The Journals of Gerontology Series A: Biological Sciences and Medical Sciences 57:B390-B398 (2002)
© 2002 The Gerontological Society of America
Different Age-Specific Demographic Profiles Are Generated in the Same Normal-Lived Drosophila Strain by Different Longevity Stimuli
Robert Arkinga,
Janna Novoseltsevab,
Dae-Sung Hwangboa,
Vassily Novoseltsevb and
Mark Lanec
a Department of Biological Sciences, Wayne State University, Detroit, Michigan
b Institute of Control Sciences, Russian Academy of Sciences, Moscow, Russia
c Nutritional and Molecular Physiology Unit, National Institute on Aging/National Institutes of Health, Baltimore, Maryland
Robert Arking, Wayne State University, College of Science, Department of Biological Sciences, Biological Sciences Building, Detroit, MI 48202 E-mail: rarking{at}biology.biosci.wayne.edu.
Decision Editor: James R. Smith, PhD
 |
Abstract
|
|---|
We review the empirical data obtained with our normal-lived Ra control strain of Drosophila and show that this one genome is capable of invoking at least three different responses to external stimuli that induce the animal to express one of three different extended longevity phenotypes, each of which arises from one of three different antagonistic molecular mechanisms of stress resistance. The phenotypes are distinguished by different age-specific mortality patterns. Depending on the selected mechanism, the genome may respond by expressing a delayed onset of senescence (type 1), an increased early survival (type 2), or an increased late survival (type 3) phenotype, suggesting their different demographic effects. We suggest that the different demographic effects stem in part from the differential ability of each selection regime to reallocate the organism's energy from reproduction to somatic maintenance. These data document the complexity of the aging process and argue for a relationship between molecular mechanisms and longevity phenotypes.
WE used artificial selection to generate long-lived strains of Drosophila, with the eventual goal of comparing them to their normal-lived progenitor strains in order to deduce what processes had been significantly altered and therefore might play an important role in the expression of the extended longevity phenotype (1). We focused our attention on one particular normal-lived wild type of strain, Ra, and have used it in three different types of experiments designed to modulate longevity (2)(3)(4), as summarized in the paragraphs that follow. These experiments show that there are multiple ways for this one wild-type organism to live longer. The different experimental procedures somehow allow the organism to selectively express certain molecular mechanisms and not others. The utilization of qualitatively different molecular mechanisms selectively modulates different parts of the adult life stages by altering age-specific mortality patterns. Depending on the selected mechanism, this may lead to a delayed onset of senescence (type 1), an increased early survival (type 2), or an increased late survival (type 3) phenotype, suggesting their different demographic effects. These data argue that aging, far from being simple, is characterized by multiple phenotypes, each of which is the outcome of a set of related mechanisms. Johnson and colleagues (5) analyzed single gene mutants in Caenorhabditis elegans and reached a similar conclusion. Our work extends their conclusion by demonstrating the genomic plasticity of one wild-type strain that, by itself, encompasses most of the extended longevity phenotypes known in the species.
|
Demographic Aspects of the Weibull Distribution
|
|---|
We base our analysis of these three phenotypes primarily on strain- and age-specific survival and mortality patterns. Pletcher and colleagues (6) have presented a cogent summary of the usefulness of these two parameters (particularly the latter) in the analysis of aging mechanisms. Wilson (7) has shown that survival data may be fitted by different distributions following changes caused by genetic and environmental modification. Vanfleteren and colleagues (8) have demonstrated that fitting survival data to these distributions may provide valuable information used to unravel the mechanisms of aging. Therefore, in our analysis, mechanistic molecular arguments play a secondary role, although they are of great explanatory value in understanding the processes underlying the expression of each phenotype.
The observed survival data and the experimental manipulations that produced them are discussed in detail in the paragraphs that follow. The classic Gompertz model was shown to be incapable of detecting the mortality deceleration observed at advanced ages in most species (9)(10). As a result, controversial opinions about what models and techniques would adequately represent survival and mortality data in animal experiments were formulated during the past decade. We prefer the Weibull distribution (Fig. 1) for the fitting of survival patterns. This model allows mortality increases to weaken at advanced ages, thus yielding a better curve fitting of the experimental survival experimental data than does the Gompertz function (8). To show that the Weibull model is a reasonable model for the survival data, we compare it with other well-known approximations. The results of the comparison confirm the premise (see the Appendix for details).
The Weibull survival is described by a distribution
(1)S(x ) = exp[-(x/a)b]
with two parameters, a and b, which can be clearly interpreted in demographic terms. An a parameter is a characteristic lifetime of a pattern, namely the age, at which 36.8% of the initial population is still alive. Its value highly correlates with a mean population life span. The other parameter, b, is a shape parameter that defines a relative steepness of a survival pattern at a characteristic age, a. Namely, the steepness at this age is
(2)dS(x)/dx | x = a = -0.368 (b/a).
The experimental data was fitted to the Weibull model by using standard Matlab least-mean-square techniques (11). The parameters of the Weibull approximations of the survival data for the seven strains are presented in Table 1 . Fig. 2 presents both the observed survival data and the Weibull approximations of the experimental data.
