Selective context, rather than persister cycling alone, drives resistance fixation in Escherichia coli

Garin Park; Hyein Kim; Sooyeon Song

doi:10.1038/s44259-026-00261-0

. 2026 Sep 8;4:57. doi: 10.1038/s44259-026-00261-0

Selective context, rather than persister cycling alone, drives resistance fixation in Escherichia coli

Garin Park ¹, Hyein Kim ¹, Sooyeon Song ^1,^2,^✉

PMCID: PMC13554269 PMID: 42711467

Abstract

Whether persister cells contribute to the evolution of antibiotic resistance and, if so, under what selective conditions this occurs, remain unresolved. Here, we examined whether repeated persister cycling itself promotes resistance evolution and how persister-associated minor variants are retained, lost, or fixed under distinct selective contexts. We compared five Escherichia coli cellular states: mutation-induced cells (M), persister-Amp cycling cells (A), persister cycling cells without selection (R), stationary-phase cells (S), and NaCl-stored persisters (P), using state-resolved whole-genome sequencing and phenotypic assays. Persister cycling without selection, stationary-phase cells, and NaCl-stored persisters maintained baseline MICs and showed no detectable high-frequency variant fixation. In contrast, mutation-induced cells fixed efflux-regulatory mutations in marR, acrR, and acrB, increasing MIC to 32 μg/mL. Persister-Amp cycling cells showed an intermediate MIC increase to 16 μg/mL without detectable quality-filtered genetic fixation, distinguishing this state from mutation-induced resistance. The stfE/stfP prophage background further shaped adaptive routes under mutation-inducing conditions. Exploratory analysis revealed sub-threshold low-frequency variant signals during persister cycling, but lineage tracking showed that these variants were not stage-specifically fixed and were instead stochastically retained or lost. These findings support a model in which persister cycling can reveal low-frequency genetic heterogeneity, but fixed resistance evolution requires selection that promotes variant retention and clonal expansion. Thus, resistance fixation was governed primarily by selective context during regrowth rather than by the persister state itself.

Subject terms: Evolution, Microbiology

Introduction

Antibiotic resistance remains a significant threat to global health¹. Under antibiotic stress, bacteria deploy distinct survival strategies, including resistance, tolerance, and persistence. Resistance arises from heritable genetic changes that enable antibiotic degradation, target modification, production of alternative binding proteins, or altered regulation of efflux pumps². By contrast, tolerance and persistence are primarily phenotypic survival strategies. Tolerant cells survive transient antibiotic exposure through reduced growth or metabolic activity, whereas persister cells represent a subpopulation that survives lethal stress without necessarily acquiring heritable resistance^3,4. Although persisters have traditionally been viewed as dormant or metabolically inactive, persistence is now recognized as a heterogeneous physiological state that can involve dynamic metabolic activity and stress-dependent survival programs^5,6. Following the consensus framework by Balaban et al.⁷, persistence can be classified into triggered persistence, induced by environmental cues such as starvation or antibiotic stress, and spontaneous persistence, which occurs stochastically during steady-state exponential growth.

A central unresolved question is whether persister cells merely survive antibiotic exposure or also contribute to the evolution of antibiotic resistance. Persisters are typically isogenic to the susceptible population and resume growth after stress removal without heritable resistance⁷. However, persistence levels can be modulated by specific mutations, such as hipA7, which increase the frequency of the persister phenotype without directly conferring fixed resistance^8,9. In addition, several studies have proposed that persisters may facilitate resistance evolution by extending survival under selective pressure^10–12. This prolonged survival may increase the opportunity for genetic change through stress-induced DNA repair^13,14, oxidative damage¹⁵, selection of tolerance-associated variants^16,17, or residual metabolic and transcriptional activity^18,19. These findings suggest that persisters may provide a physiological state in which resistance-associated variation can arise or persist, but they do not fully explain whether such variants are retained, lost, or fixed during repeated persister formation and resuscitation.

Cryptic prophages provide another potential regulatory layer in this process. These replication-defective prophage elements are integrated into bacterial chromosomes and are increasingly recognized as modulators of bacterial physiology, including stress survival, biofilm formation, SOS signaling, and genome plasticity^20,21. In Escherichia coli, the e14 prophage can undergo excision or recombination during antibiotic exposure, leading to genomic rearrangements or single-nucleotide polymorphisms that may facilitate adaptation^22,23. Cryptic prophages have also been implicated in persistence-related physiology. In E. coli, the CP4-57-encoded alpA inhibits phosphate uptake and delays persister resuscitation, whereas Gifsy-1 activation in Salmonella triggers dormancy through tRNA cleavage^24,25. These observations suggest context-dependent crosstalk among persistence, stress survival, resistance-associated adaptation, and cryptic prophage function.

