Mice with a homozygous mutation in the exonuclease domain of mitochondria DNA (mtDNA) polymerase gamma (POLG) have been used as a model of mitochondrial dysfunction and aging. These mice possess a mtDNA mutator phenotype, accumulating large numbers of deletions and point mutations in mtDNA (Kujoth et al., 2005; Trifunovic et al., 2004). Surprisingly, these mice do not display signs of elevated ROS generation, but instead exhibit increased apoptosis, a number of age-related phenotypes, and a shortened lifespan (Kujoth et al., 2005; Trifunovic et al., 2004). Subsequent studies revealed that mtDNA deletions that accumulate in the brain and heart and provide the driving force behind the progeroid-like phenotype of the mutator mice (Vermulst et al., 2007; Vermulst et al., 2008). Interestingly, catalase overexpression can attenuate the age-dependent cardiomyopathy observed in POLG mutant mice (Dai et al., 2010), suggesting that oxidative stress may in fact contribute to some of the age-related phenotypes in this mouse model. Taken together, these studies demonstrate the critical but complex connection between mitochondrial function and lifespan.

Studies in model organisms show that mtDNA mutations can both reduce or extend lifespan, depending on severity, context and developmental stage. Surprisingly, complete absence of mtDNA in yeast, the so-called petite mutation, is associated with increased lifespan (Powell et al., 2000). Similarly in worms, RNA interference (RNAi) studies showed that decreasing the expression of mitochondrial genes increased lifespan (Dillin et al., 2002; Lee et al., 2003). Furthermore, timing studies showed that respiration must be decreased during development for the life-extending benefits (Dillin et al., 2002). Rea and colleagues demonstrated the importance of the level of mitochondrial gene expression in this effect (Rea et al., 2007). Moderate inhibition extended lifespan significantly, whereas high levels of RNAi inhibition reduced lifespan. The exact mechanisms that mediate lifespan extension in these models are still not known, but may involve a mitochondrial-driven stress response that is similar to hormesis. This idea is supported by glucose restriction in worms, an intervention that increases lifespan but is associated with increased mitochondrial respiration (Schulz et al., 2007). Remarkably, in this model, increased oxidative stress is required for lifespan benefits, as antioxidant treatment of glucose-restricted worms blocks lifespan extension (Schulz et al., 2007). Along these lines, feeding worms low doses of the ROS-generating compound juglone induces small heat shock protein HSP-16.2 and increases lifespan in a DAF-16-dependent manner. Administration of higher juglone concentrations results in decreased lifespan (Hartwig et al., 2009). Furthermore, in worms, SOD2 deletion and ubiquinone-defective clk-1mutants both show increased lifespan despite elevated ROS generation and oxidative stress (Lapointe and Hekimi, 2008; Van Raamsdonk and Hekimi, 2009). Taken together, these studies demonstrate that mild forms of mitochondrial dysfunction may activate stress response pathways that promote a protective environment, conducive to long life. A greater understanding of the mechanisms involved in mitochondrial stress response pathways might provide new therapeutic opportunities for aging and age-related pathology.

Mitochondrial function is controlled by a number of signaling pathways and transcriptional regulators that sense energetic stress and contribute to lifespan regulation, including peroxisome proliferation-activated receptor coactivator 1 α (PGC-1α), sirtuins, mTOR, and AMPK. When energy is low, these pathways allow the cell to adjust fuel utilization and mitochondrial number. Many aspects of mitochondrial dysfunction associated with aging can be blocked by caloric restriction. (Hunt et al., 2006; Sohal and Weindruch, 1996). For example, caloric restriction partially prevents the age-related decline in mitochondrial gene expression in mouse heart, brain and skeletal muscle (Lee et al., 2002; Lee et al., 1999; Lee et al., 2000). These metabolic effects are mediated by a complex interplay between signaling pathways that converge on the transcriptional co-activator PGC-1 α. PGC-1 α is a central regulator of mitochondrial biogenesis and function that is induced by a variety of metabolic stressors, including low energy availability and oxidative stress (Kelly and Scarpulla, 2004). Studies in mammalian cells and tissues have shown that caloric restriction induces mitochondrial biogenesis through the up-regulation of PGC-α (Lopez-Lluch et al., 2006) and endothelial nitric oxide synthase (eNOS) (Nisoli et al., 2005). Moreover, the severe muscle wasting phenotype of mice lacking a subunit of electron transport complex IV in skeletal muscle can be partially rescued by PGC-1 α overexpression, suggesting that boosting the total number of mitochondria may compensate for a mutation in the electron transport chain (Wenz et al., 2008). Likewise, age-associated sarcopenia and metabolic dysfunction can be rescued by PGC-1α overexpression (Wenz et al., 2009). SIRT1 also activates PGC-1α by deacetylation and promotes mitochondrial biogenesis (Mattagajasingh et al., 2007; Rodgers et al., 2005). Treatment with resveratrol results in increases the lifespan of mice fed a high fat diet (Baur et al., 2006; Feige et al., 2008), and this may be mediated, in part, via activation of PGC-1α by SIRT1. PGC-1α can also be activated by phosphorylation by AMPK, increasing the expression of target genes involved in mitochondrial biogenesis and fatty acid oxidation (Jager et al., 2007; Long et al., 2005). Hence, PGC-1α is a central node of regulation for several signaling pathways that regulate both mitochondrial function and lifespan.

