Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

Mutation—The Engine of Evolution: Studying Mutation and Its Role in the Evolution of Bacteria

Ruth Hershberg

Rachel & Menachem Mendelovitch Evolutionary Processes of Mutation & Natural Selection Research Laboratory, Department of Genetics and Developmental Biology, The Ruth and Bruce Rappaport Faculty of Medicine, Technion-Israel Institute of Technology, Haifa 31096, Israel Correspondence: [email protected]

Mutation is the engine of evolution in that it generates the genetic variation on which the evolutionary process depends. To understand the evolutionary process we must therefore characterize the rates and patterns of mutation. Starting with the seminal Luria and Delbruck fluctuation experiments in 1943, studies utilizing a variety of approaches have revealed much about mutation rates and patterns and about how these may vary between different bacterial strains and species along the chromosome and between different growth condi- tions. This work provides a critical overview of the results and conclusions drawn from these studies, of the debate surrounding some of these conclusions, and of the challenges faced when studying mutation and its role in bacterial evolution.

enetic variation is a prerequisite to evolu- eral of our decades-old assumptions were shown Gtionary change. In the absence of such var- to be mistaken, in light of newly available data. iation, no subsequent change can be achieved. Genetic variation is ultimately all generated by MUTATIONS VERSUS SUBSTITUTIONS mutation. It is therefore clear that mutation is a major evolutionary force that must be studied It is important to note that, in this article, and understood to understand evolution. Yet, we will only be considering de novo point mu- often mutation is set aside and thought of as a tations. We will not discuss large insertions or random generator of variation that follows very deletions or horizontal gene transfer events. To simple and predictable rules. proceed, we must define some terms. Many reviews of mutation deal with the For the purpose of this article, we will define molecular mechanisms of mutation and repair “DNA mutations” as single nucleotide changes (e.g., Modrich 1991; Smith 1992; Lieber 2010). in the DNA sequence of an individual organ- This work, in contrast, relates to mutation as ism. These will be the end result of the molec- an evolutionary force, focusing on bacteria. ular DNA change, and of the fact that this DNA We will show that mutation is extremely diffi- change was not repaired by the cellular repair cult to study, that we do not know nearly systems. Once a mutation occurs and is present enough about mutation and that recently sev- within an individual, it will either increase in

Editor: Howard Ochman Additional Perspectives on Microbial Evolution available at www.cshperspectives.org Copyright # 2015 Cold Spring Harbor Laboratory Press; all rights reserved Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077

1 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

R. Hershberg

frequency within the population, or will vanish once it occurs (Fig. 1). Note that our definition from the population. The ultimate fate of mu- of substitutions differs somewhat from that of tations depends on a combination of natural others that sometimes define substitutions as selection and stochastic forces, such as genetic either mutations that have fixed (e.g., Gillespie drift. 1998) or a specific class of base-change muta- Wewill define “DNA substitutions” as those tion (e.g., Graur and Li 2000). mutations that we can directly observe when Wewill define a phenotypic, or marker mu- we consider DNA sequence data. The substitu- tation, as a phenotypic change occurring in an tions we observe may reflect the mutations that individual. For example, an antibiotic resistance have occurred for better or worse, depending on phenotypic mutation causes an individual bac- how natural selection has affected them. For terium to become resistant to an antibiotic. example, if when comparing sequences we ob- Similarly, we can define a phenotypic, or marker serve that a certain substitution type (e.g., C to substitution, as a phenotypic change we are able T transitions) occurs more frequently within to observe, for example, an increase in the fre- our data, this could either mean that this mu- quency of resistant mutants within a bacterial tation type occurs more frequently, or that nat- population. Such an increase can occur because ural selection tends to favor this mutation type the resistance mutation occurs more frequently

A Normal levels of selection

Mutations Substitutions

Selective sieve

B Relaxed selection

Mutations Substitutions

Selective sieve

Figure 1. Different types of mutations (represented by differently colored arrows) occur at different frequencies (represented by arrow thickness). Selection acts as a sieve and allows only a subset of these mutations to persist and become the differences we see between genomes. Such differences are referred to as substitutions. Various types of mutations have different fitness effect distributions, and will be differently affected by selection. (A) Under normal levels of selection, selection will introduce its own biases into patterns of variation. Thus, biases in the patterns of observable substitutions between genomes are likely not to reflect mutational biases. (B) When selection is extremely relaxed, it is expected to affect patterns of variation to a much lesser extent, because it will affect only mutations with very high-fitness effects. Under such conditions, observed substitutions between genomes approximate a random sample of the mutations that have occurred. Because of this, when selection is relaxed, biases in the patterns of substitutions observed between genomes will better approximate mutational biases.

2 Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

Bacterial Mutation as an Evolutionary Force

or because of natural selection favoring the re- Luria and Delbruck modeled the variance sistant mutant. expected in the number of resistant mutants Often, mutation is studied by assuming that under both these scenarios (Luria and Delbruck certain types of DNA mutations (e.g., synony- 1943). Their models showed that a much higher mous mutations) or certain marker mutations variance would be expected if the emergence of (e.g., antibiotic resistance mutations when a resistance were caused by mutations occurring bacterium is not exposed to antibiotics) evolve before exposure to viruses. If mutation is a Pois- entirely neutrally. If there is absolutely no selec- son process and if mutations occur after and in tion acting on an observed class of substitu- response to viral exposure, one would expect the tions, their patterns and rates will indeed be a number of resistant mutants following exposure derivative of the patterns and rates of mutation. to be distributed around a certain mean, with However, as we will see later in this article, it the variance equal to the mean (a known char- is rare to find cases in which DNA or marker acteristic of the Poisson distribution). If, how- mutations are totally unaffected by selection. ever, mutations occur before exposure, they can Determining mutational patterns and rates is occur in any generation of growth. Mutations therefore a tricky business that requires one to occurring in earlier generations will rise to find creative ways to eliminate or minimize the higher frequencies by the end of an experiment, effects of natural selection on observed substi- compared with mutations occurring in later tutions. generations. Therefore, the number of resistant mutants at the end of an experiment will de- pend not only on the number of mutations that LURIA AND DELBRUCK—ESTIMATING have occurred, but also on when these muta- MUTATION RATES CAN BE A NOISY tions occurred. This should greatly enhance BUSINESS the variance in the numbers of resistant mu- In their seminal 1943 “fluctuation experi- tants observed between different experiments. ments,” Luria and Delbruck showed that even Indeed, Luria and Delbruck then went on to if mutational markers truly did evolve neutrally, show that in different experiments they saw a estimates of mutation rates based on such variance that was much higher than the mean markers would be extremely noisy (Luria and number of resistant mutants. This provided Delbruck 1943). Luria and Delbruck were at- the first ever demonstration that mutations oc- tempting to understand the following phenom- curred before selection for their outcome (Luria enon. When a pure bacterial culture is exposed and Delbruck 1943). to a bacteriophage, the culture will disappear In addition to showing for the first time that because of destruction of cells sensitive to the mutation precedes selection, the Luria and Del- virus. After further incubation, the culture will bruck study also shed light on the great variance often become turbid again because of growth of in substitution rates one can expect to observe a variant that is resistant to the phage. Once the when considering phenotypic markers (Luria variant is isolated, it often remains resistant and Delbruck 1943). First, as mentioned above, even if it is cultured for many generations in they showed that the variance in marker sub- the absence of any phage. At the time Luria stitution frequency was expected to be much and Delbruck were considering this problem, higher than the mean marker substitution fre- very little was known about the molecular quency. Second, Luria and Delbruck found that mechanisms of mutation. Yet, they already un- the mean substitution frequency they estimated derstood that such a phenomenon could either by simply averaging substitution frequencies occur because of resistance mutations occurring across different experiments was much higher before the viral challenge, or because a certain than the substitution frequency estimated by proportion of sensitive cells somehow acquire assuming a Poisson distribution and consi- resistance once they are exposed to phage (Luria dering the number of experiments in which and Delbruck 1943). no resistance substitutions were observed. This

Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 3 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

R. Hershberg

exemplifies the strong effect mutations occur- is reduced across the entire genome (Fig. 1). ring early on in the experiment can have on Such genome-wide relaxations of selection can calculated average mutation frequencies. One be the result of either close relatedness (Akashi or a few experiments in which a relatively high 1995; Messer 2009) and/or small effective pop- number of mutations occurred early on, may ulation sizes (Ne) (Lynch 2007). Bacterial line- greatly skew the estimated average frequency of ages exist for which genetic variation between mutations upward. Thus, estimates of mutation members of the lineage has naturally been only frequencies and rates obtained by using mark- weakly affected by selection, probably caused by er substitutions can often be very noisy (Luria a combination of close relatedness and small Ne and Delbruck 1943). Fortunately, we can now, (Hershberg et al. 2008; Holt et al. 2008; Hersh- in many cases, move away from using markers berg and Petrov 2010; Lieberman et al. 2011). and rather use whole-genome sequencing to Large quantities of genomic data from many study mutation. members of several such lineages are publicly available. Patterns of sequence variation be- tween members of bacterial lineages evolving METHODS FOR ELIMINATING THE EFFECTS under relaxed selection can be used to charac- OF NATURAL SELECTION WHEN STUDYING terize mutational patterns (Fig. 1). MUTATION The efficiency of selection can also be arti- To be able to study different parameters of the ficially reduced in the laboratory through re- mutational process, we must be able to disen- peated single-cell bottlenecking of growing bac- tangle mutation from the effects of natural se- terial populations, which severely reduces Ne. lection. The easiest way of accomplishing this is Such experiments are called mutation accumu- by focusing on scenarios in which selection is lation (MA) experiments (Elena and Lenski expected to have less of an effect on patterns of 2003; Lind and Andersson 2008; Brockhurst et substitution (Fig. 1). A number of studies have al. 2010). It is now possible to follow up MA used pseudogenes to study mutational biases experiments with whole-genome sequencing of (e.g., see Andersson and Andersson 1999; Nach- the ancestor strain and its resulting progeny, man and Crowell 2000). Such studies assume thus allowing for the genome-wide identi- that sequence variation within pseudogenes is fication of the MA mutations. The number of unaffected by selection, because pseudogenes generations a bacterial population underwent are no longer under selection to maintain func- during an MA experiment can be easily estimat- tion. Therefore, it is assumed that patterns of ed. MA experiments therefore make it possible sequence variation within pseudogenes will be to estimate not only the relative rates with which determined solely by mutation. Although use- different classes of mutations occur, but also the ful, this approach has limitations. For one, overall, absolute mutation rates. This is a clear although pseudogenes should not be under se- advantage of MA experiments over approaches lection stemming from protein function, they that rely on sequencing data from naturally may be under selection owing to genome-wide evolving bacteria, which cannot be used to esti- factors. For example, if there is selection to mate absolute mutation rates. At the same time, maintain a certain genomic nucleotide content MA experiments are much more labor inten- (Hershberg and Petrov 2010; Hildebrand et al. sive. It is also important to note that the mu- 2010), it might affect pseudogenes as strongly as tation rates and patterns estimated through it does other sequences. Second, for most mi- MA experiments may be influenced by the crobial genomes, we can only identify a very conditions under which these experiments small number of pseudogenes, because bacterial are performed. This is a particular concern pseudogenes tend to be lost very quickly (Kuo if mutation rates and patterns change under and Ochman 2010). different growth conditions. For example, the A second approach is to focus on evolution- stress-induced mutagenesis theory suggests that ary scenarios in which the efficiency of selection mutation rates could be much higher during

