Genomic evolution of antibiotic resistance is contingent on genetic background following a long-term experiment with Escherichia coli

Antibiotic resistance is a growing health concern. Efforts to control resistance would benefit from an improved ability to forecast when and how it will evolve. Epistatic interactions between mutations can promote divergent evolutionary trajectories, which complicates our ability to predict evolution. We recently showed that differences between genetic backgrounds can lead to idiosyncratic responses in the evolvability of phenotypic resistance, even among closely related Escherichia coli strains. In this study, we examined whether a strain’s genetic background also influences the genotypic evolution of resistance. Do lineages founded by different genotypes take parallel or divergent mutational paths to achieve their evolved resistance states? We addressed this question by sequencing the complete genomes of antibiotic-resistant clones that evolved from several different genetic starting points during our earlier experiments. We first validated our statistical approach by quantifying the specificity of genomic evolution with respect to antibiotic treatment. As expected, mutations in particular genes were strongly associated with each drug. Then, we determined that replicate lines evolved from the same founding genotypes had more parallel mutations at the gene level than lines evolved from different founding genotypes, although these effects were more subtle than those showing antibiotic specificity. Taken together with our previous work, we conclude that historical contingency can alter both genotypic and phenotypic pathways to antibiotic resistance. Significance A fundamental question in evolution is the repeatability of adaptation. Will independently evolving populations respond similarly when facing the same environmental challenge? This question also has important public-health implications related to the growing problem of antibiotic resistance. For example, efforts to control resistance might benefit from accurately predicting mutational paths to resistance. However, this goal is complicated when a lineage’s prior history alters its subsequent evolution. We recently found that differences between genetic backgrounds can lead to unpredictable responses in phenotypic resistance. Here, we report that genetic background can similarly alter genotypic paths to resistance. This historical contingency underscores the importance of accounting for stochasticity, in the past as well as at present, when designing evolutionarily informed treatment strategies.


24
Antibiotic resistance is a growing health concern. Efforts to control resistance would benefit from 25 an improved ability to forecast when and how it will evolve. Epistatic interactions between 26 mutations can promote divergent evolutionary trajectories, which complicates our ability to predict 27 evolution. We recently showed that differences between genetic backgrounds can lead to 28 idiosyncratic responses in the evolvability of phenotypic resistance, even among closely related 29 Escherichia coli strains. In this study, we examined whether a strain's genetic background also 30 influences the genotypic evolution of resistance. Do lineages founded by different genotypes take 31 parallel or divergent mutational paths to achieve their evolved resistance states? We addressed this 32 question by sequencing the complete genomes of antibiotic-resistant clones that evolved from 33 several different genetic starting points during our earlier experiments. We first validated our 34 statistical approach by quantifying the specificity of genomic evolution with respect to antibiotic 35 treatment. As expected, mutations in particular genes were strongly associated with each drug. 36 Then, we determined that replicate lines evolved from the same founding genotypes had more 37 parallel mutations at the gene level than lines evolved from different founding genotypes, although 38 these effects were more subtle than those showing antibiotic specificity. Taken together with our 39 previous work, we conclude that historical contingency can alter both genotypic and phenotypic 40 pathways to antibiotic resistance. 41 42 Introduction 55 Convergent evolution is common in nature. The independent emergence of winged flight in insects 56 and mammals, and of camera-like eyes in vertebrates and cephalopod mollusks, are familiar but 57 striking examples of how evolution can drive distantly related lineages to similar phenotypic 58 outcomes (1). For over a century, biologists have sought to understand the processes underlying 59 these patterns and quantify the extent of convergent evolution in the natural world. However, 60 quantifying convergence in nature is difficult for at least two reasons. First, one typically observes 61 only a biased sample of possible outcomes. For example, extinct lineages that evolved different, 62 but ultimately unsuccessful, adaptations usually go undetected, causing one to overestimate the 63 extent of convergence (2, 3). Second, comparative studies generally cannot account for slight 64 subsequent evolution under drug selection. We found that their evolutionary potential was 111 idiosyncratic with respect to their initial genotype, such that resistance was constrained in some 112 backgrounds but not in others, indicating the role of historical contingency in this process (5, 17). 