107 - 127: Modulating mTOR in Aging and Health Open Access

The mechanistic target of rapamycin, mTOR, is a highly conserved serine/threonine kinase that plays a central role in sensing and responding to nutrient availability and growth signaling in eukaryotes. mTOR, encoded in mammals by MTOR, is an essential component of two distinct multiprotein complexes, mTORC1 and mTORC2. These signaling complexes regulate a variety of basic cellular activities including growth rate, cell size, and metabolism, and act as critical signaling hubs at the interface between nutrient or hormonal cues and cell growth and maintenance.

mTOR was first identified in yeast in studies of the immune-suppressive compound rapamycin. Rapamycin had previously been shown to act by forming a gain-of-function complex with Fpr1p, with mutations in FPR1 conferring recessive resistance to rapamycin, and expression of the human homolog FKBP12 (FK-506 binding protein 12) restoring drug sensitivity. Mutations in TOR1and TOR2, originally designated DDR1 and DDR2 (dominant rapamycin resistance), were found to confer resistance to the antiproliferative effects of rapamycin in yeast. Wild-type Tor1 and Tor2 are bound and inhibited by the Fpr1/rapamycin complex [1]. This mechanism was subsequently found to be conserved in mammals with the rapamycin-FKBP12 complex binding to and inhibiting mTOR [2].

A link between mTOR signaling and aging was first established in yeast when studies in Saccharomyces cerevisiae demonstrated that deletion of Sch9, the yeast homolog of the mTORC1 substrate S6K (see ‘mTOR Downstream Signaling'), results in a significant increase in chronological life span, defined as the duration of time that a yeast population retains viability when in a nondividing state [3]. Studies in the nematode Caenorhabditis elegans subsequently revealed that mTORC1 can negatively regulate longevity in multicellular organisms; knockdown of daf-15 (the nematode homolog of Raptor, a component of mTORC1) or let-363 (the nematode homolog of mTOR) by RNAi can extend life span in this model [4]. Reports from Drosophila melanogaster and yeast replicative aging studies further supported the role of mTOR in regulating longevity in lower eukaryotes.

Direct evidence of a role for mTOR in mammalian life span has been provided by studies showing life span extension in mice resulting from deletion of S6K, by double heterozygosity for mTOR and mlst8 (a component of mTORC1), and by treatment with rapamycin [5]. Intriguingly, life span extension in each of these studies was strongly sex-specific, with males receiving no longevity benefit from S6K deletion or double heterozygosity of mTORC1 components. Female mice also experienced a more robust response to treatment with rapamycin with an 18% increase in median life span, compared to a 10% increase in male animals [5]. Notably, intervention with rapamycin resulted in an increase in life span even when rapamycin treatment began late in life, suggesting that mTOR inhibition may prove an attractive target for intervening in human aging.

mTORC1 and mTORC2 both play essential roles in eukaryotic biology, as complete loss of either Raptor, an mTORC1 specific component, or Rictor, an mTORC2 component, results in embryonic lethality. While both are essential for development, the two mTOR complexes differ in their components, relative regulatory roles, and upstream regulation of their activity. The upstream regulators and downstream effectors of mTORC1 are generally better characterized than those related to mTORC2. Although mTOR signaling is affected by numerous intra- and extracellular growth cues and conditions, and it affects numerous downstream pathways and processes, only the best characterized of these pathways, in terms of aging and disease, are described below.

mTOR is activated by a variety of growth factors and mitogens (fig. 1). Among the canonical regulators of mTORC1 signaling in mammals are insulin and the insulin-like growth factors (IGFs). Insulin and IGFs are recognized at the cell surface by tyrosine kinase receptors and provide the primary extracellular regulation of mTOR. Signaling through insulin/IGF is partially through phosphoinositide (PI3)-mediated activation PI3-dependent kinase (PDK) and subsequent activating phosphorylation of AKT on T308 [5]. IGF-1 signaling represents a longevity-regulating pathway in its own right, acting through both mTOR and through FoxO. IGF receptor loss has been shown to increase life span in mice and worms (daf-12 is the IGF homolog in C. elegans); serum IGF levels have been shown to correlate with life span among mouse strains, and FOXO gene variants are strongly associated with extreme longevity in humans [6]. IGF-1 is the canonical and best characterized IGF activator of mTOR. AKT stimulates mTORC1 through phosphorylation of the mTORC1 inhibitor tuberous sclerosis complex protein 2 (TSC2). In its active form, the TSC1/2 complex is a GTPase-activating factor for the small guanine nucleotide-binding protein Rheb. Stimulation of Rheb by active TSC1/2 results in a conversion of loaded GTP to GDP, inactivating the protein. Active GTP-bound Rheb is a necessary component of mTORC1; thus, inhibition of TSC1/2 results in downstream activation of mTORC1 through an increase in active Rheb [7].