Despite these insights, it remains unclear whether repeated persister cycling itself promotes the evolution of fixed resistance and how persister-associated minor variants are retained, lost, or fixed under different selective contexts. Here, we addressed this question by comparing five predefined E. coli cellular states that separate physiological state from selective context: mutation-induced cells, persister-Amp cycling cells, persister cycling cells without selection, stationary-phase cells, and NaCl-stored persisters. Using state-resolved whole-genome sequencing and phenotypic assays, we tracked both high-frequency fixed variants and sub-threshold low-frequency variant signals. We found that persister cycling without selection maintained baseline MIC and showed no high-frequency variant fixation, whereas mutation-induced selection fixed efflux-regulatory mutations and increased MIC. Persister-Amp cycling produced an intermediate MIC increase without detectable quality-filtered genetic fixation. We further found that the stfE/stfP prophage axis shaped adaptive routes under mutation-inducing conditions. Together, our results support a model in which persister cycling can reveal hidden minor-variant heterogeneity, but fixed resistance evolution requires selective pressure that drives variant retention and clonal expansion.

Results

Persister cycling without selection maintains baseline MIC

To determine whether repeated persister formation and resuscitation promote resistance evolution, we compared five cellular states: mutation-induced cells (M), persister-cycling cells without selection (R), persister-Amp cycling cells (A), stationary-phase cells (S), and NaCl-stored persisters (P) (Fig. 1a; Table 1). These conditions were designed to compare physiological survival without additional selection (R, P, and S), persister-derived regrowth under sub-MIC selection (A), and repeated mutation-induced selection (M). By comparing R and A, we evaluated whether repeated persister formation and resuscitation, or antibiotic selection during regrowth, are associated with resistance-related phenotypic outcomes.

Fig. 1 — a Scheme of sampling and cycling. Mutation-induced section (M): Exponential-phase cells (E) were exposed to Amp (5 µg/mL, 30 min), washed twice in 0.85% NaCl, and resuspended in fresh LB with Amp; the cycle was repeated three times. 6 biological replicates were plated onto LB agar containing Amp (8 µg/mL), and 3 biological replicates were sequenced. Persister cycling cells (R): Persisters were generated by sequential expose to Rif (100 µg/mL, 30 min) and Amp (100 µg/mL, 3 h), washed, resuspended in fresh LB, and resuscitated; this cycle was repeated three times, and resuscitated persister cells were collected at last stage. 6 biological replicates were plated onto LB agar containing Amp (8 µg/mL), and 3 biological replicates were sequenced. Persister-Amp cycling cells (A): Persister cells were generated, resuscitated, and exposed to Amp (5 µg/mL, 30 min); this cycle was repeated three times, and resuscitated persister cells exposed to Amp were collected at the final stage. 3 biological replicates were plated onto LB agar containing Amp (8 µg/mL) and sequenced. Other states: NaCl-stored persister cells (P) were prepared by incubating persister cells in 0.85% NaCl for 4 days. 4 biological replicates of these were plated onto LB agar containing Amp (8 µg/mL), 3 biological replicates were sequenced. Stationary-phase cells (S) were obtained after 4 days in LB. 4 biological replicates of these were plated onto LB agar containing Amp (8 µg/mL), 2 biological replicates were sequenced; The viable cell counts shown in the figure represent the mean values from biological replicates at each state (Table S7). b Representative colony morphologies under antibiotic stress. Colonies from each group were grown on LB agar containing Amp (8 µg/mL). Large colonies were defined as those with a diameter ≥ 2.5 mm (yellow arrows). c Quantification of CFU and large-colony counts. Cells in different cellular states were plated on LB agar containing Amp (8 µg/mL); large-sized colonies (red) and total colonies (blue) are shown (see Table S1). Large colonies (diameter ≥ 2.5 mm) on ampicillin-containing agar were used as a phenotypic indicator of resistance-associated growth, as bacteria capable of growing above the MIC are classified as resistant based on their elevated MIC²⁶. This approach follows prior experimental evolution studies in which large-colony formation on antibiotic plates served as a readout of resistance acquisition under repeated antibiotic selection²⁷. Data are presented as mean ± s.d. Biological replicates were as follows: mutation-induced cells, n = 6; persister cycling cells, n = 6, persister-Amp cycling cells, n = 3; all other cell groups, n = 4. Post hoc analysis was performed using Tukey’s test. Differences from the exponential-phase baseline were assessed by two-way ANOVA (P < 0.05); stages labeled with a were significantly different from the baseline, corresponding to exponential-phase cells (mutation induced cell stage 0).

Table 1.

