Testing the Oxidative Stress Hypothesis of Aging in Primate Fibroblasts: Is There a Correlation Between Species Longevity and Cellular ROS Production?

Abstract

The present study was conducted to test predictions of the oxidative stress theory of aging assessing reactive oxygen species production and oxidative stress resistance in cultured fibroblasts from 13 primate species ranging in body size from 0.25 to 120 kg and in longevity from 20 to 90 years. We assessed both basal and stress-induced reactive oxygen species production in fibroblasts from five great apes (human, chimpanzee, bonobo, gorilla, and orangutan), four Old World monkeys (baboon, rhesus and crested black macaques, and patas monkey), three New World monkeys (common marmoset, red-bellied tamarin, and woolly monkey), and one lemur (ring-tailed lemur). Measurements of cellular MitoSox fluorescence, an indicator of mitochondrial superoxide $\left({\text{O}}_2^{·-}\right)$ generation, showed an inverse correlation between longevity and steady state or metabolic stress–induced mitochondrial ${\text{O}}_2^{·-}$ production, but this correlation was lost when the effects of body mass were removed, and the data were analyzed using phylogenetically independent contrasts. Fibroblasts from longer-lived primate species also exhibited superior resistance to H₂O₂-induced apoptotic cell death than cells from shorter-living primates. After correction for body mass and lack of phylogenetic independence, this correlation, although still discernible, fell short of significance by regression analysis. Thus, increased longevity in this sample of primates is not causally associated with low cellular reactive oxygen species generation, but further studies are warranted to test the association between increased cellular resistance to oxidative stressor and primate longevity.

In 1956, Denham Harman proposed (1) that organismal aging results from reactive oxygen species (ROS), generated as byproducts of mitochondrial respiration, such as superoxide $\mathrm{(}{\mathrm{O}}_2^{·\mathrm{-}})$ and hydrogen peroxide (H₂O₂), interacting with cell constituents. On the basis of the oxidative stress hypothesis of aging, it can be predicted that long-lived animals utilize a combination of strategies to limit oxidative stress–induced cellular damage. Within this conceptual framework, a series of subsidiary hypotheses are possible. For instance, one hypothesis is that cells of successfully aging animals exhibit lower steady-state mitochondrial generation of ROS, thus, taking longer to reach the critical threshold beyond which oxidative damage significantly impairs cellular function. Additionally, cells of longer-living animals could be hypothesized to exhibit less mitochondrial ROS production in response to metabolic stressors through enhanced mitochondrial coupling or superior mitochondrial antioxidant defense mechanisms. Finally, it is reasonable to predict that successfully aging species may have increased tolerance for oxidative stress–induced cell death perhaps through superior damage repair mechanisms and increased mitochondrial resistance to calcium overload (2).

Many previous studies lend support to the oxidative stress hypothesis of aging (3,4). For instance, a comparison of long-lived pigeons (maximum life span = 35 years) with shorter-lived rats (maximum life span = 4 years) found that mitochondria isolated from the heart of the long-lived species exhibit lower baseline ROS generation than those isolated form the short-lived species (5). These findings have been confirmed by other groups (3,4) and extended to other long-lived avian species (3). Moreover, a recent study by Lambert and colleagues (4) demonstrated that ROS production by heart mitochondria isolated from diverse mammalian species including muroid rodents, two bat species, naked mole rat, Damara mole rat, guinea pig, baboon, and ox tends to inversely correlate with species life span.

Although the oxidative stress hypothesis of aging continues to be among the most commonly adduced mechanistic hypotheses to explain variation in aging rate, it is also a subject of ongoing debate due to recent findings inconsistent with it in genetically manipulated laboratory mice (6–11). However, these finding should be interpreted with caution for several reasons. First, most laboratory mice used in biomedical research have undergone a century of laboratory evolution and inbreeding and thus have altered endocrine regulation, compromised mitochondrial function, metabolic defects, and impaired damage repair pathways compared with their wild progenitors (12,13). Second, laboratory mice are purposely protected from many of the vicissitudes of life, such as infectious diseases, suboptimal diets, and climatic variation. Experimental results obtained under these benign conditions may differ from those obtained under more realistic and stressful conditions. A more compelling evaluation of the oxidative stress hypothesis of aging might employ a range of species not subjected to inbreeding and laboratory selection. One approach might be a comparative assessment of cellular ROS homeostasis among wild-caught or wild-derived animals with known and reasonably disparate longevities. Such an analysis would provide valuable additional information to support or counter the oxidative stress hypothesis of aging (14–16).

