Evidence for the Adaptive Evolution of Mutation Rates

Adaptive evolution has long been regarded as the result of postmutational sorting by the process of natural selection. Mutations have been postulated to occur at random, producing genetically different individuals that then compete for resources, the result being selection of better adapted genotypes. Molecular biology has demonstrated, however, that the rate and spectrum of mutations is in large part under the control of genetic factors. Because genetic factors are themselves the subject of adaptive evolution, this discovery has brought into question the random nature of mutagenesis. It would be highly adaptive for organisms inhabiting variable environments to modulate mutational dynamics in ways likely to produce necessary adaptive mutations in a timely fashion while limiting the generation of other, probably deleterious, mutations. Have genetic systems emerged that tune mutation rates and spectra in adaptive ways? Such tuning would produce mutations with a greater chance of being adaptive than if they were completely random. This possibility has met with resistance from theorists who point out that such systems may violate a basic tenet of evolutionary theory—evolution does not involve foresight (3Dickenson W.J Seger J Nature. 1999; 399: 30Crossref PubMed Scopus (30) Google Scholar). Here, we review recent evidence for the existence of adaptively tuned mutation rates. We conclude that these mechanisms do not require any special foresight. Instead, they must have been selected for repeatedly in the past for their ability to generate genetic change. Mutational tuning does not require the specific generation of adaptive mutations (nonrandomness with respect to function) but rather the concentration of mutations under specific environmental conditions or in particular regions of the genome (nonrandomness with respect to time or location). Given a predictably variable environment, adaptively tuned mutation rates can evolve in ways completely consistent with the modern synthetic theory of evolution. Changes in mutation rate can be either environment dependent or heritable. Environment-dependent changes result from induction or suppression of mutator mechanisms that are present in all individuals and that have global (genome-wide) effects on mutation rate. Heritable mutator mechanisms are independent of the environment. They may act by altering the global mutation rate or they may be local, taking the form of gene sequences that experience unusually high or low rates of specific types of mutation. Environment-dependent global mutators are responsible for adaptive mutations of the type that are produced under starvation conditions in Escherichia coli (15Rosenberg S.M Thulin C Harris R.S Genetics. 1998; 148: 1559-1566Crossref PubMed Google Scholar). Loss-of-function mutations in genes required for growth can back-mutate, allowing survival and reproduction. The rate of these back-mutations (as well as of other, nonadaptive mutations) increases greatly with starvation. This mutator phenotype is not itself heritable—the survivors produce progeny cells with normal mutation rates. The SOS response has been implicated in such adaptive mutation (15Rosenberg S.M Thulin C Harris R.S Genetics. 1998; 148: 1559-1566Crossref PubMed Google Scholar). The SOS system is a DNA repair mechanism induced when the E. coli genome suffers physical damage such as double-strand breaks, and can be triggered by starvation. It includes alternative DNA polymerase enzymes, such as DinB and the UmuD′2C complex, which are capable of replicating badly damaged sequences. In the process of repair, these alternative polymerases generate mutations at a high rate. DinB is known to generate mutations at undamaged sites when SOS is induced. It has been suggested that DinB and similar enzymes have evolved to produce genome-wide genetic variation under stressful conditions, allowing organisms to adapt rapidly when necessary (14Radman M Nature. 1999; 401: 866-869Crossref PubMed Scopus (107) Google Scholar). Homologs of these alternative polymerases with similar functions have been identified in Saccharomyces cerevisiae, and homologs have also been identified by sequence similarity in mice and humans (for review see 4Friedberg E.C Gerlach V.L Cell. 1999; 98: 413-416Abstract Full Text Full Text PDF PubMed Scopus (120) Google Scholar). Another potential mechanism for generating variation under stressful conditions has been recently described in a eukaryote. Some of the genetic variation that is capable of affecting development in Drosophila is masked under normal circumstances. This masking is accomplished by the chaperone properties of the heat-shock protein Hsp90. This protein stabilizes signal transduction factors that regulate development. Mutational or pharmacological impairment of Hsp90 reveals otherwise unexpressed developmental variation (16Rutherford S.L Lindquist S Nature. 