| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
a Institute for Behavioral Genetics, University of Colorado, Boulder
b Max-Planck-Institut fur Demografische Forschung, Rostock, Germany
Thomas E. Johnson, Institute for Behavioral Genetics, University of Colorado, Box 447, Boulder, CO 80309 E-mail: johnsont{at}colorado.edu.
Decision Editor: John A. Faulkner, PhD
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
THE United States is undergoing a demographic revolution. Current population estimates show a larger percentage of elderly individuals (over the age of 65) now than has ever before existed in the history of the country—12.5% now versus 8.1% in 1950 and 4.1% in 1900. Many gerontologists and demographers believe that mortality at advanced ages cannot be substantially reduced (1)(2)(3)(4)(5). The canonical view has been that death rates at advanced ages in developed countries are close to limits that can only be relaxed by fundamental breakthroughs in slowing the process of aging. Recent studies in model systems such as Caenorhabditis elegans, a nematode worm, have shown that mortality at advanced ages can be dramatically reduced by simple genetic interventions. These findings offer significant challenges for the view that mortality at old age is intractable and suggest that the fraction of the elderly population could increase even more dramatically than projected.
The traditional view in biological demography is that a single, universal process of aging produces an exponential increase in mortality with age—the so-called Gompertz law (6)—and, to cite a 1991 report in Scientific American, that "like a clock" every individual "is constructed to run a certain time." The emerging evidence from our current projects and other research suggests a new paradigm of aging that recognizes a rich variety of diverse and often quite variable aging processes that can be influenced by both genetic and environmental changes. The research presented within focuses on the genetic changes in age-specific mortality of long-lived mutants in the nematode C. elegans.
C. elegans has been extensively used to identify single-gene "Age" mutants that have an increased life span (7)(8)(9)(10)(11). Mortality increases with chronological age in wild-type (N2) C. elegans, just as in most, if not all, species (12), and the rate at which mortality changes is under broad genetic control (13)(14). The mortality doubling time (MDT) of normal, wild-type worms is approximately 3 to 5 days at 20°C (14)(15)(16). The MDT of strains carrying mutations in the best studied gerontogene, age-1, is approximately twofold longer (15). The increased longevity of the Age mutants stems from any one of several types of mutational alterations. These longevity mutants fit into one of six categories (10)(17): polygenic (RI), dauer type (Daf), slow growth (Clk), fertility deficient (Spe), overexpression (Old), and slow eating (Eat). A total of 39 mutations (both published and unpublished) that extend longevity of one or both sexes in C. elegans has been reported (9), with more than a dozen still remaining to be published.
A major limitation on the trajectory of age-specific mortality is the lack of analysis of large populations of nematodes. With only two exceptions (16)(18), longevity and survival of long-lived variants have been studied only in small populations. The current study is designed to extend and deepen our understanding of the kinetic changes in mortality underlying the increased life expectancy of specific Age mutants. The focus of this paper is on detailed demographic profiles of a representative mutant of the Daf, Clk, and Spe gerontogene classes, using methodologies involving large mass cultures of 100,000 nematodes or more.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Strains were propagated at 20°C on nematode growth media (NGM) plates with Escherichia coli strain OP50 as a food source, as previously described (13). Liquid transfers of animals were performed in S basal media (13)(21). Strains used in this study were TJ1060 [spe-9(hc88) I; fer-15(b26) II], TJ1062 [spe-9(hc88) I; fer-15(b26) age-1(hx546) II], both constructed by Tom Fabian (22); TJ1067 [spe-9(hc88) I; fer-15(b26) II; clk-1(e2519) III]; and BA821 [spe-26(hc138) IV]. Most strains carried two fertility mutants [fer-15(b26) and spe-9(hc88)] that together completely eliminate the production of progeny (22); because of difficulty in strain construction, spe-26 mutants were studied without any additional sterile genes added to the background. Life spans of all of these strains have been validated by comparisons with the single gerontogene mutants alone (22); however, studies showing that different backgrounds of N2 can affect longevity (23) continue to plague the observations on BA821. Strains were constructed by using standard methods (21). Genotype was confirmed by back crossing to both parental single-mutant strains and by showing failure to complement.