It is generally accepted that exceptional longevity evolved independently many times in various mammalian orders, but it is not obvious that mechanisms of aging will be conserved among these various groups (17). Primates are among the longest lived mammals, living on average more than twice as long as a standard mammal for their body size (17,18). The present study was designed to assess whether cellular ROS production and resistance to stress-induced apoptosis might be causally involved in primate longevity. We tested this hypothesis by determining whether there were consistent patterns between ROS production, stress resistance, and longevity using primary fibroblast cultures from 13 species of phylogenetically diverse primates. To our knowledge, this is the first study comparing cellular ROS production and oxidative stress resistance in cells of a broad array of primate species.

METHODS

Species Used and Cell Culture Techniques

Primary skin fibroblast cell lines from 13 primate species (Homo sapiens [human], Pan troglodytes [chimpanzee], Pan paniscus [bonobo], Gorilla gorilla [gorilla], Pongo pygmaeus [orangutan], Macaca mulatta [rhesus macaque], Macaca nigra [black crested macaque], Papio hamadryas [baboon], Erythrocebus patas [patas monkey], Callithrix jacchus [common marmoset], Saguinus labiatus [red-bellied tamarin], Lagothrix lagotricha [woolly monkey], and Lemur catta [ring-tailed lemur]) were purchased from the Aging Cell Depository of the National Institute on Aging maintained at the Coriell Institute for Medical Research (Camden, NJ). These species span the primate phylogeny from lemurs to the great apes, range in size from 0.25 to more than 120 kg and in longevity from 20 to 90 years (Table 1). Species longevity records of animals in captivity were taken from the AnAge database (http://genomics.senescence.info/species/) compiled by de Magalhaes and colleagues (19), which is based on literature surveys covering over 100 zoos worldwide. Body mass for each species is the mean of male and female body masses reported by Rowe (20). As most primate species longevity records derive from only a few hundred to a few thousand complete and accurate individual life spans, we estimated that a similar size human sample, subject to similar living conditions, would be 90 years. None of the conclusions reached in this paper would be altered if the human data were removed. The biological age of most of the donors was less than 35% of their maximum recorded life span. The exception was L lagotricha (53% of maximal life span). Cells were cultured in Dulbecco's Modified Eagle Medium (DMEM) media supplemented with 10% fetal bovine serum plus penicillin/streptomycin/Fungizone (at 5% CO₂, at 37°C) (21). One to four cell strains per species were used after Passages 7–9. All assays were run in replicates of six.

Table 1.

Body Mass and Maximum Reported Life Span of the Species Used in This Study

Species	Common Name	Life Span (y)	Body Mass (kg)
Homo sapiens	Human	90.0	61.50
Pan troglodytes	Chimpanzee	59.4	44.80
Pan panuscus	Bonobo	55.0	35.00
Gorilla gorilla	Gorilla	53.5	120.50
Pongo pygmaeus	Orangutan	59.0	57.30
Macaca mulatta	Rhesus macaque	40.0	7.95
Macaca nigra	Crested black macaque	34.0	7.90
Papio hamadryas	Baboon	37.5	21.60
Erythrocebus patas	Patas monkey, Wadi monkey, Hussar monkey	28.3	5.50
Lagothrix lagotricha	Woolly monkey	32.0	5.85
Callithrix jacchus	Common marmoset	22.8	0.25
Saguinus labiatus	Red-bellied tamarin	20.5	0.46
Lemur catta	Ring-tailed lemur	37.3	2.69

Note : Life span data are from the AnAge database (http://genomics.senescence.info/species/), and body mass is mean of male and female masses reported in Rowe (20).

Measurement of Mitochondrial ${\mathrm{O}}_2^{·\mathrm{-}}$ Production and Cellular ROS Levels

Steady-state mitochondrial ${\mathrm{O}}_2^{·\mathrm{-}}$ (superoxide) production in cultured fibroblasts was assessed by flow cytometry (Guava Easycyte) using MitoSOX Red (Invitrogen, Carlsbad, CA), a mitochondrion-specific hydroethidine-derivative fluorescent dye, as previously reported (16,22–24). Cell debris (low forward and side scatter) and dead cells (Sytox Green positive) were gated out prior to analysis. The data are presented as mean intensity of MitoSOX fluorescence, normalized to the respective mean fluorescence intensities obtained in human fibroblasts. Cells were also grown on glass-bottom dishes, and localization of MitoSox fluorescence to mitochondria was confirmed by fluorescent microscopy. MitoTracker Green (Invitrogen) and Hoechst 33258 were used for localizing mitochondria and nuclear counterstaining, respectively.