1998; 396: 336-342Crossref PubMed Scopus (1636) Google Scholar). These laboratory manipulations mimic the potential effects of heat stress or other protein-damaging conditions in the natural environment, conditions under which Hsp90 is diverted from the proteins it normally stabilizes to other environmentally denatured proteins. Exposed developmental variants can be fixed in the population by selection. Environment-dependent mutators operate when genetic change is necessary (when the organism is maladapted to its current environment). Moreover, the offspring are not saddled with an increased rate of deleterious mutation. The essential evolutionary question is this: did systems such as SOS and Hsp90 evolve to produce or reveal genetic variation under stress, or is this potentially adaptive property an accidental byproduct of other factors? Such an evolutionary accident has been termed a spandrel, by analogy with the spaces (spandrels) that result whenever a round dome is supported by a square formed by four arches. These spaces were secondarily used as aesthetically stunning venues for mosaic portraits in St. Mark's Cathedral and other buildings. Although they provide an optimal form for these mosaics, it was not their original purpose. In evolutionary terminology, the adaptive use of a function selected for another purpose is known as an exaptation. The DNA polymerases involved in SOS-associated mutagenesis are, by necessity, less specific than their constitutive counterparts. They must be able to copy DNA that is so severely damaged that it interferes with the action of higher-fidelity constitutive polymerases. Indeed, the mutations introduced by the SOS-associated UmuD′2C are the result of repair activities. The tendency for DinB to introduce mutations at undamaged sites may also be a consequence of low fidelity that evolved to facilitate repair. Thus, the apparently adaptive hypermutagenesis of the SOS response may simply be a spandrel—the byproduct of functions evolved for repair. Similarly, it is impossible to tell whether the Hsp90 system evolved under selective pressure to limit developmental variability in the absence of stress, to provide variability in the presence of stress, or both. It is not clear whether any of the variability revealed under stress through the action (or inaction) of Hsp90 is of a potentially adaptive nature nor has the ability of heat-stressed flies to adapt more rapidly to environmental changes been tested under circumstances designed to mimic natural adaptation. Heritable global mutators have been detected in short- and long-term laboratory selection experiments (17Sniegowski P.D Gerrish P.J Lenski R.E Nature. 1997; 387: 703-705Crossref PubMed Scopus (608) Google Scholar, 10Miller J.H Suthar A Tai J Yeung A Truong C Stewart J.L J. Bacteriol. 1999; 181: 1576-1584Crossref PubMed Google Scholar), where they tend to outcompete nonmutators. Surveys of pathogenic strains of E. coli and Salmonella reveal much higher proportions of mutator strains in natural populations than would be expected to arise by chance (9LeClerc J.E Li B Payne W.L Cebula T.A Science. 1996; 274: 1208-1210Crossref PubMed Scopus (645) Google Scholar). Studies of HIV (19Wainberg M.A Drosopoulos W.C Salomon H Hsu M Borkow G Parniak M Gu Z Song Q Manne J Islam S et al.Science. 1996; 271: 1282-1285Crossref PubMed Scopus (288) Google Scholar) and Yersinia (13Najdenski H Iteman I Carniel E Contrib. Microbiol. Immunol. 1995; 13: 281-284PubMed Google Scholar) have demonstrated positive correlations between global mutation rates and virulence. Bacterial mutator phenotypes have been traced to mutations in DNA repair genes, particularly the methyl-directed mismatch repair (MMR) system. Mutators have also been found in HIV reverse transcriptase. The fixation of heritable global mutator systems requires second-order selection; the mutator can only be spread through the population through the production of linked advantageous mutations (20Weber M Biol. Phil. 1996; 11: 67-88Crossref Google Scholar). Second-order selection is facilitated by asexual reproduction, which maintains linkage. Both laboratory experiments and modeling studies have shown that increased rates of mutation are only favored when the normal mutation rate is the limiting factor in adaptation and when the selective advantage of possible adaptive mutations is high (2deVisser J.A.G.M Zeyl C.W Gerrish P.J Blanchard J.L Lenski R.E Science. 