It seems not all that long ago when almost every article on rodent aging contained a sentence that went something like, ‘dietary restriction (DR) is the only intervention known to retard aging in mammals.’ For more than 60 years, that was an accurate statement. But those days are long gone. Beginning with the discovery that a spontaneous mutation affecting pituitary development led to an even larger longevity enhancement than DR (Brown-Borg et al., 1996), the gates broke open, and we now have multiple dietary, pharmacological, and genetic means of enhancing longevity in mammals. For instance, nearly two dozen genetic manipulations have now been reported to make mice live longer (Ladiges et al., 2009).
Most of these manipulations consist of partially or completely disabling a gene, although a handful have overexpressed a human or mouse gene to achieve longer life (Ladiges et al., 2009). None, however, has enhanced the expression of multiple genes, achieved longer life, and at the same time illuminated a particularly contentious point in the biology of aging until recently. In an exceptionally intriguing report, constitutively overexpressing telomerase significantly improved health and extended survival in a mouse that had also been rendered cancer resistant by the simultaneous overexpression of a combination of tumor suppressors (Tomas-Loba et al., 2008).
As is well known, telomeres, the repetitive hexanucleotide DNA sequence and associated protein complex that protect the ends of linear chromosomes, typically shorten with each cell division in somatic cells. When telomeres reach a critically short length, a DNA damage response prevents further proliferation. Thus, telomere attrition itself is a potent tumor suppressor. Telomerase, a ribonucleoprotein that slows or prevents telomere attrition, is critical for the long-term proliferation of reproductive or stem cells, but at the same time can promote tumors if expressed at sufficient levels in proliferative tissues. This has been verified by several transgenic mouse models in which constitutive telomerase overexpression increased tumor incidence (Gonzalez-Suarez et al., 2001; Artandi et al., 2002) and by the observation that many human cancers express telomerase in tissues where it is normally silenced (Wright & Shay, 2005).
On the face of it, the observation that constitutive overexpression of telomerase leads to a long-lived mouse is more than a little paradoxical. For one thing, mouse telomeres are several times longer than human telomeres, and numerous previous reports indicate that even in highly proliferative tissues, they shorten a rather trivial amount over the course of a mouse’s life span. So what might the advantage of slowing or preventing their attrition be? In addition, mice, unlike humans, already express telomerase in many somatic tissues. These twin features of mouse telomere biology have made many investigators, including this author, skeptical about the role of telomeres in mouse aging.
But this article has altered the conceptual landscape of rodent aging research in several directions. The key feature of this particular long-lived mouse is that the researchers first enhanced its cancer resistance by the overexpression of other tumor suppressors (p53 and Ink4a/Arf) and then, in that cancer-resistant background, modestly overexpress the reverse transcriptase component of telomerase (TERT) as well. Enhancement of p53 activity has been reported by other researchers to accelerate aging despite its cancer resisting effect (Donehower, 2005), so the devil turns out as usual to be in the details. The relevant detail in this case is that these mice carry single extra copies of the tumor suppressor genes that appear to behave similarly to the endogenous genes rather than being promiscuously expressed. In this case, overexpression of p53 alone had previously been reported to have no effect on life span, but the combination of tumor suppressors lengthened life by about 16% compared to unmanipulated controls (Matheu et al., 2007).
In the current article, combining overexpression of TERT with overexpression of the tumor suppressors increased median lifespan about 40% compared to controls that overexpress only p53. Importantly, maximum longevity was also increased by this triple transgenic treatment. This is a key point because to make a credible claim that one has slowed aging rather than just extended life (related but not identical concepts); researchers typically like to verify increases not only in life expectancy but in maximum longevity too. The rationale for this is straightforward. If mean, but not maximum, longevity has been extended in the treatment group, it means that mortality rate in late life has been higher in the treatment group than controls – something difficult to reconcile with retarding aging.