113 In this study, we sequenced the complete genomes of antibiotic-resistant clones that 114 evolved from several different founding strains during our earlier experiments and used this 115 information to examine how genetic background affects the genomic evolution of antibiotic 116 resistance. First, we validated our statistical approach by demonstrating that mutations in particular 117 target genes were associated with each of the four antibiotic treatments. We then showed that 118 evolution was also contingent, albeit more subtly, on differences in genetic background, such that 119 resistant lines that evolved from the same genotype had, on average, slightly more mutational 120 targets in common. These results, taken together with our previous work, indicate that even slight 121 differences in genetic background complicate one's ability to predict phenotypic and genotypic 122 outcomes of antibiotic resistance evolution. 123

125
Genomic Evolution of Strains Evolved under Four Different Antibiotic Treatments. In our 126 previous study (17), we isolated antibiotic-resistant mutants that evolved from five different 127 parental genotypes: the LTEE ancestor and four derived clones isolated from the LTEE at 128 generation 50,000. The experiment was performed over one round of drug selection. Here, we 129 sequenced the complete genomes of 64 of the resistant clones, but we discarded 3 that were 130 identified as cross-contaminants (Materials and Methods) (Table S1). 131 The 61 remaining resistant clones had a total of 76 mutations. Forty-five genomes had a 132 single mutation, 11 others had two mutations, and 3 genomes had three mutations (Fig. 1). Two 133 other clones (both in the tetracycline treatment) had no identifiable mutations; they might have 134 had unstable genetic changes or types of mutations that could not be resolved by our analyses of 135 the short-read sequencing data (see Materials and Methods). In any case, we have excluded these 136 two clones from the analyses that follow. Twenty-seven of the 76 mutations (35.5%) were single-137 base substitutions; of these, 22 were either nonsynonymous or nonsense mutations that altered the 138 encoded protein's amino-acid sequence, 1 was synonymous, and 4 occurred in alaT, which 139 encodes a tRNA rather than a protein. Five other mutations (6.6%) were in intergenic regions 140 within 150-bp upstream of a gene, which suggests that they affect regulation. 141 at the bottom. The "Other" category represents mutations in a tRNA. Evolved genomes are labeled 147 according to their parental genetic background and replicate. Two tetracycline-selected clones 148 (Ara-5-1 and Ara+5-1) had no identifiable mutations (see Materials and Methods). 149 The largest proportion of mutations in the resistant lines were structural variants. They 150 comprise 44 (57.9%) of the observed changes ( Fig. 1). Eleven of these were IS-element insertions 151 in protein-coding genes; 8 were small insertions and deletions (indels) of less than 50 bp, 13 were 152 large deletions, and 12 were large amplifications. Twelve of the 13 large deletions and 11 of the 153 12 large amplifications were found in lines derived from the generation-50,000 backgrounds (Fig.  154 1). However, 45 of the 61 clones (73.8%) belong to that group, and neither observed distribution 155 deviates significantly from that null expectation (binomial tests, p = 0.1077 and p = 0.1368 for 156 large deletions and large amplifications, respectively). 157 158 Genomic Parallelism at the Functional Level. Antibiotic resistance can arise through mutations 159 that change gene regulation and expression, cell permeability and efflux, and metabolism (26). To 160 determine how drug selection acted on these functions in our experiment, we quantified the extent 161 of genomic parallelism in the resistant lines at the functional level (i.e., sets of genes that share 162 broadly defined functions). We used the curated descriptions of cellular processes in EcoCyc (27)  163 to match each mutated gene to an associated function. We excluded large deletions and 164 amplifications when the affected genes do not share a common function. 