Definition and characteristics of the five cellular states used in this study

Cell state	Treatment & selection	Selective pressure	MIC (µg/mL)	Survival Basis	Biological Rationale	Field classification⁷
Exponential (E)	Freshly regrown in LB broth	No	8	Baseline	To establish a genomic and phenotypic baseline representing actively growing, antibiotic-susceptible cells prior to any stress exposure.	Normal cells
Mutation-induced (M)	3× sub-MIC Amp (5 μg/mL, 30 min each) applied to exponential-phase cells to induce mutagenesis; survivors plated on LB + Amp (8 μg/mL) to select resistant mutants	Yes (sub-MIC)	32	Genetic Fixed mutations (AF = 1) detected; MIC elevated	To define the canonical resistance trajectory under repeated sub-MIC selection, serving as a comparator for persister-derived states in terms of mutational fixation and MIC elevation	Resistant mutants
Persister cycling (R)	3× cycles: Rif (100 μg/mL, 30 min) → Amp (100 μg/mL, 3 h) → resuscitation in fresh LB (30 min); no selection during resuscitation	No	8 (Stable)	No quality-filtered variants; MIC unchanged	To determine whether repeated persister formation and resuscitation cycles alone, in the absence of antibiotic selection pressure, generate heritable resistance-associated mutations or elevate MIC.	Triggered Persistence (Rif-induced, without selection)
Persister-Amp cycling (A)	3× cycles: Rif → Amp → resuscitation with sub-MIC Amp (5 μg/mL, 30 min) applied during resuscitation	Yes (Sub-MIC during resuscitation)	16	No quality-filtered variants; MIC elevated	To evaluate whether selective pressure applied during resuscitation is sufficient to drive resistance acquisition in a persister-derived population, and to directly compare the resulting adaptive trajectory with both persister cycling without selection (R) and continuous sub-MIC selection (M)	Triggered Persistence (Rif-induced, under selection)
Stationary (S)	4-day incubation in LB broth; nutrient depletion; no antibiotic exposure	No	8 (Stable)	Phenotypic (Minor variants (AF < 0.5) only; MIC unchanged throughout 4 days)	To obtain a population enriched for starvation-triggered persisters without artificial induction, serving as a comparator to Rif-induced triggered persisters (R, A, and P). To obtain a population enriched for starvation-triggered persisters without artificial induction, serving as a comparator to Rif-induced triggered persisters (R, A, and P).	Triggered persistence/ Tolerance (Starvation-induced)
Stored Persister (P)	Rif (100 μg/mL, 30 min) → Amp (100 μg/mL, 3 h) → stored in 0.85% NaCl at 37 °C for 4 days; no resuscitation	No	8 (Stable)	Phenotypic (Minor variants (AF < 0.5) only; MIC unchanged after resuscitation)	To examine variant accumulation in Rif-induced triggered persisters maintained in a nutrient-free buffer for 4 days without resuscitation, isolating the effect of prolonged arrested state from resuscitation-phase competitive dilution.	Triggered Persistence (Rif-induced, arrested without resuscitation)

Open in a new tab

Five distinct cellular states were established from Escherichia coli BW25113 as described in the Treatment & Selection column. MIC, minimum inhibitory concentration of ampicillin determined by broth microdilution after the final treatment cycle; values represent the lowest concentration preventing visible growth after 24 h incubation. For mutation-induced cells, single colonies selected on LB agar containing 8 μg/mL ampicillin at the final cycle were regrown in antibiotic-free LB medium and subsequently used as the inoculum for MIC determination. For all other states, bulk cultures of the final cycle were adjusted to the same turbidity as exponential-phase cells and subsequently used as the inoculum for MIC determination. Survival basis: Genetic, high-frequency fixed mutations (allele frequency, AF = 1) detected by whole-genome sequencing with concurrent MIC elevation; Phenotypic, minor variants only (AF < 0.5) without fixed mutations, MIC unchanged. Field classification follows the consensus framework of Balaban et al. (2019).

Survival and large-colony frequency differed markedly across conditions. M showed stepwise increases in survival over three rounds (1.8 × 10³ → 4.2 × 10³ CFU/mL), with the large-colony fraction increasing from 4.0% to 9.0% (Fig. 1b; Table S1). A also showed increased survival (2.0 × 10¹ → 8.5 × 10² CFU/mL), with the large-colony fraction increasing from 33.33% to 41.94% (Fig. 1b; Table S1). In contrast, R showed no significant increase in survival across cycles (8.0 × 10¹ → 9.3 × 10¹ CFU/mL) or large-colony frequency (Fig. 1c; Table S1). NaCl-stored persisters and stationary-phase cells showed similar total colony counts on ampicillin plates (1.3 × 10³ ± 8.9 × 10¹ vs. 1.1 × 10³ ± 2.9 × 10¹ CFU/mL), but differed in large-colony fraction (1.1% vs. 8.1%) (Fig. 1c; Table S1). Large colonies on ampicillin agar were used as a phenotypic proxy for resistance-associated growth, defined here as the ability to grow above the MIC. This approach is consistent with the resistance classification framework in which MIC elevation serves as the primary criterion for distinguishing resistance from tolerance and persistence²⁶. Large-colony formation on antibiotic plates has also been used as a readout of resistance acquisition in prior experimental evolution studies under repeated antibiotic selection²⁷.

