Over time bacteria can undergo a number of genetic mutations that allow them to evolve in response to changes in their surrounding environment. This process of ‘bacterial evolution’ is one of the major causes of antibiotics resistance, whereby disease-causing microorganisms become resistant to multiple drugs and can no longer be destroyed using antibiotic treatment. However, when bacteria become resistant to a drug this can result in an evolutionary trade-off known as ‘collateral sensitivity’ – when evolving resistance to one drug causes bacteria to gain increased sensitivity to another.

Now, Barbosa, Roemhild et al. have investigated whether this evolutionary trade-off could be exploited to tackle the antibiotic crisis and prevent bacteria adapting to different treatments. If this evolutionary trade-off is to be used medically, it must be stable long enough for the bacteria population to either become extinct, or less able to evolve multi-drug resistance. To test how stable collateral sensitivity is over time, Barbosa, Roemhild et al. studied the bacterium Pseudomonas aeruginosa which is known to evolve collateral sensitivity to certain drug treatments.

P. aeruginosa were subjected to two rounds of evolution: first, bacteria were evolved to resist ‘Drug A’ and at the same time became more sensitive to another drug, ‘Drug B’. The bacteria were then allowed to adapt to Drug B either alone or in the presence of Drug A. These evolutionary experiments revealed that the following factors affected the stability of the trade-off: the molecular structure of the antibiotic bacteria evolved sensitivity to, the strength of the original evolutionary trade-off (i.e. how sensitive bacteria became), the order drugs were administrated, and whether resistance came at a large fitness cost (i.e. when the genetic mutations promoting resistance affect bacteria’s ability to replicate and survive in normal conditions).

According to the World Health Organization, P. aeruginosa is the second most problematic multi-drug resistant bacteria. The data collected in this study could therefore be used to develop a new antibiotic therapeutic strategy for fighting this bacterium, as well as other microbes which are resistant to multiple drugs.

Treatment of infectious diseases and cancer often fail because of the rapid evolution of drug resistance (Bloemberg et al., 2015; Davies and Davies, 2010; Gottesman, 2002; Zaretsky et al., 2016). Optimal therapy should thus anticipate how resistance to treatment evolves and exploit this knowledge to improve therapy (Gatenby et al., 2009; Imamovic and Sommer, 2013). One promising strategy is based on evolved collateral sensitivity: the evolution of resistance against one drug A concomitantly causes hypersensitivity (i.e., collateral sensitivity) to a second drug B (Szybalski and Bryson, 1952). If evolved collateral sensitivity is reciprocal, it can – in theory – trap the bacteria in a double bind, thereby preventing the emergence of multidrug resistance during treatment (Baym et al., 2016; Pál et al., 2015; Roemhild and Schulenburg, 2019). Recent large-scale studies have demonstrated that evolved collateral sensitivity is pervasive in laboratory strains and clinical isolates of distinct bacterial species (Barbosa et al., 2017; Imamovic et al., 2018; Imamovic and Sommer, 2013; Jansen et al., 2016; Jiao et al., 2016; Lázár et al., 2014; Lázár et al., 2013; Oz et al., 2014; Podnecky et al., 2018) as well as cancer cells (Dhawan et al., 2017; Pluchino et al., 2012; Shaw et al., 2016; Zhao et al., 2016). More importantly, evolved collateral sensitivity can slow down resistance evolution during combination (Barbosa et al., 2018; Rodriguez de Evgrafov et al., 2015; Munck et al., 2014) and sequential therapy (Kim et al., 2014; Roemhild et al., 2015), and also limit the spread of plasmid-borne resistance genes (Rosenkilde et al., 2019).

Although collateral sensitivity appears to be pervasive, its utility for medical application is still dependent on several additional factors. Firstly, the evolution of collateral sensitivity should ideally be repeatable for a given set of conditions (Nichol et al., 2019). This means that independent populations selected with the same drug should produce identical collateral effects when exposed to a second one. Such high repeatability is not always observed. Recent work even identified evolution of contrasting collateral effects (i.e., some populations with evolved collateral sensitivity and others with cross-resistance) for different bacteria, including Pseudomonas aeruginosa (Barbosa et al., 2017), Escherichia coli (Nichol et al., 2019; Oz et al., 2014), Enterococcus faecalis (Maltas and Wood, 2019), and a BCR-ABL leukemeia cell line (Zhao et al., 2016). These patterns are likely due to the stochastic nature of mutations combined with alternative evolutionary paths to resistance against the first selective drug, subsequently causing distinct collateral effects against other drugs (Barbosa et al., 2017; Nichol et al., 2019). Secondly, the evolution of collateral sensitivity should ideally be repeatable across conditions, for example different population sizes. This is not always the case. For example, an antibiotic pair, which consistently produced collateral sensitivity in small Staphylococcus aureus populations (e.g., 10⁶), instead produced complete cross-resistance in large populations (e.g., 10⁹) and thus an escape from the evolutionary constraint, most likely due to the higher likelihood of advantageous rare mutations under these conditions (Jiao et al., 2016).

A third and largely unexplored factor is that evolved collateral sensitivity and, hence, the resistance trade-off should be stable across time. This implies that bacteria either cannot evolve to overcome collateral sensitivity and thus die out, or, if they achieve resistance to the new drug B, they should concomitantly be re-sensitized to the original drug A. Two recent studies, both with different main research objectives, yielded some insight into this topic. One example was focused on historical contingency during antibiotic resistance evolution of P. aeruginosa (Yen and Papin, 2017). As part of the results, the authors identified lineages with evolved resistance against ciprofloxacin that simultaneously showed increased sensitivity to piperacillin and tobramycin. The reverse pattern (i.e., evolved high resistance to either piperacillin or tobramycin and increased sensitivity to ciprofloxacin) was not observed and, thus, this case represents an example of uni-directional collateral sensitivity. The subsequent exposure of the ciprofloxacin-resistance lineages to either piperacillin or tobramycin led to the evolution of resistance against these two antibiotics and substantial (yet not complete) re-sensitization to ciprofloxacin. The second study focused on evolving E. coli populations in a morbidostat, in which the bacteria were exposed to repeated switches between two drugs (Yoshida et al., 2017). The evolution of multidrug resistance was only prevented in the two treatments with polymyxin B that were also characterized by evolved collateral sensitivity, although again only in one direction (Yoshida et al., 2017). To date, the general relevance of this third factor is still unclear, especially for conditions when collateral sensitivity is reciprocal and when the evolving populations are also allowed to go extinct (i.e., they cannot overcome the evolutionary trade-off).

Here, we specifically tested the potential of the model pathogen P. aeruginosa to escape reciprocal collateral sensitivity through de-novo evolution. We focused on the first switch between two drugs, because the evolutionary dynamics after this first switch will reveal the ability of the bacteria to adapt to the second drug, against which they evolved sensitivity, and, if so, whether this causes re-sensitization to the first drug. These two aspects are key criteria for applicability of a treatment strategy that exploits evolved collateral sensitivities. Our analysis is based on a two-step evolution experiment. Bacteria first evolved resistance against a first drug A and concomitant sensitivity against a second drug B. Thereafter, bacteria were subjected to a second evolution experiment, during which they were allowed to adapt to the second drug B, either alone or additionally in the presence of the first drug A. Phenotypic characterization of the evolved bacteria was combined with genomic and functional genetic analyses, in order to determine the exact targets of selection under these conditions. We finally validated the findings made by performing further, independent sets of similar two-step evolution experiments.

