Interventions that increase lifespan may be doing so by retarding aging and, thus, may provide powerful tools for analyzing the biology of aging and late-life diseases. Data showing that a drug can increase lifespan are the first step in developing the case that the drug can delay or decelerate aging. Repeatable evidence that mouse lifespan can be extended by drugs has developed only recently (Strong et al., 2008; Harrison et al., 2009; Miller et al., 2011), and the question of whether these lifespan effects do indeed reflect slower aging is under active investigation (Chen et al., 2009; Wilkinson et al., 2012). Here, we report strong evidence for lifespan extension by ACA, evidence for a possible effect of EST, and strong confirmatory evidence for benefits from NDGA. Surprisingly, each of these three agents extends lifespan either in males only or (for ACA) much more strongly in males than in females. These data thus provide new systems for exploring questions about sex-specific processes that regulate lifespan and late-life illnesses.
Lifespan studies provide a valuable screening tool to identify agents (e.g., drugs or diets) that might slow aging and thereby delay the onset of the wide range of age-related diseases, including those that lead to death. It is, however, a realistic possibility that an agent might extend lifespan by an effect on a specific kind of lethal illness, such as the neoplastic diseases that are the usual cause of death in laboratory mouse stocks. Further work on a range of age-sensitive changes, in many cells and tissues, are needed to determine whether NDGA, ACA, or EST are extending longevity by retarding basic mechanisms of aging, by postponing death from multiple forms of neoplasia, or by a combination of these. The ITP protocol uses genetically heterogeneous mice to minimize the risk of confusion due to unique characteristics of single inbred or isogenic genotypes. Further studies of ACA, EST, and NDGA in a wide variety of mouse stocks will be of high interest, to identify pathways and genetic contributors to the lifespan extension effects.
One possible reason for the larger or exclusive effects of these compounds on males is the short lifespan of the male controls at two of the three sites (medians of 704, 807, and 924 days from male controls at UT, TJL, and UM, respectively, while female control medians are 864, 918, and 887, Table 1). As in all previous ITP cohorts (Strong et al., 2008; Harrison et al., 2009; Miller et al., 2011), unknown site-specific differences lead to male control mice living longer at UM than at UT and TJL, without producing corresponding discrepancies among female controls. We do not know why this occurs, and we do not know whether the factors that distinguish TJL from UM males are the same as those that distinguish UT from UM males. If we knew the basis for the site-specific effect on control lifespan, and why it affects only males, we would be able to test the idea that the stronger (proportional) response of males to interventions such as ACA at TJL and UT represented a correction of the postulated sex-and-site-specific factor(s). We cannot rule this idea out, but we also cannot test it without a plausible hypothesis as to the nature of the factor(s) involved. The data at hand suggest that the situation may be complex and not attributable to a single factor.
Acarbose (ACA). We expected ACA to mimic many effects of diet restriction (DR). The diet used in these studies, Purina 5LG6 diet, has high starch and low sucrose. We expected that ACA, by slowing complex carbohydrate digestion, might limit absorption of sugars and simple carbohydrates, thus leading to lower overall caloric intake and producing effects similar to those of DR. DR at 2/3 normal food consumption started at 4–5 weeks of age increased survival of UM-HET3 mice, with mean lifespan increasing 40% (836–1169 days) in females and 32.5% (831–1101 days) in males; DR decreased body weight to about half in both sexes (Flurkey et al., 2010). Thus, we expected that ACA might produce some survival benefits, although not as much as produced by DR, because the weight reductions from ACA (Fig. 4) were much less than those seen in DR mice. In addition, as weight reductions (% body weight) due to ACA were greater in females than in males (Fig. 4), we hypothesized that lifespan effects would, similarly, be larger in females. The results were thus surprising: median lifespan was increased 22% in males and only 5% in females (Fig. 1).