MIC outcomes also differed after three cycles. R maintained the baseline MIC of 8 μg/mL, whereas A showed a twofold MIC increase to 16 μg/mL. In M, colony-picked isolates reached 32 μg/mL, representing a fourfold increase relative to baseline (Table 1). This is consistent with prior studies indicating that persistence is a reversible, non-genetic state that does not confer heritable resistance upon resuscitation^7,26.

To determine whether MIC changes occurred transiently during cycling, MIC was measured at each stage. R maintained an MIC of 8 μg/mL across P1, R1, R2, and R3, arguing against transient MIC elevation during persister cycling (Table S2b). A maintained an MIC of 8 μg/mL across P1, R1, and R2, but increased to 16 μg/mL at R3 (Table S2b). Stationary-phase cells and NaCl-stored persister cells also maintained MICs of 8 μg/mL at 0, 2, and 4 days, indicating no clear resistance acquisition during prolonged physiological arrest or storage (Table S2b). In M, the bulk culture MIC was 16 μg/mL, whereas single resistant colonies showed an MIC of 32 μg/mL, consistent with population heterogeneity and clonal enrichment under repeated selection (Table 1; Table S2a).

To further evaluate antibiotic susceptibility at the phenotypic level, cells from each group were collected at each stage, resuscitated in fresh medium to an OD₆₀₀ of 0.8, and then treated with ampicillin to generate time-kill curves (Figure. S1). R showed a killing profile comparable to exponential-phase cells, supporting maintained baseline susceptibility. A showed an increased MIC relative to R, but its time-kill profile remained comparable to that of exponential-phase cells. By contrast, M showed slower killing and higher residual OD₆₀₀, consistent with its increased MIC outcome. Stationary-phase cells showed a killing slope comparable to exponential-phase cells, making GASP-associated resistance unlikely under these conditions. NaCl-stored persister cells showed a delayed killing profile, likely reflecting asynchronous resuscitation kinetics; the biological basis of these physiological states is discussed further in the Discussion (Table S2b; Figure. S1).

Overall, persister cycling without selection maintained baseline MIC and antibiotic susceptibility, whereas sub-MIC Amp exposure during persister-derived regrowth was associated with an MIC increase to 16 μg/mL, and repeated mutation-induced selection was associated with an MIC increase to 32 μg/mL. Thus, R, A, and M showed distinct phenotypic outcomes: maintained susceptibility, MIC elevation during persister-derived regrowth, and stronger MIC elevation under mutation-induced selection, respectively (Fig. 1b, c; Table 1).

Quality-filtered WGS detects fixed variants only under mutation-induced selection

To determine whether the phenotypic differences described above were associated with heritable genomic changes, we performed WGS of the five treated states together with exponential-phase controls. Variants were called using a diploid variant-calling model and then analyzed after quality-based filtering (QD ≥ 2, FS < 60, SOR < 3). Fixed variants (AF = 1) were detected only in mutation-induced cells (M), including variants in acrB, acrR, marR, and xdhD (Fig. 2a–c). Because marR and acrR repress MarA and the AcrAB-TolC efflux system, frameshifts in these regulators likely derepress the efflux system, thereby enhancing antibiotic resistance^28,29, consistent with the MIC increase observed in M. A single dosC mutation (AF = 1) was detected in one exponential-phase replicate, consistent with a background variant related to c-di-GMP signaling³⁰ (Fig. 2a, d). After quality filtering, no fixed or high-frequency variants were detected in R, A, P, or S.

No variants passed the quality-filtering thresholds in persister-cycling cells (R) or persister-Amp cycling cells (A) (Fig. 2a). Notably, although A showed an MIC increase to 16 μg/mL, no detectable quality-filtered genetic fixation was observed. Thus, the MIC increase in A was not accompanied by detectable quality-filtered genetic fixation, distinguishing it from the fixed resistance observed in M. The absence of quality-filtered variants in R was consistent with the baseline MIC and antibiotic susceptibility profile observed in the phenotypic analyses above. This pattern is consistent with prior reports showing that persister cycling without selection does not generate heritable mutations^31,32.