Neither food intake or body weight was reduced, so inadvertent DR could not explain the results. In some sense, this longevity enhancement is the least interesting of the results, because these triple transgenic mice also displayed improved health by many measures even after discounting the expected delay in tumor formation. The mice were less prone to degenerative inflammatory pathologies, lost less subcutaneous fat with age, and exhibited better neuromuscular coordination and glucose tolerance than age-matched controls. On a cellular level, they also showed less telomere attrition (as expected) and fewer double-strand DNA lesions. All these health benefits occurred even though the transgenic mice had significantly higher serum IGF-1 than controls. In most mouse models of extended longevity, bioactive serum IGF-1 is suppressed (Ladiges et al., 2009). It is difficult to imagine some of these effects being due merely to longer telomeres, so these observations suggest that telomerase has cellular effects independent of its impact on telomere attrition. Unfortunately, end of life pathology was not reported, so we do not know what pathologies these long-lived, triple transgenic mice actually died from.
Another article reported the intriguing observation that dietary administration of a drug could improve mouse health without extending life (Pearson et al., 2008), a finding that if extended to humans would gladden the heart of many a pension fund bureaucrat. The drug in question was resveratrol, a naturally occurring polyphenol and putative sirtuin activator that has been reported to lengthen life in several other species. An earlier report on mice fed with a very high-fat diet (60% of calories from fat) indicated that resveratrol supplementation begun at 1 year of age increased median longevity and improved several markers of health (Baur et al., 2006). The newer study follows up on the earlier one, completing the high-fat survival study, and also reports on mice fed with a standard ad lib diet and a calorie-restricted diet, plus or minus resveratrol supplementation at several doses also beginning at 1 year of age. This study confirmed the earlier finding that resveratrol rescues compromised survival induced by a high-fat diet. However, resveratrol did not significantly affect survival of any mice fed with either standard or calorie-restricted diets. Despite the lack of a survival effect, resveratrol improved some (but not all) measures of bone health in addition to attenuating cataract formation in elderly mice fed with the higher resveratrol dose. An improvement in balance and co-ordination as revealed by the ability to remain on a rotating rod was also observed at the higher, but not the lower, resveratrol dose. Several measures of vascular health were also improved in aged mice.
The overall health profile of mice fed with standard diets supplemented with resveratrol was complex, showing improvement in some parameters but not others. Sample sizes in all cases were small. However, the suggestion that aspects of health can be improved without an impact on survival is important and provocative and deserves to be confirmed and extended in both this genotype (male C57BL/6) and a wider array of mouse genotypes.
Resveratrol has sometimes been called a DR mimetic because its impact on gene expression seems to mimic some aspects of DR. The fact that DR extends life and slows many aspects of aging in multiple genotypes of multiple species has been known for decades (Weindruch & Walford, 1988), even though the mechanism in mammals at least remains elusive. Are there new developments in the mammal DR field?
Indeed, there are. One of the most interesting has to do with whether some of the DR effect is because of reduced body fat in restricted animals. This hypothesis makes intuitive sense as high body fat has been associated with reduced life expectancy and a host of human diseases (Barzilai & Gupta, 1999). However, the conventional wisdom among DR biologists has been that body fat was not a causal player in DR’s beneficial effects. The slender portfolio of evidence supporting this conventional wisdom includes: (i) that individual longevity in F344 rats is not correlated with amount of body fat in ad lib fed animals and is actually positively correlated with body fat in restricted animals (Bertrand et al., 1980); and (ii) that restricting calories in female mice that were obese because of a disabled leptin gene (ob/ob mice) made them live even longer than nonobese, fully fed controls even though the ob/ob restricted animals had more than twice the percent body fat of controls (Harrison et al., 1984). Of course, the first point above does not directly address the issue of whether reduced body fat might be responsible for the DR effect, although it is suggestive, and the second point is weakened by the realization that fat is more than an inert energy storage organ but also has endocrine and immunological functions. Leptin-deficient fat, as in the study cited earlier, is likely to have multiple systemic effects.