4 Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

Bacterial Mutation as an Evolutionary Force

stationary phase (reviewed in Galhardo et al. longer strong enough to counteract the action 2007, and discussed in depth later in this re- of genetic drift (Lynch 2010). Supporting this view). model, Lynch was able to show that per-base mutation rates inversely correlated with effec- tive population sizes (N ) in both prokaryotes ABSOLUTE/OVERALL RATES OF MUTATION e and eukaryotes (Lynch 2010; Sung et al. 2012). One of the key parameters of the mutational Because Ne is inversely related to the power of process is the absolute rate with which muta- drift, it can therefore be said that mutation rates tions happen, on average, across all types of become higher as the power of drift relative mutations and along the entire genome. In to selection becomes stronger, congruent with 1991, based on data collected by using a com- Lynch’s model. bination of fluctuation and MA experiments, Lynch later refined his “drift-barrier” model and quantifying mutation rates based on the by showing that the regression of the mutation frequency of marker substitutions, John Drake rates versus Ne is elevated for prokaryotes com- coined “Drake’s rule” (Drake 1991). According pared with eukaryotes (Sung et al. 2012). This to this rule, per nucleotide point mutation rates finding suggested that, for a given Ne, selec- inversely correlate with genome size in mi- tion is less effective at reducing mutation rates crobes. As a result, genome-wide mutation rates in prokaryotes. To explain this phenomenon, are an approximate constant of 0.003-point Lynch suggested that the magnitude of selection mutations per genome per generation (Drake to reduce mutation rates is not just a function of 1991). These results were based on mutation the per-base mutation rate, but rather also of the rates of only seven microbes, but later results genome-wide deleterious mutation potential of from many additional microbes provided fur- the genome (Sung et al. 2012). Prokaryotes that ther support for Drake’s rule, particularly in tend to have less coding sequences in total, pro- prokaryotes and in double-stranded DNA vide a smaller target for the origin of deleterious (dsDNA) viruses (Lynch 2010). Drake argued mutations than eukaryotic genomes. Under this that such a fine-tuned mutation rate must be refined model, the strength of selection to re- an evolved trait (Drake 1991). duce per nucleotide mutation rates will scale It is generally accepted that natural selection positively with what Lynch defined as the effec- favors the lowering of mutation rates, as muta- tive genome size, which he approximated as the tions are mostly deleterious (Kimura 1967; sum of coding DNA within a genome. Fitting Drake 1991; Dawson 1998; Lynch 2010). Drake with this, Lynch observed that the effective and others postulated that reducing mutation genome-wide mutation rate, calculated as the rates comes at a certain physiological cost (Ki- per-site mutation rate multiplied by the effec- mura 1967; Drake 1991; Dawson 1998). Drake tive genome size, inversely correlated with Ne,in suggested that mutation rates reached equilib- a way that did not depend on whether an organ- rium when the benefit of further lowering mu- ism is a prokaryote or a eukaryote (Sung et al. tation rates matched the physiological cost of so 2012). doing. In other words, according to Drake, nat- Under both Drake’s and Lynch’s models, the ural selection drives both the reduction in mu- cost of deleterious mutations is what drives mu- tation rates, as well as the ultimate tapering off tation rates down (Drake 1991; Lynch 2010; of this reduction. In contrast, Michael Lynch Sung et al. 2012). Therefore, under both mod- suggested an alternative model under which els, an increase in the average cost of mutations the lower limit on mutation rates is not set by would lead to a decrease in mutation rates. To natural selection on physiological cost, but rath- examine this, Drake examined mutation rates of er by genetic drift (Lynch 2010). As per-base thermophiles and compared them to those of mutation rates become lower, selection to fur- mesophiles (Drake 2009). The rationale was that ther reduce mutation rates becomes weaker, un- many mutations that are tolerated at the stan- til a point is reached in which selection is no dard growth temperature are highly harmful

Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 5 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

R. Hershberg

when temperatures are higher. Thus, more mu- Adenine-Thymine (AT) Bias of Mutation and tations will have a fitness cost in thermophiles Bacterial Nucleotide Content Variation than in mesophiles, which should lead to lower mutation rates within thermophiles (Drake Bacterial nucleotide content is extremely vari- 2009). By again using data derived by use of able. Some bacteria have guanine-cytosine (GC) marker substitutions, Drake then showed that content ,25%, whereas the GC content of mutation rates in two different thermophilic other bacteria can reach 75%. This variation microbes were indeed much lower than in mes- was for avery long time considered to be entirely ophilic microbes and phages (Drake 2009). This neutral, and the result of extreme variation in seems to support the model under which selec- mutational biases between different bacteria tion favors lowering of mutation rates, because (Sueoka 1962; Muto and Osawa 1987). It was of the deleterious effects of mutations. thought that GC-rich bacteria were simply ones Recently, many studies have been conducted in which AT to GC mutations occurred more in which MA lines from various microbes were frequently than GC to ATmutations. The oppo- fully sequenced to determine mutation rates site pattern of mutation was thought to occur in (e.g., Lind and Andersson 2008; Lee et al. AT-rich bacteria (Sueoka 1962; Muto and Osawa 2012; Sung et al. 2012). As discussed above, 1987). However, it was more recently shown, measures of mutation rates from whole-genome using data from bacteria evolving under varying sequencing are expected to be more precise than degrees of relaxed selection, that mutation is those measured via the use of phenotypic mark- universally AT biased across both AT-rich and ers. These recent studies have shown that al- GC-rich bacteria (Balbi et al. 2009; Hershberg though the Drake rule seems to generally apply and Petrov 2010; Hildebrand et al. 2010). Given in prokaryotes and dsDNA phages, the range of that mutation is always AT biased, some other per genome mutation rates appears to be higher force must be driving elevated GC content in than originally postulated by Drake. For exam- bacteria with intermediate to high GC content. ple, Lee et al. (2012) estimated mutation rates The most obvious culprit is natural selection, for a wild-type Escherichia coli laboratory strain, favoring such higher GC content, but other non- based on whole-genome sequencing of 59 MA selective mechanisms could also be involved. lines. Based on these data, they estimated a mu- One nonselective mechanism that may be tation rate of 0.001 mutations per genome per driving GC content up in bacteria with inter- generation (lower than the 0.003 constant sug- mediate to high GC content, is biased gene con- gested by Drake) (Lee et al. 2012). Sung et al. version (BGC) (reviewed in Duret and Galtier (2012) sequenced MA lines of one of the small- 2009). It has been shown that gene conversion is est culturable bacteria, Mesoplasma florum, and GC biased in many eukaryotes, including hu- found a genome-wide mutation rate of 0.008. mans and other mammals. In other words, the probability of a GC allele to be passed on to the next generation through gene conversion is MUTATIONAL BIASES higher in these eukaryotes than that of an AT Various types of mutations may occur at dif- allele. As a result of such BGC, in these eukary- ferent rates. Such consistent variation in the otes, regions with lower recombination rates rates of different categories of mutations means tend to be more ATrich, whereas regions under- that the mutational process in itself, even in the going more recombination will tend to be more absence of any natural selection, may introduce GC rich (Fullerton et al. 2001). A relationship biases into patterns of genetic variation. Char- between levels of recombination and GC con- acterizing these biases is important for under- tent was also demonstrated for many bacteria, standing which biases in patterns of genetic suggesting that BGC, or a mechanism similar to variation are selected and thus functionally im- BGC, may affect nucleotide content in bacteria portant, and which may just be introduced by in a similar manner (Touchon et al. 2009; Las- the mutational process. salle et al. 2015).

