Experimental Gerontology 35 (2000) 167–185 Forward and reverse selection for longevity in Drosophila is characterized by alteration of antioxidant gene expression and oxidative damage patterns૾ Robert Arkinga,*, Vasanti Burdea, Kevin Gravesa, Raj Haria, Elliot Feldmana, Aaron Zeevia, Sherif Solimana, Ashesh Saraiyaa, Steven Bucka, John Vettrainoa, Kalpana Sathrasalaa, Nancy Wehrb, Rodney L. Levineb a Department of Biological Sciences, Wayne State University, Detroit, MI 48202, USA Laboratory of Biochemistry, National Heart, Lung and Blood Institute, Bethesda, MD 20892, USA b Received 21 September 1999; received in revised form 18 November 1999; accepted 20 December 1999 Abstract Patterns of antioxidant gene expression and of oxidative damage were measured throughout the adult life span of a selected long-lived strain (La) of Drosophila melanogaster and compared to that of their normal-lived progenitor strain (Ra). Extended longevity in the La strain is correlated with enhanced antioxidant defense system gene expression, accumulation of CuZnSOD protein, and an increase in ADS enzyme activities. Extended longevity is strongly associated with a significantly increased resistance to oxidative stress. Reverse-selecting this long-lived strain for shortened longevity (RevLa strain) yields a significant decrease in longevity accompanied by reversion to normal levels of its antioxidant defense system gene expression patterns and antioxidant enzyme patterns. The significant effects of forward and reverse selection in these strains seem limited to the ADS enzymes; 11 other enzymes with primarily metabolic functions show no obvious effect of selection on their activity levels whereas six other enzymes postulated to play a role in flux control may actually be involved in NADPH reoxidation and thus support the enhanced activities of the ADS enzymes. Thus, alterations in the longevity of these Drosophila strains are directly correlated with corresponding alterations in; 1) the mRNA levels of certain antioxidant defense system genes; 2) the protein level of at least one antioxidant defense system gene; 3) the activity levels of the corresponding antioxidant defense system enzymes, and 4) the ability of the organism to resist ૾ Supported by NIA AG 08834 to R.A. * Corresponding author. Tel.: ϩ1-313-577-2891; fax: ϩ1-313-577-6891. E-mail address: rarking @ biology.biosci.wayne.edu (R. Arking) 0531-5565/00/$ – see front matter © 2000 Elsevier Science Inc. All rights reserved. PII: S 0 5 3 1 - 5 5 6 5 ( 9 9 ) 0 0 0 9 4 - 7 168 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 the biological damage arising from oxidative stress. © 2000 Elsevier Science Inc. All rights reserved. Keywords: Longevity; Gene expression; Drosophila; Antioxidant genes; Oxidative damage; Catalase; CuZnSOD; MnSOD; Selection 1. Introduction We previously used artificial selection to develop several long lived strains of Drosophila (Luckinbill et al., 1984; Arking, 1987). We focused our attention on one of those strains (La) and showed that there was no significant difference in metabolic rates between the long-lived La and the normal-lived Ra strains (Arking et al.,1988), a fact that has important implications for the interpretation of subsequent data (Van Voorhies and Ward, 1999). The only factor that robustly separated out all long-lived strains from all normallived strains was paraquat resistance (Arking et al., 1991; Force et al., 1995), suggesting the possibility that an enhanced resistance to oxidative resistance might play some role in the extended longevity phenotype. Chromosome substitution studies showed that it was the 3rd chromosome, and particularly the proximal portion of the left arm (3L), that was necessary for the expression of the extended longevity (Buck et al., 1993). It was known that the structural genes for both copper-zinc superoxide dismutase (CuZnSOD) and catalase (Cat) are located on chromosome 3L (Lindsley and Zimm, 1992). Both of these genes are known to be intimately involved in the inactivation of free radicals (Feuers et al., 1995). We used biomarker analysis to determine that these two strains each underwent the same pattern of senescence but that the extended longevity of the La strain was characterized by a significant early delay in its age of onset of senescence (Arking and Wells, 1990). This finding suggested that the genetic events associated with the delayed onset of senescence had to act early in life, at about 5 to 7 days in adult life. Taken together, these data suggested that we should look in young La adults for an enhanced expression of genes associated with resistance to oxidative stress. When we did that experiment, we demonstrated a tight temporal relationship between the extended longevity of the La strain and the significant coordinate up-regulation in young adults of the CuZnSOD and Cat mRNA levels and enzyme activities (Dudas and Arking, 1995). This confirmed results independently obtained from transgenic work on wildtype Drosophila (Orr and Sohal, 1994; Parkes et al., 1998) and strengthened the supposition that there might exist a causal relationship between resistance to oxidative stress and longevity (Harman, 1959; Martin et al., 1996). Our prior study was limited to the developmental and early adult stages in one set of strains. We have now extended that investigation to include the entire adult lifespan of the same La and Ra strains. In addition, we have now assayed the levels of oxidative damage to lipids and proteins throughout the life span of these strains. To critically test the hypothesized relationship between enhanced antioxidant gene expression and extended longevity, we reverse selected the long-lived La strain for a normal (i.e. an Ra type) life span and then measured the levels of antioxidant and non-antioxidant gene and enzyme expression. Evidence for a possible causal relationship between these two factors would be supported by a specific reversal to normal levels of the elevated antioxidant levels coincident with a reversion of the longevity to normal values. R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 169 2. Materials and methods 2.1. Selection and animal culture The strains were selected and maintained as described previously (Luckinbill et al., 1984; Arking, 1987). The reverse selection experiment was done by taking an aliquot of ca. 200 young adults from the La strain at generation 50 of forward selection, and then selecting these animals and their progeny for early reproduction and short life by using the same procedure as described for the E strains (Arking, 1987). The resulting reverse selected strain was designated as RevLa. Significant decreases in longevity were noted by generation 65 of reverse selection; the stocks are now at generation 119. The La strains are now constitutively long-lived and no longer display the densitydependent aspect of the extended longevity phenotype (Buck et al., 1993). We have observed no significant alteration in any other aspect of the extended longevity phenotype as published previously by us (Arking et al., 1996). Animals were routinely raised at moderately high larval densities (ca. 250/bottle) by allowing ca. 100 to 150 males and females to lay eggs over a 3- or 4-day period and then collecting and segregating the adult cohorts throughout each day of emergence. On reaching the desired age, the cohorts were starved for a few hours in a moist environment, weighed, frozen at Ϫ70°C in an ethanol– dry ice bath, and kept at Ϫ70°C until used. Unless noted otherwise, all assays were done with mixed sex samples. Longevities are routinely measured on all stocks every three to five generations by using replicate bottles containing ca. 200 animals each. Deaths are tallied twice a week. The R and L longevities are very stable (Arking and Buck, 1995). 