View larger version (21K):
[in this window]
[in a new window]
|
Figure 2. A, Survival curves of the normal-lived Ra strain and of two long-lived strains (La and 2La) sequentially derived from it by a direct selection for delayed female fecundity. The data points represent the observed survival data and are based on the age-specific values obtained from two or three replicate cohorts consisting of at least 250 mixed-sex individuals each. The Ra, La, and 2La curves are significantly different (log-rank test = 530.16, df = 2, p < .0001). See Arking and colleagues (2) for experimental details. The continuous lines are the Weibull approximations, with the parameters presented in Table 1 , of the empirical data. Note that the control and experimental curves have different characteristic ages and slopes at these ages. B, Survival curves of the normal-lived Ra strain and the paraquat-resistant strain selected from it by direct selection for paraquat resistance. The data points represent the observed survival data and are based on the age-specific values obtained from mixed-sex cohorts of 250–450 animals each. The two curves are significantly different (log-rank test = 24.76, df = 1, p < .00005). See Vettraino and colleagues (3) for experimental details. The continuous lines are the Weibull approximations, with the parameters presented in Table 1 , of the empirical data. Note that the slopes of the two survival curves at their characteristic ages (43.12 and 49.78 days) are very close to each other (-0.0338 and -0.0390). C, Survival curves of the normal-lived Ra control strain and the longer-lived Ra heat-treated strain. The animals were subjected to a nonlethal heat shock (37°C for 90 minutes) early in life at days 5–7 after eclosion. They were then maintained under controlled optimal conditions and their survival was monitored. The two curves are significantly different (log-rank test = 17.84, df = 1, p < .00005). See Kuether and Arking (4) for experimental details. The continuous lines are the Weibull approximations with the parameters presented in Table 1 . Note that the slopes of the two survival curves at the corresponding characteristic ages (38.42 and 40.63 days) are distinctly different (-0.0347 vs -0.0284).
|
|
The logarithmic mortality patterns directly calculated from the Weibull approximating survival curves are shown in Fig. 3, together with the experimental data. Again the question arises whether this presentation is adequate, and if it gives any advantages in comparison with other well-known models, especially the Gompertz model. Indeed, the Weibull presentation of logarithmic mortality data does not allow a transparent description of mortality comparable with the widely used Gompertz-based parameters, an intercept and a slope. That is why we combine two models, Weibull and Gompertz (Fig. 3 and Fig. 4), in our mortality analysis.
View larger version (22K):
[in this window]
[in a new window]
|
Figure 3. A, Age-specific logarithmic mortality curves of the normal-lived Ra strain and of two long-lived strains (La and 2La) sequentially derived from it by a direct selection for delayed female fecundity. The data points represent the Weibull approximations and the continuous lines are calculated from the fitted Weibull approximations shown in Fig. 2. Note the long delay in the onset of mortality. B, Age-specific mortality curves of the normal-lived Ra strain and the paraquat-resistant strain selected from it by direct selection for paraquat resistance. The data points represent the Weibull approximations and the continuous lines are calculated from the fitted Weibull approximations shown in Fig. 2. A difference between the mortality curves exists mostly at early ages but vanishes with advancing age. C, Age-specific mortality curves of the normal-lived Ra control strain and the longer-lived Ra heat-treated strain. The animals were subjected to a nonlethal heat shock (37°C for 90 minutes) early in life at days 5–7 after eclosion. They were then maintained under controlled optimal conditions and their survival was monitored. The data points represent the Weibull approximations and the continuous lines are calculated from the fitted Weibull approximations shown in Fig. 2. The mortality curves are identical at early and middle ages but diverge with advancing age.
|
|
View larger version (22K):
[in this window]
[in a new window]
|
Figure 4. Age-specific logarithmic mortality curves and their Gompertz approximations. A, Age-specific logarithmic mortality curves of the normal-lived Ra strain and of two long-lived strains (La and 2La). The data points and their Gompertz approximations are presented (RaG, LaG, and 2LaG). The curves Ra and La (as well as Ra and 2La) are significantly different (p = 0), whereas La and 2La strains have no significant difference. The estimates for intercepts are as follows: for Ra, -7.506; for La, -8.596; and for 2La, -8.318. Confidence intervals (CIs) estimated by Matlab Stats Toolbox are (-6.671, -8.341), (-8.022, -9.170), and (-7.808, -8.828), respectively. The slopes Ra vs La or 2La are significantly different. Namely, for Ra, estimate = 0.120; CI = (0.097, 0.143); La, estimate = 0.076; CI = (0.057, 0.095); 2La, 0.077 and (0.068, 0.086). B, Age-specific logarithmic mortality curves of the normal-lived Ra strain and the paraquat-resistant (PQR) strain. The data points and their Gompertz approximations are shown (RaG and PQRG). Statistical analysis (Matlab Stats toolbox) confirms that the curves Ra and PQR differ significantly (p < .0036), and the intercepts are significantly different, whereas the slopes Ra and PQR have no significant difference. The estimates and CIs for intercepts are as follows: for Ra, -5.799, CI = (-6.148, -5.450); for PQR, -7.203, CI = (-7.701, -6.705). For the slopes, these are as follows: for Ra, estimate = 0.0668, CI = (0.0563, 0.0773); for PQR, estimate = 0.0856, CI = (0.0747, 0.0963). C, Age-specific logarithmic mortality curves of the normal-lived Ra control strain and the longer-lived heat-treated strain (RaHx). The experimental points and the Gompertz approximations (RaG and RaHxG) are shown. Statistical analysis (Matlab Stats toolbox) does not reveal a significant difference between the curves Ra and RaHx (p < .35). The intercepts do not differ significantly. The estimates are -6.114 (Ra) and -5.768 (RaHx). The CIs are relatively (-6.612, -5.616) for Ra and (-6.251, -5.285) for RaHx. The slopes Ra vs PQR also have no significant difference: the Ra slope is 0.0859, CI = (0.0734, 0.0984), and the RaHx slope is 0.0699, CI = (0.0569, 0.0829).
