Thank you for visiting Nature. The browser version you are using has limited support for CSS. For the best experience, we recommend that you use a newer version of the browser (or turn off the compatibility mode in Internet Explorer). At the same time, to ensure continued support, we will display sites without styles and JavaScript.

Nature Communications Volume 12, Article Number: 6092 (2021) Cite this article

Mutation hot spots can determine the outcome of evolution and make evolution repeatable. Hot spots are the product of multiple evolutionary forces including mutation rate heterogeneity, but this variable is often difficult to identify. In this work, we revealed that a near-deterministic genetic hotspot can be established and destroyed by a few silent mutations. We observed this when studying homologous immobile variants of Pseudomonas fluorescens, AR2 and Pf0-2x. AR2 restores motility through highly reproducible de novo mutations of the same nucleotide in> 95% of cell lines in minimal medium (ntrB A289C). However, Pf0-2x has evolved through many mutations, which means that there are significant differences in the adaptation process between the two strains. We determined that this evolutionary difference was only attributable to 6 synonymous variants within the ntrB locus, and we proved by exchanging sites and observing that we can destroy (>95% to 0%) and construct (0% to 80%) This point) a deterministic mutation hotspot. Our work reveals the key role of silent genetic variation in determining adaptive outcomes.

Mutation hot spots describe how independent cell lines continue to repair mutations at the same genomic site, which can make evolution remarkably repeatable. These hotspots are very important because they have been observed to promote the evolution of all areas of life, from viruses (including SARS-CoV-21) to bacteria (including MRSA2) to higher eukaryotic cell lines, including avian 3 and human cancer 4. Our understanding of evolutionary dynamics (such as competitive selection and clonal interference) can sometimes explain the emergence of hotspots, but little is known about the genetic characteristics of hotspots that are biased toward mutation rates.

There are many examples of experimental systems evolving through repeatable evolution. Microorganisms that evolve under strong selection usually quickly adopt similar new phenotypes5,6. In addition, these phenotypes are usually based on mutation hotspots, which appear in the form of clusters of genetic changes within the same region of the genome or within limited sites 9, 10, 11, and 12. Sometimes the mutations achieved are only found in genes from a single regulatory pathway13,14 or a single protein complex15. In rare cases, it can be seen that evolutionary events are repeated for a few sites within a single locus16,17. Repeatable evolution allows lines to evolve in parallel, and the degree of parallelism generally becomes less common because it drops from a wider genomic region to nucleotides 18,19. However, although repeatable evolutionary events are often described, there is often a lack of detailed understanding of the hot spots that ensure their occurrence.

The mutation hotspots that drive reproducible evolution have three main factors: (i) fixed bias, which biases evolution toward mutations that are more likely to dominate the population pool. Not all promoters of fixed bias are considered to have adaptive advantages (for example, homologous recombination events in the mammalian genome can bias gene conversion towards certain alleles20). However, when we observe fast and highly parallel scanning fixation deviations, the form of selection may be adopted, which will prompt the fixation of the most suitable competing genotype in the population 21,22). (ii) Mutation accessibility, because a genotype may only have a few easily available mutations to improve fitness23. And, (iii) mutation bias, where genetic and molecular features scattered throughout the genome cause sites to mutate at different frequencies and mutate toward certain mutation types (for example, A:T> G:C), thereby limiting Mutation profiles to support specific results 24. Previous studies have shown that mutation rate heterogeneity may be affected by the arrangement of nucleotides around specific sites, as well as genetic features such as the secondary structure of DNA26, including the formation of single-stranded DNA hairpins 27. However, the importance of gene sequence in driving the outcome of parallel evolution remains unknown.

In order to determine which mechanisms are at work, it is important to consider whether the parallel results are robust to experimental conditions such as the environment, and to consider clonal interference, which can change the chance of observing parallel evolution24,29. Clonal interference may be due to long-term genetic variation in the founder population that produces multiple adaptive genotypes in a new environment (ie, soft selective scanning), or when the mutation supply rate is high relative to the selectivity coefficient31. When cloning samples are used to establish experimental lines, perform experimental procedures in a short time, and ensure rapid fixation of adaptive mutants (for example, through spatial separation and/or introduction of artificial bottlenecks), clonal interference usually does not play an important role. However, in this case, the main effect of selection will be clonal interference, because a large number of starting populations may produce multiple adaptive genotypes, which compete for fixation throughout the experiment.

We used an ideal system to identify the key features of constructing mutation hotspots by using two engineered non-flagellate and biosurfactant-deficient soil bacteria Pseudomonas fluorescens strains: AR2, derived from SBW25, and Pf0-2x, derived from Pf0-1 (see method). These strains share a homologous genetic background, including highly similar gene regulatory structures and translated protein products, but evolutionary differences have been observed in previous work33. Both engineered strains lack the function of FleQ, the main regulator of flagella-dependent movement, and AR2 and Pf0-2x rapidly re-evolve flagella-mediated movement under strong directed selection33. In AR2, this phenotype is achieved in independent lineages through reproducible de novo mutations in the ntrB locus of the nitrogen-regulated (ntr) pathway. The parallel evolution of ntrB mutants is worth noting because this locus is always the target, and the Pf0-2x line regulates the mutational evolutionary mobility of the hierarchical structure across ntr. Since this parallel evolution between these homologs varies with scale; both are parallel to the phenotype and targeted gene regulatory network, but only one has a mutation hotspot that concentrates mutations in a single nucleus within a single locus Glycolic acid site. We conducted a series of experiments to find out why.

Here, we show that the motility in AR2 evolves in a highly reproducible manner, which is not present in Pf0-2x due to genetic characteristics based on synonymous mutations. In minimal medium (M9), more than 95% of cases found that the evolution of flagellar movement in AR2 targeted the same nucleotide substitution as the target. In the non-exercise SBW25 variant (AR2) and another SBW25 variant (SBW25 ΔfleQ) that can obtain biosurfactant-mediated motility before evolution, the results are robust in a variety of nutritional regimens . The role of selection and the number of feasible mutation paths in ensuring parallel results provides some explanation for reproducible evolution to the level of the ntrB locus, rather than nucleotides. Therefore, this means that mutational bias within the site is playing a key role. We reinforce this implication by genetically increasing the ntrB locus to indirectly attribute the mutation bias to and reveal the key potential genetic drivers of parallel evolution. Six silent nucleotide changes were introduced in the local area around frequently targeted sites, which matched the gene sequence of AR2 with Pf0-2x, but did not change the protein product. These synonymous changes reduce the parallel evolution of mutation hot spots from> 95% to 0%. In reciprocity experiments, the introduction of silent changes into the homologous strain Pf0-2x to match the native natural sequence of AR2 increased the parallel evolution of the site from 0% to 80%. These results indicate that local gene sequences can play a leading role in ensuring parallel evolutionary results, and make people pay attention to the neglected mechanism drivers behind mutation hotspots.

In order to quantify the degree of parallel evolution of flagellar motion in the fixed SBW25 model system, we placed 24 independent AR2 repeats under strong direction selection in the minimum medium environment (M9). It is easy to identify locomotor mutants by emerging locomotor regions migrating outward in concentric circles (Figure 1A). The cloned samples were isolated from the front of the region within 24 hours after the appearance, and their genotypes were analyzed by whole-genome sequencing or targeted Sanger sequencing of the ntrB locus. Motile strains evolved rapidly (Figure 1B), and it was found that each independent line was the product of a one-step de novo mutation. All 24 strains evolved in parallel at the locus level: each strain acquired a single movement-restoring mutation in ntrB (Figure 1C). What is more surprising, however, is the level of parallel evolution within the locus. The 23/24 repeat acquired a single nucleotide polymorphism at position 289, resulting in a transversion mutation from A to C (hereinafter referred to as ntrB A289C). This resulted in a T97P missense mutation in the PAS domain of NtrB. The remaining samples received a 12-base pair deletion from nucleotide positions 406-417 (Δ406-417), resulting in an in-frame deletion of residues 136-139 (ΔLVRG) in the phosphate receptor domain of NtrB.