The study of human progeroid syndromes has provided evidence for a link between genotoxic stress responses and signaling pathways that regulate the aging process. This was illustrated by a recent mouse model of a human progeroid syndrome caused by mutation of the XPF-ERCC1 endonuclease, which is involved in nucleotide excision repair. The transcriptional profile of the liver in XPF-ERCC1-deficient mice showed reduced expression of genes in the insulin/IGF-1 signaling pathway, and a shift to anabolic metabolism and increased anti-oxidant defense (Niedernhofer et al., 2006). Downregulation of insulin/IGF-1 signaling is also a feature of other mouse models with genomic instability, such as the SIRT6-knockout and a transgenic mouse with overexpression of a truncated p53 isoform (Maier et al., 2004; Mostoslavsky et al., 2006). An expression profile suggesting reduced insulin/IGF-1 signaling was also observed in cultured fibroblasts from Werner syndrome patients (Kyng and Bohr, 2005). The downregulation of this central signaling pathway may be a mechanism for shifting cellular resources from growth to repair and protection in the setting of genomic instability. Interestingly, caloric restriction or genetic mutations that extend lifespan have a similar effect on insulin/IGF-1 signaling in a variety of model organisms. Hence, genotoxic stress and aging may stimulate similar stress response pathways.

The broad spectrum of genotoxic stress responses and their potential relationship to aging is exemplified by the disorder ataxia telangiectasia (AT), which is caused by autosomal recessive mutations in the PI3 kinase-related kinase ataxia telangiectasia mutated (ATM). AT is characterized by immunodeficiency, skin lesions, pigmentary changes, neurodegeneration and a variety of neoplasms (Martin, 2005). The clinical manifestations of AT suggest that an impaired DNA damage response has different consequences in mitotic versus post-mitotic cells. Loss of efficient ATM-mediated signaling in mitotic cells leads to genomic instability, giving rise to immunodeficiency and a variety of tumors (Lombard et al., 2005). In post-mitotic neurons, however, the phenotype is one of degeneration, with a predilection for the large metabolically active Purkinje cell neurons of the cerebellum.

A recent proteomic analysis of the substrates of the ATM and ATR kinases suggests that DNA damage can activate a broad range of signaling pathways, some of which also regulate aging (Matsuoka et al., 2007). ATM is activated through autophosphorylation following interaction with the MRN complex (mre11-rad50-nbs1) at double strand breaks, whereas ATR is activated through interaction with ATR-interacting protein (ATRIP) at sites of stalled replication forks or single stranded DNA associated with double strand breaks. A proteome-wide analysis identified over 700 protein substrates of ATM and ATR that are phosphorylated in response to ionizing radiation (Matsuoka et al., 2007). In addition to replicating previously known DNA damage response proteins, many new substrates were identified including components of signaling pathways that regulate aging. For example, several proteins in the insulin/IGF-1 signaling pathway were identified, including IRS2, the kinase AKT3, and the transcription factor FOXO1. Multiple components of the protein translation regulatory pathway regulated by TOR signaling were also identified as ATM/ATR substrates, including TSC1, 4E-BP1 and p70S6K (ribosomal protein S6 kinase). Activation of these signaling pathways would be predicted to augment short-term cell survival by preventing apoptosis and increasing macromolecular biosynthesis. Surprisingly, an opposite effect, reduced insulin/IGF signaling, has been described in progeroid syndromes with chronic genomic instability. Hence, the DNA damage response network may have evolved to facilitate repair and survival in the short-term following acute DNA damage. However, repetitive or sustained activation of the DNA damage response during aging may compromise normal tissue homeostasis, leading to apoptosis or cellular senescence.