The experimental design of our two-step evolution experiment is illustrated in Figure 1a. We took advantage of previously evolved, highly resistant P. aeruginosa populations, which we obtained from serial passage experiments with increasing concentrations of clinically relevant bactericidal antibiotics (drug A, Figure 1a) (Barbosa et al., 2017). From these, we identified two cases of reciprocal collateral sensitivity, including (i) piperacillin/tazobactam (PIT) and streptomycin (STR), or (ii) carbenicillin (CAR) and gentamicin (GEN). In the current study, we now re-assessed the reciprocity of collateral effects using dose-response analysis (Figure 1b and c, Supplementary file 1–Figure 1–supplementary table 1, Figure 1—source data 1). Thereafter, we isolated resistant colonies from these populations and switched treatment to the drug, against which collateral sensitivity was observed (drug B, Figure 1a). By starting this treatment step with clonal lineages, we here tested the ability of P. aeruginosa to overcome collateral sensitivity by de-novo mutation. The evolutionary challenge was initiated at sub-inhibitory concentrations of each drug (vertical black dashed lines in Figure 1b and c, Supplementary file 1–Figure 1–supplementary table 1, Figure 1—source data 1), followed by linear concentration increases at two different rates: mild or strong (vertical orange and blue dashed lines respectively, Figure 1b and c, Supplementary file 1–Figure 1–supplementary table 1, Figure 1—source data 1). We specifically chose linear increases, because our main objective was to better understand the evolutionary dynamics occurring during the first switch of a collateral sensitivity cycling strategy. Linear increases would, in this case, facilitate evolutionary rescue and provide ample opportunity to escape the double bind, thereby yielding a conservative measure for the applicability of collateral sensitivity as a treatment strategy. We additionally considered treatments where antibiotics were also switched to collateral sensitivity, but selection by drug A was continued in combination with drug B; hereby denoted as constrained environments. Overall, four selective conditions were run in parallel: mild or strong increases of the second drug B in either the presence (constrained) or absence (unconstrained) of the first drug A (Figure 1a). We further included control experiments without antibiotics. To determine treatment success, we monitored bacterial growth with continuous absorbance measurements, quantified frequencies of population extinction, and characterized changes in antibiotic resistance of the evolved bacteria as previously evaluated for P. aeruginosa and other bacteria (Barbosa et al., 2018; Barbosa et al., 2017; Hegreness et al., 2008).

Figure 1

Download asset Open asset

Figure 1—source data 1

Mean optical density and CI95 values obtained after 12 hr of growth in minimal media and different antibiotics as reported in Figure 1.

The populations tested here include the PA14 wild type, and four resistant populations. Each value is the average of 8 technical replicates per bacterial population.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig1-data1-v2.txt

Download elife-51481-fig1-data1-v2.txt

Extinction rates were high even under mild selection regimes

The imposed antibiotic selection frequently caused population extinction (Figure 2, Figure 2—source data 1), even though sublethal drug concentrations were used. Extinction events occurred significantly more often when selection for the original resistance was maintained by the presence of both drugs (extinction in constrained treatments vs. only one drug, χ² = 12.9, df = 1, p<0.0001; Figure 2). In treatments with only the second drug B, extinction occurred significantly more often under strong, but not mild concentration increases (strong vs. mild increases in unconstrained environments, χ² = 5.5, df = 1, p=0.019). Drug switches with the antibiotic pair STR/PIT was particularly successful, with 33 extinction events (~51%, Figure 2). The results differed for the CAR/GEN pair, which produced only seven extinctions (~10%), all restricted to one drug order, GEN>CAR, suggesting asymmetry in the ability to counter collateral sensitivity. The observed differences in the extinction levels of the pairs considered here are unlikely to be the result of differences in combined synergy, since both combinations (PIT/STR and CAR/GEN) are synergistic against P. aeruginosa (Barbosa et al., 2018). From this, we conclude that strong genetic constraints against an evolutionary response to collateral sensitivity caused frequent population extinctions for STR/PIT switches, whereas evolutionary rescue was possible for the GEN/CAR pair, although influenced by drug identity and order.

Figure 2

Download asset Open asset

Figure 2—source data 1

Count data of extinction events as reported in Figure 2.

Extinct populations were determined by their inability to grow in rich media after 24 hr of incubation at 37 °C.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig2-data1-v2.txt

Download elife-51481-fig2-data1-v2.txt

Novel resistance evolved rapidly in many of the surviving populations

We subsequently focused our analysis of the evolutionary dynamics on the surviving populations of the CAR/GEN pair and identified rapid adaptive responses, especially when not constrained by the presence of the two drugs (Figure 3). We measured bacterial adaptation using relative biomass (see Materials and methods and Roemhild et al., 2018) and found it to have increased in all surviving populations (Figure 3a and b, Supplementary file 1–Figure 3–supplementary table 1, Figure 3—source data 1). For both drug orders, the increase was significantly slower in the constrained treatments, and, to a lesser extent, for the strong concentration increases (Figure 3a and b). Consistent with the asymmetry in extinction, the CAR>GEN switch (with no extinction) maintained a high relative biomass across time, while the reverse direction GEN>CAR (with extinction) produced lower relative biomass levels. These results indicate that P. aeruginosa can evolve resistance against a drug, to which it had previously shown hypersensitivity, and that such evolutionary rescue is favored for the suboptimal switch. For the STR/PIT pair, we generally obtained similar results (Figure 3—figure supplement 1a and b, Figure 3—source data 1); yet because of the few surviving populations and high variation among these rare survivors, the results remained inconclusive. We thus continued to focus on the evolved populations for the CAR/GEN pair and asked how the new adaptation influenced the original drug resistances.

Figure 3 with 2 supplements see all

Download asset Open asset

Figure 3—source data 1

Evolutionary dynamics summarized by the area under the curve (AUC) across experimental seasons relative to the reference treatment with no drugs.

Data is shown for all following experiments: CAR^R>GEN, GEN^R>CAR, PIT^R>STR and STR^R>PIT, as reported in Figure 3a and b and Figure 3—figure supplement 1a and b.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig3-data1-v2.txt

Download elife-51481-fig3-data1-v2.txt

Figure 3—source data 2

Dose-response curves data of surviving populations challenged with CAR, GEN, STR and PIT.

Data is shown in Figure 3c and d, and Figure 3—figure supplement 1c and d. Optical density values were recorded after 12 hr of incubation at 37 °C.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig3-data2-v2.txt

Download elife-51481-fig3-data2-v2.txt

Figure 3—source data 3

Growth rate estimates of the surviving population challenged with CAR or GEN, as reported in Figure 3.

Growth rate was calculated as specified in the Materials and methods section.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig3-data3-v2.txt

Download elife-51481-fig3-data3-v2.txt

Whole genome sequencing identified possible targets of antibiotic selection

We used population genomic analysis to characterize specific functional changes that were likely targeted by antibiotic selection and allowed populations to survive the second evolution experiment for the CAR/GEN pair (Figure 4, Figure 4—source data 1). In particular, we sequenced whole genomes of the resistant starting clones from the beginning and 21 surviving populations from the end of the second evolution experiment. Our results reveal that the evolution of multidrug resistance in the suboptimal switch CAR>GEN can be explained by the sequential fixation of mutations including, under unconstrained conditions, those in ptsP (Figure 4a, Figure 4—source data 1), a main component of the global regulatory system of ABC transporters and other virulence factors (Feinbaum et al., 2012). Similarly, under constrained conditions, we found mutations in the NADH-dehydrogenase genes nuoD or nuoG (Figure 4a, Figure 4—source data 1), which are known to influence proton motive force and resistance against aminoglycosides upon inactivation (El'Garch et al., 2007).

Figure 4

Download asset Open asset

Figure 4—source data 1

Genetic changes compared to Pseudomonas aeruginosa PA14 wild type strain as determined by whole-genome resequencing.

