Christopher D. Bayliss is a Research Assistant and E. Richard Moxon is a professor at Oxford University, Oxford, United Kingdom.

Links to Other ASM Pages:


• Table of Contents
• ASM News Issues

Hypermutation and Bacterial Adaptation

High rates of localized or generalized mutation help some pathogens adapt to host environments and may enhance virulence

Christopher D. Bayliss and E. Richard Moxon

Scientists widely accept the classical Darwinian concept that selection of mutant cells enables bacterial populations to adapt to changing environments. A logical, but less obvious component of this theory is that the genetic mechanisms for producing mutants are themselves under selection. Genetic mechanisms that produce a hypermutable state can result in variations in mutation rates between genes and genetic elements of a genome and also among individuals within a population.

Appreciating how these mechanisms of gene variation operate in bacteria is critical to understanding differences between commensal and virulence behaviors. To approach this task, we focused on two types of hypermutable states—one localized, and referred to as contingency loci, and the other generalized, involving mutators. We suppose that evolution of these states reflects selective pressures on the fitness of bacterial populations amid their interactions with the exceptionally dynamic spatial and temporal environments of their hosts. We also presume that hypermutation arose because classical gene regulation provides bacteria with a limited, even if often pivotal, ability to adapt to sometimes catastrophic alterations in the host environment resulting from, for example, the diverse immune responses of B and T cells, bacteriophages, and antibiotics.

Indeed, this genetic arms race between microbe and host is the context in which hypermutation evolves. For some pathogens, especially those that infect only a single host species, high rates of localized and generalized mutation are implicated in the evolution of virulence.

Constitutive Mutation Rates Are Inadequate for "Rapid" Adaptation

When we observe mutation rates, we need to consider not only the rate of intrinsic replication errors and the activities/physiological costs of repair mechanisms, but also the rate of those deleterious mutations that reduce growth rates. Jan Drake of the National Institute of Environmental Health in Research Triangle Park, N.C., argues that these factors are balanced, meaning all DNA-based microbes have a similar "constitutive genomic mutation rate" of 0.003 mutations/genome/replication.

Figure 1

Mutants arising at these rates can generate significant genetic diversity when bacterial population sizes are large. However, when bacterial pathogens reside within hosts, especially during the early stages of an infection, population sizes are limited because of bottlenecks, nutrient availability, and constraints from competing flora. Constitutive mutation rates thus may not be adequate to generate sufficient numbers of variants that, through natural selection, would enable such bacteria to adapt to the host (Fig. 1). Under these circumstances, increasing the mutation rate and the production of phenotypic variants within a population would lead to a fitness gain.

Mutation rates in bacteria may sometimes be localized to a particular region of the genome or occur throughout the genome. Both modes of hypermutation can elevate mutation rates within small populations of bacterial cells such that significant numbers of variants are produced to ensure survival under adverse environmental conditions (Fig. 1).

Regulating Hypermutation: Existing Versus Generated Variations

For organisms to survive when they encounter bactericidal environments, they may well need to arrive with resistant variants already present within the population—particularly in those cases in which effector mechanisms extinguish susceptible organisms virtually instantaneously. Because constitutive hypermutation increases the number of variants within a bacterial population, it also increases the likelihood that a population contains a resistant variant (Fig. 1A).

In contrast, Miroslav Radman of Institut Jacques Monod, Paris, France, points to the important role of regulatory pathways that enhance the mutation frequency of bacteria when their survival is threatened. This regulatory mode of adaptive behavior, through an induced hypermutable state, also enhances fitness in the face of nonlethal environmental challenges (Fig. 1B). But since there is a latent period before the response occurs, these mechanisms cannot, per se, deal with precipitous bactericidal threats (Fig. 1A).

Mechanisms of Global Hypermutation

Figure 2

Although other mechanisms may increase mutation rates throughout the genome, the best characterized mechanisms are those affecting either DNA replication or DNA repair (Fig. 2). Perturbing DNA replication increases mutation rates by elevating the replication error rate, but usually at the price of reduced doubling times. Perturbing DNA repair is more complex. Prokaryotes have multiple mechanisms for removing mutation-inducing damaged DNA and replacing it with correct nucleotides. These pathways repair many different types of DNA lesions, and some organisms encode multiple DNA repair proteins with overlapping functions. Mismatch repair (MMR) represents one major DNA repair pathway whose inactivation can increase the mutation rate 1,000-fold. MMR mutants are particularly interesting because they elevate mutation rates involving all nucleotides within a genome without reducing growth rates.