2.2. Protein carbonyl determination Frozen samples (mean ϭ 63 mg) were homogenized in cold 5% TCA, centrifuged in a microfuge at 0 to 4°C, and the TCA supernatant decanted. A volume of 6 M guanidine equal to the initial homogenate volume was added to each sample, which was then vortexed vigorously and incubated in a 37°C waterbath for 45 to 60 min, vortexed and recentrifuged. Supernatants were collected and stored at Ϫ70°C. For carbonyl analysis, samples were filtered through Millipore Ultra-free-MC 0.45 micron filter units to remove insoluble fragments. Samples were derivatized with 2,4 dinitrophenylhydrazine and analyzed by HPLC run with 6 M guanidine HCl (Levine et al., 1994) except that a Tosohaas gel filtration column (QC PAK TSK GFC 200, catalog #16215) was used to separate labeled proteins from excess reagent. A non-derivatized blank of each sample was subtracted from the derivatized sample. Protein was determined by the BCA method (Pierce and Suelter, 1977). Logistical considerations allowed us to assay only one sample of each age-strain combination and this did not allow standard statistical analysis. Accordingly, we demonstrate repeatability of the data in Fig. 3 by showing the values for the two replicates of the La strains. 2.3. Lipid peroxidation determination This procedure was done using the Lipid Hydroperoxide Assay Kit (Cayman Chemicals, Ann Arbor, cat. #705002) in which the hydroperoxides are isolated in organic solvents, reacted with ferrous ion to yield ferric ion, this product reacted with thiocyanate 170 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 and then the amount of the latter assayed by a spectrophotometer. Samples (ca. 100 mg) of the appropriate strain, age and sex were sorted, deprived of food for about 1 h, put into microfuge tubes, weighed, and frozen at Ϫ80°C in a methanol– dry ice bath and stored at Ϫ70°C. The frozen samples were later homogenized in HPLC-grade water, deproteinized and extracted with N2-saturated methanol/chloroform (1 : 2) in accordance with the kit directions. The extracts were then reacted with ferrous ions and thiocyanate as per instructions and their absorbance read in chloroform at 500 nm in quartz cuvettes in a Beckman DU650 spectrophotometer. Each sample was read in triplicate and compared to standards run with each assay. The concentration of hydroperoxide was calculated for each sample and expressed relative to mg of fly wet weight, and the mean and standard deviation calculated. 2.4. ELISA assays for CuZnSOD protein Appropriately aged and frozen cohorts (100 mg) were homogenized in 2.0 ml of 0.1 M Tris pH 8.3 by using a Pyrex homogenization tube with Teflon pestle, and centrifuged at 9000 rev./min in a microfuge at 4°C. The supernatant was decanted into new Eppendorf tubes and used within the hour. Enzyme-linked immunosorbent assays (ELISA) plates were prepared by using standard procedures (Ausubel et al., 1992). Each ELISA plate had three replicates of each sample plus at least one set of three blank controls for each strain. The controls consisted of 100 l of 1 : 500 rabbit sera/well. The experimental wells were first treated with 100 l of a 1 : 500 primary antibody (Hari et al., 1997) in TTBS, washed, blocked, washed and treated with a 1 : 15 000 secondary antibody (GAR IgG Sigma A-9119, alkaline phosphatase conjugated) in TTBS. After another round of washing and blocking, 100 l of ELISA developing buffer was added to each well. Readings were taken at 30-min intervals over a 5-h period with a BioTek Model EL340 Microplate Kinetic Reader at a wavelength of 405 nm. 2.5. Enzyme assays CuZnSOD enzyme activity was measured with a modification of the standard epinephrine assay in which the activity of the enzyme is measured by its ability to inhibit the auto-oxidation of epinephrine to adrenochrome at elevated pH (Misra and Firdovitch, 1972). In the original paper, inhibition was detected as a change in the slope of the reaction. The Beckman DU650 spectrophotometer we used has the ability to display the time course of the entire reaction on the monitor. We took advantage of this ability to measure the extent of inhibition as the time required by an experimental sample to reach half the maximum absorbance of the control sample. All other conditions are as specified by Misra and Fridovitch (1972). Mean activity values were calculated as the (difference between the experimental and control value)/control values, and expressed finally as activity units/g protein. Catalase enzyme activity was measured using the technique of Goth (1991). For determination of the 17 metabolic enzymes listed in Table 1, 5- to 7-day-old flies were collected simultaneously from all strains and processed as a group. Animals were transferred to clean, empty bottles, chilled on ice for several minutes, and 80 mixed-sex adults (1 : 1 male/female ratio) were placed in a tared microcentrifuge tube and weighed to within 0.1 mg, and a homogenate done by hand with plastic pestles and phosphatebuffered saline with 0.01 M 2-mercaptoethanol at a volume of 2 l/mg wet weight of flies. R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 171 The homogenate was flash frozen at Ϫ70°C. All homogenates were simultaneously thawed and centrifuged at 12 000 rev./min for 15 min at 4°C. The supernatants were removed, avoiding the lipid surface layer, and aliquoted to Teriyaki plates and stored under liquid N2 vapor before use. Plates were thawed and kept on ice during handling and were not refrozen. After thawing, the enzyme activity in the extracts was assayed by using an adaptation of the procedure for colorimetric assay of the enzyme activity in isozyme electrophoresis. The 2-ml reaction mix contained 8 mg of MTT (dimethylthiazole, Innovative Chemistry, Marshfield, MA, USA), 10 mg of NAD or NADP (as required), 5 to 10 mg of substrate, 50 l of a 1 mg/ml solution of Meloda Blue (6-dimethylamino-2,3benzophenoxazine), 50 l of 1 M MgCl2 (as required), and specific substrate reagents for each reaction as specified by Harris and Hopkinson (1978). All reagents were from Sigma, and the reaction mix was prepared fresh. The thawed extracts were further diluted 1 : 20 in phosphate-buffered saline and kept on ice. Reaction mixtures were aliquoted to 96-well plates at 100 l per well and 20 l of homogenate was added in triplicate, mixed, and allowed to react at room temperature for a time previously determined to incorporate the linear portion of the enzyme activity curve. Reactions were terminated by the addition of 100 ul of acidic isopropanol containing 1% (v/v) concentrated HCl. Samples were pipetted thoroughly to dissolve the formazan crystals and promptly read on a BioTech Eliza Plate Reader at 595 nm. All readings were blanked against wells containing everything except the specific substrate. The protein content of each sample was determined with