Mass Cultures
Generation of large numbers of age-synchronous young adults..--
E. coli strain OP50 was used as a food source. Each subculture was initiated by transferring agar plugs containing dauer larvae from a small NGM plate stock to three to five large (10 cm) NGM plates spread with OP50. The worms were then washed off the plates with S basal and transferred to 2% peptone plates (same as NGM but with 2% peptone) spread with RW2 (a wild-type strain of E. coli.)
Eggs were isolated by treatment with alkaline hypochlorite (24). Eggs were inoculated into sterile 10-cm glass Petri dishes at 0.1-ml egg pellet per dish, and S basal was added to a total volume of 10 ml. Eggs were allowed to hatch at 20°C over the next 24 hours in the absence of E. coli and cholesterol. First stage larvae (L1's) were then inoculated onto 2% peptone plates and allowed to develop at 25.5°C for 48 hours. More details can be found in Fabian and Johnson (22).
Maintenance of Age-Synchronous Cultures of Adults
Young adults were harvested by washing them off the plates and were inoculated at a concentration of 100 worms/ml into 2-l Erlenmeyer flasks containing 1000 ml of survival medium, which was E. coli (1 x 109/ml) and cholesterol (10 µg/ml) in S basal. Liquid cultures were maintained at 25.5°C and 250 rpm in a rotary air shaker (Series 25D Incubator Shaker, New Brunswick Scientific, Edison, NJ). At 2-day intervals, approximately one half of the medium was removed and replaced with fresh medium by allowing the worms to settle out before the medium was removed.
Assessment of Age Trajectory of Survival
Every-other-day subsampling was used to assess mortality. A subsample of approximately 2000 nematodes (20 ml) was withdrawn from the mass culture every other day and all dead worms in this sample were discarded (Fig. 1). The mortality of this subsample was then assessed over the next 24-hour period, after which the subsample was discarded and a fresh subsample taken. Our assessment of death is based on the following: lack of spontaneous movement, loss of ability to respond to touch, visible tissue degeneration, and lack of osmotic turgor. The criteria used for distinguishing live and dead worms were developed 20 years ago (13) and are the same in standard assays and mass cultures.
|
Statistical Analysis
Data were collected by hand and transferred to Dr. Deqing Wu at the Max Planck Institute in Rostock, Germany, for analysis. Some analysis was also performed at our site.
The direct calculation of the conditional probability of dying (qx) was calculated as
 |
Smoothing with locally weighted least squares was carried out as follows (25)(26). Given scatter plot data, (Xj, Yj)j = 1,...,n and associated weights Wj, the basic assumptions are as follows: (i) E (Yj | Xj = x) = f(x) with a smooth function f, and (ii) var(Yj | Xj = x) is finite. Then a consistent curve estimate of f(x) using locally weighted squares is:
 |
Here j is day, b is the bandwidth of the smoother, and K is a nonnegative kernel function. This estimate fits local linear lines by weighted least squares to the data falling into a window (x - b, x + b) around the point x where the estimate is desired. We chose the kernel function
 |
For the two-stage Gompertz model, we used the equations
 |
 |
The form of the two-stage Gompertz model (27) is
 |
 |
The logistical regression used the formula
 |
was also fitted to the data and then a likelihood test was used.
