• Open Access

The extreme longevity of Arctica islandica is associated with increased peroxidation resistance in mitochondrial membranes


Daniel Munro, Department of Biology, Université du Québec à Rimouski, 300, allée des Ursulines, C. P. 3300, succ. A, Rimouski, QC G5L 3A1, Canada. Tel.: 418 723 1986 #1566; fax: 418 724 1849; e-mail: dmunro70@hotmail.com; pierre_blier@uqar.qc.ca


The deleterious reactive carbonyls released upon oxidation of polyunsaturated fatty acids in biological membranes are believed to foster cellular aging. Comparative studies in mammals and birds have shown that the susceptibility to peroxidation of membrane lipids peroxidation index (PI) is negatively correlated with longevity. Long-living marine molluscs are increasingly studied as longevity models, and the presence of different types of lipids in the membranes of these organisms raises questions on the existence of a PI–longevity relationship. We address this question by comparing the longest living metazoan species, the mud clam Arctica islandica (maximum reported longevity = 507 year) to four other sympatric bivalve molluscs greatly differing in longevity (28, 37, 92, and 106 year). We contrasted the acyl and alkenyl chain composition of phospholipids from the mitochondrial membranes of these species. The analysis was reproduced in parallel for a mix of other cell membranes to investigate whether a different PI–longevity relationship would be found. The mitochondrial membrane PI was found to have an exponential decrease with increasing longevity among species and is significantly lower for A. islandica. The PI of other cell membranes showed a linear decrease with increasing longevity among species and was also significantly lower for A. islandica. These results clearly demonstrate that the PI also decreases with increasing longevity in marine bivalves and that it decreases faster in the mitochondrial membrane than in other membranes in general. Furthermore, the particularly low PI values for A. islandica can partly explain this species’ extreme longevity.


Polyunsaturated fatty acids (PUFA) of phospholipids that compose biological membranes are not only targets that are damaged by reactive oxygen species (ROS) but also proposed to act as amplifiers of oxidative stress. Indeed, ROS are known to trigger the self-propagating autocatalytic process of lipid peroxidation among PUFA. A likely fate of the many peroxidized PUFA involved in such a process is to break down and release reactive carbonyl species (RCS). These can be viewed as secondary ROS that severely aggravate the cellular damage that would solely be accounted for by ROS. Indeed, RCS (e.g., MDA, acrolein, and 4-hydroxy-2-nonenal) can form adducts to proteins (advanced lipoxidation end products) and lipids that impede their functionality. They can also form exocyclic adducts to the G, A, and C bases of DNA (e.g., M1dG) with mutagenic potential.

The susceptibility of individual fatty acids (FA) to peroxidation is known to increase exponentially with increasing number of double bonds on the carbon chain and has been determined empirically for pure solutions of each of them (Holman, 1954). This information allows the calculation of a single average value of susceptibility to peroxidation for any given biological membrane after conducting a lipid analysis; this value is referred to as the peroxidiation index (PI). Comparative studies in mammals and birds have shown the existence of a negative correlation between PI and a species’ longevity in the phospholipids of skeletal muscle and liver mitochondria (Hulbert et al., 2007), and the decrease in the ratio of n-3 to n-6 PUFA with increasing longevity remains valid after correcting for phylogeny nonindependence (Valencak & Ruf, 2007). Diet restriction studies also found that a lower PI value accompanies the increase in longevity (Ayala et al., 2007). Moreover, a recent phylogenomic study across 25 mammal species showed that genes controlling membrane lipid composition are positively selected in association with longevity (Richard et al., 2010). These complementary pieces of evidence are integrated into the membrane-pacemaker theory of aging (Hulbert, 2005), which is a modification of the oxidative stress theory of aging that suggests cellular damages accrued from the release of RCS play an important role in aging (Pamplona & Barja, 2011). However, nearly the entire body of evidence that substantiates this theory has been obtained from endotherms characterized by a simple phospholipid fatty acid composition. No interspecific comparisons have been made for invertebrates even though they represent more than 95% of all animal species and 99% of all animal biomass.

Many bivalve molluscs can attain old ages (often over 100 years old) and are increasingly appreciated as longevity models (Abele et al., 2009; Bodnar, 2009; Philipp & Abele, 2009). This group of invertebrates represents a very good challenge to the membrane-pacemaker theory of aging. In temperate climates, these organisms experience very low and fluctuating body temperature over the year, which places a constraint on maintaining proper membrane fluidity for enzymatic activity. Hence, their membranes may have to incorporate high and constant levels of PUFA across species (Hazel, 1995). Some bivalves also have to cycle through hypoxia and reoxygenation (e.g., intertidal species) and have additional mitochondrial electron pathways (e.g., alternative oxidase, sulfide-oxidizing electron transport systems) that allow them to sustain hypoxia–reoxygenation without a concomitant ROS formation burst (Buttemer et al., 2010). Lower susceptibility to peroxidation of membrane lipids could complement these systems in a manner that is independent of longevity.

More importantly, the membrane lipid composition of marine bivalves is largely different from that of endotherms and insects. First, they have considerable amounts of alk-1-enylglycerophospholipids, known as plasmalogens, in all organs (Kraffe et al., 2004). Plasmalogens differ from regular phospholipids in that the carbon chain of the sn-1 position is linked to the glycerol backbone by a vinyl ether linkage instead of the regular ester linkage. These alkenyl chains are often referred to as dimethyl acetal (DMA) from the name of the methylation product. In bivalves, most alkenyl chains are saturated and a few are monounsaturated (Kawashima & Ohnishi, 2003), hence they all are essentially resistant to oxidation and, as is the case for saturated fatty acids, their presence will lower the PI value of a membrane. Furthermore, the plasmalogen-specific vinyl ether linkage is known to act as a ROS scavenger and breaker of auto-oxidation, and it is believed to confer in situ antioxidant protection in membranes (Yavin et al., 2002; Engelmann, 2004; Kuczynski & Reo, 2006), providing a potential mechanism to modulate the resistance to peroxidation independently of the PI value.

Marine bivalves also have significant amounts of non-methylene-interrupted (NMI) FA in membrane lipids of all organs (Kraffe et al., 2004; Barnathan, 2009). In these diene FA, the double bonds are separated by more than one methyl group, hence the sensitive bis-allylic methylene hydrogens found on regular PUFA are absent. This structural property greatly reduces the sensitivity to oxidation (Kaneniwa et al., 1988), and recruitment of these FA may offer bivalves an alternative means by which to reduce the PI value for similar double bond indexes in the membrane (Barnathan, 2009). Significant amounts of branched fatty acids of the 16- to 18-carbon chain length are also found in marine bivalves (Kraffe et al., 2004), which can affect the physical properties of the membrane (Kaneda, 1991) without increasing the susceptibility to peroxidation. Therefore, different constraints acting on membrane lipid composition and different possible solutions to reduce the susceptibility to lipid peroxidation might result in widely different patterns of the PI–longevity relationship in marine bivalves.

