Understanding the physiological determinants of lifespan in animals has been the focus of biologists and gerontologists over the past decades. Comparative studies suggest a role for mitochondrial structure and function associated with the aging process, in particular the management of reactive oxygen species (ROS). This evidence has been integrated into one unifying hypothesis, the “mitochondrial oxidative stress theory of aging” (MOSTA, reviewed in Blier et al., 2017). It points out, among others, the importance of cell membrane composition and particularly the susceptibility of mitochondrial membrane lipids to oxidation peroxidation index (PI) as potentially causative parameters of cellular aging. In addition, modifications of the redox state and stoichiometry of mitochondrial electron transporters appear to play a crucial role for mitochondrial coupling and the rates of ROS formation (Barja, 2007), and hence are important components of the MOSTA.

Membrane lipids are particularly prone to oxidative damage (Gamliel et al., 2008), and the sensitivity of fatty acids (FA) to oxidation by ROS depends on their chemical conformation, especially with respect to the presence of conjugated double bonds. FA chains with no or only one double bond [saturated fatty acids (SFA) and monounsaturated fatty acids (MUFA)] are more resistant to peroxidation than polyunsaturated FA (PUFA), the sensitivity of which increases exponentially with the number of conjugated double bonds (Holman, 1954). The products of FA oxidation (reactive aldehydes) induce autocatalytic progression of lipid peroxidation that can further damage membrane proteins, PUFA chains and DNA. This leads to impaired protein functions and DNA mutations and underlines the crucial role of membrane lipid composition for cellular integrity, with implications for organismal lifespan (Hulbert, 2010).

The strongest association between a physiological trait and divergences in species lifespan has so far been observed for membrane structure. The early evidence of a strong negative correlation between membrane PI and increasing maximum lifespan was reported in vertebrates, specifically in mammals and birds (Hulbert et al., 2007). Later, Munro and Blier (2012) took advantage of the extreme longevity of Arctica islandica (the ocean quahog) to corroborate the same association in five species of bivalves (Munro and Blier, 2012). Among these species, the ocean quahog A. islandica is the longest-lived non-colonial animal with a record lifespan of 507 years in the North Atlantic (Butler et al., 2013). Munro and Blier (2012) compared gill membrane lipid composition of A. islandica with four shorter-lived bivalve species from the shelf areas off the coasts of eastern Canada belonging to the same subclass (heterochonchia), with lifespans ranging from 28 to 106 years. Membrane PI decreased exponentially with increasing longevity, in part due to a lower DHA (docosahexaenoic acid, 22:6, n-3) content in mitochondrial membranes of longer-lived species. Other molecules protecting bivalve cellular membranes against ROS attack include plasmalogens [dimethylacetals (DMA) in their methylated form] and non-methylene interrupted fatty acids (NMI). Plasmalogens act as a ROS scavenger, whereas NMI reduce susceptibility to peroxidation (Engelmann, 2004; Barnathan, 2009) while maintaining fluidity of the membrane. Although no consistent relationship was observed between NMI abundance and MRL, A. islandica had higher NMI and plasmalogen levels than the shortest-lived species Mya arenaria. The association previously observed in vertebrates therefore, appears to be conserved in a wide range of animal taxa and tighter, at least in bivalves, for mitochondria.

In addition to membrane structure, Munro et al. (2013) found A. islandica mantle mitochondrial isolates to exhibit lower rates of net ROS production than two shorter-lived bivalve species. These differences related to the activities of mitochondrial electron transport system (ETS) complexes, including major sites of ROS production (complex I and III) that differed between A. islandica and shorter lived species (especially M. arenaria), despite similar oxidative phosphorylation (OXPHOS) capacities. Relatively lower activities of CI (NADH oxidoreductase) and CII (succinate dehydrogenase) relative to downstream complex (cytochrome c oxidase) in A. islandica suggest a lower degree of reduction at the ETS, resulting in lower overall mitochondrial ROS production without compromising phosphorylation rates (Munro et al., 2013; Blier et al., 2017). This is reminiscent of the lower CI and CII content in calorie-restricted rats and in long-lived drosophila (Ayala et al., 2007; Neretti et al., 2009). Hence, stoichiometric arrangements between upstream and downstream ETS complexes might play an important role in lifespan determination.