A The immobile population that evolved on soft agar (left) re-evolves flagella-mediated movement through one-step de novo mutation (right). The phenotype of B on M9 minimal medium appears rapidly, usually within 3-5 days after inoculation (sample size N = 33 independent biological replicates). The median is displayed in the center, and the border of the box represents the 25th and 75th percentiles. The whiskers extend from the border of the box to the minimum and maximum values, and their length can reach 1.5 * interquartile range. The individual data points are also plotted. C. Potential genetic changes are highly parallel. All independent strains target one of the two loci (left circle, A289C and right circle Δ406-417) within the ntrB locus, while sacrificing the nitrogen (ntr) pathway Other sites. The DA single transversion mutation A289C is the most common mutation pathway, occurring in more than 95% of independent strains (N = 23/24). The source data is provided as a source data file.

Reproducible evolution may be robust or highly dependent on the environment, especially when it occurs through de novo mutations that antagonize pleiotropic effects34,35,36. However, we found that the reproducibility of the ntrB A289C mutation is robust under all tested conditions, despite the evidence that it has an antagonistic pleiotropic effect on growth. We measure the relative growth of the ancestral line and the relative growth on a minimal medium rich in lysogenic broth and containing ammonia as the sole nitrogen source or supplemented with glutamate (M9 glu) or glutamine (M9). The test environment specifically antagonizes the pleiotropic gln), both of which are naturally absorbed and metabolized by the ntr system. Although there is clearly a large fitness cost in the M9 minimal medium, supplementing M9 with glu or gln reduced the antagonistic pleiotropic levels of ntrB A289C and Δ406-417 mutants (Supplementary Figure 1). In fact, in M9 supplemented with the amino acid glutamine (M9 gln), the metabolically impaired antagonistic pleiotropic effect is low enough, and the exercise mutant increases the adaptability of the ancestral line in the static broth, which is in ntrB A289C Is significant (P = 0.0361, Supplementary Fig. 1). These findings indicate that the antagonistic pleiotropic effect is severe in M9 and is significantly reduced in other nutrient environments, so evolution in the minimal medium may limit the number of feasible adaptive pathways.

Then, we tested whether reproducible evolution is robust to different levels of antagonistic pleiotropy in our model system. Our expectation is that supplemental nutrition programs will reduce the cost of pleiotropy, thus opening up alternative ways of adaptation. We also hypothesized that strains that can migrate before mutation can also alleviate the selective pressure caused by starvation and promote more mutation pathways. Therefore, in this experiment, we used an additional immobile variant of SBW25. Unlike AR2, it does not insert a viscB transposon (see method), so it can migrate through the sliding motion before mutation (SBW25-ΔfleQ( Hereinafter referred to as ΔfleQ) 37). We observed the evolution of the "blistering" phenotype in the ΔfleQ line (Figure 1A), although they can migrate in a dendritic manner; however, we also found that the evolution of bubbles is less frequent under a richer nutritional system ( Use mucin to make populations move faster, see method). Overall, there is no evidence that the prevalence of the hotspot mutation ntrB A289C varies with nutritional conditions (gene-environment interaction: χ2 = 0.875, df = 3, P = 0.83, see Figure 2). In contrast, we observed that the ntrB A289C mutation is robust under all tested conditions, characterized by 90-100% duplication in the ΔfleQ strain and 80-100% duplication in the AR2 strain (Figure 2).

The proportion of each observed mutation is shown on the y-axis, and each unique mutation target is highlighted in color. Use 4 mM, 8 mM and 16 mM amino acid supplements glutamate (glu) or glutamine (gln; see methods) to evolve the strain. No significant relationship between supplement concentration and evolutionary goals was observed (Kruskal-Wallis chi-square test, both sides: AR2 M9 glu, df = 2, P> 0.2; AR2 M9 gln, df = 1, P> 0.23; ∆fleQ M9 gln, df = 1, P> 0.3), so they are grouped visually and used for subsequent statistical analysis. The ntrB mutation A289C is robust in both strain backgrounds (SBW25ΔfleQ-shown as ΔfleQ and AR2) and the four tested nutritional environments, and remains the main target of mutation in all cases (>88%). Therefore, no significant gene-environment interaction was observed (χ2 = 0.875, df = 3, P = 0.83). The ΔfleQ line evolved on LB can quickly migrate through sliding motion alone, concealing any potential emergency flagella (see 37). Sample size (N) of other categorical variables: ΔfleQ-M9: 25, M9 gln: 20, M9 glu: 7; AR2-LB: 5, M9: 24, M9 gln: 17, M9 glu: 18. The source data is provided as a source data file.

Our evolutionary experiments across the nutritional system revealed three new mutation pathways observed in a small number of mutants (Figure 2), indicating that mutation accessibility cannot explain the level of parallel evolution observed. The most notable is the non-synonymous AC transversion mutation at position 683 (ntrB A683C) in the ΔfleQ line evolved by M9 gln, resulting in a missense mutation in the NtrB histidine kinase domain. As a single AC transversion within the same site, we can expect A683C to mutate at a similar rate as A289C. We also observed a 12-base pair deletion at positions 410-421 (ntrB Δ410-421) in the AR2 strain evolved on M9 gln. In addition, we found a double mutant in the AR2 line that evolved on M9 glu: one mutation was a single nucleotide deletion at position 84 leading to a frameshift within glnK, and the second mutation was another A to position 688 The C transversion results in a mutation in the T230P missense RNA polymerase Sigma factor 54.

GlnK is the natural regulatory binding partner of NtrB and a repressor in the ntr pathway, which means that only frameshift mutations may explain the observed motor phenotype. However, since this mutant has undergone two independent mutations, we will not consider it in the following analysis. In addition, ntrB Δ410-421 and ntrB Δ406-417 are converted to the same protein product despite targeting different nucleotides (both compress residues LVRGL at positions 136-140 into a single L at position 136). Therefore, we also group them for the following analysis. Under the assumption of the new protein mutation pathway observed in the three remaining steps (i) may also appear in the population and (ii) may also be fixed, the original observation that ntrB A289C appeared in 23/24 cases became abnormal ( Boot test: n = 1000000, P <1 × 10−6). Therefore, it is extremely unlikely that we will accidentally observe this. This means that one or two assumptions are almost certainly incorrect. Mutation-promoted motor phenotypes may be unequal, so that clonal interference can achieve reproducible results; or adaptive mutation profiles may appear in the population at different rates, leading to mutation bias. One or two of these elements must tilt the evolution to a certain degree so that parallel evolution with nucleotide resolution becomes highly predictable.

The adaptationist explanation for parallel evolution is that the observed mutation path outperforms all other paths in a fixed process. For our experimental purposes, we define fixation as being established on the boundary of the motion area at the time of sampling. If selection by clonal interference alone will promote reproducible evolutionary results, then the superior adaptability of the ntrB A289C genotype should allow it to migrate to other sports genotypes that coexist in the population. In order to test whether the ntrB A289C mutation confers an optimal exercise phenotype, we allowed the evolved genotypes (A289C, Δ406-417, A683C, and glnK Δ84) to migrate independently under four nutritional backgrounds, and measured their migration area after 48 hours . For direct comparison, we first designed the ntrB A683C mutation that originally evolved in the ΔfleQ background as an AR2 strain. We observed that the migration speed of non-ntrB double mutant glnK Δ84 in all four nutritional backgrounds was significantly slower than that of ntrB A289C (M9: P = 0.000708, M9 gln: P = 0.032, M9 glu: P = 0.0025, LB: P = 0.0048, Figure 3). However, ntrB A289C is not significantly better than the alternative ntrB mutant line under any environmental conditions (P value range = 0.074-0.87 Figure 3). This suggests that selection may have played a role in promoting parallel evolution to the level of the ntrB locus, but it cannot explain why nucleotide position 289 is mutated so frequently.

