Factors contributing to the plasticity of the extended longevity phenotypes of Drosophila

ExperimentalGerontology,Vol. 31, No. 6, 623~43, 1996 Copyright© 1996 ElsevierScienceInc. Printed in the USA.All rights reserved 0531-5565/96$15.00 + .00 ELSEVIER PII S0531-5565(96)00096-4 Mini-Review F A C T O R S CONTRIBUTING TO THE PLASTICITY OF THE E X T E N D E D LONGEVITY P H E N O T Y P E S OF DROSOPHILA ROBERT ARKING 1, ALLAN G. FORCE,1'3 STEVEN P. DUDAS I, STEVEN BUCK 1 and GEORGE T. BAKER, III 2,4 1Department of Biological Sciences, Wayne State University, Detroit, Michigan 48202 and 2Gerontology Research Center, National Institute of Aging, Baltimore, Maryland 21224 Abstract - - A number of laboratories have constructed independently derived long-lived strains of Drosophila, each of which have similar but not identical patterns of variability in their adult longevity. Given the observed plasticity of longevity within each of these strains, it would be useful to review the operational and environmental factors that give rise to this phenotypic plasticity and ascertain whether they are common or strain specific. Our review of the more extensively analyzed strains suggests that the allelic composition of the initial genomes and the selection/transgene strategy employed yield extended longevity strains with superficially similar phenotypes but which are probably each the result of different proximal genetic mechanisms. This then offers a plausible explanation for the differential effects of various environmental factors on each strain's particular pattern of phenotypic plasticity. It also illustrates that the species has the potential to employ any one of a number of different proximal mechanisms, each of which give rise to a similar longevity phenotype. Copyright © 1996 Elsevier Science Inc. Key Words: longevity, life span, Drosophila, genetic control of aging, genetic plasticity, phenotypic plasticity, environment effects INTRODUCTION IN 1891, A u g u s t W e i s m a n n asked what factors accounted for the significant differences in life span observed b e t w e e n different species and a m o n g different m e m b e r s of the same species ( W e i s m a n n , 1891). Although R a y m o n d Pearl (1928) and others began to develop answers to his question, the task was soon a b a n d o n e d in favor of more experimentally tractable problems in genetics, evolution, embryology, biochemistry, and the like. Indeed, it was not until some insight had been attained into molecular and developmental genetic m e c h a n i s m s that the prob3Current address: Department of Biology, University of Oregon, Eugene, OR 97403 aDeceased Correspondence to: Robert Arking (Received 6 November 1995; Accepted 15 July 1996) 623 624 R. ARKING et at. lem of aging and longevity could be put in a proper perspective. Even then, the choice of an experimentally appropriate model system raised some problems. A number of investigators chose to examine the reasons underlying the shortened longevity of various mutant strains (Baker et al., 1985), believing that their analysis would provide insights into the mechanisms regulating aging and senescence in normal lived organisms. This approach was not productive, because a shortened longevity may arise from any one of a number of different developmental pathologies, none of which had much to do with normal aging (Mayer and Baker, 1985). In the last decade or so, a number of different groups decided to use artificial selection to create extended longevity strains, and use these as a model system with which to examine the mechanisms underlying aging and senescence. This approach has been more successful. One implication of this approach was the transformation of longevity into just another phenotype. As such, it is as variable as any other quantitative trait. It is the purpose of this paper to (1), catalog the sources of this phenotypic plasticity as observed within our own selected long-lived L strain; and (2), compare them to the other selected Drosophila strains and determine whether these individual sources of plasticity are strain specific or are common to the several different long-lived strains. Genetic approaches are often used to uncover a causal pathway leading from the gene to the phenotype, and vice versa. Drosophila, as the premier genetic organism, has been used in aging research since 1913 (Lints and Soliman, 1988). An impressive amount of data have been collected that bear directly on the problems of aging and longevity in this one organism. The review by Baker et al. (1985) lists almost 400 literature references, and a current comprehensive review would probably need to list several hundred more. Yet despite this effort we still do not fully understand the mechanisms of aging in Drosophila. Part of the reason for this slow progress might have to do with the variability of past results reported in different strains (Baker et al., 1985). This variability might arise from the plasticity of the aging process itself (defined as the extent to which the expression of an individual's genotype can be modified by environmental factors (Scheiner, 1993), or from the possibility that the different strains studied by different investigators might actually age in different ways, or both. In any event, the variability implies that it might be misleading to extrapolate across strains without some assurance that the strains are, in fact, comparable; that they are not affected in different ways by the same environmental variables, and/or that they are not using different mechanisms to regulate their extended longevity phenotype, as was the case with the Oregon-R and Canton-S strains studied by Ganetzky and Flannagan (1977). Two possible strategies by which a comparative genetic approach can minimize the effect of such variability are to (1), investigate some presumptive factors affecting longevity across a number of replicate strains; and/or (2), investigate the factors affecting longevity on a defined set of related strains with significantly different longevities. We shall discuss examples of both strategies. Usefulness of Drosophila as a model system for the study of aging Johnson et al. (1993) have questioned the utility of Drosophila as a model system that can afford insights into the mechanisms of longevity extension operative in other organisms including humans. These objections were based on the alleged failure to identify candidate genes, the assumed involvement of large numbers of genes with heterotic effects, the large difference in life histories between mammals and insects, and the alleged inability to ascertain causality of extended longevity in Drosophila. The data presented in this article and others suggests that these objections are not well founded. The genes comprising proximal longevity mechanisms PLASTICITY OF LONGEVITY 625 such as the antioxidant defense system have been identified as candidate genes (Dudas and Arking, 1995). The minimum number of genes necessary to significantly increase longevity is two (Orr and Sohal, 1994). There are no heterotic effects noted in our strains, the genes involved being normal recessives (Buck et al., 1993a). Causality is being addressed with respect to the antioxidant defense system and shows every sign of being successful (Arking et al., unpublished data). There is undoubtedly a large difference in life history strategies between mammals and dipteran insects; yet the disposable soma theory of aging (Kirkwood, 1990), suggests that longevity is the outcome of a series of trade-offs between reproduction and repair/defense processes. Given the fundamental similarity of basic biological processes, then it is not surprising that enhanced antioxidant defenses have been implicated in the extended longevity mechanisms probably operative in fungi, nematodes, flies, and mammals. We suggest that appropriate investigations into the extended longevity of Drosophila may provide insights useful for analysis of similar mechanisms in mammals, including humans. Sources of variation f o r any phenotype There are three sources of variation for any phenotype. There are two genetic variables, the additive and the nonadditive factors, which interact with the environmental factors so as to yield the observed final phenotype (Falconer, 1981). Scheiner (1993) has reviewed the definitions, measurements and models of phenotypic plasticity in general; the reader is referred to that article for an overall assessment of the topic. Curtsinger (1990) has estimated, on the basis of population genetic studies, that about 20% of the phenotypic variance in Drosophila longevity is due to additive variance, about 20% to nonadditive variance, and about 60% to environmental factors. We can operationally identify environmental plasticity of a phenotype by examining the data for the existence of genotype by environment interactions that result in the significant modulation of the phenotype. A complete analysis of the mechanisms regulating aging and longevity in our strains requires that we identify and characterize all components of these three variables. Our experiments have been designed with this goal in mind and our results are generally congruent with those of Curtsinger (1990). Thus, our discussion of environmental plasticity leads us to also discuss the variability of the genetic system on which it acts. Analysis of the extended longevity phenotype (ELP) in our strains of Drosophila has led us to conclude that the ELP is a genetically determined, environmentally modulated, event dependent, developmental process. We have previously determined that the critical genetic factors are located on the third chromosome (c3), but their expression is positively and negatively regulated by nonadditive factors located on the first and second chromosomes (cl and c2) (Arking et al., 1993; Buck et al., 1993a). In addition, the expression of these strain specific genetic factors is modulated by the environmental components such as larval density (Buck et al., 1993b), ambient temperature (Arking et al., 