To test this question, we used the mud clam Arctica islandica, which attracted attention when a 405-year-old individual was found, making it by far the longest living organism among metazoans (Wanamaker et al., 2008). Indeed, this maximum life span potential (MLSP) is almost double that of the second longest living species reported, the bowhead whale (Dendroica mysticetus, MLSP = 211 year) (de Magalhaes & Costa, 2009). During the writing of this article, this longevity record was increased by the discovery of another individual that had lived for more than half a millennium (507 year) (Butler et al., 2012). In this study, we compared the gill membrane lipid composition of A. islandica with four other species of marine bivalves of the heterochonchia subclass: Mya arenaria, Spisula solidissima, Mactromeris polynyma, and Mercenaria mercenaria. Together, A. islandica and these species meet four conditions that make them a proper model for aging studies (Speakman, 2005): overlapping areas of distribution, similar body sizes, members of close taxonomic groups, and greatly differing values of maximum reported longevity (MRL).

We further added a dimension to our test of the membrane-pacemaker theory of aging in bivalves by contrasting the change in lipid composition with longevity for the mitochondrial membrane vs. other cell membranes. Important developments of the oxidative stress theory of aging have indicated a major role for the accumulation of mutations to mtDNA during the onset and progression of senescence (Trifunovic et al., 2004; Kujoth et al., 2005). Within the cellular environment, the inner mitochondrial membrane is the major site of ROS production (>95% of ROS) in normal physiological conditions (Brand, 2010). For this reason, it should experience much greater lipid peroxidation than more distant cell membranes that are protected from ROS by the action of antioxidants over the diffusion distance needed to reach the target. Lipid peroxidation in the inner mitochondrial membrane releases mutagenic RCS in close proximity to or in direct contact with the mtDNA, again precluding effective scavenging by antioxidants. For the membrane-pacemaker theory of aging to be in accordance with these developments, one would expect the mitochondrial membrane to present a steeper decrease in PI with increasing longevity compared with other cell membranes.


Preliminary results and general lipid composition

While the five bivalve molluscs of the longevity model share similar body size, are part of a common taxonomic group, and have overlapping areas of distribution, they widely diverge in their MRL. Mya arenaria has the shortest MRL (28), the three intermediate species (S. solidissima, M. polynyma, and M. mercenaria) have progressively increasing MRL, and A. islandica has an extreme MRL (Fig. 1). Differential centrifugation produced a cell debris fraction (low centrifugal force) and a mitochondrial fraction (high centrifugal force). The preliminary electron transport system assay (measuring the presence of mitochondrial enzyme) showed that cell debris are essentially devoid of mitochondria compared with the mitochondrial pellet (see Experimental Procedures). For this reason, cell debris can be considered to be representative of the various cell membranes minus mitochondrial membranes and hence offer a contrast for the mitochondrial membrane. Separation of the phospholipids followed by transmethylation produced fatty acid methyl esters (FAME) and dimethyl acetals (DMA), the latter being released from the sn-1 position of plasmalogens (Morrison & Smith, 1964). Analysis by GC-MS allowed us to identify and quantify a total of 56 carbon chains (FAME and DMA); this included 27 straight-chain fatty acids, four branched fatty acids, seven NMI, and 18 DMA. Many of these carbon chains were never found in amounts > 0.5% in any species/biological fraction and were thus pooled in related groups (Tables 1 and 2). The general membrane lipid profile that we found is the same as that reported in previous bivalve membrane lipid analyses, for example, docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) are very abundant relative to what is generally found in endotherms, the 18:0 and 20:1 are the most abundant alkenyl chain (DMA) (Kawashima & Ohnishi, 2003), and the 20:2 and 22:2 dienes the most abundant NMI (Kraffe et al., 2004).

Figure 1.

 Taxonomy and life history traits for the five bivalve species representing a longevity model. MRL, maximum reported longevity.

Table 1.   Fatty acid and DMA composition (mol%) of phospholipids from gill mitochondria
  M. arenaria (n = 6)* S. solidissima (n = 8) M. polynyma (n = 9) M. mercenaria (n = 8)* A. islandica (n = 9) P-value
  1. Values are means ± SEM. Carbon chains < 0.5% are regrouped under straight-chain fatty acids < 0.5%, DMA < 0.5%, and branched fatty acids. MRL: maximum reported longevity.

  2. *Tissues from four individuals (M. arenaria) and two individuals (M. mercenaria) were pooled for each replicate.

  3. †Includes all branched fatty acids: (< 1%) 16:0 iso, 17:0 iso, and 18:0 iso; (>1%) 17:0 anteiso.

  4. ‡Includes the straight-chain fatty acids 10:0, 12:0, 20:0, 18:3 n-3, 20:3 n-6, 21:5 n-3, NMID 20:2 iso, and NMIT 20:3.

  5. §Includes the DMA 9:0, 15:0, 16:0 iso, 17:0 iso I, 18:0 iso II, 18:1, 19:0, 20:0, 20:1 iso I, and 20:2.

  6. Values with different letters are significantly different (alpha = 0.05; Tukey’s HSD); P-value from anova; NS: no significant differences.