Sequencing was performed with Illumina MiSeq2 × 150 bp PE, Nextera libraries. The data is shown in Figure 4. Isolates are coded with AA-BB-CC-: AA, antibiotic to which they are originally resistant; BB, antibiotic to which the clone shows collateral sensitivity; CC, well in the plate during experimental evolution.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig4-data1-v2.txt

Download elife-51481-fig4-data1-v2.txt

Figure 4—source data 2

MIC values for several constructed mutants against CAR and GEN, as reported in Figure 4.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig4-data2-v2.txt

Download elife-51481-fig4-data2-v2.txt

For the more effective switch GEN>CAR, multidrug resistance in the constrained treatments coincided with mutations in mexR, phoQ, and cpxS, an independent regulator of MexAB-OprM (Li et al., 2016) and two-component regulators involved in aminoglycoside resistance (Gooderham and Hancock, 2009) and envelope stress response (Roemhild et al., 2018), respectively. The re-sensitization towards the first drug in the unconstrained GEN>CAR treatments was associated with two main types of mutational changes at high frequencies across several replicates, including (i) mutations in nalC and nalD that upregulate the expression of the multidrug-efflux system MexAB-OprM in P. aeruginosa (Li et al., 2016); and (ii) large deletions in pmrB, which is part of a two-component regulatory system (Figure 4b, Figure 4—source data 1). Mutations in nalC were previously shown to mediate both resistance to CAR and hypersensitivity to GEN (Barbosa et al., 2017). Thus, re-sensitization to GEN may be caused by antagonistic pleiotropy of nalC mutations that override the resistance of the original pmrB mutation (Figure 4, Figure 4—source data 1). In addition, there may be epistasis between the two functional modules. A complementary mechanism for re-sensitization against GEN is the re-mutation of pmrB (Figure 4, Figure 4—source data 1). In three cases nalC mutations coincided with mutations in pmrB, including two deletions of 17 and 225 base pairs. Whilst the original SNP in pmrB altered gene function, the latter deletions may have suppressed the expression of the original SNP by pseudogenizing the gene. We conclude that mutations in the nalC or nalD regulators of the MexAB-OprM pump, sometimes in combination with follow up mutations in pmrB are likely to account for the re-sensitization towards the first drug GEN.

Repetition of experimental evolution revealed an influence of drug type on population extinction and drug re-sensitization

To validate and generalize our findings, we repeated evolution experiments for 14 cases of collateral sensitivity, using a total of 38 distinct resistant populations (Figure 5a; Figure 5—source data 1), obtained from independent biological replicates of our previous study (Barbosa et al., 2017). We initiated experiments with the available populations (Barbosa et al., 2017), rather than clones, and applied a less severe bottleneck of 10⁷ cells. Therefore, any adaptative changes may not exclusively rely on de-novo mutations but could also result from available genetic diversity, thus reflecting a more general scenario for evolutionary adaptation to collateral sensitivity. The 14 cases of collateral sensitivity included the same four examples tested above and 10 additional cases, encompassing both reciprocal and also uni-directional collateral sensitivities. The effect size of collateral sensitivity, measured as fold-IC₇₅, differed substantially among replicates (Figure 5a, Figure 5—figure supplement 1; Figure 5—source data 1). In the current study, we now subjected the total of 38 resistant populations to either strong or strong+constrained increases of the second drug (Figure 5a) and assessed treatment outcome by measuring population extinction rates and changes in resistance profiles of the surviving populations (Supplementary file 1–Figure 5-supplementary tables 1-2, Figure 5—figure supplement 1—source data 1).

Figure 5 with 1 supplement see all

Download asset Open asset

Figure 5—source data 1

Effect size of tested collateral sensitivities.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig5-data1-v2.txt

Download elife-51481-fig5-data1-v2.txt

Figure 5—source data 2

Extinction of experimentally evolved populations.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig5-data2-v2.txt

Download elife-51481-fig5-data2-v2.txt

Figure 5—source data 3

Evolved resistance changes of experimentally evolved populations.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig5-data3-v2.txt

Download elife-51481-fig5-data3-v2.txt

Our validation experiment now demonstrated that extinction events were frequent and occurred significantly more often when selection for the original resistance was maintained by the presence of both drugs (extinction in constrained vs. strong environments, χ² = 16.204, df = 1, p<0.0001; Figure 5b; Figure 5—source data 2). Extinction rates showed pronounced variation by drug identity and order. For example, CIP-resistant populations all survived when challenged with either of the aminoglycosides (STR, GEN), whereas more than 50% of these populations went extinct when exposed to β-lactams (CAR, PIT, Figure 5b). We specifically assessed the importance of drug target, collateral sensitivity effect size (fold-IC₇₅), and relative growth rate under drug-free conditions, using a generalized linear model (GLM; Figure 6—source data 1). We extracted data on the relative exponential growth rate of the resistant populations in drug-free media from our previous publication (Barbosa et al., 2017). Our analysis revealed that variation in extinction was significantly associated with only the molecular target of the second antibiotic (GLM, combined extinction, n = 14 drug switches, F = 11.016, p=0.011; Supplementary file 1–Figure 6–supplementary table 1, Figure 6—source data 1), but not with growth rate, extent of collateral sensitivity, or target of the first antibiotic. Extinction frequencies were on average twice as high in the six treatments that switched to β-lactam antibiotics that target the cell wall, as compared to the eight treatments that switched to aminoglycoside antibiotics that target the ribosome (Figure 6a). It thus appears that extinctions in our experiments are stochastic events that are mainly influenced by the target of the current antibiotic, but not the preceding evolutionary conditions. Taken together, we conclude that switching from one to two drugs generally increases treatment potency and that this is influenced by drug order and identity. The preferred drug order is to use a β-lactam as the second antibiotic, which is consistent with the detailed dynamics on the clonal level.

Figure 6

Download asset Open asset

Figure 6—source data 1

Summary of total extinction in the generalized experiment in relation to average initial collateral sensitivity effects and average growth rate.

Average collateral sensitivity effects are given as fold change IC₇₅, fold change IC₉₀, and AUC relative to PA14. The data is derived from that shown in Figure 5.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig6-data1-v2.txt

Download elife-51481-fig6-data1-v2.txt

Figure 6—source data 2

Summary of average resistance changes in the generalized experiment in relation to average initial collateral sensitivity effects and average growth rate.

Average initial collateral sensitivity effects are given as fold change IC₇₅, fold change IC₉₀, and AUC relative to PA14). The data is derived from that shown in Figure 5.

https://cdn.elifesciences.org/articles/51481/elife-51481-fig6-data2-v2.txt

Download elife-51481-fig6-data2-v2.txt

Resistance gains to drug B were frequently associated with re-sensitization to the original drug A in the unconstrained treatments, while resistance to the new drug B increased in all 14 tested cases (leading to changes of 2 up to >64 times the IC₉₀ of the starting populations; Supplementary file 1–Figure 5–supplementary tables 1 and 2). In detail, we observed significant antibiotic re-sensitization for 5 of the 14 cases (36%, Figure 5b, Figure 5—source data 3, Supplementary file 1–Figure 5–supplementary table 1, Figure 5—source data 1; Note that the single available replicate for STR>PIT showed a similar trend but could not be evaluated statistically). Within reciprocal collateral sensitivity, re-sensitization was not always observed in both directions, as illustrated by CAR/STR and STR/PIT. Re-sensitization did not occur in the constrained treatments. We noticed some differences between evolutionary outcomes of drug switches in clonal and population experiments. The resistance changes showed an overall categorical agreement of 75% (9/12 comparisons; no comparisons for STR>PIT, as this switch was tested with a single replicate). Yet, clonal evolution produced re-sensitization for only one sequential treatment direction for CAR/GEN, whereas population evolution showed re-sensitization for both directions. That re-sensitization occurred in both directions at the population level is expected if there was a residual frequency of cells lacking the collateral-sensitivity mutation in the inoculum. The population data further highlighted a lack of re-sensitization for PIT>STR, where we previously observed re-sensitization on the clonal level, albeit not at both concentration regimes. In addition, there is a mild significant increase of resistance to GEN in the strong constrained treatments in GEN>CAR, compared to no change. Overall our results indicate that re-sensitization is frequent for populations with collateral sensitivity, but not imperative (Figure 5b).

