Sci. Aging Knowl. Environ., 18 September 2002
Vol. 2002, Issue 37, p. pe14
[DOI: 10.1126/sageke.2002.37.pe14]


Mitigating the Tithonus Error: Genetic Analysis of Mortality Phenotypes

Scott D. Pletcher

The author is in the Department of Biology, University College London, London WC1E 6BT, UK, and the Huffington Center on Aging, Baylor College of Medicine, Houston, TX 77030, USA. E-mail: s.pletcher{at};2002/37/pe14

Key Words: aging • mortality analysis • chicoDrosophila melanogaster • biodemography

Eos, the Greek goddess of the dawn, asked Zeus to grant a prayer that her lover, Tithonus, be made immortal. Unfortunately, she forgot to ask for eternal youth, and Tithonus grew forever older and more shriveled (1).

A few years back, one of the architects of the modern view of the biology of aging, George C. Williams (see Williams Classic Paper), argued that, like Eos and Tithonus, researchers were mistakenly equating death with aging and that most research on aging was wrongly focused on the study of life-span (2). The real object of interest was senescence: the age-associated pattern of decline in physiological functions that eventually leads to the failure of individual organisms. For mammalian systems this criticism was perhaps a bit too harsh. Outward signs of aging, such as graying hair, slowing reaction time, and increasing cancer rates, are readily observed, and their age progression has been studied for many years. For researchers who study other models systems such as Saccharomyces cerevisiae, Caenorhabditis elegans, or Drosophila, however, such an argument would have landed uncomfortably close to home. Phenotypic characteristics of normal aging in these organisms were not well documented, and genetic studies often described mutations exclusively by their effect on the average or maximum life-span of a particular strain. New and exciting results derived from careful observation of cellular (3) and systemic (4) deteriorations in aging organisms threaten to change this way of thinking. For the time being, however, summarizing age at death in terms of the mean or maximum remains the most popular way to describe genetic effects on aging.

This is unfortunate, because, despite the dearth of phenotypic markers of aging in many model organisms, the ability to study the genetics of senescence, or of the aging process per se, has been there all along. Generally speaking, death is not a programmed event in an organism's life history. As physiological systems break down with advancing age, whether in a characteristic or a random fashion, the ability of an individual to avoid death through any number of different causes diminishes. Observing large cohorts of same-age genetically identical individuals allows an estimation of the age-dependent risk of death--also called the age-dependent death rate or age-specific mortality-- that is representative of that genotype (5). Changes in age-specific mortality reflect underlying physiological deterioration and provide a phenotype that is ideal for the analysis of aging (Fig. 1). This approach is nothing new. Statisticians, reliability engineers, demographers, and even a few biogerontologists (6) have employed it for years. What is new, and promises to grow increasingly useful, is the application of molecular genetic methods to the mortality phenotype.

View larger version (24K):
[in this window]
[in a new window]
Fig. 1. Age-specific mortality rates for three large cohorts of female Drosophila melanogaster cultured at different temperatures (20). The y axis is presented on the log scale, indicating that the risk of death increases roughly exponentially with age for the three treatments. Large sample sizes (on the order of several hundred individuals) are usually required for accurate estimation of the mortality trajectory, but parametric models (that is, statistical models that are described by only a few parameters), such as the Gompertz model, can be used with smaller numbers. The Gompertz model (solid lines) describes mortality as increasing exponentially with age: µ(x) = aebx, where µ(x) is the mortality function, a is the initial mortality rate, and b is the rate of increase in mortality with age (8). On the log scale, the Gompertz is a straight line, and log(a) is often called the intercept and b is the slope. The age-dependent rates of change of these curves are indicative of physiological decline and are used to characterize an aging phenotype. Even for this rather crude example, the effect of temperature on aging is clear: The decreased life-span of the high temperature cohort is brought about by an increase in the rate of aging (8).