1988), and nutrition (see below), among others . The data reviewed below will provide the factual support for this empirical definition of the aging process in these strains. Finally, the allelic composition of the initial or baseline strain must restrict the potential response of the population to a given selection pressure. Our own baseline R strain was constructed from recently caught wild flies in Michigan. Sampling this baseline strain nine times over a 10-year period with two different selection paradigms has yielded a number of long-lived strains, each of which appear to depend primarily on enhanced antioxidant resistance as their proximal mechanism. Other laboratories find a similar conformity of response among their sublines but involving some other proximal mechanism. The species response is broader than 626 R. ARKING e t al. that of any given population. Hence, the allelic composition of the initial population must predict and constrain their response to any particular selection or transgenic intervention. Different methods of constructing long-lived strains Artificial Selection The use of artificial selection as a means of constructing long-lived strains can be viewed as a proof of the correctness of the evolutionary theory of aging (Rose, 1991). Although some attempts have not been successful (i.e., Lints and Hoste, 1974; Baret and Lints, 1993), those failures cannot be logically used to disprove the general possibility of using such a procedure to generate extended longevity strains, especially in the face of the data (Arking and Buck, 1995). In most cases, different investigators have indirectly selected for extended longevity by directly selecting for delayed female fecundity (Luckinbill et al., 1984; Rose, 1984; Arking, 1987; Partridge and Fowler, 1992). However, it has also been shown possible to directly select for dessication resistance (Hoffmann and Parsons, 1989) or starvation resistance (Graves et al., 1992; Rose et al., 1992), and thus, indirectly select for extended longevity. It seems likely that these different selection scenarios would be likely to give rise to similar but not identical phenotypes, and this variability in selection paradigms might also account for some of the genetic plasticity of longevity in Drosophila. We will present data below to support this statement. Even when different investigators believe they are using the same selection procedure, they may be mistaken. For example, the different selection regimes may not differ simply in age at reproduction but may also manifest other phenotypic factors that were coselected as well. A clear example is provided in comparing the long-lived and short-lived strains from Roper et al. (1993), as well as those from our own laboratory (Buck and Arking, unpublished data). Both labs found that their strains were found to have been selected inadvertently for fast development time as well as for extended longevity. It is possible that other such traits unrelated to longevity may also have been inadvertently selected in these and other strains. This situation gives rise to two possibilities. First, if the changes in some inadvertently selected physiological parameter in both late- and early-reproduced populations occur and are not controlled, then the comparisons between such lines may give rise to spurious positive or negative relationships between any character and longevity. It seems likely that some characters may be associated with different aspects of the selection procedure but may not be associated with longevity per se. Second, the inadvertent inclusion of other phenotypic characters within the generic extended longevity phenotype may give rise to differing patterns of gene interactions between the several ostensibly identical strains. In the absence of explicit information regarding the existence of the inadvertent factors, the differing gene expression patterns may be wrongly interpreted as evidence of different genetic mechanisms. Very clearly, either of these situations may give rise to a spurious interpretation of genetic plasticity of longevity in Drosophila. On the other hand, Draye et al. (1994) have demonstrated that natural populations of Drosophila are genetically different for at least some life history traits when measured in the laboratory as soon as possible after capture. This finding implies that the species as a whole is polymorphic in these traits and, thus, that laboratory populations founded from different natural populations may respond differently to the same experimental situation because they are genetically different from one another. Such a finding gives rise to the expectation that there should be substantial genetic plasticity of longevity in laboratory strains derived from different founder populations. 627 PLASTICITY OF LONGEVITY The long-lived strains constructed by different laboratories appear to have much in common with each other. For example, the strains constructed via a direct selection for delayed female fecundity all appear to bring about the extension of longevity via a delayed onset of senescence (Luckinbill e t al., 1984; Arking and Wells, 1990; Rose, 1984; Partridge and Fowler, 1992). Such a process affects the age of onset of senescence but does not affect the rate of aging. The Gompertz curves shown in Fig. 1 illustrate this in the case of our long-lived L and normal-lived R strains, where the L strains have the same slope but a much lower intercept than do the R strains. From a demographic point of view, this means that the extended longevity arises from a lowering of the mortality rate at all ages, as was shown by Curtsinger (1995) for sister lines to our strains. Is it possible to use artificial selection to construct a strain that lives long, not because it delays the age of onset of senescence but because it slows down the rate of aging? It is clear that gene expression during the adult stage is tightly regulated and exhibits what appear to be both early life- and late life-specific gene activity patterns (Fleming e t al., 1993; Tower, 1993; Dudas and Arking, 1995; Helfand e t al., 1995). It should be possible in principle to use molecular genetic techniques to isolate genes (and mutants) with altered late life specific gene activity patterns such that they would live longer due to a slowing of the rate of senescence, in a manner similar to that used by Egilmez et al. (1989) and D ' M e l l o et al. (1994) in the isolation of the LAG-1 genes of yeast. Tower (1993) is using a similar approach in that he is using insertional mutagenesis to isolate mutants with late life gene expression patterns. A related approach was taken by Zwaan (1995), who directly selected for long-lived individuals. These strains and mutants have not yet been fully analyzed but when they are, it is likely that the mutants and/or 2, o f j -2, 0 20 40 60 80 OLa 100 AGE FIG. 1. The Gompertz plots of a normal-livedcontrol strain (RA) compared to that of a genetically selected long-lived strain (LA). The former is the normal-lived baseline strain from which the long-livedL strain was derived. Both strains were raised at high larval density. Note that the slopes of the two strains are almost parallelto one anotherbut that their interceptsare significantlydifferent. This observationsuggests that the significantfactors in the extended longevityof the L animals are a lower mortalityrate and a delayedonset of senescencebut not a decrease in their rate of aging. This conclusionis consistentwith those obtainedfrom an analysis of biomarkerdata (Arking and Wells, 1990) and from mortalitydata on sister lines (Curt.singeret al., 1995). Based on maximumlikelihood analysis of data presented in Buck et al. (1993a,b). 628 R. ARKING et al. strains isolated by these paradigms would show an extended longevity phenotype significantly different from that observed in the present long-lived L strain. Conversely, proving that two or more fundamentally different mechanisms, each derived by similar selection strategies, but which each give rise to a similar generic phenotype would also point out another source of genetic plasticity of longevity in Drosophila. Transgenic Experiments In Drosophila, transgenic experiments have been done with the specific aim of testing the antioxidant theory of aging (Harman, 1956). Reveillaud et al. (1991) made transgenic animals carrying an extra copy of the bovine Cu-Zn superoxide dismutase (SOD) gene under the control of an actin 5c promoter. The resulting animals expressed both mammalian and Drosophila SOD. Some, but not all, of the transgenic strains exhibited an increased resistance to exogenous paraquat along with a modest increase in the mean lifespan. There was no effect in the maximum lifespan observed in the experimental and control lines, suggesting that the effect of the treatment was to decrease premature mortality. This was confirmed by a later experiment showing that the bovine SOD gene had the ability to rescue a SOD null mutant from early lethality (Reveillaud et al., 1994). An essentially similar result was reported by Orr and Sohal (1992, 1993) as a result of their construction of animals transgenic for either a Drosophila SOD gene or a Drosophila catalase (CAT) gene. However, constructing an animal transgenic for both the Drosophila SOD and CAT genes, each under the control of an actin promoter but also subject to the effects of position variegation, gave rise to a strain that was not only significantly resistant to exogenous paraquat but also exhibited a significant increase in both mean and maximum longevity (Orr and Sohal, 1994). In addition, the Gompertz curve for these doubly transgenic strains had the same intercept but a lower slope compared to the controls. This indicates that the extended longevity of the transgenic animals might come about via a slower rate of aging and not via a delay in the age of onset of senescence as is the case with our selected strains (Fig. 1). This suggests that even though these two strains both appear to rely on the antioxidant defense system for