165 About 37% of the 57 mutations that fit the criteria for inclusion occurred in regulatory 166 genes, ~26% in metabolic genes, ~21% in genes that encode transporters, and ~11% in genes 167 involved in transcription or translation (Fig. 2). More mutations in some of these functional 168 categories than in others might suggest a pattern of parallel evolution. However, more E. coli genes 169 are involved in some functions than in others, and therefore a random mutation is more likely to 170 occur in those categories that constitute a larger proportion of the genome. To examine whether 171 the observed number of mutations in each category occurred more frequently than expected (28) Regulatory genes accrued mutations about 5 times more often than expected from the 181 Poisson distribution (Fig. 2), and this difference is highly significant (p < 0.0001). Genes involved 182 in transport functions had about 1.6 times more mutations than expected by chance, but this 183 difference was marginally non-significant (p = 0.0723). Genes involved in transcription or 184 translation had about 1.8 times as many mutations as expected, but this difference was also not 185 statistically significant given the small number of mutations in these targets (p = 0.1245). It should 186 be emphasized that this analysis is conservative because it lumps together all of the genes in each 187 functional category. However, mutations in only a subset of these genes are likely to cause 188 resistance. Therefore, the effective mutational target size and the resulting expected number of 189 mutations is presumably much smaller. 190 191 Specificity of Genomic Evolution in the Different Antibiotic Environments. We compared the 192 gene-level similarity of mutations between independent lines that evolved in the same antibiotic 193 treatment and across the four different treatments to evaluate the effect of the selective 194 environment on the genetic paths to increased antibiotic resistance. As described in the Materials 195 and Methods, we computed Dice's coefficient of similarity for each pair of clones using the 71 196 qualifying mutations that could be assigned to a particular gene. The average within-treatment 197 similarity was 0.089 and the average between-treatment similarity was 0.032 (Fig. 3). In other 198 words, two clones that independently evolved under the same antibiotic selection had on average 199 8.9% of their mutated genes in common, whereas those that evolved under different antibiotics 200 shared on average only 3.2% of their mutated genes. A randomization test shows that the 5.7% 201 difference in similarity is highly significant (p < 0.0001). Thus, genomic evolution was 202 demonstrably specific with respect to the antibiotic treatment. for all clone pairs evolved in the same antibiotic (Ss) and in different antibiotics (Sd), respectively. 208 Only the 71 qualifying mutations (see Materials and Methods) were included in the calculations. 209 The weighted averages of Ss and Sd are shown in the grey box. The difference between these two 210 values indicates the extent to which genome evolution was specific to the antibiotic treatment. The 211 resulting p-value was calculated using a randomization test. 212 The similarity analysis does not reveal the specific genes that contribute to the antibiotic-213 treatment specificity. To address this issue, we used Fisher's exact tests to identify genes that had 214 an excess of qualifying mutations in the replicate lines evolved under the four treatments (Fig. 4). 215 We found 5 "signature" genes in which mutations contributed significantly to antibiotic specificity 216 (Table 1, Fig. 4). The alaT gene encodes a tRNA; it was mutated in 4 of the 14 CIP-resistant lines, 217 but in none of the other 44 lines with qualifying mutations (Fig. 4). The ompR gene is part of the 218 two-component system that regulates the production of outer-membrane proteins; it was mutated 219 in 6/14 TET-resistant lines as well as in 4/44 lines that evolved resistance to other drugs. The other 220 gene in this two-component system, envZ, was mutated in 2 of the 10 TET-resistant lines that did 221 not have an ompR mutation. Two genes, ompF and hns, were associated with resistance to 222 ceftriaxone ( Table 1). The former encodes an outer-membrane porin and was mutated in 6/15 223 CRO-resistant lines along with 3/43 other lines; the latter encodes a histone-like global regulator 224 and acquired mutations in 3/15 CRO-resistant lines and 1 of the 43 lines that became resistant to 225 another antibiotic (Fig. 4). Finally, a large deletion was found in 3 of the 15 AMP-resistant lines 226 but not in any of the other 43 lines (Table 1); this deletion affects multiple genes including phoE, 227 which encodes an outer membrane porin (Fig. 4). 228 229 Fig. 4. Identity of mutated genes in antibiotic-resistant lines. A total of 58 lines (labels at left) 231 evolved from 5 different genetic backgrounds in ampicillin, ceftriaxone, ciprofloxacin, or 232 tetracycline environments. Two (TET Ara-5-1, TET Ara+5-1) had no identifiable mutations; a 233 third (AMP Ara+4-2) had no qualifying mutation that could be assigned to a specific gene (see 234 Materials and Methods). These three lines are not shown. Filled cells identify the 71 qualifying 235 mutations by the affected genes (shown along the bottom and listed in order of the total number of 236 mutations). The darkly shaded cells identify signature genes, in which mutations are significantly 237 associated with one antibiotic treatment (Table 1). A deletion or amplification spanning a given 238 genomic region is indicated when two gene names are shown. If a gene name is shown in brackets, 239 then only part of that gene is affected. If a gene name is preceded by Δ, then those genes are 240 deleted; otherwise, they are amplified. Part of the ompF gene is deleted in the CRO Ara-5-1 line. 241 antibiotic treatments and gene targets. 244

