Linking the mitochondrial genotype to the organismal phenotype


  • J. W. O. BALLARD,

    1. School of Biotechnology and Biomolecular Sciences, University of New South Wales, Sydney, NSW 2052, Australia
    2. Evolution & Ecology Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia
    Search for more papers by this author
  • R. G. MELVIN

    1. School of Biotechnology and Biomolecular Sciences, University of New South Wales, Sydney, NSW 2052, Australia
    Search for more papers by this author
    • Present address: Department of Biology, University of Minnesota, Duluth, 280 SSB, 1035 Kirby Drive, Duluth, MN 55812, USA

J. W. O. Ballard, Fax: (+61) 2 9385 1483; E-mail:


One of the grand challenges of the postgenomics era is to mechanistically link the genotype with the phenotype. Here, we consider the link between the mitochondrial genotype and the organismal phenotype that is provided by bioenergetic studies of the electron transport chain. That linkage is pertinent for the fields of molecular ecology and phylogeography as it tests if, and potentially how, natural selection can influence the evolutionary and demographic past of both populations and species. We introduce the mitochondrial genotype in terms of mitochondrial-encoded genes, nuclear-encoded genes that produce structural proteins imported into the mitochondria, and mitochondrial DNA–nuclear interactions. We then review the potential for quaternary structure modelling to predict the functional consequence of specific naturally occurring mutations. We discuss how the energy-producing reactions of oxidative phosphorylation can be used to provide a mechanistic biochemical link between genotype and phenotype. Experimental manipulations can then be used to test the functional consequences of specific mutations in multiple genetic backgrounds. Finally, we examine how mitochondria can influence the organismal mitochondrial phenotype using the examples of lifespan, fertility and starvation resistance and discuss how mitochondria may be involved in establishing both the upper and lower thermal limits of organisms. We conclude that mitochondrial DNA mutations can be important in determining aspects of organism life history. The question that remains to be resolved is how common are these adaptive mutations?


Mitochondrial DNA (mtDNA) has proven invaluable in the fields of molecular ecology and phylogeography for inferring the evolutionary and demographic past of both populations and species. In contrast, the role that mitochondria have in shaping the demographics of populations and species is poorly understood. Here, we suggest that adaptive mtDNA mutations have the potential to be important in determining aspects of the life history of an organism but their frequency in natural populations remains to be determined.

The goal of this review is to consider the link between the mitochondrial genotype and the organismal phenotype. Historically, bridging the genotype and phenotype has been notoriously difficult because simple genetic changes can result in complex phenotypic changes. Further, many genetic changes do not influence the phenotype, further obscuring the link. Here, we propose that bioenergetic studies can provide this link. As stated by Nicholls & Ferguson (2002) all biochemical reactions involve energy changes so the term ‘bioenergetics’ can validly be applied to the whole of life sciences. Bioenergetics as a discipline rose to prominence in the 1950s with the search for the mechanism(s) of energy production such as the coupling of substrates with ‘uphill’ reactions. Pathways such as the synthesis of adenosine triphosphate (ATP) from adenosine diphosphate (ADP) and inorganic phosphate (Pi) or the accumulation of ions across the inner mitochondrial membrane were discovered (Nicholls & Ferguson 2002).

We begin the review by asking the question ‘Is adaptive evolution of mtDNA negligible or not?’ We introduce the mitochondrial genotype in terms of mitochondrial-encoded genes, nuclear-encoded genes that produce structural proteins imported into the mitochondria, and mtDNA–nuclear (mitonuclear) interactions. Next, we suggest that quaternary structure modelling combined with biochemical analyses can quantify the bioenergetic efficiency of the electron transport chain. Carefully controlled experimental manipulations can then determine the functional consequences of specific mutations. Finally, we consider how mitochondria can influence aspects of the organismal phenotype using the examples of lifespan, fertility and starvation resistance and discuss how mitochondria may be involved in establishing both the upper and lower thermal limits of organisms. We conclude that mtDNA mutations have the potential to be important in determining aspects of the life history of an organism and a quantifiable subset of mutations is adaptive.

Is adaptive evolution of mtDNA negligible or not?

In independent and complementary reviews, Nachman (1998) and Rand & Kann (1998) first tested whether mtDNA is evolving in a manner consistent with a strictly neutral equilibrium model in a broad range of organisms. Both studies investigated the published data up to 1998, and screened for data sets that included multiple individuals from single species and sufficient intraspecific variation so that statistical tests could be employed. The results presented in both studies suggested that slightly deleterious mutations accumulate within species but do not go to fixation among them. Today, few biologists question that deleterious and slightly deleterious mutations are observed in mtDNA and that these changes can be interpreted as inconsistent with a strictly neutral equilibrium model.

If mtDNA is largely influenced by purifying selection, non-neutral mtDNA may alter some demographic or phylogenetic analyses. However, these alterations are expected to be relatively minor and generally quantitative and not qualitative in natural populations. In humans, deleterious mtDNA mutations are well known to cause disease. In 1988, mtDNA mutations were shown to cause three diseases: Leber’s hereditary optic neuropathy (Wallace et al. 1988), chronic progressive external ophthalmoplegia and Kearns–Sayre syndrome (Holt et al. 1988). These mutations are, however, considered to be outside of the adaptive norm and would probably be purged from humans without modern medicine. The real question is how common is adaptive evolution of mtDNA in natural populations?

If mtDNA evolution is adaptive, such that the interpretations of population histories are biased, deviations should be significant and both quantitative and qualitative. A variety of studies have concluded that mtDNA evolution is adaptive. Galtier et al. (2009b) reappraised the use of mtDNA as a marker of molecular diversity and concluded that the adaptive mtDNA evolution within species is not negligible. Bazin et al. (2006) compiled three within-species polymorphism data sets: an allozyme data set (912 species), a nuclear sequence data set (417 species) and a mitochondrial sequence data set (1683 species). They calculated the average genetic diversity in eight largely represented animal groups. The allozyme and nuclear data sets yielded highly similar results. The mtDNA data diversity, however, was highly variable between species within groups, but remarkably homogeneous among groups. Notably, insect and mollusc species did not appear more polymorphic, on average, than mammals or birds, contradicting our prior beliefs about relative population sizes in these groups. The authors considered, and rejected, the potential for variations in mtDNA mutation rate and purifying selection against deleterious mutations to generate the observed pattern. Rather, Bazin et al. (2006) suggested that selection acting on mtDNA contributes to homogenization of the average diversity among groups, in agreement with the genetic draft theory (Gillespie 2000). The authors concluded that mtDNA appears to be anything but a neutral marker and probably undergoes frequent adaptive evolution.

