In our current study, we described and provided evidence of positive selection acting on the two mitochondrial coding genes ND4 and Cytb in eight hare species (genus Lepus) from China using a wide range of selection tests, protein structure modeling and analysis of mutation effects. Positive selection was earlier recorded on different mitochondrial OXPHOS genes in a wide range of animal species (see [49] for an overview). Particularly, several studies have focused on mtDNA evolution driven by natural selection in hares and jackrabbits (genus Lepus) during the last years. The analyses of eleven mitogenomes of different hare species of temperate and arctic origins by Melo-Ferreira et al. [25] have suggested positive selection in several codons of genes of the mtOXPHOS complexes. However, the structure and the physicochemical properties of the encoded proteins seemed to be not affected by these amino-acids substitutions. The second evidence of occurrence of positive selection on mtOXPHOS genes was observed in ATP6 and ND2 sequences of hares (Lepus capensis) from Tunisia where they are continuously distributed across a steep ecological gradient and exhibit significantly varying ATP6 and ND2 protein frequencies despite high gene flow in putative neutrally evolving markers [9]. The positive selection signals were interpreted as reflecting adaptation of those hares to the different environmental conditions along the ecological cline in Tunisia [9]. The same two genes studied in 22 hare species distributed across the whole world [10] showed also positive selection with a significant climate effect for ND2 protein variants. Recently, Stefanović et al. [26] demonstrated that only one codon position has evolved under positive selection in the NADH dehydrogenase subunit 6 (mtND6) gene in brown hares (Lepus europaeus) from Europe and the Middle East. The authors suggested that two (D and F) among all observed protein variants were significantly favored under certain precipitation conditions, as proved by statistical models.
Given that the oxidative phosphorylation process produces 95% of cellular energy, it is not surprising that genes encoding for OXPHOS subunits are under adaptive selection. Giannoulis et al. [5] suggested that variations in these genes would directly influence the metabolic performance which may in turn affect the fitness of an organism. Generally, variation in mitochondrial OXPHOS genes may convey a signal of adaptation to environmental, particularly climatic, conditions (e.g., 6, 8–10, 25, 45, 46]. We have used several molecular-statistical approaches to assess the importance of natural selection in the evolution of the studied three mtDNA subunits in hares from China and to disentangle potential effects of climate conditions and evolutionary history of the species on selection acting on these genes. Our site and codon model results (Table 2) indicated that the ND4 and Cytb genes in the studied hares are globally evolving under negative constraints, i.e., purifying selection, with a small percentage of codons evolving under neutrality or positive selection, reinforcing their crucial and conserved role for the body energy production in mammals. This is also in agreement with the general tendency of the mt genome evolution in vertebrates (see e.g., [25]) where several studies identified purifying selection as the predominant force shaping the evolution of mtDNA with only few sites and loci under positive selection [25, 50–52]. Indeed, Tomasco and Lessa [53] suggested that, due to the functional importance of mitochondrial genes, purifying selection would be the dominant force in their evolution, preventing fixation of detrimental mutations.
Currently, evidence of positive selection was detected only for the ND4 and Cytb genes, but not for the COX1 gene. Positive selection was indicated by both the site model implemented in CODEML, the site specific tests implemented in Datamonkey (FEL, SLAC, MEME and FUBAR), and the Tree-SAAP analyses. Overall, ten codons were inferred to be under positive selection for both genes as suggested by more than one of the tests used in this study. Such positively selected site variation can be striking in both subunits as such amino acid changes might be critical for the optimal molecular function of ND4 as electron transporters, and as a structural role between the membrane-embedded and peripheral arms of the complex I [11], as well as for the catalytic activity (cytochrome c reduction) of the Cytb which is the only mtDNA-derived subunit of Complex III. Notably, among the currently identified positions under positive selection, positions ND4-29, ND4-187 and ND4-246 were also suggested to be under positive selection in the (much smaller) sample of hare sequences studied by Melo-Ferreira et al. [25].