Specificity of Genomic Evolution with Respect to Genetic Background. We next employed 245
Dice's coefficient of pairwise similarity to quantify the specificity of genomic evolution with 246 respect to the parental strain's genetic background. Using the ampicillin treatment as an example 247 ( Fig. 5A), the clones that evolved independently from the same founding genetic background and 248 from different backgrounds had, on average, 14.8% and 3.1% of their mutated genes in common, 249 respectively, indicating a difference of 11.7%. This trend of greater similarity for clones derived 250 from the same genetic background also occurred in the three other antibiotics (Fig. 5B-D). 251 tetracycline (D). The difference between Ss and Sd indicates the extent to which genome evolution 257 was specific to the genetic background. Two of the three replicates derived from the Ara+4 258 background in ciprofloxacin were excluded owing to cross-contamination, and Ss cannot be 259 calculated in that case (*). See Fig. 3 for additional details. 260 We performed separate randomization tests for the clones in each antibiotic treatment to 261 evaluate whether the effects of genetic background were significant. The associations between 262 genomic evolution and the identity of the parental strain were significant for the lines that evolved 263 in the ampicillin and ceftriaxone environments ( Fig. 5A and 5B), and they were marginally non-264 significant for the lines in the ciprofloxacin and tetracycline environments ( Fig. 5C and 5D). When 265 we combined the probabilities from these four independent tests of the hypothesis that differences 266 in genetic background influence the genetic basis of antibiotic resistance using Fisher's method 267 (29,30), the overall trend toward greater similarity (gene-level parallelism) of lines evolved from 268 the same founding genotype was highly significant (χ 2 = 24.67, df = 8, p = 0.0018). 269 In our analysis of genome specificity with respect to antibiotic treatment, we identified 270 several signature genes that contributed significantly to that specificity (Table 1, Fig. 4). We have 271 much less statistical power to identify particular genes that contribute to specificity with respect 272 to genetic background, because only 3 or 4 replicate lines derive from any given background in 273 each antibiotic treatment. Nonetheless, we can identify candidate loci that may contribute to that 274 specificity, which might be further studied in the future. Table 2 shows all of the genes that fulfilled 275 both of the following criteria for a given antibiotic treatment: (i) two or more lines derived from 276 the same background had mutations affecting the same gene; and (ii) that background produced at 277 least as many mutations affecting that gene as did the other four backgrounds combined. In the 278 case of each antibiotic treatment, at least two genetic backgrounds have candidate signature genes 279 that fulfill these criteria. 280 281

284
How does genetic background affect the evolution of antibiotic resistance? We previously 285 addressed this question by examining the resistance potential of the E. coli ancestor of a long-term 286 experiment and derived clones isolated from four populations after generation 50,000. We 287 challenged these strains using a series of drug concentrations, and we found that several strains 288 had a reduced capacity to evolve resistance relative to their ancestor, implicating the role of 289 historical contingency in this process (5,17,31). In this study, we asked whether genetic 290 background also influences the genomic basis of resistance by channeling evolution along different 291 mutational paths. We sequenced the complete genomes of 61 resistant lineages that evolved in our 292 earlier experiment to identify the mutations that conferred resistance. We then analyzed whether 293 there were particular signatures of (i) the antibiotic treatment and (ii) the initial background evident 294 from the identities of the mutated genes. 