|
|
|
Direct Selection for Delayed Female Fecundity: The Type 1 Phenotype
|
|---|
We produced long-lived strains by using a regime that involved a direct selection for delayed female fecundity and thus an indirect selection for extended longevity (1)(12)(13). The essential characteristic of the resulting long-lived La strain is that it lives a long time because of a delayed onset of senescence brought about by an early and specific up-regulation of the antioxidant defense system (ADS) genes (2)(13)(14). The delayed onset of senescence is characterized by a large difference in the ages at which the control and experimental strains each show a 10% mortality (LT10; i.e., 90% survival) (Table 1 , Fig. 2). The long-lived La strain maintains a low age-specific mortality (Fig. 3) and thus a high survivability for the first 8 weeks or more of its adult life. By this time, its normal-lived Ra progenitor strain has lost more than half of its cohort. The large difference between the two strains is maintained at the LT50 and LT90 ages as well (Table 1 ). There is a large and significant increase in both the mean and the maximum life span of the La strain relative to the Ra strain. A Gompertz model for the mortality data shows that the two types of curves do not differ in the intercepts but have different slopes (Fig. 4). A log of the mortality curve generated from the Weibull function generally confirms this finding (Fig. 3). Thus, the La strain appears to undergo senescence in the same manner as does the Ra strain; it has simply postponed the onset of the senescent phase by approximately 35 days. This type 1 extended longevity phenotype then is characterized by an early-acting significant delay in the age-specific mortality increase.
The response of the ADS system to this indirect selection regime is not a chance event. We repeated the experiment and generated three other long-lived strains (2La, Lb, and 2Lb) that have similar survival and mortality curves (for 2La, see Fig. 2 and Fig. 3) (2)(15). The repeatability of this phenotype suggests that the animal consistently responds in a particular manner to a specific stimulus, indicating that there are mechanistic linkages between stimulus and phenotype. However, we also know that these several phenotypically identical extended longevity strains are accompanied by significantly different metabolic alterations (16) and different patterns of ADS gene expression (17). Different organisms may decrease their reactive oxygen species (ROS) load by upregulating different components of their ADS genes, yet still provide comparable overall levels of defense, and thus show comparable age-specific mortality patterns. Phenotypic equivalence does not imply genotypic equivalence.
The type 1 phenotype arises from an early-acting delayed onset of senescence brought about by an upregulation of the ADS (2)(13). Reverse selection for normal longevity results in a specific downregulation of the ADS genes (2). Single gene mutants specifically affecting ADS expression upregulate or downregulate in the appropriate and expected manner (13). Taken together, these data led us to conclude that there was probably a causal relationship between ADS expression and extended longevity, and that the altered age-specific mortality patterns were likely the outcome of a reallocation of energy from reproduction to somatic maintenance (see the paragraphs that follow). This conclusion derived from our work is consistent with much of the independent research done on Drosophila (18)(19)(20)(21)(22)(23)(24)(25). However, this type 1 phenotype appears to be induced by other mechanisms as well. It is known that genes that downregulate the insulin-like signaling system (ISS) will yield an apparent type 1 extended longevity phenotype in Drosophila (26)(27). This presumably comes about because the downregulation of the ISS negatively regulates the forkhead transcription factor, which in turn negatively regulates the ADS genes, among others. Thus a downregulation of the ISS could result in an upregulation of the ADS genes, thereby yielding the same end result as a direct upregulation of the ADS genes. This suggestion is consistent with our own epistasis data on the ISS system in Drosophila, in which Ra animals heterozygous for a mutation in the insulin-like receptor gene (InR) express a type 1 extended longevity phenotype (28). It has been shown that feeding a particular drug (4-phenybutyrate) to Drosophila can also induce an apparent type 1 phenotype, presumably by inducing the overexpression of superoxide dismutase (SOD) and other genes (29).
Type 1 phenotypes are noted in other organisms. Mice with single gene mutations affecting the hypothalamus–pituitary axis have both survival and mortality data characteristic of a type 1 phenotype (30). SOD/catalase (CAT) mimetics are known to yield a type 1 phenotype in C. elegans (31). Although the molecular mechanism(s) underlying the effects of caloric restriction is not yet definitively known, the physiological and molecular similarities between caloric restriction effects and the operation of the ISS genes in C. elegans are such as to make it very plausible that caloric restriction works through the ISS genes in mammals as well, and this has been formally suggested by Lane (32). However, the diversity of strain-specific responses to caloric restriction in mammals makes it difficult to determine whether a type 1 phenotype is being expressed or not (33)(34)(35).
The type 1 phenotype is the only one of the three phenotypes identified here to involve the delayed onset of senescence with its concomitant increase in both mean and maximum longevities. It is most likely this phenotype that lay authors have in mind when they write about genetically altering the aging process. There does not yet seem to be any obvious human homologue to this particular longevity phenotype, even though it is well known in model systems.
|
Direct Selection for Paraquat Resistance: The Type 2 Phenotype
|
|---|
Given our success with the indirect selection strategy, which gave rise to the La strain, it seemed plausible that it should be also possible to directly select for enhanced antioxidant resistance and indirectly for long life. In view of our prior demonstration that the animal's resistance to exogenous paraquat is an excellent predictor of its antioxidant activity (36) and in view of the fact that the paraquat test is very robust and repeatable, we used the animals' resistance to exogenous paraquat as the direct selection device (3). Using the same Ra strain, we selected them over 24 generations for increased resistance to paraquat (PQR). Selection was successful, with the PQR flies having a fourfold increase in their ability to resist paraquat. Their extended longevity pattern differs from that of the La strain in that they show no sign of a delayed onset of senescence (Fig. 2 and Table 1 ). The PQR flies differ from their progenitor Ra strain only in the fact that the PQR flies have a better early survival (95% vs 91% at 20 days). This enhanced early survival gives rise to moderate but significantly enhanced LT50 and LT90 values (Table 1 ). The increased longevity does not appear to arise from any fundamental alteration in aging processes such as are seen in the type 1 phenotype. It most likely arises as a simple consequence of the decreased early deaths and the subsequently larger than normal number of organisms entering middle age. The age-specific mortality pattern is quite similar in both strains, with the PQR strain being characterized by a lower mortality rate that begins in early life and continues thereafter (Fig. 2). A Gompertz model for the mortality data shows that the two curves have different intercepts but similar slopes (see Fig. 4 for details). A log of the mortality curve generated from the Weibull function (Fig. 3) confirms that the difference existing between the two curves at early ages vanishes at the advanced ones. Thus, it represents an early-acting decrease in the rate of aging that is, however, not maintained into mid-life or late life. The decreased early-life risk results in increases in both the mean and the maximum life span, which are, however, significantly less than those noted in the type 1 phenotype. We term this early-survival-dependent longevity as a type 2 longevity phenotype.
