Organismal lifespan can be extended by genetic manipulation of cellular processes such as histone acetylation, the insulin/IGF‑1 (insulin‑like growth factor 1) pathway or the p53 system. Longevity‑promoting regimens, including caloric restriction and inhibition of TOR with rapamycin, resveratrol or the natural polyamine spermidine, have been associated with autophagy (a cytoprotective self‑digestive process) and in some cases were reported to require autophagy for their effects. We summarize recent developments that outline these links and hypothesize that clearing cellular damage by autophagy is a common denominator of many lifespan‑ extending manipulations.

The most prominent and best-studied form of autophagy is macroautophagy (hitherto referred to as ‘autophagy’), where portions of the cytoplasm are sequestered within autophagosomes and then targeted to lysosomes for digestion1 . Autophagy has often been regarded as a selfdestructive process, contributing to ‘autophagic cell death’. However, most data indicate that autophagy rather serves a cytoprotective function. When dying cells display autophagic vacuolization most die with features of (rather than by) autophagy, which has been activated as an unsuccessful homeostatic counter-reaction against cell death2 . For example, during metabolic stress triggered by nutrient depletion and hypoxia, autophagy improves cellular survival, and hence enhances cellular fitness, whereas its inhibition precipitates bioenergetic failure and cell death3 . Suppression of autophagy by knockout or knockdown of essential Atg (autophagy-related) genes often leads to apoptotic or necrotic demise of cells that would otherwise survive in conditions of elevated stress4–7. Conversely, induction of autophagy can enhance resistance of cells to distinct kinds of stress. This can even be observed at the organismal level. Fasting induces autophagy in most somatic cells8 of an organism. It also leads to an increased tolerance of flies against anoxia and of mice against the lethal side-effects of highdose chemotherapies9,10. Intriguingly, caloric restriction (while maintaining adequate nutrition) induces autophagy in non-mammalian model organisms and it is the only food-based manipulation that extends lifespan in all species tested so far. Recent data indicate that the specific induction of autophagy can increase longevity in multiple animal species, as will be discussed in this article.

 Thus, sirtuin 1 may be a major regulator of both autophagy and longevity, which seem to be interrelated. Further implicating autophagy in lifespan regulation, C. elegans with a loss of function mutation in the insulin-like signalling pathway (daf‑2) display extended lifespan and induction of autophagy24,25. The importance of this finding was later underscored by evidence that autophagy is required for caloric restriction effects on lifespan in C. elegans26,27. Simultaneous induction of lifespan extension and autophagy is also observed in C. elegans following knockdown of the p53 orthologue, CEP-1, and both effects are abolished by depleting Beclin 1/ATG628. Further examples of longevity-promoting genetic manipulations that depend on autophagy are listed in Table 1. Consistent with the idea that autophagy participates in longevity-extending processes, a screen for genes regulating clonogenic cell death during the chronological ageing process of yeast revealed that deletion of most autophagy-related genes enhances cell death and induces premature ageing19,29. As a caveat, it should be noted that in many cases autophagy has been causally linked to longevity using experimental evidence involving the deletion or depletion of essential regulators of autophagy, such as Beclin 1 (ref. 30),that may participate in diverse cellular functions. Such manipulations per se are sufficient to disturb cellular homeostasis, and thereby reduce cell survival and lifespan. Hence, even models where a treatment that prolongs the life of wild-type subjects but fails to do so in autophagy-defective subjects may not be the ultimate proof that longevity promotion depends solely on autophagy induction. 

