Maternally Inherited Differences within Mitochondrial Complex I Control Murine Healthspan

Abstract

Mitochondrial complex I—the largest enzyme complex of the mitochondrial oxidative phosphorylation machinery—has been proposed to contribute to a variety of age-related pathological alterations as well as longevity. The enzyme complex-consisting proteins are encoded by both nuclear (nDNA) and mitochondrial DNA (mtDNA). While some association studies of mtDNA encoded complex I genes and lifespan in humans have been reported, experimental evidence and the functional consequence of such variants is limited to studies using invertebrate models. Here, we present experimental evidence that a homoplasmic mutation in the mitochondrially encoded complex I gene mt-Nd2 modulates lifespan by altering cellular tryptophan levels and, consequently, ageing-related pathways in mice. A conplastic mouse strain carrying a mutation at m.4738C > A in mt-Nd2 lived slightly, but significantly, shorter than the controls did. The same mutation led to a higher susceptibility to glucose intolerance induced by high-fat diet feeding. These phenotypes were not observed in mice carrying a mutation in another mtDNA encoded complex I gene, mt-Nd5, suggesting the functional relevance of particular mutations in complex I to ageing and age-related diseases.

1. Introduction

Mitochondrial DNA encode genes for two ribosomal RNAs and the 22 transfer RNAs, as well as the 13 proteins of mitochondrial respiratory process, i.e., oxidative phosphorylation (OXPHOS) complexes [1,2]. Of the five enzyme complexes of the mitochondrial respiratory process, OXPHOS complex I (NADH:ubiquinone oxidoreductase) is the largest enzyme, and is crucial for cellular metabolism by oxidising NADH (nicotinamide adenine dinucleotide, reduced) and regenerating the NAD⁺ (nicotinamide adenine dinucleotide, oxidised) pool in the mitochondrial matrix [3]. A deficiency in complex I, caused by either the nuclear encoded genome or mitochondrial genome (mtDNA), is reportedly associated with mitochondrial disorders both in children and adults, including Leigh syndrome, cardiomyopathy and encephalomyopathies [4,5].

In addition, the link between mutations in complex I and ageing has been shown in a number of studies using different model systems. Mutations in genes encoding subunits of complex I reportedly increase reactive oxygen species (ROS) levels and extend lifespan in worms and flies by independent mechanisms [6,7]. Nuo-6 (qm200) worms, carrying a mutation in a conserved subunit of complex I, exhibit an extended lifespan by reducing oxygen consumption and decreasing complex I activities [6]. Flies with a knocked down expression of NDUFS1 (NADH: ubiquinone oxidoreductase core subunit S1), a component of complex I, exhibited a longer lifespan via an increased mitochondrial unfolding protein response and the repression of insulin signalling [7]. Other studies showed that the stability of complex I is a critical factor for the extension of lifespan of mice [8] and worms [9]. In these two studies, opposite theories are suggested. The former indicates that the instability of complex I results in reduced ROS production, which extends the lifespan in mice [8], while the latter shows that the destabilisation of the complex leads to a shortened lifespan in worms [9]. In addition, a recent study demonstrated that increased ROS production, specifically from complex I reverse electron transport, extends lifespan in flies [10].

Apart from studies of nuclear genome encoded complex I genes, mtDNA encoded complex I genes are also reportedly involved in ageing, and this is supported by several studies exhibiting an association between mitochondrial DNA polymorphisms and lifespan in different ethnic populations. These include the A variation at m.5178 in the ND2 gene (MT-ND2) accumulating in Japanese centenarians [11], and the A variation at m.9055 in the ATP6 gene (MT-ATP6) in French Caucasian centenarians [12]. More recently, a gene-wise analysis of mtDNA in a Turkish population over 90 years of age revealed that nonsynonymous mutations in complex I genes were enriched [13]. Interestingly, most of these variants exhibit protective effects against ageing and age-related diseases, such as reduced lipid levels (anti-atherogenic) [14] and protection from Parkinson’s disease [15]. In addition, variations in mtDNA encoded complex I genes are reportedly associated with age-related diseases, such as metabolic diseases (e.g., type 2 diabetes [16]) and neurodegenerative disorders (e.g., Alzheimer’s disease [17] and Parkinson’s disease [15]), in humans.