Take, for example, a comparison between longevity extension brought about by a mutation in the gene Indy with similar extension generated by laboratory selection for increased life-span. Both have been shown to increase mean longevity in flies by roughly 100% (7, 8). Rogina and colleagues suggested that the Indy mutation might be involved in nutrient uptake through its likely role as a co-transporter mediating the absorption of metabolites and intermediary metabolism (7). Specific genes directly responsible for long life in the selection lines have yet to be identified (8). Comparable effects on mean longevity might suggest similar aging phenotypes, but casual examination of the mortality patterns that characterize each of these treatments reveals very different patterns of senescence (Fig. 2). The Indy mutation extends life-span by what is clearly a slowing of the rate of increase in mortality (that is, risk of death) with age. Although the mutant flies begin to senesce soon after emergence from pupal stage, they age more slowly throughout life, with mortality doubling approximately every 5.7 days in control flies and every 10 days in the Indy mutant flies. Laboratory selection for increased life-span, achieved by collecting eggs from progressively older adults (9), generates a qualitatively different effect. Selected flies apparently delay aging until much later (Fig. 2B). It is not until roughly 40 days after emergence that demographic signs of aging (i.e., increasing death rates) are observed; afterward, mortality doubles at the same rate (maybe even slightly faster) in flies that had been selected for long-life as in their corresponding controls. This is just one of many possible examples illustrating how similar effects on mean longevity can be brought about by different patterns of aging. Clearly, a characterization of the mortality phenotype is crucial for describing the effects of single gene mutants on aging, but few studies provide this information.

View larger version (17K):
[in this window]
[in a new window]
Fig. 2. Comparison of changes in age-specific mortality brought about by (A) mutation of the gene Indy and (B) laboratory selection for increased longevity. Although both manipulations roughly double average longevity in Drosophila, their effect on the age pattern of mortality is strikingly different. The Indy mutant predominantly slows the rate of mortality increase compared to the wild type (compare the slopes of the fitted lines), whereas laboratory selection seemingly results in a delayed onset of senescent decline. Once mortality rates begin to increase in the selected flies, they do so at a rate that is equal to or faster than that seen in the controls. Data are for Indy females and for laboratory-selected males. Data for Indy were generously provided by B. Rogina and S. Helfand.

In the inaugural issue of the journal Aging Cell, work from the Tatar lab (see Tu et al.) (10) sets out to rectify this situation by describing the mortality phenotype generated by a hypomorphic mutation of chico (ch1), a gene that encodes the fly homolog of the insulin-receptor substrate (11), which functions in an insulin/insulin-like growth factor (IGF) signaling pathway (see Tatar Science Article). Tu et al. (10) used specially designed population cages to culture thousands of same-age male and female flies that were either wild-type, heterozygous for ch1, or homozygous for the mutant allele. They confirmed results previously reported by Clancy et al. (11), that the mutation extends adult life-span in both homozygous and heterozygous females and in heterozygous (but not homozygous) males. On the surface, these results seem to paint a somewhat complicated picture. Why is there no longevity phenotype in homozygous males, when homozygous females (and heterozygous flies of both sexes) are so strongly affected? Potential explanations involving sex differences in reproduction, stress response, insulin/IGF signaling, etc. could be formulated. It turns out, however, that they would all be misguided.

The mortality phenotype of ch1 flies described by Tu et al. (11) clarifies the impact of the reduction of insulin/IGF signaling on the aging process and illustrates the limitations of focusing on average or maximum life-span. While Clancy et al. found that the rate of increase in mortality with age was reduced in heterozygous females, Tu et al. showed that, in both sexes, increasing the number of mutant ch1 alleles further reduces the rate of aging. For example, wild-type females experienced a doubling of their risk of death approximately every 5 days, homozygous females decayed at a rate only half as fast, doubling in mortality every 9 days, and heterozygous females fell in between at 6.3 days. Surprisingly, this trend was also observed in males. Homozygous ch1 males aged significantly more slowly than did wild-type controls, despite the fact that mean life-span did not significantly differ between the two. The conclusion: aging is retarded similarly in both sexes by reduced insulin/IGF signaling.