To assess cellular peroxide production (derived from mitochondria, cytoplasmic sources, and cell membrane–associated oxidases), we used the cell-permeant oxidative fluorescent indicator dye C-H₂DCFDA (5 (and 6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate-acetyl ester; Invitrogen) as we previously reported (23,25,26). In brief, cells were washed with warm phosphate-buffered saline and incubated with C-H₂DCFDA (10 μM, at 37°C). For quantitative measurements, the accumulation of C-H₂DCFDA fluorescence over time (30 minutes) was compared by flow cytometry (Guava Easycyte). Fluorescent micrographs of C-H₂DCFDA–stained cells were taken to demonstrate cytoplasmic localization of C-H₂DCFDA fluorescence.

To assess mitochondrial $\left[{\mathrm{O}}_2^{·\mathrm{-}}\right]$ production under conditions of metabolic stress, fibroblasts were treated with high glucose (30 mM for 24 hours) and mitochondrial $\left[{\mathrm{O}}_2^{·\mathrm{-}}\right]$ production was assessed using MitoSOX Red by flow cytometry (Accuri C6 flow cytometer; Accuri, Ann Arbor, MI). The flow cytometer was equipped with a HyperCyt Autosampler (IntelliCyt Corporation, Albuquerque, NM), which allowed for ultra-fast automated sampling and smaller sample requirements.

Cellular H₂O₂ production was measured fluorometrically using the Amplex Red/horseradish peroxidase assay as described (21,27). The rate of H₂O₂ generation was assessed by measuring resorufin fluorescence for 60 minutes by a Tecan Infinite M200 plate reader. Hoechst 33258 fluorescence, representing cellular DNA content, was used for normalization.

Assessment of Cellular Resistance to Stressors

Cells were grown to 70% confluence in 96-well plates and treated with the following stressors: high glucose (30 mM), H₂O₂ (300 μmol/L), paraquat (50 μmol/L), tunicamycin (1 μg/mL), bacterial lipopolysaccharide (LPS, 10 μg/mL), and recombinant tumor necrosis factors α (TNF-α, 10 ng/mL) for 10–24 hours. The aforementioned stressors are known to cause apoptosis through increasing oxidative stress, mitochondrial damage, ATP depletion, and/or by damaging DNA, lipids, and proteins and are useful tools to assess cellular oxidative stress resistance (28–31). High glucose induces mitochondrial oxidative stress (16,32,33) and H₂O₂, the major form of ROS released from the mitochondria in aged cells, increases the formation of highly reactive hydroxyl radicals. The herbicide paraquat increases redox cycling and formation of intracellular $\left[{\mathrm{O}}_2^{·\mathrm{-}}\right]$. Tunicamycin causes endoplasmic reticulum stress by abolishing N-linked glycosylation of proteins and thus interferes with the assembly and transport of glycoproteins from the endoplasmic reticulum to the Golgi complex. LPS and the cytokine TNF-α are inflammatory stressors and cause oxidative stress and apoptosis by activating Toll-like receptors and TNF-α-receptors, respectively (28,29).

All incubations were at 37°C in a humidified incubator with 5% CO₂ in air. After the treatment period, the ratio of annexin V–positive cells, a marker of apoptosis, was determined by flow cytometry (Guava Easycyte) using the Guava Nexin Assay (Millipore) according to the manufacturer’s guidelines. As an independent marker of cell death, the ratio of dead cells as a percentage of total cell number was assessed using the Guava ViaCount Assay (Millipore) according to the manufacturer’s protocol. The ViaCount Assay distinguishes viable and nonviable cells based on differential permeabilities of two proprietary DNA-binding dyes, including a nuclear dye, which stains only nucleated cells, and a viability dye, which stains dying cells.

Data Analysis

Data were normalized to the respective control mean values and expressed as means ± S D. Statistical analyses of data were performed by regression analysis or by two-way analysis of variance followed by the Tukey post hoc test, as appropriate.