6 Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

Bacterial Mutation as an Evolutionary Force

A second nonselective mechanism that may study supported the results of the smaller scale be increasing GC content in bacteria relates metagenomic analysis and demonstrated that to mismatch-repair (MMR) systems. Lee et al. environment affects microbial nucleotide con- (2012) conducted MA experiments on both tent in a manner that cannot be entirely ex- wild-type E. coli and mutants deficient in MMR. plained by differences in phylogenetic compo- They found that although mutation was indeed sition. Intriguingly, the data used in the more AT biased in wild-type E. coli, it was GC biased recent study made it possible to show that en- in the absence of MMR. This suggests that the vironmental factors drive changes in nucleotide nucleotide content of genomes might be influ- content, not only between highly diverged en- enced by how well their MMR systems function vironment types (e.g., soil vs. aquatic), but also (Lee et al. 2012). Nucleotide content is a slowly between samples extracted from the guts of dif- evolving trait, because many substitutions need ferent human subjects (Reichenberger et al. to occurfor genome-wide patterns of nucleotide 2015). These results imply that the environmen- content to substantially change. Therefore, the tal factors that select for certain nucleotide com- nucleotide content of a bacterium may not be positions may be quite subtle. influenced solely by its current MMR function- The most obvious reason selection would ality. Rather, MMR function during the evo- favor high GC content in some bacteria is that lution of the lineage to which the bacterium higher GC content may provide better genome belongs may influence its current GC content. stability when temperatures are elevated. Many Fitting with this, it has been shown that the re- studies have attempted to investigate the corre- lationship between the nucleotide content of a lation between GC content and optimal growth bacterium and the current presence of MMR temperatures, with mixed results (Galtier and genes within its genome is not a straightforward Lobry 1997; Lobry 1997; Hurst and Merchant one (Garcia-Gonzalez et al. 2012). 2001; Marashi and Ghalanbor 2004; Musto et al. When it comes to selection affecting nucle- 2004, 2006; Wang et al. 2006). In the end, it is otide content, the first big question that arises very possible that growth temperature does af- concerns the nature of selection. If indeed nat- fect nucleotide content. However, high growth ural selection favors higher GC content in some temperatures are likely not the only environ- bacteria, why? What is the advantage conferred mental factors affecting nucleotide content, on these bacteria by having higher genome- and they likely do not explain why so many wide GC content? The currently available an- bacteria have high or intermediate GC content swers to this question are far from complete. in the face of universally AT-biased mutation. A study that examined metagenomic sam- Recently, Raghavan et al. (2012) have sug- ples collected from aquatic and soil environ- gested an alternative force selecting for elevated ments, was successful in demonstrating that GC content related to gene expression. Ragha- soil bacteria are substantially more GC rich van et al. inserted a plasmid containing the than aquatic bacteria, even when differences in green florescent protein (GFP) gene into strains phylogeny are accounted for (Foerstner et al. of E. coli. They generated their GFP genes to 2005). These results suggest that environmental differ in the GC content of their synonymous selection plays a role in determining nucleotide sites. This allowed them to show that strains content. The study in questionwas performed in harboring a more GC-rich GFP gene grew faster 2005 when metagenomic datawere only starting than strains harboring a more AT-rich version to become available, and used samples from of the gene, in a manner that depended on the only four different environments (Foerstner construct being expressed, at both the mRNA et al. 2005). A more recent study used a much and protein levels (Raghavan et al. 2012). They larger collection of 183 metagenomic data sets, then showed that this effect was not limited extracted from 14 environment types, to inves- to the GFP gene but also occurred when other tigate the effects of environment on nucleotide genes were so inserted into E. coli (Raghavan composition (Reichenberger et al. 2015). This et al. 2012). This finding fits the observation

Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 7 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

R. Hershberg

that bacteria with intermediate to high GC con- leads to the loss of BGC. This is currently just tent tend use GC-rich optimal codons (Hersh- an idea, and much further theoretical and ex- berg and Petrov 2009)—a trend that results in a perimental work needs to be performed to ex- much higher GC content of protein-coding amine its validity. synonymous sites, compared with noncoding intergenic sequences, within GC-intermediate Variation in Mutation Rates along and GC-rich genomes (Hershberg and Petrov the Chromosome 2009, 2012; Raghavan et al. 2012). If indeed GC-rich coding sequences are expressed more Mutation may also bias patterns of genetic var- efficiently and/or accurately, selection may in- iation if certain regions of the genome are more deed drive GC content up in coding sequences. prone to mutation than other regions. In a re- However, this suggested mechanism does not cent study, Foster et al. (2013) sequenced 24 MA explain why intergenic, noncoding regions lines of MMR defective E. coli. They found a also have higher GC content than expected at striking pattern by which mutations are not mutational equilibrium in genomes with inter- randomly distributed along the chromosome. mediate to high GC content (Hershberg and Rather, mutations fall in a wave-like pattern Petrov 2010). Thus, although selection for genes that is repeated in an almost exact mirror image to be more GC rich may contribute to elevated in the two separately replicated halves (repli- GC content, it cannot explain them in their cores) of the E. coli chromosome (Foster et al. entirety. 2013). They further showed that mutation A second question that arises when consid- density was higher in regions of the E. coli chro- ering natural selection acting on nucleotide mosome where gene expression is regulated by composition is the question of how such selec- nucleoide-associated proteins. These results tion would work. A problem arises because each were interpreted by Foster et al. (2013) to imply individual base mutation only minutely alters that mutation rates are affected by chromosome overall nucleotide content, and an enormous structure. number of mutations are needed to have any In a recent study, Martincorena et al. (2012) significant effect on overall nucleotide content. claimed to show that mutation rates are signifi- If so, how can selection on GC content affect cantly lower in highly expressed genes and genes each individual mutation? Additionally, if selec- undergoing stronger selection. They postulated tion were to affect each mutation, the associated that by lowering mutation rates, particularly in genetic load would be staggering. A possible genes that are more highly expressed and more solution to this conundrum is that natural se- important, E. coli was using an evolutionary lection may not act on individual mutations. risk-management strategy. These results were Rather if there is selection in favor of elevated obtained by analyzing patterns of synonymous GC content and there is a nonselective mecha- substitution between 34 E. coli strains, and re- nism, such as BGC, that elevates GC content lied on an assumption that these patterns of (Duret and Galtier 2009), it is possible that substitution evolved under relaxed selection, strong selection will exist on that mechanism. because of close relatedness of these strains For example, if indeed BGC affects nucleotide (Martincorena et al. 2012). It is important to content in some bacteria, as has been shown for note, however, that different E. coli strains are eukaryotes (Duret and Galtier 2009), bacteria highly diverged and that patterns of substitu- that lose the ability to carry out BGC may grad- tion between strains of E. coli are, in fact, subject ually become more ATrich. Once their GC con- to extremely strong selection (Hershberg et al. tent becomes low enough to be disfavored by 2007). It is therefore quite possible that the dif- natural selection, these bacteria will be removed ferences in the frequency of E. coli synonymous from the population. In this example, it is not substitutions between highly expressed and less each GC to AT mutation that is affected by se- highly expressed genes are because of selection, lection, but rather the mutational event that rather than mutation. Indeed, it was very re-