Two other interesting patterns emerged from the mortality analysis. First, the initial mortality rate in ch1 flies was higher than it was in the control flies. That is, young mutant flies died more often than did young wild-type flies. The high death rates were initially age-independent, which distinguished them from aging-related mortality increases. For females, the corresponding decrease in the rate of aging was sufficient to offset this initial disadvantage, and ch1 females exhibited a higher mean life-span. For males, however, early-age mortality took its toll. Despite a significantly reduced rate of aging, these early deaths were sufficient to lower average life-span to the level of the control and obscure the effect of the mutation on aging. Observations of cultures of ch1 flies in the Partridge lab suggest that homozygotes develop relatively slowly (12), leaving them to cope with a severely deteriorating larval environment (13) depleted by their larger and more rapidly developing heterozygous and wild-type siblings. As pointed out in the Tu et al. report, these pre-adult conditions might leave the ch1 homozygotes frail and more susceptible to environmental hardship.

A second intriguing observation derives from patterns of mortality in the "oldest-old" flies. Roughly a decade ago, considerable excitement was generated by reports published in Science that showed a deceleration (and possibly a decline) of mortality rate in the small fraction of medflies and Drosophila that live to very old ages (14, 15). Mortality deceleration has now been observed in many different species (16), but distinguishing between alternative hypotheses for this observation--a reduction in the rate of aging in individual organisms or selective loss of frail individuals (despite personal mortality continuing to increase exponentially)--has yet to be satisfactorily accomplished. Tu and colleagues suggest that mortality deceleration might be a phenotype amenable to genetic analysis. Wild-type flies and ch1 heterozygotes exhibited the characteristic leveling-off of mortality at older ages. Flies homozygous for ch1, however, did not. Whether this is because the mutation influences physiology at these late ages or because only the most robust homozygous individuals are capable of surviving the harsh larval environment is unknown.

It is worth taking a moment to emphasize that Tu and colleagues incorporated scrupulous controls for genetic artifacts, which have afflicted genetic studies of aging in the past (17). The chico mutation was first integrated into three independent genetic backgrounds and then rapidly backcrossed using a clever mating design. The large number of backcross generations (18) improves on the procedure in Clancy et al. (11), who report five, and provides further evidence that mutation of chico (and not of some closely linked gene) is the cause of the slowed aging. Moreover, the experiments were carried out using progeny of crosses within and between experimental backgrounds. Crosses among isogenic lines creates F1 progeny that are genetically identical, yet less likely to be burdened with the effects of fixed recessive deleterious alleles, which could simply be rescued in the chico mutant strain (see Spencer Perspective). Perhaps the only remaining limitation is the lack of multiple mutant chico alleles, as some genetic replication is desirable.

The power and insight derived from genetic analysis of age-specific mortality will only grow as more data are collected. Integrating mortality data with genetic and cellular studies might eventually allow us to distinguish major classes of longevity enhancing treatments. Contrary to summary measures of life-span, the mortality phenotype extends naturally to the analysis and interpretation of traditional genetic epistasis experiments, where combinations of mutants are used to infer whether genes act in the same pathway or process in the regulation of aging (18). Methods for detecting and quantifying genetic interaction for mortality have been developed (19) and are currently being used to investigate interventions that interact with the aging process.

Genetic analysis of mortality has one major drawback: Experiments must be large in order to accurately measure the phenotype. For each specific genetic or environmental manipulation, life-span data from hundreds of individuals are suitable for characterizing major trends in age-specific mortality (for example, the rate of increase in mortality with age according to the Gompertz model). But more detailed aspects of aging, such as the extent of mortality deceleration at the oldest ages, require significantly more. These numbers might prove prohibitive for mammalian studies, but they provide the power to study the details of senescence in smaller, less complex model systems, where the face of aging is unknown.

It is perhaps appropriate that in the end Tithonus, himself, after many long years of life as a human being, was turned into an insect. Now, if I could only find him and several hundred of his progeny, throw them into some cages, and figure out what makes them so special . . .