their extended longevity, the actual mechanisms by which this enhanced antioxidant gene activity exerts its physiological effects may well be quite different in these two strains. This difference in the phenotype is likely to manifest itself as a genetic plasticity of longevity in Drosophila. Webster (1986) first proposed that elongation factor 1~ (EF-I) was causally involved in bringing about the onset of senescence in Drosophila. Their hypothesis was initially confirmed by a transgenic experiment (Shepherd et al., 1989) in which an extra copy of the EF-I gene inserted in a wild type fly yielded a strain with a significant increase in mean and maximum lifespan relative to their controls. Subsequently, however, other more detailed analyses revealed that the effects were not as straight forward as was initially thought. Shikama et al. (1994) reported that their transgenic EF-1 flies do not synthesize more EF-1 mRNA or protein even though there still exists a reproducible difference in lifespan between the experimental and control animals. This finding suggests that the EF-I gene products themselves may not play a causal role in the longevity extension of the genetically manipulated strain. A series of EF-I transgene experiments was done by Stearns and Kaiser (1993), during which they observed negligible and/or variable effects on life span as a function of strain, sex, position, etc. It is now known that the EF-1 transgene was not expressed in these experiments and, thus, the observed effects must represent nonspecific insertional effects (Curtsinger et al., 1995). Finally, an analysis of the expression of the normal EF- 1 gene products in a genetically selected long-lived strain led Dudas and Arking (1994) to conclude that the expression of the EF-1 genes is not associated with the expression of the extended longevity phenotype in that strain. One important PLASTICITY OF LONGEVITY 629 point of these several analyses is that the interpretation of a transgene experiment may not be as simple as our preconception of the experimental situation might initially suggest. The organism represents a buffered physiological and metabolic system (Clark and Keith, 1988). Furthermore, many genetic systems are under the control of cis and/or trans regulatory loci, such as has been demonstrated for antioxidant defense systems in lower organisms (Munkres, 1990, 1992; Ames et al., 1993). Therefore, the random insertion of structural genes may disrupt regulatory homeostatic mechanisms and lead to compensatory responses that confound interpretation. For these reasons, transgenes may well have unexpected effects in different strains. Gene regulatory circuits are interconnected and can respond in unexpected ways to a physical or regulatory perturbation. The diversity of proximal mechanistic responses to such perturbations is a source of genetic plasticity for longevity in Drosophila. Genetic analysis o f different long-lived strains Transgenic Strains The transgenic experiments discussed above lead to the conclusion that altering the expression of as few as two particular genes, SOD and CAT, can lead to the expression of an extended longevity phenotype superficially similar to that produced by artificial selection. The transgenic experiment of Orr and Sohal (1994) also provides strong support for the involvement of free radicals in governing the rate of aging. Not only is their data consistent with our molecular data, but it is also consistent with our genetic estimates suggesting that, while extended longevity is a polygenic trait, it probably does not involve a large number of genes (Buck et al., 1993a). On the face of it, this experiment with its minimalist number of genes stands opposed to the concept that aging is a highly polygenic phenotype dependent on the integrated functioning of a large number of different genes, such as has been proposed by Rose and his colleagues (see below). Our own selected strains also seem to depend on an enhanced antioxidant defense system response, but as discussed below, they do this in a somewhat different mode. Some of the observed plasticity in the longevity of Drosophila may stem from the fact that there appears to be more than one way in which an organism can marshall its antioxidant defense mechanisms to yield an extended longevity phenotype. Even within a common proximal mechanism, there appears to be sufficient genetic variability that also serves as a source of additional environmental plasticity for longevity in Drosophila. Selected Strains Luckinbill et al. (1987) initially reported that at least one gene was involved in the expression of extended longevity in sister stocks to our strains. This minimum estimate was later discarded in favor of a polygenic mechanism primarily involving the third chromosome (Luckinbill et al., 1988). Although chromosome localizations were not performed, Rose and his colleagues (Hutchinson and Rose, 1991; Hutchinson et al., 1991) used two different types of quantitative genetic analyses to show that the transmission and expression of the extended longevity phenotype in their selected strains could be adequately explained as being due to the effects of additive genes averaged over some unspecified number of loci. Based on their analysis of the mobility and expression patterns of 321 proteins of the long-lived O strains of Rose (1984), Fleming et al. (1993) concluded that about six proteins had statistically different expression patterns in the long-lived and control strains, or about 2% of the total. Based on the assumption that this is a representative number that one may extrapolate to the whole genome, they then estimated that perhaps 200 to 400 loci that can postpone aging might exist in their strains. If the Orr and Sohal (1994) experiment sets a lower limit to the number of genes involved in extended longevity, then the experiment of Fleming et al. (1993) can be viewed as 630 R. ARKING et al. setting an upper limit. The resolution of this debate is important, driving as it does both our concepts and our experimental strategies of identifying the genes involved. Data obtained from the analysis of our selected strains supports the concept of a hierarchy of regulatory genes involved in the expression of a long-lived phenotype but suggests that (a), the number of structural genes involved may be substantially smaller than the estimate of Fleming et al. (1993) and (b), that selection acted so as to alter the nature of certain key regulatory genes in the experimental strain relative to its baseline control. Table 1 summarizes the results of some of the 27 isochromosomal lines constructed from the 1st, 2nd, and 3rd chromosomes (c 1, c2, c3) of a normal-lived R control strain and from a long-lived L strain selected for late-life fecundity and under high-density developmental conditions (see Wells et al., 1987; Buck et al., 1993a, 1993b; Arking et al., 1993, for details). The analysis of these and other data led to the following conclusions. First, the genes essential to the expression of the extended longevity phenotype (ELP) are located only on the L strain c3 and map somewhere to the left of ebony. Second, these genes are recessive. Third, there exists a complex pattern of epistasis such the longevity enhancing recessive genes on c3 are repressed by (unknown) genes on c2, which can, themselves, be repressed by (unknown) genes on cl. Thus, the c3 genes can be expressed either in the mutual presence of cl and c2, or in the absence of c2 (Arking et al., 1993; Buck et al., 1993). Fourth, larval density can also modulate the c3 gene expression, the critical period for its appliction being from 60 to 120 h after egg laying (Buck et al., 1993b), and is believed to act directly on the c3 (Arking et al., 1993). The regulatory circuits derived from these studies are shown in Fig. 2. The negative effect of c2 on c3 is supported by the report of Graf and Ayala (1986), in which they showed that the levels of SOD on c3 are under the control of (unknown) loci on c2. In a similar manner, Bewley and Laurie-Ahlberg (1984) have shown that the expression of CAT on c3 is also regulated by (unknown) loci on c2. The report of Graf and Ayala (1986) further suggests that one such difference between the R and L strain c3 might possibly involve a cis-acting element, such as their S O D Cat mutation, which was shown to significantly reduce TABLE I. LONGEVITY AND CHROMOSOMECOMPOSITION Mean LongeviD" ( +-SEM) Chromosome Composition Strain 000 222 001 002 220 1st 2nd 3 rd R/R L/L R/R R/R L/L R/R L/L R/R R/R L/L R/R L/L R/L LFL R/R Male 52.4 66.9 44.7 67.7 60.3 _+ 0.7 -+ 1.3 + 2.8 -+ 1.3 _+ 1.7 Female 49.2 64.3 47.8 64.3 54.6 _+0.8 _ 1. + 2.7 _+ 1.5 _+ 1.6 A n illustration of the connection between c h r o m o s o m e c o m p o s i t i o n a n d longevity. A total o f 27 i s o c h r o m o s o m a l lines, involving all possible c o m b i n a t i o n s o f the I st, 2nd, and 3rd c h r o m o s o m e s o f a short-lived R control strain and a long-lived L strain, were constructed a n d their longevity measured. The table s h o w s data o f four representative strains. The strain designation is b a s e d on a three digit n u m b e r where each place represents c l , c2, or c3 respectively; and the value at e a c h place (0, 1, or 2) represents the n u m b e r o f L type c h r o m o s o m e s present. Note that the 222 and 002 strains have identical L type life spans, a n d that the latter is significantly different f r o m the 000, 001, and 220 values. B a s e d on data presented in B u c k et al. (1993a). PLASTICITY OF LONGEVITY 631 U) ! C1 L I e2 L % FIG. 2. A summary of the chromosomaland environmental interactions involved in the expression of the extended longevity phenotype. Loci on cl L and c2e interact with c3L, both positively (cl) and negatively (c2), respectively, such that cl represses c2, which in turn, represses c3. The recessive genes on c3L are absolutely required for the delayed onset of senescence and the expression of the ELP, which may be fully expressed in the mutual presence