Previous studies using model organisms reveal that mutations in mtDNA encoded complex I genes controlled lifespan in worms [18] and fruit flies [19], while no studies have been conducted in mammalian models to date.

Therefore, we evaluated the impact of mtDNA encoded complex I genes on ageing in mice using conplastic mouse strains carrying different single point homoplasmic mutations in complex I genes on the same nuclear background of C57BL/6J, i.e., C57BL/6J-mt^ALR/LtJ (B6-mt^ALR) and C57BL/6J-mt^BPL/1J (B6-mt^BPL) [20]. The former carries a homoplasmic mutation in mtDNA encoded NADH dehydrogenase subunit 2 gene (mt-Nd2), and the latter has a homoplasmic mutation in the mtDNA encoded NADH dehydrogenase subunit 5 gene (mt-Nd5). We specifically selected these conplastic strains to compare because they differ only in complex I genes (Table S1).

2. Materials and Methods

2.1. Mice and Husbandry

Conplastic mouse strains C57BL/6J-mt^ALR/LtJ and C57BL/6J-mt^BPL/1J were previously generated [20]. C57BL/6J-mt^ALR/LtJ and C57BL/6J-mt^BPL/1J with a backcross of 13 to 18 (C57BL/6J-mt^ALR/LtJ) and 17 to 19 (C57BL/6J-mt^BPL/1J) generations were used for the survival study. To genotype nuclear genome of both C57BL/6J-mt^ALR/LtJ and C57BL/6J-mt^BPL/1J we used the MegaMUGA Mouse Universal Genotyping Array (77,808 SNPs) as described in Supplementary Materials, and greater than 99.9% of SNPs were identical to those of C57BL/6J (Table S2).

Mice had ad libitum access to filtered water and autoclaved pellet diet (Altromin, Eastern-Westphalia/Lippe, Germany). The animal facility was maintained at 21 °C on a 12 h light–12 h dark cycle. Mice were allocated into two study groups: longitudinal study group to evaluate lifespan, and cross-sectional study group to evaluate the mice at different ages.

For the high fat diet feeding experiment, eight female mice of C57BL/6J-mt^ALR/LtJ and C57BL/6J-mt^BPL/1J at four weeks of age were fed a high fat diet (EF D12492 (I) 60 kJ% fat, Ssniff, Soest, Germany), while three age- and sex-matched mice of each strain were fed a control diet (EF D12450B mod. LS 13 kJ% fat, Ssniff) over eight weeks.

Animal use and all protocols used in this study were approved by local authorities of the Animal Care and Use Committee (V242-7224. 122-5, and 5-1/16, Kiel, Germany) and performed in accordance with the relevant guidelines and regulations by certified personnel.

2.2. Lifespan Study and Determining Age at Death

Eighty-two female C57BL/6J-mt^ALR/LtJ, 82 female C57BL/6J-mt^BPL/1J, 99 male C57BL/6J-mt^ALR/LtJ and 94 male C57BL/6J-mt^BPL/1J were used to evaluate their lifespan, assuring a statistical power to detect a 10% difference in lifespan at a significance level of 0.05 and a power of 0.8 using G*Power [21].

2.3. Female Sexual Maturity

Vaginal patency was evaluated as previously described [22].

2.4. Plasma IGF-1 Levels

Plasma IGF-1 levels were measured by a commercially available ELISA kit (Mouse/Rat IGF-1 Quantikine ELISA Kit, R&S Systems, Wiesbaden-Nordenstadt, Germany), according to the manufacturer’s protocol.