8 Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

Bacterial Mutation as an Evolutionary Force

cently shown that the theory of adaptive risk that bacteria may be able to selectively increase management via lowering of mutation rates in mutation rates when they are exposed to certain highly expressed genes is theoretically untenable “stressful” or growth-limiting conditions (re- (Chen and Zhang 2013). Furthermore, the neg- viewed in Foster 2007; Galhardo et al. 2007). ative correlation suggested by Martincorena Modeling has shown that such stress-induced et al. (2012) between mutation rates and levels mutagenesis (SIM) should be highly beneficial of expression was not supported by MA studies (Ram and Hadany 2012), as it could allow in E. coli, Salmonella, and yeast (Lind and An- bacteria to transiently increase mutagenesis dersson 2008; Lee et al. 2012; Park et al. 2012; particularly when they are most pressured to Chen and Zhang 2013; Foster et al. 2013). To adapt. Yet, the study of SIM has been plagued the contrary, in some MA studies, a significant by fierce debate (e.g., Slechta et al. 2002, 2003; positive correlation is observed between levels Roth et al. 2003, 2006; Wrande et al. 2008; of expression and mutation frequencies (Lind Katz and Hershberg 2013). In this review, I and Andersson 2008; Park et al. 2012; Chen and do not have sufficient space to delve into the Zhang 2013). full debate, but will only introduce some points of contention. The strongest support of SIM, and the most CONSTITUTIVE MUTATORS AND detailed understanding of its mechanisms has STRESS-INDUCED MUTAGENESIS come from the use of a particular assay suggest- As mentioned above, natural selection is ed originally by Cairns and Foster (1991). In thought to favor the lowering of mutation rates, this assay, a special E. coli strain, deleted for its because many mutations are deleterious (Ki- chromosomal lac operon, and carrying a lacI- mura 1967; Drake 1991; Dawson 1998; Lynch lacZ fusion gene with a frameshift mutation in 2010). In sharp contrast to this expectation, it lacI on an F0 conjugative plasmid, is plated onto was observed that 1% of all natural bacterial lactose plates. On such plates, only cells that isolates are mutators that have high mutation become lac positive can form colonies, and so rates, compared with the reminder of the pop- the frequency of reversion mutants can be mon- ulation (Gross and Siegel 1981; LeClerc et al. itored. Original proponents of SIM assumed 1996). If indeed selection disfavors high muta- that growth could only be achieved on the plates tion rates, why would hypermutating bacteria by reversion mutants that corrected the frame- be present at such high frequencies? The best shift mutation in lacI. Colonies forming from explanation currently available is that mutators mutants that arose before plating were expected accelerate adaptation in asexual clonal popula- to emerge within 2 days of plating, and any tions (Sniegowski et al. 1997; Taddei et al. 1997; subsequent colonies were assumed to result Giraud et al. 2001; Notley-McRobb et al. 2002). from mutations occurring on the plates, in non- Mutator alleles may thus be linked to adaptive growing bacteria. Any such mutations were as- alleles that arise as a result of hypermutation. It sumed to be the result of SIM. is therefore thought that when bacteria are ex- Studies utilizing the Crains and Foster Lac posed to strong pressure to adapt quickly (e.g., assay suggested that the occurrence of stress-in- when they are faced with new challenges), mu- duced frameshift mutations depended on dou- tator alleles may become beneficial, which in- ble-strand breaks (DSBs), repair of these DSBs creases their frequencies (Sniegowski et al. 1997; by an error-prone polymerize, dinB, and also Taddei et al. 1997; Giraud et al. 2001; Notley- depended on the bacterial stress response, me- McRobb et al. 2002). diated by the stationary phase s factor, rpoS The mutators discussed above are constitu- (reviewed in Galhardo et al. 2007). Although tive mutators—bacteria that are defective in these results suggested a mechanism by which their repair mechanisms and that constitutively SIM could occur, use of the Lac assay was se- mutate at higher frequencies (LeClerc et al. verely debated. First, it was argued that in- 1996). However, it has also been postulated creased reversion could be caused by amplifica-

Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 9 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

R. Hershberg

tion of the inactive lac gene, slow growth of cells ments were tested for the frequency with which carrying this amplification, consequent frame- they accumulate resistance to rifampicin in shift reversion mutations, and selection for young and aging colonies. It was observed these mutants that could now grow freely on the that, to varying extents in different isolates, lactose plates (reviewed in Roth et al. 2006). the frequency of mutants resistant to rifampicin Thus, it was suggested that frameshift reversions increases in aging colonies compared with were not necessarily because of SIM. Second, it young colonies. This increase in the frequency was argued that the particular F0 conjugative of resistant mutants was a priori assumed to plasmid used was problematic, as it contained result from increased mutagenesis, resulting an extra copy of the dinB gene that was shown to from the starvation stress incurred via growth be important for increased frequency of rever- in aging colonies. Indeed the resulting paper sion (Roth et al. 2006). It was therefore argued was titled “stress-induced mutagenesis in bacte- that the results obtained using the Lac assay ria” (Bjedov et al. 2003). A subsequent study, were not general, but rather particular to the published in 2008, showed that increased fre- assay used. quency of resistance to rifampicin could also To address these concerns, Shee et al. (2011) be explained by natural selection, as it showed more recently developed an alternative chromo- that many rifampicin-resistant mutants carried somal assay for studying SIM. In this assay, the a growth advantage in aging colonies (Wrande frequency of frameshift reversions to an artifi- et al. 2008). Yetthe Bjedov et al. study continued cially introduced tetracycline resistance cassette to be very widely cited as conclusive evidence for containing a deactivating frameshift mutation the occurrence of SIM within natural bacterial is quantified. DSBs are induced artificially by populations (e.g., Bogumil and Dagan 2012; placing the tetracycline cassette 8.5 kb from an Buerger et al. 2012; Feher et al. 2012; Obolski I-sceI double-strand endonuclease cut-site. The and Hadany 2012; Rosenberg et al. 2012; Ryall cells are engineered to contain an SceI gene, et al. 2012; Sanchez-Alberola et al. 2012; Ma- controlled by a PBAD promoter, which is re- clean et al. 2013; Martincorena and Luscombe pressed when glucose is available, but dere- 2013). pressed once glucose becomes depleted and cells We have recently repeated the Bjedov et al. begin to starve. Using this assay, Shee et al. experiments on a single laboratory strain of (2011) could show that there was an increase E. coli. Consistent with their results, we were in the frequency of tetracycline-resistant rever- able to show a substantial increase in the fre- sion mutants in response to starvation. This quency of resistance to rifampicin in aging col- increase was shown to be dependent on DSBs, onies compared with young colonies. We also dinB, and rpoS (Shee et al. 2011). Shee et al. observed a sharp increase in the frequency of interpreted their results as demonstrating that resistance to a second antibiotic, nalidixic acid results obtained using the Lac assay are not spe- (Katz and Hershberg 2013). We then used cific to that assay, and that SIM indeed occurs in whole-genome sequencing to show conclusively E. coli, and depends on the stress response being that increased mutagenesis could not explain induced and on error-prone repair of DSBs. the increased frequency of resistance observed So far, I have discussed SIM as it has been to either of the two antibiotics (Katz and Hersh- studied in artificial laboratory models, but has berg 2013). Therefore, SIM cannot explain the SIM been shown to occur within natural bacte- Bjedov et al. results, and these results cannot be rial populations? Until very recently, the best, seen as evidence of SIM occurring in natural most well cited evidence for the natural occur- bacterial populations. rence of SIM came from experiments conducted We further showed that, as was previously by Bjedov et al. in 2003 (Bjedov et al. 2003). In shown for rifampicin resistance mutations these experiments, 800 natural isolates of (Wrande et al. 2008), nalidixic acid resistance E. coli, extracted from a large variety of host- mutations can also confer a growth advantage associated and non-host-associated environ- in aging colonies (Katz and Hershberg 2013).