While mTORC2 is not activated by insulin and growth factors through canonical signaling events, an intriguing mechanism has been described which couples regulation of mTORC2 activity by extracellular signals to intracellular ribosomal capacity [8]. In this pathway, mTORC2 is activated by PI3K through increased physical association with ribosomes. Activated mTORC2 at the ribosome phosphorylates S473 on a hydrophobic motif of Akt, priming it for activation by PDK1. Ribosome-associated active mTORC2 also phosphorylates a hydrophobic motif of SGK and PKC. The relationship between mTORC2 and ribosome activity links cellular translation capacity to growth signaling. Simultaneously, ribosomal biosynthesis is also regulated by TORC1 activation of p70S6K and its target, the ribosomal S6 subunit. This crosstalk between mTORC2 and mTORC1 provides another link between intracellular growth capacity sensing and the regulation of growth by mTOR.

mTORC2 thus lies both upstream and downstream of AKT, a relationship that may explain a variety of otherwise inexplicable observations related to mTOR biology. One such case is the uncoupling of mTORC1 and AKT signaling by Foxo induction. Foxo has been found to upregulate levels of Rictor, the mTORC2-specific binding partner of mTOR, and activate AKT while simultaneously causing a decrease in assembled mTORC1. This decrease has been shown to result from sequestration of mTOR to mTORC2, while increased activity of mTORC2 directly activates AKT. The downstream effects of mTORC2-specific signaling are not well characterized in the context of aging but this complex relationship with mTORC1 and AKT demonstrates the importance of mTORC2 in the outcome of mTOR signaling perturbations.

mTOR activity is modulated not only through extracellular signaling, but also by multiple intracellular energy and nutrient-sensing pathways. One well-characterized regulator of mTOR activity is the highly conserved AMP-activated kinase, AMPK. AMPK is an ancient sensor of energy status that acts as an intracellular upstream regulator of mTOR. AMPK is sensitive to the AMP:ATP status in the cell and is activated as this ratio increases. Upon activation, AMPK drives catabolic processes, such as fatty acid metabolism, and inhibits anabolic reactions, such as lipid synthesis, for the purpose of balancing intracellular energy status. AMPK acts on mTOR through multiple interactions. TSC2 is activated by AMPK through phosphorylation at T-1227 and S-1345. AMPK also appears to directly inhibit mTORC1 through phosphorylation of Raptor at serine 722 and serine 792 [9].

Multiple pharmaceutical agents that target AMPK signaling are available for research use, and AMPK activators have been used clinically in the setting of type 2 diabetes. Phenformin, a potent and direct AMPK activator, was used for the treatment of type 2 diabetics for decades before being removed from clinical use. This drug had strong beneficial effects for patients but was found to cause life-threatening, sometimes lethal, lactic acidosis following exercise [10]. The less potent and less characterized drug metformin replaced phenformin in the clinic. Metformin appears to have many of the same biological effects as phenformin (primarily inhibition of glycolysis and regulation of blood glucose) with a much lower frequency of the life-threatening lactic acidosis. The exact mechanism of function for metformin is not well understood, with some recent studies suggesting that the compound functions through inhibition of complex I of the electron transport chain, and others demonstrating an effect from the drug even in AMPK-deficient animals, together suggesting that metformin functions in large part through AMPK-independent pathways.

mTORC1 is also directly activated by amino acids through an interaction at the surface of the lysosome, allowing for intracellular amino acid abundance to regulate growth and metabolism directly through mTOR [11]. This sensing is facilitated by the recently discovered ragulator complex, a vacuolar ATPase binding complex that contains a guanine nucleotide exchange factor for Rheb and is regulated at the lysosomal surface by amino acids. Under conditions of high intracellular amino acids, mTORC1 and Rheb are recruited to the ragulator complex at the surface of the lysosome where Rheb activates mTORC1 as described above. Together, the regulatory interactions described above define mTORC1 as a central sensor of growth conditions and energy cues.

Caloric restriction (CR) is a well-documented life span-extending intervention that has been found to be effective in divergent species including yeast, flies, rodents, and nonhuman primates [12]. Defined as a reduction in nutrient intake in the absence of malnutrition, CR (also referred to as dietary restriction or DR) is simple in theory but is a complicated intervention in practice, and one that has not been rigorously standardized. CR strategies differ greatly between organisms, life span assay type, laboratory, and field of study [13]. Even within the same organism, CR can be implemented in a variety of fashions. In nematodes, CR can be accomplished by titrating down the bacterial food available on solid plates or in media or by removing bacteria altogether (referred to as bacterial deprivation or BD). A variety of CR treatments exist in yeast including variable degrees of glucose deprivation or substitution of glucose (the standard carbon source for yeast on plates) with nonfermentable carbon sources. Murine studies of CR vary greatly in terms of food composition, percent CR (mass of food provided compared to ad libitum intake), and housing conditions, such as implementing CR on singly versus multiply housed mice.