Mutants ntrB A289C, ntrB Δ406-417, ntrB A683C and glnK Δ84 were grown for 48 hours under the four environmental conditions, the surface area of ​​the motor area in the AR2 genetic background; rpoN A688C. Lysogenic broth (LB, N = 9, 9, 5, 4 independent biological replicates), ammonia-containing M9 basic medium (M9, N = 7, 6, 2, 5) and supplemented with amino acid valley Amino acid M9 (M9 glu, N = 5, 5, 5, 4) or glutamine (M9 gln, N = 5, 5, 5, 6) are used as nutritional conditions. A single data point from a biological replicate is plotted, and each migration area is normalized to the surface area of ​​AR2 ntrB A289C mutants grown in the same environment (ntrB A289C growth average = 0). The significance value of glnK Δ84; rpoN A688C: M9 P = 0.000708, M9 glu P = 0.0025, M9 gln P = 0.032, LB P = 0.0048 (Kruskal-Wallis post hoc Dunn test, one-sided). The source data is provided as a source data file.

In order to determine whether the result is still correct when the mutant line has the opportunity to perform clonal interference, we directly competed ntrB A289C with the alternative ntrB mutant lines Δ406-417 and A683C on the M9 minimal medium. In short, we co-inoculated three mutant lines on the same soft agar surface at the same concentration and allowed them to migrate competitively before sampling from the front after 24 hours of competition. In 15 independent replicates, we did not observe any significant deviation of ntrB mutation in the growth front (ntrB A289C = 4/15, ntrB Δ406–417 = 8/15, ntrB A683 = 3/15; Bootstrap test: n = 1000000, P> 0.26). We next simulated ntrB A289C that appeared in the population within a few generations after the substitution mutation, and observed that the common genotype was significantly better than the common genotype (ntrB A289C when inoculated 6 hours and 3 hours after the substitution mutation line). Establish at the leading edge = 0/16 independent repetitions (3 hours and 6 hours); Bootstrap test: n = 1000000, P <0.005). These results emphasize that ntrB A289C will not show evidence of clonal interference if the locomotor line is present in the population at the same time in the minimal medium. In addition, if the common genotype appears in the population after a few generations of its competitors, it cannot establish itself in the frontier. In addition, given that the time frame before the observed exercise phenotype may vary significantly between independent strains (see above), our data does not support that the mutation supply may be high enough to allow multiple phenotype-granting mutations to appear in as in The population is almost at the same time as explored in this measurement. Therefore, our evidence shows that the chance of clonal interference will be minimal in the short-term process of the experiment. If it occurs, there is no evidence to support that it is a pathogen that we can observe repeatedly for ntrB A289C. It is more likely that each independent strain attached to the early bird received the worm criterion, that is, the ntrB mutant that first appeared in the population was the genotype that was sampled later. Therefore, this suggests that ntrB A289C is collected so frequently when sampling is at least partly due to evolutionary power beyond selection and mutation accessibility.

Local mutation bias can play a key role in evolution24,38. This deviation can be introduced by changing the curvature of the DNA or through adjacent inverted complementary repeats (palindromes and quasi-palindromes), which have been shown to cause local mutational deviations by promoting the formation of single-stranded DNA hairpins. Therefore, we next searched for local mutation bias at ntrB position 289. Previously, we re-evolved the motility of two engineered non-motile strains of Pseudomonas fluorescens, AR2 (derived from SBW25) and Pf0-2x (derived from Pf0-1)33. Although the strains evolved in AR2 often target ntrB, the Pf0-2x strain fixes mutations in the entire ntr regulatory pathway. In addition, although Pf0-2x did acquire ntrB mutations in multiple independent strains, we did not observe evidence that ntrB site 289 was targeted. The NtrB proteins of SBW25 and Pf0-1 are highly homologous (95.57% identity), but their genetic identity is low (88.88% identity). A large part of this genetic variation is explained by synonymous genetic variation (8.36%) rather than non-synonymous variation (2.76%). Synonymous mutations can play a role in changing the bias of local mutations. This can occur by changing the nucleotide triplet to a triplet with a higher mutation rate or by changing the secondary structure of longer DNA fragments through the mechanisms described above. It has been observed that nucleotides that remain unpaired when adjacent nucleotides form hairpins with nearby reverse complementary bundles show an increased mutation rate39. It was found that both SBW25 and Pf0-1 have short inverted complement bundles on both sides of position 289. However, due to the synonymous variation, the composition called hairpin is not exactly the same (Supplementary Figure 2). In general, there are 6 synonymous nucleotide substitutions ±5 codons on both sides of position 289 (C276G, C279T, C285G, C291G, T294G, and G300C), which may have affected this hairpin structure and affected The local mutation rate.

In order to test whether the synonymous sequences are biased against the evolutionary results, we replaced the 6 synonymous sites in the AR2 strain with those from the Pf0-1 background (hereinafter referred to as AR2-sm). Not all of these sites constitute part of the theoretically predicted stem that overlaps with position 289, but since they are very close to this site, all of these sites have become targets. This ensures that these changes capture any secondary structure formed in the local area around nucleotide position 289. The AR2-sm line was placed under exercise selection, and we observed that the motility evolution of these lines was significantly slower (Figure 4A), all in the M9 minimum medium and LB (Wilcoxon rank sum test with continuity correction: M9, W = 44.5, P <0.001; LB, W = 22, P <0.001). The evolutionary AR2-sm lines that re-evolved movement within 8 days were sampled, and their ntrB locus was analyzed by Sanger sequencing (Figure 4B). We observed some ntrB mutations similar to those previously identified: ntrB A683C mutation was observed in an independent line of LB evolution, and ntrB Δ406-417 was also observed in the background of two strains. However, the most common genotype of ntrB A289C dropped to 0% from that observed in more than 95% of independent lines in M9. In addition, we observed multiple previously unseen ntrB mutations, and quite a few lines reported wild-type ntrB sequences instead of targeting another gene of the ntr pathway (glnK) or unidentified targets that may be outside the network (Figure 4B).

Histograms of the appearance time of independently replicated locomotor phenotypes of non-motile SBW25 (AR2, gray) and AR2 strains with six synonymous substitutions in the ntrB locus (AR2-sm, white) under two nutritional conditions. The evolved lines after the experimental time frame (10 days) are merged into bin 11. B. The target of mutation observed after directed evolution of synonymous variants is performed in two environments. Each unique mutation is highlighted in an identifiable color. (Sample size (N): AR2: LB N = 5, M9 N = 24; AR2-sm: LB N = 8, M9 N = 8). Please note that the characteristic genotypes were sampled within 8 days of the start date of the experiment. Unidentified mutations cannot be distinguished from wild-type sequences of genes belonging to nitrogen-regulated pathways (ntrB, glnK, and glnA) analyzed by Sanger sequencing. ntrB Δ406-417 is the only mutation target shared by two strains in the same nutritional environment. The source data is provided as a source data file.