RESULTS

Inverse Relationship Between Mitochondrial ROS Production and Species Longevity

In primate fibroblasts cell lines, steady-state mitochondrial ${\mathrm{O}}_2^{·\mathrm{-}}$ production was assessed by measuring cellular MitoSox fluorescence intensity by flow cytometry. We found that there was an inverse relationship (p = .04, r ² = .32) between steady-state mitochondrial ROS production and maximal species longevity (Figure 1A). Representative fluorescent images of MitoSox-loaded cells are shown in Figure 1B. For each cell line, perinuclear MitoSox fluorescence was co-localized with Mitotracker green fluorescence. Merged images revealed that there was a greater MitoSox fluorescence per unit mitochondria in fibroblasts of shorter-lived species than in those of longer-living species.

Figure 1.

(A)Inverse relationship between maximum life span and mitochondrial ${\mathrm{O}}_2^{·\mathrm{-}}$ production in primate fibroblasts. Cellular MitoSox fluorescence intensities were assessed using flow cytometry, as described in the Methods. Data are means ± S E M. The regression is significant (p = .04, r ² = .32). Mus musculus data (red symbol) are shown for reference but not included in the statistical analysis. (B) Representative fluorescent images showing stronger perinuclear MitoSox staining (red fluorescence) in primary Macaca nigra fibroblasts (bottom) than in Homo sapiens fibroblasts (top). Hoechst 33258 (blue fluorescence) and Mitotracker (green fluorescence) were used for nuclear staining and visualizing mitochondria, respectively. Overlaying the fluorescent images show that MitoSox fluorescence localizes to the mitochondria (original magnification: 20×). (C) Representative fluorescent images showing stronger C-H₂DCFDA staining (representing cellular peroxide production, see Methods) in primary M nigra fibroblasts (right) than in H sapiens fibroblasts (left). Hoechst 33258 (blue fluorescence) was used for nuclear staining. (D) Positive correlation (p = .04, r = .89) between cellular MitoSox fluorescence (representing mitochondrial reactive oxygen species production) and C-H₂DCFDA staining (representing cellular peroxide levels) in primate fibroblasts. Data are means ± S E M (n = 4–6 for each data point).

Because mitochondria-derived ROS (mostly H₂O₂) are membrane permeable, we also assessed cellular ROS production by C-H₂DCFDA fluorescence. Representative fluorescent images of C-H₂DCFDA–loaded cells are shown in Figure 1C. For each cell line, C-H₂DCFDA fluorescence was localized to the cytoplasm. Analysis of cellular C-H₂DCFDA fluorescence intensities by flow cytometry revealed that there is an inverse relationship (p = .01, r ² = .72) between cellular ROS production and maximal species longevity. Cellular ROS levels also show a positive correlation (p = .04, r = .89) with steady-state mitochondrial ROS production (Figure 1D).

Body mass correlates with many cellular and physiological parameters, including longevity (34), and it is often informative in assessing relationships among variables that persist after the effects of body size are statistically removed (35). Thus, we analyzed the relationship between life span and mitochondrial ROS production after removing the effects of body mass on the two traits by residual analysis. The logarithmic relationship between maximum life span and body mass was constructed from the life span and body mass data for all primate species from the AnAge database (19), including the data in Table 1, and a line was fit to the data by least squares regression (Figure 2A). Residuals were then calculated from the regression equation. This procedure was repeated for body mass versus relative mitochondrial ROS production. We did not find a significant correlation between these two parameters (Figure 2B). The negative correlation between mitochondrial ${\mathrm{O}}_2^{·\mathrm{-}}$ production and the residuals for maximal life span, which are now independent of body mass, did not reach statistical significance (Figure 2C, p = .10, r ² = .21). Thus, body size represents a significant confounder of the relationship between species longevity and ROS production.

Figure 2.

Analysis of the effects of body mass and shared evolutionary history on the relationship between maximum life span and mitochondrial ${\mathrm{O}}_2^{·\mathrm{-}}$ production in primate fibroblasts. (A) Ln body mass vs Ln maximum life span (crosshairs represent data for all primate species from the AnAge database for comparison; p = .003, r² = .78). (B) Ln body mass vs Ln MitoSox fluorescence (p = .13, r ² = .2). (C) Lack of significant correlation between residual Ln maximum life span and MitoSox fluorescence (p = .1, r ² = .21). (D) Phylogenetic tree (with 13 species resulting in 12 contrasts at nodes numbered as 1–12) used in the phylogenetically independent contrast (PIC) analysis for mitochondrial reactive oxygen species production. (E) PIC of residual ln maximum life span vs PIC of MitoSox fluorescence with the x-axis “positivized” (36).