Given that CR varies widely in implementation practices, it is almost surprising that CR generally works, suggesting that the downstream factors regulating this intervention are robust, well conserved, and may hold real promise in human aging and disease. There has been a significant effort to identify downstream effectors of CR, and mTOR has been often identified as a pathway likely to regulate a portion of the CR response. Given the role of mTOR in nutrient signaling and response to growth cues, it makes sense that mTOR would be involved in the CR response. mTORC1 activity has been shown to be reduced by CR in invertebrate and mammalian studies. Additional evidence implicating mTORC1 in CR comes from reports in which CR fails to additively extend yeast replicative life span in mutants for mTOR or S6K, as well as from studies in C. elegans showing a similar lack of additive extension between a genetic model for CR (the eat-2 mutant) and RNAi against mTOR [14]. S6K and translation initiation factors were, however, found to be additive with CR in this model.

It has become clear over the past decade of aging research in invertebrates, and more recently in murine studies, that classical genetic complementation experiments are often difficult to interpret and are prone to overinterpretation when longevity is the phenotypic readout. Given that longevity is an extraordinarily complex trait and that the regulators of maximum longevity are still poorly understood, it has proven difficult to demonstrate true complementarity in life span studies. There always exists a possibility that some unrelated factor or process is limiting additive life span extension in experiments that appear to demonstrate complementation. Even considering this caveat, there is a large body of literature arising from multiple model organisms that link mTORC1-regulated processes, including autophagy and mRNA translation, to the beneficial effects of CR; together, these strongly suggest a role for mTOR in the CR response. Given the clear links of CR to processes downstream of mTOR, the observed decrease in mTORC1 signaling on CR, and the role of mTOR in nutrient signaling, there is a general consensus that altered mTOR signaling plays a significant role in CR, though additional pathways undoubtedly contribute to the overall effects of the intervention.

As a major energy and growth signaling sensor, mTORC1 acts as a central coordinator of proliferative and maintenance programs. mTORC1 activity drives growth through activation of mRNA translation, regulation of metabolic pathways including glycolysis and fatty acid metabolism, and repression of cellular catabolic pathways, primarily the autophagy/lysosomal degradation pathway. Inhibition of mTORC1 results in reduced mRNA translation, increased catabolic processes, and a shift in metabolic substrate preference. Many of the effects of mTORC1 activity or inhibition are mediated by activation or suppression of downstream transcriptional regulators and the complex crosstalk between these factors. This is particularly apparent in the effects of mTORC1 modulation on metabolism, with the outcomes being highly context dependent on organism, tissue type, duration of intervention, type and severity of intervention, and complex interactions with extracellular or extraorganismal environment. Thus, dissecting out the pathways and targets of key importance in aging and disease is a significant challenge. This is a context in which future work using systems biology approaches may play an especially important role.

While mTOR signaling is complex, some highly conserved and well-defined pathways have been identified downstream of mTORC1 and mTORC2. mTORC1 has been studied more extensively, and mTORC1-regulated processes are generally better described. The best-described mTOR-driven processes are briefly addressed below.

Hormonal signaling and abundant nutrient availability promote mTORC1 activation which upregulate a variety of cellular processes necessary for growth. One critical process driven by active mTORC1 signaling is mRNA translation, required for protein synthesis and cell growth. mTORC1 kinase activity is known to promote translation through at least two distinct substrates [15]. mTORC1 phosphorylates p70S6K, the 70-kDa ribosomal S6 kinase, which is an activator of ribosome biogenesis. The interaction of this level of regulation of translation with the control of ribosomal biogenesis by TORC2 has been previously mentioned. Eukaryotic translation initiation factor 4E-binding protein 1 (eIF4E-BP1 or 4E-BP1) is also directly phosphorylated by mTORC1. Phosphorylation of eIF4E-BP1 results in its release from the eukaryotic translation initiation factor 4E (eIF4E), allowing eIF4E to associate with mRNA cap binding proteins and form the cap-dependent translation initiation complex. The formation of this cap-binding complex is a key translation initiation event in eukaryotes.

Activation of mRNA translation is a critical function of mTORC1 and likely accounts for many of the phenotypes associated with mTOR-driven disease, while decreased translation likely mediates many of the positive effects of mTOR inhibition. The antiproliferative effects of mTOR inhibitors in cancer and immune diseases may largely be attributed to decreased rates of protein synthesis due to reduced translation. Decreased mRNA translation has been identified as a major pro-longevity intervention in multiple organisms, and CR is thought to largely act through decreased translation. Deletion or knockdown of ribosomal components or translation initiation factors has been clearly demonstrated to increase life span in yeast, flies and nematodes. Furthermore, deletion of S6K extends life span and decreases body size in mice [6], though rates of translation have not yet been directly examined in these animals.