September 18, 2002
  1. T. Bulfinch, The Age of Fable (Review of Reviews, New York, 1913).
  2. G. C. Williams, The 1999 Crafoord Prize lectures: The Tithonus error in modern gerontology. Q. Rev. Biol. 74, 405-415 (1999).[CrossRef][Medline]
  3. L. Guarente, C. Kenyon, Genetic pathways that regulate ageing in model organisms. Nature 408, 255-262 (2000).[CrossRef][Medline]
  4. D. Garigan, A. L. Hsu, A. G. Fraser, R. S. Kamath, J. Ahringer , C. Kenyon, Genetic analysis of tissue aging in Caenorhabditis elegans. A role for heat-shock factor and bacterial proliferation. Genetics 161, 1101-1112 (2002).[Abstract/Free Full Text]
  5. S. D. Pletcher, D. Houle, J. W. Curtsinger, Age-specific properties of spontaneous mutations affecting mortality in Drosophila melanogaster. Genetics 148, 287-303 (1998).[Abstract/Free Full Text]
  6. C. Finch, Longevity, Senescence and the Genome (Univ. of Chicago Press, Chicago, IL, 1990).
  7. B. Rogina, R. A. Reenan, S. P. Nilsen, S. L. Helfand, Extended life-span conferred by cotransporter gene mutations in Drosophila. Science 290, 2137-2140 (2000).[Abstract/Free Full Text]
  8. S. D. Pletcher, A. A. Khazaeli, J. W. Curtsinger, Why do life spans differ? Partitioning mean longevity differences in terms of age-specific mortality parameters. J. Gerontol. A Biol. Sci. Med. Sci. 55, B381-B389 (2000).[Abstract/Free Full Text]
  9. L. S. Luckinbill, R. Arking, M. J. Clare, W. C. Cirocco, S. A. Buck, Selection for delayed senescence in Drosophila melanogaster. Evolution 38, 996-1003 (1984).[CrossRef]
  10. M.-P. Tu, D. Epstein, M. Tatar, The demography of slow aging in male and female Drosophila mutants for the insulin-receptor substrate homologue chico. Aging Cell 1, 75-80 (2002).[CrossRef][Medline]
  11. D. J. Clancy, D. Gems, L. G. Harshman, S. Oldham, H. Stocker, E. Hafen, S. J. Leevers, L. Partridge, Extension of life-span by loss of Chico, a Drosophila insulin receptor substrate protein. Science 292, 104-106 ( 2001).[Abstract/Free Full Text]
  12. W. Mair, personal communication.
  13. A. M. Leroi, A. K. Chippindale, M. R. Rose, Long-term laboratory evolution of a genetic life-history trade-off in Drosophila melanogaster. 1. The role of genotype-environment interaction. Evolution 48, 1244-1257 (1994).[CrossRef]
  14. J. R. Carey, P. Liedo, D.Orozco, J. W. Vaupel, Slowing of mortality rates at older ages in large medfly cohorts. Science 258, 457-461 (1992).[Abstract/Free Full Text]
  15. J. W. Curtsinger, H. H. Fukui, D. R.Townsend, J. W. Vaupel, Demography of genotypes: failure of the limited life-span paradigm in Drosophila melanogaster. Science 258, 461-463 (1992).[Abstract/Free Full Text]
  16. J. W. Vaupel, J. R. Carey, K. Christensen, T. E. Johnson, A. I. Yashin, N. V. Holm, I. A. Iachine, V. Kannisto, A. A. Khazaeli, P. Liedo, V. D. Longo, Y. Zeng, K. G. Manton, J. W. Curtsinger. Biodemographic trajectories of longevity. Science 280, 855-860 (1998).[Abstract/Free Full Text]
  17. M. Tatar, Transgenes in the analysis of life span and fitness. Am. Nat. 154, S67-S81 (1999).[CrossRef]
  18. D. Gems, S. Pletcher, L. Partridge, Interpreting interactions between treatments that slow aging. Aging Cell 1, 1-9 (2002).[CrossRef][Medline]
  19. S. Pletcher, L. Partridge, unpublished data.
  20. S. Pletcher, unpublished data.
Citation: S. D. Pletcher, Mitigating the Tithonus Error: Genetic Analysis of Mortality Phenotypes. Science's SAGE KE (18 September 2002),;2002/37/pe14

A Reassessment of Genes Modulating Aging in Mice Using Demographic Measurements of the Rate of Aging.
J. Pedro de Magalhaes, L. Thompson, I. de Lima, D. Gaskill, X. Li, D. Thornton, C. Yang, and D. Palmer (2018)
Genetics 208, 1617-1630
   Abstract »    Full Text »    PDF »

Science of Aging Knowledge Environment. ISSN 1539-6150