In order to test that A289C transversion is still a viable mutation target in the AR2-sm genetic background, we subsequently designed an AR2-sm strain with this kind of locomotor mutation. We observed that AR2-sm ntrB A289C is motor and phenotypically comparable to the ntrB A289C mutant that evolved in the ancestral AR2 genetic background (Supplementary Figure 3). We also found that AR2-sm ntrB A289C retains the same motility as other ntrB mutants evolved from AR2-sm (Supplementary Figure 3). Therefore, we can determine that the AR2-sm genetic background will not prevent the motility after the mutation at ntrB position 289, nor will it make this mutation uncompetitive. Therefore, it is inferred that the only variable (six synonymous changes) that changed between the two strains excludes the mutation at position 289. In summary, these results strongly indicate that the synonymous sequence immediately adjacent to ntrB site 289 contributes to its location as a local mutation hotspot, and that local mutation bias is essential for achieving highly parallel evolution in our model system.

Since the previous results exemplified the power of synonymous mutations in breaking mutation hotspots, we next assumed that mutation hotspots could be established with the same number of mutations. To achieve this, we designed a synonymous variant of the immobilized Pf0-2x strain (Pf0-2x-sm6). This strain is a reciprocal mutant of AR2-sm, because it has synonymous mutations at the same six positions in ntrB, but it is replaced with the natural sequence of AR2 (G276C, T279C, G285C, G291C, G294T and C300G) Match. We put Pf0-2x and Pf0-2x-sm6 under the directional selection of motion, and observed that the movement evolution speed of Pf0-2x is slower than that of Pf0-2x-sm6 (Figure 5A), and is targeted at multiple sites. Multiple sites (Figure 5B)). In sharp contrast, Pf0-2x-sm6 evolves faster (Figure 5A; Wilcoxon rank sum test with continuity correction: M9, W = 239.5, P <0.001; LB, W = 461.5, P <0.001 ) And a lot more parallel to its local counterparts. Pf0-2x-sm6 fixed ntrB A289C (8/10 independent line) in 80% of M9 instances, although this de novo mutation did not occur once in the Pf0-2x evolution line (0/22 independent line, Figure 5B) . The significant difference between the two strains from the Pf0-2x genetic background (Figure 5) clearly reflects the results observed in the AR2 genetic background (see above). This indicates that a small amount of synonymous mutations will seriously affect the results of mutations across genetic backgrounds and homologous strains.

Pseudomonas fluorescens strains Pf0-1 (Pf0-2x33, gray) and Pf0-2x strains independently replicated the histogram of the appearance time of the motor phenotype, in the ntrB locus (Pf0-2x-sm6, Pf0-2x-sm6 , White) under two nutritional conditions. B. The target of mutation observed after directed evolution of synonymous variants is performed in two environments. Each unique mutation is highlighted with an identifiable color. (Sample size (N): Pf0-2x: LB N = 29, M9 N = 22; Pf0-2x-sm: LB N = 6, M9 N = 10). Unidentified mutations cannot be distinguished from wild-type sequences of genes belonging to nitrogen-regulated pathways (ntrB, glnK, and glnA) analyzed by Sanger sequencing. The mutation ntrB A289C was not observed in a single instance in the evolved Pf0-2x strain, but it became a strongly preferred target after synonymous substitution. The source data is provided as a source data file.

Understanding the evolutionary forces that form mutation hotspots and repeatedly drive certain mutation fixation remains a huge challenge. This is true even in the simple system used in this study, in which the cloned bacterial population has evolved under strong directed selection of very few phenotypes (ie, motility and nitrogen metabolism). Here, we used immobile variants of Pseudomonas fluorescens SBW25 (AR2) and Pf0-1 (Pf0-2x), which have been observed to repeatedly target the same gene regulatory pathway during the re-evolution of motility . We found that in more than 95% of cases in M9 minimal medium, the evolutionary population of AR2 adapts through de novo substitution mutations in the same site (ntrB) and the same nucleotide site (A289C). The AR2 population is restricted. What genetic pathways can they take to access the selected phenotype, but the availability of mutations and clonal interference alone cannot explain such a high degree of parallel evolution. Pf0-2x is unique in that it does not evolve in parallel with nucleotide or locus resolution. We observed that by introducing synonymous changes around the mutation hotspot (ntrB site 289) in AR2 and Pf0-2x in order to exchange their local gene sequences, we can push the evolving AR2 population away from the parallel path and move Pf0 The -2x line is pulled to a parallel path. This work shows that synonymous sequences are an indispensable factor in achieving highly reproducible evolution and establishing mutation hotspots in our system.

Recent studies have shown that the impact of synonymous changes on fitness through interference before and during translation is underestimated. Synonymous sequence variation can affect fitness by changing the stability of mRNA 40, 41, and 42 and changing the codons to disrupt or better match the codon-anticodon ratio. We show here that synonymous sequences are also essential to ensure parallel evolutionary results across genetic backgrounds. Our results strongly infer that this is due to its effect on local mutation bias, which may be mechanistically due to the formation of single-stranded hairpins between short inverted repeats on the same DNA strand27,44. The formation of these secondary DNA structures provides a mechanism for mutational bias within the site, which can work under local influence and depends on DNA sequence variation, because the introduction of synonymous changes can easily disrupt adjacent inverted repeats. Complementarity (eg Supplementary Figure 2). In addition, the discovery of only six synonymous mutations that have a significant impact on DNA structure does not represent a surprising result, because the secondary structure can be changed by a single mutation45.

We can confidently assert that since the 6 synonymous sites are located within 14 bases of either flanking position 289, the altered mutation bias is due to intra-site effects. This powerful mutation bias needs further study. We know that at least some of the six substituted nucleotide positions are essential for parallel genetic results, but we do not yet know local neighborhoods or broader other nucleotide features (such as strand direction 46 or distance replication Distance from the starting point 25) Is it possible to use local sequences to force mutation deviations. Interestingly, our data shows that mutation hotspots usually mutate so quickly that they mask mutations that appear elsewhere and outside of the nitrogen-regulated pathway, and these mutations only appear when the hotspot is disturbed (Figure 4 and 5). Therefore, this provides an additional opportunity to quantify the difference in mutational deviation caused by the secondary structure.

Our research results show that the existence of mutation hotspots is a stronger deterministic evolutionary force in our system than other variables (such as nutrition system, starvation-induced selection, and genetic background). We expect that the selective environment will have some influence on the outcome of evolution18 mainly due to the different levels of antagonism pleiotropic, which has been found to be a key driver of similar exercise research. Similarly, while parallel evolution is sometimes impressive in genetic backgrounds47, some innovations depend to a large extent on the evolutionary history of organisms48. Genome variation is often combined with environmental differences, prompting populations to take different paths49. However, in our experiments, strains with the same 6 synonymous sites evolved more similarly than strains with the same broader genetic background (Figure 4B and 5B). These results indicate that strains can not only share a high degree of global homology, but also share similar genome structures—including translated protein structure and gene regulatory network organization—but due to synonymous mutations, it is possible to select exactly the same traits. Will produce very different mutation results. This raises an interesting question as to whether neutral changes can promote the advantage of genotype in the adaptation process due to previously obtained mutation hotspots, and asked whether these mutations can be selectively enforced Hot spot.

Models seeking to describe the drivers of adaptive evolution usually prioritize fitness and the number of accessible adaptive routes50,51, but rarely pay attention to local mutation bias (but, see 36). However, the heterogeneity of mutations becomes critical when the system follows the strong selection weak mutation model (SSWM), which describes a situation where a favorable mutation undergoes a hard scan fixation before another beneficial mutation appears52. In this case, the relative fitness value between the adaptive genotypes is downgraded to secondary importance after the likelihood of the adaptive genotypes appearing in the population. In fact, experimental systems that comply with the SSWM guidelines have been observed to evolve in parallel, although multiple mutation pathways can be selected to improve fitness47. This suggests that uneven mutation bias may be a key driving factor for the formation of mutation hotspots and parallel evolution. This conclusion has been theoretically strengthened24 although there is still a lack of empirical data. Therefore, if we are to determine the existence of mutation hotspots that allow accurate prediction of evolution, then understanding the mechanism of mutation rate heterogeneity will be essential38,53. The challenge remains to determine what these mechanical quirks might be, where they might be found, and how they affect the outcome of evolution.