We have placed special attention to the candidate sites for positive selection in order to assess the potential impact of the observed amino acid substitutions considering their location relative to known functional domains of the proteins and the physicochemical properties of the amino acid, such as size and charge. Sixty percent of the sites under positive selection were located in the transmembrane regions. Moreover, among the positively selected sites, we observed distinct amino acid substitutions at the sites ND4-101, ND4-305, and Cytb-194, which are suggested to be lined up along the proton translocation channel. Indeed, the amino acids located in the transmembrane domain of all OXPHOS complexes were suggested to play essential structural and functional roles related to the proton transport across the membrane [54–56]. Subunits ND2, ND4, and ND5 of the mammalian OXPHOS complex I were suggested to be proton-pumping devices which are related to Na+/H + antiporters of the Mrp family [11]. In cytochrome b (Complex III), the transmembrane domain is often functionally conserved, being involved in the creation of the proton gradient and the transfer of electrons to Complex IV [54]. Consequently, for mammals and other vertebrates, mutations in the mtDNA subunits may interfere with the efficiency of the proton-pumping process and could hinder or improve the proton translocation [11, 58]. These domains are less variable and likely constrained by stronger purifying selection, so amino acid replacements in these domains may suggest a change in OXPHOS protein function that could be subject to positive selection [11, 58–60].
The analyses of the vibrational entropy change upon mutation by the DYNAMUT software identified three amino acid replacements in ND4 and one replacement in Cytb that affect protein stability. On the other hand, among all positively selected sites, two in ND4 were suggested to be deleterious as indicated by the PROVEAN software. Notably, for the sites ND4-425 and Cytb-356 disease related mutations (LHON disease) were described in humans [61, 63]. Azevedo et al. [64] suggested that the deleterious effect of a mutation can be compensated by a second-site interacting residue which explains why mutations that are deleterious in some species are tolerated in phylogenetically related lineages, rendering evidence that those mutations are, by all means, only deleterious in the species-specific context. Our results are concordantly indicating that amino acid changes at specific sites are having a strong effect on protein function. Such protein alterations (i.e., change in stability and disease liability) at amino-acid positions that were conserved over large evolutionary timescales might be slightly deleterious or/and counteracted by compensatory changes in the nuclear-coded mitochondrial proteins [63] or may truly reflect adaptation [65, 66].
Hypothesizing that positively selected sites are relevant for adaptation to different climate conditions, we applied statistical models to test this hypothesis. Our PCA of the climate variables were summarized successfully in four (statistically independent) climate factors, and three of them (factor 1, 2, and 4) could be successfully interpreted in climatological terms. Our results showed that two and three climate factors had a significant effect on the occurrence of certain protein variants of ND4 and Cytb, respectively (i.e., amino acid changes at the positively selected sites in ND4 and Cytb). This suggested adaption to climatic/environmental conditions of the OXPHOS genes in the currently studied hares from China. Concordantly, our TreeSAAP analysis for both ND4 and Cytb showed that amino acid changes altered mainly the equilibrium constant (ionization of COOH) property (Table 1). This property was suggested to influence the protein efficiency reducing ROS (reactive oxygen species) production while increasing individual longevity [67]. Romero et al. [68] suggested that alterations in the equilibrium constant allow organisms to better cope with abiotic stress conditions (which could be imposed by ambient climatic conditions). Indeed, the activation of the antioxidant metabolism reducing a ROS excess has been linked to desiccation tolerance in the algae Mastocarpus stellatus and Porphyra columbina occurring in the upper intertidal zone [69]. Since abiotic stress in general is linked to metabolic activity and ROS production [68, 70], this directly affects the distribution of a biological species and its success to occupy new ecological niches. Moreover, increased metabolic efficiency has been also related to the capability of diverse animals to invade new ranges [68] and George and Blieck [71] detected significant changes in the equilibrium constant (ionization of COOH) property affecting similar regions in the genes of amphibians, lungfishes, and coelacanths which was suggested as an adaptation to increased oxygen levels and changing metabolic requirements.
Positive selection on diverse mt OXPHOS genes has been suggested by in silico analyses in a wide range of animal species in the contexts of varying ecological and climate conditions. However, only few experimental studies were able to assess the adaptive value of mtDNA variations. A clinal variation of mitochondrial mitotypes along temperature gradients and associations between mitotype and climate have been observed for numerous metazoan species, including humans [18, 45]. Experiments in invertebrates have demonstrated directly that different mitotypes can alter temperature tolerance [72, 73] and that the mitotype was associated with adaptation to temperature in natural environments [18, 74]. Recently, Lajbner et al. [17] used laboratory-based experimental evolution in the fruit fly, Drosophila melanogaster, to test whether thermal selection can shift population frequencies of two mtDNA haplogroups whose natural frequencies exhibit clinal associations with latitude along the Australian east-coast. They found experimental evidence that the thermal regime in which the laboratory populations were maintained drove changes in haplogroup frequencies across generations. The authors suggested that adaptation to novel environments might routinely involve selection of mitochondrial polymorphisms that optimize thermal performance in those environments, and this process might be relevant to all metazoans, both poikilothermic and homeothermic, and indeed to all eukaryote life [17].