Mitochondrial dysfunction and longevity in animals: Untangling the knot

See allHide authors and affiliations

Science  04 Dec 2015:
Vol. 350, Issue 6265, pp. 1204-1207
DOI: 10.1126/science.aac4357


Mitochondria generate adenosine 5′-triphosphate (ATP) and are a source of potentially toxic reactive oxygen species (ROS). It has been suggested that the gradual mitochondrial dysfunction that is observed to accompany aging could in fact be causal to the aging process. Here we review findings that suggest that age-dependent mitochondrial dysfunction is not sufficient to limit life span. Furthermore, mitochondrial ROS are not always deleterious and can even stimulate pro-longevity pathways. Thus, mitochondrial dysfunction plays a complex role in regulating longevity.

The primary and most essential function of mitochondria is to produce energy for the cell. The oldest explanation for aging, the rate-of-living theory, postulates that aging and life span are regulated by the rate of energy metabolism, with lower rates leading to longer life spans. However, the appeal of the rate-of-living theory has been weakened by its failure to accurately predict the observed relationships between energy expenditure and life span. This is not to say that mitochondrial and energy metabolism don’t play a crucial role in aging, but their relationship to aging might not be simple. Mitochondria do much more than produce energy. Particularly relevant to aging, the mitochondrial electron transport chain (ETC) leaks electrons and generates reactive oxygen species (ROS) during normal respiration. Thus, a potentially harmful elevation of ROS production occurs when the ETC function is perturbed. The mitochondrial free-radical theory of aging posits that biological aging results from the production of ROS and the ensuing damage. However, direct manipulation of cellular ROS levels within the biologically meaningful range does not accelerate aging or decrease life span. Here we review the relationships between normal mitochondrial function, mitochondrial dysfunction, ROS generation, and life span.

Deleterious mitochondrial dysfunction

Numerous studies have described damage to mitochondria in aged cells and organisms, including in human samples. This damage includes a gradual decline in respiratory chain capacity, decreased activities of individual ETC complexes, elevated oxidative damage, decreased mitochondrial content, morphological abnormalities in mitochondrial structure, and increased fragility of aged mitochondria during experimental isolation (Fig. 1) (1). In exploring the implications of these observations for the aging process, a key question is whether the observed damage and dysfunction are severe enough to cause the other degenerative phenotypes of aging.

Fig. 1 Age-dependent gradual mitochondrial dysfunction.

Various mitochondrial defects are found to accompany aging. However, their role in causing aging is unclear.

There is no doubt that mitochondrial dysfunction can severely damage the organism. Human patients with mutations in mitochondrial DNA (mtDNA) or in nuclear genes coding for proteins that function in the mitochondrial ETC are generally severely affected. They often show multisystem disorders that include myopathy, encephalopathy, stroke, and hearing loss (2). Most mitochondrial disorders present with neurological and muscular symptoms. It is thus generally postulated that cells with high energy demands, such as those in the central nervous system and muscles, are more susceptible to the reduced energy output of defective mitochondria and are consequently more strongly affected by mitochondrial impairment. There is, however, considerable clinical variability among mitochondrial disease patients, and some mutations only affect particular tissues, reflecting a diversity of distinct disease mechanisms that are still poorly understood. To understand these conditions, a variety of mouse knockout (KO) models have been developed for nuclear-encoded mitochondrial proteins (3). These include mutants carrying KO mutations in genes that are required for the assembly and function of ETC complexes, mutants with defects in the production of mobile electron carriers [cytochrome c and ubiquinone (UQ)], and mutants lacking necessary factors for the maintenance of mitochondrial dynamics or the integrity of mtDNA. In virtually every case, complete germline KO causes embryonic to perinatal lethality. Tissue-specific conditional KOs, mostly targeted to neurons or muscles, result in abnormal mitochondria with severe deficits in respiratory chain function, giving rise to a variety of disease phenotypes. Most show severe progressive loss of tissue function, such as progressive skeletal muscle weakening, movement impairment, and neurobehavioral abnormalities. All result in death within the first year of life, with life spans reduced to less than 40% of the normal (3).