The PQR animals have a lower level of ADS activity and an increased total P450 enzyme activity level relative to controls (3). This pattern of molecular changes is quite different from that observed in the ADS and ISS mutants. We interpret these findings as suggesting that the activation of different molecular mechanisms (i.e., ADS vs P450) during the same (early) phase of the adult life cycle may nonetheless generate different types of extended longevity phenotypes. We suggest in the paragraphs that follow that this might be related to their different effect on energy metabolism.
One example of a human homologue to the type 2 phenotype is the change in the U.S. white female population between 1900 and 1960, which was due almost entirely to the reduction of premature (early) mortality (37). Another example would be the reduced early mortality and increased mean life span that is characteristic of exercising humans versus their sedentary controls (38).
|
Nonlethal Heat Shock and Hormesis: The Type 3 Phenotype
|
|---|
It has been demonstrated in several animal models that a brief nonlethal application of high temperature is capable of inducing an increased longevity (39)(40), possibly by means of an ADS-dependent increase in the expression of the heat-shock protein (Hsp) genes (41)(42)(43)(44). In fact, there is substantial evidence showing that exposure of Drosophila to low doses of a wide variety of environmental insults can significantly increase their longevity (45)(46)(47).
Using the Ra strain, we have demonstrated that a nonlethal heat shock early in adult life causes a significant extension in the life span of the treated animals (4). The data in Fig. 2 and Table 1 show that the LT10 and LT50 values of the control and treated Ra lines are statistically identical, but that the LT90 values are significantly different. We constantly observed such advanced-age divergence in survival patterns in heat-shock experiments (not presented). This divergence can be explained as a result of a slight mortality decrease in the middle-age interval (approximately 35–50 days), which is hard to observe and evaluate in mortality patterns. A log of the mortality curves generated from the Weibull function plotted for the data of Fig. 2 also show that the difference between the two curves arises at advanced ages (Fig. 3). Summarizing the results presented in Fig. 2, Fig. 3, and Fig. 4, we hypothesize that this form of extended longevity is not marked by a delayed onset of senescence as is characteristic of the type 1 phenotype, nor is it distinguished by an enhanced early survival, as is characteristic of the type 2 phenotype. It is characterized by an increase in late-life survival (although the Gompertz analysis does not confirm this observation clearly, because, the confidence intervals overlap; see Fig. 4), and it thus constitutes a different extended longevity phenotype, which we term a type 3 longevity phenotype. Thus, this phenotype arises from an increased late survival, probably arising from a hormetic effect of Hsp or ADS gene expression on presently unknown but presumably critical systems of the organism (39)(42)(43). The effect of mild heat shock on longevity has been well studied in Drosophila and C. elegans, and the available information suggests that mild stress induces the accumulation of stress response proteins, such as molecular chaperones, that retard late aging (43)(47). The increased longevity does not seem to arise from a fundamental alteration of the aging processes such as are seen in the type 1 phenotype, but rather from a slowing down of the late-life age-specific mortality.
A human homologue to the type 3 phenotype is the change in the U.S. white female population between 1960 and 1980, which was due almost entirely to the reduction of late-life mortality (37). Another human example would be the enhanced late survival of centenarians, caused by their compression of morbidity, relative to their control populations as studied by Perls and his colleagues (48)(49). A genetically based human example would be the enhanced late life survival of klotho FV heterozygotes (50).
|
Extended Longevity, Stress Resistance, and Energy Allocations
|
|---|
The disposable soma theory suggests that the manner in which organisms apportion their available energy into reproduction and somatic maintenance will have consequences on their longevity and fecundity (51). We used our empirical data on mean-population longevity, age-related fecundity, and patterns of oxygen consumption and subjected it to mathematical modeling so as to determine if the observed pattern of these traits is close to those theoretically expected by various optimality theories (52). By describing these traits in terms of the ratio of the oxygen consumption spent on reproduction to that spent on maintenance, we were able to calculate the metabolic reproductive efficiency (MRE). We found that the traits experimentally observed in the Ra strain can be regarded as being very close to the theoretically predicted optimum. Thus we have a confirmed theoretical basis for analyzing these animals in terms of energy allocations and MRE. Recent papers illustrating the intimate role of the reproductive system in bringing about certain of these allocations present a possible mechanism (53)(54)(55), but they do not otherwise disturb our view of the MRE as an evolutionary constraint in the Ra female flies.
The data to support the empirical application of this concept to the ability of the Ra strain to express the type 1 longevity phenotype in response to selection is presented in more detail in Buck and Arking (17), whereas a detailed analysis of the energy shifts required to generate an extended longevity phenotype is presented in Arking and colleagues (56). In summary, the latter paper demonstrates the existence of metabolic and mitochondrial differences between normal- and long-lived organisms, and it presents data supporting the observation that the animals selected for extended longevity are both more fecund and longer lived than their progenitor control animals. A revised interpretation, made on the basis of new data, of the events underlying the selection process indicates that there is a two-step change in energy allocations leading to the type 1 phenotype. Initial selection first allows the upregulation of the ADS genes and a shift to the use of the pentose shunt, which is later followed by alterations in mitochondrial fatty acid composition and other changes necessary to reduce the leakage of H2O2 from the mitochondria into the cytosol. The recaptured energy available from the latter step is diverted from somatic maintenance back into reproduction, resulting in animals that are both long lived and fecund.