Our work clarifies the ability of silencing genetic variation to establish mutation hotspots with functionally significant evolutionary consequences. This hot spot is constructed by an adaptive site under strong directional selection. The site enjoys biased mutations and promotes highly reproducible evolution when the mutational bias and selection are aligned. Mutation is essentially a random process, but not all sites in the genome have the same fixation potential. Most changes will not improve the phenotype under selection, and those changes will not necessarily mutate at the same rate. Therefore, we can significantly improve our ability to predict the location of mutation hotspots, allowing us to understand in detail the evolutionary variables that are at play. Considerable progress has been made in achieving this goal. When looking for adaptive targets, it has been emphasized that loss-of-function mutations are the most commonly observed mutation types under selection54,55 and the wider position of genes in their regulatory networks determines their propensity to transmit phenotypic changes56. When looking for mutational deviations, it has been shown that parallel evolution at the locus level is partly determined by gene length53 and molecular devices involved in replication and repair can strongly influence the possibility of a given nucleotide substitution57, 58. Here, we emphasize the influence of synonymous sequences on the realization of highly repeatable evolution, showing that synonymous sequences are worth considering together with other variables.

Our model system uses strains of soil microorganisms P. fluorescens SBW25 and Pf0-1, which lack motility through partial gene deletion or destruction of fleQ (the main regulator of flagellar movement 37, 59). In the absence of FleQ, motility can be restored after remutation, which allows the recruitment of homologous response modifiers, of which NtrC of the nitrogen-regulated pathway is the easiest to target. The initial mutations that promote NtrC recruitment occur at other sites in the nitrogen pathway, leading to hyperphosphorylation of NtrC33. In this study, two SBW25 derivative strains were used as ancestors: SBW25 ΔfleQ (hereinafter referred to as ΔfleQ) and a ΔfleQ variant with functional viscB knockout isolated from the transposon library (SBW25ΔfleQ IS-ΩKm-hah : PFLU2552, hereinafter referred to as AR237. The ΔfleQ on AR237 can migrate. Soft agar (0.25%) is due to the ability of the strain to produce mucin before it is mutated by sliding motion. AR2 cannot produce mucin, so it is completely immobile before mutation Pf0-1 is a natural gacA mutant 60 and therefore does not produce mucin, so its ΔfleQ variant Pf0-2x becomes completely immobile after fleQ is interrupted. All cells are grown at 27°C and used throughout the study All strains (ancestors, evolution and engineering) are stored at -80 °C in 20% glycerol. The nutrient conditions used in the whole work are lysogen broth (LB) and M9 basic medium containing glucose and 7.5 mM NH4 The minimal medium is used alone or supplemented with glutamate (M9 glu) or gl. Unless otherwise specified, the final supplementation concentration of utamine (M9 gln) is 8 mM. The complete list of primers used throughout the study can be found in Supplementary Table 1. Found in.

Use LB and M9 soft agar (0.25%) exercise plates to select non-exercise variants for flagella-mediated exercise. The details of agar preparation are in 37. In short, an aliquot of melted 30 ml soft agar is added to a petri dish with a diameter of 88 mm and cooled at room temperature for four hours. Thereafter, the condensate on the agar surface and the lid was removed by drying the plate in a laminar flow hood for 30 minutes. The supplementary concentration of glutamate (glu)/glutamine (gln) in M9 soft agar was expanded to include final concentrations of 4 mM, 8 mM, and 16 mM because of the observation of dendritic movement mediated by biosurfactants in the ΔfleQ system The concentration is enhanced at higher supplements, and any bubbles that appear are masked. Decreasing the concentration of gln supplements increases the possibility of observing flagellar vesicles in the M9 gln exercise board (16 mM: 4/12, 8 mM: 9/20, 4 mM: 7/12 independent lines). However, in all M9 glu supplements and continuous blister masking (16 mM: 2/12, 8 mM: 3/20, 4 mM: 2/11 independent lines), dendritic movement is still high. Although supplementation of gln/glu has no effect on the motility of AR2 strains, the supplementary conditions of gln/glu have been expanded for consistency. Each sports plate is inoculated with a single monoclonal colony, which is derived from a striped plate prepared from a cloned cryogenic stock solution. Using colonies to start the test minimizes the number of generations from the cloning of low-temperature ancestors to the beginning of the test, helping to ensure the cloning of the starting population. Use a sterile pipette tip to inoculate a single colony into the center of the agar and monitor it daily until an area of ​​active bubbles appears (as shown in Figure 1A). Separate the sample from the leading edge, select the strongest motor phenotype on the plate within 24 hours after its appearance, and streak it on LB agar (1.5%) to obtain a cloned sample. Since the ΔfleQ line moves through dendritic movement before re-evolving flagellar movement and can visually mask the flagella-mediated movement area, the sample was placed for 120 hours before sampling from the leading edge of growth. An exception is the case where the bubbling movement area is observed only in the growth area, in which case the area is preferentially sampled.

Through PCR amplification and sequencing of ntrB, glnK, and glnA genes, the changes that promote exercise are determined. Polymerase chain reaction (PCR) products and plasmids were purified using Monarch® PCR & DNA Cleanup Kit (New England Biolabs), and Sanger sequencing was performed by Eurofins Genomics. Milner Genomics Center and MicrobesNG (LB: n = 5, M9: n = 6, M9 gln: n = 6, M9 glu: n = 7). This allows us to screen for potential secondary mutations and identify rare changes in sports strains with wild-type ntrB sequences. We did not observe adaptive secondary mutations in the sports lines that received WGS. However, all AR2 derivative strains shared the mutations from the assembled genome of SBW25 in the same 5 positions: 45877 A> AG, 985332 G> GC, 1786536 A> G, 3447980 TCC> T and 3694384 A> G. The commonality of these mutations strongly suggests that the background AR2 line differs from the reference genome at these positions. We sent another 6 Pf0-2x samples that contained unidentified WGS campaign granting mutations, and these lines did not share secondary mutations that deviated from the reference genome. P. fluorescens SBW25 genome was used as an assembly template derived from the ancestral strain (NCBI assembly: ASM922v1, GenBank sequence: AM181176.4), P. fluorescens Pf0-1 was used as an assembly template for Pf0-2x strain (NCBI) Assembly: ASM1244v1 , GenBank sequence: CP000094.2). Use Snippy with default parameters to call SNPs through the Microbial Bioinformatics Cloud Infrastructure (CLIMB62). In the case that the coverage of the called site is low (≤10x), the change of the call is confirmed by Sanger sequencing.

Cryopreserved samples of AR2 and derived ntrB mutants were streaked and grown on LB agar (1.5%) for 48 hours. Then pick out three colonies, inoculate them in LB broth, and grow them overnight under stirring at 180 rpm to create biological triplicates for each sample. The overnight cultures were pelleted by centrifugation, their supernatants were removed, and the cell pellets were resuspended in phosphate buffered saline (PBS) to a final concentration of OD595 1 cell/ml. Inoculate each replicate of 1 μl into the soft agar by piercing the top of the agar with a pipette tip and spraying the culture into the cavity when the tip is withdrawn. Incubate the plate for 48 hours and take pictures. Calculating the diameter of the concentric circle growth laterally and longitudinally allows us to calculate the average total surface area using A = πr2, and then perform a root transformation. This process was repeated because several independent strains underwent the second step mutation in the 48-hour test. This phenotype is easy to observe because the bubbles appearing on the leading edge along part of the circumference distorted the expected concentric circles of the colony migration population. Therefore, these boards were discarded from the study. By completing an additional set of three biological replicates, we ensure that each sample has at least three biological replicates for analysis, except for A683C under M9 conditions, which includes two biological replicates.