We suggest that mtDNA cannot be assumed to evolve in a manner consistent with a strictly neutral equilibrium model of evolution, without statistically testing this assumption in molecular ecology studies. A battery of useful statistical tests is reviewed elsewhere (Ballard & Whitlock 2004) and is available in freeware such as DnaSP (Rozas & Rozas 1999) ( and Arlequin ( (Excoffier et al. 2005). If purifying selection is the predominant evolutionary force, and nuclear markers corroborate the story, mtDNA can be a very useful marker for molecular ecology. If, on the other hand, mtDNA evolution is rejected in the direction of positive selection, mtDNA will not be a robust marker to interpret demographic and evolutionary history. In this latter case, mtDNA may give insight into the adaptive processes that have shaped the evolutionary history of species. Here, we propose that linking mitochondrial bioenergetics with experimental manipulations can provide a testable mechanistic link between adaptive mitochondrial mutations and the organismal phenotype.

Mitochondrial genotype

In this section, we consider variability in mtDNA-encoded genes, variability in nuclear-encoded genes that produce proteins imported into the mitochondria and variability in mitochondrial and nuclear interactions. For the purpose of this review, we restrict our discussion to the union of the set of genes located on the mtDNA with the set of nuclear genes that encode the set of structural proteins involved in oxidative phosphorylation (OXPHOS). mtDNA contains 37 genes, all of which are essential for normal mitochondrial function. In addition, about 75 nuclear genes encode the set of structural proteins that are translocated to the mitochondria and are involved in mitochondrial OXPHOS. In this article, we do not discuss all the nonstructural factors that are required at the transcriptional, translational or post-translational level or the gene products involved in the assembly of the OXPHOS complexes (Smeitink et al. 2001). These nonstructural factors probably influence mitochondrial bioenergetics in natural populations and are well known to cause human disease (Perez-Martinez et al. 2008).

The apparent simplicity of the mitochondrial molecule has led some experimentalists to postulate that gene frequencies are governed primarily by migration and genetic drift (reviewed in Ballard & Whitlock 2004). However, factors other than genetic drift are thought to be important determinants governing the fate of mutations. While mtDNA may recombine in some cases (Awadalla et al. 1999; Eyre-Walker & Awadalla 2001; Innan & Nordborg 2002), the lack of normal recombination in metazoan mitochondria means that each genome has a single genealogical history and all genes will share that history. Any evolutionary force acting at any one site will equally affect the history of the whole molecule. Thus, the fixation of an advantageous mutation by selection, for example, will cause the fixation of all other polymorphisms by a process known as genetic hitchhiking (Maynard-Smith & Haigh 1974; Kaplan et al. 1989) (Fig. 1). In the case of an advantageous mutation, all linked changes will increase in frequency. Hitchhiking of mtDNA with a maternally transmitted parasite, such as the alpha proteobacteria Wolbachia, may also be quite common, particularly in insects (Hurst & Jiggins 2005). However, in the latter case the specific mtDNA haplotype that increases in frequency may simply be randomly associated with the initial infection and may not reflect any selective advantage of that particular molecule. In both cases, however, a loss of mtDNA diversity is expected (Fig. 1).

Figure 1.

 Genetic hitchhiking in the completely linked mtDNA molecule. Horizontal lines represent six mtDNA molecules sampled from a population. Grey circles represent neutral and slightly deleterious mutations. The region inside the box represents the portion of the mtDNA molecules surveyed by sequencing. The thick black arrow represents passage of time. Consider the case of a selectively advantageous mutation depicted as black circle. Over time the selectively advantageous mutation can rise to high frequency in the population and will carry the associated neutral or slightly deleterious mutation with it. As an alternative, consider the influence of a maternally inherited parasite. In this case, the black circle represents the infection. If the parasite causes an advantage, the specific mtDNA haplotype that it was first associated with can increase in frequency. In this latter case, the increase in frequency of that mtDNA type does not reflect any selective advantage of that particular molecule but rather the combined effect of the mtDNA and the parasite. In both cases, however, a loss of mtDNA diversity is expected.

Variability in mtDNA-encoded genes

In metazoans, the high mutation rate of the mitochondrial genome may be caused by a more mutagenic intracellular environment or by low efficiency of DNA repair pathways. This results, for example, in a mean mtDNA divergence at synonymous sites between species of vertebrates that is 5–50 times higher than in the nuclear genome (Lynch 2007). In humans, the mitochondrial mutation rate may be even higher than interspecific divergences suggest. Human pedigree studies suggest a 10-fold higher rate than divergence-based estimates, based on the appearance of de novo mtDNA variants (Howell et al. 2003). This high rate of mutations is probably influenced by both somatic and germline changes. Specifically, somatic changes that occur in the grand maternal ancestor can, in some cases, be inappropriately interpreted as germline mutations. This has the potential to artificially elevate the mutation rate based on pedigree studies.

Several workers have made direct estimates of the mtDNA mutation rates in birds (Lambert et al. 2002), nematodes (Denver et al. 2000), humans (Howell et al. 1996, 2003; Parsons et al. 1997) and the fly Drosophila (Haag-Liautard et al. 2008). In the fly, Haag-Liautard et al. (2008) directly estimated the mutation rate by scanning the mitochondrial genome of lines that had undergone ∼200 generations of spontaneous mutation accumulation. They detected a total of 28 point mutations and eight insertion–deletion mutations, yielding an estimate for the single-nucleotide mutation rate of 6.23 × 10−8 per site per fly generation. This was in the virtual absence of effective natural selection. They showed that the mitochondrial mutation rate was about ten times higher than the nuclear DNA mutation rate. In a mutation accumulation study, a direct estimate of the mtDNA mutation rate was also carried out in the worm Caenorhabditis. The estimate for the single-nucleotide mutation rate was also substantially higher than the corresponding estimate for the nuclear genome (Denver et al. 2000). However, of the 16 single-nucleotide mutations detected, only four would increase A/T content. The corresponding figure for the fly was 24/28. In this respect, therefore, the outcomes of the two studies were quite different, suggesting that selection and mutation may be of different relative strengths in the two species.

Variability in nuclear-encoded genes that produce proteins imported into the mitochondria

Analysis of nuclear-encoded genes involved in OXPHOS is a fertile area for investigating selection and evolutionary processes because the bioenergetic function of the genes is well established and sex-specific affects have been documented. Despite the continued movement of genes from the mitochondrial to the nuclear genome in many species (Bensasson et al. 2001), functional nuclear-encoded OXPHOS genes appear to be less likely to form duplicates (or preserve them) than other nuclear genes. De Grassi et al. (2005, 2008) have studied the evolution of gene families encoding subunits of the five OXPHOS complexes, which are fundamental and highly conserved in all respiring cells. The authors concluded that the average size of all OXPHOS complexes and accessory proteins is smaller than the average size of nuclear gene families in both vertebrates and invertebrates. They suggested that this result is in line with the hypothesis that duplicated genes coding for multi-protein complexes are negatively selected, since their expression may influence the complex formation and thereby fitness of the organism.