1999; 283: 404-406Crossref PubMed Scopus (268) Google Scholar). Pathogenic prokaryotes may be in a particularly good position to take advantage of mutator phenotypes. Their populations experience extreme bottlenecks that reduce existing genetic variability, and individual mutations can have exceptional advantages. Nonetheless, in asexual populations there is a limit to the contribution that increases in mutation rate can make to the rate of adaptation. This is because multiple adaptive mutants coexisting in the same asexual population but in different individuals cannot be simultaneously brought to fixation. They compete with each other for fixation, a phenomenon known as clonal interference (2deVisser J.A.G.M Zeyl C.W Gerrish P.J Blanchard J.L Lenski R.E Science. 1999; 283: 404-406Crossref PubMed Scopus (268) Google Scholar). Heritable global mutators, like environment-dependent mutators, would appear to be adaptive under certain circumstances. While heritable mutators are likely to arise entirely by chance, they may be carried to high frequency through second-order selection acting through valuable mutations that they generate. Modeling indicates that this process can fix mutators (18Tenaillon O Toupance B Nagard H.L Taddei F Godelle B Genetics. 1999; 152: 485-493PubMed Google Scholar). In the real world, however, complete fixation of mutators is rare. Bacterial global mutators are usually found in populations consisting primarily of nonmutators. HIV provides further evidence for the optimization of global mutation rates through selection. When HIV-1 infections are treated with the nucleotide analog 3-TC, the virus evolves to a drug-resistant state through mutations that increase the fidelity of reverse transcriptase and lower the mutation rate of the virus. 3-TC resistance arises readily, suggesting that the mutations responsible for resistance (and therefore for decreased rates of mutation) are common, yet these mutations are neither fixed nor maintained in the absence of 3-TC selection (19Wainberg M.A Drosopoulos W.C Salomon H Hsu M Borkow G Parniak M Gu Z Song Q Manne J Islam S et al.Science. 1996; 271: 1282-1285Crossref PubMed Scopus (288) Google Scholar). In fact, there is evidence for a selective force that maintains a nonminimal mutation rate in HIV-1. Strains resistant to 3-TC are less virulent and less able to evolve resistance to other antiviral drugs than are 3-TC-sensitive strains. Even in this case, the causal link between a need for rapid adaptation and the evolution of nonminimal mutation rates is circumstantial. The evidence for the adaptive optimization of mutation rates is much stronger in the case of heritable local mutators. All known local mutators are the result of unique sequence characteristics that predispose specific regions of the genome to particular types of mutation. The best examples come from investigations of pathogenesis, presumably because pathogenesis results in very strong selective pressures favoring high rates of specific types of mutations. Hypermutable sequences generate antigen diversity in pathogens and antibody diversity in hosts. The best known of these are the “contingency loci” of pathogenic prokaryotes, which produce highly specific tuning of mutation rates in particular genes. Contingency loci are antigen- and phase-determinant genes in pathogenic bacteria that switch between functional (on) and nonfunctional (off) states at very high rates (Figure 1). In many cases, the genes being switched on and off are primary antigenic determinants, resulting in a pathogen that can vary its antigenic signature abruptly without suffering an increased rate of mutation elsewhere. This switching behavior results from the mutational properties of tandem repetitive DNA (microsatellites) located within the gene or its associated controlling elements (12Moxon E.R Rainey P.B Nowak M.A Lenski R.E Curr. Biol. 1994; 4: 24-33Abstract Full Text Full Text PDF PubMed Scopus (559) Google Scholar). Microsatellites experience high rates of single-motif insertion and deletion mutations through replication slippage. This generates a high rate of alternating loss-of-function and reversion mutations, which is highly adaptive for the pathogens that carry them. Although the genes that are controlled by contingency repeats in different organisms are often involved in anti-genicity, the proteins they code for are very diverse, and the repetitive DNAs driving hypermutability are diverse as well (11Moxon E.R Wills C Sci. Am. 1999; 280: 72-77Crossref Scopus (82) Google Scholar). Most contingency repeats are the longest microsatellite DNAs in their respective prokaryotic