the dye binding assay (kit #500-0002, BioRad Labs, Richmond, CA, USA). The data from the three replicates of each sample was tallied and enzyme activity expressed as the mean (Ϯ SD) absorbance units/g protein/min. 2.6. mRNA Analysis Total RNA was isolated and used for both Northern and dot blot construction by using the TriR reagent (guanidine thiocyanate and phenol, catalog #TR-118, Molecular Research Center Inc., Cincinnati, OH, USA) and followed the protocol provided by the company. Ten micrograms of total RNA from each strain and age sample was applied to a 1.5% agarose gels made with a 1 ϫ MESA buffer and 0.66 M formaldehyde. Electrophoresis was at 200 volts for 2 to 3 h with a 1 ϫ MESA/0.66 M Formaldehyde running buffer. RNA then was transferred to neutrally charged GeneScreen nylon membrane NEF 985 (DuPont–NEN) via capillary action. Dot blots were made by spotting two to four 10-g samples of total RNA from each strain-age combination to a GeneScreen membrane. The various probes used for hybridization were amplified and isolated as previously described (Dudas and Arking, 1995) and labeled using the Renaissance Random Primer Extension System NEP 103 (DuPont–NEN). Prehybridizations and hybridizations were done in Northern Max Prehyb/Hyb solution (Ambion catalog #8677). Prehybridizations were performed for 3 to 18 h at 42 to 46°C in a hybridization oven with rotation. The used prehybridization solution was then drained and replaced with 5 to 10 ml (depending on blot size) of hybridization solution (5 ϫ SSC, 1 ϫ Denhardt’s, 100 g/ml salmon sperm DNA, 31.25% formamide, 0.1% SDS and 1% dextran sulfate). The blots were hybridized in an oven with rotation at 42 to 43°C for up to 48 h. The blots were then washed in Wash Solution I (2 ϫ SSC, 0.1% SDS) and Wash Solution II (0.1 ϫ SSC, 0.1% SDS) until the signal-to-background ratio for the blots was high. Autoradiography was done at Ϫ70°C by using an intensifying screen. Autoradiograms were quantitated using the Ambis Radioanalytic Imaging System. Between hybridizations, the blots were stripped of probe by 172 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 boiling for 15 to 30 min in a solution consisting of 10 mM Tris HCl, pH8.0; 1 mM EDTA; and 1% SDS. The CuZnSOD and Cat probes were a gift of Dr. W. Orr (Dept. of Biological Sciences, Southern Methodist University). The MnSOD probe was a gift of Dr. John Phillips (Dept of Molecular Biology and Genetics, University of Guelph). The rRNA probe was a gift of Dr. Charles Rozek (Case Western University). The data for any one Northern band or dot blot was expressed as the ratio of the number of O.D. units of the specific gene probe relative to the number of O.D. units of rRNA measured in that same dot or lane. The data from the replicates was tallied and expressed as the mean Ϯ SD of this ratio. Normalization of data across different blots or the same blot using independently labeled probes of the same type was accomplished by including 10 g total RNA each of 5 day old Canton-S and Oregon-R on each blot as an internal control and adjusting the data according to the intensity of the control signal. All such control RNA was obtained from the same extraction of the same population. 2.7. Statistical analysis Statistical analysis of survival curves was done using the Kaplan–Meier Survival procedure (Log-rank test) in SPSS for Windows, v.7 (SPSS Inc., Chicago, IL, USA). Differences between the mean values of assayed cellular components (such as mRNA levels, CuZnSOD protein levels, enzyme activity levels and oxidative damage levels) of two strains at a given age were done using the Student’s t-test procedure for independent paired or unpaired samples as appropriate (Zar, 1974), without correction for multiple comparisons (Harp, 1963, pp. 487– 489; Keppel and Zedeck, 1989, pp. 172–179). For the three strains involved in the CuZnSOD enzyme data of Fig. 6, a one-way ANOVA was performed followed by a post-hoc Tukey HSD test by using the procedures in SPSS for Windows, v.7 (Zar, 1974; Keppel and Zedeck, 1989; SPSS Inc., Chicago, IL, USA). 3. Results 3.1. Effects of forward selection 3.1.1. Longevity Forward selection for extended longevity was carried out as previously described (Arking, 1987Fig. 1). In brief, the normal lived Ra strain is the progenitor of the other strains and is maintained under conditions such that they are under no overt selection pressure for alteration in longevity. Selection for long life span was done indirectly via the manipulation of the age at which reproduction was allowed to occur. Selection for extended longevity (La strain) was very effective, bringing about significant alterations in mean and maximum longevity within ca. 22 generations (Arking, 1987). 3.1.2. Oxidative damage Protein carbonyl has been shown by Stadtman (1992) to be a good indicator of oxidative damage to proteins. When we determined the age-related changes in the amount of protein carbonyl present in males of each of these two strains (Ra and La), we found that there were obvious differences in the level of protein carbonyl (Fig. 2). The La strain seems to enter a protective phase in which it shows no net increase in its level of protein carbonyl from early adulthood through Day 30 (Fig. 2). After that time, they enter a R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 173 Fig. 1. Survival curves of representative generations of the selected strains used in this study. The long lived La (F80) strain was measured in February 1996; the normal Ra (F151) strain was measured in September 1995; the reverse selected Rev-La strain (F96) was measured in February 1997. There is a significant difference between the Ra and La curves, and between the La curves and the RevLa curves. Each survival curve is based on the age specific values obtained from two or three replicate cohorts consisting of at least 250 mixed sex individuals each. damage phase in which their levels of oxidative damage show a sharp net increase and climb to levels as high as those seen in the Ra strain do. In contrast, the Ra strain seems to have no protective phase and instead moves straight into its damage phase that lasts for at least the first 24 days of its adult life. Because of logistical reasons, there were not enough samples on which to do valid statistics; however, the data for the independent replicates of the La strain shows that the measurements are repeatable and seem to be different from the Ra values. Fig. 2. The age-related changes in the protein carbonyl content of the La and Ra series of strains. There is a clear difference between the La and Ra strains such that the long-lived strains have lower levels of oxidative damage between Days 13 through 30. The repeatability of the data are suggested by the La values that were obtained from two independent replicate lines. See text for discussion. 