What are the drivers of the observed resistance gains and re-sensitization? Our statistical analysis revealed that the extent of re-sensitization in the unconstrained treatments was significantly associated to equal degrees with the target of drug A (GLM, resistance against drug A, n = 14, F = 15.27, p=0.0019; Supplementary file 1–Figure 6–supplementary table 2, Figure 6—source data 2), the effect size of collateral sensitivity (F = 15.91, p=0.004), and growth rate in drug-free media (F = 18.38, p=0.0027), while target of drug B did not have a significant influence. Re-sensitization was stronger in drug switches that started with an aminoglycoside, as compared to β-lactam, and fluoroquinolone (Figure 6b). This result is consistent with the drug-dependent negative epistasis (Figure 4), and validates the relative stability of collateral sensitivity switches from aminoglycosides to β -lactams. Growth rate in drug-free conditions and collateral sensitivity effect size showed small negative effects, indicating that they decelerate re-sensitization. In contrast, the extent of resistance gains in unconstrained treatments was not associated with drug targets of either drug A or B, but instead significantly associated with both the initial effect size of collateral sensitivity (GLM, resistance against drug B, n = 14, F = 6.71, p=0.0321; Supplementary file 1-Figure 6–supplementary table 2, Figure 6—source data 2), and the initial growth rate of the resistant population in drug-free media (GLM, resistance against drug B, n = 14, F = 7.90, p=0.0228). Collateral sensitivity effect size showed a negative association with the resistance gains to drug B (Figure 6c). High sensitivity thus appears to be prone to more rapid resistance gains, possibly because it opens a fitness space for rapid adaptation. The population resistance changes against drug B were generally larger than the initial collateral sensitivity (Figure 5b; Figure 5—source data 3); Supplementary file 1-Figure 5-supplementary tables 1-2; evolved lineages showed higher resistance than the PA14 wild type), indicating that they were not solely due to competition of pre-existing sub-populations, but involved new genetic adaptations. As observed above, the presence of a general adaptation trade-off (measured as reduction in growth rate relative to ancestral PA14 in drug-free environments) inhibited rapid adaptation (Figure 6d). Taken together, our results at the population level suggest that the stability of collateral sensitivity switching is shaped predominantly by drug targets, but additionally influenced by the effect size of collateral sensitivity and a general adaptation trade-off. The most stable collateral sensitivity was identified for the switch from aminoglycoside to β-lactam. Within a particular drug switch, the highest stability was achieved for low collateral sensitivity effect sizes. Large adaptation trade-offs tended to inhibit evolutionary change, that is both desirable re-sensitization, and new resistance. The conclusions are consistent with the data obtained from the clonal evolution experiments.

Collateral sensitivity is a pervasive feature of resistance evolution, but its potential for medical application is currently debated (Nichol et al., 2019; Podnecky et al., 2018; Roemhild and Schulenburg, 2019). Its promise as a treatment focus is dependent on the premise that the exploited trade-off is evolutionarily stable and cannot be easily overcome. As a consequence, it should either drive bacterial populations to extinction or minimize the emergence of multidrug resistance by re-sensitization to one of the antibiotics. We here tested the validity of these key predictions with the help of evolution experiments and the model pathogen P. aeruginosa. We found that the effective exploitation of evolved collateral sensitivity in sequential therapy is contingent on drug order and combination, collateral sensitivity effect size, general adaptation trade-offs, and also epistatic genetic interactions.

Evolved reciprocal collateral sensitivity generally limited bacterial adaptation. Adaptation was particularly constrained in treatments that switched to a β-lactam, as reflected by the elevated extinction rates. The effect was strongest when the first antibiotic was maintained and a second was added. This finding may point to a promising, yet currently unexplored treatment strategy, namely single-drug therapy followed by combination therapy (for example the addition of a β-lactam), that can maximize exploitation of the evolutionary trade-off. Yet, extinction rates were still high under unconstrained conditions, when drugs were replaced, and in spite of a relatively mild selection intensity. In the detailed analysis with the CAR/GEN drug pair, we observed higher extinction and slower growth improvements in strong, compared to mild dose increases. This finding is generally consistent with previous studies, performed in different context, in which narrowed mutation space upon fast environmental deterioration increased extinction frequencies (Bell and Gonzalez, 2011; Lindsey et al., 2013). Interestingly, extinction rates are often not reported as an evolutionary outcome in related studies, possibly because of a different main focus of the study (Yen and Papin, 2017), or because extinction could not be recorded due to the particular experimental set-up (i.e., usage of a morbidostat; Yoshida et al., 2017). Considering that antimicrobial therapy usually aims at elimination of bacterial pathogens and extinction frequencies are known from previous evolution experiments to vary among treatment types (Barbosa et al., 2018; Hansen et al., 2017; Roemhild et al., 2018; Torella et al., 2010), their consideration should help us to refine our understanding of treatment efficacy.

In the treatments that replaced drugs, evolutionary stability of the resistance trade-off was determined by drug order. In both datasets (clonal and population-level evolution), we observed high relative stability for collateral sensitivity treatments switching from an aminoglycoside to a β-lactam. Our detailed analysis for the CAR/GEN pair at the clonal level demonstrates that this stability may be caused by slower adaptation (Figure 3a), and efficient re-sensitization (Figure 3c). Also, the slow adaptation rates may be the result of selection favoring the reversal of two evolutionary trade-offs (Figures 3e and 6d) together with a relative paucity of accessible resistance mutations. The latter is highlighted by several instances of re-mutation of pmrB, where the effect of the ancestral collateral-sensitivity SNP was countered by additional larger deletions, potentially leading to a loss of function of the same gene (Figure 4b). The efficient re-sensitization that was observed during switches from an aminoglycoside to a β-lactam may be explained by pleiotropy and drug-specific negative epistasis (Figure 4c and d). On the other hand, collateral sensitivity treatments that switch to aminoglycosides, tended to show lower stability, as reflected by the lower levels of extinction (Figure 6a) and the lack of re-sensitization (Figure 6b) at the population level. For instance, the collateral sensitivity treatments CAR>STR and CIP>STR were frequently countered by the evolving bacteria. These treatments hardly resulted in extinction but allowed for rapid loss of susceptibility to both drugs (Figure 5b). While our data cannot fully explain the high instability, it appears that small values for a general adaptation trade-off (i.e., indicated by reduced growth rates in drug-free environments), which are also frequently observed for CIP resistance mutations (Huseby et al., 2017), accelerated adaptation (Figure 6d). Overall, our finding of strong variation in the stability of collateral sensitivity treatments in pathogenic P. aeruginosa indicate the importance for a careful evaluation of new treatment options.

Our detailed analyses for GEN>CAR emphasize the importance of epistasis for the stability of collateral sensitivity. A recent publication confirmed that the expression of particular collateral sensitivity mutations strongly depended on the genetic background and could even cause opposite effects in the closely related species of E. coli and Salmonella enterica due to epistasis (Apjok et al., 2019). Our work on the GEN>CAR switch now shows, that epistasis is of high importance also for the temporal stability of antibiotic sensitivity. Here, drug re-sensitization in the unconstrained treatments was likely dependent on negative epistasis among pleiotropic resistance mutations. Mutations in pmrB and the efflux regulators nalC and nalD interacted negatively with each other and caused a complete re-sensitization of bacteria that were previously resistant against GEN. While re-sensitization reliably occurred for the GEN>CAR treatment, it did not occur in the reverse case. Similar examples of antibiotic re-sensitization were previously reported for E. coli and P. aeruginosa, but these relied on different mechanisms. For E. coli, repeated alternation between two antibiotics led to re-sensitization as a consequence of clonal interference between variants in two genes, secD and/or basB (Yoshida et al., 2017). The change between drugs prevented fixation of the competing variants, thus maintaining pleiotropic alleles and thereby the allele causing resistance to one drug and hypersensitivity to the other (Yoshida et al., 2017). In the previous example for P. aeruginosa, hypersensitivity to a β-lactam depended on an expression imbalance of the MexAB-OprM and the MexEF-OprN efflux systems after exposure to a fluoroquinolone (Maseda et al., 2004; Sobel et al., 2005; Yen and Papin, 2017). Interestingly, partial re-sensitization against the aminoglycoside tobramycin was dependent on inducible resistance, a phenomenon mediated by the MexXY-OprM efflux pump, whereby expression, and consequently resistance, is induced by the presence of the drug, but then reverted after its removal (Hocquet et al., 2003; Yen and Papin, 2017). We conclude that our finding of negative epistasis between pleiotropic resistance mutations is a previously unknown mechanism underlying re-sensitization. Whilst positive epistasis can substantially amplify resistance gains (Wistrand-Yuen et al., 2018), negative epistasis can limit evolutionary trajectories (Weinreich et al., 2006), thus possibly contributing to efficacy of treatment in our case.