2.5. Statistical Analysis

Survival curves were estimated using the Kaplan–Meier method, and median lifespans with 95% confidence intervals were calculated. The differences of longevity between the strains were analysed using the log-rank method, as sensitivity analysis, Peto and Peto modification of the Gehan test. The R package survival in R version 3.5 was utilized for this [23]. Assuming that a most pronounced effect is visible in female mice and our previous studies demonstrating the impact of mtDNA variants on lifespan is more prominent in females [24], the difference between strains in females was tested first. In a hierarchical way, difference between strains in males and in the sex-combined samples was tested for significance only in case that the previous test was significant at a significance level of 0.05.

Differences in glucose between strains and across time points were compared using a nonparametric analysis of variance for repeated in measures using the package nparLD in R, version 3.5. The effect of strain and age on IGF-1 was investigated in a linear regression model.

Statistical analyses for other functional studies were performed using GraphPad Prism (GraphPad Software, San Diego, CA, USA), and statistical tests used for analysis are indicated in the figure legends.

Statistical tests were performed only for descriptive purposes and descriptive p-values are reported.

3. Results

3.1. A Mutation in the mt-Nd2 Gene Results in a Shorter Lifespan in Mice

First, to evaluate the impact of mutations in complex I genes on lifespan in mice, we performed a longevity study using a large cohort of the two conplastic mouse strains carrying a mutation in mt-Nd2 (B6-mt^ALR) and mt-Nd5 and (B6-mt^BPL). The female B6-mt^ALR mice lived approximately 60 days shorter than the B6-mt^BPL mice did (median survival: 829 days in B6-mt^BPL and 772 days in B6-mt^ALR; p = 0.0400, log-rank test; Figure 1A and Table S3), while this significant difference was not observed in males or a sex-mixed analysis (p = 0.4923, and p = 0.1532, respectively; Figure 1B,C). In both strains, the incidence of spontaneous age-related diseases (i.e., tumours, ulcerative dermatitis and arthritis, Figure S1A) and the ageing score (Figure S1B) were comparable. The days of vaginal patency and plasma IGF-1 levels in differently aged females showed no difference between the strains (Figure S1C,D).

3.2. The mt-Nd2 Mutant Mice Exhibited Mitochondrial Functional Differences under Stress Conditions

Next, we investigated the functional consequence of the mt-Nd2 variant in the liver mitochondria obtained from the two conplastic mouse strains. Because a significant effect on lifespan was observed only in female mice, only female mice were used in the mitochondrial functional study exclusively. The mitochondrial OXPHOS complex activities values were normalised to the individual values of the citrate synthase (CS) activities. No difference was observed in the levels of the OXPHOS complex enzyme activities (values normalized with CS) between the B6-mt^BPL and B6-mt^ALR mice in both age groups, except that there was an age-dependent increase in complex III activity levels in both strains (descriptive p = 0.0011, young B6-mt^ALR vs. aged B6-mt^ALR; descriptive p = 0.0004, young B6-mt^BPL vs. aged B6-mt^BPL; one-way ANOVA, Figure 2A). A Western blot analysis of the liver and heart proteins prepared from the young (3–4 months of age) and aged (18–22 months of age) mice revealed unaltered levels of the OXPHOS complex subunit proteins between the strains in both age groups (Figure 2B,C). The protein levels of three other subunits of mitochondrial complex I protein levels were also evaluated, and no differences were observed between the strains regardless their age group (Figure 2C, right panel).

Primary skin fibroblasts were isolated from the B6-mt^BPL and B6-mt^ALR mice, and skin fibroblasts cell lines (B6-mt^BPL and B6-mt^ALR) were generated from these primary fibroblasts. Each cell line carried the distinct mtDNA mutations as those in parental mouse strain (Table S1). A cellular flux analysis of the conplastic fibroblast cell lines demonstrated that the levels of maximal respiration and the spare capacity in B6-mt^ALR fibroblasts exhibited a decreasing trend compared to those in the B6-mt^BPL fibroblasts (Figure 2D).