From these studies, we can suspect that a low membrane PI and the associated high robustness to ROS attack are prerequisites for reaching long lifespan. On an evolutionary perspective, one obvious question arising from these observations is to which extent membrane variability among populations of one species could set the plasticity and potential evolution of longevity, or associated life history characters. At the intraspecific level, most studies linking lipid composition to metabolism or lifespan have focused on genetically manipulated strains (Brzȩk et al., 2007) with relatively small differences in maximum lifespan, or on laboratory bred animal lines (Valencak and Ruf, 2013). Haddad et al. (2007) found a lower PI in longer-lived honeybee queens than shorter-lived workers, but otherwise no studies have looked at the importance of membrane composition inside a species in a natural setting. The bias of comparing controlled laboratory experiments with physiological processes occurring in nature, could potentially blur some of the real age-related changes (Austad, 2018). In order to test to what extent mitochondrial characters associated to lifespan divergences among species govern the rate of mitochondrial aging within a species, studying different natural populations displaying a wide range of contrasting average lifespans is necessary.

Considerable population-specific variations in MRL are seen in A. islandica, both in the western (Kilada et al., 2007) and eastern (Basova et al., 2012) parts of the North Atlantic. One individual of 507 years was found in Icelandic Coast (IC) waters where animals more than 300 years old are regularly encountered (Strahl et al., 2007; Butler et al., 2013). All other populations have much shorter reported lifespan maxima with the shortest MRL of only 40 years reported for the well-studied Baltic Sea population (Basova et al., 2012, and see Table 1). The different A. islandica habitats are characterized by wide thermal and salinity windows, but there is no consistent evidence for specific metabolic rate or antioxidant activities in animals from any population of these regions associated to environmental conditions (Basova et al., 2012). On the other hand, environmental factors such as food and sunlight availability (Moss et al., 2017), as well as dredging effort (Thorarinsdóttir et al., 2009) could help explain differences in lifespan. Begum et al. (2018) used one mitochondrial locus (cytochrome b) to prove that, in spite of their vastly differing MRL, all these populations belong to the same species and hence provide an applicable system to test longevity-associated characters at the intraspecific level.

The main objective of the present paper is to determine if characters associated to divergences in lifespan among species can dictate the pace of aging among populations of one species. We therefore compared membrane lipid composition, PI, and ETS complex activities across the same A. islandica populations from the Atlantic and the brackish coastal seas. We hypothesized that ETS activities and lipid parameters in the populations vary as a function of MRL in the same fashion as found between different bivalve species (Munro and Blier, 2012). We thus predicted to find (1) a decrease in mitochondrial membrane PI with increasing longevity, (2) a higher %PUFA and %DHA in shorter-lived populations, (3) more plasmalogens in longer-lived populations, and (4) a decrease in enzymatic activity for the CI + CIII relative to COX as longevity increases.

Specimens of A. islandica from six populations with contrasting MRLs were collected across a geographic gradient of European coastal seas (KB and WS) and from the North Atlantic (see Table 1 for details). Approximately 500 mg of gills and mantle tissues were dissected, and immediately frozen in liquid nitrogen and kept at −80°C before enzymatic and lipid analyses were undertaken. The mantle was chosen as it is the largest somatic tissue and as a means of comparing with a wealth of other studies, while the gills were chosen on the basis of their important respiratory and environmental sensory role. Mitochondrial isolation was carried out on 250 mg of frozen mantle and gills, as previously described (Munro and Blier, 2012). The resulting mitochondrial pellet was resuspended in 150 μl of buffer, and both mitochondria and cellular debris fractions were conserved at −80°C after nitrogen flush for lipid analysis. The abundance of mitochondrial ETSs in both biological fractions was assessed using the NADH-INT (iodonitrotetrazolium) oxidoreduction assay, targeting complexes I and III. The values for the cellular debris pellets were below 1/20th of those in the mitochondrial pellets, and were therefore considered essentially devoid of mitochondria.