and the mutual absence of cl L and c2L. The positive and negative roles of larval density are indicated by HD and LD. Based on data presented in Buck et al., (1993a,b) and in Arking et al. (1993). See text for discussion. SOD protein levels. Finally, while the molecular genetic regulation of anioxidant systems is one important and confirmed approach to extended longevity, other hierarchies of regulatory genes controlling systems that may act in conjunction with antioxidant systems are likely to exist. Indeed, caloric restriction is a good example of an experimental paradigm known to enhance longevity and overall physiological vigor, and which undoubtedly has its own regulatory hierarchy. The effectiveness of different proximal mechanisms suggests additional sources of genetic and environmental plasticity affecting the expression of the extended longevity phenotype. Molecular analysis o f different long-lived strains In nonselected standard laboratory strains, transgenic techniques have been used to increase selectively the dosage of certain antioxidant defense system (ADS) genes and thereby discern its effects on longevity. As discussed above, increasing the gene dosage of the SOD gene alone has only minimal effects on longevity (Seto et al, 1990; Reveillaud et al., 1991; Orr and Sohal, 1992). However, Orr and Sohal (1994) found that simultaneously increasing the dosage of both the SOD and CAT ADS genes brings about an increase in longevity comparable to that observed in our genetically selected strains. Their data certainly make the free radical theory a probable proximal theory of aging in Drosophila. That proximal explanation of extended longevity is applicable to some but not all selected long-lived strains. We have assayed the changes in the m R N A and/or enzyme activity levels of a number of loci during the development and early adult life of our normal-lived R and long-lived L strains (Dudas and Arking, 1995). The m R N A data, shown in Table 2, demonstrates that, at day 5 in the L strain, there appears to be a coordinately regulated significant increase in the m R N A levels of CuZnSOD, MnSOD, CAT, and xanthine dehydrogenase (XDH). These increases in m R N A levels are accompanied by significant increases in the enzyme activity of CuZnSOD, CAT, and glutathione-S-transferase (GST) (Table 2). In addition, our ongoing work shows that the SOD-specific protein increases proportionately as well (Burde, Haft, and Arking, unpublished data). Thus, it seems reasonable to conclude that these alterations in gene expression are probably the result of a transcriptional level change. It is well known that all four of these gene products are involved in antioxidant defense, and that null mutants at each locus 632 R. ARKING et ~tl. TABLE 2. RELATIVE LEVELS OF THE EXPRESSION OF ANTIOXIDANT GENES 1N NORMAL-LIVED AND LONG-LIVED STRAINS OF Drosophila AT DAY 5 OF ADULT LIFE* Relative Level ~[' lz)-pression in: Gene Product ADS mRNAs: CuZnSOD MnSOD CAT GST XDH A D S enzymes: SOD CAT GST Non-ADS mRNAs ADH haywire R strain L strain Significant (p < 0.05) 1.00 1.00 1.00 1.00 1.00 1.88 1.52 1.27 1.17 2.45 yes yes yes no yes 1.00 1.00 1.00 1.95 1.13 2.49 yes yes yes 1.00 1.00 0.90 0.98 no no *Data taken f r o m D u d a s and A r k i n g (1995), D u d a s (1993), and J u n g and A r k i n g (unpublished). render the organism very sensitive to paraquat dependent oxygen stress (see Phillips and Hilliker, 1990; Weinhold et al., 1990, for review and references). The mRNA's levels for other gene loci assayed in these same strains, such as the EF-1 (Dudas and Arking, 1994), the alcohol dehydrogenase (ADH) or haywire loci (Table 2), show different patterns of expression through the same developmental stages, thus indicating that the phenomenon observed is specific for the antioxidant defense genes. That these changes in ADS gene expression are important to the expression of the ELP is suggested by the following four facts. First, a twofold higher resistance to exogenous paraquat, a known free-radical generator, is observed in the young adults of our long-lived L strains relative to their normal lived controls (Arking et al., 1991). Second, the long-lived L strains have a enhanced temporal expression of paraquat resistance relative to the controls, maintaining significant levels of resistance for the first four to five weeks of adult life (Soliman and Arking, 1996). Third, long-lived L strains have a delayed onset and a lower level of oxidative protein damage relative to their normal lived controls (Soliman and Arking, unpublished). In addition, the long-lived L strain has a significantly lower rate of lipid peroxidation than does the R strain, and the final level of lipid peroxidation attained is also significantly lower in the L strain relative to the R strain (Soliman and Arking, unpublished data). Finally, it has been found that feeding the long-lived L strain animals with aminotriazole, a known specific inhibitor of CAT, makes the treated L animals respond to paraquat as if they were the low resistant R types (Dudas and Arking, 1995). Electrophoretic studies have shown that there is no such allozyme difference for SOD or CAT in our selected long-lived strains relative to the normal controls (Arking et al., 1993; Dudas, 1993). Such evidence suggests that the ADS structural genes in the L A and Ra strains are probably identical to one another (Dudas and Arking, 1995). It then follows that the selection process may have involved the fixation in the L A strain of regulatory gene sequences different PLASTICITY OF L O N G E V I T Y 633 from those commonly found in the R A strain. Operationally, we then would expect to find a number of polymorphic regulatory sites that differ between the two strains and which, when isolated, would be candidate longevity assurance regulatory genes, Such projects are now underway. In addition, the fact that the XDH gene is coordinately activated along with the other ADS genes suggests that the regulatory genes known to modulate XDH activity in terms of its effects on eye color and purine metabolism may also be operative in regulating its activity in terms of the ELP. If confirmed, then this suggestion would significantly extend and complicate the antioxidant gene regulatory system and expand the sources of additional genetic plasticity. On the other hand, Rose and his colleagues (Tyler et al., 1993) found that their selected long-lived strains contained a known high-activity allele of the SOD gene. However, although enzyme activity measurements have been reported for this allele in other genetic backgrounds, there has been no report of their activity in the O strains background. As described elsewhere in this paper and in their writings, their analyses have led them to consider other metabolic and biochemical factors as constituting proximal explanations for the expression of extended longevity in Drosophila. It is entirely possible that artificial selection operating on two different progenitor stocks may have operated so as to bring about a generically similar extended longevity phenotypes each dependent on different proximal mechanisms. Some of the observed plasticity of longevity may have its origins in this situation. Environmental factors affecting the expression of the extended longevity phenotype Some Environmental Factors Have A Minor Effect on Expression of the ELP It has long been known that the life span of poikiothermic animals can be significantly affected by changes in the adults' ambient temperature (see David, 1988, for review and references). Such environmental manipulations were once extensively used as a means for the experimental manipulation of longevity. We examined the effects of ambient temperature on adult life span to determine if manipulation of this factor would make the R strain control animals more similar to the long-lived L strain experimental animals (Arking et al., 1988). The results were instructive. Animals raised at 18°C have a mean and maximum life span that is about twice as great as it is for genetically identical animals raised at 28°C. Thus, the adult life span for both strains is changed by ambient temperature in the qualitative manner that one would expect on the basis of past work, both strains yielding an inverse relationship between the two variables with roughly equivalent slopes (see Fig. 1 of Arking et al., 1988). However, following a recent detailed reanalysis of these data, it is now quite apparent that the two strains also differ quantitatively in their response to ambient temperature. Computing the change in life span per degree C between different measured temperature regimes, as listed in Table 3, shows that these two genotypes differ in their response to ambient temperature in several important ways, as follows: First, the L strain animals show a larger response to temperature over the entire 18-28°C tested range than do the R strain controls, the actual value depending on other genetic (i.e., sex) and environmental (i.e., developmental temperature) factors, but ranging from a 13 to a 39% increase. Second, within the 18-28°C range, the L strain animals consistetly show their largest response in the 22-25 ° segment, while the R strain controls show their largest response in the 18-22 ° segment. 634 R. ARKING et al. TABLE3. GENOTYPE BY ENVIRONMENT EFFECTS OF AMBIENT TEMPERATURE ON ADULT LONGEVITY* Change in the M e a s u r e d Lifespan/1 °C Over the M e a s u r e d Temperature Differential Female Strain LE RE LI Rl LTso LT9o LTso LT9o LTso LT9o LTso LT9o Male 18-28 ° 1~22 ° 22 25 ° 22-28 ° 18-28 ° 18-22 ° 22-25 ° 5.7 8.7 4.6 6.7 7.0 7.8 4.3 5.6 4.2 9.5 10.0 2.7 3.0 5.5 9.2 6.4 8.1 7.0 7.8 4.3 5.8 3.2 8.5 7.8 13.7 6.5 6.7 8.2 3.3 4.2 6.2 8.7 3.0 4.0 10.5 8.3 8.5 6.3 8.0 16.3 8.3 10. 7 8.3 3.7 11.7 11.5 8.0 7.0 8.2 8.5 10. 