To test whether the daily mortality between two groups is significantly different, we use the normally distributed test statistic Zq(x) (28):
 |
 |
Significance tests of mortality in total between two groups were performed by using the k-sample test (29), which can be described as follows: suppose that, in the pooled sample, failures occur at times t1 < t2 < ... < tN. Let nij be the number of individuals in sample j still at risk just prior to ti, and let dij be the number of failures in sample j at time ti. Let the corresponding numbers in the pooled sample be ni+ and di+, respectively (ni+ = 
jnij, and di+ = jdij). Under the hypothesis of no difference between samples, the conditional expected value of dij, given nij, can be estimated as Eij = nij(di+/ni+). Let E+j = iEij. Then the test statistic is
 |
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
300), the rapid loss of carcasses after death, presumably as a result of rapid digestion by the E. coli in which these animals are raised, and, most importantly, the effort involved in daily censuses of large populations. To solve the problem of large numbers of progeny, we incorporated strains that were made completely sterile by using two mutations ( fer-15 and spe-9) blocking sperm development (22). Such strains contain two fertility mutants that completely sterilize the worms but do not cause any alteration in life span (22). (Note that no single temperature-sensitive mutation that we tested was completely effective at eliminating progeny production; absolute sterility is essential to ensure that the longest-lived individuals in a mass culture are not offspring of the original cohort.) The problem of loss of dead worm carcasses was addressed in preliminary experiments in which worms were killed and then followed under conditions similar to those used for the mass culture where little loss of carcasses was observed after 2 days and probably up to 1 week. The problem of counting 100,000 or more worms per day was solved by using a sampling strategy such that we counted a smaller number of worms each day and ascertained mortality in this subpopulation, as in Brooks and colleagues (16). A subsample of approximately 2000 nematodes (20 ml) was withdrawn from the mass culture every other day. All dead worms in this sample were discarded (Fig. 1) and the mortality of this subsample assessed over the next 24-hour period, after which the subsample was discarded. When only approximately 2000 worms, more or less, were left alive, we harvested and counted the entire population.
We have collected at least partial data from seven distinct cohorts on 700,000 wild-type worms of the strain TJ1060 (22). Five such experiments have been successfully carried to completion out of a total of seven attempted (Fig. 2). In the wild-type TJ1060, the mortality profiles were remarkably homogenous between replicates (Fig. 2) and showed two phases: an exponential and a plateau. To further estimate replicability, we split a population and assessed mortality independently, but simultaneously, in both subpopulations (Fig. 2, 10-7-96A and 10-7-96B). There was complete homogeneity in the daily force of mortality, with these replicates showing no significant difference. Although some daily variation in force of mortality is observable, this was not statistically significant. Note especially that variability of small peaks is not reliable or reproducible. Two-stage Gompertz modeling (Table 1 ), which provides a much better fit than does a one-stage Gompertz model, is used throughout. In the five different wild-type populations, variation in estimates of the Gompertz parameters is small for b but much larger for ß (Table 1 ). The slopes of the first stage of the Gompertz plots ranged between 0.334 and 0.638 and varied only approximately 25%. Stage 2 slopes varied from 0 to 0.115. Thus, the genetic specification of longevity more rigidly determines early mortality than the later "plateau" stage. The age at which the inflection point occurred differed only slightly between experiments on TJ1060 (Table 1 ), varying from 13.5 to 19.5 days of age with only 2.43 days as a standard error. The mortality level at which the inflection occurred was more variable, but the mass summary statistic suggests that the inflection occurred at 19.5% mortality. Thus, the family of five best-fit curves for the two-stage Gompertz model for the wild-type strain TJ1060 (Fig. 2) are remarkably consistent among themselves. We saw considerable variation among replicates in the number of worms alive when the second stage of the Gompertz was reached (Table 1 ). This variation was not explained by any easily assessed parameter. Because of the environmental variation between genetically identical replicates, multiple replicates of each mutant strain provide a better estimate of the genetic effects by allowing multiple sampling of these inconsistent environmental variations across replicates.
|
|
|
In contrast, clk-1 started at a tenfold higher initial mortality but then had a twofold slower rate of increase in mortality in both stages of the two-stage Gompertz (Table 1 ). For clk-1, the age at which mortality levels off as well as the level of mortality at the inflection point are approximately the same as wild type. clk-1 shows decreased mortality only late in life (Fig. 3).
The spe-26 mutant strain seems to start at a sevenfold lower mortality than the wild type and to then show a wild-type rate of mortality increase until approximately 1 week later in life than wild type (Table 1 , Fig. 3). Then, at 3 weeks of age, mortality tends to level off but at a higher mortality than that of wild type. Thus spe-26 mutants show lower mortality only early in life.
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
We observed that mortality tends to level off (always increasing slightly with increasing age) at a high level in all strains and experiments. This is consistent with many experiments (see 18 for a review) in a variety of species, showing a similar tendency to level off or even decrease. Thus, mortality does not increase indefinitely but levels off at a rate that is under genetic control, as shown by the results of mutants described within. Mortality kinetics are well described by a two-stage Gompertz model.