174 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 Fig. 3. Age-and strain-related differences in the mean (Ϯ SD) lipid hydroperoxide levels of the La/Ra strain, based on the assay in triplicate of 3 to 6 samples per data point. Note that the long-lived La strains have lower mean values during the first 3 weeks of life. The L and R strains show significant differences (t-test, P Ͻ .05) at days 5, 9, 21 and 67. See text for discussion. Lipid peroxidation is a good measure of the molecular damage done to lipids by oxidative stress. We examined the levels of lipid peroxidation in the La and Ra strains. There is an obvious and statistically significant difference between the mean lipid hydroperoxide levels present in the La animals relative to their respective controls (Fig. 3). Although both strains show an age-related increase in their lipid hydroperoxide levels, the patterns are quite different. The La animals show a gradual increase in lipid hydroperoxide levels from Day 5 to Day 51 followed by a slow but statistically significant decrease through Day 61. The Ra animals, on the other hand, show a rapid increase in oxidative damage early in life (between Days 5 and 11), followed by a plateau period (from Day 11 to Day 22), succeeded by a possible decrease in lipid hydroperoxides at Day 31 through 42 followed by a late-life increase through Day 60. There are statistically significant differences between the mean values of lipid hydroperoxides in the two strains at Days 5, 9, 21, and 67. 3.2. Antioxidant defenses 3.2.1. CuZnSOD protein and enzyme activity levels We measured the actual amount of CuZnSOD protein present in the La and Ra animals throughout most of their life span by ELISA (Fig. 4), using a newly developed antibody specific to this Drosophila protein (Hari et al., 1998). During early life, the La animals have about twice as much CuZnSOD protein as do the Ra animals, and they maintain this high and statistically significant ratio for at least 14 days. Note that at or sometime before Day 33, the Ra strain shows an unexpected but repeatable increase in its CuZnSOD protein content that is still elevated at Day 44. This (apparent) Day 33 increase in R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 175 Fig. 4. The age-related changes in the mean (Ϯ SD) absolute levels of CuZnSOD protein in the La and Ra strains as determined with a newly developed antibody in an ELISA assay, based on the analysis of three independent assays at the indicated ages. Note that the La animals have a significantly higher amount of protein than do the Ra animals over the first two weeks of adult life, and these differences are statistically significant (t-test, P Ͻ .05) at Days 9 and 14. At 33 days and later, the protein level in the Ra strain increases, thereby decreasing the relative level of the protein in the La strain. See text for discussion linking together the data of Figs. 3 through 6. CuZnSOD protein level may correlate with the decrease in the Ra strain’s level of protein carbonyl at Day 31 (Fig. 2) and its level of lipid hydroperoxides at Day 31 (Fig. 3). It remains to be seen whether this apparent correlation is real or not. Thus the Ra strain’s early life low levels of CuZnSOD protein are correlated with high levels of oxidative damage whereas its mid-life increased CuZnSOD protein levels are correlated with a decreased level of oxidative damage. And there is a good correlation between the La strains generally higher levels of CuZnSOD protein and its generally lower levels of oxidative damage. An apparent relationship between these two variables therefore can be detected both within and between strains. Fig. 5 shows the CuZnSOD enzyme activity measured throughout the lifespan of La males and their progenitor Ra strains. Relative to the Ra strain, the La strain shows a large and significant increase in enzyme activity from Day 5 through Day 20, an equivalent amount at Day 30 followed by a second significant relative increase on Days 40 and 50. These mean values are significantly different between the two strains at all ages save for Day 30. A one-way ANOVA (F ϭ 12.649, df ϭ 37, P ϭ 0.0005) indicated that the three strains differed significantly in their CuZnSOD enzyme activities. The Tukey HSD test showed that the Ra and La strains are significantly different from each other at ages 5 to 20 days (q ϭ 4.961, P ϭ 0.0005) whereas the Ra and RevLa strains are not statistically different from each other at ages 5 to 20 days (q ϭ 1.868, P ϭ 0.163). At the later ages, the differences between the La and Ra strains are statistically significant (t-test, P Ͻ 0.05) at all ages except Day 30. Note that the CuZnSOD protein levels (Fig. 4) and CuZnSOD enzyme activity levels (Fig. 5) for each strain are in good agreement with one another for each of the corresponding ages of adults assayed in these two independent experiments. 176 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 Fig. 5. The age-related changes in the mean (Ϯ SD) CuZnSOD enzyme activity measured in males of the Ra, La, and reverse selected Rev-La strains. The La strains generally have a higher specific activity than do their Ra controls, although the early and late periods of elevated activity are obvious in both strains. The Rev-La strains have a shortened lifespan (Fig. 1); reverse selection has brought about a decrease in this ADS enzyme activity such that it is now comparable to their normal-lived Ra strain animals. Each point is the mean of 3 to 5 replicates; points without visible error bars are those where the variance is so small that it is hidden by the data symbol. Note that there is a good correlation between the amount of CuZnSOD protein present (Fig. 5) and the corresponding age-strain enzyme activity (these data). 3.2.2. Changes in gene expression Three important antioxidant defense genes are CuZnSOD, MnSOD and Cat. We assayed the relative age-specific levels of each of these mRNAs in the La and Ra strains (Fig. 6a– c). In the Ra strain, each of the three antioxidant gene products is present at a more or less constant level throughout life (Fig. 6a– c); there is no obvious stage specificity of gene expression in this normal-lived animal. In the La strain, the levels of all three antioxidant gene products increase significantly during the first half of life. The CuZnSOD mRNA levels are equivalent in the two strains at Day 5 but increase thereafter and reach a significantly high plateau from Days 30 through 60 before undergoing a terminal decline (Fig. 6a). The MnSOD mRNA levels show a similar pattern in the La animals, being significantly higher than the controls for Day 9 onwards, except that they plateau from Days 42 to 60 before undergoing a terminal decline which is nonetheless significantly higher than the levels seen in the comparably aged Ra strain animal (Fig. 6b). The Cat mRNA levels in the La animals follow a somewhat different pattern, being higher than the Ra controls from Day 1 onward (Fig. 6c). They gradually increase to a significantly higher peak level between Days 20 to 40 and then plateau at a lower level from Days 50 to 70 (Fig. 6c). However, Cat shows the smallest overall increase in the La relative to the Ra control and it seems to have a high variance. The net effect of these changes in gene expression is that the La strain has a much higher amount of these three antioxidant mRNAs than does the Ra strain, particularly the CuZnSOD and MnSOD mRNAs. These “surplus” antioxidant gene products are first available to the La organism at Day 9; and this situation continues throughout the rest of the life span. R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 177 Fig. 6. The mean (Ϯ SD) levels of CuZnSOD (a), MnSOD (b), and Cat (c) gene expression patterns in the La and Ra animals as determined by a quantitative analysis of the mRNA prevalence for these three gene products. Note that the Ra animals have a more or less constant level of these gene products over their entire measured life span. The La animals, on the other hand, not only have significantly higher levels for each gene product but also show an apparently coordinate regulation of the three ADS gene products such that they begin increasing at Days 5 through 9 and peak at Days 31 through 40. Two to six independent samples (median ϭ three) were used for the assay of each of the 108 different age-strain combinations. Data points without visible error bars are those where the standard deviation is so small that it is hidden by the data symbol. The asterisks indicate the ages at which the difference between the two strains is statistically significant (t-test, P Ͻ .05). 