Next, primary lymphocytes were isolated from the B6-mt^BPL and B6-mt^ALR mice, and the mitochondrial superoxide levels were measured by flow cytometry using MitoSOX at the basal (i.e., immediately after the preparation) and immunologically activated (i.e., 24 h after culturing with anti-CD3/anti-CD28 antibodies) status. The activated cells from the B6-mt^ALR mice produced less mitochondrial superoxide than did those from B6-mt^BPL mice (descriptive p < 0.0001, two-way ANOVA; Figure 2E, left). The fold change of increase for superoxide production was lower in the B6-mt^ALR lymphocytes than that in the B6-mt^BPL cells (descriptive p = 0.0005, t-test; Figure 2E, right). The mitochondrial membrane potential (MMP) was measured using the same batch of primary lymphocytes used for the superoxide assay. The MMP was increased by the activation, while no difference in the levels of MMP was observed between the strains (descriptive p = 0.1457, two-way ANOVA; Figure 2F, left). When the fold change of the MMP levels upon the activation was compared, the B6-mt^ALR primary lymphocytes were less abundant than were the B6-mt^BPL lymphocytes (descriptive p = 0.0445, t-test; Figure 2F, right).

3.3. Higher Levels of Tryptophan Were Observed in the Cells Carrying the mt-Nd2 Mutation

Regeneration of NAD⁺ is one of the critical functions of mitochondrial complex I. To evaluate whether mutations in complex I affect this function, we determined the NAD⁺ and NADH levels in the liver tissues obtained from the B6-mt^BPL and B6-mt^ALR female mice. No difference was observed in the NAD⁺/NADH ratio between the strains (Figure 3A, Figure S2A). To also obtain insights into the biosynthetic pathways leading to NAD⁺, the levels of tryptophan—an essential amino acid that is degraded into NAD⁺ [25]—was determined by NMR in the skin fibroblast conplastic cell lines carrying the mt-Nd5 mutation or the mt-Nd2 mutation (Figure 3B). The levels of tryptophan in the cells carrying the mt-Nd2 mutation were identified to be higher than those in mt-Nd5 mutant cells (descriptive p = 0.0295, t-test; Figure 3B,C), suggesting the m.4738C>A mutation may have an impact on tryptophan metabolism, without altering NAD⁺ levels.

3.4. Ageing-Related Pathways and Mitochondrial Functional Pathways Are Altered in mt-Nd2 Mutant Mice

Next, we performed RNA-seq on the liver-isolated RNA obtained from the female B6-mt^BPL and B6-mt^ALR mice at the age of 3 to 4 months to elucidate the pathways involved in the mt-Nd2-related phenotypes. Of the 13,967 expressed genes, 35 genes were differentially expressed between the B6-mt^BPL and B6-mt^ALR mice (q < 0.05; Figure 4A, Data S1). The genes upregulated in the B6-mt^ALR mice included G0s2, Leap2, Usmg5, Chchd1 and ribosomal protein genes (Data S1). G0s2, a G0/G1 switch gene 2, shows an inhibitory capacity for the lipolytic enzyme adipose triglyceride lipase (ATGL) [26], and the deletion of G0s2 ameliorates high-fat diet induced body weight gain and insulin resistance [27]. Leap2 (liver-expressed antimicrobial peptide 2) was recently revealed as an endogenous antagonist of the ghrelin receptor, controlling the blood glucose levels depending upon the nutrient status [28], and Usmg5 (upregulated during skeletal muscle growth 5) is also increased in B6-mt^ALR mice. The gene encodes the protein USMG5, which is also called DAPIT (diabetes-associated protein in insulin-sensitive tissues). This protein was initially discovered in insulin-sensitive tissues of the streptozotocin-induced diabetic rats [29], and its expression is reportedly higher in cells with a highly aerobic metabolism [30]. These findings indicate that the mutation in mt-Nd2 has an influence on lipid and glucose metabolism, as well as ATP production in the respiratory chain. Chchd1 (coiled-coil-helix-coiled-coil-helix domain-containing 1) is one of the recently identified mitochondrial ribosomal proteins, MRPS37 [31], suggesting that the mt-Nd2 variant affects mitochondrially encoded protein expression as well.