Analysis of the phospholipid composition of the mitochondrial and cellular debris fractions was done following the protocol established by Munro and Blier (2012). Separation of phospholipids was done using solid-phase extraction columns (Bond Elut-NH₂, 500 mg, 3 mL), and trans-methylated overnight at 55°C in a 1% H₂SO₄ solution in methanol. The resulting fatty acid methyl esters (FAME) and DMA were recovered in 100 μl hexanes and analyzed by GC-FID (Agilent Trace Ultra 100, Thermo Fisher Scientific, Waltham, MA, United States) using a column with a high polarity stationary phase (HP-88, 60 m, 0.25 mm × 0.20 μm, Agilent Technologies Canada, Mississauga, ON, Canada). All solvents used were of ultra-pure grade. Conveniently, calibration of the system was previously done by GC-MS using common FA mix standards (SUPELCO 37 FAME) marine FA standards (SUPELCO) and DMA standards (SIGMA). FA composition-longevity relationships were tested using relative abundance in mol% of individual FA, FA classes (total saturated, monounsaturated, polyunsaturated, DMA and NMI), as well as of different indexes as per Hulbert et al. (2007). The PI was calculated as PI = (0.025 × % monoenoics) + (0.258 × % 20:2 NMI) + (0.32 × % 22:2 NMI) + (1 × % dienoics) + (2 × % trienoics) + (4 × % tetraenoics) + (6 × % pentaenoics) + (8 × % hexaenoics). The unsaturation index (UI) represents the number of double bonds per 100 acyl chains and was calculated as UI = (% monoenoics) + (2^*% dienoics) + (3^*% trienoics) + (4^*% tetraenoics) + (5^*% pentaenoics) + (6^*% hexaenoics). Results will be presented as means ± SEM.

Each gill and mantle sample was weighed and homogenized in 6 volumes of cold homogenization buffer (20 mM Tris (hydroxymethyl) aminomethane, 1 mM EDTA, 0.1% Tween 20, pH 7.4) in Precellys Homogenizer 24 (Bertin Technologies) with 2 times 15 s at 5000 rpm at 8°C. Enzymatic activities of CI + III, COX and CS were measured at 8°C using Plate Reader TriStar (Berthold Technologies) and expressed in U.g^–1 tissue fresh mass. All assays were run in duplicate. All protocols were adapted from Breton et al. (2009). We chose to present CI + III, rather than using the previous annotation “ETS,” since this assay does not measure COX activity, which is a major component of the system.

Differences between groups were assessed using Systat v. 13 software. Variables were first tested for normality using the Shapiro-Wilk test. A between-group principal analysis (PCA) was used to assess the variation in enzymatic activities and membrane lipid composition for individuals belonging to the six populations studied. This analysis consists to running PCA on a dataset, where observations are gathered by user-defined groups to emphasize the between-groups variability in the analysis process (Dolédec and Chessel, 1989). Here groups were built by distinguishing specimens accordingly to their sampling localization. The significance of the proportion of the variability explained by the grouping factor (e.g., population) in between-groups PCA was tested by Monte-Carlo permutation (999 permutations) and were considered significant when the simulated p value = 0.001. Between-groups PCA was performed using “ade4” library (Chessel et al., 2004) implemented in R 3.2.1 (R Development Core Team, 2012).

Mitochondrial resistance to peroxidation and ETS activity linked to ROS production are two important tenets of the MOSTA at the interspecific level (Barja, 2013; Blier et al., 2017). Our study aimed at assessing whether these characters were related to longevity at the intraspecific scale, by taking advantage of the extraordinary MRL diversity across well-studied regional populations of the ocean quahog, A. islandica.