A diet enriched in physiologically relevant polyamines (that is, putrescine, spermidine and spermine) also increases lifespan and health span in mice, although the requirement for autophagy in this case has not yet been investigated45. In yeast, spermidine causes global hypo-acetylation of histone H3, but selective histone acetylation at the promoter region of the ATG7 gene19. Thus, transcription of ATG7 is enabled in the context of general gene silencing resulting in a relative upregulation of ATG7 (and also of other ATG genes). Spermidine-mediated transcription of autophagy-relevant genes may account for the observed induction of autophagy. Spermidine was shown to inhibit histone acetyltransferases in cell-free assays. In support of a role for histone acetyltransferases in the lifespan of yeast, deletion of essential subunits in two histone acetyltransferases increased lifespan, and although treatment of this strain with polyamines further increased longevity, it was not to the same extent as that seen in wild-type cells19. However, by comparison to sirtuin 1, which affects autophagy through deacetylation of essential autophagy-relevant proteins21, spermidine may induce autophagy through a combination of nuclear (transcriptional) and cytoplasmic (transcription-independent) mechanisms (Fig. 1). An exciting speculation is that inhibition of protein acetyltransferases (by spermidine) and activation of deacetylases (by resveratrol) might converge on similar autophagy regulators or effectors. However, the precise identity of these targets remains elusive and should thus constitute a focus of future research. in yeast, rapamycin extends lifespan only under conditions in which autophagy can be induced28,33,35.

 Rapamycin-induced autophagy is independent of sirtuin 1 in human cells and in C. elegans23, suggesting that rapamycin and sirtuin 1 promote autophagy through distinct, non-overlapping mechanisms (Fig. 1). In mice, whether rapamycin prolongs longevity by autophagy remains to be determined, as autophagy has not been monitored or blocked in these experiments34. Thus, it is possible that rapamycin-fed mice display lifespan extensions through anti-inflammatory functions that are beneficial under laboratory conditions. Promoting those functions might become deleterious in the wild. Resveratrol, a polyphenol present in red wine and an indirect activator of sirtuin 1, prolongs the lifespan of yeast36, worms and flies37. In mice, resveratrol extends lifespan only if animals are kept on a high-calorie diet (and hence become obese)38. Resveratrol is also a potent inducer of canonical and non-canonical autophagy39. It only prolongs C. elegans lifespan if the animals express a functional sirtuin 1 gene23. Knockdown of Beclin 1/ATG6 abolished the beneficial effects of resveratrol on longevity22. Taken together, the data indicate that resveratrol increases lifespan of C. elegans through sirtuin 1-dependent induction of autophagy. However, it is noteworthy that sirtuin 1 overexpression or resveratrol fails to extend the lifespan of mice on a normal diet. The ability of sirtuins to increase lifespan in lower eukaryotes is controversial, as in yeast they can also shorten chronological lifespan40. 

Inhibition of autophagy through brain-specific knockout of essential autophagy genes (such as atg5 and atg7) in mice leads to the accumulation of neuronal inclusion bodies containing polyubiquitylated proteins. The appearance of these inclusion bodies correlates with premature neurodegeneration47,49,50, though it is unclear whether the inclusion bodies are protective or promote the disease. Autophagy thus seems to reduce the accumulation of potentially toxic aggregates and, if proteotoxicity partly accounts for the ‘normal’ (non-diseaseassociated) ageing process, autophagy could mitigate the effects of ageing through this effect. Damaged mitochondria containing mutated mitochondrial DNA accumulate during ageing and are a production site of harmful reactive oxygen species; autophagy targets mitochondria that have lost their transmembrane potential4,51, which may facilitate the removal of such damaged mitochondria. Autophagy directly facilitates the destruction of intracellular infectious pathogens, a process referred to as ‘xenophagy’, wherein autophagy invigorates the innate immune response against microbial agents1 . Autophagy is also required at multiple levels for the normal function of the immune system and for avoiding unwarranted inflammatory reactions. For example, pre-apoptotic autophagy in antigen donor cells — either virus-infected or cancer cells — is necessary for the optimal transfer of antigens to dendritic cells and/or their subsequent presentation to T lymphocytes52,53.