Pathways upregulated in the B6-mt^ALR mice included the OXPHOS, MYC target, fatty acid metabolism, E2F target and mTORC1 signalling pathways (Figure 4B). The oncogene MYC reportedly targets genes involved in mitochondrial biogenesis, i.e., protein import, complex assembly and mitochondrial transcription/translation [32]. Recent studies showed that E2F regulates genes involved in mitochondrial functions via directly binding to their promoter regions (e.g., COX8 and CYB5-M), and through interactions with key regulatory factors of mitochondrial biogenesis, such as NRF1/2 and PGC-1 beta [33]. These pathway analysis data are in line with the phenotypes observed in the B6-mt^ALR mice, i.e., shorter lifespan (Figure 1) and altered mitochondrial functions (Figure 2). An additional pathway analysis using a different database also pointed in the same direction, i.e., the upregulation of mitochondrial bioenergetics in B6-mt^ALR mice (Figure S3A,B)

3.5. Earlier Onset of Glucose Intolerance Is Induced by High-Fat Diet Feeding in B6-mt^ALR Mice

Lastly, to investigate whether the lifespan-linked mt-Nd2 mutation also contributes to other age-related phenotypes, we induced a diet-induced diabetes model in the female B6-mt^ALR and B6-mt^BPL mice. The high-fat diet (HFD) or control diet (CD) was started when mice were 4 weeks old. The HFD-fed mice acquired more body mass than did those fed the CD, while there was no strain difference in the HFD- nor CD-fed mice (Figure 5A). After 8 weeks of HFD feeding, an intraperitoneal glucose tolerance test (IPGTT) was performed, and the levels of glucose intolerance in the B6-mt^ALR mice were higher than those in the B6-mt^BPL mice, with a different time course in both strains (main effect of the strains: descriptive p = 0.0238, interaction between strains and time: descriptive p < 0.0001, two-way nonparametric analysis of variance; Figure 5B).

The levels of glucose and insulin in the morning-fasted serum samples were comparable between the two strains in the HFD-fed (Figure 5C), CD-fed groups (Figure S4A), and in random-fed groups (Figure S4B). The fructosamine levels were also unaltered between the random-fed B6-mt^BPL and B6-mt^ALR mice (Figure S4B), confirming that the basal glucose levels are similar between the strains. The lipid parameter assays revealed that the levels of total cholesterol in the HFD-fed mice mildly increased (Figure 5D) compared with those in the CD-fed mice (Figure S4C) and random-fed mice (Figure S4D). Interestingly, the levels of total cholesterol and high-density cholesterol (HDL) were lower in the HFD-fed B6-mt^ALR mice than those in the HFD-fed B6-mt^BPL mice (descriptive p = 0.0057 and 0.0038, respectively, t-test; Figure 5D), while the levels of low-density cholesterol (LDL) were unaltered, resulting in a lower ratio of HDL to LDL in the B6-mt^ALR mice compared to that in the B6-mt^BPL mice (descriptive p = 0.0054, t-test; Figure 5D). Changes in total cholesterol and the ratio of HDL to LDL were not detected in the CD-fed groups and random-fed groups (Figure S4C,D). Despite the unchanged triglyceride levels between the HFD-fed B6-mt^ALR and HFD-fed B6-mt^BPL mice, we observed that HFD-fed B6-mt^ALR mice exhibited lower levels of free fatty acid than did the HFD-B6-mt^BPL mice (descriptive p = 0.0115, t-test; Figure 5C), but this did not occur in the CD-fed mice (Figure S4C), suggesting impaired beta oxidation in the B6-mt^ALR mice under metabolic stress.