To address this issue experimentally, Muzumdar et al. and colleagues surgically removed visceral fat (the presumptive ‘bad’ fat because it is responsible for compromising insulin action) from young ad lib-fed rats (VF group) and tracked their survival in comparison with ad lib-fed controls and a food restricted group fed 60% the calories consumed by the controls (Muzumdar et al., 2008). Fourteen months later, visceral fat was still significantly reduced in the VF group, and in fact was statistically indistinguishable from the amount in the DR group – about one-third that of the controls. As expected, DR rats lived about 35% longer than fully fed controls. Interestingly, the VF group also lived longer than the controls (mean and median longevity about 10% longer), although not as long as DR animals. These results suggest that reduction of visceral fat may play at least a partial role in the life extension observed under DR.
The past year also marked the publication of the first survival analysis from a long-term monkey DR study (Colman et al., 2009). These results have been eagerly anticipated because the health and longevity effects of DR have never been investigated in a long-lived species, much less a close human relative, as it has been repeatedly in rodents. The one exception was a study in which dogs’ (Labrador retrievers) caloric intake was reduced by 25% (Lawler et al., 2005). Although longevity was significantly increased in the restricted group, this dog breed is exceptionally prone to bone and joint disease and many (half) of the animals had to be euthanized for nonlife-threatening but painful conditions. This makes interpretation of the study problematic and complicates comparisons with rodent studies, in which only moribund animals are euthanized.
The primate study is one of two long-running DR studies of rhesus macaques that were initiated in the late 1980s (Roth et al., 1991; Wanagat et al., 1999). In this recent article, 38 control monkeys were compared with the same number of DR animals, who had been subjected to 30% calorie-restriction begun in early adulthood and continued for 15–20 years (two study cohorts included in the analysis were initiated 5 years apart). The study has been nicely designed and performed, particularly given the special challenges of working with nonhuman primates. The results, however, with more than half of the control animals dead, still seem to this observer to be equivocal, particularly in comparison with laboratory rodent studies. For instance, although a statistically significant difference in survival between the two groups was reported (P = 0.03); this result depends on the exclusion of a substantial number of deaths. One-third (7 of 21) of the deaths in the control group were excluded compared with nearly two-thirds of deaths (9 of 14) in the DR group. The rationale for censoring these deaths is that some animals died because of accidents or other causes deemed not to be age related and thus unrelated to the focus of the study. I would be sympathetic to this perspective if we could be certain that the nonage-related deaths were not themselves affected by the dietary regime – something that is not clear. If all deaths are included, the statistical survival difference between the groups disappears (P = 0.16), although the trend is still in the expected direction. Few, if any, deaths are censored in rodent DR studies, but mortality differences by mid-life are nonetheless substantial (Turturro et al., 1999). Thus, in terms of mortality, it is difficult to compare the monkey DR results with those of laboratory rodents.
But mortality statistics are only part of the story. Of equal importance are the effects of DR on health and on the severity and timing of degenerative diseases. Here, the results are unequivocal, at least with respect to the development of diabetes and prediabetic glucose dysregulation that dominate the age-related clinical profile of these monkeys. Among control animals 42% (16 of 38) developed glucoregulatory problems by midlife, compared with not even a single case among the DR animals. This seems a strikingly high rate for this disease in control monkeys, and may reflect the sedentary life enforced by laboratory husbandry, but in any case it is clear that DR completely abolishes the problem. This observation is consistent with what is known in humans, where reduced weight because of either DR or to endurance exercise improves insulin sensitivity and glucoregulatory control (Weiss et al., 2006), although some indices of glucoregulation respond better to exercise alone (Fontana et al., 2009).