9:00.1 ± 0.00.1 ± 0.00.5 ± 0.20.3 ± 0.10.3 ± 0.1(NS)
14:00.9 ± 0.1bc1.1 ± 0.1b2.8 ± 0.3a0.3 ± 0.0c0.4 ± 0.1c< 0.0001
15:00.5 ± 0.0b0.4 ± 0.0b0.9 ± 0.1a0.2 ± 0.0c0.3 ± 0.0bc< 0.0001
16:012 ± 0.3a5.9 ± 0.3c8.3 ± 0.6b7.0 ± 0.3bc6.4 ± 0.4c< 0.0001
16:1 n-72.9 ± 0.3a0.9 ± 0.1bc1.2 ± 0.1b1.1 ± 0.1b0.5 ± 0.0c< 0.0001
17:01.5 ± 0.1a0.8 ± 0.1b0.9 ± 0.1b1.4 ± 0.1a1.4 ± 0.1a< 0.0001
18:03.4 ± 0.0a4.9 ± 0.1b4.5 ± 0.4b5.4 ± 0.2bc6.3 ± 0.3c< 0.0001
18:1 n-110.2 ± 0.0a0.1 ± 0.0a0.2 ± 0.0a0.1 ± 0.0a4.9 ± 0.3b< 0.0001
18:1 n-93.3 ± 0.1a1.3 ± 0.1bc1.7 ± 0.2b0.9 ± 0.1c1.2 ± 0.1c< 0.0001
18:1 n-70.9 ± 0.1a0.4 ± 0.0b0.3 ± 0.0b1.1 ± 0.1a1.1 ± 0.1a< 0.0001
18:2 n-60.7 ± 0.1a0.3 ± 0.0b0.3 ± 0.0b0.2 ± 0.0b0.2 ± 0.0b< 0.0001
20:1 n-114.0 ± 0.3b3.9 ± 0.1b8.4 ± 0.4a1.1 ± 0.0c0.1 ± 0.0c< 0.0001
20:1 n-92.1 ± 0.2b2.2 ± 0.1b2.6 ± 0.1a0.5 ± 0.1c0.5 ± 0.0c< 0.0001
20:1 n-72.3 ± 0.1b1.4 ± 0.0c2.7 ± 0.1b1.0 ± 0.1c3.1 ± 0.2a< 0.0001
20:2 n-60.3 ± 0.0a0.5 ± 0.0b0.5 ± 0.0b0.6 ± 0.0b0.6 ± 0.0b< 0.0001
20:4 n-67.3 ± 0.2a2.5 ± 0.1cd2.2 ± 0.2d4.2 ± 0.2b2.9 ± 0.1c< 0.0001
20:5 n-39.5 ± 0.3ab8.7 ± 0.4ab10.9 ± 0.8a9.6 ± 0.5ab7.6 ± 0.3b  0.0006
22:4 n-61.9 ± 0.5ab2.6 ± 0.1a0.8 ± 0.0ab2.3 ± 0.2a0.5 ± 0.0b< 0.0001
22:5 n-62.4 ± 0.1a2.6 ± 0.2a0.9 ± 0.1c1.5 ± 0.1b0.9 ± 0.0c< 0.0001
22:5 n-30.6 ± 0.0a3.0 ± 0.2b3.0 ± 0.2b1.5 ± 0.2a1.1 ± 0.0a< 0.0001
22:6 n-319.0 ± 0.6a14.8 ± 0.6b15.4 ± 0.9b15.9 ± 0.6b12.0 ± 0.4c< 0.0001
NMID 20:2 (Δ5,11)3.4 ± 0.2b2.3 ± 0.1b3.6 ± 0.1b0.0 ± 0.0d9.8 ± 0.4a< 0.0001
NMID 20:2 (Δ5,13)1.0 ± 0.1a0.5 ± 0.1a0.5 ± 0.1a0.5 ± 0.1a4.2 ± 0.2b< 0.0001
NMID 22:2 (Δ7,13)0.2 ± 0.1a6.8 ± 0.2c1.5 ± 0.2b1.4 ± 0.1b0.0 ± 0.0a< 0.0001
NMID 22:2 (Δ7,15)0.0 ± 0.0a2.4 ± 0.1b0.0 ± 0.0a9.2 ± 0.4d4.5 ± 0.2c< 0.0001
NMIT 22:3 (7,13,16)0.0 ± 0.0a1.2 ± 0.1b0.1 ± 0.0a0.9 ± 0.1b0.0 ± 0.0a< 0.0001
DMA 16:00.6 ± 0.0b0.2 ± 0.0c0.1 ± 0.0d1.2 ± 0.0a0.1 ± 0.0d< 0.0001
DMA 17:0 iso II0.4 ± 0.0b0.1 ± 0.0bc0.0 ± 0.0c2.3 ± 0.1a0.0 ± 0.0c< 0.0001
DMA 17:00.3 ± 0.0b0.2 ± 0.0bc0.1 ± 0.0c1.0 ± 0.0a0.1 ± 0.0c< 0.0001
DMA 18:0 iso I0.5 ± 0.1a1.2 ± 0.1b0.5 ± 0.0a3.0 ± 0.1d1.4 ± 0.1b< 0.0001
DMA 18:02.3 ± 0.1a9.3 ± 0.3b2.6 ± 0.2a11.5 ± 0.7c7.6 ± 0.3b< 0.0001
DMA 19:0 iso0.1 ± 0.0a0.3 ± 0.0b0.2 ± 0.0a0.7 ± 0.0c0.8 ± 0.0c< 0.0001
DMA 20:1 iso II0.1 ± 0.0a2.5 ± 0.1b0.1 ± 0.0a0.0 ± 0.0a0.0 ± 0.0a< 0.0001
DMA 20:111.8 ± 0.6b11.5 ± 0.4b18.5 ± 0.5d2.4 ± 0.1a16.2 ± 0.5c< 0.0001
Branched FA†1.0 ± 0.1bc0.6 ± 0.0c0.7 ± 0.0bbc2.4 ± 0.1a1.1 ± 0.1b  0.0001
Fatty acids < 0.5%‡1.3 ± 0.11.2 ± 0.11.5 ± 0.21.2 ± 0.11.0 ± 0.1NS
SFA19.7 ± 0.5a13.9 ± 0.5b19.2 ± 1.6a17.1 ± 0.5ab16.6 ± 0.9ab  0.0023
UFA63.0 ± 0.6a59.2 ± 0.4ab57.6 ± 1.6bc54.4 ± 0.1c56.5 ± 0.6bc< 0.0001
MUFA15.6 ± 0.2a10.1 ± 0.2c17.1 ± 0.5a5.7 ± 0.2d11.5 ± 0.3b< 0.0001
PUFA47.4 ± 0.6ab49.1 ± 0.6a40.5 ± 1.6c48.7 ± 0.9ab44.9 ± 0.7b< 0.0001
n-6 PUFA12.6 ± 0.6a8.7 ± 0.4b4.8 ± 0.2c9.1 ± 0.3b5.1 ± 0.1c< 0.0001
n-3 PUFA29.3 ± 0.8a27.0 ± 0.5a29.9 ± 1.4a27.4 ± 1.1a21.0 ± 0.5b< 0.0001
n-3 PUFA (% PUFA)61.9 ± 1.5b55.0 ± 0.6c73.6 ± 1.1a56.1 ± 1.3c46.7 ± 1.0d< 0.0001
PUFA (without NMI)41.9 ± 0.7a35.7 ± 0.6b34.6 ± 1.5b36.4 ± 1.3b26.1 ± 0.6c< 0.0001
NMI total5.5 ± 0.3a13.4 ± 0.3b5.9 ± 0.3a12.2 ± 0.4b18.8 ± 0.6c< 0.0001
NMI (% PUFA)11.5 ± 0.7a27.3 ± 0.7b14.6 ± 0.8a25.3 ± 1.4b41.9 ± 1.1c< 0.0001
DMA < 0.5%§1.2 ± 0.1bc1.4 ± 0.1b1.1 ± 0.1a6.4 ± 0.3a0.8 ± 0.0c< 0.0001
Branched DMA (iso)1.8 ± 0.1ab4.7 ± 0.2c1.2 ± 0.1a11.6 ± 0.4d2.4 ± 0.1b< 0.0001
DMA total17.3 ± 0.7a26.8 ± 0.5c23.2 ± 0.5b28.5 ± 1.2c27.0 ± 0.7c< 0.0001
Unsaturation index255.5 ± 3.1a236.9 ± 3.3ab230.0 ± 8.2b222.1 ± 6bc202.3 ± 2.6c< 0.0001
Peroxidation index267.9 ± 5.1a232.7 ± 4.6b229.3 ± 10.6b236.5 ± 8.4ab175.2 ± 3.7c< 0.0001
MRL (years)283792106507 
Table 2.   Fatty acid and DMA composition (mol%) of phospholipids from gill cell debris
  M. arenaria (n = 6)* S. solidissima (n = 8) M. polynyma (n = 8) M. mercenaria (n = 8)* A. islandica (n = 8) P-value
  1. Values are means ± SEM. Carbon chains < 0.5% are regrouped under straight-chain fatty acids < 0.5%, DMA < 0.5%, and branched fatty acids. MRL: maximum reported longevity.

  2. *Tissues from four individuals (M. arenaria) and two individuals (M. mercenaria) were pooled for each replicate.