The fact that female control lifespans do not differ at the three sites, but are still increased significantly by ACA, suggests that ACA may affect basic mechanisms of aging that affect survival in both sexes. In addition, it is noteworthy that the effects of ACA in male mice at UM, where male controls are longest-lived, approached or met conventional criteria for statistical significance (P = 0.054 for log-rank test; P = 0.04 for Wang/Allison test of 90th percentile survival) despite the limited statistical power at any single test site. The idea that ACA has effects on aging and survival independent of site-specific environmental factors is further supported by the fact that in both males and females, maximum lifespan (90th percentile) is increased significantly and similarly in the pooled data for both males (11%, P < 0.001) and females (9%, P = 0.001) (Table 1). Maximum lifespan often indicates effects on mechanisms of aging more reliably than mean or median lifespan, as the latter may be altered by early deaths unrelated to aging.
Acarbose reduced weight more in females than in males (Fig. 4). Thus, the lengthened survival for ACA-treated males vs. ACA-treated females cannot be explained by changes in body weight or seen simply as the effect of overall caloric restriction. It is not currently known whether slower weight gain is a direct effect of loss of caloric content absorbed or represents a modulation of CNS and gastrointestinal endocrine circuits, perhaps with modification of appetite and associated metabolic set points. The data in Fig. 3 provide an initial indication of key physiological parameters in young adult ACA-treated mice. The lower fasting glucose values, in combination with the unaltered HbA1c levels, are consistent with the idea that ACA may diminish the amplitude of postprandial spikes in plasma glucose levels, with lower peak levels but higher trough levels in both sexes. In both blood levels of FGF21 and activity, effects of ACA differed from effects of DR (Fig 3C,F), and thus benefits of ACA on lifespan may not be attributed simply to diminished caloric intake. The male-specific decline in fasting insulin level hints that this altered diurnal pattern of glucose spikes may produce higher insulin sensitivity in males, with less effect in females, which might in turn contribute to the sexual dimorphism in longevity effect. Furthermore, differences between ACA and CR may be due to carbohydrate vs. total diet restriction or to changes in the microbiome. Each of these ideas warrants more thorough evaluation at different ages and in different tissues involved in glucose homeostasis. ACA-treated mice also had a mild reduction in plasma IGF1 levels, of equivalent size in both sexes. Yuan et al. (2012) showed a strong association among mouse strains between levels of IGF1 at 6 months of age and lifespan, but this association disappeared when IGF1 was tested at 12 and 18 months of age. There is some evidence that IGF1 levels at 6 months may predict lifespan in F1 hybrids (Harper et al., 2006) as well as in UM-HET3 mice, where low levels at 15 months predicted increased lifespan (Harper et al., 2003).
17 α-estradiol (EST) was suggested as an anti-aging intervention on the grounds that it might mimic, in male mice, the beneficial effects produced by estrogen in control females and might avoid the serious adverse effects of chronic estrogen treatment mediated by activation of estrogen receptors. Thus, rather than using hormonally active estrogen, we used a nonfeminizing estrogen with reduced binding for estrogen receptors. This class of nonfeminizing estrogen has shown many health benefits (Dykens et al., 2005; Liu et al., 2005; Perez et al., 2005), and non-feminizing estrogens should not cause deleterious effects seen in chronic estrogen therapy (Wang et al., 2006). EST feeding was started at 10 months of age in mice, at the age when estrus cycling and reproduction begin to slow down in long-lived females (Nelson et al., 1982). Evidence for effects on aging was inconclusive. EST feeding had no effect on lifespan in females, as expected. Although the pooled data showed a significant longevity benefit in males, the magnitude of this effect was far stronger at UT than at the other two test sites (Table 1, Fig. 5). Interestingly, weights were affected by EST to a similar degree at TJL and UT (Fig. S2), although a large increase in survival only occurred at UT. Thus, the reduction in weight in males does not consistently explain the great effect of EST on male survival at UT. The test of the hypothesis that 17-α-estradiol might improve health in males thus gives equivocal results—a dramatic benefit in males at one site but no significant benefits at the other two sites. More work will be needed to see whether or not the EST result is a site-specific observation. If the health benefits males at UT are replicated, perhaps by other groups, on other stocks, or at higher doses of 17-α-estradiol, this should motivate follow-up studies to learn which cell type(s) and which estrogen-dependent/independent events are responsible for the benefit.