There are six or more classes and a total of 39 reported genes in C. elegans in which mutations result in longer adult life expectancy and maximum life span (9). We have examined representative mutants from three of these classes: age-1, representing the dauer-constitutive class, clk-1, representing the clock class of mutants, and spe-26, representing the fertility-defective class of mutants. Each of these three classes (and the individual mutant representing each class) is thought to represent a quite distinct mechanism for life prolongation. Our demographical results bear out this prediction in that each mutant alters the demographic profile in a characteristic fashion. The first class consists of those loci where mutants have a dauer-constitutive (Daf) phenotype (10)(11). In addition to the age-1 mutant, described within, mutants in daf-2 result in a more than twofold extension of life expectancy in the adult phase and daf-2 interacts with daf-12 to cause an almost fourfold increase in life expectancy. Finally, pdk-1 mutants show a life-extension phenotype. All of the effects (dauer formation, life extension, and stress resistance) of age-1 and daf-2 are suppressed by mutant alleles of daf-16, a gene necessary for dauer formation. So far we have only assessed mortality in mass cultures of one representative of this class of genes, age-1, and that mutation lowers mortality at all ages.
A second class (Fertility deficient, or Fer) contains only two genes: spe-26 and spe-10. Two of the six mutant alleles of spe-26, a gene specifying proper segregation of cellular components affecting sperm activation, were reported (35) to result in life extensions of approximately 80% for the hermaphrodite and the mated male, although the details are contentious (36). Results presented here show that life extensions fall short of that previously reported for spe-26. The spe-10 mutant is also long lived, perhaps more long lived than spe-26 (37).
The third class studied is the "clock" class (Clk; 9–11), where two of four alleles of clk-1, both of which have altered cell cycle and developmental timing, also have increased life expectancy. Other genes in this class (clk-2, clk-3, and gro-1) all have modest (typically 20–30%) extensions of life span, but only for some alleles. Other classes are the Eat (eat-1, -2, -3, -6, -13, -18, and unc-26) that slow the consumption of food and also result in life extension (9)(10)(11), presumably through a self-imposed, dietary restriction; hypermorphic (Old for overexpression longevity determinants) (38), a new class of gerontogene that increases life span when overexpressed; and polygenic mutants, in which life prolongation results from the combined action of many genes (quantitative trait loci; 14,(39)(40)(41)).
We assessed detailed mortality kinetics in representative mutants from each of the first three classes of mutants listed above. As researchers found previously in insects (42)(43), we observed a deflection of the Gompertz curve at late ages. This is more pronounced than that previously reported in C. elegans (27). We confirmed that all three mutants (age-1, clk-1, and spe-26) are longer lived. Maximum life spans of all three were also greater than that of the wild-type TJ1060, where the last living worm typically died at approximately day 30. In contrast, the last living age-1 mutant worm died at 50 to 55 days. For clk-1 mutants the last worm died at approximately day 33 to 40 in most experiments, and for spe-26 only approximately a 3-day life extension was observed (last dead at day 28 or 33). This modest increase in life expectancy of spe-26 is consistent with that previously reported (35)(36).
Each mutant had a distinct, statistically different pattern of mortality that accounted for its extended longevity. By far, the most impressive were the age-1 mutants, which showed extremely significant, lower mortality throughout life (almost tenfold at some ages). In contrast, clk-1 and spe-26 mutants showed age-specific (but highly significant statistically) alterations in mortality that we refer to as "segmental." The spe-26 mortality was reduced only early in life, prior to day 18, whereas clk-1 had consistently lower mortality only after approximately 20 days. It could be that the life extension of spe-26 results from the elimination of some cause of death related to reproduction, which occurs only early in life and has been shown to cause increased mortality in Drosophila (44), or that uncontrolled background effects (see 23) played a role. However, these assessments were made in a sterile background, so it cannot be that reproduction per se underlies the difference in survival of spe-26. In contrast, clk-1, which may have reduced metabolic rates (45)(46), only lowers mortality later in life. For clk-1, early mortality is actually higher than in the wild type. Because large numbers of animals at risk are necessary to accurately estimate such low mortality rates early in life and because we have only assessed three representative mutant alleles in these large culture conditions, it is not clear whether the generalizations distinguishing each mutant can be extended to include other members of the same class of mutant.