7 6.5 6.8 8.3 3.0 3.3 22-28 ° 7.0 9.7 3.7 5.8 6.2 7.3 3.3 3.7 *Based on a reanalysis of data originally presented in Fig. 1 of Arking et al. (1988). E strains were kept at the indicated temperature from oogenesis to death; I strains were kept at 25°C until eclosion when they were transfered to the indicated temperature. The italicized numbers identify the temperature range in which that genotype exhibited its m a x i m u m response. See text for discussion. Third, both R and L strain animals raised at the same temperature throughout their entire life (i.e., the E series of data in Arking et al., 1988) usually showed a larger response to ambient temperature than did their sibs raised at 25°C during the developmental periods and then switched to the appropriate ambient temperature for the rest of their adult life span (i.e., the I series of data in Arking et al., 1988). Thus, stage specific temperature treatments may have delayed and cumulative effects on the adult. The two strains obviously react differently to environmental temperatures. This different genotype by environment interaction leads to a significant increase in the L strain lifespan at lower temperatures relative to the R strain. Although both the L and the R strain display an inverse relationship between their life spans and the ambient temperature, the actual response of the two genotypes is quite different in detail. Temperature is thought to exert its effects through a generalized alteration of the metabolic rate. There is an obvious inverse relationship between ambient temperature and mean daily metabolic rate (MDMR) such that animals raised at 18°C have a significantly lower M D M R than do animals of the same strain raised at 28°C (Arking et al., 1988). However, there is no statistically significant difference in the M D M R between the R and L strains raised at the same temperature even though there are such differences in longevity. The enhanced antioxidant defense of the L strain may explain part of this apparent paradox. The differential temperature dependency of longevity may come about, not because the temperature dependent rate of free radical production is different, but because the intrinsic activity of the ADS is different in these two strains. If temperature modulates the life span of Drosophila by affecting the metabolic rate, then the presumed increased production of free radicals will have a more deleterious effect in the R strain relative to the L strain because it is known that the latter strain has a more active ADS (Table 2; Dudas and Arking, 1995). This is consistent with the report of Helfand et al. (1995) suggesting that temperature can affect the timing but not the pattern of differential gene expression in the adult. PLASTICITY OF LONGEVITY 635 Some Environmental Factors Have Very Strong Effects on Expression of the ELP Larval density is known to affect life span in wild strains of D. melanogaster (Miller and Thomas, 1958; Lints and Lints, 1971; Economos and Lints, 1984; Zwaan et al., 1991), as well as in selected long-lived strains (Clare and Luckinbill, 1985; Buck et al., 1993b; Graves and Mueller, 1993). The data suggest that there is a generalized response of Drosophila longevity to larval density that alters adult longevity. Our selected strains also exhibited an obvious density dependent longevity (Clare and Luckinbill, 1985). The tested isogenic lines showed a mean increase of 16.9 days in their life span (Buck et al., 1993b), a response that was much greater than the five to seven day increase observed in wild strains. This suggests that selection may have acted so as to increase the magnitude of the density response in these animals. The genetic variability uncovered by larval crowding must be related to the genetic variability for longevity, presumably via some common genetically controlled factor(s) such as oxidative stress or developmental time. The L strain animals do develop faster that the R strain animals under both HD and LD conditions (Buck and Arking, in preparation), although we must point out that this may have been an unintended consequence of the design of our original selection experiment (see Arking, 1987, for discussion). Thus, the two strains react differently to this environmental variable as well. This constitutes another source of plasticity, at least to the extent that developmental time has an effect on longevity. Tigerstedt (1969) has shown that selection for fast development increases population fitness and liberates additive genetic variance, a finding not fully supported by Mueller et al. (1991). Dominguez and Albernoz (1987) were able to draw no general conclusions about the effect of density on the five strains they tested, suggesting the possible existence of extensive species plasticity for this environmental variable. Our use of density dependent selection to create the L strain also leads to the prediction that the density dependent aspect of the phenotype has the potential to become genetically assimilated (Waddington, 1940), thereby decreasing the environmental plasticity of the phenotype by incorporating the environmental effect into the genetic component. We have used density shift experiments (Buck et al., 1993b) to define the existence of a critical period in larval life that begins no later than 60 h after egg laying and that ends no later than 120 h after egg laying, and during which the developing larvae must be exposed to high density conditions if the extended longevity phenotype is to be expressed. The nature of the inductive event implied by the existence of the critical period is still unknown but it might well involve the antioxidant defense system. During this same time period, we have found that the HD food has an increased redox potential, relative to the LD food, as indicated by nitroblue tetrazolium measurements on food samples (Force and Arking, unpublished data). We have also observed that caloric restriction in the larvae is responsible for the induction of a significantly higher level of adult paraquat resistance (see Table 4). These observations suggest a role for oxidative stress during the larval stages as well as in the adult stages as discussed above and in Table 2. This presents a very interesting developmental problem regarding environmental induction of specific gene activities--an obvious and potentially useful source of plasticity for the extended longevity phenotype. Although there exists a strong density dependent genotype by environment interaction for lifespan in the B and O stocks of Rose (1984), this interaction has not been shown to involve as obvious a density threshold as was shown by Clare and Luckinbill (1985) for the L strains. The differences between these and other long-lived strains suggest that the presence of this density effect is not essential to the expression of the extended longevity phenotype. We have confirmed this supposition by manipulating one of our density dependent long-lived strains so as to produce a density independent (i.e., constitutive) long-lived strain that has an identical extended longevity phenotype (Arking, unpublished data). Larval density may be viewed as a condition that is 636 R. ARKING et al. TABLE 4. Group No. 1 2 3 4 5 6 7 8 9 EFFECT OF LARVAL YEAST RESTRICTION ON BODY WEIGHT AND ON PARAQUAT RESISTANCE n mg Yeast per 10 Eggs 203 208 208 208 196 184 206 209 207 0.01 0.025 0.05 0.10 0.25 0.50 0.75 1.0 2.0 mg Body Weight of 5-Day Adult on 10 mM 1.40 1.41 1.42 1.41 1.44 1.44 1.39 1.42 1.44 1.41 +_0.01 1.43 +_0.02 Median Survival Time Paraquat 43.0 42.8 42.7 42.4 37.4 38.6 40.6 34.8 40.9 42.7 +_0.03 38.5 +_2.5 Longevity 48.3 51.0 47.4 50.6 42.0 46.5 44.5 43.0 44.7 Median 49.2 +_1.6 44.1 +_1.7 All animals wee raised under LD conditions (10 eggs/vial) on standard sucrose agar media supplemented with top yeast as indicated. Our standard nutrition conditions are approximately equivalent to group 7 or 8. There is a discontinuous effect of diet such that feeding less than 0.1 mg/yeasWl0eggs significantly extends the paraquat response and the median longevity. This threshold is suggested by the dotted line. The responses of groups 14 (indicated by the italicized means and standard deviations) are significantly different from those of groups 5-9 (also indicated by italicized numbers) by the Kolmorov-Smirnovtest, except for the effect of diet restriction on adult body weight, which is not significant. associated with, but not always essential for, the expression of the extended longevity phenotype. Certainly, some o f the plasticity associated with longevity in different strains of D r o s o p h i l a may be attributed to the alterations in gene expression patterns associated with this environmental factor, some of which may have effects on longevity as well as on other density dependent traits. Finally, Bakker (1961) has shown that larval density had no effect on adult body weight in his stocks when compensation was made for the effects of competition for food. This implies that larval density must be considered in the context of dietary restriction effects. Dietary restriction is known to be a highly effective method of bringing about extended longevity in mammals and other forms (Austad, 1989; Finch, 1991). Economos and Lints (1984) have investigated the effect of larval dietary restriction on the adult longevity of the wild type (i.e., normal lived) S N O W strain. They found that the strain's maximum longevity of ca. 63 days was obtained at a submaximal level of ca. 40 mg yeast per 120 eggs. We have observed a similar effect in our long-lived strains. We have found (unpublished data) that varying the amount of yeast available to the larvae affects their performance on our standard bioassay of longevity. L strain animals were raised under LD conditions and allowed to feed on standard sucrose/yeast/ agar media containing concentrations of " t o p y e a s t " ranging from 2.0 mg to 0.01 mg per vial. The resulting adults were then divided into two subsets for each group. One subset was tested at five days of age for their paraquat resistance, a known bioassay of longevity in our strains (Arking et at., 1991); and the other subset was assayed for their adult longevity. We observed a consistent and statistically significant trend such that larvae