There were no statistically significant differences in risks of other major age-related diseases in the study, but there were too few cases to yield much statistical power for any single disease. Those differences that were seen were all in the expected direction (eight cases of neoplasia in controls versus 4 under DR, four cases of cardiovascular disease in controls versus two under DR). A previous report from the same animals indicated that DR also attenuated age-related muscle loss (Colman et al., 2008). Age-related loss of gray matter in some regions of the brain appeared reduced, but whether this has cognitive significance is unknown, because no cognitive tests were reported.
In sum, this primate DR study points to a clear reduction in health problems related to glucose regulation relative to sedentary controls with less clear but promising effects on some other health conditions. However, the effect of DR on monkey survival is more ambiguous and certainly seems less dramatic than in most rodent studies.
Another recent study highlighted complications that may lie below the surface of many aging studies because of the propensity of those in the field to rely too heavily on results from a single genetic background. The complication in this case involves the spontaneous mutation in the Prop1 gene (Prop1df), a hypomorphic allele that confers roughly 50% greater longevity on the Ames dwarf mouse compared with littermate controls. This was the first longevity mutation discovered in mammals (Brown-Borg et al., 1996) as well as the one with the biggest effect on life span to date. Moreover, the lifespan of Ames dwarf mice can be extended still further by 30% caloric restriction (CR); on CR diets, Ames dwarf mice live about 75% longer than ad lib-fed controls (Bartke et al., 2001).
Garcia et al. (2008) set out to evaluate the eminently plausible hypothesis that the rate of accumulation of somatic mutations contributes significantly to longevity (Vijg, 2007). In order to bestow a reasonable degree of generality on their results, these authors decided to increase longevity by one environmental manipulation, i.e. 40% DR, and one genetic manipulation, i.e. the Prop1df Ames mutation. To measure mutation rate, this group employed a genetic construct in which multiple copies of a lacZ mutation reporter gene were maintained in an inbred C57BL/6J background. To produce their experimental animals, they crossed these mice to Prop1df animals from the heterogeneous Ames stock. To increase the homogeneity of the genetic background, they made repeated backcrosses with the C57BL/6 lacZ mice, thus placing the Prop1df allele into a background that included increasing proportions of C57BL/6 genes. After five such backcross generations, however, they discovered that a substantial fraction of pups born died prior to reaching adulthood (Garcia et al., 2008). They concluded, as others had previously noted (Nasonkin et al., 2004), that the Prop1df/df mutation has deleterious side-effects in the context of C57BL/6 background genes.
To pursue their original goals, Garcia and colleagues then made use of mice that placed the lacZ reporter construct onto a genetically heterogeneous stock with a greater proportion of Ames background genes, and used these mice for their experiments. They found, as expected, that both DR and the Prop1df mutation increased longevity compared with controls, Surprisingly, however, DR of the Prop1df mice not only failed to further extend their lives, but shortened their lives to no longer than the ad lib-fed wild-type controls. Concerned that 40% restriction was leading to starvation in the Prop1df mice, the authors reduced the restriction level to 30% when the dwarfs were 19 months of age, but this did not decrease their higher mortality compared with fully fed dwarfs. Thus, a subtle alteration of genetic background appears to reverse the survival effect of DR, at least in this long-lived mutant mouse. In more general terms, it remains to be seen to what extent the favorable effects of pituitary dwarfing genes, and related endocrine mutants, may be blunted or reversed on specific genetic backgrounds.
The mutational analyses performed with these mice failed to provide much support for the hypothesis that somatic mutation frequency correlates well with variation in lifespan. As expected, mutation frequency increased with age in all three tissues measured (liver, kidney, intestine), although only two relatively young ages (7 and 15 months) were measured. However, although the longer-lived mice bearing the Prop1df mutation or subject to DR generally exhibited lower mutation frequency relative to fully fed controls, this was not universally true. At 7 months of age, for instance, mutation rate in dwarf intestine was significantly higher than controls, although this effect was reversed by 15 months. Moreover, although DR reduced the longevity of the dwarf genotype, it nonetheless had no effect or even lowered mutation frequency compared to that of fully fed dwarfs. Data from mice at older ages might have helped to clarify this rather inconclusive picture. The data on C57BL/6 mice were also inconsistently related to the somatic mutation hypothesis. DR lowered mutation rate in liver, although that difference waned with age and this pattern was reversed in the intestine.