The species A. islandica has evolved a long-life phenotype with adjustment of mitochondrial membrane composition, low metabolic activity and control of cellular waste products (Strahl et al., 2007; Strahl and Abele, 2010; Munro et al., 2013; Moss et al., 2016). Our results show that at the intraspecific level, there is no direct relationship between two important mitochondrial components: FA composition of the membrane, enzymatic organization at the level of CI + III, COX and CS, and population-specific MRL. Nonetheless, it appears that contrary to the ratio of upstream to downstream ETS complexes (which is conserved among populations), the enzymatic activities and aerobic capacities (as citrate synthase activity per mg of proteins) are plastic traits among populations. Brackish, coastal environments in KB and WS is associated to an increase of citrate synthase maximal activity relative to ETS capacities suggesting elevated rates of substrate oxidation. This might better support both osmoregulation and anaerobic energy metabolism under fluctuant salinity and temperature conditions in these habitats.

Differences in population longevity appear to be independent of sampling effort (see for example Gruber et al., 2015) and, although no precise record of age at maturity from our sites could be found, data from the western Atlantic suggests that it is variable among populations (Thompson et al., 1980) and likely correlated to longevity as it is in different bivalve species (Ridgway et al., 2011). As our understanding of oxidative stress increases, it appears that mitochondrial ROS management is at the center of modulation of life-history traits (such as reproduction and growth, see Costantini, 2018). We then suggest that the divergences in observed MRL among populations should not only be the consequence of detrimental effects of stressful conditions, but could also result from the adjustment of maturation to insure completion of the life cycle in the different environmental conditions experienced by the populations. It seems that A. islandica populations are able to deal with a large range of environmental conditions at the expense of MRL. We suspect that all A. islandica, independently of the population they belong to, appear to have the potentiality to reach extreme longevity. However, we cannot exclude that reaching such high longevity as 507 years old would require very special dispositions, even for a centenarian bivalve, as particular environmental or ecological conditions (such as predation pressure, see Moss et al., 2017), specific mitochondrial phenotypes and/or peculiar genetic predispositions.

The results of the present study clearly reveal that a characters that have been tightly associated to divergences of lifespan among species namely PI of mitochondrial membranes cannot explain the observed divergences among populations of the longest-lived species A. islandica. It therefore rules out the proposal that variation in PI could alone manage pace of aging or dictate the expressed lifespan of populations. High metabolic demands requested by stressful and viable environments could partly explain these populations’ longevity divergences, but they cannot only result from the impact of stress on physiological conditions, since divergences in lifespan occur with adjustments of the age at reproduction. These bivalves are thus a useful system to disentangle the metabolic characters modulating reproduction and linking maturation to lifespan and provide support or challenge aging theories. Integrating field observation and experimentation with results obtained in the laboratory is the only way to tackle this crucial question (Austad, 2018).

Future studies shall investigate the biochemical adjustments in mitochondria under range edge conditions, including their response to fluctuating salinities and hypoxia in terms of ROS production and the use of alternative electron transport pathways. Other mitochondrial determinants of these lifespan differences have been proposed in recent studies and should be explored. These potential candidates include different phospholipid species such as cardiolipin (reviewed in Paradies et al., 2010), the stability of ETS supercomplexes (Gómez and Hagen, 2012), the control of electron flux and ROS management (Blier et al., 2017; Munro and Treberg, 2017), and the rates of mtDNA mutations (reviewed in Pinto and Moraes, 2015).

All datasets generated for this study are included in the manuscript and/or the Supplementary Files.

ER and CD performed the experiments, analyzed the data, and drafted the manuscript. PB, DA, and TH aided in interpreting the results and writing the manuscript. ER and PB planned the experimentation. ER performed lipid biochemistry, statistical, and enzymatic analyses, wrote the first version of the manuscript, and participated in the corrections. CD participated in the sampling, and performed some of the enzymatic analysis. DA provided the samples. CD, TH, DA, and PB participated in the redaction and correction of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.