Grossman et al. (2001) investigated changes in the biochemical machinery for aerobic energy metabolism. They found that protein subunits of two of the electron transfer complexes (i.e. complex III and complex IV) and cytochrome c (i.e. the protein carrier that connects them) have all undergone a period of rapid protein evolution in the anthropoid lineage that ultimately led to humans. Indeed, subunit IV of complex IV (cytochrome c oxidase; COX) provides a strong example of positive selection. The rate of subunit IV evolution accelerated in our catarrhine ancestors in the period between 18 and 40 Ma and then decelerated in the descendant hominid lineages. This pattern of rate changes is strongly suggestive of positive selection followed by purifying selection acting against further changes.

A number of studies have attempted to document the intraspecific genetic variation in nuclear-encoded genes that produce proteins imported into the mitochondrion (Ballard et al. 2002; Willett & Burton 2004; Melvin et al. 2008; Willett & Ladner 2009). To investigate the genetic variability at a nuclear-encoded OXPHOS gene with that observed in the mtDNA, Willett & Ladner (2009) examined sequence variation in both the rieske iron–sulphur protein (RISP, nuclear) and cytochrome b (mtDNA) in Tigriopus californicus. The RISP protein interacts structurally with cytochrome b in complex III and functionally through its iron–sulphur redox centre (which is involved in the transfer of electrons in this complex) (Saraste 1999). The amino acid sequence of RISP was highly conserved across sites and regions in this data set. In contrast, the cytochrome b gene was highly variable. These results strongly suggest that structural co-evolution between these two proteins is not occurring within a region or between relatively closely related regions. These results do not rule out the possibility of coadaptation with other subunits of complex III.

Ballard and colleagues have investigated the genetic variation in nuclear-encoded OXPHOS genes in Drosophila simulans. Ballard et al. (2002) compared the variability in a portion of the NADH: ubiquinone reductase 75 kD subunit precursor (ND75) with intron 1 of the alcohol dehydrogenase-repeated (Adhr) locus. ND75 forms part of the NADH dehydrogenase complex in the inner mitochondrial membrane while Adhr has considerable variation and the pattern is consistent with a neutrally evolving locus (Sumner 1991). Results of HKA tests (Hudson et al. 1987) were consistent with the hypothesis that ND75 and Adhr are evolving in a manner consistent with a neutral equilibrium model of evolution. The HKA test is a conservative test of an equilibrium-neutral model’s prediction that polymorphism within species and divergence between species will be positively correlated. Melvin et al. (2008) analyzed the variation within the nine reported nuclear-encoded complex IV genes, and their isoforms, from 12 D. simulans strains and did not find any compelling evidence for selection. Investigating the selective forces that influence the evolution of nuclear-encoded genes that produce proteins targeted to the mitochondria is of general interest because such studies have the potential to explore functionally significant interactions.

Variability in mitochondrial and nuclear interactions

The evolutionary forces responsible for the movement of genes from the mitochondrion to the nucleus are better understood in plants than in animals, but the movement of genes appears to be an ongoing process (Adams et al. 2002; Bergthorsson et al. 2003). Brandvain & Wade (2009) simulated the transfer of mitochondrial genes to the nucleus in clonal and self-fertilizing plants. They showed that the relative mutation rates of mitochondrial and nuclear genes and epistatic selection influence mitochondrial-to-nuclear mutation rates. If mitochondrial mutational pressure drives the functional transfer of mitochondrial genes to the nucleus, then plants with higher mitochondrial mutation rates are expected to have transferred more mitochondrial genes to their nuclei than plants with lower mitochondrial mutation rates. As suggested previously, plausible explanations for higher mutation rates are a more mutagenic mitochondrial environment or a low efficiency of DNA repair pathways in mitochondria.

Sex-by-genotype interactions for mtDNA type, cytoplasms and X chromosomes are an emerging feature of mitochondrial evolution. In most diploid sexual species, the patterns of joint cytonuclear chromosomal transmission are different for the X chromosome compared with the autosomes. Where the female is the homogametic sex, half of the copies are cotransmitted through the female with the mtDNA. For the X chromosomes, however, two-thirds of the copies within a mating pair are cotransmitted through the female with the mtDNA. This difference in the patterns of cotransmission for X chromosomes and autosomes motivated Rand et al. (2001) to re-examine models of cytonuclear fitness interactions that were based on autosomal loci. They developed a model of joint transmission of X chromosomes and cytoplasms and showed that cytonuclear polymorphisms can be maintained by selection on X–cytoplasm interactions. The study further demonstrated significant sex-by-genotype interactions for mtDNA type, cytoplasms and X chromosomes.

Mitochondrial bioenergetics

Currently, it is not feasible to ‘do genetics’ on mtDNA, because the recombination rates are very low and genetic hitchhiking makes it difficult to determine those specific mutations that are slightly deleterious or slightly advantageous. We propose that quaternary structure modelling combined with biochemical analyses can predict and then quantify the bioenergetic efficiency of the electron transport chain and of specific complexes in the chain.

Mitochondria are the Dr Jekyll and Mr Hyde of life and death. As Dr Jekyll, they are the ‘powerhouse of the cell’ and produce about 80% of our cells’ energy in the form of ATP, which is essential for reproduction. As the misanthropic Mr Hyde, mitochondria produce reactive oxygen species (ROS) as a by-product of normal metabolism. ROS damage DNA, cell membranes and lipids, decrease bioenergetic efficiency as we age, and are a major contributor to cell and organismal death (Harman 1956, 1998). Certainly, biochemists have developed sophisticated strategies for identifying the functional consequences of specific human mitochondrial mutations that cause disease (Wallace 1992; Ruiz-Pesini et al. 2000; Smeitink et al. 2001; Taylor & Turnbull 2005; Kirby et al. 2007). Currently, however, the importance of mutations in natural populations remains underexplored. Significant questions that critically need study are how many of these mutations are adaptive and how can they be identified?

Quaternary structures

The structures of complexes II, III and IV are known in great detail through X-ray crystallography analyses (reviewed in Rich 2003). Parts of complexes I and V have also been resolved by this experimental strategy (Dickson et al. 2006; Sazanov & Hinchliffe 2006). Despite being considered of great functional importance (Dudkina et al. 2008), their supramolecular organization is still largely unknown.