A literature review suggests that the involvement of mitochondrial and antioxidant changes are probably widespread in the expression of the type 1 extended longevity phenotype by various species. However, this does not mean they are universal. The long-lived O strains of Rose are believed to attain their type 1 survival curve by means of a decrease in the rate of aging (6)(57), not by means of a delay in the onset of senescence as in our La–2La strains. It would be interesting to submit both of these strain sets to a detailed analysis so as to identify the genes and traits associated with each of these two different antiaging strategies. Their characterization might be of general interest because caloric restriction is known to use both strategies to increase life span in different vertebrate and invertebrate species (35).
Direct selection for paraquat resistance yielded a PQR strain with an enhanced detoxification (by means of an enhanced P450 activity), a decreased ADS activity, and a type 2 longevity phenotype. This suggests that the metabolic changes associated with detoxification are not of the sort necessary to bring about a fundamental alteration of energy metabolism. In fact, a detailed comparison of the La metabolic enzyme activity data contained in Table 5 of Arking and colleagues (2) with the PQR metabolic enzyme activity contained in Table 1 of Vettraino and colleagues (3) shows that of the 15 enzymes analyzed, 9 change in qualitatively different directions in the two strains, 5 change in the same direction but have significantly different percent activity changes, and only 1 (phosphoglucomutase) has the same qualitative and quantitative change in both strains. We interpret this post hoc data analysis as strongly supporting the idea that the two different selection regimes imposed on the Ra strain brought about different metabolic alterations, one of which was capable of postponing the age of onset of senescence and inducing the expression of the type 1 extended longevity phenotype; the other was capable only of slowing down the rate of aging. A detailed analysis of all the metabolic changes is presented elsewhere (17).
The molecular mechanisms underlying the expression of the type 3 longevity phenotype are not known with certainty. However, the probable mechanisms discussed herein are quite different from those operative in the type 1 animals and give no indication that the metabolic differences induced by heat shock are at all similar to those known to occur in the type 1 animals.
Although all three longevity phenotypes involve some form of stress resistance, we believe the data presented here strongly suggest that the basic difference between the type 1 and the type 2 or type 3 phenotypes has to do with the wide-ranging energy reallocation and metabolic reorganization undergone by the former but not the latter.
|
Usefulness of Multiple Extended Longevity Phenotypes
|
|---|
The important point of these data is that they demonstrate that one wild-type genome can, when subjected to certain selection pressures, yield at least three different extended longevity phenotypes with different demographic effects. This finding shows that there are multiple ways for this one normal-lived organism to live longer, and that it is the nature of the selective pressure applied that triggers the organism's response.
Different patterns of gene expression underlie each particular extended longevity phenotype. All of these genes are known to have homologues in humans. Only two of these phenotypes are known to exist in humans. It seems reasonable to conclude that the failure to induce the type 1 phenotype in humans represents a transient limitation of our knowledge rather than a permanent limitation imposed by human biology. Predictions that the type 1 phenotype will never be expressed in humans may be regarded as personal opinions based on past events rather than on a scientific analysis of future probabilities.
There are at least six other major benefits to be gained from this sorting out and classifying of different extended longevity phenotypes.
First, it will help test the interactions of the molecular mechanisms involved in longevity extension and foster a better understanding of their limitations and advantages. Such hypothesis testing involves two separate questions: first, Are there multiple mechanisms that give rise to each longevity phenotype and do they represent complementary or antagonistic pathways? and second, Are the mechanism(s) involved in one extended longevity phenotype complementary or antagonistic to those involved in another longevity phenotype? We have data that bear on these two points. The type 1 phenotype can be produced in Drosophila by ADS upregulation (2) or by ISS downregulation (26)(27)(28). Constructing a long-lived ADS animal that is also heterozygous for an InR mutation allows the double mutant animal to live longer than either of its progenitors (28). But subjecting a type 1 long-lived ADS animal to heat shock (type 3) results in a significant shortening of longevity (4). Thus it tentatively seems as if molecular pathways leading to the same longevity phenotype might be complementary whereas those leading to different longevity phenotypes might be antagonistic. Adaptation to express one particular phenotype seemingly inhibits the ability to express another. These data also imply that the type 2 and type 3 phenotypes cannot be viewed as being mere subsets of the type 1 phenotype, a conclusion that is, of course, confirmed by age-specific mortality patterns (Fig. 3 and Fig. 4). Should this preliminary description of pathway relationships be confirmed by future tests, it then might offer an objective means by which to rank certain longevity extension strategies over others.
Second, the phenotypic and genomic plasticity discussed here provides the beginning of a mechanistic basis for the heterogeneity observed in various populations and that is believed to have important implications for longevity and mortality studies (58). The combination of chance developmental variations (59) with the intrinsic genomic and phenotypic plasticity discussed here might one day explain much of the demographic heterogeneity in longevity and aging (60)(61).
Third, the conclusions derived from an analysis of single gene mutations in C. elegans are consistent with these conclusions derived from an analysis of selected strains in Drosophila: namely that extended longevity is brought about by genes with characteristic age-specific mortality patterns and thus particular demographic effects on the affected populations. In addition, we point out that the different age-specific mortality patterns likely arise as a result of the Ra genome's heterogeneous response to selection. Only the type 1 longevity phenotype appears to reallocate energy in such a way as to delay the increase in mortality and thus delay senescence. The other two longevity phenotypes arise by means of an age-specific slowdown in the rate of mortality and thus slow the rate of aging. Aging is not a simple or single phenotype.