295 The populations that evolved resistance to the four different drugs in our study exhibited 296 divergent underlying genetic changes (Fig. 3). This result was expected given that bacteria 297 generally evolve resistance through mutations specific to a drug's mechanism of action (12, 15, 298 21). This specificity was driven by parallel mutations in several genes (Table 1). Overall, ompR 299 and ompF had more mutations than any other genes (Fig. 4). OmpR is a transcriptional regulator 300 involved in responses to osmotic and acid stress; mutations to OmpR also contribute to antibiotic 301 resistance by altering the expression of the OmpF major porin (32-34). A recent study showed that 302 ompF deletions reduce the permeability of β-lactams (e.g., ampicillin and ceftriaxone) across the 303 outer membrane, thus increasing resistance (34). We found that ompF mutations were strongly 304 associated with ceftriaxone-resistant lines, consistent with this prior study. However, the evolution 305 of ampicillin resistance occurred through more diverse mutational paths in our experiment. 306 Although there were some mutations in ompF and ompR in the ampicillin treatment, we saw a 307 significant association of that treatment with deletion of a different outer-membrane porin, PhoE. 308 The down-regulation of this porin also partially modulates the cell's response to osmotic stress 309 (35). In the tetracycline treatment, mutations were more common in ompR than in ompF, which 310 suggests that altering the expression of other genes in this regulon also contributes to resistance. 311 Mutations in hns were associated with ceftriaxone-resistant lines. Nishino and Yamaguchi (36) 312 showed that deletion of this global transcriptional regulator increases resistance to multiple drugs 313 because it causes overexpression of several efflux pumps. In contrast to the results for the other 314 three antibiotics in our study, the evolution of ciprofloxacin resistance was not associated with 315 mutations in genes related to outer-membrane proteins. Instead, mutations in alaT, which encodes 316 an alanine tRNA, were a signature of this treatment. The mechanism behind this resistance is 317 unknown. One possibility is that these mutations modulate interactions that have been reported 318 between this tRNA and tmRNA, which rescues stalled ribosomes from aberrant translational 319 events (37). Enhanced rescue might directly promote survival or indirectly affect the expression 320 of other vital genes, when cells are treated with this antibiotic. 321 The signature genes that we observed for each treatment are not the canonical resistance 322 genes for the respective antibiotics (26). Ampicillin and ceftriaxone irreversibly bind to 323 transpeptidases and disrupt cell-wall synthesis; ciprofloxacin targets topoisomerase and inhibits 324 DNA replication; and tetracycline targets the ribosome and hinders protein synthesis. Drug 325 resistance often arises through modifications to these targets, yet these changes rarely occurred in 326 our study. This discrepancy may reflect two factors, one environmental and the other genetic. First, 327 altering a drug target often confers high-level resistance, but at the expense of bacterial growth 328 rate (10, 38). We used moderate drug concentrations to select for mutants (17), and the observed 329 resistance rarely reached levels defined as clinically relevant (39). This moderate environment 330 should favor mutations that provide sufficient resistance at a low fitness cost, because they will 331 leave more descendants during population growth before treatment, and consequently they will be 332 seen more often after the antibiotic challenge. Second, the E. coli used in our experiments are all 333 derived from a B strain that differs in important ways from the K-12 strains that are more widely 334 used in studies of antibiotic resistance (40). In particular, E. coli K-12 has two major porins, OmpC 335 and OmpF, whereas E. coli B expresses only OmpF (41, 42). Thus, the use of the E. coli B strain 336 background may well have influenced which genes could mutate to yield resistance in our 337 experiments. 338 We also found genomic signatures of adaptive divergence associated with differences in 339 genetic background, and these differences are far smaller than those between E. coli B and K-12. 340 We sequenced and analyzed resistant lines that evolved from five backgrounds that were separated 341 in time by only a few decades, and which differed only in the mutations that had accumulated in 342 the antibiotic-free environment of the LTEE (Fig. 1). Three or four resistant lines independently 343 evolved from each parental background for each of the four antibiotics studied, allowing us to 344 assess the genomic specificity of resistance with respect to the genetic background (Fig. 5). 345 Although these background effects were more subtle than those showing antibiotic specificity, 346 they are compelling when taken together. Various factors might contribute to the genetic 347 background specificity. Most broadly, epistatic interactions can cause the same mutation to have 348 different effects on resistance, or on its fitness costs, in different backgrounds. The rates at which 349 particular resistance mutations arise may also vary between different genetic backgrounds. 