The notion that chronic activation of the DNA damage response may contribute to the aging process is supported by the phenotypes of different p53 gain of function mouse models. It was originally reported that a mutant p53 allele with enhanced tumor suppressor activity accelerated aging in a variety of different tissues in a mouse model (Tyner et al., 2002). However, accelerated aging was not observed in a different mouse model that overexpressed wild-type p53 under the control of the endogenous promoter (Garcia-Cao et al., 2002). A salient difference between the two models was the constitutive expression of the gain of function p53 allele versus the regulated expression of the allele under the control of the endogenous promoter, which was overexpressed under stress-related conditions, but not in the absence of stress. The constitutive overexpressing mutant mouse showed accelerated aging, whereas the stress-related expressor had a normal lifespan with augmented tumor suppressor activity. Hence, persistent stimulation of the DNA damage response resulting in chronic p53 activation, even at a low level, may be deleterious, potentially leading to apoptosis or cellular senescence (Lombard et al., 2005). Moreover, recent studies suggest that chronic DNA damage and ATM signaling in senescent cells leads to secretion of pro-inflammatory cytokines, possibly through activation of the ATM target NF-κB (Rodier et al., 2009). A persistent DNA damage response may therefore contribute to systemic inflammation, a known contributory factor for many age-related degenerative disorders.

Genetic studies in C. elegans suggest that aging and proteostasis are closely linked and coordinately regulated. Aging worms show impaired activation of heat shock and unfolded protein responses and the accumulation of aggregated proteins (Ben-Zvi et al., 2009; David et al., 2010). Proteostasis in aging worms can be restored by manipulations that also extend lifespan, such as downregulation of insulin/IGF-1 signaling by RNAi or activation of the transcription factors DAF-16 and HSF-1 (David et al., 2010; Hsu et al., 2003; Morley et al., 2002). HSF-1 is a member of a family of transcription factors that act in worms and mammals to transcriptionally activate the proteostasis network. Lifespan extension from reduced insulin/IGF-1 signaling is suppressed by RNAi for HSF-1 or chaperones that are transcriptional targets of HSF-1, suggesting that proteostasis is a major component of lifespan regulation in the worm (Morley and Morimoto, 2004). In mammalian cells, the related transcription factor HSF is subject to feedback inhibition through interaction with shock proteins and by direct acetylation. Deacetylation of HSF by the stress resistance protein SIRT1 potentiates the transactivation of heat shock genes (Westerheide et al., 2009). Proteostasis is also a component or hormesis, in which mild stress results in a long-term increase in stress resistance. In worms, hormetic heat shock in young adult animals extends lifespan through the induction of heat shock proteins (Olsen et al., 2006).

Loss of function in the protein quality control network contributes to the pathogenesis of a group of age-dependent disorders known as conformational diseases that are exemplified by Alzheimer’s and Huntington’s disease. A characteristic feature is age-dependent accumulation of protein aggregates in the brain that may contribute to neurodegeneration and neurological decline. Huntington’s disease (HD) is the prototype of a group of human genetic diseases caused by polyglutamine-containing proteins that aggregate, leading to neurodegeneration. The length of the polyglutamine tract is inversely related to the age of onset of disease, suggesting a critical interplay between aging, proteostasis, and neurodegenerative pathology. Moreover, signaling pathways that regulate aging also control polyglutamine-related pathology in C. elegans models; loss of function mutants in the insulin/IGF-1 regulated transcription factors DAF-16 and HSF-1 reduce lifespan and accelerate polyglutamine aggregation and toxicity (Hsu et al., 2003). Conversely, the AGE-1 mutation that extends lifespan in worms reduces polyglutamine aggregation and toxicity (Morley et al., 2002). Conformational diseases also illustrate the fragility of the proteostasis network during aging. Misfolding and aggregation of a single metastable protein, such as mutant huntingtin, can impair the folding of other proteins (David et al., 2010; Gidalevitz et al., 2006) and globally inhibit the ubiquitin proteosome system (Bennett et al., 2007).

