View metadata, citation and similar papers at core.ac.uk brought to you by CORE providedDispatch by ElsevierR487 - Publisher Connector

Evolution: Setting the mutation rate Paul Sniegowski

A recent study of X-chromosome and autosome genes A prediction of either trade-off theory outlined above is in mammals suggests that selective trade-offs are that a lower equilibrium mutation rate should evolve if the important in the long-term evolution of mutation rates; deleterious effect of mutation is increased. A recent paper but recent studies with bacteria show that high by McVean and Hurst [5] has provided evidence consis- mutation rates can nonetheless evolve in the short term tent with this prediction by comparing inferred rates of in clonal populations. mutation on X chromosomes and autosomes in mouse and rat. Because the X chromosome is hemizygous in mam- Address: Department of Biology, University of Pennsylvania, Philadelphia, Pennysylvania 19104, USA. malian males, deleterious recessive or partially recessive mutations arising on it will have stronger immediate Current Biology 1997, 7:R487–R488 effects on fitness than those arising on the autosomes. http://biomednet.com/elecref/09609822007R0487 Thus, the X chromosome might be predicted to have © Current Biology Ltd ISSN 0960-9822 evolved a lower mutation rate than the autosomes. To test this idea, McVean and Hurst compared the average rate of The great majority of spontaneous mutations are likely to synonymous nucleotide substitutions (those that do not be deleterious, suggesting that the genomic mutation rate change an encoded amino acid) in 33 X-linked and 238 is under constant pressure to decrease by natural selection. autosomal genes previously sequenced in mouse and rat. Why, then, does the mutation rate not evolve to zero [1]? If neutral with regard to fitness, synonymous substitutions One explanation might be that zero mutation rates are accumulate at a rate equal to the mutation rate and can be unattainable given universal physical constraints. That is, used to infer rates of mutation. McVean and Hurst found perhaps random processes at the molecular or submolecu- that X-linked genes showed significantly lower rates of lar level make it impossible to replicate hereditary mole- synonymous substitution than autosomal genes, and they cules with absolute fidelity or to repair them completely. concluded that the mutation rate on the X chromosome This simple non-evolutionary explanation, however, is has been reduced by natural selection. unable to account for the observation that per-nucleotide mutation rates vary widely among different species [2]; McVean and Hurst [5] considered two alternative explana- clearly, universal physical constraints do not prevent lower tions for their results. One is that synonymous substitu- mutation rates from evolving in at least some species. tions in rodents might not be neutral: instead, they might be slightly deleterious. If this were the case, then purify- It seems that the mutation rate is set by a trade-off between ing selection would tend to remove such mutations from natural selection favouring lower mutation rates and oppos- the X chromosome more efficiently than from the auto- ing selective forces favouring higher mutation rates. The somes, leading to an incorrect inference of lower mutation two possible outcomes of this trade-off have been termed rate on the X chromosome. Evidence from previous ‘minimal’ and ‘optimal’ mutation rates [3]. Mechanisms that studies, however, has strongly suggested that synonymous maintain the fidelity of replication presumably operate at substitutions in rodents are effectively neutral [6], and so some physiological cost to individuals. The mutation rate this explanation seems unlikely. The alternative explana- may evolve to a minimal level that is constrained by this tion is that the inference of lower mutation rates on the X cost of fidelity, such that individuals with mutation rates chromosome is misleading because males have higher higher and lower than this minimal level are less fit. Alter- mutation rates than females. If this were the case, then as natively, selection at the level of populations or species the X chromosome spends less time in males than the might favour the maintenance of a certain rate of mutation, autosomes it would only appear to have evolved a lower because lineages that produce more variation have an adap- rate of mutation. However, male mutation rates far in tive advantage over the evolutionary long run. Thus, popu- excess of previous estimates for rodents would be required lations with lower mutation rates would perish for lack of to explain the observed difference in synonymous substi- sufficient adaptive variation, while those with higher muta- tution rates between the X and the autosomes, so this tion rates would suffer from an excessive load of deleterious explanation also seems unlikely. mutations [4]. Although the notion that mutation rates might be adjusted optimally for adaptation is attractive, it is It will be interesting to see whether similar evidence for quite plausible that the ‘minimal’ rate, as set by the cost of selectively favourable reductions in mutation rate emerges fidelity in most species, is higher than the ‘optimal’ rate from studies of other species for which sequence infor- required for long-term adaptive success [4]. mation is available. There is very little other empirical R488 Current Biology, Vol 7 No 8