350 Imagine that the same mutation arises in separate populations founded from two distinct 351 backgrounds. If the mutation confers less resistance in one background than the other, then it may 352 go undetected when those populations are challenged at a high drug concentration. This type of 353 epistasis could therefore generate a signature of genomic specificity of resistance mutations with 354 respect to the genetic background. It is also possible that different genetic backgrounds affect the 355 evolution of resistance by changing the likelihood of certain genomic amplifications or deletions. 356 These types of structural mutations often occur by homologous recombination between IS 357 elements, and they can confer resistance by altering the number of membrane transporters or drug 358 targets (43). Such mutations can also occur spontaneously at very high rates in comparison to point 359 mutations (43,44). In our study, many of the resistance mutations were mediated by IS elements, 360 including new copies in the derived backgrounds that previously arose during the LTEE (24). The 361 evolution of resistance in these cases is therefore influenced, at least in part, by changes in the rates 362 at which certain types of mutations arise in the derived genetic backgrounds. 363 Antibiotic resistance is a growing public-health concern. If the most likely evolutionary 364 paths to resistance can be accurately predicted, then there exists a potential opportunity to control 365 the emergence of resistance through rational treatment strategies. However, to predict the 366 evolution of resistance with accuracy, we must understand and integrate information about many 367 factors, including a bacterial lineage's evolutionary history, and how that history may potentiate 368 or constrain its future evolution. The results from this study, together with our previous findings, 369 demonstrate the importance of historical contingency in the evolution of drug resistance at both 370 the phenotypic and genotypic levels. This contingency underscores the importance of accounting 371 for stochasticity, in the past as well as at present, when designing evolutionarily informed 372 treatment strategies. 373 374

375
Evolution Experiments and Bacterial Strains. The LTEE has been described in detail elsewhere 376 (3, 23). Briefly, 12 replicate populations of E. coli were founded from a common ancestral strain 377 called REL606. These populations have been propagated for more than 70,000 bacterial 378 generations by daily 1:100 transfers in a glucose-limited Davis minimal (DM) medium without 379 antibiotics. 380 In a previous study (17), we measured the intrinsic resistance of the LTEE ancestor and 381 derived clones isolated from four populations (designated Ara-5, Ara-6, Ara+4, and Ara+5) at 382 generation 50,000 to the antibiotics ampicillin (AMP), ceftriaxone (CRO), ciprofloxacin (CIP), 383 and tetracycline (TET). We also quantified these strains' capacities for evolving resistance by 384 challenging them across a range of concentrations to these same drugs during one round of 385 selection. In this study, we sequenced the complete genomes of a subset of the resistant mutants 386 that evolved during these experiments, and we examined whether the genetic targets of the 387 resistance mutations systematically differed between the four antibiotics and five genetic 388 backgrounds. Specifically, for each antibiotic treatment we sequenced 4 mutants that 389 independently evolved from the ancestral background, and 3 mutants from each derived 390 background, for a grand total of 64 sequenced mutants (16 mutants × 4 antibiotics) (Table S1) Each population evolved unique substitutions in pykF during the LTEE that distinguish 410 them from one another (24). Therefore, we first compared this locus for each resistant clone against 411 its corresponding parental strain to test for possible external and cross-contamination. Strains 412 KJC184 and KJC217 from the CIP and CRO treatments, respectively, were supposed to derive 413 from the Ara+4 and Ara+5 parental backgrounds, respectively, but they had pykF alleles 414 corresponding to other backgrounds used in this study. Also, strain KJC152 from the CIP treatment 415 was supposed to derive from the Ara+4 background, but its genome was identical to a resistant 416 mutant derived from the ancestral clone. We discarded these three cross-contaminants from our 417 study. 