The OD corrected biological triplicate of the ntrB mutant line was prepared as described above. For three replicates of each biology, mix one OD595 unit of ntrB Δ406-417 and ntrB A683C at the same cell density (to provide a total of two OD595 units), pellet and resuspend one OD595 unit of ntrB A289C in the same volume As a mixed culture. Use 1μl of each biological replicate of the resuspended mixed culture to inoculate the above four soft agar plates and incubate at the allocated time (0 hours, 3 hours, and 6 H). At 0 hour inoculation, biological copies of ntrB A289C were added to the plate immediately after ntrB Δ406-417 and A683C, and each copy was inoculated on four soft agar plates. In the case where ntrB A289C is added to the plate 3 or 6 hours after ntrB Δ406–417 and A683C, the culture can be avoided by incubating the ntrB A289C culture at 22°C for 0 hours until the cells settle and resuspend for approximately 1 hour Overgrowth before inoculation. The same "angle of attack" is used for both inoculation examples (ie the side of the plate that the pipette tip passes through as it reaches the center), because a small amount of liquid falling from the tip onto the plate may cause local satellite growth. To avoid the risk of satellite growth affecting the results, separate samples were collected from the leading edge at an angle of attack of 180° after 24 hours. The ntrB locus of one sample per replicate was determined by Sanger sequencing to establish a dominant genotype at the growth front.

Using vector pTS1 as a template, use overlap extension PCR (oePCR) cloning (see 63 for details) to assemble the pTS1 plasmid containing ntrB A683C. ntrB synonymous mutants (AR2-sm and Pf0–2x-sm6) and AR2-sm ntrB A289C pTS1 plasmids use oePCR to construct inserts for allelic exchange, then use nested primers for amplification and restrict annealing to pTS1 vector-ligation. pTS1 is a suicide vector that can replicate in E. coli but not in Pseudomonas, and contains a tetracycline resistance cassette and an open reading frame encoding SacB. The cloned plasmid was introduced into the SBW25 strain of Pseudomonas fluorescens by mating with the water hole of the auxotrophic E. coli donor strain ST18. Using the method outlined by Hmelo et al. 64, the mutation was integrated into the genome through a two-step allelic exchange with the following adjustments: (i) Pseudomonas fluorescens cells were grown at 27°C. (ii) An additional passaging step was introduced before partial diploid selection to allow colonies of Pseudomonas fluorescens cells incorporated with plasmids (partial diploids) to grow overnight in LB broth without selection , Thereby gaining extra generation time to expel plasmids from the genome. (iii) The overnight culture was then serially diluted and spotted on NSLB agar 15% (wt/vol) sucrose for AR2 strain and NSLB agar 5% (wt/vol) sucrose for Pf0-2x strain. The positive mutant strains were identified by targeted Sanger sequencing of the ntrB locus. Merodiploids have only undergone a recombination event and will have mutant and wild-type alleles of the target locus, as well as the sacB locus and the tetracycline resistance cassette. However, after a successful two-step recombination, the wild-type allele, sacB and tetracycline resistance will subsequently be lost. Therefore, we also screened these mutants for counter-selection escape through PCR amplification and sacB locus sequencing and growth on tetracycline. The mutants are considered successful only when there is no product on the agarose gel after amplification of sacB with appropriate controls, these lines are sensitive to tetracycline, and the PCR results at the target site report the expected changes at the target site .

All statistical tests and figures are generated in R65. The graph was created using the ggplot package. The simulated data set for the Bootstrap test is randomly selected from a pool of n values, these values ​​have the same weight x times, for one million iterations. Note that for tests that check mutation profiles when discussing mutation accessibility, the simulation data set is drawn from a pool of three values, so it is possible to encode no mutation paths other than the three observed. Therefore, the derived statistics are underestimated, and any additional route of weight will reduce the possibility of repeated observation of a single value. Except for the Dunn test performed using the FSA package, all other tests are done using functions in base-R. In addition to the Bootstrap test, the statistical tests used throughout the study include: Kruskal-Wallis chi-square test, Kruskal-Wallis post hoc Dunn test, and Wilcoxon rank sum test with continuity correction.

For more information on the research design, please see the abstract of the nature research report linked to this article.

The data used to generate this manuscript is public and can be accessed [https://doi.org/10.17605/OSF.IO/VUYWP]68. The source data of the graph. 1, 2, 3, 4, 5 and supplementary diagrams. 1 and 3 are provided with the paper. Partial sequences in the form of a single locus with mutations can be found in the data repository, and a wiki page detailing the file types and nomenclature of these partial sequences can be found on the repository homepage. Access the publicly accessible Pseudomonas sequence through the Pseudomonas genome database (https://www.pseudomonas.com/), and access the SBW25 genome assembly through NCBI (NCBI assembly: ASM922v1, GenBank sequence: AM181176.4). The whole genome has been deposited in GenBank, and the access codes are: JAIOKC000000000, JAIOKD000000000, JAIOKE000000000, JAIOKF000000000, JAIOKG000000000, JAIOKH00000000000000000. This article provides source data.

The MATLAB script for ntrB sequence analysis is available on GitHub and can be accessed via the following link: https://github.com/JS-Horton/Syn-sequence-parallel-evolution. https://doi.org/10.5281/zenodo.510998469.

Weber, S., Ramirez, C. and Doerfler, W. The signaling hotspot mutations in the SARS-CoV-2 genome have evolved with the spread and active replication of the virus in different parts of the world. Virus Research 289, 198170 (2020).

CAS PubMed PubMed Central Article Google Scholar 

Sekowska, A., Wendel, S., Fischer, EC & Nørholm, MHH Generation of mutation hot spots in aging bacterial colonies. science. Representative 6, 4–10 (2016).

Galen, SC etc. Contribution of mutation hotspots to hemoglobin adaptation of Andean house wrens at high altitudes. Process National Academy of Sciences. science. United States 112, 13958–13963 (2015).

ADS CAS PubMed PubMed Central Article Google Scholar 

Trevino, V. HotSpotAnnotations — cancer hotspot mutation and annotation database. Database 1-8 (2020) https://doi.org/10.1093/database/baaa025.

Fong, SS, Joyce, AR & Palsson, B. Ø. Parallel adaptive evolution culture of Escherichia coli resulted in convergent growth phenotypes with different gene expression states. Genome research. 1365–1372 (2005) https://doi.org/10.1101/gr.3832305.15.

Ostrowski, EA, Woods, RJ & Lenski, RE The genetic basis of parallel and different phenotypic responses in the evolutionary population of Escherichia coli. Process R. Soc. B biological. science. 275, 277–284 (2008).

Riehle, MM, Bennett, AF & Long, AD The genetic structure of Escherichia coli heat adaptation. Process National Academy of Sciences. science. America. 98, 525–530 (2001).

ADS CAS PubMed PubMed Central Article Google Scholar 

Frebel, DT et al. The environment determines the evolutionary trajectory in the restricted phenotype space. Elife 6, e24669 (2017).

PubMed PubMed Central article Google Scholar 

Bull, JJ etc. The abnormal convergence and evolution in the virus. Genetics 147, 1497–1507 (1997).

CAS PubMed PubMed Central Article Google Scholar 

Wichman, HA, Badgett, MR, Scott, LA, Boulianne, CM & Bull, JJ Different trajectories of parallel evolution during virus adaptation. Science 285, 422–424 (1999).