  3. †Includes all branched fatty acids: (< 1%) 16:0 iso, 17:0 iso, and 18:0 iso; (>1%) 17:0 anteiso.

  4. ‡Includes the straight-chain fatty acids 10:0, 12:0, 20:0, 18:3 n-3, 20:3 n-6, 21:5 n-3, NMID 20:2 iso, and NMIT 20:3.

  5. §Includes the DMA 9:0, 15:0, 16:0 iso, 17:0 iso I, 18:0 iso II, 18:1, 19:0, 20:0, 20:1 iso I, and 20:2.

  6. Values with different letters are significantly different (alpha = 0.05; Tukey’s HSD); P-value from anova; NS: no significant differences.

9:00.1 ± 0.01.0 ± 0.50.7 ± 0.20.2 ± 0.00.8 ± 0.2(NS)
14:01.1 ± 0.1c2.1 ± 0.1b3.6 ± 0.1a0.3 ± 0.00.5 ± 0.1d< 0.0001
15:00.5 ± 0.0b0.7 ± 0.0c1.0 ± 0.0d0.2 ± 0.0a0.4 ± 0.0b< 0.0001
16:012.9 ± 0.2a8.7 ± 0.4b9.1 ± 0.4b7.8 ± 0.2b8.9 ± 0.4b< 0.0001
16:1 n-73.2 ± 0.1a1.3 ± 0.1b1.2 ± 0.2b1.6 ± 0.1b1.1 ± 0.2b< 0.0001
17:01.2 ± 0.0ab0.9 ± 0.1b0.9 ± 0.1b1.2 ± 0.1b1.7 ± 0.1a< 0.0001
18:04.0 ± 0.2a5.9 ± 0.2b5.3 ± 0.2b4.5 ± 0.1a7.6 ± 0.3c< 0.0001
18:1 n-110.1 ± 0.1a0.1 ± 0.0a0.3 ± 0.0a0.1 ± 0.0a3.9 ± 0.1b< 0.0001
18:1 n-93.8 ± 0.0a1.9 ± 0.1b1.8 ± 0.1b1.0 ± 0.1c1.9 ± 0.2b< 0.0001
18:1 n-70.8 ± 0.0a0.5 ± 0.0a0.4 ± 0.0a1.3 ± 0.1ab2.1 ± 0.2b< 0.0001
18:2 n-60.5 ± 0.0a0.4 ± 0.0a0.2 ± 0.0b0.2 ± 0.0b0.3 ± 0.0b< 0.0001
20:1 n-114.8 ± 0.1b3.5 ± 0.1c7.4 ± 0.2a1.2 ± 0.0d0.1 ± 0.0 e< 0.0001
20:1 n-92.7 ± 0.0a2.5 ± 0.1a2.5 ± 0.1a0.6 ± 0.1b0.6 ± 0.0b< 0.0001
20:1 n-73.1 ± 0.0ab1.8 ± 0.1b3.8 ± 0.1a1.3 ± 0.1c4.0 ± 0.2a< 0.0001
20:2 n-60.3 ± 0.0a0.6 ± 0.0c0.4 ± 0.0b0.7 ± 0.0d0.6 ± 0.0c< 0.0001
20:4 n-68.3 ± 0.1a2.6 ± 0.1c2.6 ± 0.2c4.6 ± 0.2b2.9 ± 0.1c< 0.0001
20:5 n-37.1 ± 0.1a6.8 ± 0.5a9.9 ± 0.3b7.1 ± 0.3a8.0 ± 0.4a< 0.0001
22:4 n-61.0 ± 0.1b2.3 ± 0.2a0.7 ± 0.0bc2.7 ± 0.2a0.4 ± 0.0c< 0.0001
22:5 n-62.0 ± 0.1a1.9 ± 0.1a0.7 ± 0.1c1.5 ± 0.0b0.6 ± 0.1c< 0.0001
22:5 n-31.3 ± 0.0a3.4 ± 0.1c3.5 ± 0.2c2.2 ± 0.1b1.2 ± 0.1a< 0.0001
22:6 n-314.8 ± 0.2a14.5 ± 0.7a15.4 ± 0.6a13.3 ± 0.3ab11.6 ± 0.7b  0.002
NMID 20:2 (Δ5,11)3.3 ± 0.1b2.0 ± 0.1c3.1 ± 0.1b0.0 ± 0.0d8.1 ± 0.2a< 0.0001
NMID 20:2 (Δ5,13)1.0 ± 0.1a0.7 ± 0.1a0.7 ± 0.1a0.7 ± 0.1a3.4 ± 0.2b< 0.0001
NMID 22:2 (Δ7,13)0.0 ± 0.0a5.1 ± 0.2c1.2 ± 0.1b1.3 ± 0.1b0.1 ± 0.0a< 0.0001
NMID 22:2 (Δ7,15)0.0 ± 0.0a1.8 ± 0.1b0.0 ± 0.0a9.0 ± 0.3d3.4 ± 0.5c< 0.0001
NMIT 22:3 (7,13,16)0.0 ± 0.0a0.9 ± 0.1b0.1 ± 0.0a1.0 ± 0.1b0.0 ± 0.0a< 0.0001
DMA 16:00.8 ± 0.0b0.3 ± 0.0c0.1 ± 0.0d1.4 ± 0.0a0.1 ± 0.0d< 0.0001
DMA 17:0 iso II0.5 ± 0.0b0.1 ± 0.0c0.0 ± 0.0c2.5 ± 0.1a0.1 ± 0.0c< 0.0001
DMA 17:00.3 ± 0.0b0.2 ± 0.0c0.1 ± 0.0d1.1 ± 0.0a0.1 ± 0.0cd< 0.0001
DMA 18:0 iso I0.7 ± 0.0b1.1 ± 0.1c0.5 ± 0.1a3.2 ± 0d1.2 ± 0.1c< 0.0001
DMA 18:02.5 ± 0.1a7.6 ± 0.2b2.6 ± 0.2a11.8 ± 0.5c6.1 ± 0.2b< 0.0001
DMA 19:0 iso0.1 ± 0.0a0.3 ± 0.0b0.1 ± 0.0a0.7 ± 0.0c0.7 ± 0.0c< 0.0001
DMA 20:1 iso II0.0 ± 0.0a2.3 ± 0.2b0.1 ± 0.0a0.0 ± 0.0a0.0 ± 0.0a  0.0002
DMA 20:113.7 ± 0.3b10.4 ± 0.4c16.5 ± 0.4a2.6 ± 0.1d14.3 ± 0.5b< 0.0001
Branched FA†1.0 ± 0.0a0.7 ± 0.1a0.9 ± 0.1a2.8 ± 0.1c1.4 ± 0.1b< 0.0001
Fattyacids < 0.5%‡1.2 ± 0.11.7 ± 0.21.2 ± 0.21.2 ± 0.01.3 ± 0.1NS
SFA20.9 ± 0.5ab20.5 ± 1.3ab22.1 ± 0.8a17.3 ± 0.3b21.8 ± 1.0a  0.0026
UFA59.1 ± 0.2a55.8 ± 1.0b56.8 ± 0.6ab52.4 ± 0.5c54.8 ± 0.8bc< 0.0001
MUFA18.5 ± 0.2a11.6 ± 0.2c17.5 ± 0.3a7.1 ± 0.3d13.6 ± 0.4b< 0.0001
PUFA40.7 ± 0.2a44.2 ± 0.9ab39.3 ± 0.6a45.3 ± 0.4b41.2 ± 0.8ab  0.0003
n-6 PUFA12.1 ± 0.2a7.9 ± 0.3c4.8 ± 0.1d9.9 ± 0.3b4.8 ± 0.2d< 0.0001
n-3 PUFA23.4 ± 0.3bc25.4 ± 0.9b29.2 ± 0.7a23.1 ± 0.3bc21.1 ± 0.8c< 0.0001
n-3 PUFA (% PUFA)57.6 ± 0.6b57.4 ± 1.4b74.1 ± 0.7a51.1 ± 0.5b51.2 ± 1.4c< 0.0001
PUFA (without NMI)35.5 ± 0.2a33.4 ± 1a33.9 ± 0.7a33.0 ± 0.6a25.9 ± 0.9b< 0.0001
NMI total5.2 ± 0.1a10.9 ± 0.3b5.4 ± 0.2a12.3 ± 0.3b15.3 ± 0.6c< 0.0001
NMI (% PUFA)12.7 ± 0.2a24.6 ± 0.9b13.7 ± 0.6a27.1 ± 0.8b37.1 ± 1.6c< 0.0001
DMA <  0.5%§1.3 ± 0.1bc1.4 ± 0.0b1.1 ± 0.1cd7.0 ± 0.1a0.8 ± 0.0d< 0.0001
Branched DMA (iso)2.2 ± 0.1b4.2 ± 0.2c1.0 ± 0.1a12.6 ± 0.2d2.1 ± 0.1b< 0.0001
DMA total20.0 ± 0.3a23.6 ± 0.6b21.1 ± 0.5a30.3 ± 0.6c23.4 ± 0.5b< 0.0001
Unsaturation index223.6 ± 1.3a219.6 ± 5.3a224.3 ± 3.8a203.4 ± 2.3b194.1 ± 4.6b< 0.0001
Peroxidation index221.8 ± 1.5a216.8 ± 6.8a225.3 ± 5.2a209.3 ± 2.9a172.4 ± 6.2b< 0.0001
MRL (years)283792106507 

