Bioenergetics in human evolution and disease: implications for the origins of biological complexity and the missing genetic variation of common diseases

The striking correlation between mtDNA variation and geographic locale of indigenous populations stands in stark contrast to nDNA variation, where the common allelic variants are dispersed across most populations, differing primarily in allele frequencies [9–11]. For the mtDNA, each indigenous population has a limited number of clusters of related mtDNA haplotypes. Each of these haplotype clusters is generally descended from a founder mtDNA that harboured one or more functional variants. Therefore, the founder mtDNAs must have arisen in the population and, as the population grew, the descendants of the founding mtDNA acquired additional variants creating population-specific groups of related mtDNA haplotypes, designated haplogroups.

Haplogroups are also related to each other in higher-order clusters [43]. All of the haplogroups in Africa, designated ‘L’ haplogroups, are components of the wider African mtDNA lineage designated macrohaplogroup L that arose between 130 000 to 200 000 years before the present (YBP) (figure 1). Of the L haplogroups, haplogroup L3 spawned two new mtDNAs, designated M and N, in the Ethiopian area. About 65 000 YBP, only mtDNAs derived from M and N successfully departed Africa and colonized the rest of the world, generating macrohaplogroups M and N. Macrohaplogroup M moved along tropical Southeast Asia, ultimately reaching Australia, and later moved north out of Southeast Asia to form a plethora of central and eastern Asian mtDNA haplogroups, including C, D, G and M1–M20. By contrast, macrohaplogroup N went in two directions. In one, N moved along Southeast Asia to Australia and from southern Asia north into central Asia to generate haplogroups A and Z. In the second, N moved due north out of Africa to form the European haplogroups I, X and W and also into western Eurasia to form sub-macrohaplogroup R. R then gave rise to the European-specific haplogroups H, J, Uk, T, U and V, and R also moved east to produce the Asian mtDNA haplogroups B and F (figure 1).

Of all of the Asian mtDNA variants, only A, C and D became enriched in northeastern Siberia and were in a position to cross the Bering Land Bridge about 20 000 YBP to give rise to the Paleo-Indians. Later, A, C and D were joined by additional migrations bringing B and X to round out the Paleo-Indian mtDNA lineages. Subsequently, additional migrations brought haplogroups A and A + D to produce the Na-Dene and the Eskimos and Aleuts, respectively. Lastly, individuals carrying mtDNA haplogroup B mtDNA moved out from the Asian coast to colonize all of the Pacific Islands (figure 1) [47,48].

This regionality of the mtDNA haplogroups is extraordinary in several ways. First, of all of the African diversity, only two mtDNA lineages (M and N) colonized the rest of the world. This striking fact was missed in the first paper to report an African origin of the mtDNA [49], since these authors included African-American mtDNAs in their ‘African’ sample and about 30 per cent of African-American mtDNAs are of European, Native American and Asian origin introduced after the arrival of Africans in the Americas [50]. Second, of all of the Asian mtDNAs, only three mtDNA lineages (A, C and D) moved to extreme northeast Siberia to found the Paleo-Indians. Third, and most surprising, the mtDNA sequence evolution rate is such that it produced important mtDNA evolutionary changes that coincide with the major human geographical transitions. It is hard to imagine that such associations occurred by chance. It is much more probable that mtDNA variation was intimately involved in facilitating the human adaptation to different regional environments.

Support for the hypothesis that the major human population transitions were facilitated by mtDNA variation comes from the observations that mtDNA lineages that appear together with human transitions often harbour important mtDNA functional changes [34,35,51]. For example, the founding mtDNA for macrohaplogroup N, which moved directly from sub-tropical Africa north into the European temperate zone, harboured two polypeptide variants: ND3 nt 10398 G>A (A114T) and ATP6 nt 8701 G>A (A59T)[47,48], which changed the mitochondrial membrane potential and Ca2+ metabolism [52]. By contrast, macrohaplogroup M, which remained in semi-tropical Southeast Asia, was not founded by distinctive polypeptide variants. Significant mitochondrial biochemical differences have also been reported between other haplogroups, for example, European haplogroups H and Uk [53,54].