Another potential confounder of comparative analyses is nonindependence among species values due to shared evolutionary ancestry. To account for this fact, we generated phylogenetically independent contrasts (PIC, from the residuals of maximum life spans and the MitoSox fluorescence data) (4) using the methods of Garland and colleagues (36). A phylogenetic tree was derived from molecular sequence data from various sources (37–41) and is shown in Figure 2D. The tree has 13 “tips”, corresponding to the 13 primate species, and 12 “nodes”. The tip values for life span were subtracted, giving a contrast for this trait. The same procedure was applied to the residuals of each oxidative stress parameter as well. The process was repeated for the internal nodes, which are estimates of the ancestral state. These values are the calculated averages of the daughter nodal values, weighted according to the daughter branch lengths (representing evolutionary time). Figure 2E shows the relationship between mitochondrial ROS production and maximum life span for the PIC in our data set (the nodes in Figure 2D). A linear regression through the points in Figure 2E did not give a significant correlation (p = .3, r ² = .07). Thus, the raw data appear to be confounded by phylogenetic influences.

To assess mitochondrial ROS production under conditions of metabolic stress, we exposed cultured fibroblasts to high glucose. We found that an inverse correlation exists between high glucose–induced mitochondrial oxidative stress and maximal species life span (Figure 3A). We found that the magnitude of high glucose–induced mitochondrial oxidative stress showed a positive correlation with the steady-state mitochondrial ROS generation (p = .049, r ² = .37). Using the Amplex Red/horseradish peroxidase assay, we found that high glucose–induced cellular H₂O₂ production (assessed by measuring resorufin fluorescence) also showed a significant inverse correlation (p = .004, r ² = .60) with maximal species life span (data not shown), similar to the results of the mitochondrial ROS measurements.

Figure 3.

(A)Inverse relationship (p = .03, r ² = .40) between maximum life span and mitochondrial ${\mathrm{O}}_2^{·\mathrm{-}}$ production in primate fibroblasts stressed by high glucose (30 mmol/L, for 24 hours). Cellular MitoSox fluorescence intensities were assessed using flow cytometry, as described in the Methods section. Data are means ± S E M (n = 4–6 for each data point). (B) Ln (body mass) vs Ln (MitoSox fluorescence) in high glucose–treated cells (p = .024, r ² = .38). (C) Rresidual Ln (maximum life span) vs residual Ln (MitoSox fluorescence; p = .15, r ² = .17). (D) Phylogenetically independent contrast (PIC) of residual Ln maximum life span vs PIC of residual Ln (MitoSox fluorescence) with the x-axis “positivized” (36). The relationship is not significant by regression analysis.

Because the logarithmic relationship between body mass and high glucose–induced mitochondrial oxidative stress was significant (p = .024, r²= .38, Figure 3B), we also analyzed the relationship between life span and high glucose–induced mitochondrial ROS production after removing the effects of body mass on the two traits by residual analysis (Figure 3C). The correlation between the two sets of residuals, which are now independent of body mass, did not reach statistical significance (Figure 3C, p = .15, r ² = .17). Thus, body size represents a significant confounder of the relationship between species longevity and mitochondrial ROS production both at baseline and under metabolic stress conditions. To remove the effect of shared evolutionary history, we generated PIC from the residuals in Figures 2B and 3B. Figure 3D shows the relationship between high glucose–induced mitochondrial ROS production and maximum life span for the PIC in our data set. A linear regression through the points in Figure 3D did not give a significant correlation, suggesting that the raw data are confounded by phylogenetic influences.

Association Between Oxidative Stress Resistance and Species Longevity

To assess the relationship between cellular stress resistance and species longevity, cultured fibroblasts were treated with a range of stressors—high glucose, H₂O₂, paraquat, tunicamycin, LPS, and TNF-α. The level of annexin V–positive cells, indicating incipient apoptosis, was low in untreated samples.