Longevity and peroxidation-sensitive fatty acids

Interestingly, the abundance of the highly unsaturated DHA markedly decreased with increasing longevity. This FA is the single most unsaturated, with six double bonds on the carbon chain and is eight time more susceptible to peroxidation than fatty acids with two double bonds and 320 times more susceptible than fatty acids with one double bond (Holman, 1954). In the mitochondrial fraction, DHA is significantly more abundant in the short-lived M. arenaria and significantly less abundant for the extremely long-lived A. islandica compared to the other species (Table 1, Fig. 2A). In contrast, this trend is much less pronounced in the cell debris fraction: the value for M. arenaria is not different from those species with intermediate longevity, and A. islandica is no longer different from at least one species with intermediate longevity (M. mercenaria). We plotted the mean DHA abundance for each species against their MRL and found a negative linear correlation for the cell debris (R2 = 0.774, = 0.049; Fig. 3A). However, the abundance of DHA decreases exponentially in the mitochondrial fraction and is best described by a negative correlation with the logarithm of the MRL (R2 = 0.688, = 0.0824; Fig. 3A).

Figure 2.

 Key lipid composition indicators in mitochondrial membrane phospholipids (black bars) and cellular debris phospholipids (white bars) of the five bivalve species. % DHA, docosahexaenoic acid (A), % PUFA, polyunsaturated fatty acids (B), PI, peroxidation index (C), % NMI, non-methylene-interrupted fatty acids (D), %NMI + 20:1n−11 FA (E), and % plasmalogens (F). Species are presented in the order of increasing reported maximum longevity from left to right. Capital and lower-case letters indicate significant differences among species for mitochondria and cell debris, respectively. Asterisks represent significant within-species differences between mitochondria and cellular debris. See Tables 1 and 2 for sample sizes and species longevity. Values are means ± SEM.

Figure 3.

 Relationships between % docosahexaenoic acid (A), % peroxidation-sensitive polyunsaturated fatty acids (B), and peroxidation index (C) and maximum reported longevity. The peroxidation index is also plotted against the longevity known to be attained in at least two populations of the species in (D). Data for gill mitochondrial membranes (black symbols and solid lines) and cellular debris phospholipids (white symbols and dashed lines) are presented with their best fitting model. See Tables 1 and 2 for sample sizes. Values are means ± SEM.

Taken separately, neither the n-6 PUFA nor the n-3 PUFA show any relationship to longevity, but their sum decreases with increasing MRL. In other words, the sum of all peroxidation-sensitive PUFA (FA class above the line in Fig. 4), which exclude NMI, decreases with increasing longevity. Similar to what was noted for DHA, this decrease is linear for the cell debris (R2 = 0.961, = 0.003; Fig. 3B) and exponential for mitochondria (R2 = 0.850, = 0.026; Fig. 3B). Individual comparisons reveal that in the mitochondrial fraction, the sum of all peroxidation-sensitive PUFA is significantly higher for the short-lived M. arenaria, while it is significantly lower for the extremely long-lived A. islandica (= 0.014 or lower in all cases; Fig. 2B). In contrast, for the cell debris fraction, the difference is no longer significant for the comparison between the short-lived M. arenaria and the intermediate species but remains significant for the extremely long-lived A. islandica (< 0.0001 in all cases; Fig. 2B).

Figure 4.

 Carbon chain composition of mitochondrial (A) and cellular debris (B) phospholipids from bivalve gills. Species are presented in the order of increasing reported maximum longevity from left to right. Letters indicate significant differences between species for each carbon chain classes. Carbon chain classes are indicated on the figure: DMA, dimethyl acetals, SFA, saturated fatty acids, MUFA, monounsaturated fatty acids, NMI, non-methylene-interrupted fatty acids, and PUFA, polyunsaturated fatty acids. The line delimits carbon chains that are resistant to peroxidation (below) and those that are peroxidation-sensitive (n-6 PUFA + n-3 PUFA; above). See Tables 1 and 2 for sample sizes.

A lower abundance of DHA and peroxidation-sensitive PUFA in general will result in a lower PI value with increasing longevity. This is clearly what is found for the mitochondrial fraction, although the relationship is less pronounced for the cell debris fraction. Indeed, the resulting PI for the mitochondrial membrane is significantly higher for the short-lived M. arenaria compared to all other species, with the exception of the comparison with M. mercenaria, where it just fails to reach significance (= 0.058; Fig. 2C). It is also significantly lower for the extremely long-lived A. islandica compared with all other species (< 0.0001; Fig. 2C,). However, for the cell debris fraction, there are no differences between the short-lived M. arenaria and the intermediate species, and only the value for the extremely long-lived A. islandica is significantly reduced compared with the other species (< 0.0001; Fig. 2C). Not surprisingly, when the mean PI for each species is plotted against their MRL, the decrease is linear for the cell debris (R2 = 0.924, = 0.009; Fig. 3C) and exponential for the mitochondrial membrane (R2 = 0.855, = 0.024; Fig. 3C).