178 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 Fig. 7. The mean (Ϯ SD) levels of CuZnSOD and Cat mRNA levels were measured in the Rev-La strains, and compared to their respective Ra and La controls. Reverse selection results in a shorter lifespan (Fig. 1) and significantly decreased mRNA levels for each of these ADS genes. Note that in the Rev-La strain, the CuZnSOD but not the Cat genes show significant decreases (Days 20 to 50) relative to its progenitor La strain. Three samples were assayed at each point for the Rev-La CuZnSOD but only one sample for the Rev-La Cat that was inadvertently destroyed. With the exception of the Rev-La Cat data, points without visible error bars are those where the variance is so small that it is hidden by the data symbol. The differences in the patterns of CuZnSOD mRNA levels (Fig. 6) versus that of the CuZnSOD protein levels (Fig. 4), CuZnSOD enzyme activities (Fig. 5) and the organisms’ lifetime response to paraquat (Arking et al., 1991) suggests that significant post-transcriptional regulation of the CuZnSOD protein turnover and enzyme activities may be taking place. 3.3. Effects of reverse selection 3.3.1. Loss of extended longevity Aliquots of the La strains were subjected to reverse selection for shortened longevity, using the same procedure as described previously for the early reproduced and short-lived E strains (Luckinbill et al., 1984; Arking, 1987). Reverse selection was effective in abolishing the extended longevity phenotype characteristic of the progenitor La strain, reducing the mean life span from ca. 74 days to ca. 51 days within 65 generations (Fig. 1). The reverse selected La strain now has an intermediate longevity that is slowly moving back to the Ra level and that is significantly different from the La survival curve (Log-rank test ϭ 165.89, P Ͻ 0.00005). Reverse selection for decreased longevity has been successful. These strains have been termed the Rev-La lines. 3.3.2. Loss of enhanced antioxidant defenses We determined the age-specific levels of the CuZnSOD and Cat mRNA levels in the Rev-La strain (Fig. 7). The expression patterns of these two antioxidant defense genes are significantly different from that characteristic of the L strains, and have reverted to the low and constant levels generally characteristic of the R normal lived strains (Fig. 6a and c). However, note that the major effect of reverse selection at this level has been to R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 179 Table 1 Effect of forward and reverse selection on enzymes involved in glucose metabolism and antioxidant defense Enzyme Enzyme Activity in 5 day old: Difference from: a R3L L3Rev-L 2.62 Ϯ 0.14 3.87 Ϯ 0.09 5.30 Ϯ 0.13 7.01 Ϯ 0.09 6.99 Ϯ 0.14 4.63 Ϯ 0.01 2.57 Ϯ 0.05 2.49 Ϯ 0.03 3.22 Ϯ 0.08 6.19 Ϯ 0.07 9.87 Ϯ 0.27 2.18 Ϯ 0.04 3.70 Ϯ 0.03 5.66 Ϯ 0.05 11.99 Ϯ 0.24 4.62 Ϯ 0.06 3.70 Ϯ 0.08 Ϫ14.4% ϩ4.1 ϩ7.4 Ϫ9.9 Ϫ7.6 ϩ2.8 ϩ23.1 Ϫ9.5 ϩ6.1 Ϫ7.50 ϩ9.6 ϩ15.5 ϩ15.7 Ϫ12.1 ϩ21.84 ϩ17.1 ϩ12.0 ϩ4.2% Ϫ9.4% Ϫ6.9 Ϫ10.6 Ϫ11.0 Ϫ7.2 Ϫ2.2 Ϫ7.5 Ϫ3.4 Ϫ11.1 Ϫ10.1 Ϫ6.0 Ϫ14.2 Ϫ18.5 ϩ15.5 Ϫ24.5 Ϫ22.3 Ϫ27.6 Ϫ10.5% 61.0 Ϯ 2.0 45.6 Ϯ 3.6 ϩ49.1% Ϫ60.2% Ϫ25.6% ϩ50.1% Ra La Rev-Lab 3.38 Ϯ 0.17 3.99 Ϯ 0.18 5.51 Ϯ 0.12 8.74 Ϯ 0.16 8.16 Ϯ 0.26 4.61 Ϯ 0.03 2.26 Ϯ 0.04 2.84 Ϯ 0.09 3.42 Ϯ 0.12 7.44 Ϯ 0.06 9.58 Ϯ 0.18 2.20 Ϯ 0.04 3.92 Ϯ 0.06 5.57 Ϯ 0.04 13.29 Ϯ 0.25 5.08 Ϯ 0.18 4.56 Ϯ 0.14 Mean change 2.89 Ϯ 0.13 4.16 Ϯ 0.05 5.92 Ϯ 0.03 7.87 Ϯ 0.03 7.53 Ϯ 0.19 4.73 Ϯ 0.02 2.78 Ϯ 0.06 2.58 Ϯ 0.07 3.63 Ϯ 0.08 6.88 Ϯ 0.10 10.50 Ϯ 0.30 2.53 Ϯ 0.01 4.54 Ϯ 0.04 4.90 Ϯ 0.04 16.04 Ϯ 0.33 5.95 Ϯ 0.10 5.11 Ϯ 0.06 55.3 Ϯ 8.7 76.3 Ϯ 3.2 82.2 Ϯ 10.7 30.4 Ϯ 3.0 b General metabolic enzymes Alcohol dehydrogenase Glucose dehydrogenase Glyceraldehyde PDH Glycerol 3PDH Lactic DH NADH diaphorase Phosphoglucomutase Phosphogluconate DH Xanthine DH Malic DH Adenylate kinase Glucose-6-phosphate DH Malic enzyme NADPH diaphorase Glucose Phosphate Isomerase Hexokinase Isocitrate dehydrogenase Antioxidant defense enzymesc CuZnSOD: Catalase a Note that the RevL values are based on a combined (1 : 1) population of Rev-La and Rev-Lb. Values expressed as mean absorbance units/g protein/min Ϯ SD; three replicates/sample. c Values expressed as specific activity/g protein; 3-5 replicates/sample; CuZnSOD activities ϫ 10Ϫ4; Catalase activities ϫ 10Ϫ2. b significantly decrease the CuZnSOD mRNA prevalence in the Rev-La strain relative to the La control, the effect on the Cat mRNA prevalence being proportionally much smaller (Fig. 7). We also determined the age-specific levels of the CuZnSOD enzyme activity in the Rev-La strain (Fig. 5). Reverse selection applied to the La strain resulted in the Rev-La strain having CuZnSOD enzyme levels that are significantly lower than those of its progenitor long-lived strain (Fig. 5). 3.3.3. Specificity of the Selection Effects on Antioxidant Defenses It was necessary to determine whether the changes in the mRNA and enzyme activity levels in the selected long-lived lines reflect specific alterations in the animals’ antioxidant enzyme activities or only a general and non-specific alteration of the organisms’ metabolism. To ascertain this point, we measured the specific activity of the 17 enzymes listed in Table 1 in the Ra, La and Rev-La strains. Forward selection (R– Ͼ L) for increased longevity caused a ϩ4.2% mean increase in these specific enzyme activities, whereas reverse selection (L– Ͼ Rev-L) for decreased longevity brought about a mean decrease of Ϫ10.5%. By contrast, the change brought about by forward selection on the CuZnSOD specific activity of the La lines was ϩ49.1%, whereas reverse selection brought about a 25.6% decrease in this antioxidant enzyme activity. These alterations are 2- to 10-fold larger than the values reported for forward and reverse selection on the general metabolic 180 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 enzymes in these same lines. Similarly large changes were brought about in the activity levels of the catalase enzyme but it is important to note that they are opposite in direction to that of the CuZnSOD enzymes (Table 1). The large difference between the mean changes in both antioxidant defense enzymes when compared to the 17 tested metabolic enzymes indicates that the CuZnSOD and Cat enzyme activities are specifically altered by selection for longevity and this happens independently of the other enzymes. 