Cellular flux analysis was assessed in the primary hepatocytes isolated from the mice fed with HFD over 8 weeks, indicating the lower levels of basal respiration (descriptive p = 0.0389, t-test; Figure 5E) and OXPHOS-linked ATP levels (descriptive p = 0.0446, t-test; Figure 5F) in the cells prepared from the B6-mt^ALR mice than those in the cells from the B6-mt^BPL mice.

We also observed a mildly delayed glucose clearance in the CD-fed B6-mt^ALR compared to that of the CD-fed B6-mt^BPL mice again with a different time course in both strains (main effect of the strains: descriptive p = 0.0062, interaction between the strains and time: descriptive p = 0.0017, two-way nonparametric analysis of variance; Figure S4E). Furthermore, two independent IPGTT experiments, using regular chow-fed B6-mt^BPL and B6-mt^ALR mice, demonstrated a pattern of glucose clearance that was similar to that of the CD-fed mice (main effect of the strains: descriptive p = 0.3062, interaction between the strains and time: descriptive p = 0.0056, two-way nonparametric analysis of variance, Figure S4F), confirming that the impaired glucose metabolism was caused by the mutation in mt-Nd2. In line with this result, an indirect calorimetric cage analysis of the regular-chow-fed B6-mt^BPL and B6-mt^ALR mice showed lower levels of respiratory exchange ratio (RER) in the B6-mt^ALR mice compared with those in the B6-mt^BPL mice, indicating the preference of fat to glucose as an energy source in the B6-mt^ALR mice (descriptive p = 0.0382, t-test, Figure S4G). We did not observe any difference in energy expenditure, locomotor activity, or food intake levels other than water intake (Figure S4H–K).

4. Discussion

In the present study, we presented experimental evidence that variations in mtDNA encoded complex I genes, i.e., mt-Nd2 and mt-Nd5, differentially affected lifespan and metabolic phenotypes in mammals. Our results are in line with previously reported studies using Drosophila carrying a mutation in the ND2 gene (ND2^del1), i.e., shorter lifespan as well as impaired fat storage, apart from Leigh syndrome-like (or spontaneous) neurological dysfunction [19,34]. These studies suggested that ND2 mutations led to the shorter lifespan.

Mitochondrial functional studies using liver-mitochondria, cellular and tissue proteins did not exhibit major differences at the basal levels between the B6-mt^ALR and B6-mt^BPL mice. In fact, the impact of the mt-Nd2 variant on the complex I activity was minimal, with only a tendency of reduced activity in young B6-mt^ALR mice compared to age-matched B6-mt^BPL mice. Other OXPHOS complexes activities (complex III, IV and V; values were normalised to the individual CS activity) were also comparable between the two strains. At the same time, when normalised to complex I activity, the activity levels of these three OXPHOS complexes exhibited a tendency towards higher levels in B6-mt^ALR mice compared to those in B6-mt^BPL mice (CIII/CI, 3.76± 0.3156 in B6-mt^BPL vs. 5.028± 0.4504 in B6-mt^ALR, descriptive p = 0.0396; CIV/CI, 2.652 ± 0.4682 in B6-mt^BPL vs. 3.139 ± 0.5759 in B6-mt^ALR, descriptive p = 0.5340; CV/CI, 3.142 ± 0.3037 vs. 4.735 ± 0.5585, descriptive p = 0.0265; unpaired t-test), suggesting that these OXPHOS complexes increase their activities to compensate the mildly reduced complex I activity caused by the mt-Nd2 variant. This observation may explain why B6-mt^ALR mice showed a mild increase of superoxide levels and slightly higher levels of mitochondrial membrane potential at the basal levels, but no large statistically significant differences compared to B6-mt^BPL mice. However, once the cells received additional stimuli, i.e., immunological- (for lymphocytes) or metabolic stress (for hepatocytes), the levels of oxygen consumption, OXPHOS-linked ATP production and mitochondrial superoxide were significantly lower in the B6-mt^ALR mice than those in the B6-mt^BPL mice. This effect is likely caused by the lower levels of spare capacity and maximal respiration in the B6-mt^ALR mice. In addition, glucose and lipid metabolism were skewed in this mouse strain. These alterations potentially resulted in a shorter lifespan and a higher susceptibility to diet-induced glucose intolerance in the B6-mt^ALR mice. Interestingly the RNA-seq data pointed that the pathways involved in the mt-Nd2 mutant mice were mitochondrial pathways, e.g., OXPHOS pathways and metabolic pathways as well as the ageing-related mammalian target of rapamycin (mTOR) pathway.