fed on yeast concentration of 0.10 mg/vial or less resulted in a higher paraquat resistance and an increased adult life span (Table 4). The data suggest the existence of a dietary threshold below which the antioxidant defense system genes may possibly be activated. It would be instructive to determine whether the environmental triggers of larval dietary restriction and larval density exerted their effects PLASTICITYOFLONGEVITY 637 trhough similar or different mechanisms. There was no observable effect of the yeast concentration on body weight, suggesting that this dietary effect may be rather specific in its effects. One complication of doing feeding experiments on larvae is that of separating out the direct effects of nutrition on growth of the adult imaginal discs and internal organs from those affecting adult longevity via some other means. For example, the metabolic fate of ethanol derived carbons has been determined in D. melanogaster (Frerikesen et al., 1991, 1994; Heinstra and Geer, 1991; Geer et al., 1993). These data suggest that the larval ADH activity exerts an almost complete control over the flux of carbons from ethanol into lipid, its flux control coefficient being approximately 1.0 (Freriksen et al., 1991; but see Freriksen et al., 1994). This finding implies that there should be a direct correlation between ADH activity and lipid content. It also implies that larval dietary restriction and/or ADH gene activity changes may inadvertently affect the adult lipid content and body weight. Because these latter factors have been hypothesized to consititute part of the causal nexus underlying extended longevity (Service et al., 1985), then it follows that larval dietary restriction might affect the longevity of the adult by altering the animal's biochemistry and, thus, its potential ability to resist stresses such as starvation (Graves et al., 1992). On the other hand, however, our L strain animals are associated with a significantly reduced larval ADH mRNA and enzyme activity relative to the R strain (Dudas, 1993; Dudas and Arking, 1995). Accordingly, one might predict that the resulting adults should have a relative decrease in their lipid content which, as summarized below, is exactly what we have observed. Thus, comparatively small difference in the activity of stage- and locus-specific gene products can bring about substantial changes in the resulting adult; changes that are thought by some to be causally involved in the determination of adult longevity. It is reasonable to believe that these different genotypes will respond quite differently to identical environmental factors such as larval dietary restriction, yeast, and alcohol content of the food, and so forth. Thus, an understanding of the processes involved in larval stage metabolism should enhance our understanding of the mechanisms modulating the plasticity of the adult longevity phenotype. Chippindale et al. (1993) have investigated the role of dietary restriction during the adult stages, and report that dietary restriction enhanced the mean and maximum life spans of both control and long-lived strains by between 7.1 to 13.3 days. This increase is similar to that seen in our L and R strains as a result of larval density, although the two environmental factors presumably impinge on different regulatory gene hierarchies. The effect of dietary restriction in the adult is a very interesting result in view of the fact that there is very little mitotic activity in the adult; thus, one should be able to use this tool to investigate metabolic effects on longevity without the complications of cell replacement. If dietary restriction in the fly works in a manner similar to that observed in vertebrates, then one can predict that adult dietary restriction should alter the patterns of adult gene expression such as has been reported by Dudas and Arking (1995) and by Helfand et al. (1995). However, it should be noted that Le Bourg and Medioni (1991) have shown that, in their hands, adult dietary restriction had no effect on the longevity of their wild-type laboratory strain. This result suggests that different genotypes may respond differently to this environmental variable as well. Finally, it must be noted that Leroi et al. (1994) have reported that strong genotype by environment interaction can obscure what was thought to be a fundamental trade off of certain life-history characters (early fecundity and extended longevity), a finding that suggests that the continued presence of negative genetic correlations in a selected population may depend on the particularities of the genotype by environment interactions. If this turns out to be generally true, then whether a particular environmental factor is judged to have a major or minor effect upon the expression of the ELP may depend on when the question is asked. 638 R. ARKING et al. Comparison o f extended longevity mechanisms in different long-lived strains Several studies by different laboratories on different strains have implicated various biochemical and stress resistant features as being positively associated with increased longevity (Service et al., 1985, 1987; Graves et al., 1988, 1992; Luckinbill et al., 1988, 1989; Service et al., 1988; Hoffmann and Parsons, 1989). Service et al. (1985) and Service (1987) found that increased lipid content as well as increased starvation and dissiccation resistance were associated with increased longevity in their strains. It has been shown that sister lines to one of our long-lived L strains exhibited increased flight duration relative to a related but short-lived (E) strain (Graves et al., 1988). In a separate study using the independently derived long-lived O strain of Rose (1984), Graves et al. (1992) noted that increased flight duration and higher glycogen content were positively correlated with increased longevity. The authors concluded that there were at least two distinct physiological mechanisms that extend longevity in Drosophila: one associated with starvation resistance and lipid content, and one associated with dessication resistance, increased flight endurance, and higher glycogen content. This conclusion suggests that the extended longevity phenotype can arise from any one of several mechanisms, each of which act so as to reduce mortality associated with specific environmental stresses. We have examined various biochemical (protein, lipid, and glycogen content) and stress resistance (ability to survive starvation, dessication, and exogenous paraquat) parameters of 10 sister lines of five of our different Drosophila strains, four pairs of which were deliberately selected so as to express either a short-lived (M, 2E) or a long-lived (L, 2L) phenotype, while the fifth pair (R) was deliberately maintained in a nonselected state and served as the baseline strain to which all others were compared (Force et al., 1995). Our analysis of the data obtained from this comparative survey led us to the following conclusions regarding the biological processes underlying the expression of this differential longevity in our strains, as follows: First, the protein content is not significantly correlated with the strains selected for either early or late age of reproduction in our strains. Second, a reduced lipid content and a lower body weight are statistically associated with the expression of extended longevity in our strains, in contrast to the reports of other laboratories using other strains (Service, 1987). Third, starvation resistance is not correlated with the strains selected for either early or late age of reproduction in our strains, in contrast to the reports of other laboratories using other strains (Graves et al., 1992). Fourth, glycogen content and dessication resistance are positively correlated with longevity in our strains. However, this association does not appear to be diagnostic for either trait, because identical values for either trait may be found in strains that have been subjected to diametrically opposed selection regimes. Fifth, an enhanced resistance to exogenous paraquat is clearly and significantly associated with extended longevity in our strains, a finding that confirms and extends our previous findings regarding the coordinate upregulation of antioxidant gene activities in our long-lived strains as discussed above (Dudas and Arking, 1995). Paraquat resistance is a reliable diagnostic character, clearly separating extended longevity strains from all others. All of the available data strongly suggests that our L strains delay the onset of senescence due, at least in part, to their having a more effective antioxidant defense system. It is entirely possible that artificial selection operating on these two different progenitor stocks may have operated so as to bring about two superficially similar extended longevity phenotypes each dependent on different proximal mechanisms. Some of the observed plasticity of longevity PLASTICITY OF LONGEVITY 639 may have its origins in this situation, whereby the variety of operative genetic mechanisms, each with its own set of genotype-environment interactions, greatly increases the longevity in Drosophila and allows the population as a whole to respond appropriately to multiple environmental signals. However, it is not yet possible to determine whether the presumably independent proximal mechanisms involved in the expression of extended longevity in these different strains may actually have some deep genetic identity. Such a possibility is not totally implausible, given the data of Clark and Keith (1988) showing the interrelated expression of genes coding for various metabolically important enzymes. CONCLUSIONS We have demonstrated that different long-lived strains of Drosophila show a complex but understandable pattern of variability in their adult longevity as a result of manipulation of particular genetic and/or environmental factors. Within each strain, each of these factors must be properly defined, and their interactions understood, if we are to attain a proper understanding of the mechanisms that