The mutations observed in this study are commonly associated with variants observed in clinical isolates, particularly in those obtained from cystic fibrosis patients (Hancock and Speert, 2000; Jansen et al., 2016; Marvig et al., 2015; Tueffers et al., 2019). Both efflux regulators (including nalC and nalD) and two-component regulatory systems (mainly pmrAB and phoQF) were repeatedly reported to be associated with intermediate and highly resistant isolates of P. aeruginosa, E. coli, Acinetobacter baumannii and other pathogenic species against colistin and aminoglycosides (Cao et al., 2004; Gerson et al., 2019; Sato et al., 2018). This overlap suggests that the negative epistasis between the genes involved in resistance against β-lactam and aminoglycosides observed here could also be encountered and exploited in clinical settings.

We anticipate that the findings of our study could help to guide the design of sustainable antibiotic therapy that controls the infection, whilst reducing the emergence of multidrug resistance. In principle, the refined exploitation of collateral sensitivity could represent a promising addition to new evolution-informed treatment strategies, including as alternatives specific combination treatments (Barbosa et al., 2018; Chait et al., 2007; Rodriguez de Evgrafov et al., 2015; Gonzales et al., 2015; Munck et al., 2014), fast sequential therapy (Nichol et al., 2015; Yoshida et al., 2017), or treatments utilizing negative hysteresis (Roemhild et al., 2018). The success of this treatment strategy depends on several key factors. One critical prerequisite is that collateral sensitivity does evolve in response to treatment, which may vary depending on alternative evolutionary paths to resistance against the initially used drug A (Barbosa et al., 2017; Nichol et al., 2019). Our new data additionally suggest that treatment can be further optimized by switching from an aminoglycoside to a β-lactam and/or by focusing on drugs, for which resistance is associated with large general adaptation trade-offs, and/or by using drug combinations with small collateral sensitivity effect sizes. These new insights clearly warrant further evaluation, for example by using clinical isolates of P. aeruginosa and/or patient-related test conditions.

Sequencing data have been deposited at NCBI under the BioProject number PRJNA524114. All other data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for all figures.

1. 1. Barbosa C
   2. Roemhild R
   3. Rosenstiel P
   4. Schulenburg H

   (2019) NCBI Bioproject

   ID PRJNA524114. Evolutionary stability of antibiotic collateral sensitivity.

   https://www.ncbi.nlm.nih.gov/bioproject/?term=PRJNA524114

1. 1. Abyzov A
   2. Urban AE
   3. Snyder M
   4. Gerstein M

   (2011) CNVnator: an approach to discover, genotype, and characterize typical and atypical CNVs from family and population genome sequencing

   Genome Research 21:974–984.

   https://doi.org/10.1101/gr.114876.110

   • PubMed
   • Google Scholar
2. 1. Apjok G
   2. Boross G
   3. Nyerges Á
   4. Fekete G
   5. Lázár V
   6. Papp B
   7. Pál C
   8. Csörgő B

   (2019) Limited evolutionary conservation of the phenotypic effects of antibiotic resistance mutations

   Molecular Biology and Evolution 36:1601–1611.

   https://doi.org/10.1093/molbev/msz109

   • PubMed
   • Google Scholar
3. 1. Barbosa C
   2. Trebosc V
   3. Kemmer C
   4. Rosenstiel P
   5. Beardmore R
   6. Schulenburg H
   7. Jansen G

   (2017) Alternative evolutionary paths to bacterial antibiotic resistance cause distinct collateral effects

   Molecular Biology and Evolution 34:2229–2244.

   https://doi.org/10.1093/molbev/msx158

   • PubMed
   • Google Scholar
4. 1. Barbosa C
   2. Beardmore R
   3. Schulenburg H
   4. Jansen G

   (2018) Antibiotic combination efficacy (ACE) networks for a Pseudomonas aeruginosa model

   PLOS Biology 16:e2004356.

   https://doi.org/10.1371/journal.pbio.2004356

   • PubMed
   • Google Scholar
5. 1. Baym M
   2. Stone LK
   3. Kishony R

   (2016) Multidrug evolutionary strategies to reverse antibiotic resistance

   Science 351:aad3292.

   https://doi.org/10.1126/science.aad3292

   • Google Scholar
6. 1. Bell G
   2. Gonzalez A

   (2011) Adaptation and evolutionary rescue in metapopulations experiencing environmental deterioration

   Science 332:1327–1330.

   https://doi.org/10.1126/science.1203105

   • PubMed
   • Google Scholar
7. 1. Bloemberg GV
   2. Keller PM
   3. Stucki D
   4. Trauner A
   5. Borrell S
   6. Latshang T
   7. Coscolla M
   8. Rothe T
   9. Hömke R
   10. Ritter C
   11. Feldmann J
   12. Schulthess B
   13. Gagneux S
   14. Böttger EC

   (2015) Acquired resistance to bedaquiline and delamanid in therapy for tuberculosis

   New England Journal of Medicine 373:1986–1988.

   https://doi.org/10.1056/NEJMc1505196

   • Google Scholar
8. 1. Bolger AM
   2. Lohse M
   3. Usadel B

   (2014) Trimmomatic: a flexible trimmer for illumina sequence data

   Bioinformatics 30:2114–2120.

   https://doi.org/10.1093/bioinformatics/btu170

   • PubMed
   • Google Scholar
9. 1. Cao L
   2. Srikumar R
   3. Poole K

   (2004) MexAB-OprM hyperexpression in NalC-type multidrug-resistant Pseudomonas aeruginosa: identification and characterization of the nalC gene encoding a repressor of PA3720-PA3719

   Molecular Microbiology 53:1423–1436.

   https://doi.org/10.1111/j.1365-2958.2004.04210.x

   • PubMed
   • Google Scholar
10. 1. Chait R
    2. Craney A
    3. Kishony R

    (2007) Antibiotic interactions that select against resistance

    Nature 446:668–671.

    https://doi.org/10.1038/nature05685

    • PubMed
    • Google Scholar
11. 1. Cingolani P
    2. Platts A
    3. Wang leL
    4. Coon M
    5. Nguyen T
    6. Wang L
    7. Land SJ
    8. Lu X
    9. Ruden DM

    (2012) A program for annotating and predicting the effects of single Nucleotide Polymorphisms, SnpEff: snps in the genome of Drosophila Melanogaster strain w1118; iso-2; iso-3

    Fly 6:80–92.

    https://doi.org/10.4161/fly.19695

    • PubMed
    • Google Scholar
12. 1. Davies J
    2. Davies D

    (2010) Origins and evolution of antibiotic resistance

    Microbiology and Molecular Biology Reviews 74:417–433.

    https://doi.org/10.1128/MMBR.00016-10

    • PubMed
    • Google Scholar
13. 1. Dhawan A
    2. Nichol D
    3. Kinose F
    4. Abazeed ME
    5. Marusyk A
    6. Haura EB
    7. Scott JG

    (2017) Collateral sensitivity networks reveal evolutionary instability and novel treatment strategies in ALK mutated non-small cell lung Cancer

    Scientific Reports 7:1232.

    https://doi.org/10.1038/s41598-017-00791-8

    • PubMed
    • Google Scholar
14. 1. El'Garch F
    2. Jeannot K
    3. Hocquet D
    4. Llanes-Barakat C
    5. Plesiat P

    (2007) Cumulative effects of several nonenzymatic mechanisms on the resistance of Pseudomonas aeruginosa to aminoglycosides

    Antimicrobial Agents and Chemotherapy 51:1016–1021.

    https://doi.org/10.1128/AAC.00704-06

    • Google Scholar
15. 1. Feinbaum RL
    2. Urbach JM
    3. Liberati NT
    4. Djonovic S
    5. Adonizio A
    6. Carvunis AR
    7. Ausubel FM

