Is mitochondrial DNA a strictly neutral marker?

REVIEWS 13 14 15 16 17 18

19 20 21

22 23 24 25

Seger, J. (1988) Philos. Trans. R. Sot. London Ser. B 319,541-555 Gavrilets, S. and Hastings, A. PKIC. R. Sot. London Ser. B (in press) Bell, G. and Maynard Smith, J. (1987) Nature 328,66-68 Namkoong, G. and Selgrade, J.F. (1986) Theor. Popul. Biol. 29,64-86 Auslander, D., Guckenheimer, J. and Oster, G. (1978) Theor. Popul. Biol. 13,276-293 Asmussen, M.A. (1979) Theor. Popul. Biol. 16,172-190 Gouyon, P-H.,Vichot, F. and van Damme, J.M.M. (1991)Am. Nat. 137, 498-514 May, R.M. and Anderson, R.M. (1983) Proc. R. Sot. London Ser. B 219, 281-313 Holt, R.D. (1983) in Aduances in Herpetology and Euolutionary Biology (Rhodin, A.G.J. and Miyata, K., eds), pp. 680-694, Harvard University Museum of Comparative Zoology Nee, S. and May, R.M. (1993) J. Theor. Biol. 161,95-109 Hastings, A., Horn, C.L., Ellner, S.,Turchin, P. and Godfray, H.C.J. (1993) Annu. Reu. Ecol. Syst. 24,1-33 Hassell, M.P., Comins, H.N. and May, R.M. (1991) Nature 353, 255-258 Boerlijst, M.C., Lamers, M.E. and Hogeweg, P. (1993) froc. R. Sot. London

Ser. B 253,15-18

26 Lloyd, D., Lloyd, A.L. and Olsen, L.F. (1992) Biosystems 27, 17-24 27 Dennis, B., Desharnais, R.A., Cushing, J.M. and Costantino, R.F. Ecol. Monogr. (in press) 28 Ferriere, R. and Gatto, M. (1993) Proc. R. Sot. London Ser. B 251,33-38

29 Gatto, M. (1993) Theor. Popul. Biol. 43,310-336 30 Gould, S.J.and Lewontin, R.M. (1979) Proc. R. Sot. London Ser. B 205, 581-598 31 Conrad, M. (1986) in Chaos (Holden, A.V., ed.), pp. 3-14, Manchester University Press 32 Ferriere, R., Cazelles, B., Cezilly, F. and Desportes, J.P.Anim. Behau. (in press) 33 Allen, J.C.,Schaffer, W.M. and Rosko, D. (1993) Nature 364,229-232 34 Thomas, W.R., Pomerantz, M.J. and Gilpin, M.E. (1980) Eco/ogy 61, 13-17 35 Berryman, A.A. and Millstein, J.A. (1989) Trends Ecol. Euol. 4, 26-28 36 Metz, J.A.J.,Geritz, S.A.H.,Iwasa, Y. and Meszena, G. J. Math. Biol. (in press) 37 Ferriere, R. and Gatto, M. Theor. Popul Biol. (in press) 38 Metz, J.A.J.,Nisbet, R.M. and Geritz, S.A.H.(1992) Trends Ecol. Euol. 7, 198-202 39 Rand, D.A. and Wilson, H.B. (1993) Proc. R. Sot. London Ser. B 253, 137-141 40 Foster, D. and Young, P. (1990) Theor. Popul Biol. 38,219-232 41 Zeeman, E.C. (1992) in Understanding Catastrophe (Bourriau, J., ed.), pp. 83-101, Cambridge University Press 42 Green, D.M. (1991) Trends Ecol. Euol. 6,333-337 43 Thomas, R.D.K.and Reiff, W.E. (1993) Euolution 47,341-360 44 Wesson, R. (1991) Beyond Natural Selection, MIT Press

Is mitochondrial DNA a strictly neutral marker? J. William 0. Ballard and Martin Kreitman

M

Variation and change in mitochondrial itochondriaf genes have because of its linkage to the rest of DNA (mtDNA) is often assumed to been employed extenthe genome. Selection need not sively in evolutionary conform to a constant mutation rate even be acting on the mitochonstudies because of their equilibrium neutral model of molecular drial genome itself: any maternally evolution. Recent evidence, however, uniparental mode of inheritance, inherited factor could potentially high rate of evolution and relative indicates that the assumptions underlying influence haplotypic diversity. One simplicity of enzymatic amplifithis model are frequently violated. The such factor is the maternally incation using ‘universal’ primerG. mitochondrial genome may be subject herited rickettsia Wolbachiu, deThey have also been widely used in to the same suite of forces known to be scribed below. The vast majority population studies owing to the acting in the nuclear genome, including of studies employing mtDNA as general belief that gene frequenhitchhiking and selection, as well as an evolutionary marker have not cies are governed primarily by forces that do not affect nuclear variation. attempted to test the basic asWherever possible, evolutionary studies migration and genetic drift, and sumptions of the neutral model: a that most of the variation within involving mtDNA should incorporate constant mutation rate, a stationa species is selectively neutral. statistical tests to investigate the forces ary allele frequency distribution, However, factors other than gen- shaping sequence variation and evolution. and a correlation between polyetic drift are expected to be impormorphism levels and divergence. tant determinants governing the We will first review the evidence fate of mutations. The lack of nor- William Ballard is at The Field Museum, Roosevelt Rd leading to the widespread belief at Lake Shore Drive, Chicago, IL 60605.2496, USA; mal recombination in mitochonthat mtDNA conforms to the neuMartin Kreitman is at the Dept of Ecology and dria means that each genome has tral model, and then discuss recent Evolution, University of Chicago, 1101 E57th St, a single genealogical history and studies in humans, rodents and Chicago, IL 60637, USA. all genes will share that history. Drosophila, where the observed patAny evolutionary force acting at terns of variation have been tested one site will equally affect the history of the whole molecule. against these neutral theory predictions (Table 1). Thus, the fixation of an advantageous mutation by selection, for example, will cause the fixation of all other polymorEvidence for neutrality phisms by a process known as genetic hitchhikings. Even A review of the literature leads us to conclude that the the quickly evolving noncoding origin of replication region widespread acceptance of the selective neutrality of mtDNA cannot be assumed to have neutral allele frequencies follows from a series of plausibility arguments connecting TREE

001.