Nordihydroguaiaretic acid (NDGA) has been shown in a previous cohort to increase lifespan in male but not female UM-HET 3 mice (Strong et al., 2008). We tested it a second time because NDGA serum concentrations were, in the original study using 2500 ppm NDGA, much lower in females than in males, which might explain the absence of lifespan benefit in females. In the current study, we evaluated a higher dose of NDGA (5000 ppm) in an attempt to increase blood levels to see whether these would increase female lifespan. The blood level was indeed increased in females receiving 5000 ppm to a value similar to that in males receiving 2500 ppm NDGA (Fig. S5), but we saw no evidence for an increase in female lifespan with 70% of the survival curve complete (Fig. 6). Effects in males (Fig. 6) were similar to those reported in our earlier study (Strong et al., 2008), and the medium dose of NDGA used previously was the most consistent at increasing lifespan (Figs. 6 and S4). The new data confirm the original report in an independent cohort, show that the lack of effect in females is not due simply to altered pharmacodynamics, and provide helpful information on the dose–response relationship. Although the current NDGA survival experiment is not complete, it is apparent that NDGA, such as ACA and EST, has a significant benefit in males but not in females. In males, the low and middle doses did not affect body weight, but the high dose caused a small reduction (Fig. S6). The high dose reduced female body weight about 14%, 22%, and 23% at 12, 18, and 24 months, respectively, while male weight was only reduced 3–6% (Fig. S6B); thus, the lack of effect of NDGA on female lifespan cannot be explained by concentration in the plasma or by effects on body weight. It is possible that NDGA produces beneficial effects in both sexes but produces harmful effects in females that prevent lifespan extension. Further data, in both sexes, on age-dependent changes, including those related to overall health, will be needed to determine whether NDGA modulates multiple aspects of aging in either or both sexes. It is noteworthy that several of the agents found to extend lifespan in the ITP series, that is, aspirin, NDGA, and rapamycin, inhibit some forms of inflammatory response, consistent with suggestions that late-life inflammation is an important aging process (Jenny, 2012). A mutation that blocks production of the pro-inflammatory cytokine MIF (migration inhibition factor) also extends mouse lifespan (Harper et al., 2010).
Methylene blue (MB) extends proliferative lifespan of human embryonic fibroblasts (IMR90) and also increases the activity of mitochondrial complex IV as well as mitochondrial heme synthesis in these fibroblasts; it reverses accelerated senescence in vitro due to oxidative stress and induces antioxidant defense enzymes in HepG2 cells (Atamna et al., 2008). Despite this, and the protection against irradiation, brain damage, poisoning, and chemotherapy in vitro, MB did not lead to strong or consistent effects on lifespan in UM-HET 3 mice at the dose used, although it did lead to a statistically significant improvement in female maximum lifespan (Fig. 5F), as measured by the proportion of surviving mice at the 90th percentile age in the pooled data set. Specific effects on median lifespan at a single site suggest damage in males (TJL) or benefits in females (UM), but these effects disappear in pooled data (Table 1). There is a 6% increase in female maximum 90% lifespan that is significant (Table 1). However, because the effect is so small, and limited to one sex, the data fail to support the hypothesis that MB is an effective anti-aging intervention at the dose used.
We evaluated slope and intercept of the Gompertz distribution for males and females exposed to EST, ACA, and MB as previously described (Miller et al., 2011). For females, pooling across sites, there were no significant effects of any of these agents on the slope or intercept coefficients. For males exposed to ACA, slope is increased and intercept is decreased, although interpretation of these effects is complicated by the high sensitivity of both parameters to early deaths whose relationship to aging processes is obscure. Neither MB nor EST led to a significant alteration in either Gompertz parameter.