4. Discussion 4.1. Oxidative stress and longevity Many laboratory manipulations can lead to significant life span extension but all of these extended longevity phenotypes are associated with an increased resistance to various environmental stresses. Parsons (1995) pointed out that evolutionary considerations predict that selection for stress resistance should always be accompanied by changes in longevity. In fact, it has been explicitly stated that .“..the ability to respond to stress is the rate determining factor leading to aging and senescence” (Johnson et al., 1996, p. B393). This paper presents data demonstrating that selection of our normal-lived Ra strain for extended longevity resulted in a long-lived La strain that was distinguished from its progenitor by; 1) a significantly enhanced expression of certain antioxidant defense system (ADS) mRNAs such as CuZnSOD, MnSOD, and (to a lesser extent) Cat; 2) an enhanced level of the CuZnSOD protein; 3) a significantly increased activity of the corresponding CuZnSOD and Cat enzymes; and 4) a significantly enhanced resistance of the organism to the biological damage arising from oxidative stress. In addition, we have previously reported that the only factor robustly associated with extended longevity in all of our long-lived strains is resistance to paraquat such that the long lived La strains have a significantly greater paraquat resistance than does the normal-lived (Ra) strain for at least the first three weeks of adult life, with the maximum differences being seen at Day 5 (Force et al., 1995). Reverse selection of the La strain for a decreased longevity results in 1) a significantly decreased expression of these same ADS mRNAs; and, 2) a significantly decreased activity of the corresponding ADS enzymes. The large and statistically significant changes in the ADS enzyme activities associated with both forward and reverse selection are specifically restricted to those ADS enzymes and are not observed in 17 other ‘ordinary’ metabolic enzymes; thus they are not the result of some non-specific and general metabolic reorganization of the organism. A reasonable interpretation of the data presented here suggests that alterations of longevity in these strains is accompanied by, and might well be due to, corresponding alterations in the expression of specific ADS genes at the mRNA, protein and enzyme levels. It is these changes in gene expression that confer on the organism the ability to significantly and specifically resist the deleterious effects of oxidative stress. Resistance to oxidative stress translates into the organism suffering lower levels of oxidative damage to lipids and proteins, as shown above, and it is likely that it is this significantly lower level of cellular damage that may be responsible for the observed delayed onset of senescence. Thus, the extended longevity, observed in the La animal relative to the Ra animal (Arking and Wells, 1990). This interpretation is supported by the work of Curtsinger (1998) who have done quantitative trait loci (QTL) mapping on a sister line to our La strain and demonstrated R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 181 that a statistically significant QTL for longevity (as well as one for paraquat resistance) is localized to a region that includes the CuZnSOD locus. Thus the same strain independently analyzed by two laboratories using different approaches, nonetheless, yields complementary sets of data. Each set of data is fully consistent with the interpretation that the extended longevity is dependent on an over expression of certain ADS genes that gives rise to an increased resistance to oxidative stress. 4.2. Can changes in longevity be explained by the altered expression of a small number of genes? There has been a long standing assumption that quantitative traits such as longevity must be under the control of a large number of genes; that it is a continuously varying polygenic trait in the classical sense. Fleming et al. (1993) operate on the assumption stated above and have estimated that there are 200 to 400 genes involved in their extended longevity phenotype. But there are both theoretical and empirical evidence that do not support this assumption. Thompson (1975) has shown that quantitative traits that are generally considered to be due to the effects of many genes each of which has only a small effect, can just as accurately and empirically be considered to be due to the effects of only a few genes, each of which has a large effect. Evidence to support this contention is being obtained from the genetic studies on longevity being conducted in various model organisms in which a characteristically small number of genes are demonstrated to cause significant increases in longevity. Both Curtsinger and his colleagues (Curtsinger et al., 1998; Resler et al., 1998) and Mackay and her colleagues (Nuzhdin et al., 1997), using different sets of strains, each identified a relatively small number (ca. 5ϩ) of statistically significant QTLs. A more direct demonstration involves the use of transgenes. For example, Orr and Sohal (1994) used transgenic techniques to increase the expression of the CuZnSOD and Cat genes by ca. 50% in their wildtype strain. Three of their transgenic lines yielded significant increases in mean and maximum longevity coupled with a decreased level of oxidative damage. Parkes et al. (1998) showed that over expression of only the CuZnSOD gene in the motor neurons of transgenic Drosophila led to a 40% extension of normal life span. And Sun and Tower (1998) have shown that controlled induction of a CuZnSOD transgene results in significant extension of the organisms’ longevity, and that the amount of CuZnSOD induction correlates with the amount of life span extension. The assumption that longevity must be under the control of a very large number of genes does not seem to be supported by the data so far available. Thus, given the theoretical and experimental data supporting the causal role of small numbers of particular genes in bringing about the extension of longevity, we interpret the data presented herein as demonstrating that the specifically enhanced expression of a few different antioxidant defense genes plays a major, if not primary, role in bringing about a delayed onset of senescence and an extended longevity in our La strain. Our data does suggest the possible existence of post-translational regulation of the ADS enzyme activity levels (e.g. the high steady level of La CuZnSOD mRNA at Day 60 (Fig. 6a) is not reflected in the decreasing La CuZnSOD enzyme activity (Fig. 3) noted at the same time). The ADS enzyme activity levels may be regulated at any or all of several different organizational levels, thus possibly allowing for complex evolutionary relationships. 182 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 4.2.1. Reversion leads to loss of enhanced antioxidant gene expression This chain of events works both ways. When the La strain is reverse selected for a normal lifespan, it loses its characteristic extended longevity. It also loses its characteristic La type pattern of ADS mRNA prevalence and displays an ADS gene activity pattern similar to that of the Ra strain (Fig. 7). The specific activity of the CuZnSOD enzyme significantly decreases as well (Fig. 4). In the selection process that gives rise to the Rev-La strain, the CuZnSOD and catalase enzyme activities change in opposite directions (Table 1); as a consequence, the Rev-La strain has antioxidant enzyme activity levels similar to that of the Ra strain. Forward selection (Fig. 1) showed significant results in ca. 22 generations of selection (Luckinbill et al., 1984; Arking, 1987). The reverse selection (Fig. 1) required at least 65 generations to only partially (but significantly) decrease the lifespan of the long-lived La strain. These observations suggest that the reverse selection response may have been due to new mutations affecting the ADS gene expression rather than to lack of fixation of the selected alleles in the long-lived La strain. 