Complexes 1–IV generate the proton gradient in the intermembrane space of mitochondria. Complex I, or NADH dehydrogenase, is the main entrance point of electrons and is the largest complex of the respiratory chain. It catalyzes the transfer of two electrons from NADH to ubiquinone and translocates four protons across the inner mitochondrial membrane. The extensively studied bovine heart mitochondria complex I is made of 45 different protein subunits (Carroll et al. 2003). Fourteen homologous ‘core subunits’ are present in all organisms, including prokaryotes and chloroplast complex I, and constitute the ‘minimal enzyme’. Additional subunits have been depicted as ‘peripheral’. Structural studies describe complex I as an L-shaped molecule consisting of a large membrane arm and hydrophilic arm that protrudes into the matrix (reviewed in Yagi & Matsuno-Yagi 2003).

Complex II, or succinate dehydrogenase, represents an alternative entrance point of electrons to the respiratory chain. This complex transfers electrons from succinate to ubiquinone and directly connects the Krebs cycle to the respiratory chain. Complex II consists of four protein subunits and contains a bound flavin adenine dinucleotide cofactor, iron–sulphur clusters, and a heme group that does not participate in electron transfer to coenzyme Q. In metazoans, complex II genes reside in the nucleus, however, genes encoding subunits of respiratory complex II are present in the mtDNA of three phylogenetically distant eukaryotes (Leblanc et al. 1995; Burger et al. 1996).

The central component of the OXPHOS system, cytochrome c reductase or complex III, is a functional dimmer (reviewed in Scheffler 1999). It transfers electrons from reduced ubiquinone (which is referred to as ‘ubiquinol’) to cytochrome c—a small mobile electron carrier associated with the outer surface of the inner membrane. In mammals, each subunit complex contains 11 protein subunits (10 nuclear-encoded genes that produce proteins imported into the mitochondrion and one mtDNA), an [2Fe-2S] iron–sulphur cluster and three cytochromes (one cytochrome c1 and two b cytochromes) (Iwata et al. 1998).

Complex IV, cytochrome coxidase or COX, represents the terminal complex of the respiratory chain. It catalyzes electron transfer from cytochrome c to molecular oxygen thereby reducing the latter to water. There are usually 10 nuclear-encoded genes (nine in Drosophila) and three mtDNA genes in the complex. In metazoans, a 243 amino acid region of cytochrome oxidase I (COI) in this complex is a common DNA barcoding region. Melvin et al. (2008) constructed a complex IV quaternary structure model from the inferred amino acid sequence of each COX gene for Drosophila based on the bovine model. Figure 2 shows the COI barcoding region. Of the 243 amino acids in the barcoding region, 40% (97) interact with another protein in complex IV while 60% (146) do not interact with another protein in the complex (Fig. 3). In combination, these data show that the COI barcoding region has the potential to be influenced by mutations in nuclear-encoded genes. This result has the potential to significantly influence estimates of diversity gleaned from the region. Schmidt et al. (2001) observed that mtDNA-encoded residues in close physical proximity to nuclear DNA-encoded residues evolved more rapidly than the noncontact mtDNA-encoded residues in 26 species of mammals.

Figure 2.

 Cytochrome c oxidase monomer showing the region of subunit I that is translated from the DNA barcode sequence in red. The amino acid sequence inferred from Drosophila DNA sequences was modelled on the structure of bovine cytochrome c oxidase (PDB 1v54).

Figure 3.

 Cytochrome oxidase I barcode interactome showing the amino acid interactions. A ‘1’ denotes that the amino acid is with 4 Å of another amino acid in a different protein. A ‘0’ denotes that it is more than 4 Å of an amino acid in a different protein.

Complex V, or ATP synthase, is the final enzyme in the OXPHOS pathway. This enzyme is found in all forms of life and functions in the same way in both prokaryotes and eukaryotes. The enzyme uses the energy stored in a proton gradient across a membrane to drive the synthesis of ATP from ADP and phosphate (Pi). It is composed of two domains: a hydrophobic F0 membrane part that is connected to a water-soluble F1-headpiece by two stalks. Knowledge of the structure and assembly of the ATP synthase subunits has grown largely in recent times (Walker & Dickson 2006; Strauss et al. 2008). The mammalian enzyme complex contains 16 subunits: 14 nuclear-encoded subunits and two mtDNA-encoded subunits. Phylogenetic analysis of complex V mtDNA-encoded subunits showed that they are among the most evolutionarily divergent components of the mitochondrial genome between the orangutan and the other apes (Bayona-Bafaluy et al. 2005). Another fascinating study of this complex involves the unique capacity for camels to live in extreme thermal environments. Di Rocco et al. (2009) sequenced and analyzed the ATP6 and ATP8 genes and then compared the sequence data with other mammals and Escherichia coli. In ATP6, they found that two specific amino acid substitutions occur in highly conserved sites and suggest that these are likely to be functionally important. Studies such as this example are interesting and suggestive, but care needs to be taken in how these inferences are made. Camelid-specific substitutions are to be expected, even in conserved sites, if camelids share a unique evolutionary history. That camels live in arid environments does not establish that these synapomorphic residues are associated with that adaptation.

Efficiency of OXPHOS

The efficiency of OXPHOS may be assessed from the top-down or from the bottom-up. A top-down approach is essentially breaking down a system to gain insight into its compositional subunits. In a top-down approach, an overview of the system is first formulated, specifying but not detailing any first-level subunits. Each subunit is then refined in greater detail, sometimes in many additional subunit levels, until the entire specification is reduced to base elements. Biochemically, the whole chain can be assayed using the ADP:O ratio (Hafner et al. 1990; Nicholls & Ferguson 2002). The ADP:O ratio is defined as the moles of ADP phosphorylated to ATP divided by the moles of atomic oxygen consumed by mitochondria. If the top-down approach shows that the efficiency of the chain is compromised, then the chain may be broken down into functional blocks to determine the specific complexes that are responsible for the decline. The activity of each complex within the chain can then be quantified with relatively standard biochemical tests (reviewed by Kirby et al. 2007). A limitation of this approach is that the ADP:O ratio derived from an organism requires highly specialized equipment and has the potential to be influenced by a variety of factors, including the tissue from which the mitochondria were extracted (Katewa et al. 2006) and the geographic locality from which the organism was collected (Hochachka & Somero 2002).

An alternative to the top-down approach is bottom-up analysis. A bottom-up approach is piecing together subunits that give rise to the original systems. In a bottom-up approach, the individual base elements of the system are first specified in great detail. These elements are then linked together to form larger subsystems until a complete top-level system is formed. In terms of OXPHOS, this approach examines the effects of titrating specific mitochondrial enzymes and transporters with irreversible inhibitors and determining the effects on respiratory rate and ATP synthesis (Nicholls & Ferguson 2002). This approach is appropriate if the underlying genetic mutations are known. However, if the efficiency of the whole chain is not reconstructed, the approach has the potential to generate misleading information, particularly when the efficiency of a single complex or one enzyme does not equate with that of the whole electron transport chain.