While mRNA translation is globally decreased in the setting of CR, and in at least some reports of mTOR inhibition, multiple studies in model systems suggest that the beneficial effects of reduced translation may result from differential translation of a subset of mRNAs rather than simply being a consequence of reduced global translation. This model has been best established in budding yeast, where life span extension resulting from ribosomal protein subunit deletion has been linked to an increase in translation of the transcription factor Gcn4 [16]. Gcn4 regulates a variety of genes including genes coding for proteins necessary for response to low nutrient conditions and genes encoding proteins involved in stress response. In yeast, Gcn4 has been found to be necessary for life span extension by ribosomal mutants, and appears necessary for full life span extension by mTOR or S6K deletion. Similar observations have been described in nematodes and flies, but this model has not yet found support in mammalian systems. A recent report using the mTORC1-specific catalytic inhibitor Torin 1 in p53^-/- mouse embryonic fibroblasts suggested that differential effects on translation resulting from mTOR inhibition in mammalian cell culture could be largely explained by the presence of a 5′ terminal oligopyrimidine motif, though the study could not rule out the existence of less abundant differentially regulated 5′ or 3′ elements [17]. The interpretation of this study is complicated by the authors' claim that all of the effects of mTOR inhibition were mediated through the 4E-BPs. Deletion of individual 4E-BPs has not been reported to alter translation and has no obvious effect on body size in mice, although it does appear to alter adipose tissue mass [18], while deletion of S6K results in a marked reduction in body mass, demonstrating that S6K regulates growth rate (presumably through direct effects on translation) in vivo. These apparent discrepancies may be a result of the cell culture system or mode of mTOR inhibition - it may be that catalytic inhibition of mTORC1 by Torin 1 does not accurately model mTOR inhibition by rapamycins or through genetic modulation.

Global reductions in mRNA translation may directly contribute to the beneficial effects of mTOR inhibition on age-related diseases involving proteotoxic stress. The reduction of translation rates may directly enhance the fidelity of translation [19], and it is widely accepted that decreases in protein synthesis rates may allow for improved cellular proteostasis through a decreased workload on endogenous protein repair and degradation. A decreased steady state requirement for protein repair and degradation machinery may result in an increased capacity for cells to respond to transient stresses such as oxidative damage, protein aggregation, and heat or cold shock. Loss of proteostasis is a critical component of a number of age-related diseases (see ‘mTOR and Age-Related Diseases') and maintaining proteostasis is crucial for organism survival. It seems that decreased mRNA translation may be promoting longevity at least partially through improved proteostasis and increased protein degradation, though it is difficult to dissect this phenotype away from elevations in autophagy (see below), antioxidant defense, or other biological effects of mTOR inhibition.

In addition to promoting protein synthesis and cell growth, active mTOR inhibits the intracellular catabolic process of autophagy. As a major intracellular recycling pathway in eukaryotes, the autophagy-lysosomal pathway plays an essential role in degrading damaged organelles and macromolecules. Nutrient deprivation decreases mTOR activity, relieving the inhibition of autophagy by active mTOR and resulting in an increase in the catabolism of proteins and organelles. This increased catabolic activity provides amino acids and allows for cell survival when nutrients are limiting.

The observed accumulation of damaged and aggregated proteins, oxidized lipids, and damaged organelles with age suggest that basal levels of autophagy decline or are insufficient to prevent the accumulation of damaged macromolecules associated with aging. Lipofuscin, the complex granular pigment that accumulates in aged tissue, is a highly conserved phenotype of cellular aging that has been observed in all multicellular eukaryotes [20]. While the exact composition and functional consequences of lipofuscin remain to be determined, it has become clear that longevity-promoting interventions also slow the rate of lipofuscin accumulation. Thus, lipofuscin is often used as a biomarker of relative age. Given the close correlation between longevity and damaged macromolecule accumulation, the obvious prediction is that accumulated damaged macromolecules are a driving factor in aging and modulation of this accumulation could attenuate aging. While this hypothesis has proven difficult to test directly given the challenges in selectively inducing the autophagy-lysosomal system, a large body of evidence from yeast and C. elegans supports the model that induction of autophagy is a necessary downstream effector of mTORC1 inhibition in mediating life span extension [21]. Induction of autophagy has also been shown to be necessary for CR-mediated longevity promotion, potentially through mTOR [22]. While the necessity of autophagy for the success of these interventions is broadly accepted, it is not clear whether induction of autophagy alone is sufficient to increase life span.

In addition to the role of autophagy in promoting longevity, dysfunction of this pathway has been implicated in a variety of pathologies, and activation of autophagy has been demonstrated to attenuate a variety of age-related diseases. The induction of autophagy has been directly implicated as a potential clinical target for treatment of cardiovascular disease, age-related macular degeneration, diabetes, and a variety of neurodegenerative disorders including Parkinson's disease (PD) and Alzheimer's disease (AD) [23]. The nervous system appears particularly sensitive to the accumulation of damaged macromolecules and protein aggregates, and increasing autophagic degradation has been shown to prevent neurodegeneration in models of AD and PD as well as in models of Huntington's disease (HD), a progressive neurodegenerative disease directly associated with proteotoxic insult.

Mitochondria are key organelles in metabolism, disease, and aging. These organelles are the major producers of energy for most cell types, a primary site of metabolic reactions, a major source of toxic products (both reactive oxygen species and toxic intermediates of metabolism) and provide key cellular signaling regulators. Given the multifaceted role that mitochondria play in eukaryotic biology, it is unsurprising that mitochondria have been linked to a variety of pathological states, diseases, and aging [24]. mTORC1 appears to influence mitochondrial function through multiple mechanisms and downstream regulatory factors.