Another potential functional consequence of the mt-Nd2 variant in the B6-mt^ALR mice was the altered levels of tryptophan, but not NAD⁺ levels. Tryptophan is an essential amino acid that is metabolised by the kynurenine pathway through a series of metabolic reactions, and consequently NAD⁺ is synthesised [35]. Tryptophan is only obtained through dietary intake, and no difference was observed in the levels of food intake between the strains in our indirect calorimetric cage data. This result suggests that the higher levels of tryptophan were due to the intrinsic metabolic alteration in the B6-mt^ALR mice. A previous study demonstrates that tryptophan induces the phosphorylation of the mTOR, and accelerates non-alcoholic fatty liver disease in mice [36], suggesting that the higher levels of tryptophan may contribute to shorten the lifespan in the B6-mt^ALR mice by activating the mTOR pathway, which is in line with our RNA-seq data.

Interestingly, the m.4738C>A mutation, which is carried by the B6-mt^ALR mice, causes the amino acid substitution of leucine to methionine at the 276th peptide (Leu276Met). The 276th peptide of human ND2 is also leucine, and this residue is known as the binding site for 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (https://www.ebi.ac.uk/pdbe/entry/pdb/5xtc/protein/17), suggesting that the mutation could potentially affect the binding capacity and/or affinity to glycerolipids, which are the components of the mitochondrial inner membrane. It is tempting to speculate that the mt-Nd2 mutation alters the structure of mitochondrial complex I, or supercomplex, which may change its function, particularly under stress conditions, without influencing the protein levels of complex I subunits. A study to confirm the structural alteration by the mt-Nd2 mutation (m.4738C>A), e.g., complex I assembly analysis, will be performed in the future. We also evaluated the potential impact of m.11902T>C variant in B6-mt^BPL mice on the protein structure using the same database, however, the amino acid position affected by the variant (Phe54Leu) has not been reported as a potential binding site for any proteins.

In humans, variants in the MT-ND2 gene, particularly m.5178C>A, exhibit a link with longevity [11] as well as lipid metabolism [14] and the incidence of age-related diseases, e.g., Parkinson’s disease [37]. For this particular variant, the A allele is associated with extreme longevity and is protective for the abovementioned diseases, suggesting that lifespan-associated mtDNA variants are also responsible for age-related disease susceptibility. This effect was particularly observed in female mice. This is consistent with the above-discussed previous report of Drosophila carrying the ND2 mutation, i.e., only female ND2^del1 mutant flies exhibited the behavioural phenotype [19]. However, we cannot exclude the possibility that there may be other phenotypes we did not investigate in this study, which this specific variant may affect only in males. For example, we recently reported that mice harbouring another maternally inherited natural mtDNA variant, m.15124A>G in the mitochondrially encoded cytochrome b gene (mt-Cytb) in complex III, developed spontaneous middle-aged obesity only in males [38]. In the present study, we demonstrated that the B6-mt^ALR mice, carrying the mt-Nd2 (m.4738C>A) variant, showed a slightly but significantly shorter lifespan and a higher susceptibility to diet-induced glucose intolerance than did the B6-mt^BPL mice, which carry the wild type C allele at m.4378 in mt-Nd2 and the mutant C allele at m.11902 in mt-Nd5. While the effect of each SNP appears to be variable in different species, it is clear that variations in the ND2 genes are linked with ageing and metabolic conditions both in mice and humans.