evidence on the role of trade-offs in the evolution of very rare mutators will sometimes hitch-hike to high fre- mutation rates. Drake [2] has noted a remarkable invari- quencies in asexual populations [10]. This has been cor- ance of genomic mutation rate in microbes ranging from roborated by an evolution experiment in which a subset phages to yeast, but a selective explanation for this of replicate E. coli populations fixed mutator alleles that pattern is not obvious. A more direct indication of trade- had themselves arisen as a consequence of mutation [11]. offs comes from the less well-known results of an evolu- Furthermore, mutator alleles have been recently associ- tionary experiment in Drosophila. Nöthel [7] exposed ated with pathogenicity in E. coli and Salmonella isolates laboratory D. melanogaster populations to X-irradiation for [12] and with certain cancers [13]. In this regard, the evo- up to 600 generations, thereby creating a situation in lution of mutation rates—a seemingly esoteric problem which increased repair of X-ray-induced DNA damage in population genetics—appears to have important health would be favoured by natural selection. Irradiated popula- implications [14]. tions evolved lower rates of X-ray-induced lethal mutation than control (unirradiated) populations, and populations References exposed to higher X-ray dosages evolved lower mutation 1. Sturtevant AH: Essays on evolution. I. On the effects of selection on mutation rate. Q Rev Biol 1937, 12:464-467. rates than those exposed to lower dosages. Furthermore, 2. Drake JW: A constant rate of spontaneous mutation in DNA-based when irradiation was stopped, some populations evolved microbes. Proc Natl Acad Sci USA 1991, 88:7160-7164. 3. Maynard Smith J: The Evolution of Sex. Cambridge: Cambridge back toward their previous, higher levels of mutability, University Press; 1978. suggesting that a fitness cost was associated with decreas- 4. Kimura M: On the evolutionary adjustment of spontaneous ing this type of mutation in the normal environment. mutation rates. Genet Res Camb 1967, 9:23-34. 5. McVean GT, Hurst LD: Evidence for a selectively favourable reduction in the mutation rate of the X chromosome. Nature 1997, A key feature of population genetic models for the evolu- 386:388-392. tion of mutation rates is the presence or absence of recom- 6. Wolf KH, Sharp PM: Mammalian gene evolution: nucleotide sequence divergence between mouse and rat. J Mol Evol 1993, bination between alleles that modify the mutation rate — 37:441-456. for example, alleles of loci involved in DNA replication, 7. Nöthel H: Adaptation of Drosophila melanogaster populations to high mutation pressure: evolutionary adjustment of mutation proofreading and repair — and fitness mutations at other rates. Proc Natl Acad Sci USA 1987, 84:1045-1049. loci in the genome. With no recombination, the fate of a 8. Leigh EG: Natural selection and mutability. Am Nat 1970, modifier allele conferring a higher mutation rate (a 104:301-305. 9. Chao L, Cox EC: Competition between high and low mutating mutator) is determined by the fitness effects of the spe- strains of Escherichia coli. Evolution 1983, 37:125-134. cific mutations with which it is associated. Under these 10. Taddei F, Radman M, Maynard Smith J, Toupance, B, Gouyon, PH, circumstances, a mutator associated with a rare beneficial Godelle B: Role of mutator alleles in adaptive evolution. Nature, 387:700-702. mutation can rise to high frequency — ‘hitch-hike’ — 11. Sniegowski PD, Gerrish PJ, Lenski RE: Evolution of high mutation along with that beneficial mutation, provided that the rates in experimental populations of Escherichia coli. Nature, 387:703-705. fitness advantage of the beneficial mutation outweighs 12. LeClerc JE, Li B, Payne WL, Cebula T: High mutation frequencies any disadvantage of the mutator. By repeatedly generating among Escherichia coli and Salmonella pathogens. Science 1996, new beneficial mutations at a higher rate than the wild 274:1208-1211. 13. Modrich P: Mismatch repair, genetic stability and tumour type, a mutator can retain its advantage and remain at high avoidance. Phil Trans Roy Soc Lond Ser B 1995, 347:89-95. frequency. Recombination changes the situation, 14. Cox EC: mutS, proofreading, and cancer. Genetics 1997, however, by randomizing associations between mutators 146:443-446. and fitness mutations. With free recombination, selection on all copies of a mutator comes to reflect the overall dele- terious effect of increasing the mutation rate, rather than the effects of particular fitness mutations. The modifier models thus predict that optimal mutation rates will evolve only in asexual populations, whereas in sexual pop- ulations a minimal mutation rate will generally be favoured, regardless of what the optimal rate might be [8].

Whether mutation rates are optimal or minimal in general remains unresolved, but experimental observations have clearly demonstrated the potential for high mutation rates to evolve over the short term in asexual populations. Studies using laboratory Escherichia coli populations [9] have shown that mutator strains, when introduced into populations at relatively high frequencies, will displace wild types by hitch-hiking with beneficial mutations. Recent computer simulations have predicted that even