Hypoxia-inducible factor 1, Hif-1, is a transcription factor that promotes glycolytic processes, and can be activated through mTOR signaling in mammals [25]. This factor is linked to longevity in model organisms through somewhat unclear mechanisms and to vascular tumor growth, ‘wet' macular degeneration [26], and rheumatoid arthritis [27] in mammals through its positive effects on the angiogenic factor VEGF. At the intracellular level, Hif-1 promotes glycolysis, downregulates mitochondrial oxygen consumption, and at least partially mediates the Warburg effect in mammalian neoplasia. Decreased mTOR signaling is thought, therefore, to directly influence tumor vascularization, tumor metabolism, and the Warburg effect at least partially through decreased activation of Hif-1. This pathway may act independently of, or cooperatively with the general anti-proliferative effects of mTOR inhibition.

Consistent with these effects, mTOR inhibition has been associated with increased mitochondrial respiration in yeast and worms, and CR has been associated with increased mitochondrial content and respiration in a variety of organisms. This effect has been directly associated with longevity in yeast, with adaptive signaling resulting from increased mitochondrial superoxide production being implicated in this response [28]. Mice lacking mTORC1 components in white adipose tissue also show an increase in mitochondrial content and respiration, suggesting that this may be conserved in mammals [29]. Mitochondrial metabolism and cellular mitochondrial mass have also been reported to be increased by mTORC1 inhibition, at least in certain conditions, through a downstream activation of PGC-1α and the transcription factor Ying-Yang 1 [30]. The precise role of mitochondrial metabolism as a downstream mediator of mTOR is far from clear, but available data suggest that this is a critical component of aging and disease, and thus warrants further attention.

In addition to the above, mTOR affects mitochondrial function through autophagic degradation of mitochondria, a process termed mitophagy. The mitochondrial-lysosomal axis theory of aging suggests that proper maintenance of a functional pool of mitochondria depends on continuous successful removal of damaged and dysfunctional mitochondrial components through fission and mitophagy of fission products [31]. This theory predicts that reducing the rate of turnover of mitochondria, as may occur in aging, would result in an accumulation of dysfunctional mitochondria. This accumulation could lead to an increase in basal ROS production, damage accumulation, loss of tissue homeostasis, and potentially cellular senescence or death. mTORC1 inhibition increases basal autophagy, as described above, and would thus be predicted to preserve or restore mitochondrial function with age and thus potentially improve overall mitochondrial function. While this model remains to be directly addressed, it provides an attractive link between mitochondrial function and mTOR signaling in aging and disease.

Stem cell loss and dysfunction are likely a significant factor in mammalian aging and age-related diseases. This is particularly likely in proliferative tissues such as dermis, the immune and gastrointestinal system, as well as in wound repair or response to ischemic injury in which proliferative capacity is required. While the exact role for stem cells in aging is currently unknown, evidence suggests that mTORC1 is central in the maintenance of stem cells with age. As discussed below, mTOR inhibition has been shown to protect immune function with age in murine models of infection, and this has been attributed to enhanced hematopoietic stem cell capacity. mTOR inhibition with rapamycin has also been recently reported to improve intestinal stem cell function, although in this case the improvement was linked to alterations in mTOR signaling in the adjacent Paneth cells, which are responsible for maintaining the stem cell niche, rather than a direct effect on intestinal stem cells. CR has been shown to enhance the function of skeletal muscle stem cells, presumably related, at least in part, to the concomitant decrease in mTOR activity. The role of stem cells in aging and of mTOR in regulating their function remains an exciting and largely uncharted avenue of research.

Extension of health span, defined as the duration of life for which an organism is free from major age-related disease or loss of function, is considered by many to be the critical goal of aging research. Longevity studies in model organisms can intrinsically include health span components. C. elegans viability determination in standard plate or liquid-based life span studies is based on the ability of the animals to respond to mechanical stimulus. Yeast studies, both replicative and chronological, depend on the cell capacity to successfully produce progeny. In both cases, the individual organism may remain viable beyond the point that they are considered deceased by the assay standards. Thus, the nematode and yeast models are tied to neurological, muscular health, and reproductive capacity, respectively. In the murine model animal welfare regulations generally prevent expiration of mice by natural causes, requiring euthanasia if animals decline past a set cutoff in body mass, appear immobile, hunched, or in pain, or if they show signs of severe and untreatable diseases, such as ulcers or cancers. These restrictions may complicate accurate determination of life span in a longevity study but they also compel longevity-promoting interventions to be those that, at least in part, also protect health span.