Among quality-filtered minor variants (AF < 0.5), stfE and BW25113_RS25865 were detected in stress-exposed states (M, P, and S), but not in E, R, or A (Fig. 2a, b). In exponential-phase cells, only a single pbpC variant was detected, distinguishing this profile from the stfE and BW25113_RS25865 variant pattern observed in stress-exposed states. This stress-associated insertion pattern is examined further in the following section.

Thus, the quality-filtered variant profiles differed by cellular state. Fixed, function-altering mutations were detected in M, stress-associated minor variants were detected in M, P, and S, and no variants passed the quality-filtering thresholds in R or A (Fig. 2a, b). These results show that mutation-induced selection was associated with fixed resistance-associated variants, whereas persister cycling and persister-Amp cycling showed no detectable genetic fixation by quality-filtered WGS.

Prophage-linked insertions are associated with stress-exposed states

Quality-filtered variants were annotated using SnpEff (v4.3t)³³ and mapped to functional classes using EcoCyc, EggNOG, and the Comprehensive Antibiotic Resistance Database^34–36. Because no quality-filtered variants were detected in persister-cycling cells (R) or persister-Amp cycling cells (A), these two groups were not included in the functional-category analysis (Fig. 2e). The remaining groups showed state-dependent functional profiles.

Mutation-induced cells (M) showed proportions of 29.31% for energy metabolism-related genes, 5.19% for antibiotic resistance-related genes, and 21.82% for prophage-related functions (Fig. 2e; Table S3). NaCl-stored persisters (P) showed 29.73% energy/metabolism-associated genes and 13.41% prophage-related functions, with no detectable resistance-related genes. Stationary-phase cells (S) showed the highest proportion of prophage-associated variants among the analyzed groups (32.35%). In exponential-phase cells (E), the resistance-related category signal was attributable to the single dosC variant detected in this group (Fig. 2e; Table S3).

WGS identified prophage-linked insertional events at two loci. First, a stfP-derived fragment was inserted into stfE, resulting in frameshift variants; two independent insertion variants were detected in a subset of M, P, and S samples (Fig. 3a). In contrast, these variants were not detected in E, R, or A in the quality-filtered variant set. However, in the exploratory unfiltered analysis, sub-threshold stfE variant calls were also observed in R. These calls did not pass the quality-filtering thresholds and were therefore excluded from the quality-filtered variant set; these sub-threshold signals are analyzed separately in the section on low-frequency variant dynamics. Second, prophage-derived fragments from DLP12 and CP4-6 were inserted into BW25113_RS25865. Three independent insertion variants were detected in a subset of M, P, and S samples and were likewise absent from E, R, and A in the quality-filtered variant set (Fig. 3b).

Fig. 3 — a Frameshift variants from sequence insertion in *stfE*. Inserted sequences are labelled as inserted; substituted sequences are shown above the original and struck through; boxes marked with an × indicate deleted sequences. Each numbered entry represents an independent variant event as called by GATK HaplotypeCaller⁵⁸. b Frameshift variants from sequence insertion in *BW25113_RS25865*. Notation follows the same scheme as in (a).

These patterns suggest that prophage-linked insertions are associated with stress-exposed cellular states and may represent a state-dependent source of genomic variation.

The stfE/stfP background alters survival, MIC outcomes, and mutational trajectories under mutation-induced selection

To evaluate whether the stfE/stfP prophage background influences resistance-associated outcomes, we compared mutation-induced and persister-cycling cells in WT, ΔstfE, and ΔstfP backgrounds. Total survival and large-colony counts were measured on LB agar containing 8 μg/mL Amp (Fig. 4; Table S4).

Fig. 4 — a Representative colony morphologies under antibiotic stress. WT, Δ*stfE*, and Δ*stfP* strains were grown on LB agar containing Amp (8 µg/mL) at each stage of mutation-induced selection and persister cycling. Large colonies were defined as those with a diameter ≥ 2.5 mm (yellow arrows). b Quantification of total CFU and large-colony counts. Mutation-induced cells or persister cycling cells were plated on LB agar containing Amp (8 µg/mL); large-sized colonies (red) and total colonies (blue) are shown for each genotype (See Table S4). Data are presented as mean ± s.d. Biological replicates were as follows: WT, n = 4; Δ*stfE* and Δ*stfP*, n = 3. Differences were assessed by two-way ANOVA within each stage, followed by Tukey’s post-hoc test (P < 0.05); stages labeled with a were significantly different from the baseline, corresponding to exponential-phase cells (mutation induced cell stage 0) within the same genotype, and stages labeled with b were significantly different from WT at the same stage.