DNA polymerases that copy DNA less accurately than do replicative DNA polymerases represent another global hypermutation pathway (Fig. 2). In Escherichia coli, the bacterial enzymes Pol IV and Pol V, along with Pol II, are induced during an SOS response (a response to DNA damage). Their primary function apparently is to repair damaged DNA. But these enzymes can also introduce mutations into undamaged DNA. For instance, overexpression of Pol IV elevates frameshift mutation rates, most commonly in mononucleotide repeat tracts.

Does Constitutive Global Hypermutation Accelerate Adaptation?

Figure 3

Constitutive global hypermutation results when a DNA metabolism gene is either reduced in activity or inactivated. The increased rate of mutations in these "mutators" may facilitate a bacterial population adapting to novel changes in its environment. However, mutators generally produce more deleterious than beneficial mutations (Fig. 3). What is the evidence that mutators contribute positively to fitness?

Modelling shows that mutators can become the dominant phenotype in a population when the mutator mutation coexists with one or more beneficial mutations—termed hitchhiking. When genetic exchange is incorporated into such models, the rate at which mutators become fixed is reduced because the mutator separates from other beneficial mutations. Moreover, recombination accelerates adaptation because beneficial mutations recombine into a single genome. In small populations consisting of 10⁵-10⁶ cells, however, high global mutation rates have a larger impact on adaptation time than do high rates of recombination.

In experiments conducted about five years ago, mutators generated in batch cultures of E. coli became fixed between about 2,000 and 10,000 generations. Paul Sniegowski of the University of Pennslyvannia, Philadelphia, and Philip Gerrish and Richard Lenski of Michigan State University, East Lansing, attributed these results to "chance hitchhiking events." Subsequently, deleterious mutations were demonstrated to accumulate at a higher rate within a constitutive mutator strain compared to a wild-type E. coli strain during passage through single-cell bottlenecks. When mice were infected with constitutive mutator strains of E. coli, the mutators adapted more rapidly to mice guts during the initial stages of infection than did wild-type strains. However, transmission of such mutator strains to uninfected animals was significantly reduced compared to wild-type strains, with the presumed route of transmission being through fecal material.

Mutator strains are relatively common in natural populations of E. coli, Salmonella typhimurium, and Neisseria meningitidis. Intriguingly, cystic fibrosis patients who were treated with a rigorous regime of antibiotics were colonized with strains of Pseudomonas aeruginosa harboring mutations in MMR genes. Other bacterial species, such as Campylobacter jejuni and Helicobacter pylori, lack recognizable MMR pathways.

Analyses of mutS sequences from diversely representative strains of E. coli indicate a large amount of heterogeneity, which is indicative of frequent recombination. Apparently, mutS has undergone repeated cycles of inactivation and restoration within E. coli, according to Radman of Institut Jacques Monod and Thomas Cebula of the Center for Food Safety and Applied Nutrition at the Food and Drug Administration (FDA) in Washington, D.C. They speculate that mutS mutations produce a hypermutable state that accelerates adaptation and evolution.

Theoretical models, experiments, and observations of bacteria from natural sources indicate that constitutive global hypermutators can help bacterial populations adapt to environmental fluctuations. However, constitutive mutators might arise and persist only on a limited basis, namely due to hitchhiking in those populations that are under constant selection. Mutators are also at a disadvantage during transmission events involving different environments and/or small subpopulations because of the accumulation of deleterious mutations (Fig. 3).

Is Transient Global Hypermutation Environmentally Regulated?

Nonheritable mechanisms may generate transient increases in bacterial mutation rates—these may include exposure to mutagens, elevations in DNA replication error rates when precursors are depleted, and nonregulated, environmentally driven decreases in levels of critical DNA replication/repair proteins. Cells may enter such a transient mutator state stochastically, which would lead individual cells within a population to exhibit differences in mutation rates. However, whether nonheritable hypermutation mechanisms could significantly alter mutation rates to generate adaptive mutations is not known.

Evidence of heritable mechanisms for generating transient states of hypermutation has come from studies of lac mutant E. coli cells starved on lactose medium. In these nongrowing (stationary-phase) cells, the MutL protein becomes limiting, resulting in global hypermutation and adaptive mutations in the lac gene. Overexpression of this protein reduces the number of these adaptive mutations. Similarly, when the SOS response is induced in these cells, leading to overexpression of Pol IV, more of these adaptive mutations are produced. These examples of environmentally driven and, presumably, heritable alterations in protein expression leading to increased mutation rates indicate that nutrient deprivation can induce a transient hypermutable state. Up-regulating mutation rates within a bacterial population facing stressful environments increases that population's "options for survival," according to Susan Rosenberg of Baylor College of Medicine in Houston, Tex.