Considering these restrictions, it is perhaps unsurprising that current data suggest longevity-enhancing interventions extend the health of populations and decrease or delay the incidence of age-related disease, rather than increase survival of unhealthy individuals. It appears that the regulation of health span and life span are at least closely linked. Concordantly, relatively few examples exist in the literature where health span is benefited in the absence of longevity benefits, perhaps a result of using life span as a primary end point. A noted exception is the recent (2013) NIA study of CR in rhesus monkeys, which observed a significant decrease in the appearance of age-related diseases without a change in survival [32]. While this, and studies like it, show that it may be possible to uncouple health span and longevity, they are generally limited by a lack of positive controls and/or a clear definition of what constitutes baseline health span. The rhesus study, for example, stands in contrast to a prior study that reported an increase in both life span and health span [33]. Differences in diet, housing conditions, and severity of the CR protocol may be factors distinguishing the two studies and the lack of a positive control limits the conclusions that can be drawn from this work. It is possible that longevity-promoting interventions may more sensitively and broadly affect health span than life span.

There has been a recent growth in emphasis on efforts to define and characterize the effects of longevity-promoting interventions on age-related health parameters. Each model used for aging studies has a set of health parameters that have been used to explore the relationship between life span, health span, and aging interventions. In yeast, the primary health span parameters are replicative capacity (in this case the actual readout for life span), mitochondrial function, and cell morphology. Nematodes have typically been used to study proteostasis and clearance of damaged macromolecules during the aging process as well as neurological decline, diseases of proteotoxicity, and age-related changes in muscle function. D. melanogaster has been useful in studying neurological function, muscle function, sensory function, stem cell function, and cardiac function with age. Mammalian systems have been used to examine a variety of age-related physiological and health parameters relevant to the biology of human aging.

Concurrent with demonstrations of a role for mTOR in regulating longevity, it became increasingly apparent that mTOR signaling plays a central role in regulating health span and a variety of age-related and non-age-related pathologies (table 1). mTOR inhibition attenuates specific age-related changes in lower organisms, as noted above. In addition, mTOR inhibition slows or delays many age-related and age-associated changes that are broadly conserved from lower eukaryotes to mammals. Included among these are lipofuscin accumulation, DNA damage accumulation, age-related mitochondrial dysfunction, and loss of proteostasis. Cardiac, neuronal, and stem cell functions are also improved with age in multicellular invertebrates and mammals by inhibition of mTOR. While the efficacy of mTOR inhibitors in attenuating age-related pathologies in humans is yet to be determined, there is an abundance of literature suggesting that mTOR is a clear potential target for diseases of aging.

There is evidence that mTOR inhibition may be generally protective against cardiomyopathies, including age-related cardiomyopathy [34]. Zebrafish heterozygous for mTOR were protected against two different models of cardiomyopathy. Administration of rapamycin markedly suppresses cardiac hypertrophy in the trans-aortic constriction model of pressure overload-induced heart failure, and rapamycin treatment has been shown to result in regression of established pressure overload-induced cardiac hypertrophy, fibrosis, and dysfunction. Perhaps the greatest impact of mTOR-targeted pharmacotherapy in cardiac disease has emerged through the widespread use of stents that elute rapamycin or rapamycin derivatives (e.g. everolimus, temsirolimus, ridaforolimus, umirolimus, zotarolimus, collectively referred to as ‘rapamycins') to inhibit cell proliferation and restenosis following angioplasty, with significant decreases in major adverse cardiovascular events during the first few years after implantation [35].

While decreased mTOR signaling has been clearly demonstrated to attenuate a variety of cardiac myopathy and failure models, the precise mechanisms of importance are less clear [34]. Decreased mRNA translation, inflammation, and hypertrophic growth signaling, increased autophagy, improved mitochondrial function or content, and altered metabolic preference have all been independently linked to improved outcome resulting from mTOR inhibition in cardiac models making it difficult to parse out the key functions downstream of mTOR crucial to the benefits observed. The complex nature of the cardiac system and the multifaceted role of mTOR inhibition in attenuating cardiac dysfunction make age-related cardiac dysfunction a particularly attractive model for a systems approach. Much of the recent and ongoing work in the cardiac aging field relies on systems biology techniques, such as shotgun proteomics, RNA sequencing, and metabolomics, to demonstrate a shift in organ state in aging and an attenuation with treatment.

These stand-alone systems analyses are proving highly informative for understanding the effects of interventions, but a full mechanistic picture of the role of mTOR in age-related cardiac disease and rescue will likely require a combination of multiple approaches. Systems analyses will likely reveal multiple distinct functional states in cardiac tissue where interventions affecting any of the individual downstream mediators of mTOR actually lead to a similar shift in the steady state through feedback. Decreased translation and increased autophagy could, in this scenario, converge in their functional consequence by causing an overall shift in the functional, proteomic or metabolic state of cardiac cells. Thus, the cardiac aging paradigm seems particularly amenable to a systems biology perspective.

As noted above, it has been suggested that mTORC1 inhibition-mediated enhancement of autophagy may lead to improved degradation of aberrant or misfolded proteins and reduced proteotoxic stress in neurodegenerative diseases such as PD, AD, and HD [36]. Thus, inhibition of mTOR could prove to be a successful therapeutic strategy in neurodegenerative disease. Evidence of such effects has been seen in fly and murine models of PD as well as in fly, murine, and cell culture models of HD. Positive effects of rapamycin on disease progression have been reported in two different mouse models of AD.