Under mutation-induced selection, WT survival increased progressively across cycles (3.4 × 10² → 1.5 × 10³ CFU/mL), accompanied by increased large-colony counts (Fig. 4; Table S4a). ΔstfE showed lower initial survival than WT (4.5 × 10¹ CFU/mL; 13% of WT at stage 0), but survival increased after repeated sub-MIC Amp exposure (7.0 × 10¹ → 2.2 × 10² CFU/mL) (Fig. 4; Table S4b). ΔstfP showed higher initial survival than ΔstfE (4.2 × 10² CFU/mL; comparable to the WT level at stage 0), but survival declined across cycles to 7.3 × 10¹ CFU/mL (Fig. 4; Table S4c).

In persister-cycling conditions, ΔstfE and ΔstfP formed persisters at levels comparable to WT (3.9 × 10⁸ ± 7.1 × 10⁶ and 4.1 × 10⁸ ± 5.9 × 10⁷ vs. WT 5.2 × 10⁸ ± 1.2 × 10⁸ CFU/mL) (Table S4d). However, survival after resuscitation and Amp challenge was reduced in both deletion backgrounds. ΔstfE decreased from 3.0 × 10¹ to 0 CFU/mL, with no large colonies detected. ΔstfP also showed reduced survival (5.7 × 10¹ → 1.0 × 10¹ CFU/mL), without large-colony formation. In contrast, WT maintained detectable survival across cycles (6.5 × 10¹ → 1.5 × 10¹ CFU/mL), with large colonies observed at low frequency (Fig. 4; Table S4a–c).

MIC outcomes further distinguished the genotypes. In exponential-phase and persister-cycling states, WT, ΔstfE, and ΔstfP all maintained an MIC of 8 μg/mL (Table S2a). Under mutation-induced selection, WT reached 32 μg/mL, whereas ΔstfE and ΔstfP reached 16 μg/mL (Table S2a). Plasmid complementation of stfE or stfP restored survival and large-colony formation, whereas MIC remained at 16 μg/mL (Fig. S2; Table S2a, Table S4e–i).

These results indicate that deletion of stfE or stfP alters survival after resuscitation and Amp challenge, as well as MIC outcomes under mutation-induced selection.

To determine whether the altered survival and MIC outcomes observed in ΔstfE and ΔstfP were associated with genotype-specific mutational trajectories, one large colony from each of three independent parallel cultures recovered after repeated exposure to 5 μg/mL Amp was subjected to WGS (Fig. 5). High-frequency variants (AF = 1) differed among WT, ΔstfE, and ΔstfP backgrounds, with limited overlap in fixed variants across the three genotypes (Fig. 5d).

Fig. 5 — a Variant loci and allele-frequency spectra in mutation-induced Δ*stfE* and Δ*stfP* cells. Allele frequencies (AF) are shown for three biological replicates (n = 3). For each genotype, one colony from each of three independent parallel cultures was selected after the final sub-MIC Amp exposure cycle and subjected to WGS. AF was calculated as alternate-allele depth divided by total read depth using the diploid setting of GATK HaplotypeCaller⁵⁸. Intergenic variants are denoted with a slash (/) between adjacent genes. b Fixed mutations in mutation-induced Δ*stfE* cells. Amino-acid changes are shown for WT and Δ*stfE* mutation-induced cells; yellow highlights mark variant residues. Asterisks (*) indicate stop codons; dashes (–) indicate absence of corresponding sequences. c Fixed mutations in mutation-induced Δ*stfP* cells. Amino-acid changes are shown for WT and Δ*stfP* mutation-induced cells with yellow highlights marking variant residues; asterisks (*) indicate stop codons and dashes (–) indicate absence of corresponding sequences. For intergenic variants with AF = 1, the altered nucleotide sequence is shown. d Shared fixed mutations across mutation-induced WT, Δ*stfE*, and Δ*stfP* cells. e Functional classification of variant-affected genes in mutation-induced WT, Δ*stfE*, and Δ*stfP* cells. Genes predicted to be functionally affected by fixed or minor mutations, as annotated by SNPEff³³, were classified into functional categories. Bar heights represent the proportion of variant-affected genes per functional category relative to the total variant-affected genes within each genotype, averaged across biological replicates in which variants were detected. Category-level summaries are provided in Table S5.