    (2012) Genome-wide identification of Pseudomonas aeruginosa virulence-related genes using a Caenorhabditis elegans infection model

    PLOS Pathogens 8:e1002813.

    https://doi.org/10.1371/journal.ppat.1002813

    • PubMed
    • Google Scholar
16. 1. Gatenby RA
    2. Silva AS
    3. Gillies RJ
    4. Frieden BR

    (2009) Adaptive therapy

    Cancer Research 69:4894–4903.

    https://doi.org/10.1158/0008-5472.CAN-08-3658

    • PubMed
    • Google Scholar
17. 1. Gerson S
    2. Betts JW
    3. Lucaßen K
    4. Nodari CS
    5. Wille J
    6. Josten M
    7. Higgins PG

    (2019) Investigation of novel pmrB and eptA mutations in isogenic Acinetobacter baumannii isolates associated with colistin resistance and increased virulence

    Antimicrobial Agents and Chemotherapy 63:e01586.

    https://doi.org/10.1128/AAC.01586-18

    • Google Scholar
18. 1. Gonzales PR
    2. Pesesky MW
    3. Bouley R
    4. Ballard A
    5. Biddy BA
    6. Suckow MA
    7. Wolter WR
    8. Schroeder VA
    9. Burnham CA
    10. Mobashery S
    11. Chang M
    12. Dantas G

    (2015) Synergistic, collaterally sensitive β-lactam combinations suppress resistance in MRSA

    Nature Chemical Biology 11:855–861.

    https://doi.org/10.1038/nchembio.1911

    • PubMed
    • Google Scholar
19. 1. Gooderham WJ
    2. Hancock REW

    (2009) Regulation of virulence and antibiotic resistance by two-component regulatory systems in Pseudomonas aeruginosa

    FEMS Microbiology Reviews 33:279–294.

    https://doi.org/10.1111/j.1574-6976.2008.00135.x

    • Google Scholar
20. 1. Gottesman MM

    (2002) Mechanisms of Cancer drug resistance

    Annual Review of Medicine 53:615–627.

    https://doi.org/10.1146/annurev.med.53.082901.103929

    • PubMed
    • Google Scholar
21. 1. Hancock REW
    2. Speert DP

    (2000) Antibiotic resistance in Pseudomonas aeruginosa: mechanisms and impact on treatment

    Drug Resistance Updates 3:247–255.

    https://doi.org/10.1054/drup.2000.0152

    • Google Scholar
22. 1. Hansen E
    2. Woods RJ
    3. Read AF

    (2017) How to use a chemotherapeutic agent when resistance to it threatens the patient

    PLOS Biology 15:e2001110.

    https://doi.org/10.1371/journal.pbio.2001110

    • PubMed
    • Google Scholar
23. 1. Hegreness M
    2. Shoresh N
    3. Damian D
    4. Hartl D
    5. Kishony R

    (2008) Accelerated evolution of resistance in multidrug environments

    PNAS 105:13977–13981.

    https://doi.org/10.1073/pnas.0805965105

    • Google Scholar
24. 1. Hocquet D
    2. Vogne C
    3. El Garch F
    4. Vejux A
    5. Gotoh N
    6. Lee A
    7. Lomovskaya O
    8. Plesiat P

    (2003) MexXY-OprM efflux pump is necessary for adaptive resistance of Pseudomonas aeruginosa to aminoglycosides

    Antimicrobial Agents and Chemotherapy 47:1371–1375.

    https://doi.org/10.1128/AAC.47.4.1371-1375.2003

    • Google Scholar
25. 1. Huseby DL
    2. Pietsch F
    3. Brandis G
    4. Garoff L
    5. Tegehall A
    6. Hughes D

    (2017) Mutation supply and relative fitness shape the genotypes of ciprofloxacin-resistant Escherichia coli

    Molecular Biology and Evolution 34:msx052.

    https://doi.org/10.1093/molbev/msx052

    • Google Scholar
26. 1. Imamovic L
    2. Ellabaan MMH
    3. Dantas Machado AM
    4. Citterio L
    5. Wulff T
    6. Molin S
    7. Krogh Johansen H
    8. Sommer MOA

    (2018) Drug-Driven phenotypic convergence supports rational treatment strategies of chronic infections

    Cell 172:121–134.

    https://doi.org/10.1016/j.cell.2017.12.012

    • PubMed
    • Google Scholar
27. 1. Imamovic L
    2. Sommer MO

    (2013) Use of collateral sensitivity networks to design drug cycling protocols that avoid resistance development

    Science Translational Medicine 5:204ra132.

    https://doi.org/10.1126/scitranslmed.3006609

    • PubMed
    • Google Scholar
28. 1. Jansen G
    2. Crummenerl LL
    3. Gilbert F
    4. Mohr T
    5. Pfefferkorn R
    6. Thänert R
    7. Rosenstiel P
    8. Schulenburg H

    (2015) Evolutionary transition from pathogenicity to commensalism: global regulator mutations mediate fitness gains through virulence attenuation

    Molecular Biology and Evolution 32:2883–2896.

    https://doi.org/10.1093/molbev/msv160

    • PubMed
    • Google Scholar
29. 1. Jansen G
    2. Mahrt N
    3. Tueffers L
    4. Barbosa C
    5. Harjes M
    6. Adolph G
    7. Friedrichs A
    8. Krenz-Weinreich A
    9. Rosenstiel P
    10. Schulenburg H

    (2016) Association between clinical antibiotic resistance and susceptibility of Pseudomonas in the cystic fibrosis lung

    Evolution, Medicine, and Public Health 2016:182–194.

    https://doi.org/10.1093/emph/eow016

    • PubMed
    • Google Scholar
30. Preprint

    1. Jiao YJ
    2. Baym M
    3. Veres A
    4. Kishony R

    (2016) Population diversity jeopardizes the efficacy of antibiotic cycling

    bioRxiv.

    https://doi.org/10.1101/082107

    • Google Scholar
31. 1. Kim S
    2. Lieberman TD
    3. Kishony R

    (2014) Alternating antibiotic treatments constrain evolutionary paths to multidrug resistance

    PNAS 111:14494–14499.

    https://doi.org/10.1073/pnas.1409800111

    • Google Scholar
32. 1. Lázár V
    2. Singh GP
    3. Spohn R
    4. Nagy I
    5. Horváth B
    6. Hrtyan M

    (2013)

    Bacterial evolution of antibiotic hypersensitivity

    Molecular Systems Biology 1:.

    • Google Scholar
33. 1. Lázár V
    2. Nagy I
    3. Spohn R
    4. Csörgő B
    5. Györkei Á
    6. Nyerges Á
    7. Horváth B
    8. Vörös A
    9. Busa-Fekete R
    10. Hrtyan M
    11. Bogos B
    12. Méhi O
    13. Fekete G
    14. Szappanos B
    15. Kégl B
    16. Papp B
    17. Pál C

    (2014) Genome-wide analysis captures the determinants of the antibiotic cross-resistance interaction network

    Nature Communications 5:4352.

    https://doi.org/10.1038/ncomms5352

    • PubMed
    • Google Scholar
34. 1. Li H
    2. Handsaker B
    3. Wysoker A
    4. Fennell T
    5. Ruan J
    6. Homer N
    7. Marth G
    8. Abecasis G
    9. Durbin R
    10. 1000 Genome Project Data Processing Subgroup

    (2009) The sequence alignment/Map format and SAMtools

    Bioinformatics 25:2078–2079.

    https://doi.org/10.1093/bioinformatics/btp352

    • PubMed
    • Google Scholar
35. Book

    1. Li XZ
    2. Elkins CA
    3. Zgurskaya H

    (2016)

    Efflux-Mediated Antimicrobial Resistance in Bacteria: Mechanisms, Regulation and Clinical Implications

    New York: Springer.