The evidence of a benefit of rapamycin in neurodegenerative diseases raises the question of whether mTORC1 inhibition might also attenuate age-related declines in cognitive function in the absence of a more severe neurological disorder. Recent studies assessing the effects of chronic mTOR inhibition on cognitive function during aging in mice have reported that old animals treated with rapamycin performed substantially better on tasks measuring spatial learning and memory than did untreated, age-matched animals [37]. Intriguingly, there were indications that rapamycin also enhanced cognitive function in young mice and had anti-anxiety and antidepressive effects at all ages. mTOR has also been identified as a potential target for treatment of seizures [38], suggesting that growth signaling inhibition may have broad neurological benefits. The complexity of the central nervous system makes it a challenge to investigate the mechanisms underlying the beneficial effects of decreased mTOR signaling, but the clear therapeutic potential make it a very attractive target for testing mTOR inhibitors in a clinical setting.

A majority of tumors show evidence for activation or upregulation of mTOR signaling, and mTOR inhibition has been studied extensively as a potential therapy for a wide variety of cancers. Rapamycins potently inhibit growth of solid tumor cell lines but have shown disappointing efficacy in several clinical trials, though certain rare cancers, including renal cell carcinomas and glioblastomas, may respond to mTOR inhibition. Mutations in TSC1 and TSC2, which are upstream inhibitors of mTOR, cause a variety of hyperplastic diseases including tuberous sclerosis, directly linking hyperplasia to mTOR overactivity. Dysplasias driven by TSC1 or TSC2 loss are currently being evaluated as clinical targets for mTOR inhibitors [39]. Rapamycin has been shown to delay or reduce deaths due to age-related and age-associated cancers in several studies in mice.

mTOR signaling has been implicated in the development of age-associated metabolic disorders such as obesity and type 2 diabetes. Inhibition of mTORC1 inhibits, while mTORC1 activation stimulates, adipogenesis in mice [40]. Obesity results in chronic activation of mTOR in adipose tissue, a state that has been linked to obesity-associated cancers, inflammation, β-cell adaptation preceding type 2 diabetes, nonalcoholic fatty liver disease, and many other complications [41]. Multiple downstream effectors of mTOR signaling, such as S6K-1, 4EBP1, and SREBP, act as mediators between nutrient signaling and the development of obesity and type 2 diabetes. SREBP, a transcription factor that induces the expression of lipogenic genes, is activated by mTORC1 but not mTORC2 [41]. S6K-1 null mice display reduced body fat mass and resistance to diet-induced obesity, while mice lacking 4EBP1 show increased sensitivity to diet-induced obesity and adipogenesis, possibly through hyperactivation of S6K-1 [41]. Mice lacking Raptor in adipose tissue are lean with fewer and smaller adipocytes, have increased insulin sensitivity, and show resistance to diet-induced obesity [29]. Conversely, mice with adipose-specific knock out of Rictor have normal body fat mass and glucose tolerance but are hyperinsulinemic [42]. Mice lacking S6K1 in all tissues or lacking RAPTOR specifically in adipose tissue show a profound resistance to diet-induced obesity [29].

The relationship between mTOR signaling and age-related metabolic disorders, including type 2 diabetes and obesity, is complicated. Rapamycin can protect mice from diet-induced obesity through the inhibition of adipocyte differentiation [43]. However, mice and rats chronically treated with rapamycin demonstrate altered metabolic homeostasis, observed as altered insulin sensitivity and glucose tolerance [44]. These effects of chronic rapamycin have recently been attributed to effects on mTORC2. Blood hyperlipidemia is also observed with chronic rapamycin treatment in mouse and human studies [40]. The S6K1 knockout mice are also hypoinsulinemic and glucose intolerant, apparently due to a decrease in β-cell size and function. Despite these diabetes-like symptoms, both S6K1 knockout mice and rapamycin treated mice are long-lived, suggesting that these effects are not so detrimental as to limit survival. It has also been pointed out that the ‘starvation-induced diabetes' associated with mTORC1 inhibition differs substantially from type 2 diabetes, which is caused by insulin resistance resulting from overnutrition in association with mTOR activation. Thus, additional study is needed to determine whether targeted inhibition of mTORC1 and the observed changes in lipid profile, insulin sensitivity, and glucose homeostasis represent a health risk or an altered metabolic state consistent with the promotion of longevity and health span.

Rapamycins are used clinically as immunosuppressive or immunomodulatory drugs. There are, however, also reports that rapamycins can enhance immune system efficacy in certain settings, including tuberculosis, anti-tumor vaccine responses in mice, and vaccinia vaccination in nonhuman primates [45]. In the context of age-related immune function, treating 22- to 24-month-old mice with rapamycin for only 6 weeks doubled the percentage and number of B (but not T) cells in the bone marrow, and restored the capacity of the aged animals' immune system to mount an effective response to influenza vaccination, which was protective against subsequent infection. The apparent contradiction between these observations and the use of rapamycins as immunosuppressive drugs may be explained by observations that mTOR can exert divergent immunoregulatory functions during immune cell activation and differentiation, depending on the cell subset type. Furthermore, while rapamycins may limit immune activation and proliferation in the setting of an immunogenic insult, they appear to have a robust effect in preventing age-related declines in immune function, thus preserving immune function later in life. Thus, rapamycins' functions in immune biology are more complex than previously recognized, with outcomes depending on dose, duration of treatment, immune cell type, and specific immune challenge. Furthermore, the long-term effects of rapamycins on age-related declines in immune function stand in stark contrast to those of short-term responses.