In WT mutation-induced cells, fixed variants included frameshifts in marR and acrR, a missense mutation in acrB; a minor mutation corresponding to the stfP→stfE insertion identified in the analysis of prophage-linked insertions was also detected (Fig. 2c and Fig. 3). In ΔstfE, a marR frameshift was detected as the major fixed mutation in two of three colonies, whereas acrR, acrB, and stfP→stfE events were not detected (Fig. 5a, b, d). In ΔstfP, fixed variants included a cyoB stop-gain mutation, an rpoD missense mutation, and intergenic variants between kgtP–rrfG and BW25113_RS22180–BW25113_RS23025; these variants were not detected in WT or ΔstfE (Fig. 5a, c, d).

Functional profiles of variant-affected genes also differed by genotype. WT was enriched for energy metabolism (29.31%), prophage-associated functions (21.82%), and antibiotic resistance-related functions (5.19%). ΔstfE showed enrichment in stress response (25.16%) and energy metabolism (34.97%), with low prophage representation (0.98%). ΔstfP showed the highest proportions of cell growth-related functions (33.33%) and energy metabolism (38.24%), with lower prophage-associated (2.94%) and antibiotic resistance-related (1.96%) representation (Fig. 5e; Table S5).

Taken together, WT, ΔstfE, and ΔstfP followed distinct mutational trajectories under mutation-induced selection, with limited overlap in fixed variants. The efflux-associated route observed in WT was not detected in either deletion background. These findings indicate that the stfE/stfP prophage background shapes adaptive route selection under mutation-induced selection, with WT favoring an efflux-associated trajectory and ΔstfE and ΔstfP following alternative routes.

Persister-associated low-frequency variants are stochastically retained or lost during cycling

The MIC and quality-filtered WGS analyses showed that no variants passed quality-filtering thresholds in persister-cycling cells. To examine whether low-frequency variant signals were present below the filtering threshold and whether such variants were fixed in a stage-specific manner during persister formation and resuscitation, we performed an exploratory analysis using unfiltered variant calls and single-colony lineage tracking.

In the exploratory analysis without quality filtering, low-frequency variant signals were detected across multiple cellular states; group-level details are provided in Fig. S3a. Variants in yfgM, yjfM, pcnB, and rlmF were detected across M, P, S, E, and R. This pattern suggests that these variant signals were not specific to persister cycling, but instead represented low-frequency signals observed across multiple cellular states. Group-associated variants included sbcB in R, nrfA and nrfB in A, and stfP in S (Fig. S3a).

Because these variants did not pass the quality-filtering thresholds, they were treated as exploratory sub-threshold signals rather than quality-filtered evidence of genetic fixation.

To explore the functional relevance of genes carrying recurrent or commonly detected sub-threshold variants, we characterized deletion mutants for growth, antibiotic response, and resuscitation kinetics (Fig. S3b). The tested mutants included ΔyfgM, ΔyjfM, ΔypdE, ΔrlmF, ΔydjY, and ΔpcnB. These mutants were selected because the corresponding genes were recurrently detected as sub-threshold variant signals across multiple cellular states and represented known or putative functions related to stress survival or regrowth. These functions included Sec-associated protein translocation and periplasmic folding (yfgM), predicted envelope-stress adaptation (yjfM), predicted c-di-GMP phosphodiesterase activity (ypdE), 23S rRNA modification and ribosome assembly (rlmF), poorly characterized inner-membrane function (ydjY), and poly(A)-dependent RNA turnover (pcnB).

In growth assays, ΔypdE, ΔyfgM, ΔpcnB, ΔrlmF, and ΔydjY grew faster than WT in both LB and M9 media (Fig. S3b). Under ampicillin stress (100 μg/mL), ΔyfgM, ΔyjfM, ΔydjY, ΔrlmF, and ΔypdE showed slower OD₆₀₀ decline than WT (Fig. S3b). During resuscitation, ΔyfgM, ΔyjfM, ΔypdE, ΔmetE, and ΔstfE resumed growth comparably to WT (Fig. S3b). In repeated cycling, variants consistently detected after resuscitation were limited to yfgM, yjfM, and stfE, whereas other variant signals decreased below detection (Fig. S3a).

These results indicate that several sub-threshold variant-target genes are associated with measurable growth, antibiotic-response, or resuscitation phenotypes. However, these deletion-mutant phenotypes do not by themselves indicate genetic fixation during persister cycling.

To determine whether sub-threshold variants observed in the exploratory analysis accumulated in a stage-specific manner during persister formation and resuscitation, we performed serial single-colony lineage tracking across P1 → R1 → P2 → R2 (Fig. S4a). Two independent lineages were initiated from single colonies, and two single-colony isolates were sequenced at each stage.

No consistent stage-specific bias in variant emergence or loss was observed (Fig. S4b). fimA and glnH were maintained across all stages, whereas ykgH, aceK, and smrB appeared and disappeared across stages without a consistent persister- or resuscitation-specific pattern (Fig. S4b). Thus, the variants observed in single-colony lineages were maintained or lost without evidence of stage-specific fixation. This stochastic pattern is consistent with population bottleneck effects and outgrowth competition during resuscitation-phase regrowth^37,38.