regulate the expression of the extended longevity phenotype. The genetic factors responsible for the extended longevity phenotype(s) are under investigation in several laboratories and presumably will be soon understood. The environmentally based phenotypic plasticity may itself reflect the different genotypes involved (Thompson, 1991). Scheiner and Lyman (1991) have shown that under these conditions, the genetic basis of plasticity appears to arise from epistatic gene interactions. The question arises as to whether the phenotype of extended longevity observed in different long-lived strains within the species may actually arise via different mechanisms. Do similar phenotypes arise via similiar or dissimiliar mechanisms? A comparative analysis of existing data suggests that the latter is the case. However, there seems to be an assumption in the literature (e.g., Parsons, 1995) that generalized stress resistance may be the main mechanism by which adult longevity is extended throughout the entire species. There is a problem here in the induction of a general conclusion from fairly specific research. This proposal forces one to lump together mechanisms as disparate from one another as resistance to oxidative stress and resistance to starvation. There is no known generalized stress resistance mechanism in Drosophila. Expression of the extended longevity phenotype by unrelated strains developed in different laboratories appear to involve different genetic, molecular, and physiological mechanisms. Thus, it seems prudent not to extrapolate data between unrelated strains without the support of empirical data. The existence of different mechanisms in independent strains derived from independent progenitor stocks would do much to explain the obvious disagreements between the data obtained from several laboratories each using strains that appear to be expressing the same generic extended longevity phenotype. The assumption that all extended longevity strains of Drosophila are dependent upon the same proximal mechanisms is not supported by the available data, but it appears to have been uncritically accepted (e.g., Dixon, 1993). This is unfortunate to the extent that it inhibits future experimental strategies. Rose (1991, and elsewhere) has emphasized the importance of replicated selection lines in aging research. We wish to emphasize the importance of doing comparisons between selection systems. All long-lived stocks of Drosophila are not the same. Our future work will involve the experimental verification of the genetic-environmental circuitry discussed here, using molecular and mutational techniques to define, characterize, and isolate the genes involved in the expression of the extended longevity phenotype in this strain. It will be of great interest to determine if the regulatory and structural genes operative in our strain are also operative in other long-lived strains, and, thus, provide a deep genetic identity. 640 R. ARKINGet al. It is p o s s i b l e t h a t e n e r g y a l l o c a t i o n m e c h a n i s m s m a y p l a y s u c h a role. S u c h a p h e n o m e n o n w o u l d f o r c e u s to r e c o n s i d e r i f w h a t w e n o w v i e w as d i f f e r e n t p r o x i m a l m e c h a n i s m s variations on a common are but t h e m e . I f s u c h a d e e p g e n e t i c i d e n t i t y d o e s not, in f a c t , e x i s t , t h e n w e m a y c o n s i d e r t h e s e g e n e s to b e u n i q u e s t r a i n - s p e c i f i c i d e n t i f i e r s o f q u a l i t a t i v e l y d i f f e r e n t e x tended longevity phenotypes. Acknowledgments--This paper was initially presented in a Gerontological Society of America sponsored Symposium on Genetic Plasticity of Longevity in 1991. It has been substantially revised and updated to December 31, 1995. We acknowledge the helpful comments of Michael R. Rose and an anonymous reviewer on an earlier version of the manuscript. We also acknowledge the assistance of Dr. Elaine Hochman (WSU Research Support Lab) with the statistical analysis. Portions of the work described were supported by a Shock Foundation Fellowship to S.P.D., by a Howard Hughes Underagraduate Research Fellowship to A.G.F., by a WSU President's Excellence Award to R. A., and by NIH grant AG 08834 to R.A. REFERENCES AMES, B.N., SHIGENAGA, M.K., and HAGEN, T.M. Oxidants, antioxidants, and the degenerative diseases of aging. Proc. Natl. Acad. Sci. USA 90, 7915-7922, 1993. ARKING, R., BUCK, S.A., WELLS, R.A., and PRETZLAFF, R. Metabolic rates in genetically based long lived strains of Drosophila. Exp. Gerontol. 23, 59-76, 1988. ARKING, R. and BUCK, S.A. 1995. Selection for increased longevity in Drosophila: A reply to Lints. Gerontology 41, 69-76, 1994. ARKING, R. and WELLS, R.A. Genetic alteration of normal aging processes is responsible for extended longevity in Drosophila. Dev. Genet. 11, 141-148, 1990. ARKING, R., DUDAS, S.P., and BAKER, G.T., III. Genetic and environmental factors regulating the expression of an extended longevity phenotype in a long lived strain of Drosophila. Genetica 91, 127-142, 1993. ARKING, R., BUCK, S.A., BERRIOS, A., DWYER, S., and BAKER, G.T., lIl. Elevated paraquat resistance can be used as a bioassay for longevity in a genetically based long-lived strain of Drosophila. Dev. Genet. 12, 362 370, 1991. ARKING, R. 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, 1987. AUSTAD, S.N. Life extension by dietary restriction in the bowl and doily spider, Frontilnela pyramitela. Exp. Gerontol. 24, 83-92, 1989. BAKER, G.T., III, Jacobson, M., and Mokrynski, G. Aging in Drosophila. In: Handbook of Cell Biology' of Aging, V. Cristofalo (Editor), CRC Press, Boca Raton, FL, 1985. BAKKER, K. An analysis of factors which determine success in competition for food among larvae of Drosophila melanogaster. Arch. Neerland. Zool. 14, 200-281, 1961. BARET, P. and LINTS, F.A. Selection for increased longevity in Drosophila melanogaster: A new interpretation. Gerontology 39, 252-259, 1993. BEWLEY, G. and LAURIE-AHLBERG, C.C. Genetic variation affecting the expression of catalase in Drosophila melanogaster: Correlations with rates of enzyme synthesis and degradation. Genetics 106, 4 3 3 5 4 4 8 , 1984. BUCK, S., WELLS, R.A.. DUDAS, S.P., BAKER, G.T., III, and ARKING, R. Chromosomal localization and regulation of the longevity determinant genes in a selected strain of Drosophila melanogaster. Heredity 71, 11-22, 1993a. BUCK, S., NICHOLSON, M.; DUDAS, S.P., BAKER, G.T., Ill, and ARKING, R. Larval regulation of adult longevity in a genetically selected long lived strain of Drosophila melanogaster. Heredity 71, 23-32, 1993b. CHIPPINDALE, A.K., LEROI, A.M., KIM, S.B., and ROSE, M.R. Phenotypic plasticity and selection in Drosophila life-history evolution. I. Nutrition and the cost of reproduction. J. Evol. Biol. 6, 171-193, 1993. CLARE, M. and LUCKINBILL, L. The effects of gene-environment interaction on the expression of longevity. Heredity' 55, 19-29, 1985.CLARK, A.G. and KEITH, L.E. Variation among extracted lines of Drosophila melanogaster in triacylglycerol and carbohydrate storage. Genetics 119, 909-923, 1988. CURTSINGER, J. Genetic and enviromnental components in variance of longevity in Drosophila males. Talk presented at 43rd annual meeting of Gerontology Society of America, 18 November, 1990. CURTSINGER, J., FUKUI, H.H., KHAZAELI, A.A., KIRSCHER, A., PLECHER, S.D., PROMISLOW, D.E.L., and TATAR, M. Genetic variation and aging. Annu. Rev. Genet. 29, 553- 575, 1995. DAVID, J.R. Temperature. In: Drosophila as a Model Organism.[br Ageing Studies, Blackie, Glasgow, 1988. DIXON, L.K. Use of recombinant inbred strains to map genes of aging. Genetica 91, 151-165, 1993. PLASTICITYOF LONGEVITY 641 D'MELLO, N.P., CHILDRESS, A.M., FRANKLIN, D.S., KALE, S.P., PINSWASDI, C., and JAZWINSKI, S.M. Cloning and characterization of LAG 1, a longevity assurance gene in yeast. J. Biol. Chem. 269, 15451-15459, 1994. DOMINGUEZ, A. and ALBORNOZ, 1. Environment-dependent heterosis in Drosophila melanogaster. Genet. Sel. EvoL 19, 37-48, 1987. DRAYE, X., BULLENS, P., and LINTS, F.A. Geographic variations of life history strategies in Drosophila melanogaster. I. Analysis of wild caught populations, Exp. Gerontol. 29, 205-222, 1994. DUDAS, S.P. Molecular genetic investigation of the extended longeveity phenotype of a long lived strain of Drosophila melanogaster. A dissertation submitted to the Graduate School of Wayne State University, May 1993. DUDAS, S.P. and ARKING, R.A coordinate upregulation of the 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, 1995. DUDAS, S.P. and ARK1NG, R. The expression of the EF1 genes of Drosophila is not associated with the extended longevity phenotype in a selected long lived strain. Exp. Gerontol. 29, 645-657, 1994. ECONOMOS, A.C. and LINTS, F.A. Growth rate and life span in Drosophila. III. Effect of body size and developmental temperature on the biphasic relationship between growth rate and life span. Mech. Ageing Dev. 27, 153-160, 1984. EGILMEZ, N.K. and JAZWINSKI, S.M. Evidence for the involvement of a cytoplasmic factor in the aging of the yeast Saccharomyces cerevisiae. J. Bacteriol. 171, 37-42, 1989. FALCONER, D.S. Introduction to Quantitative Genetics, 2nd ed., Longman, London, 1981. FINCH, C.E. Longevity, Senescence, and the Genome. Univ. of Chicago Press, Chicago, 1991. FLEMING, J.E., SPICER, G., GARRISON, R.C,, and ROSE, M.R. Two-dimensional protein electrophoretic analysis of postponed aging in Drosophila. Genetica 91, 183-193, 1993. FORCE, A.G., STAPLES, T., SOLIMAN, S., and ARKING, R. A comparative biochemical and stress analysis of genetically selected Drosophila strains with different longevities. Dev. Genet. 17, 340-351, 1995. FRERIKSEN, A., SEYKENS, D., SCHARLOO, W., and HEINSTRA, P.W.H. Alcohol dehydrogenase controls the flux from ethanol into lipids in Drosophila larvae: A 13C NMR study. J. Biol. Chem. 266, 21399-21403, 1991. FRERIKSEN, A., DE RUITER, B.L.A., SCHARLOO, W., and HIENSTRA, P.W.H. Drosophila alcohol dehydrogenase polymorphism and carbon- 13 fluxes: Opportunites for epistasis and natural selection. Genetics 137, 1071-1078, 1994. GANETZKY, B. and FLANAGAN, J.R. On the relationship between senescence and age-related changes in two wild-type strains of Drosophila melanogaster. Exp. Gerontol. 