IO, no.

I2 December

I995

0 1995, Elsevier

Science

Ltd

485

REVIEWS The perceived uniformity in the substitution rate in Table 1. Approaches that can be employed to test the evolutionary dynamics of mtDNA mtDNA also provides sup Approach Test Prediction under neutrality Refs port for the neutrality of mutation+. As mentioned Direct Competition between mitochondrial Haplotype frequency will not change in any 4-i above, mtDNA mutation rates predictable or repeatable way. haplotypes. are not uniformly high, as Phylogenetic Rates of evolution along different The variance should equal the mean after taking 83 thought initially. Gillespie27 into account possible ‘lineage’ effects, such as lineages. compared the rates of evoludifferences in mutation rate. tion of five mitochondrial and Types of mutational changes along The type or pattern of substitution should be the 10,ll four nuclear loci in mammals. different branches of a tree. same on all branches. He noted that while the subWith no recombination and no selection, haplotype Statistical Frequency spectrum of haplotypes. 12-15 stitution rates in the mtDNA frequencies should conform to the neutral infinite alleles distribution. exceeded those in the nuclear genome, there was no eviDistribution of polymorphism within The level of polymorphism and divergence, governed 16 dence for a speed up in the species and divergence between only by genetic drift and selective constraints, will be positively correlated. species for two loci. rate of protein evolution in mitochondrial genes, which Distribution of synonymous and If synonymous and replacement variation is neutral, 17,18 the ratio of polymorphism:divergence will be the replacement changes within and suggests an uncoupling of between species. same when the species compared are closely related. silent site and protein evolution. However, because amino acid replacement rates vary features of mtDNA evolution with (mis)conceptions of considerably among proteins, Gillespie’s conclusion will be neutral theory. One example is the proposal that the transsensitive to the choice of nuclear loci examined. lation apparatus of a small genome, such as mtDNA, encodThere is compelling evidence for variation in evolutioning only 13 polypeptides, is expected to function under ary rates within a lineage. In the lineage to higher primates, more-relaxed constraints than a system that translates the mitochondrially encoded subunit 11of cytochrome oximany thousands of mRNA 19. The primary evidence for dase (COI) has undergone at least a fivefold acceleration in relaxed constraints is the high rate of evolution of mito- the rate of protein evolution relative to rat, mouse and cowlg. chondrially encoded ribosomal and tRNA genes in mam- A parallel rate of acceleration is observed for somatic cytomals. In addition, restriction fragment length polymorphism chrome c, a nuclear encoded protein that interacts directly (RFLP) analysis of 112 human mtDNAs indicates a signifi- with this oxidase subunit in the electron transport chainlg. cantly higher polymorphism level in the tRNA genes recog- This result has lead to the suggestion that the fast protein nizing fourfold degenerate codons than those that recognize evolution is selectively driven in the primate lineage. twofold degenerate codons. This is interpreted as indicating a greater tolerance of translational errors for fourfold Human variation redundant codons compared to twofold redundant Whittam and colleagues28studied haplotype frequencies codonslg. The relaxed constraints hypothesis, however, at 28 ‘loci’ in 145 individuals representing five geographic does not require that genetic drift be the principle mecha- populations. They identified 35 frequency distributions that nism governing variation and change. Rather, relaxed con- they could compare to that expected under neutral mutation straints only mean that fewer mutations are subject to theory assuming an infinite alleles model. Twenty nine perstrong purifying selection. cent of these comparisons led to a rejection of the neutral model at the 2.5% level. An excess of singletons was obAdditional support for the neutrality of mtDNA comes from evidence that the mean rate of divergence averaged served within populations, and may be evidence for expandover the whole molecule is 5-10 times greater (about 2% per ing populations. However, common alleles also occurred in million years) than nuclear DNA (nDNA) in a variety of animals higher than expected frequencies, suggesting positive seincluding primates’, rodent9 and Hawaiian Drosophila21. lection but not population expansion. Excoffierzg compared This apparent support for neutrality, however, is based on mtDNA haplotype frequencies in 31 human populations to another misappropriation of a neutral theory tenet. Under that simulated under the sampling theory of selectively neuneutrality, the expected rate of evolution will be equal to the tral alleles. The data for Oriental, but not African, popumutation rate to neutral alleles. Thus, the rate of neutral lations contained one common allele. Excoffier was careful molecular evolution is expected to be inversely related to not to favor a selectionist hypothesis over alternatives such the functional importance of a particular nucleotide site or as population history. However, selection against deleterigene.*By this argument, Kimura22 showed that the rates of ous mutations or positive selection favoring the common evolution of first (kl), second (k2) and third (k3) positions of allele could have contributed to the non-neutral frequency codons fit the neutral theory prediction, k2 < kl