The use of MRL may introduce a bias in estimating the relative longevity differences between species. MRL is known to differ between populations of bivalves of the same species (Basova et al., 2012), and efforts devoted to finding each species’ longevity records are unequal. Shells of long-lived A. islandica individuals are valuable tools for reconstructing the ocean climate (Wanamaker et al., 2008; Butler et al., 2012), which has led to the search for the longest lived individuals among populations of this species. Similarly, the growing interest in using bivalves as a longevity model has led to the search for the longest lived M. mercenaria individual (Ridgway et al., 2011). In contrast, the three other species of bivalves that we used to build the longevity model have only been studied in the context of gathering information on life history traits such as estimating maturity at age, often for the purpose of fisheries management. This problem can be partly circumvented by searching for the MRL in different populations and using the second value in decreasing order, a variable hereafter referred to as the shared maximum longevity (SML) because it represents the maximum longevity reported to be attained in at least two populations. After further examination of the literature, this value was found to be 17, 35, 46, 52, and 267, respectively, for M. arenaria, S. solidissima, M. mercenaria, M. polynyma, and A. islandica. Note that M. mercenaria and M. polynyma now exchange their position in the gradient of increasing longevity. This new method of evaluating the maximum longevity should be preferable because the interest of this study is not to find the best estimate of the maximum life span potential of each species, but rather to investigate the existence of a correlation between membrane PI and the frequently reached longevity in average populations such as those from which our individuals were sampled. Figure 3D shows the results of such a re-examination. The cell debris PI still shows a linear decrease and the model fit is not much affected. The mitochondrial PI still decreased exponentially; however, the model fit is greatly improved, giving a remarkable correlation (R2 = 0.977, = 0.001). This might be much closer to the true relationship between longevity and membrane properties in the populations that we studied.

Non-methylene-interrupted fatty acids

The abundance of NMI FA is included in the calculation of the PI value; however, this group of FA deserves particular attention because they have been suggested to offer antioxidant protection in membranes (Barnathan, 2009). Indeed, empirically determined oxygen consumption (oxidation) for the most abundant NMI of our species translates into a much reduced individual PI value as compared to normal diene FA. From the results of Kaneniwa et al. (1988), we estimated the PI for NMI dienes to be 0.258 and 0.320 for the 20- and 22-carbon chains, respectively, while it is one for regular dienes. Therefore, NMI have the potential to lower the susceptibility to peroxidation for a minimal loss of membrane fluidity; hence, one could expect their recruitment to compensate for a decrease in peroxidation-sensitive PUFA in long-living species. We found the 20:2 NMI to predominate over the 22:2 in M. arenaria, M. polynyma, and A. islandica, while the opposite was found for S. solidissima and M. mercenaria. When all NMI are considered as a group, their abundance in both biological fractions is relatively low in the short-lived M. arenaria and significantly higher for the extremely long-lived A. islandica compared with all other species (Fig. 2D). Despite this trend, no significant correlation between NMI and MRL or SML was found across all five species because the low values for M. polynyma do not fit. However, M. polynyma have significantly higher abundances of 20:1n-11 compared with all other species (Tables 1 and 2), and in endotherms, where NMI are absent, monoenes are often found to be recruited to compensate for the decrease in highly PUFA in long-living species. For this reason, we calculated the sum of all NMI and 20:1n-11 to see whether this new variable would better correlate with longevity. In both biological fractions, the addition of 20:1n-11 FA does indeed improve the consistency of the trend (Fig. 2E), but the models for positive correlation still do not reach significance for either MRL or SML. Nonetheless, for the mitochondrial fraction, NMI of either the chain length or the 20:1n-11 FA appears to be the sole potential candidate acyl chains recruited to compensate for the decrease in highly polyunsaturated FA with increasing longevity in bivalves.


Plasmalogen abundance (see online supporting information for quantification method) is not integrated into the calculation of the PI, and these lipids should be viewed as offering additional protection against peroxidation (Engelmann, 2004; Kuczynski & Reo, 2006). Within the mitochondrial fraction, plasmalogen abundance is significantly lower for the short-lived M. arenaria than for any other species, while it is among the highest for A. islandica (Fig. 2F). However, this trend for an increasing abundance of plasmalogens with increasing longevity does not reach significance when modeled using either MRL or SML. Within the cell debris fraction, plasmalogen abundance does not show any relationship with longevity.


We found that the peroxidation index (PI) of gill membrane lipids significantly decreases with increasing longevity in marine bivalves ranging from 28 to 507 year in MRL. Furthermore, when using a more conservative approach (maximum longevity shared by at least two populations) to estimate the maximum longevity attainable by our bivalve species, the relationship becomes remarkably precise for the mitochondrial fraction. We base these conclusions on a strong longevity model. First, all species of the model share similar body sizes, avoiding the possible allometric confounding factor that has been revealed for mammals (Couture & Hulbert, 1995; Hulbert et al., 2002). Second, individuals were maintained for at least 4 months on a common diet and temperature regime. Third, the five species are from closely related taxonomic groups, with four from the same order and the fifth (M. arenaria) from the same subclass. This is particularly important because no data are currently available to build a proper phylogenetic tree (appropriate topology and realistic branch length) for these five species to apply corrections for phylogenic nonindependence (Garland et al., 2005). Nonetheless, the selection of closely related species is a way to better respect the assumptions of classical statistics that does not rely on any assumption in itself. First, it tends to reduce the amplitude of the phylogenetic signal to be corrected, and secondly, it tends to equilibrate the phylogenic distance between species (tend toward star phylogeny). For these reasons, we remain confident in the conclusions of our study, which are based on strong P-values. It is important to note, however, that the significant correlation between DHA and longevity for the cell debris (P < 0.049) could very well become nonsignificant after IC correction.

Pattern of lipid remodeling with increasing longevity in bivalves

The PI decrease appears to be achieved by precisely controlling the ratio of peroxidation-sensitive PUFA (PUFA with two or more double bonds except NMI) to peroxidation-resistant alkenyl and acyl chains, the latter comprising saturated FA, monoenes, and NMI. Previous studies in mammals and birds established the existence of an inverse correlation between PI and longevity for both the liver mitochondria and the total phospholipids in skeletal muscle. In these species, a recurring and important component of the shift for less unsaturated PUFA is the exchange of DHA for the less peroxidizable monounsaturated 18:1n-9 FA. Monoenes are a proper choice to replace highly PUFA: the addition of the first double bond to a carbon chain only causes a limited increase in the susceptibility to peroxidation because the highly fragilized bis-allylic hydrogen atoms are only found when a methyl group is immediately flanked by two double bonds. On the other hand, by introducing a kink in the carbon chain, the addition of a first double bond brings the greatest contribution to the fluidity of the membrane while additional double bonds will have a lesser effect. Because lowering membrane fluidity is known to lower the activity of transmembrane enzymes and affect the maintenance of ion gradients across membranes (Else & Hulbert, 2003), exchanging DHA for a monoene represents an efficient compromise for achieving a reduced PI while maintaining functionality.