Fourth, an awareness of the existence of multiple longevity phenotypes may also improve our estimates of the heritability of longevity in humans. Most studies show this to be relatively modest, with a mean value of 0.2–0.3 and rarely exceeding 0.5 (62). Yet a recent populationwide survey of human longevity found that there were significant reductions in late-life mortality among the offspring of long-lived parents relative to normal-lived parents (48), a finding consistent with a strong hereditary component. Similar findings were reported in a recent mouse study (63) . This apparent contradiction may well be due to the confounding of data from type 1 long-lived families with those of type 2 or type 3 families. Separating the data may also allow us to identify the environmental factors involved in the latter two longevity phenotypes.
Fifth, the organism's different responses to stress yield different types of extended longevity. This conclusion marks a significant modification of the stress theories of longevity (64)(65), for it both imposes limits on the types of competent mechanisms and offers insight into their nature.
Finally, the recognition that there are multiple extended longevity phenotypes will improve our ability to more accurately forecast future changes in human longevity. Some recent demographic forecasts (9)(66) predict limited increases in human longevity. Other nondemographer gerontologists predict much larger increases (67)(68). Why the discrepancy? Both of the demographers limit their analysis to the type 2 and type 3 extended longevity phenotypes known to exist in the human data. Although prudent, this assumes that the future will be just like the past. This may be too conservative. In contrast, the scientists working with model organisms are making their predictions based on the type 1 phenotypes they observe in their genetically altered organisms but that are presently absent in the human data. This assumes that the laboratory data will be able to be translated directly into human events. This may be too optimistic. Knowing, however, that there are at least three different extended longevity phenotypes with distinctly different molecular underpinnings would enable both sets of investigators to clearly see the limitations of their data and their forecasts.
When our understanding of the molecular mechanisms gets to the point where we can significantly alter mammalian nonhuman aging—and some might argue that such a time is rapidly approaching, such as de Grey and colleagues (69)—then the demographers will have to incorporate the type 1 extended longevity phenotype into their mathematical models, or at least into their more speculative ones. Failure to do so might result in a large disparity between predicted and actual longevities. Given the important public policy implications of these demographic forecasts, it is imperative that they be as accurate as the available data permit so that we may have time to adjust to the coming realities.
View larger version (29K):
[in this window]
[in a new window]
|
Figure 5. Four different statistical models used for an age-related survival pattern S(x) of a single nonlethal heat shock strain (RaHx). A, The Weibull model, S(x) = exp[-(x/a)b]. B, The Gompertz model, S(x) = exp[-(a/b)(ebx - 1)]. C, The two-parameter logistic model, S(x) = 1/[1 + (x/c)b]. D, The three-parameter logistic model, S(x) = exp(-bx - (b/k) log{[a+(b - a)e-kx]/b}). It is seen that the Gompertz approximation poorly catches both early and late survival, whereas the two-parameter logistic model overestimates heavy-tail survival probability at advanced ages. The Weibull and three-parameter logistic approximations adequately describe the survivals both at early and at late ages.
|
|
|
Acknowledgments
|
|---|
R. Arking was the recipient of a Korean Science and Engineering Foundation "Expert Brain Pool" Visiting Research Professorship at Pusan National University during the writing of this manuscript and appreciates the congenial and profitable interactions with Professors M. A. Yoo and H. Y. Chung during that time.
We thank Professors M. Friedrich, A. Popadic, and M. VanBerkum for their constructive criticisms.
Received January 25, 2002
Accepted August 6, 2002
|
Appendix
|
|---|
To show that the Weibull model is a reasonable one for the survival data, we compared it with three other well-known statistical models. These models are the Gompertz, the two-parameter logistic, and the three-parameter logistic functions. We used a standard mean-square-error technique (11) to fit each one of our survival curves shown in Fig. 2. The typical fitting results are shown in Fig. 5, in which the four models are applied for survival data from a nonlethal heat shock treatment. The six other curves (not presented) yield similar results. The standard errors for all seven cases are given in Table 2 .
It is seen that, on average, the Gompertz model gives the worse results; the best results are from the three-parameter logistic approximation. The Weibull model is the second best. Thus we prefer the Weibull model as it yields a transparent parametric presentation of the survival data (as it is shown in Fig. 1), whereas the three parameters fitted in three-logistic presentation give no straightforward and understandable information on the survival pattern.
|
References
|
|---|
- Arking R, 1987. Successful selection for increased longevity in Drosophila: analysis of the survival data and presentation of a hypothesis on the genetic regulation of longevity. Exp Gerontol 22:199-220. [Medline]
- Arking R, Burde V, Graves K, et al. 2000. Selection for longevity specifically alters antioxidant gene expression and oxidative damage patterns in Drosophila. Exp Gerontol 35:167-185. [Medline]
- Vettraino J, Buck S, Arking R, 2001. Direct selection for paraquat resistance in Drosophila results in a different extended longevity phenotype. J Gerontol Biol Sci 56A:B415-B425. [Abstract/Free Full Text]
- Kuether K, Arking R, 1999. Drosophila selected for extended longevity are more sensitive to heat shock. Age 22:175-180.
- Johnson TE, Wu D, Tedesco P, Dames S, Vaupel JW, 2001. Age-specific demographic profiles of longevity mutants in Caenorhabditis elegans show segmental effects. J Gerontol Biol Sci 56A:B331-B339. [Abstract/Free Full Text]
- Pletcher SD, Khazaeli AA, Curtsinger JW, 2000. Why do life spans differ? Partitioning mean longevity differences in terms of age-specific mortality parameters. J Gerontol Biol Sci 55A:B381-B389. [Abstract/Free Full Text]
- Wilson DL, 1994. The analysis of survival (mortality) data: fitting Gompertz, Weibull, and logistic functions. Mech Ageing Dev 74:15-33. [Medline]
- Vanfleteren JR, De Vreese A, Braeckman BP, 1998. Two-parameter logistic and Weibull equations provide better fits to survival data from isogenic populations of Caenorhabditis elegans in axenic culture than does the Gompertz model. J Gerontol Biol Sci 53A:B393-B403. [Abstract]
- Wilmoth JR, 2000. Demography of longevity: past, present, and future trends. Exp Gerontol 35:1111-1129. [Medline]
- Vaupel JW, Carey JR, Christensen K, et al. 1998. Biodemographic trajectories of longevity. Science 280:855-860. [Abstract/Free Full Text]
- Matlab: The Language of Technical Computing. Natick, MA: MathWorks, Inc.; 1998.