In summary, different genetic variants, each identified initially as a prolongevity mutant, affect different parts of the nematode life span, segmentally. One possible implication is that differential gene action or the accumulation of toxic gene products or other toxic agents could differentially accumulate among the different types of mutant strains. Clearly clk-1, which may have a severe metabolic defect, perhaps involving coenzyme Q biosynthesis (47), is at higher risk early in life, consistent with its delayed development and altered cyclicity (48). However, those worms surviving early mortality are then destined to age slower and show dramatic decreases in mortality approaching that of age-1. Because spe-26 shows lower early mortality, but does not affect late-life mortality, it is tempting to suggest that this segmental pattern has something to do with the fertility defect presumably responsible for the increased longevity of spe-26 (and perhaps of spe-10 as well). However, the lack of consistent effects of all spe-26 alleles on longevity suggest that other factors are at work and the fact that the increased longevity and stress resistance of spe-26 mutants are suppressed by daf-16 (49) suggests a direct role in dauer formation. The greatest impact on longevity and mortality is seen in mutants in age-1. This is the first gerontogene mutant identified in any species (50)(51) and until last year was the only mutant ever identified solely by its longevity phenotype. In the only other published study on mortality in these mutants, Johnson (15) reported a good fit to a one-stage Gompertz model and found a twofold change in the slope but no change in the initial mortality rate. The studies shown here extend those observations and show little effect on initial mortality but reduced increases in mortality with age. Mortality in age-1 is drastically reduced at all ages after the initial time point, suggesting either continued need for the AGE-1 protein through all of adult life or long-lasting effects of an early-life alteration in AGE-1 activity. Mutants of age-1 are healthy and move vigorously for much longer than do wild type (52)(53), suggesting that it is not just longevity but also health span that is altered.
In considering several potential complications of these assessments, the most important is possible problems from contamination by progeny. Adults were completely sterilized by constructing strains carrying two temperature-sensitive, fertility-defective mutations that together completely block fertility at 25.5°C. In a previous study (22), we found that most single fertility mutants allowed some leakage of small numbers of offspring. For three of the assessments reported here (wild type, age-1, and clk-1), we were successful using this protocol. For the third mutant (spe-26), we relied on only a single mutation and found that spe-26 mutants produced few progeny under the conditions used in the assessments. Small animals (<3/4 adult size, which represented immature or developmentally abnormal worms) were few, usually 2% or less, and did not increase in number during the experiment, consistent with a complete block of reproduction.
We have run several controls to ask if this assessment protocol accurately assesses the underlying mortality of the mass culture. One way in which estimates may be biased is if dead worms are lost at higher rates than are the living. We performed defined experiments in which we either killed worms artificially and asked how long we could still recover their carcasses or in which we kept those dying a natural death for several more days, counting them daily. Both studies showed that carcasses survive for several days under these conditions. Alternately, we asked if living worms are more likely to be lost in the flask; again, we found that this was not likely to be true.
Another possible confound is the effect of the day at which assessment protocols were changed (Big Count Day, Table 1 ). By this day, most of the worms in the culture were dead or had been removed by sampling so that only a completely countable number of worms were still available (Table 1 ). In the wild-type cultures of TJ1060, this day varied between 16 and 22 days. All three gerontogenes mutants had later Big Count Days: spe-26 and clk-1, approximately 7–8 days later, whereas for age-1 there was a 16- to 24-day delay in the age at which this assessment occurred. It should also be noted that there is no association between Big Count Day and the inflection point in the two-stage Gompertz model.
It is our hypothesis that the leveling off of the mortality functions in later life observed in previous experiments (18) is driven by acquired heterogeneity, that is, differential mortality between individuals or sub cohorts that was not present in the initial populations. In genetically identical cohorts, we hypothesize that acquired heterogeneity is a major source of variation in subsequent survival (54)(55).