Much research on why mitochondrial dysfunction gradually develops with time has focused on mtDNA. Mutations and deletions in mtDNA increase with age, and clonally expanded mtDNA damage is more abundant in those areas of aged tissue that also show mitochondrial ETC dysfunction (4). These findings and many earlier studies led to the notion that continuous accumulation of mtDNA damage may play a causal role in the aging process. One particular model that has been used to study this is the “mutator” mouse. In these mice, the proofreading function of the mtDNA polymerase gamma (Polg) is defective, which leads to the accumulation of random mutations and deletions in mtDNA (5, 6). Decreased life span and an array of phenotypes reminiscent of normal aging have been observed in homozygous mutator mice (PolgD257A/D257A). The mice exhibit decreased content and activity of ETC complexes, lower respiratory chain capacity, and lower ATP (adenosine 5′-triphosphate) levels, as well as an activation of apoptotic pathways (5, 6). However, most studies of the mutator mice have reported negligible increases in oxidative stress, which often accompanies disruption of ETC function in the KO mouse models described above (6). This is of interest because oxidative stress, whether causative of aging or not, has been commonly regarded as a reliable biomarker of aging. In fact, it has often been suggested that ROS could be responsible for aging by acting as mutagens on mtDNA in somatic cells, inducing a vicious cycle in which the mtDNA mutations lead to defective ETC function, thereby producing even more mtDNA-damaging ROS. It is thus noteworthy that a recent study using a mitochondria-targeted mass spectrometry probe detected an increase in hydrogen peroxide in the mutator mice close to the end of their life span (7). However, no increase was found in young mutator mice, despite their already elevated level of mtDNA mutations, nor was any increase found in old wild-type mice by this technique. Thus, although high mitochondrial ROS may contribute to the shorter life span of mutator mice, a hypothesis that is supported by the beneficial effects of some antioxidant interventions (8), its role in normal aging is probably not of great importance. In Drosophila and human brain tissue, age-related increases in mtDNA mutations are not caused by oxidative stress (9, 10), further weakening the vicious-cycle hypothesis of mitochondrial aging.

Beyond this, the quantitative findings with the mutator mouse argue against the notion that damage to mtDNA causes aging. In the homozygous mutator mice, life span is shortened, but the mtDNA mutation load is much higher than that detected in aged animals or elderly humans (11). Heterozygous mutator mice are born with a mutation burden 30 times higher than that of aged wild-type mice, yet they lack overt phenotypes and have a normal life span (11). This calls into question whether the naturally occurring slow accumulation of age-related mtDNA mutations has a leading role in causing aging, rather than representing only one of the types of damage accumulation that accompany aging. A different mouse strain, the mtDNA deletor mouse, is also relevant in this context. These mice accumulate large-scale mtDNA deletions in postmitotic tissues but do not have a shortened life span, although they exhibit late-onset respiratory dysfunction in a subset of tissues (12). These findings further undermine the notion that damage to mtDNA or mitochondrial dysfunction is sufficient to accelerate aging.

Uncoupling mitochondrial dysfunction and aging

Not all partial losses of mitochondrial function result in shortened life span, and some can even result in increased life span (13). UQ is an obligate electron carrier in the ETC, and MCLK1 (also known as COQ7) is the penultimate enzyme in the mitochondrial UQ biosynthetic pathway. Mice missing one copy of Mclk1 appear superficially normal and live longer than their wild-type littermates, despite markedly reduced mitochondrial respiration. Overall UQ concentrations in whole mitochondria extracts are normal in these heterozygous mice, but they are lower in the inner membrane fraction. This causes a decrease in respiratory chain capacity, which in turn results in low ATP generation. Production of mitochondrial ROS (mtROS) appears to be increased in the mutant, whereas overall ROS levels are not. The extended longevity of these mutants is also associated with slow development of the biomarkers of aging, high macrophage expression of HIF-1α (hypoxia-inducible factor 1α), and an enhanced immune response (13). However, it is not known whether these phenotypes are necessary or sufficient for the observed increase in longevity.