Treatment with H₂O₂ significantly increased the number of apoptotic fibroblasts derived from shorter-living species, whereas the number of apoptotic cells remained low in H₂O₂-treated cells from longer-living primates (Figure 4A). The inverse correlation between the magnitude of H₂O₂-induced (Figure 4A, p = .015, r ² = .74) apoptosis and species longevity was significant. When the ratio of viable and apoptotic/dead cells was compared using the ViaCount assay, we also detected an inverse correlation between species longevity and decreases in cell viability induced by H₂O₂ (p = .041, r ² = .32; data not shown). There was a significant correlation between Ln (body mass) and Ln (number of annexin V–positive cells) in the H₂O₂-treated cohort (Figure 4B, p = .035, r ² = .34). Analysis of body size residuals showed a highly significant inverse correlation between the magnitude of H₂O₂-induced apoptosis (p = .005, r ² = .52) and maximum life span (Figure 4C). Analysis of PIC also suggested that the magnitude of H₂O₂-induced apoptosis and maximum life span inversely correlate, although the association reached only marginal statistical significance (p = .059, r ² = .34, Figure 4D).

Figure 4.

(A)Inverse relation between maximum life span and H₂O₂-induced (300 μmol/L, 24 hours) apoptotic cell death. The ratio of annexin V–positive cells was assessed by flow cytometry. The correlation is significant (p = .015, r ² = .74). Data are means ± S E M (n = 4–6 for each data point). (B) Ln (body mass) vs Ln (number of apoptotic H₂O₂-treated cells) p = .035, r ² = .34. (C) Analysis of body size residuals shows a significant inverse correlation between the magnitude of H₂O₂-induced apoptosis and maximum life span (p = .005, r ² = 0.52). (D) Phylogenetically independent contrast (PIC) analysis of residual ln maximum life span vs PIC of residual ln rate of H₂O₂-induced apoptosis. The numbers correspond to the nodes of the phylogenetic tree presented in Figure 2. Eleven of the 12 points show that sensitivity to H₂O₂-induced cell death associates negatively with maximum life span, the dashed line at y = 0 is to emphasize this fact. The presence of the outlier at node six renders the relationship nonsignificant (p = .059) by regression analysis. (E) Inverse relation between maximum life span and high glucose–induced (30 mmol/L, 24 hours) apoptotic cell death. The correlation is significant (p = .041, r ² = .34). Data are means ± S E M (n = 4–6 for each data point). (F) Ln (body mass) vs Ln (number of apoptotic high glucose–treated cells) p = .045, r ² = .31. (G) Residual Ln (maximum life span) vs residual Ln (number of apoptotic high glucose–treated cells; p = .15, r ² = .17). (H) PIC of residual ln maximum life span vs PIC of residual Ln (number of apoptotic high glucose–treated cells). The relationship is not significant by regression analysis.

Treatment with high glucose also significantly increased the number of apoptotic fibroblasts derived from shorter-living species, whereas the number of annexin V–positive cells remained low in high glucose–treated cells from longer-living primates (Figure 4E). The inverse correlation between the magnitude of high glucose–induced apoptosis (Figure 4E, p = .041, r ² = .34) and species longevity was significant. When the ratio of viable and apoptotic/dead cells was compared using the ViaCount assay, we also detected an inverse correlation between species longevity and decreases in cell viability induced by high glucose (p = .045, r ² = .34, data not shown). There was a significant correlation between Ln (body mass) and Ln (number of annexin V–positive cells) in the high glucose–treated cohorts (Figure 4F, p = .045, r ² = .31). Analysis of body size residuals (Figure 4G) and analysis of PIC (Figure 4H) showed no significant correlation between the magnitude of high glucose–induced apoptosis and maximum life span.

Treatment with paraquat, tunicamycin, LPS, and TNF-α also resulted in an increase in the level of annexin V–positive cells (Figure 5A–D, respectively). However, there was no significant correlation between species life span and resistance to apoptosis induced by these stimuli. Results from the ViaCount assay yielded identical conclusions, showing no significant correlation between species life span and resistance to apoptosis induced by paraquat (p = .68, r ² = .09), tunicamycin (p = .15, r ² = .17), LPS (p = .70, r ² = .02), and TNF-α (p = .9, r ² = .0013).

Figure 5.