That mtDNA variants might have been adaptive has been substantiated by studying mtDNA variation in Tibetans, who are adapted to high altitude and thus low oxygen tension. In Tibetans, the ratio of macrohaplogroup M to N mtDNAs is greater than in low-land Chinese. Furthermore, an mtDNA ND1 nt 3394C (Y30H) variant has been found to have arisen three independent times on different macrohaplogroup M mtDNAs in Tibetans, but never to have become established on any Tibetan macrohaplogroup N mtDNAs. The 3394C variant progressively increases in frequency in villages with increasing altitude. For example, haplogroup M9 mtDNA, which encompasses the 3394C variant, is present at less than 2 per cent of the mtDNAs at sea level but increases to almost 35 per cent of the mtDNAs in the highest Tibetan villages.

Biochemical analysis has shown that when the 3394C variant arises on macrohaplogroup N mtDNAs, specifically haplogroups B or F, it results in a 15–28% reduction in complex I-specific activity. However, when the 3394C allele is present on haplogroup M9 mtDNA, it is associated with a complex I activity that is equal to or greater than any of the macrohaplogroup N mtDNAs with the wild-type 3394T variant [40]. Hence, the 3394C variant is adaptive at high altitude when arising on macrohaplogroup M mtDNAs.

mtDNA haplogroups have also been correlated with predisposition to a wide range of metabolic and degenerative diseases, cancers and longevity [48,53,55,56]. For example, the ND1 nt 3394C (Y30H) variant, when arising on macrohaplogroup N mtDNAs in European populations, is associated with increased penetrance of the milder primary mtDNA mutations that cause Leber Hereditary Optic Neuropathy (LHON) [57,58]. Haplogroup F mtDNAs are associated with a predilection to diabetes [59] and analysis of the complex I-specific activities between mtDNA haplogroups B and F has revealed a 30 per cent lower complex I activity in haplogroup F [40]. Consequently, adaptive and clinically relevant mtDNA variants are also associated with functional changes in OXPHOS.

These data prove that at least some mtDNA variation is not neutral, but rather is functional and can be adaptive when present on the appropriate mtDNA background and environment. Yet the same variant in a different mtDNA background and/or environment can be deleterious and predispose to metabolic and/or degenerative diseases [34,35,51].

According to Kimura [12], the high mtDNA mutation rate should create many deleterious mutations resulting in a genetic load that could destroy the species. This concern is particularly apt for the mtDNA since every one of the genes of the mtDNA is critical for life. However, this concern is unfounded since the mammalian ovary has a pre-fertilization selection system that removes those proto-oocytes that harbour severe mtDNA mutations before they can be ovulated and fertilized [60,61]. Hence, these mutations never contribute to the genetic load of the species. The human female generates millions of proto-oocytes, but only successfully ovulates about 400. All the rest are eliminated by atresia. This apparent inefficiency might exist specifically to permit selection against severe mtDNA mutations, thus permitting a high mtDNA sequence evolution rate.

The high mtDNA mutation rate paired with the pre-fertilization selection means that de novo mtDNA mutations that subtly alter mitochondrial energy metabolism are constantly being introduced into the mammalian germline. These mtDNA variants then produce bioenergetic diversity that lets subpopulations within the species physiologically adapt to and occupy slightly different region energetic environments. Hence, mtDNA variation is an important component of the missing functional variation that allows subpopulations of humans and other animals to adapt to local environments.

Why does this system not work for nDNA anatomical gene mutations? The difference is that mitochondrial bioenergetics is manifest at the individual cellular level, so natural selection can act directly on the metabolism of the unfertilized oocyte and/or its associated nurse cells. By contrast, to express anatomical genes, the oocyte must be fertilized and progress through development to generate a fully formed individual before the mutation can be manifest and acted on by natural selection. Therefore, even the most deleterious nDNA anatomical mutations are passed into the population where they can contribute to the genetic load. The only way of minimizing the lethal effects of deleterious anatomical mutations is to keep the nDNA mutation rate very low. This means that nDNA mutations are generally too rare to be the initiating factors in intra-specific functional radiation.