10 Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

Bacterial Mutation as an Evolutionary Force

An additional study showed that a mutation Balbi KJ, Rocha EP,Feil EJ. 2009. The temporal dynamics of conferring resistance to streptomycin can also slightly deleterious mutations in Escherichia coli and Shi- gella spp. Mol Biol Evol 26: 345–355. improve growth when bacteria are grown on Bjedov I, Tenaillon O, Gerard B, Souza V,Denamur E, Rad- poor carbon sources (Paulander et al. 2009). man M, TaddeiF,Matic I. 2003. Stress-induced mutagen- Combined, these results show that using anti- esis in bacteria. Science 300: 1404–1409. biotic resistance as a marker for the study of Bogumil D, Dagan T. 2012. Cumulative impact of chaper- mutation in general and SIM in particular one-mediated folding on genome evolution. Biochemis- try 51: 9941–9953. may be highly problematic. Brockhurst MA, Colegrave N, Rozen DE. 2010. Next-gener- ation sequencing as a tool to study microbial evolution. Mol Ecol 20: 972–980. CONCLUDING REMARKS Buerger S, Spoering A, Gavrish E, Leslin C, Ling L, Epstein SS. 2012. Microbial scout hypothesis, stochastic exit from Much remains to be understood about the rates dormancy, and the nature of slow growers. Appl Environ and patterns of mutation and about how these Microbiol 78: 3221–3228. vary between different bacterial isolates, within Cairns J, Foster PL. 1991. Adaptive reversion of a frameshift populations, as a factor of growth conditions, mutation in Escherichia coli. Genetics 128: 695–701. and along the chromosome. Mutation is diffi- Chen X, Zhang J. 2013. No gene-specific optimization of mutation rate in Escherichia coli. Mol Biol Evol 30: 1559– cult to study because it is a highly noisy process 1562. and because it affects variation in a manner that Dawson KJ. 1998. Evolutionarily stable mutation rates. J is highly entangled with the effects of natural Theor Biol 194: 143–157. selection. To characterize the effects of muta- Drake JW.1991. A constant rate of spontaneous mutation in tion, we need to acknowledge these complica- DNA-based microbes. Proc Natl Acad Sci 88: 7160–7164. tions and find creative ways to address them. Drake JW. 2009. Avoiding dangerous missense: Thermo- philes display especially low mutation rates. PLoS Genet Future studies will undoubtedly take advantage 5: e1000520. of our increasing ability to examine variation at Duret L, Galtier N. 2009. Biased gene conversion and the the whole-genome level to reveal much more evolution of mammalian genomic landscapes. Annu Rev about mutation and how it acts as an engine Genomics Hum Genet 10: 285–311. of evolution in bacteria and beyond. Elena SF, Lenski RE. 2003. Evolution experiments with mi- croorganisms: The dynamics and genetic bases of adap- tation. Nat Rev Genet 4: 457–469. ACKNOWLEDGMENTS Feher T, Bogos B, Mehi O, Fekete G, Csorgo B, Kovacs K, Posfai G, Papp B, Hurst LD, Pal C. 2012. Competition I thank Sophia Katz, Wesley Field, and Talia between transposable elements and mutator genes in bacteria. Mol Biol Evol 29: 3153–3159. Karasov for their helpful comments. R.H. is Foerstner KU, von Mering C, Hooper SD, Bork P. 2005. supported by a European Research Council Environments shape the nucleotide composition of ge- (ERC) FP7 CIG Grant (No. 321780), by a BSF nomes. EMBO Rep 6: 1208–1213. Grant (No. 2013463), by a Yigal Allon Fellow- Foster PL. 2007. Stress-induced mutagenesis in bacteria. ship awarded by the Israeli Council for Higher Crit Rev Biochem Mol Biol 42: 373–397. Education, and by the Robert J. Shillman Career Foster PL, Hanson AJ, Lee H, Popodi EM, Tang H. 2013. On the mutational topology of the bacterial genome. G3 (Be- Advancement Chair. Workby R.H. is performed thesda) 3: 399–407. in the Rachel & Menachem Mendelovitch Evo- Fullerton SM, Bernardo Carvalho A, Clark AG. 2001. Local lutionary Process of Mutation & Natural Selec- rates of recombination are positively correlated with GC tion Research Laboratory. content in the human genome. Mol Biol Evol 18: 1139– 1142. Galhardo RS, Hastings PJ, Rosenberg SM. 2007. Mutation as REFERENCES a stress response and the regulation of evolvability. Crit Rev Biochem Mol Biol 42: 399–435. Akashi H. 1995. Inferring weak selection from patterns of Galtier N, Lobry JR. 1997. Relationships between genomic polymorphism and divergence at “silent” sites in Droso- GþC content, RNA secondary structures, and optimal phila DNA. Genetics 139: 1067–1076. growth temperature in prokaryotes. J Mol Evol 44: 632– Andersson JO, Andersson SG. 1999. Insights into the evo- 636. lutionary process of genome degradation. Curr Opin Ge- Garcia-Gonzalez A, Rivera-Rivera RJ, Massey SE. 2012. The net Dev 9: 664–671. presence of the DNA repair genes mutM, mutY, mutL, and

Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 11 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

R. Hershberg

mutS is related to proteome size in bacterial genomes. Lieber MR. 2010. The mechanism of double-strand DNA Front Genet 3: 3. break repair by the nonhomologous DNA end-joining Gillespie J. 1998. Population genetics, a concise guide, 1st ed. pathway. Annu Rev Biochem 79: 181–211. The Johns Hopkins University Press, Baltimore. Lieberman TD, Michel JB, Aingaran M, Potter-Bynoe G, Giraud A, Radman M, Matic I, Taddei F. 2001. The rise and Roux D, Davis MR Jr, Skurnik D, Leiby N, LiPuma JJ, fall of mutator bacteria. Curr Opin Microbiol 4: 582–585. Goldberg JB, et al. 2011. Parallel bacterial evolution with- in multiple patients identifies candidate pathogenicity Graur D, Li W. 2000. Fundamentals of molecular evolution. genes. Nat Genet 43: 1275–1280. Sinauer Associates, Sunderland, MA. Lind PA, Andersson DI. 2008. Whole-genome mutational Gross MD, Siegel EC. 1981. Incidence of mutator strains in biases in bacteria. Proc Natl Acad Sci 105: 17878–17883. Escherichia coli and coliforms in nature. Mutat Res 91: Lobry JR. 1997. Influence of genomic G C content on 107–110. þ average amino-acid composition of proteins from 59 bac- Hershberg R, Petrov DA. 2009. General rules for optimal terial species. Gene 205: 309–316. codon choice. PLoS Genet 5: e1000556. Luria SE, Delbruck M. 1943. Mutations of bacteria from Hershberg R, Petrov DA. 2010. Evidence that mutation is virus sensitivity to virus resistance. Genetics 28: 491–511. universally biased towards AT in bacteria. PLoS Genet 6: Lynch M. 2007. The origins of genome architecture. Sinauer e1001115. Associates, Sunderland, MA. Hershberg R, Petrov DA. 2012. On the limitations of using Lynch M. 2010. Evolution of the mutation rate. Trends Genet ribosomal genes as references for the study of codon us- 26: 345–352. age: A rebuttal. PLoS ONE 7: e49060. Maclean RC, Torres-Barcelo C, Moxon R. 2013. Evaluating Hershberg R, Tang H, Petrov DA. 2007. Reduced selection evolutionary models of stress-induced mutagenesis in leads to accelerated gene loss in Shigella. Genome Biol 8: bacteria. Nat Rev Genet 14: 221–227. R164. Marashi SA, Ghalanbor Z. 2004. Correlations between ge- Hershberg R, Lipatov M, Small PM, Sheffer H, Niemann S, nomic GC levels and optimal growth temperatures are Homolka S, Roach JC, Kremer K, Petrov DA, Feldman not “robust.” Biochem Biophys Res Commun 325: 381– MW, et al. 2008. High functional diversity in Mycobacte- 383. rium tuberculosis driven by genetic drift and human de- Martincorena I, Luscombe NM. 2013. Non-random muta- mography. PLoS Biol 6: e311. tion: The evolution of targeted hypermutation and hy- Hildebrand F, Meyer A, Eyre-Walker A. 2010. Evidence of pomutation. BioEssays 35: 123–130. selection upon genomic GC-content in bacteria. PLoS Martincorena I, Seshasayee AS, Luscombe NM. 2012. Evi- Genet 6: e1001107. dence of non-random mutation rates suggests an evolu- Holt KE, Parkhill J, Mazzoni CJ, Roumagnac P, Weill FX, tionary risk management strategy. Nature 485: 95–98. Goodhead I, Rance R, Baker S, Maskell DJ, Wain J, et al. Messer PW. 2009. Measuring the rates of spontaneous mu- 2008. High-throughput sequencing provides insights tation from deep and large-scale polymorphism data. into genome variation and evolution in Salmonella Ty- Genetics 182: 1219–1232. phi. Nat Genet 40: 987–993. Modrich P.1991. Mechanisms and biological effects of mis- Hurst LD, Merchant AR. 2001. High guanine-cytosine con- match repair. Annu Rev Genet 25: 229–253. tent is not an adaptation to high temperature: A compar- Musto H, Naya H, Zavala A, Romero H, Alvarez-Valin F, ative analysis amongst prokaryotes. Proc Biol Sci 268: Bernardi G. 2004. Correlations between genomic GC lev- 493–497. els and optimal growth temperatures in prokaryotes. Katz S, Hershberg R. 2013. Elevated mutagenesis does not FEBS Lett 573: 73–77. explain the increased frequency of antibiotic resistant Musto H, Naya H, Zavala A, Romero H, Alvarez-Valin F, mutants in starved aging colonies. PLoS Genet 9: Bernardi G. 2006. Genomic GC level, optimal growth e1003968. temperature, and genome size in prokaryotes. Biochem Kimura M. 1967. On the evolutionary adjustment of spon- Biophys Res Commun 347: 1–3. taneous mutation rates. Genet Res 9: 23–34. Muto A, Osawa S. 1987. The guanine and cytosine content Kuo CH, Ochman H. 2010. The extinction dynamics of of genomic DNA and bacterial evolution. Proc Natl Acad bacterial pseudogenes. PLoS Genet 6: e1001050. Sci 84: 166–169. Lassalle F,Pe´rian S, Bataillon T,Nesme X, Duret L, Daubin V. Nachman MW, Crowell SL. 2000. Estimate of the mutation 2015. GC-content evolution in bacterial genomes: The rate per nucleotide in humans. Genetics 156: 297–304. biased gene conversion hypothesis expands. PLoS Genet Notley-McRobb L, Seeto S, Ferenci T.2002. Enrichment and 11: e1004941. elimination of mutY mutators in Escherichia coli popula- LeClerc JE, Li B, Payne WL, Cebula TA.1996. High mutation tions. Genetics 162: 1055–1062. frequencies among Escherichia coli and Salmonella path- Obolski U, Hadany L. 2012. Implications of stress-induced ogens. Science 274: 1208–1211. genetic variation for minimizing multidrug resistance in Lee H, Popodi E, Tang H, Foster PL. 2012. Rate and molec- bacteria. BMC Med 10: 89. ular spectrum of spontaneous mutations in the bacte- Park C, Qian W, Zhang J. 2012. Genomic evidence for ele- rium Escherichia coli as determined by whole-genome vated mutation rates in highly expressed genes. EMBO sequencing. Proc Natl Acad Sci 109: E2774–E2783. Rep 13: 1123–1129.