CAS PubMed article Google Scholar 

Herron, MD & Doebeli, M. Parallel evolutionary dynamics of adaptive diversification of Escherichia coli. Public Science Library Biology. 11, e1001490 (2013).

CAS PubMed PubMed Central Article Google Scholar 

Kram, KE etc. Adaptation of Escherichia coli to long-term serial passages in complex media: evidence of parallel evolution. mSystems 2, 1–12 (2017).

Notley-McRobb, L. & Ferenci, T. Adaptive mgl in the glucose-restricted E. coli population regulates mutation and evolution of genetic diversity. environment. microorganism. 1, 33–43 (1999).

CAS PubMed article Google Scholar 

Miller, C. et al. The adaptation of Enterococcus faecalis to daptomycin reveals the orderly development of resistance. Anti-microbial. Acting for Chemother. 57, 5373–5383 (2013).

CAS PubMed PubMed Central Article Google Scholar 

Avrani, S., Boltin, E., Katz, S., and Hershberg, R. Rapid genetic adaptation during the first four months of survival under resource depletion. Mole. biology. evolution. 34, 1758–1769 (2017).

CAS PubMed PubMed Central Article Google Scholar 

Meyer, JR etc. The reproducibility and contingency of the key innovation evolution of lambda phage. Science 335, 428–432 (2012).

ADS CAS PubMed PubMed Central Article Google Scholar 

Van Ditmarsch, D. et al. The convergent evolution of Hyperswarming leads to impaired formation of pathogenic bacterial biofilms. Cell Representative 4, 697–708 (2013).

PubMed PubMed Central article CAS Google Scholar 

Bailey, SF, Rodrigue, N. & Kassen, R. The influence of selection environment on the probability of parallel evolution. Mole. biology. evolution. 32, 1436–1448 (2015).

CAS PubMed article PubMed Central Google Scholar 

Tenaillon, O. et al. Molecular diversity of adaptive convergence. Science 335, 457–461 (2012).

ADS CAS PubMed article PubMed Central Google Scholar 

The evolution of Eyre-Walker, A. & Hurst, LD, etc. Nat. Pastor Genette. 2, 549–555 (2001).

CAS PubMed article Google Scholar 

Wood, TE, Burke, JM and Rieseberg, LH Parallel genotype adaptation: when evolution repeats itself. Genetics 123, 157–170 (2005).

PubMed PubMed Central article Google Scholar 

Woods, R., Schneider, D., Winkworth, CL, Riley, MA & Lenski, RE Testing of parallel molecular evolution in long-term E. coli experiments. Process National Academy of Sciences. science. United States 103, 9107–9112 (2006).

ADS CAS PubMed PubMed Central Article Google Scholar 

Weinreich, DM, Delaney, NF, De Pristo, MA & Hartl, DL Darwinian Evolution Only a few mutation paths can find a more suitable protein. science. 312, (2006).

Bailey, SF, Blanquart, F., Bataillon, T. & Kassen, R. What drives parallel evolution? : How population size and mutations contribute to repeated evolution. Biology Paper 39, 1-9 (2017).

Long, H. et al. The mutation rate, spectrum, topology, and context dependence of DNA mismatch repair defects in Pseudomonas fluorescens ATCC948. Genomic biology. evolution. 7, 262–271 (2014).

PubMed PubMed Central article CAS Google Scholar 

Paragraph, C. etc. The decrease in intrinsic DNA curvature leads to an increase in mutation rate. Genomic biology. 19, 1-12 (2018).

De Boer, JG & Ripley, LS demonstrates the generation of frameshift and base substitution mutations through quasi-palindromic DNA sequences. Process nails. Akkad. science. United States 81 (1984).

Turner, CB, Marshall, CW & Cooper, VS Parallel genetic adaptation across different environments of growth patterns or resource availability. evolution. Wright. 2, 355–367 (2018).

PubMed PubMed Central article Google Scholar 

Lässig, M., Mustonen, V. and Walczak, AM predicts evolution. Nat. Ecology. evolution. 1, 1-9 (2017).

Hermisson, J. & Pennings, PS Soft Scan: Molecular population genetics adapted from standing genetic variation. Genetics 169, 2335–2352 (2005).

CAS PubMed PubMed Central Article Google Scholar 

Barrett, RDH, M'Gonigle, LK & Otto, SP Distribution of beneficial mutant effects under strong selection. Genetics 174, 2071–2079 (2006).

CAS PubMed PubMed Central Article Google Scholar 

Jerison, ER & Desai, MM Genome research on evolutionary dynamics and epistasis in microbial evolution experiments. Curry opinions. Gene. Development. 35, 33–39 (2015).

CAS PubMed PubMed Central Article Google Scholar 

Taylor, TB, etc. Reconnect the nitrogen regulation system to revive the evolution of flagellar movement. Science 347, 1014–1017 (2015).

ADS CAS PubMed article PubMed Central Google Scholar 

McGee, LW etc. Synergistic pleiotropicity transcends the complexity cost of virus adaptation. Genetics 202, 285–295 (2016).

CAS PubMed article PubMed Central Google Scholar 

McGrath, PT, etc. The parallel evolution of the domesticated Caenorhabditis species targeted the pheromone receptor gene. Nature 477, 321–325 (2011).

ADS CAS PubMed PubMed Central Article Google Scholar 

Sackman, AM etc. Parallel evolution driven by mutation during virus adaptation. Mole. biology. evolution. 34, 3243–3253 (2017).

CAS PubMed PubMed Central Article Google Scholar 

In addition, AS et al. The biosurfactant viscose produced by Pseudomonas fluorescens SBW25 helps spread movement and promote plant growth. environment. microorganism. 16, 2267–2281 (2014).

CAS PubMed article PubMed Central Google Scholar 

Lind, PA, Libby, E., Herzog, J. & Rainey, PB predicting mutation pathways for new adaptive phenotypes. Elife 8, e38822 (2019).

PubMed PubMed Central article Google Scholar 

Wright, BE, Reschke, DK, Schmidt, KH, Reimers, JM & Knight, W. Predicting the frequency of mutations in the stem-loop structure of disinhibitory genes: impact on evolution. Mole. microorganism. 48, 429–441 (2003).

CAS PubMed article Google Scholar 

Kudla, G., Murray, AW, Tollervey, D. & Plotkin, JB Coding-sequence determinants of gene expression in Escherichia coli. Science 324, 255–258 (2009).

ADS CAS PubMed PubMed Central Article Google Scholar 

Kristofich, J. et al. When growth is restricted by weak ligases, synonymous mutations can make a huge contribution to health. PLOS gene. 14. e1007615 (2018).

PubMed PubMed Central article CAS Google Scholar 

Lebeuf-Taylor, E., McCloskey, N., Bailey, SF, Hinz, A. & Kassen, R. The distribution of fitness effects between synonymous mutations in selected genes. Elife e45952 (2019) https://doi.org/10.1101/553610.

Frumkin, I. etc. The codon usage of highly expressed genes affects the translation efficiency in the proteome range. Process National Academy of Sciences. science. United States 115, E4940–E4949 (2018).

PubMed PubMed Central article Google Scholar 

Fieldhouse, D. & Golding, B. The source of small repetitive sequences in genomic DNA. Genetics 129, 563–572 (1991).

CAS PubMed PubMed Central Article Google Scholar 

Dong, F., Allawi, HT, Anderson, T., Neri, BP & Lyamichev, VI Secondary structure prediction and structure-specific sequence analysis of single-stranded DNA. Nucleic acid research. 29, 3248–3257 (2001).

CAS PubMed PubMed Central Article Google Scholar 

Merrikh, CN and Merrikh, H. Gene inversion enhances the evolutionary ability and virulence of bacteria. Nat. Community. 9, 10 (2018).