What about mutations in nDNA-coded bioenergetic genes such as those detected by GWAS? As Kimura [12] predicted, most nuclear mutations result in lethal phenotypes, for example, those that cause Maturity Onset Diabetes of the Young (MODY) [31]. Less severe but still functionally significant metabolic gene mutations might be beneficial in peripheral energetic environments and become locally enriched by adaptive selection. However, if the energetic environment changes or the individuals migrate to a new environment these functional variants become maladaptive and are removed by purifying selection causing metabolic disease. Only the mildest phenotypic variants in bioenergetic loci escape selection and can become dispersed throughout populations and detected by GWAS [13,14].

Animals thus have two primary evolutionary strategies: an energetic strategy that changes rapidly and is optimal for intra-specific radiation into various regional energetic environments and an anatomical strategy that changes slowly and is optimal for inter-specific structural radiation for exploitation of new niches. The short-term energetic adaptive strategy is primarily driven by the high mtDNA mutation rate and the resulting rapid intra-specific energetic radiation. The long-term anatomical adaptive strategy is rooted in the low nDNA mutation rate, leading to subtle inter-specific anatomical alterations [7,32].

The continuously arising mtDNA functional variation allows subpopulations of animal species to occupy a range of energetic environments within the species' niche. Those subpopulations that occupy peripheral energetic environments for prolonged periods will amass multiple mtDNA variants permitting the adjustment to local energy resources and demands. As the mtDNA sequence evolves, the slower-evolving nDNA bioenergetic genes will follow to consolidate survival in the peripheral energetic environment and to maintain optimal nDNA and mtDNA interactions [62].

The nDNA variants can have a range of phenotypic consequences from large to small. Variants that are of physiological significance and thus phenotypic importance in moderately divergent environments include the specific genetic variants in PPARγ [22], PGC-1α [23,24] and UCP1-3 [16–19]. Examples of nDNA variants that permitted human adaptation to extreme environments are found among Tibetans, who have adapted to low oxygen tensions for tens of thousands of years. Adaptive Tibetan nDNA loci include an intronic SNP within endothelial Per-Arnt-Sim (PAS) domain protein 1 (EPAS1), also known as hypoxia-inducible factor 2α (HIF-2α). HIF-2α is a transcription factor that responds to hypoxia by inducing glycolysis and modifying mitochondrial energetics. The Tibetan SNP shows a 78 per cent frequency difference between Tibetan and Han samples [63]. Additional hypoxia pathway genetic variants found in Tibetans include the HIF-1α modifying prolyl hydroxylase domain-containing protein 2 (PHD2)(EGLN1) and the HIF-1α target gene PPARA (PPARα) [64].

However, when an environment changes those loci that were adaptive in the previous environment can become incompatible with the new environment. This can result in pathological conditions such as diabetes and obesity and these alleles are then removed by purifying selection. nDNA variants with more modest phenotypic effects can be retained and dispersed by stochastic processes.

Associations between mtDNA haplogroups and diabetes have been found when studying well-defined human populations [59]. However, studies attempting to correlate mtDNA SNPs to diabetes using an aggregate of multiple Europe and American populations have not been successful [65].

These considerations then provide a logic for linking the disparate observations about human genetic variation and the origin of intra-specific adaptive variation. The high mutation rate of the mtDNA together with the pre-fertilization ovarian selection provide a rapid adaptation system for each species to adjust to regional environmental changes. These variants have a very high functional effect and hence are relatively specific to particular regional environments and thus subpopulations. As a result, even relatively closely related populations have very different adaptive mtDNA variants and thus different mtDNA haplotypes, haplogroups and associated mtDNA SNPs. The high mtDNA mutation rate also means that the same adaptive variant can arise multiple independent times in disparate populations. However, in different genetic and environmental contexts the phenotypic effect of a particular mtDNA variant can be very different.

Once mtDNA variation has permitted a species' subpopulation to become established in a marginal energetic environment for prolonged periods, mutations that augment energetic adaptation also accumulate in the nDNA-coded bioenergetic genes. If the subpopulation remains indefinitely in the marginal environment, then sufficient nDNA changes will accumulate to render the mtDNA and nDNA variants of the isolated population incompatible with those of the parental population, causing speciation [62,66–73]. Consistent with this supposition, a number of nDNA OXPHOS gene variants have been found to correspond to primate speciation events [34,62,74–77].