In bivalves, we also found that DHA systematically decreases with increasing MRL or SML, contributing approximately 46% to the total decrease in peroxidation-sensitive PUFA in gill mitochondria from the shortest-lived to the longest lived species. However, the 18:1n-9 does not increase with increasing longevity and was never found in concentrations >3.3%. Nonetheless, we found that the sum of NMI and 20:1n-11 FA presents a loose positive correlation with longevity in both fractions. As is the case for 18:1n-9, both the NMI and the 20:1n-11 represent an interesting compromise between fluidity loss and reduced susceptibility to peroxidation. The species-specific recruitment of one or a mix of these acyl chains might serve to compensate for the loss of highly PUFA with increasing longevity in long-living bivalves, much as the recruitment of 18:1n-9 seems to do in endotherms.

Mitochondrial membrane vs. other cell membranes

The parallel analysis of mitochondrial membranes and of other cell membranes revealed a very interesting pattern. Even though the decrease in PI with increasing longevity is linear for the cell debris, an exponential decrease best describes the relation for mitochondria. A greater emphasis on increasing resistance to lipid peroxidation with longevity in the mitochondrial membrane as compared to other cell membranes would not be surprising given some basic cellular morphology and metabolic considerations. In normal physiological conditions, ROS are mainly produced by the electron transport system (mostly complex I and III) located in the inner mitochondrial membrane, and most ROS are released to the matrix side of this membrane (Brand, 2010). In theory, this should subject the mitochondrial membrane to an immensely higher rate of lipid peroxidation than remote membranes, for which peroxidation can only be initiated by those ROS that can diffuse through the membrane, have a sufficient half-life to migrate a certain distance, and have not been intercepted by water-soluble antioxidants. The consequences of an elevated level of ROS attack to the inner mitochondrial membrane are not limited to the loss of functionality such as the loss of fluidity and increased permeability. The deleterious RCS released as the breakdown product of oxidized lipids have a much longer half-life than ROS and can diffuse through the membrane to reach remote targets. For these reasons, the mitochondrial membrane may be the key membrane acting as an amplifier of primary ROS by producing secondary ROS (RCS) (Hulbert, 2005).

Furthermore, mtDNA is in close proximity or even in direct contact with the inner mitochondrial membrane and will consequently suffer a greater rate of attack by both ROS and RCS (Barja, 2000). Mutations are known to accumulate on mtDNA with age and are believed to result in the gradual loss of ATP production capacity of mitochondria, which would account for a major part of the gradual and irreversible loss of physiological capacities that characterizes aging (Barja, 2000; Pamplona, 2011). This hypothesis has received strong support from a transgenic mouse model that has a proofreading–deficient version of the mtDNA polymerase gamma (PolgAD257A mutants), which causes it to accumulate mtDNA mutations at a faster rate and display accelerated aging (Trifunovic et al., 2004; Kujoth et al., 2005). Like bivalve mitochondrial PI, the PI of mammalian membranes also show an exponential decrease with increasing longevity. Interestingly, when modeled with a power function, the liver mitochondrial membrane PI decreases with an exponent of −0.40 while that of the skeletal muscle phopholipids, which includes mitochondria as well as other cell membranes, decreases with an exponent of −0.30. This represents a 24% decrease in PI with every doubling in maximum life span for liver mitochondria compared with a 19% decrease in the skeletal muscle phospholipids (Hulbert et al., 2007). Although these are two distinct organs, it is nonetheless tempting to conclude that the PI decreases more abruptly in the mitochondrial membrane than in the other cell membranes in mammals. Furthermore, one study analyzed the fatty acid composition of brain mitochondria and microsomes by comparing a senescence-accelerated mouse strain to a senescence-resistant one (Choi et al., 1996). While mitochondrial PI was found to be higher for the senescence-accelerated strain, there were no differences between microsomes of either strain, suggesting that the reduction in PI with increasing longevity does not occur in the endoplasmic reticulum’s membrane.

Bivalves vs. mammals

It is interesting to compare the gill mitochondrial membrane PI value that we found for bivalves to the predicted value for liver mitochondria of similarly long-lived mammals using the equation in Hulbert et al. (2007). Considering M. arenaria (MRL = 28) and A. islandica (MRL = 507), which are the extremes of the longevity gradient, we found values of 268 and 175, respectively. The PI for the liver mitochondrial membranes of mammals is calculated as PI = 475*maximum longevity -0.40, which gives values of 125 and 39, respectively, for the corresponding longevity. A higher PI value in bivalves than in mammals of similar longevity should not be interpreted as a denial of the membrane-pacemaker theory of aging for many reasons. The most obvious one is the presence of a high relative proportion of plasmalogens in bivalve tissues that will reduce susceptibility to peroxidation independently of the PI value. Plasmalogens are glycerophospholipids displaying a vinyl ether linkage at the sn-1 position instead of the regular ester linkage. This linkage is susceptible to oxidation; however, when oxidized, it does not participate in the propagation of the chain reaction of lipid peroxidation. A review article (Engelmann, 2004) on plasmalogens concludes that they satisfy the criteria for the definition of a lipophilic antioxidant by protecting membranes from lipid peroxidation. It has been proposed that the relatively high abundance of plasmalogen in the human brain and nervous tissues in general compared with other organs could explain the tolerance for a concomitant high level of DHA in neuron membranes (Yavin et al., 2002), and there is in vivo evidence of a protective effect against lipid peroxidation in the rat brain (Kuczynski & Reo, 2006). Furthermore, plasmalogen deficiency in the human brain is associated with premature aging, Down syndrome, and Alzheimer’s disease (Nagan & Zoeller, 2001; Grimm et al., 2011). Bivalve membranes are similar to mammalian neuron membranes in that they are rich in DHA and have elevated levels of plasmalogens in general (Kraffe et al. 2004), particularly in the gills (Nevenzel et al., 1985). We found an average of 49.1% plasmalogen glycerophospholipid in the gills of our bivalves. Compared with the average 20% proportion found in the adult human brain (Farooqui & Horrocks, 2001), this abundance suggests an important additional protection against lipid peroxidation in the membranes of these organisms.

Mammalian liver mitochondria also differ from bivalve gill mitochondria in the rate of decline of the PI value with increasing longevity. When the relation between mitochondrial membrane PI and SML of our bivalves is modeled with a power function, we find an exponent of −0.15 while that of the liver mitochondria of mammals reported in Hulbert et al. (2007) is −0.40. Thus, with every doubling in longevity, the PI of the mitochondrial membrane of the bivalve gills is reduced by 11%, while it is reduced by 24% for mammal liver mitochondria. Once again, plasmalogens can account for this difference because the content of these lipids increases irregularly but markedly in gill mitochondria from the short-lived M. arenaria to the extremely long-lived A. islandica, ranging from 34.6% to 54%. Therefore, the increasing abundance of plasmalogens might compensate for a lower rate of decrease in PI to achieve the same resulting decrease in susceptibility to peroxidation with increasing longevity. Unfortunately, we are not aware of any common denominator by which to compare the influence of the PI and of the abundance of plasmalogens on the resulting global susceptibility to peroxidation of a membrane. Interestingly, when the long-living naked mole rat (Heterocephalus glaber) was compared with the laboratory mouse (Mus musculus), a higher abundance of the plasmalogen form of phosphatiditylcholine was found in the skeletal muscle, heart, liver, and liver mitochondria glycerophospholipids, and the same was also true for the phosphatidylethanolamine fraction in skeletal muscle and liver (Mitchell et al., 2007). Future research on the PI–longevity relationship should include a complete integration of plasmalogens when using such organs as the heart and brain or when using longevity models that possess high amounts of plasmalogens in all organs. Failure to do so would result in the biased estimate of the actual PI by failing to include the alkenyl chains (DMA) and would ignore the additional protection conferred by the vinyl ether linkage. For instance, this omission may partly explain why a consistent relationship between membrane PI and longevity was not always found in previous studies addressing this question for the mammalian brain.