No significant correlation between maximum life span and apoptotic cell death induced by (A) paraquat (p = .25, r ² = .11), (B) tunicamycin (p = .20, r ² = .10), (C) lipopolysaccharide (LPS, p = .21, r ² = .13), or (D) recombinant tumor necrosis factors α (TNF-α, p = .81, r ² = .0004). The ratio of annexin V–positive cells was assessed by flow cytometry. Data are means ± S E M (n = 4–6 for each data point).

DISCUSSION

On the basis of the oxidative stress hypothesis of aging, it is predicted that if mitochondrial ROS production is a determinant in the rate of aging, then mitochondria of long-lived animals should produce less ROS. Our data gave initial support to this hypothesis showing that steady-state mitochondrial ROS production in primate skin fibroblasts inversely correlates with species longevity (Figure 1A and B). As predicted, we also found that increased mitochondria-derived ROS production results in higher peroxide levels in the cytoplasm of cells from shorter-living primates (Figure 1C and D). These findings in primate fibroblasts are consistent with the recent findings of Lambert and colleagues (4) and others (21,42–45), indicating that in a range of vertebrate homeotherms, there seems to exist an inverse correlation between free radical production by isolated mitochondria and maximum life span. These observations also accord with our recent data obtained in fibroblasts from short-lived and long-lived muroid rodents (2). Similar conclusions were reached by studies showing that oxidative damage to mitochondrial DNA is inversely related to maximum life span in the heart and brain of mammals (46,47). At present, the mechanisms underlying interspecies differences in mitochondrial ROS production are not well understood (4,48–50). The mechanisms may include differences in the efficiency of the electron transport chain, uncoupling proteins, mitochondrial membrane composition, mitochondrial thiol redox state, amounts of coenzyme Q associated with mitochondrial membrane proteins, and differential regulation of the entry of electrons into the cytochrome chain. Furthermore, the efficiency of mitochondrial antioxidant systems may differ between longer-living and shorter-living species. This view is supported by the findings that a positive correlation exists between superoxide dismutase activity-specific activity and maximum life span in primates (51).

Since the original work of Sacher (52), a large number of studies have been published confirming that, on average, larger mammals live longer than smaller animals (19,53,54). There is also a highly significant correlation between body mass and maximum species life span throughout the entire primate order (Figure 2A). The mechanisms underlying this general trend for an association between larger body size and longevity are a subject of ongoing debate (19,55–57), but the current view is that interspecies differences in basal metabolic rate do not explain the variation in maximum longevity (19). A plausible explanation for the allometry of life span in mammals invokes evolutionary ecological factors: small-bodied animals tend to have higher extrinsic mortality rates due to predation (58). In contrast, greater survivorship in large-bodied animals leads to longer life spans and, according to evolutionary theory, the evolution of cellular mechanisms that regulate the process of aging and determine longevity (59). Because the covariation of physiological traits and life span with body mass may be a problem in comparative studies, earlier studies (35) advocate the removal of the confounding effects of body mass. We found that after correction for body mass, the inverse correlation between maximum life span and mitochondrial ROS production was significantly weakened (Figure 2C). Thus, body size appears to have a major influence on mitochondrial ROS production in primates.

It is also clear that the data in Figures 2A and 3A are clustered by their order of origin. Although there seems to be no progressive evolution of increased longevity among the primate superfamilies, the greater apes are substantially longer-lived than the rest of the primates (18) and their cells tend to produce less ROS than cells from other monkeys. When we used PIC analysis after residual analysis to remove the effects of body mass to account for the phylogenetic dependence of the comparative data, the negative correlation between life span and mitochondrial ROS production was lost (Figure 2D). Interestingly, a similar negative conclusion was reached when the data of Lambert and colleagues (4) on mitochondrial ROS production in a wide range of vertebrate species were analyzed using PIC. Taken together, it appears that body size and phylogeny have major impact on mitochondrial redox homeostasis and thus our data do not support a direct cause-and-effect relationship between low mitochondrial ROS production and increased longevity in primates.