- Luckinbill LS, Arking R, Clare MJ, Cirocco WC, Buck S, 1984. Selection for delayed senescence in Drosophila melanogaster. Evolution 38:996-1004.
- Arking R, 2001. Gene expression and regulation in the extended longevity phenotypes of Drosophila. Ann NY Acad Sci 928:157-167. [Medline]
- Arking R, Wells RA, 1990. Genetic alteration of normal aging processes is responsible for extended longevity in Drosophila. Dev Genet 11:141-148. [Medline]
- Dudas SP, Arking R, 1995. A coordinate upregulation of antioxidant gene activities is associated with the delayed onset of senescence in a long-lived strain of Drosophila. J Gerontol Biol Sci 50A:B117-B127.
- Arking R, Burde V, Graves K, et al. 2000. Identical longevity phenotypes are characterized by different patterns of gene expression and oxidative damage. Exp Gerontol 35:353-373. [Medline]
- Buck SA, Arking R, 2001. Metabolic alterations in genetically selected Drosophila strains with different longevities. J Am Aging Assoc 24:151-162.
- Harshman LG, Haberer BA, 2000. Oxidative stress resistance: a robust correlated response to selection in extended longevity lines of Drosophila melanogaster?. J Gerontol Biol Sci 55A:B415-B417.
- Orr WC, Sohal RS, 1984. Extension of life-span by overexpression of superoxide dismutase and catalase in Drosophila melanogaster. Science 263:1128-1130.
- Sohal RS, Mockett RJ, Orr WC, 2000. Current issues concerning the role of oxidative stress in aging: a perspective. Hekimi S, , ed.The Molecular Genetics of Aging: Vol. 29 Results and Problems in Cell Differentiation 45-66. Springer-Verlag, Berlin.
- Sun J, Tower J, 1999. FLP recombinase-mediated induction of Cu/Zn-superoxide dismutase transgene expression can extend the life span of adult Drosophila melanogaster flies. Mol Cell Biol 19: (1) 216-228. [Abstract/Free Full Text]
- Parkes TL, Elia AJ, Dickinson D, Hilliker AJ, Phillips JP, Boulianne GL, 1998. Extension of Drosophila lifespan by overexpression of human SOD1 in motorneurons. Nat Genet 19:171-174. [Medline]
- Phillips JP, Parkes TL, Hilliker AJ, 2000. Targeted neuronal gene expression and longevity in Drosophila. Exp Gerontol 35:1157-1164. [Medline]
- Lin Y-J, Seroude L, Benzer S, 1998. Extended life-span and stress resistance in the Drosophila mutant methuselah. Science 282:943-945. [Abstract/Free Full Text]
- Novoseltsev VN, Carey J, Liedo P, Novoseltseva J, Yashin AI, 2000. Anticipation of oxidative damage decelerates aging in virgin female medflies: hypothesis tested by statistical modeling. Exp Gerontol 35:971-987. [Medline]
- Clancy DJ, Gems D, Harshman LG, et al. 2001. Extension of life-span by loss of CHICO, a Drosophila insulin receptor substrate protein. Science 292:104-106. [Abstract/Free Full Text]
- Tatar M, Kopelman A, Epstein D, Tu M-P, Yin C-M, Garofalo RS, 2001. A mutant Drosophila insulin homolog that extends life-span and impairs neuroendocrine function. Science 292:107-110. [Abstract/Free Full Text]
- Hwangbo DS. Life-Span Extension in Drosophila Brought About by a Down-Regulation of the Insulin-like Signaling System [master's thesis]. Detroit, MI: Wayne State University; 2002.
- Kang H-L, Benzer S, Min K-T, 2002. Life extension in Drosophila by feeding a drug. Proc Natl Acad Sci USA 99:838-843. [Abstract/Free Full Text]
- Flurkey K, Papaconstantinou J, Miller RA, Harrison DE, 2001. Lifespan extension and delayed immune and collagen aging in mutant mice with defects in growth hormone production. Proc Natl Acad Sci USA 98:6736-6741. [Abstract/Free Full Text]
- Melov S, Ravenscroft J, Malik S, et al. 2000. Extension of life-span with superoxide dismutase/catalase mimetics. Science 289:1567-1569. [Abstract/Free Full Text]
- Lane M, 2000. Metabolic mechanisms of longevity: caloric restriction in mammals and longevity mutations in Caenorhabditis elegans; a common pathway. J Amer Aging Assoc 23:1-7.
- Weindruch R, Walford RL. The Retardation of Aging and Disease by Dietary Restriction. Springfield, IL: Charles C Thomas; 1988.
- Fishbein DH. Biological Effects of Dietary Restriction. Berlin: Springer-Verlag; 1991.
- Yu BP. Modulation of Aging Processes by Dietary Restriction. Boca Raton, FL: CRC Press; 1994.
- Arking R, Buck S, Berrios A, Dwyer S, Baker GT, 1991. Elevated antioxidant activity can be used as a bioassay for longevity in a genetically based long-lived strain of Drosophila. Dev Genet 12:362-370. [Medline]
- Myers GC, Manton KG, 1984. Recent changes in the U.S. age at death distribution: further observations. Gerontologist 24:572-575. [Medline]
- Holloszy JO, Kohrt WM, 1995. Exercise. Masoro EJ, , ed.Handbook of Physiology. Section 11: Aging 633-666. Oxford University Press, New York.