The efficiency of OXPHOS is influenced by loss of electrons from the OXPHOS complexes to form ROS, conductance of protons across the inner mitochondrial membrane (proton leak) and the retrograde response to the nucleus. Increased ROS production causes a decrease in mitochondrial efficiency and likely an increase in oxidative stress. The common forms of ROS in cells are the superoxide free radicals (O2), hydroxyl radical (OH) and hydrogen peroxide (H2O2). Highly reactive O2, formed when O2 is reduced by the loss of single electrons from the OXPHOS complexes, can damage mitochondrial and cellular proteins, lipid membranes and DNA. Oxidative stress results when there is an imbalance between the production of ROS and a biological system’s ability to detoxify the reactive intermediates or easily repair the resulting damage. As discussed below, accumulating evidence suggests that the capacity to withstand oxidative stress plays an important role in shaping individual longevity and fecundity.

A second mechanism that reduces OXPHOS efficiency is proton leak (Brand 1995). The potential energy lost by proton leak is dissipated as heat. The biological explanation for proton leak is not known. It is hypothesized to act somewhat like a safety valve by reducing the backup of electrons in the electron transport chain and thereby minimizing ROS production (Brand 2000). In mammals, proton leak may account for as much as 20% of mitochondrial respiration (Stuart et al. 1999). Proton leak-related heat production has been argued to influence the distribution of mammalian mtDNA haplogroups (Mishmar et al. 2003; Ruiz-Pesini et al. 2004) and may influence the whole-organism thermal tolerance in ectotherms (Pörtner 2002; Pörtner et al. 2007).

A third mechanism that modulates OXPHOS efficiency is the retrograde response. This response is a signalling pathway from the mitochondria to the nucleus. Retrograde signalling activates nuclear and mitochondrial transcription factors and results in increased numbers of mtDNA, OXPHOS complexes and whole mitochondria (Passos et al. 2007). Retrograde response is well known in the yeast Saccharomyces cerevisiae and to a lesser extent in mammalian cells (Butow & Avadhani 2004). Conceptually, overcompensation for the decrease in function of a specific OXPHOS complex, through a retrograde response, has the potential to bias estimates of OXPHOS efficiency so this possibility should be considered when bioenergetic test results appear counter-intuitive.

Experimental manipulations

Quaternary structure studies have the potential to predict the functional consequences of specific mtDNA mutations, and mitochondrial bioenergetics can quantify the efficiency of the whole chain and specific complexes. However, it is necessary to place the mutations in one or more nuclear backgrounds to experimentally test these predictions. Currently, at least three techniques are available to achieve this goal. In this section, we review the three major techniques, present a subset of examples and discuss some potential limitations of each approach.

In the near future, an additional experimental manipulation methodology is likely to be possible. Specifically, it may soon be possible to colonize ethidium bromide-treated mitochondria with synthesized genomes of known DNA sequence. These mitochondria may then be microinjected directly into embryos. If these embryos develop to maturity, and the artificial mtDNA can outcompete the resident mtDNA, the bioenergetic and phenotypic consequences of any induced mutations could be tested.


Backcrossing is the crossing of a hybrid with one of its parents, or an individual genetically similar to its parent, to achieve offspring with a genetic identity that is closer to that of the parent. If a female hybrid is repeatedly backcrossed to the paternal genotype, the resulting offspring are expected to have an increased proportion of the paternal nuclear genotype but maintain the maternal mtDNA (assuming paternal leakage does not occur).

Ron Burton and his collaborators have developed one of the best-studied systems of mitonuclear coadaptation in the marine copepod Tigriopus californicus utilizing backcrossing. Much of the work has focused on nuclear and mitochondrial components of the electron transport system emphasizing interactions between the nuclear-encoded cytochrome c and complex IV that contains both mitochondrial- and nuclear-encoded subunits. In a subset of interpopulation crosses, mitonuclear hybrids created by repeated backcrossing exhibited COX activity levels consistent with mitochondrial–nuclear coadaptation (Edmands & Burton 1999). In vitro studies showed higher COX activity when cytochrome c and COX were from the same population (Rawson & Burton 2002; Harrison & Burton 2006). However, the highest COX activity occurred with cytochrome c variants (generated by site-directed mutagenesis) that were a mosaic of amino acids from populations that did not match the cytoplasmic background (Harrison & Burton 2006). Subsequently, Ellison & Burton (2006) also showed that mitonuclear mismatch reduced activity of complexes I and III, as well as ATP production in hybrids between allopatric populations. Strong evidence for mitonuclear interaction in Tigriopus also comes from Ellison & Burton (2008). In this latter case, the fitness of F3 hybrids was restored in maternal backcrosses that had a full haploid nuclear genome matching the mitochondrial genome. Fitness levels were not restored in paternal backcrosses, which had greater mitonuclear mismatch.

A limitation of the backcrossing strategy is that mitonuclear interactions may influence the randomness of the introgressions. As a consequence, the robustness of replacing specific genes using this strategy needs to be tested. In theory, 99.976% of the nuclear genome is replaced after 12 generations of backcrossing. However, it is unlikely that this level of replacement is ever achieved. Dermitzakis et al. (2000) crossed Drosophila sechellia males with Drosophila simulans females in 221 independent lines and found that some parts of the D. sechellia genome were never detected in the F1 hybrids. This could be attributed to incompatibilities of epistatic genes and, in particular, genes that are associated from the D. simulans mitochondrial and the D. sechellia nuclear genomes.


Microinjection of mitochondria into zygotes is an established technique for transferring mtDNA into distinct genetic backgrounds. If the desired mitochondria drifts to fixation, or can be selected upon, the incumbent mitochondria may be replaced.

Microinjection of cytoplasm, containing mitochondria, plus a sperm has been used clinically in humans, but the technique is currently banned in many countries including the United Kingdom and United States (Spikings et al. 2006). The rationale for employing the technique is that the microinjected cytoplasm has the potential to contain younger, nonmutated mtDNA and will possibly contain other important cytoplasmic components to compensate for the defective cytoplasm in the oocyte (Cohen et al. 1997). Oocytes with higher copy numbers of mtDNA are known to be associated with improved fertilization rates, with ‘good quality’ human oocytes having a higher mtDNA copy number than ‘poor quality’ oocytes (mean 255 000 vs. 152 000 copies) (Reynier et al. 2001). In humans, cytoplasmic transfer has led to the birth of nearly 30 babies worldwide (Barritt et al. 2001). Not surprisingly, this technique has caused significant clinical concern from a variety of perspectives, including the possibility that heteroplasmy, resulting from two wild-type mtDNA molecules, has the potential to be detrimental to the offspring and result in mitochondrial disease (Spikings et al. 2006). An alternative possibility, which is worth considering, is that the fitter mtDNA haplotype will go to fixation in each tissue (Jenuth et al. 1997).