Localized Hypermutation Mechanisms

Phenotypic consequences of localized hypermutation include changes in either the structure or expression of specific determinants. Many mechanisms lead to localized hypermutation, including those dependent on homologous recombination, site-specific recombination, or simple sequences (Fig. 2B).

Homologous recombination drives localized hypermutation when there are multiple copies of a gene present in a single genome. Recombination between identical copies of a duplicated gene produces variants containing extra or fewer copies of a gene, leading either to higher or lower expression of the gene product. Intrachromosomal recombination (or gene conversion) between nonidentical copies of a gene generates new or mutant versions of a gene and an altered gene product. The multiple structural alterations generated by this mechanism enable bacteria to escape immune responses to particular determinants.

In site-specific recombination, specialized recombinases mediate genetic exchanges between specific sequences. This form of recombination results in localized hypermutation when it involves inversions of DNA sequences within a genome and consequent alterations in expression of genes. Rates of site-specific recombination are controlled by the amounts of recombinases and accessory proteins, providing the potential for environmental regulation of adaptive responses mediated by these pathways.

The third mechanism involves "simple sequences," which are DNA segments containing multiple repeats of an identical sequence, often termed microsatellites or minisatellites, with unit sizes ranging from one to many nucleotides. These sequences are prone to ``slippage'' mutations that result in gains or losses of repeat units. These mutations mediate changes in gene expression by altering promoter activity or the reading frame of a particular gene. Mutation rates of simple sequences depend on the length of the repeat tract, status of the replication machinery, and whether slippage mutations are subject to repair pathways such as MMR. This dependency of mutation rate on tract length means that mutation rates of individual loci might undergo significant changes.

Is Localized Hypermutation Limited in Generating Adaptation?

Localized hypermutation occurs in loci that are under strong selective pressure to mutate. Crucially, localized hypermutation often results from mutations with a high frequency of reversion. Therefore, it can mediate ON/OFF switches of gene expression in time scales that, although longer than those of gene regulation, affect small bacterial populations. This reversible phenotype appears to be a potent mechanism for generating adaptive mutations, but has this benefit actually been demonstrated?

Phase variation rates for individual loci suggest that large numbers of variants can be generated in small bacterial populations. Our mathematical model indicates that experimentally measured phase variation rates will generate small populations (1 x 10⁶ cells) containing organisms that have undergone switches in two different loci and also larger populations (1 x 10⁹ cells) with switches in three loci. These population sizes equate with those postulated as being necessary for some bacteria to invade, colonize, and persist in normal habitats. These results and analyses suggest that localized hypermutation contributes significantly to adaptation by small populations.

In experiments, showing that localized hypermutation is an adaptive mechanism depends primarily on inactivating or deleting hypermutable loci to reduce virulence. In many cases, however, hypermutable loci exhibit some degree of redundancy. That is, genomes often contain several loci encoding proteins with similar functions. This phenomenon often makes it difficult to prove whether a particular locus is required for pathogenesis.

Hence, attention focuses on certain hypermutable loci that mediate switching between two different phenotypes, whose presence implies the existence of selective pressures for rapid switching between those two states. For example, the capsule of N. meningitidis serotype B mediates resistance to complement but also inhibits its invasion of eukaryotic cells.

Meanwhile, phase variation mediated by these hypermutable loci occurs in infected host animals. For instance, when Haemophilus influenzae infects the middle and inner ear and nasal cavity of chinchillas, variants found in these different sites differ in expression of an adhesin as a result of changes in the number of heptamer repeats in the gene's promoter. Phase variants in alternate states are also found in different anatomic compartments in human hosts. For example, only N. meningitidis capsule B ON variants are isolated from invasive sites, but both ON and OFF variants are found within the nasal tract. These results suggest that localized hypermutation contributes critically to adaptation. However, no one has proved that the high mutation rates of these loci are essential for virulence.

Analyzing genome sequences can help in judging the extent to which localized hypermutation serves as a bacterial response to environmental change. For example, 27 tracts of simple sequences (mainly mononucleotide repeats) can be readily identified in Campylobacter jejuni and 12 tetranucleotide repeat tracts occur in H. influenzae.

Hypermutable loci ordinarily encode outer membrane proteins or proteins that modify outer membrane determinants. In some bacterial species, the majority of outer membrane determinants are subject to localized hypermutation—demonstrating the potency of this adaptive strategy. For instance, 1/1,700th of the N. meningitidis genome (1,200 bp) may cause phase variable expression of 65 proteins, a much more economic use of coding capacity than occurs in two-component gene regulation systems.