We believe our new data provide several important steps toward the understanding of aging mechanisms and the search for clinical strategies to retard chronic diseases. ACA represents the second agent, among those tested by the ITP, that significantly increases both male and female lifespan. Compared with rapamycin, ACA produces larger median effects and similar 90th percentile effects in males; effects in females are much less than produced by rapamycin (Harrison et al., 2009; Miller et al., 2011). The ACA result draws attention to the possibility that transient fluctuations in plasma glucose, thought to be blunted in ACA-treated mice, may play a role in aging and in cancer biology. Comparisons of ACA to various food restriction protocols are likely to be quite informative: both ACA and DR lead to longevity, some change in body mass (much more severe in DR mice), and alterations in glucose/insulin physiology, but the two protocols differ in fasting glucose, stimulation of spontaneous activity and perhaps other CNS responses, post-prandial glucose pulse height, and in the sexual specificity of the lifespan response. Our data show that drugs can lead to dramatically different effects on lifespan in the two sexes, with ACA, NDGA, and (perhaps) EST showing benefits in males, and rapamycin showing consistently stronger effects in females (Harrison et al., 2009; Miller et al., 2011, and unpublished results). The hints of benefits from MB suggest that MB might lead to lifespan and anti-aging benefits using other doses or alternate dosage schedules or both. More generally, the growing arsenal of drugs that extend lifespan, perhaps by modulation of aging, cancer, or both, will complement work performed using mutant stocks and dietary interventions to delineate the factors that control aging rate in mammals and link aging to multiple forms of illness.
Mouse production, maintenance, and estimation of lifespan
UM-HET3 mice were produced at each of the three test sites as previously described (Strong et al., 2008; Harrison et al., 2009; Miller et al., 2011), where environmental conditions are presented in detail. The dams of the test mice were CByB6F1/J, JAX stock #100009 (dams, BALB/cByJ; sires, C57BL/6J). The sires of the test mice were C3D2F1/J, JAX stock #100004 (dams, C3H/HeJ; sires, DBA/2J). In each site, breeding mice were fed LabDiet® 5008 mouse chow (PMI Nutritional International, Bentwood, MO, USA). As soon as mice were weaned, they were fed LabDiet® 5LG6 from the same source.
Details of the methods used for health monitoring were provided previously (Strong et al., 2008; Harrison et al., 2009; Miller et al., 2011). In brief, each of the three colonies was evaluated four times each year for infectious agents. All such surveillance tests were negative for pathogens at all three sites throughout the entire study period.
Removal of mice from the longevity population
Mice were removed from the study because of fighting or accidental death (e.g., during chip implantation) or chip failure, or because they were used for another experimental purpose, such as testing immune responses. For survival analyses, all such mice were treated as alive at the date of their removal from the protocol and lost to follow-up thereafter. These mice were not included in calculations of median longevity. Overall, 3–5% of the mice were removed from the longevity populations reported here, with no significant site differences, except that about 20% of female controls were removed at TJL to be used for a separate investigation.
Estimation of age at death (lifespan)
At UM and UT, mice were examined daily for signs of ill health from the time they were set up in the experiment. At JAX, once mice were marked as ill, they were examined daily for signs of ill health. Mice were euthanized for humane reasons if so severely moribund that they were considered, by an experienced technician, unlikely to survive for more than an additional 48 h. A mouse was considered severely moribund if it exhibited more than one of the following clinical signs: (i) severe lethargy, as indicated by reluctance to move when gently prodded with a forceps; (ii) inability to eat or to drink, for example due to a severe balance or gait disturbance; (iii) rapid weight loss over a period of 1 week or more; or (iv) a severely ulcerated or bleeding tumor. The age at which a moribund mouse was euthanized was taken as the best available estimate of its natural lifespan. Mice found dead were also noted at each daily inspection.