Perhaps the most dramatic discovery with respect to mammalian aging over the past year or so is further evidence for the involvement of the mTOR network in longevity modulation. TOR (Target Of Rapamycin) is a serine/threonine kinase that is at the center of a complex cellular signaling network that interacts with the insulin/IGF signaling pathway among other pathways with potential effects on health and aging. TOR is conserved across all eukaryotes studied to date (Wullschleger et al., 2006). The mammalian TOR, designated mTOR, acts as a nutrient sensor. Activation of TOR by nutrients or growth factors has generally anabolic effects, stimulating protein synthesis, ribosome biogenesis, transcription, as well as inhibiting autophagy and promoting cell growth and proliferation. Stresses, such as nutrient deprivation, hypoxia, or DNA damage, inhibit TOR and consequently reverse the effects described earlier, stimulating cellular stress resistance and deterring cell growth and proliferation. Inhibition of TOR signaling had previously been shown to lengthen life in yeast, worms, and flies (Vellai et al., 2003; Kapahi et al., 2004; Kaeberlein et al., 2005), possibly by mimicking the cellular effects of DR.
Therefore, the National Institute on Aging’s Interventions Testing Program (ITP) evaluated the potent TOR inhibitor rapamycin for its effect on mouse longevity (Harrison et al., 2009). Three aspects of the study’s results make it particularly striking. First, the rapamycin intervention did not begin until the mice were 20 months old, which is the rough demographic equivalent of 60 years old in human. No other successful intervention in rodent longevity has been initiated at such an advanced age. From the time they began receiving rapamycin in their food, life expectancy increased 28% for males and 38% for females compared with controls fed the same diet without the drug. The original experimental design involved earlier exposure of the mice to rapamycin, but technical difficulties in drug delivery took some time to resolve, so that the mice were 20 months old before the dietary intervention could be initiated. However, another cohort of mice, begun on rapamycin at 9 months of age, also showed a statistical improvement in survival.
Second, rapamycin intervention was performed simultaneously on three independent groups of mice at three research sites (The Jackson Laboratory [TJL], University of Michigan [UM], and the University of Texas Health Science Center San Antonio [UT]). I have previously lamented that important findings in mammalian aging results are only infrequently replicated in second or third laboratories or in additional genetic backgrounds (Austad, 2008). The design of the ITP obviates this concern and gives the study an immediate high degree of credibility and generality. Each site obtained similar qualitative results, but there were some differences worth mentioning. At two of the sites (UM and UT), males in the group destined to be given rapamycin had lower mortality rates compared to controls even prior to initiation of the rapamycin treatment, indicating that these mice may have been healthier before drug treatment began. This effect may have been because of differences in the control diets they received prior to the initiation of rapamycin exposure. At TJL, this complication did not exist. A significant difference in rapamycin-fed male survival was observed at all three sites, however. Moreover, in the cohort of males that got rapamycin from age 9 months, survival was equivalent at the time the drug treatment was initiated, and these mice also showed a significant enhancement of survival.
Among females, there was no such complication in any group. Survival between control and experimental groups was identical at 20 months of age when the supplementation began, and there was a significant increase in longevity at all sites from that point on. Female survival was also statistically increased in the group begun at 9 months.
As mentioned previously, it is important to also note whether maximum longevity – often operationally defined as the longevity of the longest-lived 10% of animals – is also extended, to make a credible claim that aging has been retarded. At all sites and in both sexes, the age at 90th percentile survival in rapamycin-treated animals was above the 95% confidence interval of the age of 90th percentile survival in controls.