However, some localized hypermutation mechanisms are limited to changing levels of protein expression—a potential disadvantage. Bacterial genomes circumvent this problem by having multiple genes of related function. H. influenzae genomes, for example, encode three- or four-phase variable surface-exposed iron acquisition proteins, permitting switches in expression that enable cells to escape up to three rounds of antibody responses without losing the ability to acquire iron. This genomic information system is consistent with the notion that localized hypermutation can enable cells to adapt to environmental changes.

Global and Localized Hypermutation Contribute to Adaptation, Evolution

Global hypermutation can provide a bacterial cell a short route to reaching a large repertoire of potential mutations within its genome. Because genetic variation correlates with the potential for adaptation, these mechanisms appear to be powerful driving forces behind bacterial adaptative responses. However, this adaptive advantage costs cells because global hypermutation also generates deleterious mutations that reduce fitness. Transient global hypermutation can reduce this cost, by generating mutations only when an environmental fluctuation is encountered, but only at the price of a reduced ability to respond to and survive bactericidal environments.

In contrast, localized hypermutation occurs within specific loci at high frequencies in small populations of cells without increasing deleterious mutations (Fig. 3). The limited number of adaptive mutations produced by localized hypermutation is countered by its concentration in loci encoding determinants under strong selective pressure. Although we doubt whether global hypermutation provides cells with a viable long-term strategy for adapting to environmental fluctuations, localized hypermutation appears to be compatible with long-term survival.

Even so, there are gray areas. Global hypermutation mechanisms tend to increase mutation rates of some DNA sequences more than others. Hence, genomes and global hypermutation-causing mutations might coevolve, leading potentially to overlaps between global and localized hypermutation mechanisms. For example, some strains of N. meningitidis have high hemoglobin receptor phase variation rates compared to strains with equivalent repeat tracts, but low global mutation rates, according to I. Stojiljkovic of Emory School of Medicine in Atlanta, Ga. Similarly, hypermutable sites are unmasked in E. coli mismatch repair mutants, according to Jeffrey Miller of the University of California, Los Angeles.

Such results highlight potential synergies between global and localized hypermutation. Assessing whether such synergies contribute to the biology of hypermutation will require careful interrogation of genomes, detailed experimental investigations of mechanisms of global hypermutation, and an appraisal of how particular mutations help each bacterial species adapt to its environment.

SUGGESTED READING

Bayliss, C. D., D. Field, and E. R. Moxon. 2001. The simple sequence contingency loci of Haemophilus influenzae and Neisseria meningitidis. J. Clin. Invest. 107:657-662.

De Bolle, X., C. D. Bayliss, D. Field, T. van de Ven, N. J. Saunders, D. W. Hood, and E. R. Moxon. 2000. The length of a tetranucleotide repeat tract in Haemophilus influenzae determines the phase variation rate of a gene with homology to type III DNA methyltransferases. Mol. Microbiol. 35:211-222.

Drake, J. W. 1991. A constant rate of spontaneous mutation in DNA-based microbes. Proc. Natl. Acad. Sci. USA 88:7160-7164.

Funchain, P., A. Yeung, J. L. Stewart, R. Lin, M. M. Slupska, and J. H. Miller. 2000. The consequences of growth of a mutator strain of Escherichia coli as measured by loss of function among multiple gene targets and loss of fitness. Genetics 154:959-970.

Giraud, A., I. Matic, O. Tenaillon, A. Clara, M. Radman, M. Fons, and F. Taddei. 2001. Costs and benefits of high mutation rates: adaptive evolution of bacteria in the mouse gut. Science 291:2606-2608.

McKenzie, G. J., and S. M. Rosenberg. 2001. Adaptive mutations, mutator DNA polymerases and genetic change strategies of pathogens. Curr. Opin. Microbiol. 4:586-594.

Moxon, E. R., P. B. Rainey, M. A. Nowak, and R. E. Lenski. 1994. Adaptive evolution of highly mutable loci in pathogenic bacteria. Curr. Biol. 4:24-33.

Radman, M. 1999. Enzymes of evolutionary change. Nature 401:866-869.

Richardson, A. R., and I. Stojiljkovic. 2001. Mismatch repair and the regulation of phase variation in Neisseria meningitidis. Mol. Microbiol. 40:645-655.

Tenaillon, O., H. Le Nagard, B. Godelle, and F. Taddei. 2000. Mutators and sex in bacteria: conflict between adaptive strategies. Proc. Natl. Acad. Sci. USA 97:10465-10470.