4.2.2. Selection on longevity specifically affects only antioxidant gene expression patterns Our data show that the effects of selection on longevity have specific effects on the expression of at least these three antioxidant genes. It is important, however, to address the question as to whether the selection effects are restricted to the antioxidant genes alone, or are generalized throughout the metabolic landscape (e.g. Clark et al., 1995). The latter circumstance would make it difficult to develop a mechanistic linkage between selection for longevity and resistance to oxidative stress. To answer this question, we assayed how selection alters the specific activities of a large number of enzymes involved in different aspects of carbohydrate metabolism and compared these results with that for CuZnSOD (Table 1). Selection for extended longevity results in a 4.2% overall increase in the general metabolic activities, whereas reverse selection for shortened longevity results in a 10.5% overall decline in these same values. In contrast, CuZnSOD shows a 49.1% increase in its specific activity as a function of selection for extended longevity, and a 25.6% decrease when selected for shortened longevity. This large difference in the responses of the two sets of gene products suggests that the coordinate changes in the expression of the antioxidant genes in the La strain is the result of a selective effect acting on some specific process(es) of antioxidant gene regulation and is not the result of a non-specific stimulation of the metabolism. 4.3. Other genes whose expression may be correlated with the ADS genes If the ADS gene products are indeed a critical component determining life span, then it follows that other specific enzymes and enzyme systems that support the ADS might also be affected by selection for longevity. One example of such a supporting metabolite is NADPH, which is required by both catalase and glutathione-S-reductase for the detoxification of peroxides formed after the dismutation of superoxide by CuZnSOD (Feuers et al., 1995). Elevated ADS activities require correspondingly elevated NADPH levels. The enzymes glucose-6-phosphate dehydrogenase (G6PDH) and isocitrate dehydrogenase (IDH) both generate NADPH and both have been implicated in influencing life span in Drosophila (Riha and Luckinbill, 1996; da Cunha et al., 1996). R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 183 Classic insect biochemistry studies have shown that insects usually have a deficient NADPH reoxidizing system (Chefurka, 1965); thus the activity of the pentose shunt depends on the rate of reoxidation of NADPH. The ADS enzymes could function as a NADPH reoxidizing system, and their increased operation would stimulate the pentose shunt at the expense of the TCA cycle. This efficient production of ATP without concomitant production of an “oxygen debt” might underlie the La strain’s extraordinary flight endurance (Graves et al., 1988). This possible explanation ties together two otherwise disparate aspects of the phenotype. The last six enzymes listed in Table 1 are involved either directly in the production of NADPH (i.e. isocitrate dehydrogenase) or indirectly via their ability to control the flux through the glycolytic or pentose shunt pathways (i.e. glucose-6-phosphate dehydrogenase). All six of these enzymes show large increases in specific activity correlated with the direction of selection (R– Ͼ L ϭ ϩ11.4%; L– Ͼ RevL ϭ Ϫ15.4%) and which are at least twice as large as the changes shown by the other 11 enzymes of Table 1 (R– Ͼ L ϭ ϩ0.4%; L– Ͼ Rev ϭ Ϫ7.5%). It is reasonable to postulate that the large and correlated responses in the latter set of six enzymes might well arise because of the metabolic alterations necessary to produce high NADPH levels for the support of the ADS enzymes (Chefurka, 1965; da Cunha et al.,1996). Then, in our view, the metabolic alterations described above play an important role in supporting the high ADS enzyme activities but do not seem to play a primary role by themselves in longevity extension as suggested by others (Luckinbill et al., 1990). Several authors have suggested that a generalized stress resistance is a prerequisite for the expression of the extended longevity phenotype (e.g. Johnson et al., 1996; Djawden et al., 1998). However the empirical data suggests that it is specifically the resistance to oxidative stress that constitutes the operative factor (Sohal et al., 1993; Sohal and Weindruch, 1996; Martin et al., 1996; Taub et al., 1999; this report). We have other data that bears on this point. As noted above, the La animals were derived from the Ra strain and are resistant to paraquat, have enhanced ADS enzyme activities, and live long. When we employed a direct selection regime on the Ra animals, we created a new strain (PQR) extraordinarily resistant to paraquat but which had elevated P450 enzyme levels and depressed ADS enzyme activity levels (Arking, 1998; Vettraino and Arking, in preparation). This PQR strain show a normal longevity and is statistically indistinguishable from the Ra strain. We believe that the data allows us to conclude that it is probably the specific protective effects of the enhanced ADS enzymes that actually produce the extended longevity in this model system. Acknowledgments We especially thank David Mancini and Mark Oliecwicz for their capable assistance with the work. We also thank Dr Trudy Mackay for her helpful comments on the manuscript. References Arking, R. (1987). Successful selection for increased longevity in Drosophila: analysis of the survival data and presentation of a hypothesis on the genetic regulation of longevity. Exp Gerontol 22, 199 –220. 184 R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 Arking, R. (1998). Molecular basis of extended longevity in selected Drosophila strains. Curr Science 74, 859 – 864. Arking, R. & Buck, S. (1995). Selection for increased longevity in Drosophila melanogaster: A reply to Lints. Gerontology 41, 69 –76. Arking, R., Buck, S., Berrios, A., Dwyer, S., & Baker, G. T., III. (1991). Elevated antioxidant activity can be used as a bioassay for longevity in a genetically based long-lived strain of Drosophila. Develop Genetics 12, 362–370. Arking, R., Buck, S., Wells, R. A., & Pretzlaff, R. (1988). Metabolic rates in genetically based long-lived strains of Drosophila. Exp Gerontol 23, 59 –76. Arking, R., Force, A. G., Dudas, S. P., Buck, S., & Baker, G. T., III. (1996). Factors contributing to the plasticity of the extended longevity phenotypes of Drosophila. Exp Gerontol 31, 623– 643. Arking, R. & Wells, R. A. (1990). Genetic alteration of normal aging processes is responsible for extended longevity in Drosophila. Develop Genetics 11, 141–148. Ausabel, F. M., Brent, R., Kingston, R., Moore, D., Seidman, J.G., Smith, J.A., & Struhl, K. (1992). Short