13, 189-196, 1977. GEER, B.W., HIENSTRA, P.W.H., and MCKECHNIE, S.W. The biological basis of ethanol tolerance in Drosophila. Comp. Biochem. Physiol. 105B, 203-229, 1993. GRAF, J.-D. and AYALA, F.J. Genetic variation for superoxide dismutase level in Drosophila melanogaster. Biochem. Genet. 24, 153-168, 1986. GRAVES, J.L. and MUELLER, L.D. Population density effects on longevity. Genetica 91, 99-110, 1993. GRAVES, J.L., LUCKINBILL, L.S., and NICHOLS, A. Flight duration and wing beat frequency in long- and shortlived Drosophila melanogaster. J. Insect Physiol. 34, 1021-1026, 1988. GRAVES, J.L., TOOLSON, E.C., JEONG, C., VU, L.N., and ROSE, M.R. Dessication, flight, glycogen, and postponed • senescence in Drosophila melanogaster. Physiol. Zool. 65, 268-286, 1992. HARMAN, D. Aging: A theory based on free radical and radiation chemistry. J. Gerontol. 11, 298-300, 1956. HEINSTRA, P.W.H. and GEER, B.W. Metabolic control analysis and enzyme variation: Nutritional manipulation of the flux from ethanol to lipids in Drosophila. Mol. Biol. Evol. 8, 703-708, 1991. HELFAND, S.L., BLAKE, K. J., ROGINA, B., STRACKS, M. D., CENTURION, A., and NAPTRA, B. Temporal patterns of gene expression in the antenna of the adult Drosophila melanogaster. Genetics 140, 549-555, 1995. HOFFMANN, A. and PARSONS, P. Selection for increased desiccation resistance in Drosophila melanogaster: Additive genetic control and correlated responses for other stresses. Genetics 122, 837-845, 1989. HUTCHINSON, E.W. and ROSE, M.R. Quantitative genetics of postponed aging in Drosophila melanogaster. I. Analysis of outbred populations. Genetics 127, 719-727, 1991. HUTCHINSON, E.W., SHAW, A.J., and ROSE, M.R. Quantitative genetics of postponed aging in Drosophila melanogaster. II. Analysis of selected lines. Genetics 127, 729-737, 1991. JOHNSON, T.E., TEDESCO, P.M., and LITHGOW, G.J. Comparing mutants, selective breeding, and transgenics in the dissection of aging processes of Caenorhabditis elegans. Genetica 91, 65-78, 1993. KIRKWOOD, T.B.L. The disposable soma theory of aging. In: Genetic Effects on Aging. H Harrison, D.E. (Editor), pp. 9-19, Telford Press, Caldwell, NJ, 1990. LE BOURG, E. and MEDIONI, J. Food restriction and longevity in Drosophila melanogaster. Age Nutr. 2, 90-93, 1991. LEROI, A.M., CHIPPINDATE, A.K., and ROSE, M.R. Long-term laboratory evolution of a genetic life-history trade-of in Drosophila melanogaster. I. The role of genotype-by-environment interaction. Evolution 48, 1244-1257, 1994. 642 R. ARKINGet al. LINTS, F.A. and HOSTE, C. The Lansing effect revisited: I. Life span. Exp. Gerontol. 9, 51- 69, 1974. LINTS, F.A. and LINTS, C.V. Influence of preimaginal environment on fecundity and ageing in Drosophila melanogaster hybrids. II. Developmental speed and life span. Exp. Gerontol. 6, 4 2 7 4 4 5 , 1971. LINTS, F.A. and SOLIMAN, M. H. Drosophila as a Model Organism for Ageing Studies, Blackie, Glasgow, 1988. LUCKINBILL, L.S., ARKING, R., CLARE, M.J., CIROCCO, W.C., and BUCK, S.A. Selection for delayed senescence in Drosophila melanogaster. Evolution 38, 996-1004, 1984. LUCKINBILL, L.S., CLARE, M.J., KRELL, W.L. CIROCCO, W.C., and RICHARDS, P. Estimating the number of geneic elements that defer senescence in Drosophila. Evolut. Ecol. 1, 37-46, 1987. LUCK1NBILL, LS., GRAVES, J.L., REED, A.H., and KOETSAWANG, S. Localizing genes that defer senescence in Drosophila melanogaster. Heredity 60, 367-374, 1988. LUCKINBILL, L.S., GRUDZIEN, T.A., RHINE, S., and WEISMAN, G. The genetic basis of adaptation to selection for longevity in Drosophila melanogaster. Evolut. Ecol. 3, 31-39, 1989. MAYER, P.J. and BAKER, G.T., IlI. Genetic aspects of Drosophila as a model system of eucaryotic aging. Int. Rev. Cytol. 95, 61-102, 1985. MILLER, R.S. and THOMAS, J.L. The effects of larval crowding and body size on the longevity of adult Drosophila melanogaster. Ecology 39, 118-125, 1958. MUELLER, L.D., GUO, P.Z., and AYALA, F.J. Density-dependent natural selection produces trade-offs in life history traits. Science 253, 433-435, 1991. MUNKRES, K.D. Genetic coregulation of longevity and anitoxienzymes in Neurospora crassa. Free Radic. Biol. Med. 8, 355-361, 1990. MUNKRES, K. Selection and analysis of superoxide dismutase mutants of Neuro,spora. Free Radic. Biol. Med. 13, 305-318, 1992. NUSBAUM, T.J., MUELLER, L.D., and ROSE, M.R. Evolutionary patterns among measures of aging. Exp. Gerontol. 31, 507-516, 1996. ORR, W.C. and SOHAL, R.C. The effects of catalase gene overexpression on life span and resistance to oxidative stress in transgenic Drosophila melanogaster. Arch. Biochem. Biophys. 297, 35-41, 1992. ORR, W.C. and SOHAL, R.C. Effects of Cu/Zn superoxide dismutase overexpression on life span and resistance to oxidative stress in transgenic Drosophila melanogaster. Arch. Biochem. Biophys. 301, 34-40, 1993. ORR, W.C. and SOHAL, R.J. Extension of life-span by overexpression of superoxide dismutase and catalase in Drosophila melanogaster. Science 263, 1128-1130, 1994. PARSON, P.A. Inherited stress resistance and longevity: A stress theory of ageing. Heredity 75, 216-221, 1995. PARTRIDGE, L. and FOWLER, K. Direct and correlated responses to selection on age at reproduction in Drosophila melanogaster. Evolution 46, 76-91, 1992. PEARL, R. The Rate of Living, University of London Press, London, 1928. PHILLIPS, J.P. and HILLIKER, A.J. Genetic analysis of oxygen defense mechanisms in Drosophila melanogaster. Adv. Genet. 28, 43-71, 1990. REVEILLAUD, I., NIEDZWIECKI, A., BENSCH, K.G., and FLEMING, J.E. Expression of bovine superoxide dismutase in Drosophila melanogaster augments resistance to oxidative stress. Mol. Cell, Biol. 11, 632-640, 1991. REVEILLAUD, I., PHILLIPS, J., DUYF, B., HILLIKER, A., KONGPACHITH, A., and FLEMING, J.E. Phenotypic rescue by a bovine transgene in a Cu/Zn superoxide dismutase-null mutant of Drosophila melanogaster. Mol. Cell. Biol. 14, 1302-1307, 1994. ROPER, C, P., PIGNATELLI, X. and PARTRIDGE, L. Evolutionary effects of selection on age at reproduction in larval and adult Drosophila melanogaster. Evolution 47, 445-455, 1993. ROSE, M.R. Laboratory evolution of postponed senescence in Drosophila melanogaster. Evolution 38, 1004-1010, 1984. ROSE, M.R. Evolutionary Biology of Aging, Oxford University Press, New York, 1991. ROSE, M.R., VU, L.N., PARK, S.U., and GRAVES, J.L., JR. Selection on stress resistance increases longevity in Drosophila melanogaster. Exp. Gerontol. 27, 241-250, 1992. SCHE1NER, S.M. Genetics and evolution of pheotypic plasticity. Annu. Rev. Ecol. Syst. 24, 35-68, 1993. SCHEINER, S.M. and LYMAN, R.F. The genetics of phenotypic plasticity. II. Response to selection. J. Evol. Biol. 4, 23-50, 1991. SERVICE, P.M., HUTCHINSON, H.W., MACK1NLEY, M.D., and ROSE, M.R. Resistance to environmental stress in Drosophila melanogaster selected for postponed senescence. Physiol. Zool. 58, 380-389. 1985. SERVICE, P.M. Physiological mechanisms of increased stress resistance in Drosophila melanogaster selected for postponed senescence. Physiol. Zool. 60(3), 321-326, 1987. PLASTICITyOF LONGEVITy 643 SERVICE, P.M., HUTCHINSON, E.W., and ROSE, M.R. Multiple genetic mechanisms for the evolution of senescence in Drosophila melanogaster. Evolution 42, 708-716, 1988. SETO, N.O.L., HAYASHI, S., and TENER, G.M. Overexpression of Cu-Zn superoxide dismutase in Drosophila does not affect life-span. Proc. Natl. Acad. Sci. USA 87, 4270--4274, 1990. SHEPHERD, J.C.W., WALLDORF, U., HUG, P., and GEHR1NG, W.J. Fruit flies with additional expression of the elongation factor EF-la live longer. Proc. Natl. Acad. Sci. USA 86, 7520-7521, 1989. SHIKAMA, N., ACKERMANN, R., and BRACK, C. Protein synthesis elongation factor E F l a expression and longevity in Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 91, 4199-4203, 1994. SOLIMAN, S. and ARKING, R. The relationship between paraquat resistance and oxidative damage in genetically selected long lived and short lived strains. The Gerontologist (in press). STEARNS, S.C. and KAISER, M. The effects of enhanced expression of elongation factor EF-lct on lifespan in Drosophila melanogaster. Genetica 91, 167-182, 1993. THOMPSON, J.N. Phenotypic plasticity as a component of evolutionary change. Trends Ecol. Evolut. 6, 246-249, 1991. TIGERSTEDT, P.M.A. Experiments on selection for developmental rate in Drosophila melanogaster. Ann. Acad. Sci. Fenn. Series A, IV, 1-58, 1969. TOWER, J., WHEELER, J., KURAPATI, R., and BIESKE, E. Novel promoter elements direct aging--Specific transcriptional regulation of heat shock and other genes. Proc. 34th Drosophila Research Conference, 1993:310. TYLER, R.H., BRAR, H., SINGH, M., LATORRE, A., GRAVES, J.L., MUELLER, L.D. ROSE, M.R., and AYALA, F.J. The effect of superoxide dismutase alleles on aging in Drosophila. Genetica 91, 143-149, 1993. WADDINGTON, C.H. The Strategy of the Genes, Allen and Unwin, London, 1940. WEBSTER, G.C. Effects of aging on the components of the protein synthesis system. In: Insect Aging, Collatz, K.G. and Sohal, R.S. (Editors), pp. 207-216, Springer Verlay, Berlin, 1986. WEINHOLD, L.C., AHMAD, S., and PARDINI, R.S. Insect glutathione-S-transferase: A predictor of allelochemical and oxidative stress. Insect Biochem. Physiol. 95B, 355-363, 1990. WEISMANN, A. Essays Upon Heredity and Kindred Biological Problems, 2nd ed., vol. 1, Clarendon Press, Oxford, 1891. WELLS, R.A., BUCK, S., ALl, R., MARZOUQ, O., and ARKING, R. Localization of the longevity genes in D. melanogster. Gerontologist 27, 149A; 1987. ZWAAN, B., BHLSMA, R., and HOEKSTRA, R. Direct selection on lifespan in Drosophila melanogaster. Evolution 49, 649~559, 1995. ZWAAN, B., BIJLSMA, R., and HOEKSTRA, R. On the developmental theory of ageing. I. Starvation resistance and longevity in Drosophila melanogaster in relation to pre adult breeding conditions. Heredity 66, 29-39, 1991.