12 Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

Bacterial Mutation as an Evolutionary Force

Paulander W, Maisnier-Patin S, Andersson DI. 2009. The Slechta ES, Liu J, Andersson DI, Roth JR. 2002. Evidence fitness cost of streptomycin resistance depends on rpsL that selected amplification of a bacterial lac frameshift mutation, carbon source and RpoS (sS). Genetics 183: allele stimulates Lacþ reversion (adaptive mutation) 539–546. with or without general hypermutability. Genetics 161: Raghavan R, Kelkar YD, Ochman H. 2012. A selective force 945–956. favoring increased GþC content in bacterial genes. Proc Slechta ES, Bunny KL, Kugelberg E, Kofoid E, Andersson DI, Natl Acad Sci 109: 14504–14507. Roth JR. 2003. Adaptive mutation: General mutagenesis Ram Y, Hadany L. 2012. The evolution of stress-induced is not a programmed response to stress but results from rare coamplification of dinB with lac. Proc Natl Acad Sci hypermutation in asexual populations. Evolution 66: 100: 12847–12852. 2315–2328. Smith KC. 1992. Spontaneous mutagenesis: Experimental, Reichenberger ER, Rosen G, Hershberg U, Hershberg R. genetic and other factors. Mutat Res 277: 139–162. 2015. Prokaryotic nucleotide composition is shaped by both phylogeny and the environment Genome. Biol Evol Sniegowski PD, Gerrish PJ, Lenski RE. 1997. Evolution of 7: 1380–1389. high mutation rates in experimental populations of E. coli. Nature 387: 703–705. Rosenberg SM, Shee C, Frisch RL, Hastings PJ. 2012. Stress- induced mutation via DNA breaks in Escherichia coli:A Sueoka N. 1962. On the genetic basis of variation and het- erogeneity of DNA base composition. Proc Natl Acad Sci molecular mechanism with implications for evolution 48: 582–592. and medicine. BioEssays 34: 885–892. Sung W,Ackerman MS, Miller SF,Doak TG, Lynch M. 2012. Roth JR, Kofoid E, Roth FP,Berg OG, Seger J, Andersson DI. Drift-barrier hypothesis and mutation-rate evolution. 2003. Regulating general mutation rates: Examination of Proc Natl Acad Sci 109: 18488–18492. the hypermutable state model for Cairnsian adaptive mu- tation. Genetics 163: 1483–1496. Taddei F, Radman M, Maynard-Smith J, Toupance B, Gouyon PH, Godelle B. 1997. Role of mutator alleles in Roth JR, Kugelberg E, Reams AB, Kofoid E, Andersson adaptive evolution. Nature 387: 700–702. DI. 2006. Origin of mutations under selection: The adap- Touchon M, Hoede C, Tenaillon O, Barbe V, Baeriswyl S, tive mutation controversy. Annu Rev Microbiol 60: 477– Bidet P,Bingen E, Bonacorsi S, Bouchier C, Bouvet O, et 501. al. 2009. Organised genome dynamics in the Escherichia Ryall B, Eydallin G, Ferenci T. 2012. Culture history and coli species results in highly diverse adaptive paths. PLoS population heterogeneity as determinants of bacterial Genet 5: e1000344. adaptation: The adaptomics of a single environmental Wang HC, Susko E, Roger AJ. 2006. On the correlation transition. Microbiol Mol Biol Rev 76: 597–625. between genomic GþC content and optimal growth Sanchez-Alberola N, Campoy S, Barbe J, Erill I. 2012. Anal- temperature in prokaryotes: Data quality and con- ysis of the SOS response of Vibrio and other bacteria with founding factors. Biochem Biophys Res Commun 342: multiple chromosomes. BMC Genomics 13: 58. 681–684. Shee C, Gibson JL, Darrow MC, Gonzalez C, Rosenberg SM. Wrande M, Roth JR, Hughes D. 2008. Accumulation of 2011. Impact of a stress-inducible switch to mutagenic mutants in “aging” bacterial colonies is due to growth repairof DNA breaks on mutation in Escherichia coli. Proc under selection, not stress-induced mutagenesis. Proc Natl Acad Sci 108: 13659–13664. Natl Acad Sci 105: 11863–11868.

Advanced Online Article. Cite this article as Cold Spring Harb Perspect Biol doi: 10.1101/cshperspect.a018077 13 Downloaded from http://cshperspectives.cshlp.org/ on September 28, 2021 - Published by Cold Spring Harbor Laboratory Press

Mutation−−The Engine of Evolution: Studying Mutation and Its Role in the Evolution of Bacteria

Ruth Hershberg

Cold Spring Harb Perspect Biol 2015; doi: 10.1101/cshperspect.a018077

Subject Collection Microbial Evolution

Not So Simple After All: Bacteria, Their Population Genome-Based Microbial Taxonomy Coming of Genetics, and Recombination Age William P. Hanage Philip Hugenholtz, Adam Skarshewski and Donovan H. Parks Realizing Microbial Evolution Horizontal Gene Transfer and the History of Life Howard Ochman Vincent Daubin and Gergely J. Szöllosi Thoughts Toward a Theory of Natural Selection: Early Microbial Evolution: The Age of Anaerobes The Importance of Microbial Experimental William F. Martin and Filipa L. Sousa Evolution Daniel Dykhuizen Coevolution of the Organization and Structure of Microbial Speciation Prokaryotic Genomes B. Jesse Shapiro and Martin F. Polz Marie Touchon and Eduardo P.C. Rocha Mutation−−The Engine of Evolution: Studying The Evolution of Campylobacter jejuni and Mutation and Its Role in the Evolution of Bacteria Campylobacter coli Ruth Hershberg Samuel K. Sheppard and Martin C.J. Maiden The Origin of Mutants Under Selection: How Paleobiological Perspectives on Early Microbial Natural Selection Mimics Mutagenesis (Adaptive Evolution Mutation) Andrew H. Knoll Sophie Maisnier-Patin and John R. Roth Evolution of New Functions De Novo and from Preexisting Genes Dan I. Andersson, Jon Jerlström-Hultqvist and Joakim Näsvall

For additional articles in this collection, see http://cshperspectives.cshlp.org/cgi/collection/

Copyright © 2015 Cold Spring Harbor Laboratory Press; all rights reserved