    • Google Scholar
36. 1. Li H
    2. Durbin R

    (2010) Fast and accurate long-read alignment with Burrows-Wheeler transform

    Bioinformatics 26:589–595.

    https://doi.org/10.1093/bioinformatics/btp698

    • PubMed
    • Google Scholar
37. 1. Lindsey HA
    2. Gallie J
    3. Taylor S
    4. Kerr B

    (2013) Evolutionary rescue from extinction is contingent on a lower rate of environmental change

    Nature 494:463–467.

    https://doi.org/10.1038/nature11879

    • PubMed
    • Google Scholar
38. 1. Maltas J
    2. Wood KB

    (2019) Pervasive and diverse collateral sensitivity profiles inform optimal strategies to limit antibiotic resistance

    PLOS Biology 17:e3000515.

    https://doi.org/10.1371/journal.pbio.3000515

    • Google Scholar
39. 1. Marvig RL
    2. Sommer LM
    3. Molin S
    4. Johansen HK

    (2015) Convergent evolution and adaptation of Pseudomonas aeruginosa within patients with cystic fibrosis

    Nature Genetics 47:57–64.

    https://doi.org/10.1038/ng.3148

    • PubMed
    • Google Scholar
40. 1. Maseda H
    2. Sawada I
    3. Saito K
    4. Uchiyama H
    5. Nakae T
    6. Nomura N

    (2004) Enhancement of the mexAB-oprM efflux pump expression by a Quorum-Sensing autoinducer and its cancellation by a regulator, MexT, of the mexEF-oprN efflux pump operon in Pseudomonas aeruginosa

    Antimicrobial Agents and Chemotherapy 48:1320–1328.

    https://doi.org/10.1128/AAC.48.4.1320-1328.2004

    • Google Scholar
41. 1. Munck C
    2. Gumpert HK
    3. Wallin AI
    4. Wang HH
    5. Sommer MO

    (2014) Prediction of resistance development against drug combinations by collateral responses to component drugs

    Science Translational Medicine 6:262ra156.

    https://doi.org/10.1126/scitranslmed.3009940

    • PubMed
    • Google Scholar
42. 1. Nichol D
    2. Jeavons P
    3. Fletcher AG
    4. Bonomo RA
    5. Maini PK
    6. Paul JL
    7. Gatenby RA
    8. Anderson AR
    9. Scott JG

    (2015) Steering evolution with sequential therapy to prevent the emergence of bacterial antibiotic resistance

    PLOS Computational Biology 11:e1004493.

    https://doi.org/10.1371/journal.pcbi.1004493

    • PubMed
    • Google Scholar
43. 1. Nichol D
    2. Rutter J
    3. Bryant C
    4. Hujer AM
    5. Lek S
    6. Adams MD
    7. Jeavons P
    8. Anderson ARA
    9. Bonomo RA
    10. Scott JG

    (2019) Antibiotic collateral sensitivity is contingent on the repeatability of evolution

    Nature Communications 10:334.

    https://doi.org/10.1038/s41467-018-08098-6

    • PubMed
    • Google Scholar
44. 1. Oz T
    2. Guvenek A
    3. Yildiz S
    4. Karaboga E
    5. Tamer YT
    6. Mumcuyan N
    7. Ozan VB
    8. Senturk GH
    9. Cokol M
    10. Yeh P
    11. Toprak E

    (2014) Strength of Selection Pressure Is an Important Parameter Contributing to the Complexity of Antibiotic Resistance Evolution

    Molecular Biology and Evolution 31:2387–2401.

    https://doi.org/10.1093/molbev/msu191

    • Google Scholar
45. 1. Pál C
    2. Papp B
    3. Lázár V

    (2015) Collateral sensitivity of antibiotic-resistant microbes

    Trends in Microbiology 23:401–407.

    https://doi.org/10.1016/j.tim.2015.02.009

    • PubMed
    • Google Scholar
46. 1. Pluchino KM
    2. Hall MD
    3. Goldsborough AS
    4. Callaghan R
    5. Gottesman MM

    (2012) Collateral sensitivity as a strategy against Cancer multidrug resistance

    Drug Resistance Updates 15:98–105.

    https://doi.org/10.1016/j.drup.2012.03.002

    • PubMed
    • Google Scholar
47. 1. Podnecky NL
    2. Fredheim EGA
    3. Kloos J
    4. Sørum V
    5. Primicerio R
    6. Roberts AP
    7. Rozen DE
    8. Samuelsen Ø
    9. Johnsen PJ

    (2018) Conserved collateral antibiotic susceptibility networks in diverse clinical strains of Escherichia coli

    Nature Communications 9:3673.

    https://doi.org/10.1038/s41467-018-06143-y

    • PubMed
    • Google Scholar
48. 1. Rahme LG
    2. Stevens EJ
    3. Wolfort SF
    4. Shao J
    5. Tompkins RG
    6. Ausubel FM

    (1995) Common virulence factors for bacterial pathogenicity in plants and animals

    Science 268:1899–1902.

    https://doi.org/10.1126/science.7604262

    • PubMed
    • Google Scholar
49. 1. Rodriguez de Evgrafov M
    2. Gumpert H
    3. Munck C
    4. Thomsen TT
    5. Sommer MO

    (2015) Collateral resistance and sensitivity modulate evolution of High-Level resistance to drug combination treatment in Staphylococcus aureus

    Molecular Biology and Evolution 32:1175–1185.

    https://doi.org/10.1093/molbev/msv006

    • PubMed
    • Google Scholar
50. 1. Roemhild R
    2. Barbosa C
    3. Beardmore RE
    4. Jansen G
    5. Schulenburg H

    (2015) Temporal variation in antibiotic environments slows down resistance evolution in pathogenic Pseudomonas aeruginosa

    Evolutionary Applications 8:945–955.

    https://doi.org/10.1111/eva.12330

    • Google Scholar
51. 1. Roemhild R
    2. Gokhale CS
    3. Dirksen P
    4. Blake C
    5. Rosenstiel P
    6. Traulsen A
    7. Andersson DI
    8. Schulenburg H

    (2018) Cellular hysteresis as a principle to maximize the efficacy of antibiotic therapy

    PNAS 115:9767–9772.

    https://doi.org/10.1073/pnas.1810004115

    • PubMed
    • Google Scholar
52. 1. Roemhild R
    2. Schulenburg H

    (2019) Evolutionary ecology meets the antibiotic crisis: can we control pathogen adaptation through sequential therapy?

    Evolution, Medicine, and Public Health 2019:37–45.

    https://doi.org/10.1093/emph/eoz008

    • PubMed
    • Google Scholar
53. 1. Rosenkilde CEH
    2. Munck C
    3. Porse A
    4. Linkevicius M
    5. Andersson DI
    6. Sommer MOA

    (2019) Collateral sensitivity constrains resistance evolution of the CTX-M-15 β-lactamase

    Nature Communications 10:618.

    https://doi.org/10.1038/s41467-019-08529-y

    • PubMed
    • Google Scholar
54. 1. Sato T
    2. Shiraishi T
    3. Hiyama Y
    4. Honda H
    5. Shinagawa M
    6. Usui M
    7. Kuronuma K
    8. Masumori N
    9. Takahashi S
    10. Tamura Y
    11. Yokota S-ichi

    (2018) Contribution of Novel Amino Acid Alterations in PmrA or PmrB to Colistin Resistance in mcr -Negative Escherichia coli Clinical Isolates, Including Major Multidrug-Resistant Lineages O25b:H4-ST131- H 30Rx and Non-x

    Antimicrobial Agents and Chemotherapy 62:e00864.

    https://doi.org/10.1128/AAC.00864-18

    • Google Scholar
55. 1. Shaw AT
    2. Friboulet L
    3. Leshchiner I
    4. Gainor JF
    5. Bergqvist S
    6. Brooun A
    7. Burke BJ
    8. Deng YL
    9. Liu W
    10. Dardaei L
    11. Frias RL
    12. Schultz KR
    13. Logan J
    14. James LP
    15. Smeal T
    16. Timofeevski S
    17. Katayama R
    18. Iafrate AJ
    19. Le L
    20. McTigue M
    21. Getz G
    22. Johnson TW
    23. Engelman JA

    (2016) Resensitization to crizotinib by the lorlatinib ALK Resistance Mutation L1198F

    The New England Journal of Medicine 374:54–61.

    https://doi.org/10.1056/NEJMoa1508887

    • PubMed
    • Google Scholar
56. 1. Sobel ML
    2. Neshat S
    3. Poole K

    (2005) Mutations in PA2491 (mexS) promote MexT-dependent mexEF-oprN expression and multidrug resistance in a clinical strain of Pseudomonas aeruginosa

    Journal of Bacteriology 187:1246–1253.

    https://doi.org/10.1128/JB.187.4.1246-1253.2005

    • PubMed
    • Google Scholar
57. 1. Szybalski W
    2. Bryson V

    (1952)

    Genetic studies on microbial cross resistance to toxic agents I. cross resistance of Escherichia coli to fifteen antibiotics

    Journal of Bacteriology 64:489–499.