Mouse studies demonstrate the central importance of autophagy in mammalian aging and age-related neurodegeneration. Mice deficient for autophagy-related genes (ATGs) exhibit reduced autophagy, high levels of protein inclusion bodies, and neurodegenerative pathology (Hara et al., 2006; Komatsu et al., 2006). These phenotypes are also common features of human neurodegenerative diseases associated with protein aggregation, such as Alzheimer’s disease and Huntington’s disease (Bishop et al., 2010; Cuervo et al., 2010), suggesting a close relationship between the regulation of autophagy, aging and neurodegeneration. Moreover, blocking mTOR in a mouse model of AD decreases Aβ levels and improves cognitive function, and these effects may be due, in part, to increased autophagy (Spilman et al., 2010). Thus, the lifespan extension observed in mammals treated with mTOR inhibitor rapamycin (Harrison et al., 2009) may also be mediated, in part, by the up-regulation of autophagy. Taken together, these studies suggest a role for autophagy in the coordinated regulation of aging and age-related stress responses, as well as the pathogenesis of neurodegenerative disorders.

Activation of autophagy is fundamentally linked to the nutrient status of the cell and impacts metabolism and stress response pathways. Autophagy increases during times of nutrient deprivation, such as in fasting or calorie restriction. This is likely to provide cells with an additional source of energy during times of nutrient scarcity; generating free amino acids can be utilized for protein synthesis. Autophagy is also important for mobilizing lipids in the liver during nutrient deprivation (Singh et al., 2009a), which can be metabolized as an alternate energy source, or used for new membrane biosynthesis. Strikingly, overexpression of LAMP-2A in mouse liver in the late stages of adult life prevents an age-related decline in autophagy, reducing the associated accumulation of damaged proteins and leading to improved liver function (Zhang and Cuervo, 2008).

The mammalian sirtuin SIRT1 stimulates rapamycin- and starvation-stimulated autophagy in cultured cells via deacetylation and activation of autophagy proteins Atg5, Atg7 and Atg8 (Lee et al., 2008), suggesting that at least some of the effects of SIRT1 on metabolism and stress responses could be due to regulation of this clearance pathway. SIRT1 was also required for the CR-mediated increase in mitochondrial autophagy in mouse kidney observed during hypoxia (Kume et al., 2010), demonstrating multiple levels by which protein turnover can be regulated by this sirtuin. Hence, autophagy may be a major component of age-dependent stress response pathways that are activated by central regulators such as SIRT1 and mTOR.

The necessity for post-mitotic neurons to survive an entire lifespan imposes a major challenge for protein quality control, in particular autophagic clearance of damaged or aggregated proteins. Age-related diseases of the nervous system are characterized by the accumulation of aggregated proteins in affected neurons and glial cells that may contribute to dysfunction and neurodegeneration. Autophagy can become impaired in the aging brain at a number of levels. The initial formation of the autophagosome is mediated by the Class III PI3 kinase complex composed of beclin-1, Vps15 and Vps34. Beclin-1 levels decline in the normal aging brain (Shibata et al., 2006) and fall further in early AD, potentially reducing autophagy and predisposing to Aβ aggregation (Pickford et al., 2008). Autophagosome formation may also be impaired at the level of cargo recognition. This has been observed in models of Huntington’s disease, resulting in impaired clearance of aggregated proteins despite normal autophagosome formation (Martinez-Vicente et al., 2010). There is also evidence of impaired autophagosome clearance in neurodegenerative disorders, resulting in the accumulation of autophagosomes that are not efficiently processed further (Wong and Cuervo, 2010). This may relate, in part, to disruption of microtubular or actin cytoskeletal elements that are required for autophagosome transport and fusion. Autophagosome accumulation may contribute directly to neurodegeneration by serving as a source of cytotoxic proteins, such as the 42 amino acid form of Aβ, or by disrupting cellular trafficking (Yu et al., 2005). Presenilin-1 mutations that cause familial AD have recently been shown to compromise both delivery to the lysosome and acidification of lysosomes, compromising the clearance of autophagic substrates at several levels (Lee et al., 2010). Deficits in autophagy can also occur in specific subcellular compartments. Two genes associated with familial Parkinson’s disease, PINK1 and parkin, may function together to promote autophagy of failing mitochondria, a process known as mitophagy (Geisler et al., 2010; Narendra et al., 2008). Disease-causing mutations in PINK1 and parkin impair this process, potentially leading to the persistence of dysfunctional mitochondria that generate high levels of reactive oxygen species and increase oxidative stress. These observations provide evidence for a broad range of mechanisms leading to impaired autophagy in age-related neurodegenerative disorders.