A limitation with the microinjection technique is that it may not be possible to construct all possible combinations of mitochondrial and nuclear genomes. Indeed, it may not be possible in all experimental systems. In an elegant study, de Stordeur (1997) conducted microinjection studies between eggs carrying the three D. simulans mtDNA types (siI, -II and -III) and assayed the frequencies of the foreign injected mtDNA. He demonstrated that it is straightforward to microinject and then allow selection to fix siII and siIII mitochondria in flies natively harbouring siI mtDNA. However, he was not able to successfully transfer siI or siIII mtDNA into flies natively harbouring siII mtDNA, or siI mtDNA into flies natively harbouring siIII mtDNA.

Cybrid cells

Cybrid cell lines are eukaryotic cell lines that are produced by the fusion of a whole cell with a cytoplast. Cybrid technology, now mainstream in the biochemical and bioenergetics fields, is poised to enter the fields of molecular ecology and evolution. King & Attardi (1989) isolated two derivatives of a human cell line that had been entirely depleted of mtDNA by long-term exposure to low concentrations of ethidium bromide. These cell lines (termed p0) were then used for mitochondrial transformation studies. Cybrid transformants obtained with various mitochondrial donors exhibited a respiratory phenotype that was in most cases distinct from that of the p0 parent or the donor, indicating that the genotypes of the mitochondrial and nuclear genomes, as well as their specific interactions, play a role in the respiratory competence of a cell.

In two controversial studies, Wallace and colleagues (Mishmar et al. 2003; Ruiz-Pesini et al. 2004) proposed that human mtDNA haplogroups A, C, D and X had decreased coupling efficiency and increased heat production that were both beneficial for humans moving out of Africa into colder climates. To overcome effects of different nuclear backgrounds, Amo & Brand (2007) fused mitochondria obtained from healthy human volunteers with cells derived from human lung carcinoma A549. The resultant cybrids had mtDNA from the different donors, but their nuclear DNA was identical and derived from human lung carcinoma A549. Contrary to the predictions of Wallace and colleagues (Mishmar et al. 2003; Ruiz-Pesini et al. 2004), mitochondria from Arctic haplogroups had similar or even greater coupling efficiency than mitochondria from tropical haplogroups.

Despite the apparent elegance of the experimental cybrid test, a significant factor surrounding the value of these studies is that the A549 cell line was taken from a 58-year-old Caucasian male (Lieber et al. 1976). This is a concern because the lack of mtDNA transmission in males prevents direct selection on mtDNA haplotypes with male-specific phenotypes, and further stifles the evolution of nuclear gene mutations that might suppress male-specific mitochondrial defects (Frank & Hurst 1996). Consistent with this genetic argument, there is evidence that mitochondrial diseases can be more severe, or more prevalent, in males than in females (Frank & Hurst 1996; Chinnery & Turnbull 2001). Further, in Drosophila, sex-specific fitness effects of mtDNA haplotypes have been described (Rand et al. 2001). When the mtDNA from D. simulans is introgressed onto the nuclear background of Drosophila mauritiana, the disruption of COX activity is more severe in males than in females (Sackton et al. 2003). A more robust test of the predictions of Wallace and colleagues (Mishmar et al. 2003; Ruiz-Pesini et al. 2004) would be to fuse mitochondria obtained from healthy human volunteers with cells derived from multiple females. The mitochondrial bioenergetics of these cybrids could then be quantified.

Mitochondrial phenotype

Understanding why and how particular individuals pass on a greater number of gene copies to the following generation is a hub of evolutionary ecology. Most generally, the greater lifetime reproductive success of some individuals over others is usually explained by positive covariation between major life-history traits, such as lifespan and fecundity (Stearns 1992). In this section, we consider the links between mitochondrial genotype and the three phenotypic life-history traits of lifespan, fecundity and starvation stress. We then discuss how mitochondria may be involved in determining the distribution of a species by establishing both the upper and lower thermal limits of organisms.

In the case of mitochondrial metabolism, bioenergetic differences may underpin significant evolutionary trade-offs. Ballard and colleagues have examined trade-offs in the three Drosophila simulans mtDNA haplotypes. Combined, the data suggest a classical life-history trade-off with D. simulans siII flies having an advantage when resources are continuous and abundant. They have greater egg size and fecundity, are more cold-tolerant and have higher rates of hydrogen peroxide production from complex III of the electron transport chain (Ballard et al. 2007). In contrast, D. simulans siIII flies are likely to have an advantage when resources are discontinuous. Flies with this mtDNA type are more starvation-resistant and have more efficient mitochondrial metabolism with higher COX activity and lower proton leak. A likely trade-off of the low proton leak is an increased cold sensitivity that effectively restricts the biogeographic distribution of this mtDNA type (Dean et al. 2003; Ballard 2004; Ballard et al. 2007; Katewa & Ballard 2007). Island endemic, D. simulans siI flies are fastest developing and have the lowest probability of surviving to the three ages tested (James & Ballard 2003). These data explain, at least in part, the observations that D. simulans siII flies are fitter in the laboratory (Ballard & James 2004) but D. simulans siIII-harbouring flies occur at intermediate frequency in Reunion Island, Madagascar and continental east Africa (Dean et al. 2003; Ballard 2004).

Mitochondrial genotype and lifespan

The mitochondrial theory of ageing (Boveris & Cadenas 1982; Foreman & Boveris 1982) is alive and well, probably more so now than at any time since its original formulation. The core of the mitochondrial theory of ageing is that mutations in the mitochondrial machinery result in greater production of ROS as by-products of normal bioenergetics. These ROS are well known to damage DNA, lipids and cell membranes. Damage to mtDNA can lead to the production of mutant forms of mitochondrial proteins, leading to increased ROS production, oxidative stress and a vicious cycle of damage that will eventually lead to loss of function (Passos et al. 2007). Recent findings have, however, added complexity to the mitochondrial theory of ageing and, in particular, thrown up some interesting puzzles concerning the connections between the amino acid substitution rate in mitochondrial proteins and the ageing process.