    • PubMed
    • Google Scholar
58. 1. Torella JP
    2. Chait R
    3. Kishony R

    (2010) Optimal drug synergy in antimicrobial treatments

    PLOS Computational Biology 6:e1000796.

    https://doi.org/10.1371/journal.pcbi.1000796

    • PubMed
    • Google Scholar
59. 1. Trebosc V
    2. Gartenmann S
    3. Royet K
    4. Manfredi P
    5. Tötzl M
    6. Schellhorn B
    7. Kemmer C

    (2016) A novel Genome-Editing platform for Drug-Resistant Acinetobacter baumannii reveals an AdeR-Unrelated tigecycline resistance mechanism

    Antimicrobial Agents and Chemotherapy 60:7263–7271.

    https://doi.org/10.1128/AAC.01275-16

    • Google Scholar
60. 1. Tueffers L
    2. Barbosa C
    3. Bobis I
    4. Schubert S
    5. Höppner M
    6. Rühlemann M
    7. Franke A
    8. Rosenstiel P
    9. Friedrichs A
    10. Krenz-Weinreich A
    11. Fickenscher H
    12. Bewig B
    13. Schreiber S
    14. Schulenburg H

    (2019) Pseudomonas aeruginosa populations in the cystic fibrosis lung lose susceptibility to newly applied β-lactams within 3 days

    Journal of Antimicrobial Chemotherapy 74:2916–2925.

    https://doi.org/10.1093/jac/dkz297

    • PubMed
    • Google Scholar
61. 1. von der Schulenburg JHG
    2. Hancock JM
    3. Pagnamenta A
    4. Sloggett JJ
    5. Majerus MEN
    6. Hurst GDD

    (2001) Extreme length and length variation in the first ribosomal internal transcribed spacer of ladybird beetles (Coleoptera: coccinellidae)

    Molecular Biology and Evolution 18:648–660.

    https://doi.org/10.1093/oxfordjournals.molbev.a003845

    • Google Scholar
62. 1. Wei Z
    2. Wang W
    3. Hu P
    4. Lyon GJ
    5. Hakonarson H

    (2011) SNVer: a statistical tool for variant calling in analysis of pooled or individual next-generation sequencing data

    Nucleic Acids Research 39:e132.

    https://doi.org/10.1093/nar/gkr599

    • PubMed
    • Google Scholar
63. 1. Weinreich DM
    2. Delaney NF
    3. Depristo MA
    4. Hartl DL

    (2006) Darwinian evolution can follow only very few mutational paths to fitter proteins

    Science 312:111–114.

    https://doi.org/10.1126/science.1123539

    • PubMed
    • Google Scholar
64. 1. Wistrand-Yuen E
    2. Knopp M
    3. Hjort K
    4. Koskiniemi S
    5. Berg OG
    6. Andersson DI

    (2018) Evolution of high-level resistance during low-level antibiotic exposure

    Nature Communications 9:1599.

    https://doi.org/10.1038/s41467-018-04059-1

    • PubMed
    • Google Scholar
65. 1. Wong A

    (2017) Epistasis and the evolution of antimicrobial resistance

    Frontiers in Microbiology 8:246.

    https://doi.org/10.3389/fmicb.2017.00246

    • PubMed
    • Google Scholar
66. 1. Ye K
    2. Schulz MH
    3. Long Q
    4. Apweiler R
    5. Ning Z

    (2009) Pindel: a pattern growth approach to detect break points of large deletions and medium sized insertions from paired-end short reads

    Bioinformatics 25:2865–2871.

    https://doi.org/10.1093/bioinformatics/btp394

    • PubMed
    • Google Scholar
67. 1. Yen P
    2. Papin JA

    (2017) History of antibiotic adaptation influences microbial evolutionary dynamics during subsequent treatment

    PLOS Biology 15:e2001586.

    https://doi.org/10.1371/journal.pbio.2001586

    • Google Scholar
68. 1. Yoshida M
    2. Reyes SG
    3. Tsuda S
    4. Horinouchi T
    5. Furusawa C
    6. Cronin L

    (2017) Time-programmable drug dosing allows the manipulation, suppression and reversal of antibiotic drug resistance in vitro

    Nature Communications 8:15589.

    https://doi.org/10.1038/ncomms15589

    • Google Scholar
69. 1. Zaretsky JM
    2. Garcia-Diaz A
    3. Shin DS
    4. Escuin-Ordinas H
    5. Hugo W
    6. Hu-Lieskovan S
    7. Torrejon DY
    8. Abril-Rodriguez G
    9. Sandoval S
    10. Barthly L
    11. Saco J
    12. Homet Moreno B
    13. Mezzadra R
    14. Chmielowski B
    15. Ruchalski K
    16. Shintaku IP
    17. Sanchez PJ
    18. Puig-Saus C
    19. Cherry G
    20. Seja E
    21. Kong X
    22. Pang J
    23. Berent-Maoz B
    24. Comin-Anduix B
    25. Graeber TG
    26. Tumeh PC
    27. Schumacher TN
    28. Lo RS
    29. Ribas A

    (2016) Mutations associated with acquired resistance to PD-1 blockade in melanoma

    New England Journal of Medicine 375:819–829.

    https://doi.org/10.1056/NEJMoa1604958

    • PubMed
    • Google Scholar
70. 1. Zhao B
    2. Sedlak JC
    3. Srinivas R
    4. Creixell P
    5. Pritchard JR
    6. Tidor B
    7. Lauffenburger DA
    8. Hemann MT

    (2016) Exploiting temporal collateral sensitivity in tumor clonal evolution

    Cell 165:234–246.

    https://doi.org/10.1016/j.cell.2016.01.045

    • PubMed
    • Google Scholar

Author details

1. Camilo Barbosa

   Department of Evolutionary Ecology and Genetics, University of Kiel, Kiel, Germany

   Contribution

   Conceptualization, Investigation, Visualization, Writing—original draft, Writing—review and editing

   Contributed equally with

   Roderich Römhild

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-0433-9311

2. Roderich Römhild

   1. Department of Evolutionary Ecology and Genetics, University of Kiel, Kiel, Germany
   2. Max Planck Institute for Evolutionary Biology, Plön, Germany

   Present address

   Department of Medical Biochemistry and Microbiology, Uppsala University, Uppsala, Sweden

   Contribution

   Conceptualization, Investigation, Visualization, Writing—original draft, Writing—review and editing

   Contributed equally with

   Camilo Barbosa

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-9480-5261

3. Philip Rosenstiel

   Institute of Clinical Molecular Biology, UKSH, Kiel, Germany

   Contribution

   Investigation, Methodology

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-9692-8828

4. Hinrich Schulenburg

   1. Department of Evolutionary Ecology and Genetics, University of Kiel, Kiel, Germany
   2. Max Planck Institute for Evolutionary Biology, Plön, Germany

   Contribution

   Conceptualization, Supervision, Funding acquisition, Writing—original draft, Writing—review and editing

   For correspondence

   hschulenburg@zoologie.uni-kiel.de

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-1413-913X

• 4,662

  views

• 681

  downloads

• 44

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.