SURF1 is an inner mitochondrial membrane protein required for the assembly of complex IV (cytochrome c oxidase), a protein complex needed for oxidative phosphorylation. A knockout model of Surf1, in which a premature stop codon was inserted into exon 7, resulted in viable mice with increased life span (14). These mice exhibited the anticipated decrease in complex IV activity, which was 30 to 50% of normal. Mitochondrial respiration was mildly affected in some tissues, but no change in mtROS production was detected. Other phenotypic features include lower fat mass, elevated protein expression of the mitochondrial biogenesis regulator PGC-1α (peroxisome proliferator–activated receptor gamma coactivator 1α) (Box 1), increased insulin sensitivity, resistance to calcium-induced neuronal death, and an increase in the expression of some of the proteins involved in the mitochondrial unfolded protein response (UPRmt) (1416). There are other indications that the UPRmt could participate in life-span determination in mammals: Conserved longevity-promoting interventions, such as overexpression of SIRT1 and rapamycin and resveratrol treatments, induce the UPRmt in mammalian cells (17, 18). However, it remains unclear how the loss of SURF1 extends life span.

Box 1

Benefits of preserving or boosting mitochondrial function.

PGC-1 is a transcription coactivator of mitochondrial biogenesis, oxidative metabolism, and ROS-handling enzymes. Overexpression of PGC-1α has been used to preserve or boost mitochondrial function. In mice, increased muscle PGC-1α leads to preservation of muscle function during aging (40). In fruit flies, overexpression of PGC-1α in intestinal stem and progenitor cells leads to a longer life span (41). The mechanisms underpinning these effects are uncertain.

NAD+ (nicotinamide adenine dinucleotide) is a coenzyme for many reactions in oxidative phosphorylation and the tricarboxylic acid cycle. NAD+ levels decline with age. Boosting NAD+ levels has been shown to increase life span in C. elegans and improve health indicators in old mice (17, 42). The effects are mediated by sirtuins and are associated with more and healthier mitochondria.

SOD2 is a mitochondrial matrix superoxide dismutase that serves as a first line of defense against oxidative stress by converting superoxide to hydrogen peroxide. Sod2 heterozygous (Sod2+/−) mice show increased oxidative stress, as indicated by inactivation of ROS-sensitive enzymes, higher sensitivity to oxidative stress, impaired mitochondrial respiration, and higher levels of DNA oxidative damage in both the nucleus and mitochondria (19). Despite this, Sod2+/− mice appear normal and have a wild-type life span. Thus, mitochondrial damage induced by a decrease in antioxidant defenses is not sufficient to compromise longevity in mice.

It is conceivable that impaired mitochondrial function does not shorten life span until it reaches a critical threshold. This argument provides a potential rationale for why, despite the prime importance of mitochondria for many cellular functions, some mitochondrial defects, such as those mentioned above, are not associated with a shortened life span. However, even mice that have suffered severe and prolonged mitochondrial dysfunction are able to live as long as control mice when mitochondrial function is partially restored at mid-life (20). Mutant mice in which the Mclk1 gene was globally deleted in 2-month-old adults (aogMclk1 KO) showed a severe loss of UQ and impaired mitochondrial respiration. Heart, kidney, and skeletal muscles had only 50% of the normal respiratory rate. The mutant mice died at around 9 months of age with severe phenotypes, including small size, absence of fat, hair loss, low blood glucose, elevated lactate, low triglycerides, intervals of catatonia, and severe neurological symptoms (20). UQ biosynthesis can be restored in the absence of the MCLK1-catalyzed step by treatment with an appropriate unnatural biosynthetic precursor, 2,4-dihydroxybenzoate (2,4-DHB) (21). Treatment of KO mice with 2,4-DHB shortly before death led to virtually full phenotypic recovery, with animals looking essentially wild-type (except for a small deficit in weight), despite only partial normalization of mitochondrial function. The treatment resulted in full restoration of a normal life span, even though the mice had lived almost their entire lives with mitochondrial dysfunction (20). Thus, neither chronic nor acutely severe mitochondrial dysfunction is sufficient to produce irreversible phenotypes that limit life span.

