoupled. The oxidative damage model is not independent of the rate of living model as the major site for the formation of ROS is the mitochondrion and production is related to the level of metabolic activity. Some Mit mutants (for example, lrs-2 and isp-1), in common with IGF/insulin signaling pathway mutants daf-2 and age-1, show increased lifespan and resistance to oxidative stress , reviewed in . oxa-1, as presented here, appears to fall into this category. By contrast, gas-1, nuo-1 and mev-1 mutations confer reduced lifespan associated with a hypersensitivity to free radicals [16, 42, 43], reviewed in . Resistance to oxidative stress is suggested to extend lifespan by decreasing damage caused by accumulation of free radicals in the cell.
On the other hand, mutants such as clk-1 and sod-2 display a hypersensitivity to oxidative stress but are also long-lived [18, 22], suggesting that oxidative stress per se is not always a primary cause of aging. Indeed, it has been suggested that ROS generated inappropriately in some mitochondrial mutants might actually extend lifespan by signaling various adaptive responses [12, 44]. This proposed preconditioning of mitochondria has been termed mitohormesis . Intriguingly, though, there do not appear to be mutants that are highly resistant to oxidative stress yet display a shortened lifespan, supporting the idea that oxidative stress resistance (or adaptation to oxidative stress) is associated with a long life.
“Oxidative stress resistance” is rather an umbrella term, as different Mit mutants display different sensitivities to a spectrum of oxidative stresses. For example, in one study of 10 different mutant/RNAi strains, most of the long-lived worms with compromised mitochondria displayed marked resistance to hydrogen peroxide, yet were not resistant (or even displayed increased sensitivity) to paraquat . The authors of this study proposed that as paraquat triggers the production of superoxide in a NADPH-dependent reaction, paraquat hypersensitivity of worms with mitochondrial dysfunction might be due to an increase in NADPH associated with a particular metabolic response to reduced electron transport. However, the data that we present here indicate that oxa-1(RNAi) worms, in contrast to those analyzed in the Lee study, exhibit marked resistance to paraquat. This suggests that it is very difficult to generalize about the effects of mitochondrial dysfunction, even within the group of mutants/RNAi-treated worms that display both prolonged lifespan and heightened stress resistance. Furthermore, a recent study has found that pathways of lifespan extension can differ depending on whether RNAi or mutation for the same gene is used to reduce gene function . Mutations in sft-1 or oxa-1 were not analyzed here, but these would form an interesting subject for future analysis.
It has been previously suggested that long-lived Mit mutants utilize a novel metabolism, and that longevity in these animals may be dependent on this altered metabolic state. For example, it has been proposed, albeit using a limited number of mutants, that long-lived Mit mutants up-regulate fermentative malate dismutation, where fumarate is terminally reduced at complex II to succinate, generating fewer radical species overall . This is an anaerobic metabolic pathway thought to be unique to nematodes, and normally up-regulated during dauer formation. However, other studies argue that although metabolic restructuring does indeed occur in Mit mutants, the restructuring per se does not cause lifespan extension . Recent data, however, challenge this view. For example, the importance of the alternative glyoxylate pathway in Mit mutant lifespan extension has been investigated by knocking down gei-7, which encodes the main glyoxylate shunt enzyme in C. elegans. gei-7 mutation was found to suppress the enhanced longevity of clk-1 mutants  and has also been found to reduce the lifespan of cyc-1 Mit mutants .
Strikingly, mitochondrial respiratory complex dysfunction models being developed in other systems display many of the same features as C. elegans Mit mutants. For example, a partial deficiency of Mclk1, the mouse clk-1 ortholog, increases average lifespan by 30% . Even more relevant to this study, increased lifespan following inactivation of Surf1 in mice has been recently demonstrated , and similarly, Surf1 knockdown in the central nervous system of Drosophila melanogaster has also been shown to induce longevity . This suggests that the function of sft-1 in regulating lifespan is likely to be widely conserved, and it will be very interesting to discover the extent to which precise mechanisms of lifespan extension are conserved between disparate species. For example, it would be interesting to examine whether lifespan extension in sft-1 knockdown animals is dependent on gei-7 and thus a shift to the glyoxylate pathway. It is not clear, however, how such metabolic restructuring might proceed in other organisms where this alternative pathway is not thought to operate.
Whatever the precise mechanisms, it is clear that sft-1 and oxa-1 influence longevity through distinct molecular pathways, despite the fact that both genes encode factors required for COX assembly.