 Autophagy is also indispensable for dying cells to emit the ‘find-me’ signal, lysophosphatidylcholine (a chemo-attractant for macrophages), as well as to efficiently expose the apoptotic ‘eat-me’ signal, phosphatidylserine, on the plasma membrane surface (which facilitates the engulfment of dying cells). Insufficient autophagy thus reduces the efficacy of corpse removal: this causes cells that would otherwise be engulfed during apoptosis to undergo secondary necrosis1,54,55, which subsequently triggers inflammatory responses. In this context, it seems particularly intriguing that spermidine was first identified by its ability to alleviate programmed, age-associated necrosis in yeast cells, as well as in human peripheral blood mononuclear cells19. Whether this ‘anti-necrotic’ action contributes to the anti-ageing effects of autophagy inducers remains an open possibility. Autophagy can prevent apoptotic and necrotic cell death in multiple cellular models; we have yet to understand whether this cytoprotection is linked to autophagy-dependent longevity. The inhibition of apoptotic or necrotic cell death can extend longevity in various lower eukaryotes. For example, deletion of the sole caspase-like protease prolongs lifespan of both Podospora anserina and S.cerevisiae56, 57. However, whether apoptosis is a pro- or anti-ageing mechanism in higher organisms is a subject of considerable debate. The Bax protein is a central regulator of apoptosis that causes mitochondrial outer membrane permeabilization, a likely ‘point of no return’ in apoptosis execution. 

Age-induced increase in Bax levels could contribute to ageassociated atrophy in skeletal and heart muscle of rats58,59. Also, an intriguing link between anti-ageing mechanisms and apoptosis inhibition may connect Bax activation to Sirtuins. Sirtuin 1 deacetylates the DNA repair factor Ku70 causing dissociation of Bax from mitochondria, thus inhibiting apoptosis16. However, the effects of autophagy on Bax have not been studied in detail. During ageing, a growing number of haematopoietic (including immune) cells lose their capacity to proliferate and function properly due to senescence. Conditional deletion of the mTOR negative regulator, Tsc1, from haematopoietic stem cells accelerates senescence and thus mimics the phenotype of aged mice. Conversely, in old mice mTOR inhibition (for example, by rapamycin) can restore the selfrenewal of haematopoietic stem cells, as well as their immunocompetence, as evidenced by restored haematopoiesis and effective responses against potentially lethal infection with influenza virus60,61. In vitro, rapamycin administration prompts cell cycle re-entry of human and rodent arrested cells62. Whether this anti-senescence action of rapamycin is mediated by autophagy remains to be investigated. Moreover, it should be noted that studies performed in cancer cells suggest that autophagy is required for the acquisition of the senescent phenotype63,64. Senescence and programmed cell death, both of which can be modulated by autophagy, constitute the major endogenous barriers against oncogenesis. Depletion of several autophagy-relevant genes (such as Beclin 1 and its interacting partners, UVRAG and Bmf1) enhance carcinogenesis1,65. 

 For instance, brain-specific overexpression of Atg8 in Drosophila extends longevity by 50%68. Unless such tissue-specific expression of Atg8 can stimulate autophagy all over the body — perhaps through neuroendocrine mechanisms — these results suggest that autophagy processes within the central nervous system are particularly important for determining longevity. Similarly, improving protein homeostasis in neurons of C. elegans through expression of a heat shock factor (HSF-1) is sufficient to promote longevity69 and in mice deletion of the huntingtin polyglutamine stretch enhances both neuronal autophagy and longevity70. Thus, it remains to be determined whether longevity-extending drugs should be designed as agents that mimic the natural pattern of autophagy induction by fasting (affecting most somatic cells) or that preferentially induce autophagy in the central nervous system. Future investigations will settle the question of whether such drugs could be effective or whether fasting should be promoted as the sole strategy for increasing health and lifespan. Regular periods of fasting could become a burgeoning trend of modern civilization, perhaps resetting our eating behaviour to our ancient roots, to times when nature, not culture, forced us to eat less.