In comparing our studies with those of other studies on genetic effects on mortality kinetics, several things are noticeable. First is the major impact of single-gene mutants, especially age-1, as compared with relatively minor effects and small heritabilities seen in studies of polygenic variations (53). This also should be compared with fairly small narrow-sense heritabilities in Drosophila for longevity (6–9%) versus 19% (narrow sense) to 44% (broad sense) heritability estimates in nematodes (13). The range of differences for age-specific mortality seen in Drosophila is not that different from those observed in comparing age-1 mutants with wild type.
The identification and cloning of genes that play a role in increasing longevity in a variety of species, but especially in C. elegans, and the recent demonstration that human genes can play a similar role in model systems (S. Murakami and T.E. Johnson, unpublished data, 1999) clearly demonstrates that longevity is remarkably responsive to genetic interventions. The results presented within show that huge changes in age-specific mortality can be achieved by altering only one gene. In human populations, a drug treatment, instead of a genetic alteration, could be envisioned that would similarly inactivate one protein or gene function. Whether such drug interventions are achievable or desirable is debatable.
|
Acknowledgments |
|---|
We thank Kendra Robinson, Ibett Ramirez, and numerous other members of the Johnson lab over the years for their help in collecting this data. We wish to thank Chris Link and members of his group for input as well as advice from Jim Carey, Jim Curtsinger, and Anatoli Yashin in the interpretation of demographic data. We also thank the two anonymous reviewers for several suggestions that helped the final presentation.
Deqing Wu is currently at the Institute of Population Research, Chinese Academy of Social Science, Jian Guo Men Nei Dajie 5, 100732, Beijing, P. R. China.
Received September 21, 2000
Accepted November 28, 2000
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  D. Wu, J. R. Cypser, A. I. Yashin, and T. E. Johnson The U-Shaped Response of Initial Mortality in Caenorhabditis elegans to Mild Heat Shock: Does It Explain Recent Trends in Human Mortality? J. Gerontol. A Biol. Sci. Med. Sci., July 1, 2008; 63(7): 660 - 668. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Mulder, J. Hendriks, R. Baerselman, and L. Posthuma Age Structure and Senescence in Long-Term Cohorts of Eisenia andrei (Oligochaeta: Lumbricidae) J. Gerontol. A Biol. Sci. Med. Sci., December 1, 2007; 62(12): 1361 - 1363. [Full Text] [PDF]
|
|
|
|
|
|
J. Chen, D. Senturk, J.-L. Wang, H.-G. Muller, J. R. Carey, H. Caswell, and E. P. Caswell-Chen A Demographic Analysis of the Fitness Cost of Extended Longevity in Caenorhabditis elegans J. Gerontol. A Biol. Sci. Med. Sci., February 1, 2007; 62(2): 126 - 135. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. P. Caswell-Chen, J. Chen, E. E. Lewis, G. W. Douhan, S. A. Nadler, and J. R. Carey Revising the Standard Wisdom of C. elegans Natural History: Ecology of Longevity Sci. Aging Knowl. Environ., October 5, 2005; 2005(40): pe30 - pe30. [Abstract] [Full Text]
|
|
|
|
|
|
T. E. Johnson Genes, Phenes, and Dreams of Immortality: The 2003 Kleemeier Award Lecture J. Gerontol. A Biol. Sci. Med. Sci., June 1, 2005; 60(6): 680 - 687. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. E. Morley Editorial: Hot Topics in Geriatrics J. Gerontol. A Biol. Sci. Med. Sci., January 1, 2003; 58(1): M30 - 36. [Full Text] [PDF]
|
|
|
|
|
|
R. Arking, J. Novoseltseva, D.-S. Hwangbo, V. Novoseltsev, and M. Lane Different Age-Specific Demographic Profiles Are Generated in the Same Normal-Lived Drosophila Strain by Different Longevity Stimuli J. Gerontol. A Biol. Sci. Med. Sci., November 1, 2002; 57(11): B390 - 398. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. J. Olshansky, L. Hayflick, and B. A. Carnes Position Statement on Human Aging J. Gerontol. A Biol. Sci. Med. Sci., August 1, 2002; 57(8): B292 - 297. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. J. Olshansky, L. Hayflick, and B. A. Carnes Position Statement on Human Aging Sci. Aging Knowl. Environ., June 19, 2002; 2002(24): pe9 - 9. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