Inflammation is strongly associated with aging and drives a multitude of age-associated disorders. Cardiovascular disease, obesity and metabolic disorders, cancer, and neurodegenerative diseases all include inflammatory components, and attenuation of inflammation has been implicated as a clinical target in each of these disease settings. Hyperactive mTOR has been linked to inflammation, and inhibition of mTOR by rapamycins has been demonstrated to be anti-inflammatory in renal disease, lung infection, and in vascular inflammation in atherosclerosis and following angioplasty. Furthermore, CR strongly attenuates age-related inflammatory signaling, and this effect appears to be at least partly mediated through mTOR. Reduced mTOR thus seems a good candidate for treating or preventing age-related inflammatory processes, while reduced inflammation seems to play significant mechanistic role in the pro-longevity effects of mTOR inhibitors.

Rapamycins are used clinically to reduce nephrotoxicity in chemotherapy, prevent allograft rejection, and as a treatment for renal cell carcinoma. Activation of mTOR signaling has been associated with several common forms of kidney disease, suggesting that inhibition of mTOR might have broad therapeutic benefits for renal health. Consistent with this, rapamycins have been shown to reduce kidney fibrosis, attenuate diabetic nephropathy, and improve outcome in animal models of polycystic kidney disease.

Age-related macular degeneration is the leading cause of blindness in Western countries. Capillary overgrowth in the choroid layer of the eye, which is a contributing factor, has been attributed to excessive production of VEGF. Rapamycin has been shown to reduce VEGF expression in retinal pigment epithelium and inhibit angiogenesis in vitro [46]. In a rat model of age-related macular degeneration, rapamycin decreased the incidence and severity of retinopathy [47] and in human patients rapamycin appeared to decrease the need for anti-VEGF intravitreal injections by approximately half [48]. Thus, age-related macular degeneration appears to be a promising clinical target for mTOR-inhibiting interventions.

Hutchinson-Gilford progeria syndrome (HGPS) is typically caused by a de novo mutation in the lamin A/C gene (LMNA) that activates a cryptic splice site, producing an abnormal lamin A protein termed progerin. Accumulation of progerin leads to aberrant nuclear morphology in vitro, and is believed to be the causal factor in the pathogenesis of disease. The precise mechanism linking progerin accumulation to the phenotypes associated with this disease is unclear, but it is generally thought to involve disruption of nuclear DNA binding proteins, including transcription factors and DNA repair components, as a result of aberrant nuclear scaffold structure. It has been reported that treatment of cells from HGPS patients with rapamycin corrects the nuclear morphology defect, delays the onset of cellular senescence, and enhances the clearance of progerin through autophagic degradation [49]. No effective treatment for HGPS currently exists, and these data provide hope that rapamycins might slow disease progression in HGPS patients. Any success in HGPS would strongly suggest that rapamycins might show efficacy in patients diagnosed with atypical Werners' syndrome, often caused by non-HGPS mutations in LMNA, as well as in patients with muscular dystrophies resulting from lamin mutations or other laminopathies.

Genome-wide approaches, such as yeast single-gene mutant screens and RNAi screens in nematodes, have been critical to uncovering genes involved in the regulation of life span. The advent of proteomics, metabolomics, whole-genome sequencing, and RNAseq has fundamentally altered the way that aging studies are designed and executed. As the accessibility of these high-throughput methods has improved, these techniques are being increasingly utilized. The study of mTOR in aging has historically been largely led by reductionist experiments, with researchers focusing primarily on the use of single gene or small molecule perturbations to study mTOR and aging in model organisms. While these approaches have been, and continue to be, very fruitful in defining the regulation of aging and age-related processes by mTOR, they are limited in their ability to model the complexities of aging.

As large datasets produced using these methods become more widely available, it becomes increasingly important that the systems biology paradigm is applied to produce a comprehensive picture of the information (fig. 2). Modern aging research will greatly benefit from collaborations with bioinformatics experts that extend beyond analysis of high-throughput data and into systems-based integration of diverse datasets into models that combine genetic, proteomic, transcriptomic, and metabolomics data into informative and approachable descriptions of aging processes. These models should provide novel targets and strategies for intervention in the aging process as new nodes are described in genetic, proteomics, transcriptional, and metabolic paradigms. In addition, the systems approach to aging research may provide clear answers to difficult scientific queries, such as the nature of the similarities and differences between CR and CR mimetics (such as rapamycin). Thus, while classic methods for studying aging are far from exhausted, it is clear that systems approaches have a role to play in our understanding of aging, the influence of genotype on aging, and the mechanisms of interventions, such as mTOR inhibition, on the aging process.