Insights from invertebrate studies

Several of the studies with vertebrates described above fail to support a causal role for mitochondrial dysfunction in the aging process. Moreover, recent work with invertebrate animal model systems suggests that mitochondrial dysfunction in fact can lead to the generation of intracellular signals that stimulate anti-aging processes. It is widely assumed that the mechanisms of aging are conserved and can be studied in model organisms, including the nematode Caenorhabditis elegans and the fruit fly Drosophila melanogaster. In C. elegans, clk-1 is the ortholog of the mouse Mclk1 gene discussed above. Twenty years ago, the clk-1 mutant was the first long-lived mutant to be described in which increased longevity was associated with mitochondrial dysfunction (22). Several more long-lived C. elegans mutants were subsequently found to be associated with mitochondrial dysfunction (23, 24). In addition, it was found that using RNA interference (RNAi) to knock down C. elegans genes whose products function in mitochondria frequently resulted in increased life spans (25, 26), a phenomenon that also appears to be conserved in Drosophila (27) and possibly mice (18). At least some of the mutations appear to increase life span by mechanisms that are distinct from those that underlie the effects of RNAi (23). However, whether distinct mechanisms are at work, and even which mechanisms are engaged by RNAi knockdowns to induce longevity, remain controversial.

Long-lived electron transport chain mutants

A point mutation in the C. elegans isp-1 gene, which encodes the iron sulfur protein of mitochondrial respiratory complex III, leads to mitochondrial dysfunction but also to a dramatically increased life span (24). The mitochondrial dysfunction in the isp-1 and similar long-lived ETC mutants, such as the nuo-6 mutant, elevates mitochondrial superoxide generation (28). This elevation appears to be necessary for extended longevity, which is suppressed by treatment with antioxidants. A pro-longevity role for superoxide generation is further supported by the observation that treatment with very low concentrations (0.1 mM) of the mitochondrial superoxide generator paraquat can dramatically increase the life span of wild-type animals without impairing their mitochondrial function, but this treatment has no effect on the long-lived ETC mutants (28). This supports the notion that increased mtROS generation with aging does not mean that ROS cause aging, but rather that mtROS are part of a stress response that combats the damage accumulation that accompanies aging (Box 2) (29).

Box 2

ROS in other mechanisms of life-span extension.

ROS acting as signals have been linked to the longevity resulting from disturbed insulin and insulin-like growth factor 1 signaling in C. elegans (43). Mitochondria also have frequently been implicated in the longevity associated with caloric restriction (CR). In C. elegans, a CR regime imposed by inhibition of glycolysis via 2-deoxy-d-glucose promotes ROS formation and subsequent longevity (44). However, mtROS generated by paraquat can further extend the life span of the eating-defective eat-2 mutant, suggesting that the mechanism of CR is distinct from the life-span extension initiated by mtROS (28). The C. elegans transcription factor SKN-1 (homolog of vertebrate NRF1/2), which is crucial to the response to oxidative stress, also has recently been implicated in linking mitochondrial function to CR (45).

“There is no doubt that mitochondria wear down with age. However, by itself, this functional decline appears to be insufficient to cause aging.”

Recent findings suggest that one of the principal mechanisms by which mtROS act as prolongevity signaling molecules is via the intrinsic apoptosis signaling pathway, by activating a specific pattern of changes in gene expression without inducing cell death (Fig. 2A) (30). The intrinsic pathway of apoptosis is physically associated with mitochondria and is sensitive to mtROS in vertebrates, where it contributes to homeostasis by eliminating unwanted or dysfunctional cells. In C. elegans, this signaling pathway can be used in two ways: either to stimulate apoptosis when it is triggered by expression of EGL-1 or to stimulate greater survival when it is triggered by expression of CED-13 and mtROS. The mtROS presumably act on CED-9 (BCL2-like), which is tethered to the mitochondrial outer membrane, or CED-4 (APAF1-like), which is bound by CED-9 (Fig. 2A). The expression of the ced-13 gene is known to be controlled by the p53 C. elegans homolog CEP-1, which also affects mtROS pro-longevity signaling (31, 32). Studies in yeast have suggested that mtROS can also affect longevity by triggering a noncanonical activation of DNA damage pathways (33).