genomes, suggesting that they did not arise at these locations by chance. Thus, repetitivity itself is evolving in these sequences, and microsatellites have independently been selected for their mutability many times in response to similar selective pressures. Local mutator systems have evolved in eukaryotic organisms, but some are very different from contingency loci (1Deitsch K.W Moxon E.R Wellems T.E Microbiol. Mol. Biol. Rev. 1997; 61: 281-293Crossref PubMed Scopus (196) Google Scholar). Many of these systems do not generate heritable, germline mutations but rather employ mutational mechanisms that effectively decouple the consequences of hypermutation from the process of long-term evolution. The best characterized example is mammalian antibody diversification by somatic hypermutation. Other systems, such as the cassette-switching system involved in trypanosome antigenic hypervariability, do give rise to heritable mutations, but these mutations too are strictly confined to certain parts of the genome (Figure 2). These systems are much more complex and highly regulated than bacterial contingency genes, and it is clear that they have evolved specifically to increase the rate of mutation in localized parts of the genome. Other eukaryotic local mutators, such as the CpG island-associated gene conversions generating diversity at the MHC locus, generate germline genetic changes in a manner more similar to that seen in contingency loci (Figure 2). The existence of all of these systems demonstrates that adaptive evolution can produce highly specific mutators in eukaryotes. The high mutation rates experienced in prokaryotic contingency genes and in their eukaryotic counterparts are determined only by the unique sequence characteristics of the genes themselves. The evolution of local mutators does not depend on linkage or population size, factors that can interfere with fixation of global mutators. Local mutators also carry no cost resulting from increased rates of deleterious mutation elsewhere in the genome. The evolution of adaptively variable antigenic determinants in pathogens generates tremendous selective pressure on their hosts to evolve adaptively variable immune defenses, and vice versa. This is an optimal situation for the evolution of mutators (6Haraguchi Y Sasaki A J. Theor. Biol. 1996; 183: 121-137Crossref PubMed Scopus (40) Google Scholar). Factors other than host-pathogen interactions may be spurring the evolution of local mutators. Coding-region triplet-repeat microsatellites, which are both common and hypervariable in higher eukaryotes, may provide genetic variability that allows selective fine tuning of protein function (8King D.G Soller M Kashi Y Endeavour. 1997; 21: 36-40Crossref Scopus (97) Google Scholar). Single-repeat length mutations in these microsatellites alter the length of repetitive amino acid tracts within protein products without generating frameshift mutations. Among humans, some of the longest coding triplet-repeat microsatellites occur in or near proteins that affect neural function, and some of these are long enough to become highly unstable, resulting in genetic disorders such as Fragile X syndrome, Huntington's disease, and spinocerebellar ataxia. Because the human brain has recently undergone considerable directional selection, it is possible that the deleterious effects of these long microsatellites have been counterbalanced, either by second-order selection acting through highly adaptive mutations generated by these hypermutable repeats or by hitchhiking with strongly selected adaptations generated independently of the microsatellites but tightly linked to them. It may soon be feasible to distinguish between these possibilities, through comparisons between the human and chimpanzee genomes. Local mutators do not predict the future, even though they do preferentially produce the very types of mutations likely to increase survival. Hypervariability evolved in antigenic contingency genes because it allowed the ancestors of today's pathogens to repeatedly escape immune clearance by their hosts. Because host immune responses confront pathogens with variable environments that have strong elements of predictability, mutational mechanisms in the pathogens have evolved to generate an appropriately variable antigenic repertoire. The hosts' immune systems have also evolved local mutators of various kinds, in response to pathogen variability. The key is the element of predictability in these interactions. If general environmental change also has elements of predictability, global mutation rates may have evolved toward optimal levels, but such optimization has not yet been clearly demonstrated.