- Lithgow GJ, White TM, Melov S, Johnson TE, 1995. Thermotolerance and extended life span conferred by single-gene mutations and induced by thermal stress. Proc Natl Acad Sci. USA 92:7540-7544. [Abstract/Free Full Text]
- Khazaeli AA, Tater M, Pletcher SD, Curtsinger JW, 1997. Heat induced longevity extension in Drosophila I. Heat treatment, mortality, and thermotolerance. J Gerontol Biol Sci 52A:B48-B52.
- Yanase S, Cypser J, Johnson TE, Ishi N, 2000[abstract]. Adaptive response extends the life span through signaling to DAF-16 in Caenorhabditis elegans. J Radia Res 41: (4) 373
- Yanase S, Hartman PS, Ito A, Ishi N, 2000[abstract]. Oxidative stress pretreatment increases the X-irradiation resistance of the nematode Caenorhabditis elegans. J Radia Res 41: (4) 373
- Walker GA, White TM, McColl G, et al. 2001. Heat shock protein accumulation is upregulated in a long-lived mutant of Caenorhabditis elegans. J Gerontol Biol Sci 56A:B281-B287. [Abstract/Free Full Text]
- Lithgow GJ, 2000. Stress response and aging in Caenorhabditis elegans. Results Prob Cell Diff 29:131-148.
- Minois N, 2000. Longevity and aging: beneficial effects of exposure to mild stress. Biogerontology 1:15-29. [Medline]
- Arking R, Giroux CN, 2001. Antioxidant genes, hormesis, and demographic longevity. J Anti-Aging Med 4:125-136.
- Tatar M, Khazaeli AA, Curtsinger JW, 1997. Chaperoning extended life. Nature 390:30[Medline]
- Perls TT, 1997. Centenarians prove the compression of morbidity hypothesis but what about the rest of us who are genetically less fortunate?. Med Hypoth 49:405-407. [Medline]
- Perls TT, Bochen K, Freeman M, Alpert L, Silver MH, 1999. Validity of reported age and centenarian prevalence in New England. Age Ageing 28:193-197. [Abstract/Free Full Text]
- Arking DE, Krebsova A, Macek M, Sr et al. 2002. Association of human aging with a functional variant of klotho. Proc Natl Acad Sci USA 99:856-861. [Abstract/Free Full Text]
- Kirkwood TBL, 1987. Immortality of the germ-live versus disposability of the soma. Woodhead AD, Thompson KH, , ed.Evolution of Longevity in Animals: A Comparative Approach 209-218. Plenum, New York.
- Novoseltsev VN, Arking R, Novoseltseva JA, Yashin AI. Evolutionary optimality applied to Drosophila experiments: hypothesis of constrained reproductive efficiency. Evolution. In press.
- Patel MN, Knight CG, Karageorgi C, Leroi AM, 2002. Evolution of germ-line signals that regulate growth and aging in nematodes. Proc Natl Acad Sci USA 99:769-774. [Abstract/Free Full Text]
- Sgro CM, Partridge L, 1999. A delayed wave of death from reproduction in Drosophila. Science 286:2521-2524. [Abstract/Free Full Text]
- Hsin H, Kenyon C, 1999. Signals from the reproductive system regulate the lifespan of C. elegans. Nature 399:362-366. [Medline]
- Arking R, Buck S, Hwangbo D-S, Lane M, 2001. Genomic plasticity, energy allocations, and the extended longevity phenotypes of Drosophila. Aging Res Rev. 1:209-228.
- Service PM, Micheieli CA, McGill K. Experimental evolution of senescence: an analysis using a heterogeneity mortality model. Evolution. In press.
- Carnes BA, Olshansky SJ, 2001. Heterogeneity and its biodemographic implications for longevity and mortality. Exp Gerontol 36:419-430. [Medline]
- Finch CE, Kirkwood TBL. Chance, Development, and Aging. New York: Oxford University Press; 2001.
- Lints FA, Lints CV, Bullens P, Bourgois M, Delince J, 1989. Unexplained variations in the life span of the Oregon-R strain of Drosophila melanogaster over a four year period. Exp Gerontol 24:265-271. [Medline]
- Izmaylov DM, Obukhova LK, Okladnova OV, Akifyev AP, 1993. Phenomenon of life span instability in Drosophila melanogaster: I. Nonrandom origin of life span variations in successive generations. Exp Gerontol 28:169-180. [Medline]
- Finch CE, Tanzi RE, 1997. The genetics of aging. Science 278:407-411. [Abstract/Free Full Text]
- Klebanov S, Flurkey K, Roderick TH, et al. 2000. Heritability of life span in mice and its implication for direct and indirect selection for longevity. Genetica 110:209-218. [Medline]
- Parsons PA, 1995. Inherited stress resistance and longevity: a stress theory of ageing. Heredity 75:216-221.
- Johnson TE, Lithgow GJ, Murakimi S, 1996. Hypothesis: interventions that increase the response to stress affect the potential for effective life prolongation and increased health. J Gerontol Biol Sci 51A:B392-B395.
- Olshansky SJ, Carnes BA, Desesquelles A, 2001. Prospects for human longevity. Science 291:1491-1492. [Abstract/Free Full Text]
- Kirkwood TBL. Time of Our Lives: The Science of Human Aging. London: Phoenix Press; 2000.
- Rose MR, 1999. Can human aging be postponed?. Sci Am 281: (6) 106-111. [Medline]
- de Grey ADNJ, Ames BN, Andersen JK, et al. 2002. Time to talk SENS: critiquing the immutability of human aging. Ann NY Acad Sci 959:452-462. [Medline]
Top
Abstract
Demographic Aspects of the...