Fig. 2 Pro-longevity responses to mitochondrial stress.

(A) In C. elegans, pro-longevity signaling by mtROS acts through the intrinsic apoptosis pathway. This is further modulated by key stress-response pathways. (B) The UPRmt has been suggested to link mitochondrial stress to life-span extension, but it is still uncertain whether its activation alone is sufficient to extend life span.

Altered energy metabolism in the long-lived ETC mutants (low oxygen consumption and ATP levels) might also play a role in their longevity, because mtROS in these mutants alters ATP-dependent behaviors and developmental processes, possibly by redirecting ATP use toward protective processes (30). This is consistent with findings that the metabolic regulators adenosine 5′-monophosphate (AMP) kinase and HIF-1α modulate the effects of the mtROS pathway (34, 35).

Activating the UPRmt

The UPRmt allows cells to cope with the presence of unfolded or misfolded proteins in mitochondria by conveying a stress signal to the nucleus and up-regulating mitochondrial chaperones and proteases (36). It has been proposed that UPRmt activation promotes longevity and is responsible for the life-span extension induced by resveratrol and rapamycin treatments (17), as well as for the life-span extension that follows the mitochondrial dysfunction induced by RNAi knockdown of ETC components (23, 37) or components of the mitochondrial translation machinery (Fig. 2B) (18). However, the mechanisms induced by RNAi and by the long-lived ETC point mutations appear to be fully distinct, in particular because their effects on life span are additive (23).

Nevertheless, the question of whether activation of the UPRmt is sufficient for life-span extension remains open. Activation of the UPRmt can be uncoupled from life-span extension in C. elegans (38). For example, gain-of-function alleles of atfs-1, which encodes the nuclear transcription factor that turns on the UPRmt by sensing mitochondrial stress, do not lengthen life span. Furthermore, loss-of-function atfs-1 alleles do not always prevent the longevity induced by RNAi knockdown of ETC subunits.

Even though the ATFS-1–dependent UPRmt and other mechanisms of mitochondrial protein homeostasis (39) might not confer longevity by themselves, they might be needed to permit life-span extension in long-lived mutants with stressed mitochondria, such as clk-1 and isp-1 mutants (37, 39). The extended life spans of the mutants were abolished when the activation of these pathways, which shield mitochondria from the consequences of dysfunction, was prevented, because this intervention resulted in very severe synthetic phenotypes.

There are thus two distinct mechanisms of life-span extension by dysfunctional mitochondria: one mechanism induced by point mutations in ETC subunits, which increases mtROS generation and engages the apoptotic pathway, and one mechanism induced by RNAi knockdown of components of the ETC and of the mitochondrial translation machinery. The role of mitochondrial protein homeostasis in either mechanism is less clear, but it might be protective and thus facilitate life-span extensions induced by mitochondrial stress (Fig. 2B).


Studies in both vertebrates and invertebrates demonstrate the intimate connection between mitochondria and longevity. On the one hand, there is no doubt that mitochondria wear down with age. However, by itself, this functional decline appears to be insufficient to cause aging. On the other hand, some deviations from normal mitochondrial states can elicit responses that are protective and extend longevity. These findings point to unexpectedly complex links between mitochondria and longevity.


  1. ACKNOWLEDGMENTS: We thank C. Yee for useful discussions and R. Branicky for reading the manuscript. Our laboratory is funded by grants from the Canadian Institutes of Health Research (grants MOP-114891, MOP-123295, and MOP-97869) and by McGill University. S.H. holds the Strathcona Chair of Zoology and the Campbell Chair in Developmental Biology.
View Abstract

Stay Connected to Science

Navigate This Article