As a subpopulation drifts away from the parent population energetically, new environmental energy reservoirs become accessible, frequently in significantly different physical forms. This selects for anatomical changes that optimize exploitation of the new energy reservoir.

Once the new species has arisen to exploit a new energy reservoir, the species will begin to expand its range to overlap that of the new energy resource. This will require the population to expand back into more median global energetic environments. The resulting reduction in the extreme environmental constraints will select for the return of the mitochondrial bioenergetic metabolism back to its more energy-efficient state. This revision back to the more energy-efficient mtDNA genotypes can occur relatively rapidly due to the high sequence evolution rate of the mtDNA and the strong selection for the most efficient OXPHOS system when calorie limitation is the primary selective factor.

Only a portion of the mtDNA polypeptide amino acids are amenable to adaptive mutations, since these mutations must alter the efficiency of the OXPHOS enzyme complex without altering the enzyme's structure or assembly. Since these amino acid substitutions do not affect the capacity of individuals within the species to successfully mate with each other, these mtDNA variants can be maintained within a species and are not the final arbiters of speciation. Thus, these adaptive mtDNA variants can be polymorphic within a species but be essentially invariant between species. The reversion of these adaptive polypeptide variants to a more universal mtDNA genotype following speciation may explain why the mtDNAs from one species can introgress into another very closely related species through a hybridization zone [78].

In addition to the adaptive mtDNA polypeptide variants, the high mtDNA mutation rate also generates mtDNA variants that alter the interaction between the mtDNA and nDNA subunits of the same OXPHOS complex. This class of mtDNA variants can also arise in marginal energy environments and may be beneficial in certain contexts. However, since they reduce the stability of the OXPHOS complexes, they produce a strong selective pressure for the appearance of nDNA mutations in the OXPHOS enzyme polypeptide that interacts with the mtDNA mutant polypeptide to permit optimal enzyme assembly and function. Once such pairs of complementary mtDNA and nDNA OXPHOS polypeptide variants become established in a peripheral environment, they will begin to limit the success of matings of individuals from the peripheral environment with those of the parental environment. Hence, these mtDNA variants can drive speciation via nuclear–cytoplasmic incompatibility [79].

In summary, there are two major classes of evolutionarily relevant mtDNA variants, one that permits expansion of the species range into energetically peripheral environments and another that leads to the genetic isolation of the peripheral population through the accumulation of complementary mtDNA and nDNA OXPHOS enzyme variants. As more of the nuclear–cytoplasmic subunit interaction variants accumulate, the genetic barrier between successful mattings of the parental and peripheral population increases until speciation becomes established. As the new species expands out from its peripheral energy environment, the adaptive mtDNA variants revert to permit the new species to optimally exploit its new energetic niche.

The neutralist–selectionist debate arose because studies of nDNA genetic variation between human populations could not identify the required adaptive variation in the nDNA. The discovery and characterization of the mtDNA evolutionary system resolves this dilemma since it is mtDNA variation that provides the initial high impact variation for intra-specific adaption to different energy environments. The high mtDNA mutation rate permits physiological adaptation in the time span of thousands of years, which is the time frame for intra-specific radiation. Hence, it was mtDNA variation that permitted our ancestors to occupy new energetic environments, which is why mtDNA variation is regional and corresponds to major human geographical transitions.

Occupation of alternative energetic environments then enriched for rare nDNA variants, which consolidated the bioenergetic adaptation of the isolated animal populations to different energetic environments within the species' niche. Subsequent environmental changes and/or migration created gene–environment mismatches and bioenergetic stress that in humans manifests as metabolic disease. The most phenotypically significant adaptive loci generate the greatest functional disconnect and most severe disease when the environment changes. Hence, these are the risk factors for common diseases, not the much milder and ubiquitous variants detected across all populations by GWAS.

The clarification of the bioenergetic genetics of common metabolic diseases indicates that to identify high impact nDNA genetic loci for common diseases it will be necessary to sequence the entire genome of patients and controls from regional populations and look for novel regional genetic variants that correlate with the disease. With these new approaches, high impact loci should be found that will permit presymptomatic diagnosis in regional populations. Such advances will suggest new therapeutic approaches for the treatment and prevention of human disease [48].