The third aspect of this experiment that made the results striking was that it was performed in a genetically heterogeneous mouse stock created from interbreeding of four inbred laboratory strains (Nadon et al., 2008); thus, the results are unlikely to be because of some idiosyncrasy of a single mouse genotype. Overall, the distribution of causes of death in the control mice did not differ significantly from the rapamycin-treated mice.
A second study of rapamycin treatment suggests that it may extend life even if begun even later. Chen et al. administered rapamycin beginning at age 22–24 months to a small number of male C57BL/6 mice and found a statistically significant impact on remaining life span (Chen et al., 2009). Moreover, in this study a short (6 week) bout of rapamycin given to even older (26 month old) mice boosted their immune response to vaccination, so that they survived experimental infection with influenza virus significantly better than age-matched mice that had only been treated with vehicle. This result is particularly intriguing as it has been previously observed that DR of this same mouse genotype decreases survival in response to experimental influenza infection (Gardner, 2005).
Now that it has been firmly established that rapamycin treatment extends life in mice even when begun late in life, it will be of interest to see whether even greater life extension might be achieved with other doses of the drug. After all, only a single dose has so far been evaluated. It will also be important to assess whether in addition to extending life, rapamycin also improves measures of health. Finally, it will be useful to investigate in more depth the detailed molecular pathway or network by which rapamycin has its effect.
It has been assumed that rapamycin has its effects through inhibition of mTORC1, one of two multiprotein complexes that regulate different branches of the mTOR network. Indeed, the bioactivity of rapamycin was verified in the ITP study by documenting inhibition of the kinase activity of mTORC1 via reduced phosphorylation of S6, a target of one of mTORC1’s downstream effectors S6K1 (Harrison et al., 2009). A strong suggestion that inhibition of S6K1 contributed to the rapamycin effect was the publication several months later that deletion of S6K1 gene enhances longevity and improves other markers of aging in female, although not male, C57BL/6 mice (Selman et al., 2009).
The sex specificity of longevity-retarding treatments has been observed previously (Holzenberger et al., 2003; Selman et al., 2008; Strong et al., 2008) but almost never investigated. Indeed, in some of the studies discussed in this article, only one sex was used (Pearson et al., 2008) or the sex of animals was not specified (Tomas-Loba et al., 2008). Opportunities for learning about sex-specific mechanisms of aging might be lost by continuing to ignore these differences.
In the case of S6K1 gene deletion, the sex specificity is particularly striking. Mean longevity of S6K1−/− females was increased by 20%, and maximum lifespan by 10% compared to controls, whereas not a trace of such an effect was seen in males. Moreover, despite greater food consumption, the S6K1−/− females weighed less and had less body fat than controls. At 600 days of age, genetically manipulated females also exhibited greater activity, improved measures of bone health, lower plasma glucose and leptin, and better insulin sensitivity than controls. S6K1−/− females were also better at remaining on a rotating rod than controls, although the interpretation of this test is somewhat problematic, because body weight affects performance in this test (McFadyen et al., 2003). There was no indication that reduced circulating IGF-1 played a role in the aforementioned effects.
The apparently lack of a sex-specific effect on survival in rapamycin-treated mice compared with an emphatic effect in S6K1 knockout mice suggest either than there are sex-specific genetic background effects or that these two life-extending treatments may be working by mechanisms that overlap only partially. The complexity of the mTOR network is just beginning to be appreciated. S6K1 has complex effects, including interactions with the activity of adenosine monophosphate (AMP)-activated protein kinase (AMPK) and the insulin-IGF signaling pathway. There are multiple outputs from mTORC1 in addition to S6K1 as well (Wullschleger et al., 2006) and evidence that chronic rapamycin treatment can also inhibit mTORC2, in addition to mTORC1, in many cell types (Sarbassov et al., 2006).
The past year or so has firmly established the mTOR network as an important modulator of mammalian aging. The next few years should lead to an explosion of knowledge about the precise health effects of manipulating specific components of this network.