Vogwill, T., Kojadinovic, M., Furió, V. & Maclean, RC used comparative experimental evolution of antibiotic resistance to test the role of genetic background in parallel evolution. Mole. biology. evolution. 31, 3314–3323 (2014).

CAS PubMed PubMed Central Article Google Scholar 

Blount, ZD, Barrick, JE, Davidson, CJ & Lenski, RE Genomic analysis of a key innovation in an experimental E. coli population. Nature 489, 513–518 (2012).

ADS CAS PubMed PubMed Central Article Google Scholar 

Spor, A. etc. The fusion of phenotype and genotype is influenced by historical accident in yeast and environment. Evolution (New York). 68, 772–790 (2014).

Orr, HA probability of parallel evolution. Evolution (New York). 59, 216 (2005).

Zagorski, M., Burda, Z. & Waclaw, B. Beyond the hypercube: The evolutionary accessibility of a fitness landscape with a real mutation network. PLoS calculation. biology. 12, 1-18 (2016).

Gillespie, JH Molecular evolution of mutant landscapes. Evolution (New York). 38, 1116 (1984).

Bailey, SF, Guo, Q. & Bataillon, T. Identifying the drivers of parallel evolution: regression model methods. Genomic biology. evolution. 10, 2801–2812 (2018).

CAS PubMed PubMed Central Article Google Scholar 

Kimura, M. Evolutionary rate at the molecular level. Nature 217, 624–626 (1968).

ADS CAS PubMed article Google Scholar 

Lind, PA, Farr, AD & Rainey, PB Experimental evolution reveals the hidden diversity in evolutionary pathways. Elife 4, e07074 (2015).

McDonald, MJ, Gehrig, SM, Meintjes, PL, Zhang, XX, and Rainey, the PB adaptive divergence of the experimental population of Pseudomonas fluorescens. Fourth, genetic constraints guide the evolutionary trajectory in parallel adaptive radiation. Genetics 183, 1041–1053 (2009).

CAS PubMed PubMed Central Article Google Scholar 

Lind, PA & Andersson, DI Genome-wide mutation bias in bacteria. Process National Academy of Sciences. science. United States 105, 17878-17883 (2008).

ADS CAS PubMed PubMed Central Article Google Scholar 

Stoltzfus, A. & McCandlish, DM mutation bias affects parallel adaptation. Mole. biology. evolution. 34, 2163–2172 (2017).

CAS PubMed PubMed Central Article Google Scholar 

Robleto, EA, López-Hernández, I., Silby, MW & Levy, SB Genetic analysis of the AdnA regulator in Pseudomonas fluorescens: the unnecessary role of flagella in attachment to sand and biofilm formation. J. Bacteria. 185, 453–460 (2003).

CAS PubMed PubMed Central Article Google Scholar 

Seaton, SC, Silby, MW & Levy, SB The pleiotropic effect of Gaca on Pseudomonas fluorescens pf0-1 in vitro and in soil. Application environment. microorganism. 79, 5405–5410 (2013).

ADS CAS PubMed PubMed Central Article Google Scholar 

Seemann, T. Snippy: Quickly identify bacterial variants from NGS readings. (2015).

Connor, TR, etc. CLIMB (Cloud Infrastructure for Microbial Bioinformatics): An online resource for the medical microbiology community. microorganism. Genomics 2, 6 (2016).

Bryksin, AV & Matsumura, I. Overlap extension PCR cloning: a simple and reliable method to create recombinant plasmids. Biotechnology 48, 463–465 (2010).

CAS PubMed PubMed Central Article Google Scholar 

Hmelo, LR, etc. Precision engineering of the genome of Pseudomonas aeruginosa through two-step allele exchange. Nat. protocol. 10, 1820–1841 (2015).

CAS PubMed PubMed Central Article Google Scholar 

R core team. R: A language and environment for statistical calculations. R Statistical Computing Foundation, Vienna, Austria http://www.r-project.org/. (2014).

Wickham, H. ggplot2: Elegant graphics for data analysis. ISBN 978-3-319-24277-4 (2016).

Ogle, DH, Wheeler, P. & Dinno, A. FSA: Fishery inventory analysis. (2020).

Horton, JS, Flanagan, LM, Jackson, RW, Priest, NK & Taylor, TB A mutation hotspot that determines highly reproducible evolution can be established and destroyed by silent genetic changes. Synchronous sequences evolve in parallel. Open science. frame. (2021) https://doi.org/10.17605/OSF.IO/VUYWP.

Horton, JS, Flanagan, LM, Jackson, RW, Priest, NK & Taylor, TB A mutation hotspot that determines highly reproducible evolution can be established and destroyed by silent genetic changes. GitHub (2021) https://doi.org/10.5281/zenodo.5109984.

We thank Lawrence Hirst for his comments on an earlier version of this manuscript. In addition, we thank Matthew Shepherd, a member of Taylor Labs, for his insightful comments and discussions, and Mark Silby for his contribution to the ancestral Pseudomonas fluorescens Pf0-2x strain used in the study. This work was supported by the University of Bath Research Student Account (URSA) awarded to TBT and NKP; the Royal Society Dorothy Hodgkin Research Fellowship (DH150169) awarded to TBT; and the JABBS Foundation of RWJ. The bioinformatics analysis of the paper was performed by the Microbe Bioinformatics Cloud Infrastructure (CLIMB) of the Medical Research Council (MRC) and Illumina whole-genome sequencing by the Milner Genomics Center in Bath, UK and MicrobesNG in Birmingham, UK.

These authors jointly supervise: Nicholas K. Priest, Tiffany B. Taylor.

Milner Center for Evolution, Department of Biology and Biochemistry, University of Bath, Claverton Down, Bath, BA2 7AY, UK

James S. Horton, Louise M. Flanagan, Nicholas K. Priest and Tiffany B. Taylor

School of Biological Sciences and Birmingham Forest Research Institute (BIFor), University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

You can also search for this author in PubMed Google Scholar

JH and LF contributed to data collection and analysis. JH, RJ and TT contributed to the project conception and research design. JH wrote the manuscript. JH, RJ, NP and TT revised the manuscript.

Correspondence with James S. Horton or Tiffany B. Taylor.

The author declares no competing interests.

Peer review information Nature Communications thanks Rees Kassen, Philip Ruelens and another anonymous reviewer for their contributions to the peer review of this work. Peer review reports are available.

The publisher states that Springer Nature remains neutral on the jurisdiction claims of published maps and agency affiliates.

Open Access This article has been licensed under the Creative Commons Attribution 4.0 International License Agreement, which permits use, sharing, adaptation, distribution and reproduction in any media or format, as long as you appropriately indicate the original author and source, and provide a link to the Creative Commons license , And indicate whether any changes have been made. The images or other third-party materials in this article are included in the article’s Creative Commons license, unless otherwise stated in the material’s credit line. If the article’s Creative Commons license does not include the material, and your intended use is not permitted by laws and regulations or exceeds the permitted use, you need to obtain permission directly from the copyright owner. To view a copy of this license, please visit http://creativecommons.org/licenses/by/4.0/.

Holden, JS, Flanagan, LM, Jackson, RW etc. A mutation hotspot that determines highly reproducible evolution can be established and destroyed by silent genetic changes. Nat Commun 12, 6092 (2021). https://doi.org/10.1038/s41467-021-26286-9

DOI: https://doi.org/10.1038/s41467-021-26286-9

Anyone you share the following link with can read this content:

Sorry, there is currently no shareable link in this article.

Provided by Springer Nature SharedIt content sharing program

By submitting a comment, you agree to abide by our terms and community guidelines. If you find content that is abusive or does not comply with our terms or guidelines, please mark it as inappropriate.

Nature Communications (Nat Commun) ISSN 2041-1723 (online)