Rottenberg (2006, 2007a, b) analyzed the evolution of protein sequences in mammals and birds. He reported an increased amino acid substitution rate in long-lived species, which he interpreted as an adaptive process. Somewhat controversially, he suggested that the mitochondrial proteins of long-lived organisms would have accumulated advantageous mutations that optimized the respiratory chain to reduce the rate of ROS production. This hypothesis is controversial because it is contrary to the tenet that purifying selection is the predominant force in shaping the evolution of mtDNA. Specifically, as the effective population size declines, some slightly deleterious nonsynonymous changes that were initially suppressed by purifying selection become effectively neutral and can reach fixation (Popadin et al. 2007). As a consequence of the apparent conflict, Galtier et al. (2009a) re-analyzed this relationship in various mammalian lineages using a Bayesian phylogenetic analysis of amino acid sequences. As expected, a negative relationship between protein evolutionary rate and species longevity was reported for all OXPHOS complexes. A detailed analysis of cytochrome b in 528 mammals reinforced this result. Galtier et al. (2009a) explained the discrepancy between the results by their improved taxon sampling and more appropriate methodology that accounted for the high variation of substitution rate across amino acid sites.

To examine the effects of natural mtDNA variation on Drosophila melanogaster lifespan, Clancy (2008) used an elegant crossing scheme to control the nuclear genotype (w1118) whilst allowing retention of the original mtDNA haplotype. Flies were collected from a variety of localities including Dahomey, Alstonville, Mysore and Japan. The mtDNA protein-coding regions of all mtDNA genomes was then sequenced and the male lifespan quantified. The standard mtDNA genome of w1118 differed from Alstonville by six amino acids but median lifespan did not differ. The mtDNA from Dahomey differed from that collected in Alstonville by three amino acids and males had a 4% reduction in lifespan; Mysore differed from Alstonville by four amino acids and males had a 9% reduction in lifespan; Japan differed from Alstonville by a single amino acid and lifespan was reduced by 15%. The bioenergetic efficiency and fecundity of these fly lines has yet to be determined.

Mitochondrial genotype and fecundity

Evidence is growing that the capacity to withstand oxidative stress plays an important role in shaping individual fecundity (Beckman & Ames 1998; Finkel & Holbrook 2000; Barja 2004). In agreement with the hypothesis that oxidative stress incurs costs to the organisms in terms of reduced fecundity, Alpine Swift females with lower resistance to oxidative stress laid clutches that were smaller and less likely to hatch than did females with higher resistance to oxidative stress (Bize et al. 2008). Further, males that survived up to the next breeding season were more resistant to oxidative stress (Bize et al. 2008).

Mitochondrial mutations have been shown to influence mitochondrial bioenergetics, which in turn can influence reproductive success in two mutually exclusive ways; namely, effects on germinal and somatic cells. Mitochondria are of major importance in the oocyte and early embryo, particularly as a source of ATP generation, and perturbations in their function have been related to reduced embryo quality. Higher quality oocytes, assessed by morphology, contain significantly higher ATP levels and produce significantly higher blastocyst rates after fertilization (Stojkovic et al. 2001). In this respect, human oocytes require at least 2 pmol ATP at fertilization to enhance development and implantation (Van Blerkom et al. 1995). Furthermore, a correlation between mitochondrial distribution in individual blastomeres and blastomere ATP content suggests that the ATP contained within early embryos is likely to be derived from mitochondria (Van Blerkom et al. 2000).

Mitochondrial DNA mutations may also accumulate in oocytes. Unlike nuclear DNA which replicates only once during each cell cycle, mtDNA is continuously turning over in nondividing tissues with a half-life of between 0.7 and 30 days in humans (Gross et al. 1969). Although the rate of mtDNA turnover in human oocytes is unknown, it is likely that their mitochondrial genomes will have been turned over many times before human menopause, and it is theoretically possible that they may acquire mtDNA deletions. Chen et al. (1995) showed that 50% of apparently healthy human oocytes (harvested for IVF) harbour mtDNA mutations. Although there is some evidence that the level of these mutations increases with maternal age (Keefe et al. 1995), other studies did not report any relationship between the amount of mutated mtDNA in oocytes from young and old women (Muller-Hocker et al. 1996). It remains to be determined whether the mutations are present in oocytes that are capable of fertilization, or whether the affected oocytes are not capable of forming a healthy embryo. One report indicated that as many as 40% of human oocytes that fail to fertilize may have mtDNA deletions (Reynier et al. 1998), and that each oocyte collected from the same donor displays its own distinctive deletion.

There is also strong evidence for an association between mtDNA mutations and male fertility (Frank & Hurst 1996). The association occurs because mtDNA is maternally inherited and mutations that are neutral in females but slightly deleterious in males can accumulate in populations. Further, nuclear mutations affecting the activity of the mammalian complex I have been mapped to the mammalian X chromosome (Zhuchenko et al. 1996). Nakada et al. (2006) showed that a 696-bp deletion in the mitochondrial genome could cause male infertility in humans, while Holyoake et al. (1999) showed that a single amino acid change in human ATPase 6 was sufficient for male infertility. The cause of reduced fertility is likely due to a reduction in sperm activity that is dependent on OXPHOS (Ruiz-Pesini et al. 2000).

Mitochondrial DNA type also influences sperm motility and respiratory chain complex enzyme activities in the mussel Mitilus edulus (Jha et al. 2008; Breton et al. 2009). Bivalves of the families Mytilidae, Unionidae and Veneridae have an unusual mode of mtDNA transmission called doubly uniparental inheritance (DUI). A characteristic feature of DUI is the presence of two gender-associated mtDNA genomes that are transmitted through females (F-type mtDNA) and males (M-type mtDNA), respectively. Female mussels are predominantly homoplasmic whereas males are heteroplasmic, with the F type in somatic tissue and the M type expressed in the gonad. Collectively, males are polymorphic for two distinct classes of M-type mtDNA known as the standard male (SM) and recently masculinized (RM) types. Because the SM and RM types exhibit considerable amino acid sequence divergence, Breton et al. (2009) hypothesized that these differences could affect mitochondrial respiratory chain complex enzyme activities. To test this hypothesis, the activity of the major mitochondrial respiratory chain complexes (complexes II, I+III and IV) as well as the activity of citrate synthase was measured on gonad samples from males containing either the SM or RM mtDNA. The data demonstrated that the mitochondrial subunits encoded by the RM mtDNA are associated with higher enzymatic activities than the gene products of the SM mtDNA. This may influence sperm activity and the multiple evolutionary replacements of standard M-types by masculinized M-types that have been hypothesized for the mytilid lineage (Jha et al. 2008).

Male fertility may also be influenced by mtDNA genotype in a variety of other organisms, including some birds (Gemmell et al. 2004). In domestic chickens, Gallus gallus domesticus, sperm motility and fertilization success is strongly influenced by maternal genes (Birkhead et al. 1999; Froman et al. 2002). The maternal mtDNA is strongly implicated because in birds females are the heterogametic sex. Therefore, the only sex-specific molecule that can be transmitted from mother to son is mtDNA. Mitochondrial type does not, however, influence sperm velocity in the Zebra Finch Taeniopygia guttata (Mossman et al. 2009).