Protocols in Molecular Biology, 2nd ed. New York: John Wiley and Sons. Buck, S., Wells, R. A., Dudas, S. P., Baker, G. T., III, & Arking, R. (1993). Chromosomal localization and regulation of the longevity determinant genes in a selected strain of Drosophila melanogaster. Heredity 71, 11–22. Chefurka, W. (1965). Intermediary metabolism of carbohydrates in insects. In M. Rockstein (Ed.). The Physiology of Insecta, Vol. II (pp. 669 –768). New York: Academic Press. Clark, A. G., Wang, L., & Hulleberg, T. (1995). P-element induced variation in metabolic regulation in Drosophila. Genetics 139, 337–348. Curtsinger, J. (1998). Quantitative trait loci mapping and cloning of longevity assurance genes in Drosophila (abstr.). Talk given at GSA national meeting, Philadelphia, PA, Nov. 22 1998. The Gerontologist 38 (special issue):92;and personal communication. Curtsinger, J. W., Fukui, H. H., Resler, A. S., Kelly, K., & Khazaeli, A. A. (1998). Genetic analysis of extended life spans in Drosophila melanogaster. I. RAPD screen for genetic divergence between selected and control lines. Genetica 104, 21–32. da Cunha, G. L. & de Oliveira, A. K. (1996). Citric acid cycle: A mainstream metabolic pathway influencing life span in Drosophila melanogaster? Exp Gerontol 31, 705–715. Djawdan, M., Chippendale, A. K., Rose, M. R., & Bradley, T. J. (1998). Metabolic reserves and evolved stress resistance in Drosophila melanogaster. Physiol Zool 71, 584 –594. Dudas, S. P. & Arking, R. (1995). A coordinate upregulation of antioxidant gene activities is associated with the delayed onset of senescence in a long-lived strain of Drosophila. J Gerontol: Biol Sci 50A, B117–B127. Feuers, R. J., Duffy, P. H., Chen, F., Desai, V., Oriaku, E., Shaddock, J. G., Pipkin, J. W., Weindruch, R., & Hart, R. W. (1995). Intermediary metabolism and antioxidant systems. In R. W. Hart, D. A. Neumann, R. T. Robertson (Eds.), Dietary Restriction, Implications for the Design and Interpretation of Toxicity, and Carcinogenicity Studies (pp. 181–196). Washington, DC: ILSI Press. Fleming, J. E., Spicer, G. S., Garrison, R. C., & Rose, M. R. (1993). Two-dimensional protein electrophoretic analysis of postponed aging in Drosophila. In M. R. Rose, C. E. Finch (Eds.), Genetics and Evolution of Aging (pp. 199 –214). The Netherlands: Kluwer Academic Publishers, Dordrecht. Force, A. G., Staples, T., Soliman, S., & Arking, R. (1995). Comparative biochemical and stress analysis of genetically selected Drosophila strains with different longevities. Develop Genetics 17, 340 –351. Goth, L. (1991). A simple method for determination of serum catalase activity and revision of reference range. Clin Chim Acta 196, 143–51. Graves, J. L., Luckinbill, L. S., & Nichols, A. (1988). Flight duration and wing beat frequency in long- and shortlived Drosophila melanogaster. J Insect Physiol 34, 1021–1026. Hari, R., Burde, V., & Arking, R. 1998. Preparation of a synthetic antibody specific for CuZnSuperoxide Dismutase protein of Drosophila melanogaster. Exp Gerontol 33, 227–237 Harmon, D. (1956). Aging: a theory based on free radical and radiation chemistry. J Gerontol 11, 298 –300. Harp, W. L. (1963). Statistics for Psychologists. New York: Holt, Rinehart and Winston. Harris, H. & Hopkinson, D. A. (1978). Handbook of Enzyme Electrophoresis in Human Genetics (Suppl.). Amsterdam: North–Holland Publishing Co. Johnson, T.E., Lithgow, G.J., & Murakami, S. (1996). Hypothesis: interventions that increase the response to stress after the potential for effective life prolongation and increased health. J Gerontol Biol Sci 51A, B392–B395. R. Arking et al. / Experimental Gerontology 35 (2000) 167–185 185 Keppel, G. & Zedek, S. (1985). Data analysis for research designs: analysis of variance and multiple regression/correlation approaches. New York: W. H. Freeman and Company. Lindsley, D. L. & Zimm, G. G. (1992).The Genome of Drosophila Melanogaster. San Diego: Academic Press. Levine, R. L., Williams, J. A., Stadtman, E. R., & Schacter, E. (1994). Methods in Enzymology 233, 346 –357. Luckinbill, L. S., Arking, R., Clare, M. J., Cirocco. W. C., & Buck, S. (1984). Selection for delayed senescence in Drosophila melanogaster. Evolution 38, 996-1004. Luckinbill, L. S., Riha, V., Rhine, S., & Grudzien, T. A. (1990). The role of glucose-6-phosphate in the evolution of longevity in Drosophila melanogaster. Heredity 65, 29 –38. Martin, G. M., Austad, S. N., & Johnson, T. E. (1996). Genetic analysis of ageing: role of oxidative damage and environmental stresses. Nature Genet 13, 25–34. Misra, H. P. & Fridovitch, I. (1972). The role of superoxide anion in the autoxidation of epinephrine and a simple assay for superoxide dismutase. J Biol Chem 256, 8506 – 8509. Nudzhin, S. V., Pasyukova, E. G., Dilda, C. L., Zeng, Z.-B., & Mackay, T. F. C. (1997). Sex specific quantitative trait loci affecting longevity in Drosophila melanogaster. Proc Natl Acad Sci USA 94, 9734 –9739. Orr, W. C. & Sohal, R. S. (1994). Extension of life-span by over expression of superoxide dismutase and catalase in Drosophila melanogaster. Science 263, 1128 –1130. Parkes, T. L., Elia, A. J., Dickinson, D., Hilliker, A. J., Phillips, J. P., & Boulianne, G. L. (1998). Extension of Drosophila lifespan by over expression of human SOD1 in motorneurons. Nature Genet 19, 171–174. Parsons, P. A. (1995). Inherited stress resistance and longevity: a stress theory of ageing. Heredity 75, 216 –221. Pierce, J. & Suelter, C. H. (1977). An evaluation of the Coomassie brilliant blue G-250 dye binding method for quantitative protein determination. Anal Biochem 81, 478 – 480. Resler, A. S., Kelly, K., Kantor, G., Khazaeli, A. A., Tatar, M., & Curtsinger, J. W. (1998). Genetic analysis of extended life spans in Drosophila melanogaster. II. Replication of the backcross test and molecular characterization of the N14 locus. Genetica 104, 33–39. Riha, V. F. & Luckinbill, L. S. (1996). Selection for longevity favors stringent metabolic control in Drosophila melanogaster. J Gerontol: Biol Sci 51, B284 –294. Sohal, R. J., Agarwal, S., Dubey, A., & Orr, W. C. (1993). Protein oxidative damage is associated with life expectancy of houseflies. Proc Natl Acad Sci USA 90, 7255–7259. Sohal, R. J. & Weindruch, R. (1996). Oxidative stress, caloric restriction, and aging. Science 273, 59 – 63. Stadtman, E. (1992). Protein oxidation and aging. Science 257, 1220 –1224. Sun, J. & Tower, J. (1999). FLP recombinase-mediated induction of Cu/ZnSOD transgene expression can extend the life span of adult Drosophila. Mol Cell Biol 19, 216 –228. Taub J, Lau, J. F., Ma, C., Hahn., J. H., Hoque, R., Rothblatt, J., & Chalfie, M. (1999). A cytosolic catalase is needed to extend adult lifespan in C. elegans daf-C and clk-1 mutants. Nature 399, 162–166. Thompson, J. N., Jr. (1975). Quantitative variation and gene number. Nature 258, 665– 668. van Voorhies, W. A. & Ward, S. (1999). Genetic and environmental conditions that increase longevity in Caenorhabditis elegans decrease metabolic rate. Proc Nat Acad Sci USA 96, 11399 –11403. Vettraino, J. & Arking, R. Direct selection for paraquat resistance in Drosophila yields reciprocal effects on the P450 system and the antioxidant defense system. In preparation. Zar, J. H. (1974). Biostatistical Analysis (2nd ed). Englewood Cliffs, NJ: Prentice–Hall.