418 The breseq results for each of the other 61 sequenced resistant clones gave information on 419 both its genetic background and the mutations that evolved during our previous antibiotic-selection 420 experiments. We manually curated the results by removing the background-specific mutations 421 (i.e., those that arose during the LTEE), which we did by comparing each resistant clone to its 422 parental strain. We also excluded expansions and contractions of hypermutable short sequence 423 repeats that are unlikely to contribute to stably inherited resistance, and mutations within multi-424 copy elements (e.g., ribosomal RNA operons and insertion sequences) that may result from gene 425 conversions but cannot be fully resolved using short-read sequencing data. In addition, we resolved 426 numerous structural variants by manually examining the depth of read coverage across the genome 427 and predictions of new sequence junctions from split-read mapping for each clone (47). To verify 428 the predicted mutations, we applied the genomic changes in each parental background to the 429 REL606 reference genome and reran breseq. 430 In total, we identified mutations in 59 of the 61 antibiotic-resistant clones. Two clones 431 (KJC65 and KJC66) had no clear genetic changes despite exhibiting increased phenotypic 432 resistance in our earlier study (17). This discrepancy suggests at least two possible explanations. 433 First, these resistant lines might have mutations that could not be adequately resolved by short-434 read sequencing, including amplifications of genes or chromosomal regions and inversions 435 bounded by identical sequences (e.g., multi-copy IS elements) (49). Second, these lines might have 436 had unstable genetic changes, including copy number changes in homopolymeric tracts and gene 437 amplifications, which are often unstable and can lead to hypermutability, phase variation, and other 438 complications (43,(50)(51)(52). To look for amplifications that might have been missed by the breseq 439 pipeline, we also used another pipeline (53) to examine the two genomes without identifiable 440 mutations for evidence of regions with above-average read coverage. However, this analysis did 441 not reveal any amplifications in these two clones. Additional details can be found in the R 442 within 150-bp upstream of the start of a gene. However, we modified their approach to also include 452 large deletions and amplifications if at least one of the affected genes was also found to be mutated 453 in another clone or if there were parallel changes across lines. We excluded from these analyses 454 synonymous mutations, the two clones with no identified genetic changes, and a third clone with 455 only a large amplification that was unique and could not be assigned to any particular gene. A total 456 of 71 mutations qualified based on these criteria. 457 We then calculated Dice's coefficient of similarity, S, for each pair of evolved clones, where 458 = 2| ∩ |/(| | + | |). Here, |X| and |Y| represent the number of genes with qualifying 459 mutations in each clone, and | ∩ | is the number of mutated genes in common between them. S 460 therefore ranges from 0, when the pair of clones have no mutated genes in common, to 1, when 461 both have mutations in exactly the same set of genes (30,53,54). Finally, we calculated the 462 average of these coefficients for all pairs of clones evolved within the same treatment or from the 463 same parental genotype, Ss, and for all pairs of clones evolved across different treatments or 464 different genotypes, Sd. The difference between these two values serves as a test statistic for the 465 specificity of genomic evolution. 466 The observed outcome can be seen as one of many possible but equally likely outcomes 467 that could have arisen by chance. One can therefore perform a randomization test to evaluate the 468 significance of the test statistic associated with the observed outcome (30). To do so, we repeatedly 469 rearranged the clones associated with each antibiotic treatment, or the clones within each treatment 470 when testing for background specificity, while maintaining the number and identity of the 471 mutations in any clone (54). For example, if mutations A and B were found together in the same 472 sequenced clone, we retained their association throughout the procedure but randomly assigned 473 the set to a different clone label. We calculated the specificity test statistic for each of 10,000 474 permutations of the clone labels. This procedure yields the expected distribution of the test statistic 475 under the null hypothesis that the similarity among lines is independent of the antibiotic treatment 476 or founding genotype. We then calculated an approximate p-value for rejecting this null hypothesis 477 from the proportion of permutations in the expected distribution with a specificity statistic value 478 greater than or equal to the observed value. 479 To quantify the specificity of genomic evolution with respect to genetic background, we 480 performed an independent randomization test for each of the four antibiotics. Because these tests 481 address the same null hypothesis, we combined the resulting p-values using Fisher's method with 482 2k = 8 degrees of freedom, where k is the number of comparisons (29, 30). We provide the datasets 483 and details of our statistical analyses in an R Notebook on GitHub 484