Mitochondrial genotype and starvation resistance

Individuals within many species must survive periods of starvation or exposure to suboptimal diets in order to successfully reproduce. As a consequence, positive selection for resistance to starvation stress is expected in localities where food is likely to be less abundant or temporarily less reliable. Three mechanisms have been proposed to increase starvation resistance (Rion & Kawecki 2007) and each has the potential to be influenced by mitochondrial variation. First, greater energy reserves, in particular lipid, may be stored (reviewed in (Hoffmann et al. 2005; Hoffmann & Harshman 1999; Rion & Kawecki 2007). Starvation-resistant insects may carry greater lipid reserves at eclosion (Aguila et al. 2007) or continue to increase their lipid reserves during early adulthood (Chippindale et al. 1996). Ballard et al. (2007, 2008) showed that D. simulans siIII flies, were more starvation resistant than D. simulans siII flies and starvation resistance was positively correlated with body lipid proportion. However, increasing lipid reserves does not guarantee an increase in starvation resistance in all cases (Hoffmann et al. 2001; Baldal et al. 2005).

An alternative mechanism to increase starvation resistance may be to reduce the rate at which reserves are utilized. Evidence supporting this alternative is equivocal but may involve a reduction in energy expenditure (Rion & Kawecki 2007). Drosophila lines selected for starvation resistance have also been reported to show lower locomotor activity under some but not all conditions (Hoffmann & Parsons 1993; Williams et al. 2004). Increasing locomotor activity is expected to require additional ATP to fuel the energetic expenditure and this, in turn, is expected to increase the production of ROS as a by-product of normal OXPHOS. Increasing ROS is likely to cause oxidative stress unless antioxidant levels are also elevated (Beckman & Ames 1998; Finkel & Holbrook 2000).

A final mechanism that may, at least in theory, increase starvation resistance is the lowering of the minimal resources required for survival (Rion & Kawecki 2007). In terms of mitochondrial biogenesis, it may occur by increasing OXPHOS efficiency by decreasing ROS and heat production (Mishmar et al. 2003; Ruiz-Pesini et al. 2004).

Demographic patterns

The pervasive effects of temperature on biochemical and physiological processes are thought to play a fundamental role in establishing biogeographic patterns (Hochachka & Somero 2002), but precise mechanisms involved in establishing both the upper and lower thermal limits of organisms are still poorly understood (Pörtner et al. 2006). Niki et al. (1989) constructed heteroplasmic flies possessing both endogenous mitochondrial DNA and foreign mitochondrial DNA by intra- and interspecific transplantation of germplasm. During the maintenance of these heteroplasmic lines, flies of D. melanogaster were produced that no longer possessed their own mtDNA but retained the foreign mtDNA from Drosophila mauritiana. Interestingly, temperature was shown to influence the fate of the microinjected mitochondria and was subsequently used as a strong selective agent (Nagata & Matsuura 1991; Matsuura et al. 1993, 1997).

The concept of oxygen and capacity-limited thermal tolerance (OLTT) has recently been proposed as a unifying principle for understanding the mechanistic basis of whole-organism thermal tolerance in ectotherms (Pörtner 2002; Pörtner et al. 2007). The OLTT hypothesis generates several specific predictions with respect to the relationship between mitochondrial properties and whole-organism responses to temperature: (i) cold-adapted or -acclimated organisms should have greater mitochondrial capacity (either increased mitochondrial amount or increased functional activity); (ii) cold-adapted or -acclimated organisms with increased mitochondrial content or function should have higher standard (resting) metabolic rate relative to warm-adapted organisms when measurements are made at the same temperature in both cold- and warm-adapted organisms; and (iii) the increased metabolic rate in cold-adapted or -acclimated organisms should be associated with a decrease in thermal tolerance.

To test the OLTT hypothesis, Fangue et al. (2009) assayed whole-organism metabolic rate and mitochondrial properties of killifish from the northern and southern subspecies acclimated to a range of temperatures. The most striking observation was that acclimation temperature level had a different effect on the acute response of mitochondria from each subspecies of killifish. There were few functional differences between subspecies in mitochondria isolated from fish acclimated to 25 °C. In contrast, mitochondria from fish acclimated to 5 °C had a much more complex response to acute temperature change, and the shape of the response differed between subspecies. These data demonstrate that there is plasticity in the function of mitochondria and that this plasticity involves changes in the way in which the mitochondria respond to acute temperature change.


The influence of genotypic variation on the phenotype of an organism has been a longstanding question in biology, and it is now one of the grand challenges of the postgenomics era. Mitochondria play a unique role in animal biology, but this role has largely been studied—in the evolutionary biology and ecology communities at least—as a means to demographic and historical inference. Here, we suggest that mitochondria have the potential to shape the demographics of populations and species rather than simply reflect history. A central question that remains to be resolved is the frequency of adaptive mutations in mtDNA.

The lack of recombination in mtDNA means that it is not possible to do traditional genetic studies. It is, however, possible to investigate the importance of specific substitutions by linking advances in quaternary structure modelling of each OXPHOS subunit with bioenergetic studies to predict the functional consequences of naturally occurring changes. Experimental manipulations can then test these predictions. Several evolutionary studies have documented fitness consequences arising from the mitochondrial genetic variation that exists naturally within (Rand et al. 2001; Dowling et al. 2007b) and between populations (see, for example, Clark 1985; James & Ballard 2003; Ballard & James 2004; Dowling et al. 2007a; Clancy 2008) but few studies have fully explored a mechanistic link. Given the fundamental role that the mitochondrion plays in metabolism, it is easy to imagine that there are important ecological differences mediated by mitochondrial changes that may even involve such basic processes as speciation (Gershoni et al. 2009). In this review, we considered the links between mitochondrial genotype and ageing, fecundity, starvation stress and thermal tolerance. Focused studies of natural populations are required to determine whether adaptive mutations in mtDNA are the rule, or the exception to the rule.


We wish to acknowledge constructive comments made by Nicolas Galtier, David Rand, Mike Whitlock, Kirrie Ballard and Lou Puslednik. For inspiring the past 15 years of research, J.W.O.B. would also like to thank the individual who came up to him at the 1994 Evolution meeting at the University of Georgia, Athens, GA and crossed his index fingers saying ‘so you are the person who suggests that mitochondrial DNA is not a neutral evolutionary marker’.

Bill Ballard is a Professor of the School of Biotechnology and Biomolecular Sciences at the University of New South Wales, Sydney, Australia. His goal is to link the genotype with the phenotype using comparative genomics, population genetics, biochemical analyses and life-history trait analyses. His system of choice is mitochondrial genome in Drosophila. Rich Melvin is now a Postdoctoral Fellow, UMD-Biology in Duluth, Minnesota. His research blends biochemistry and population genetics to study evolution in structured populations.