Control and experimental diets
TestDiet®, Inc., a division of Purina Mills (Richmond, IN, USA), prepared batches of LabDiet® 5LG6 food containing each of the test substances, as well as control diet batches, at intervals of approximately 4 months, and shipped each batch of food at the same time to each of the three test sites. Acarbose was purchased from Spectrum Chemical Mfg. Corp., Gardena, CA. It was mixed at a concentration of 1000 mg of ACA per kilogram of diet (1000 ppm); mice were fed continuously from 4 months of age. 17 α-estradiol (EST) was purchased from Steraloids, Inc. (Newport, RI, USA) and mixed at a dose of 4.8 mg kg−1 of diet (4.8 ppm); mice were fed continuously from 10 months of age. Methylene blue (MB) was purchased as methylene blue hydrate (Cat # 66720) from Sigma-Aldrich (St Louis, MO) and mixed at 28 mg kg−1 of diet (28 ppm); mice were fed continuously from 4 months of age. Nordihydroguaiaretic acid (NDGA) was purchased from Cayman Chemicals (Ann Arbor, MI, USA) and was mixed at concentrations of 800, 2500, or 5000 mg kg−1 of diet (800, 2500, 5000 ppm); mice were fed continuously from 6 months of age. Lifespan data for NDGA are not complete because these studies were started a year later than the others reported. NDGA treatment was in Cohort 2010 mice, while ACA, EST, and MB treatments were included in Cohort 2009.
Measurement of NDGA
NDGA was quantified in mouse blood using HPLC with electrochemical detection. Briefly, 100 μL of calibrators, controls, and unknown samples were mixed with 0.9 mL of a 50% ACN/25% IPA/25% EtOH/0.1% ascorbic acid solution. The samples were vortexed vigorously and then centrifuged at 3200 g for 20 min. Supernatants were transferred to glass test tubes and dried to residue under a gentle stream of nitrogen. The residues were redissolved in 100 μL of mobile phase and then filtered using a microfilterfuge tube. Then, 50 μL of the final samples was injected into the HPLC-EC system. The peak area of NDGA was compared against a linear regression of peak areas of calibrators at concentrations of 0, 10, 50, 100, 500, and 1000 ng mL−1 to quantify NDGA in the blood samples. NDGA concentration in blood was reported in ng mL−1.
The HPLC-electrochemical system consisted of an Alltima C18 column (4.6 × 150 mm, 3 micrometer) heated to 36 °C, ESA Coulochem II Detector with 5011 detector cell, Waters 717 autosampler, and a Waters 515 HPLC pump. The mobile phase was 50% MeOH, 49.5% Milli-Q water, 0.1 mm EDTA, and 0.5% phosphoric acid (pH 2.5). The flow rate of the mobile phase was 0.75 mL min−1, and the detector settings were E1: +450 mV, R1: 100 nA, E2: −350 mV, R2: 100 nA, Guard Cell: +300 mV.
Each mouse originally entered into the study was, at the time of analysis, considered to be in one of two categories: either dead (from natural causes) or censored. Mice were censored at the age when they were no longer subjected to the mortality risks typical of unmanipulated mice. In some cases, this was because the mouse was removed because of fighting; in other cases, mice died as the result of an accident (e.g., death when anesthetized for implantation of a radio-emitting chip). In still other cases, mice were considered censored on the day in which they received an experimental treatment (such as blood sampling or tests of immune response) to which the control mice were not exposed. Kaplan–Meier analysis and log-rank comparisons among groups considered censored mice to be lost from follow-up on the day at which they were removed from the longevity protocol. There were no mice remaining alive in the ACA, EST, and MB groups at the time of the analyses reported here. Only about 70% of the mice given NDGA and controls had died when these analyses were performed, so only median data are reported.
Unless stated otherwise, all significance tests about survival effects are based upon the two-tailed log-rank test at P < 0.05, stratified by test site, with censored mice included up until their date of removal from the longevity population. Other statistical tests are described in the text; all P-values are two-tailed and reported without adjustment for multiple comparisons. Statistical claims related to maximum lifespan are based on the procedure of Wang et al. (2004), which uses the Fisher exact test to compare the proportions of surviving mice, in Control and Test groups, at the age corresponding to the 90th percentile for survival in the joint distribution of the Control and Test groups together. For the pooled data sets, surviving mice were enumerated at the 90th percentile age for each site separately, and these counts were combined for the overall Fisher exact test.