A lower rate of ROS production in marine bivalves can also account for the difference in PI values for similar longevities. The membrane-pacemaker theory of aging does not propose that increased longevity is achieved by a lower PI value per se, but rather by the resulting lower damage accrual to membranes and a lower rate of RCS production that should accompany lower PI values when everything else is equal. However, Buttemer et al. (2010) reviewed the literature and found that the ROS production rate of isolated mitochondria is on average an order of magnitude lower for marine bivalves compared with mammals. For instance, it is reported that the succinate-driven state III rate of H2O2 production is 0.144 ± 0.14 nmol H2O2*min−1*mg prot−1 for M. arenaria mantle mitochondrial and even lower for A. islandica mantle mitochondria, while it ranges between 2 and 2.8 nmol H2O2*min−1*mg prot−1 in rat heart mitochondria. Therefore, during normal physiological conditions, a lower body temperature coupled with a lower ROS production rate might result in a lower realized peroxidation rate in bivalve mitochondria despite higher PI values.

Conclusion and perspectives

We have shown that PI values for bivalve gill mitochondrial membranes follow an exponential decrease with increasing species longevity, as was found for liver mitochondria and skeletal muscle phospholipids in mammals and birds; this finding definitely excludes the possibility of a mechanism that is limited to endotherms or vertebrates. Moreover, the fact that (i) the PI–longevity relationship also exists in bivalves even though they face additional constraints believed to act on membrane lipid composition and that (ii) the lipid remodeling is partly achieved using different types of lipids (and hence different enzymatic pathways) in these organisms are two important observations. They support the membrane-pacemaker theory of aging as a general and fundamental determinant of longevity. The distinct PI–longevity relationship found for mitochondria (exponential decrease) and other membranes (linear decrease) suggests that an increase in longevity primarily results from decreasing lipid peroxidation in the mitochondrial environment. In addition, our results suggest that high levels of plasmalogens could provide DHA-rich tissues with a protection against lipid peroxidation in a similar manner for bivalve and mammalian nervous tissues. Finally, the significantly lower PI and elevated levels of plasmalogens found for A. islandica mitochondria should be a prime factor contributing to this species’ extraordinary longevity. Further research is necessary to incorporate the PI value and plasmalogen abundance into a single metric representing the overall susceptibility to peroxidation of a membrane in order to include various organs and organisms within a single means of evaluating the susceptibility to peroxidation. Survey on other species of veneridae and construction of robust phylogenetic trees will allow delineating the phylogenetic signal in the establishment of the relationship between structural properties of mitochondrial membranes and animal longevity. Our laboratory is currently working to sequence the whole mitochondrial genomes of veneridae species used in the present study as well as other representative species for further studies.

Experimental procedures

Bivalves used in this study were from populations located between the latitude 47°15′ and 48°35′ in the provinces of Quebec and New Brunswick, Canada. Individuals were obtained during summer 2009 and kept at the ISMER marine research station in Pointe-au-Père, Québec, Canada, in an open flow-through, two-tank system. In addition to the natural phytoplankton remaining in the semi-filtred (50 μm) water of the station, bivalves were fed a live microalgal diet (Nannochloropsis sp., Isochrysis galbana, Pavlova lutheri; 12% to 44% to 44% in cell numbers) at 1% body mass/day following a Q10 of 2.5 adjusted at 8 °C between April and December.

Individuals were sampled between December 2009 and April 2010, after water temperature in the tanks had stabilized at 4 °C. Gills were removed from bivalves, rinsed with saline water, and maintained at 4 °C while being thoroughly minced with a razor blade. Homogenization was achieved using three passes of a loose-fitting precooled glass/teflon pestle in six parts ice-cold homogenization buffer. The homogenate was centrifuged at 1250 g for 10 min and the supernatant was removed and conserved. The pellet was resuspended in 15 mL of homogenization buffer and centrifuged at 1000 g for 10 min; this second supernatant was removed and the final pellet of cellular debris was conserved. The two supernatants were combined and a first centrifugation was run at 1250 g for 10 min to eliminate residual debris. The final supernatant was considered free of unbroken cells or cell debris and was centrifuged at 10 500 g for 15 min. The final mitochondrial pellet was resuspended in 150 μL of homogenization buffer. All centrifugation steps were carried out at 4 °C. Mitochondria and cellular debris were held at −80 °C after nitrogen flush and until lipid extraction. We compared the abundance of mitochondrial electron transport systems in both biological fractions (mitochondrial pellet and cell debris) using the NADH-INT (iodonitrotetrazolium) oxidoreduction assay, which targets complexes I and III. Cell debris always provided values of mitochondrial abundance that were below 1/20 and usually around 1/100 (or no signal) of those obtained for the mitochondrial pellet. We thus consider the cell debris to be essentially devoid of mitochondria and to represent other cell membranes.

Glycerophospholipids were separated from total lipid extracts using silica gel (60, mesh 150–230) microcolumns and trans-methylated using 10% boron trifluoride–methanol at 90 °C for 45 min. All solvents used were ultra pure grade.

The resulting fatty acid methyl esters (FAME) and dimethyl acetals (DMA) were analyzed by GC-MS using a column with a high-polarity stationary phase (HP-88 60 m, 0.25 mm × 0.20 μm). Calibration of the system was performed using regular FA mix standards, marine FA standards, and DMA standards. The preservation of intact DMA was verified using a standard of the pure plasmalogen PE (AVANTI Polar Lipids Inc, Alabaster, AL, USA) subjected to the same protocol. The peroxidation index (PI) is 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), including a single case of DMA among monoenes (Hulbert et al., 2007). Individual PI values for NMI are estimated from Kaneniwa et al. (1988).

Significant effects (alpha = 0.05) were determined by ANOVA. Where ANOVA revealed a significant effect, Tukey–Kramer HSD was used to provide P-values for the differences between species. Homogeneity of variance was estimated using the Brown–Forsythe test, and variables were log10 transformed when necessary. Results are presented as means ± SEM.


We are grateful to Dr Richard St-Louis from ISMER, Dr Edouard Kraffe, and Diane Bérubé for their help in the development of the lipid analysis method. Daniel Munro is supported by a scholarship grants from NSERC and the FONCER programme from Réseau Aquaculture Québec. This project was supported by an NSERC Discovery grant to Pierre Blier.

Authors contribution

Daniel Munro was responsible for acquiring and raising the individuals, developing the lipid analysis method, acquiring the data, and writing the first draft. Pierre Blier contributed to writing the manuscript.