We also investigated whether cells of longer-living primates were more resistant to a range of stressors. Initial analysis of the data showed that there was indeed significant association between longevity and stress resistance but only for oxidative stressors (glucose, hydrogen peroxide). We used high glucose as a metabolic stressor because hyperglycemia plays a central role in accelerated aging in diabetes mellitus, as shown by prevention or retardation of the development of complications of diabetes by strict metabolic control (60). The harmful cellular effects of hyperglycemia have been attributed to, at least in part, an increased oxidative stress (61). Enhanced glucose flux through glycolysis results in increased generation of pyruvate, which shuttles into the mitochondria for subsequent oxidation through the tricarboxylic acid cycle generating NADH (62,63). This in turn causes accelerated electron flow through the respiratory chain, which increases mitochondrial superoxide production. Previously, we documented an inverse relationship between metabolic stress–induced mitochondrial ROS production and longevity in small muroid rodents (16). In the present study, we found that mitochondrial ROS generation induced by metabolic stress is blunted in cells of longer-living primates as compared with those of shorter-living ones (Figure 3). Yet, after corrections for body mass and lack of phylogenetic independence, the correlation was removed (Figure 3B–D). It is possible that cells that were derived from primates born in the relatively stress-free environment of a zoo may exhibit a stress response phenotype that is different from that of cells derived from wild-caught animals (64). However, in the present study, we could not control for this variable.

Accumulating empirical data suggest that resistance to the aging process at the organismal level is often reflected in resistance to oxidative stressors at the cellular level (14,65). A causal link between cellular stress resistance and longevity is also suggested by findings that cellular stress resistance phenotypes can be modulated by both life span–extending dietary interventions and treatment with caloric restriction mimetics (66,67). In the present study using fibroblasts as a model system, we found that there was an association between longevity and resistance to metabolic stress- and H₂O₂-induced cell death in primates (Figure 4). Importantly, after correction for body mass and the lack of independence in the data, the correlation between life span and resistance to H₂O₂-induced apoptosis (but not to metabolic stress-induced apoptosis) was still discernible (Figure 4). Our findings extend the original observations of Kirkwood and coworkers showing a relative resistance to oxidative stress–induced cytotoxicity in human fibroblasts as compared with cells of shorter-living species, including marmosets (65). These results are also in line with the previous findings showing that cells of longer-living Peromyscus leucopus are significantly more resistant to the apoptotic effects of high glucose and other oxidative stressors than cells of shorter-living Mus musculus (21). Similarly, increased cellular resistance to oxidative stress was also found in blood vessels (68) and fibroblasts (69) from long-lived bats. Moreover, a recent interspecies comparative study demonstrated that longevity is associated with resistance to H₂O₂-induced death in fibroblast cell lines from nine species of rodents (69). Previous studies suggest that the transcription factor Nrf2 is one of the main regulators of intracellular redox balance and a major determinant of fibroblast oxidative stress resistance (70,71), but other cellular pathways (72,73), including factors regulating cellular senescence (74), may also contribute to interspecies differences in cellular oxidative stress resistance. In our current study, there was no association between life span and resistance to the pro-apoptotic effects of paraquat, tunicamycin, or the inflammatory stressors, LPS and TNF-α (Figure 5). These findings extend those of Harper and colleagues (69) showing that cells of long-lived rodent species also exhibit differential resistance to oxidative stressors and other cytotoxic agents. The mechanisms underlying the quantitative differences in cellular oxidative stress resistance between short-lived and long-lived primates are likely multifaceted and need to be elucidated in subsequent studies. We propose that the role of superior free radical detoxification systems and more efficient repair of free radical–mediated macromolecular damage should be simultaneously considered in future studies.

Conclusions

To our knowledge, this is the first study comparing cellular ROS production and oxidative stress resistance in cells of shorter-living and longer-living primates. We found that in cultured primate fibroblasts, the apparent inverse correlation between species longevity and mitochondrial production of ROS in this cohort of primates may be primarily due to the impact of body mass and phylogeny. In contrast, there seems to be an association between longevity and resistance to oxidative stress–induced cell death, which is in accordance with the predictions based on the oxidative stress hypothesis of aging and provides justification for evaluation of pathways that are master regulators of cellular resistance to oxidative stressors (such as the Nrf2/ARE pathway (24)) in primate species.

Funding

This work was supported by grants from the American Diabetes Association (to Z.U.), American Federation for Aging Research (to A.C.), the American Heart Association (to A.C.), the Oklahoma Center for the Advancement of Science and Technology (to A.C. and Z.U.), the University of Oklahoma College of Medicine Alumni Association (to A.C.), the San Antonio Area Foundation (to A.P.), and the National Institutes of Health (AG031085 to A.C., AT006526 to Z.U., AG022873, and AG025063 to S.N.A.).