Overall, the exploratory analysis revealed sub-threshold low-frequency variant signals during persister cycling, and several variant-target genes showed altered phenotypes in deletion-mutant assays. However, single-colony lineage tracking did not support stage-specific fixation of these variants during persister cycling. Together, these results indicate that persister cycling maintains phenotypic susceptibility and lacks detectable quality-filtered fixation, while sub-threshold variants show stochastic rather than stage-driven dynamics. In contrast, mutation-induced selection was associated with fixed resistance-associated variants and MIC elevation, with the stfE/stfP prophage background shaping which adaptive routes emerged under selection.

Discussion

Persister cells have generally been understood as antibiotic-surviving cells that retain genetic susceptibility after antibiotic removal. Although persisters were historically described as non-dividing dormant survivors, recent studies indicate that persistence can involve dynamic physiological activity and heterogeneous survival states rather than strict dormancy^5,6. Several studies have suggested links between persistence and the evolution of resistance^11,27,39. However, whether repeated persister cycling itself promotes the evolution of fixed resistance, or whether selective context during regrowth drives variant retention and fixation, remains unclear.

To address this question, we compared five cellular states that differed in physiological state and selective context (Table 1). Mutation-induced cells (M) carried fixed resistance-associated variants in acrB, acrR, marR, and xdhD, with MIC increasing to 32 μg/mL. This is consistent with AcrAB-TolC efflux derepression through loss of marR/acrR function²⁸. In contrast, persister cycling without selection (R) maintained an MIC of 8 μg/mL and showed no detectable genetic fixation by quality-filtered WGS. Stage-wise MIC measurements across P1, R1, R2, and R3 excluded transient MIC elevation during cycling, and time-kill curves confirmed baseline antibiotic susceptibility phenotypically (Table S2b; Fig. S1). Persister-Amp cycling (A) showed an MIC increase to 16 μg/mL, but this was not accompanied by detectable quality-filtered genetic fixation, distinguishing it from the fixed resistance observed in M. Therefore, the MIC increase in A should be interpreted cautiously and may reflect non-fixed adaptation, heteroresistance, or variation below the WGS detection threshold⁴⁰. These findings indicate that adaptive outcomes in persister-derived populations are closely linked to selective context during regrowth rather than to the persister state alone.

To understand the meaning of low-frequency variant signals observed during persister cycling, we performed an exploratory analysis of variants below the quality-filtering threshold. Several recurrent sub-threshold variant-target genes mapped to cellular processes relevant to stress survival and regrowth. For example, yfgM and yjfM may be linked to membrane-associated or poorly characterized envelope-related functions, rlmF is involved in ribosome modification, and pcnB is involved in RNA turnover through poly(A)-dependent regulation of mRNA stability^34,41–43. Consistent with these functional contexts, deletion mutants of variant-target genes showed altered growth, antibiotic-response, or resuscitation phenotypes (Fig. S3b). However, single-colony lineage tracking revealed no consistent stage-specific bias in variant emergence or loss, with variants being retained or lost across lineages and stages (Fig. S4). Thus, although these minor variants may have functional relevance, they are not automatic precursors of fixed resistance during persister cycling.

Based on these results, we propose that resuscitation-phase population dynamics may shape the fate of persister-associated minor variants. Population bottlenecks and outgrowth competition during resuscitation may contribute to stochastic retention or loss of low-frequency variants, limiting reproducible fixation in the absence of antibiotic selection^37,38. However, because only a limited number of single-colony isolates were analyzed at each stage, the persistence or loss of rare variants may have been under- or over-estimated⁴⁴. For stationary-phase cells (S) and NaCl-stored persisters (P), stage-wise MIC measurements and time-kill curves provided evidence against detectable resistance acquisition during prolonged physiological arrest or storage (Fig. S1; Table S2b).

Prophage-linked insertion events were also observed in stress-exposed states. A stfP-derived fragment inserted into stfE, and DLP12/CP4-6-derived fragments inserted into BW25113_RS25865, were detected as quality-filtered minor variants in M, P, and S, but not in E, R, or A (Fig. 3a, b). stfE and stfP reside within the e14 cryptic prophage, which has been associated with SOS-induced excision, stress resistance, and biofilm formation in E. coli K-12^20,45. Under mutation-induced selection, WT followed an efflux-associated trajectory involving marR, acrR, and acrB, reaching an MIC of 32 μg/mL; this route is consistent with the established role of AcrAB-TolC in resistance, whose biased partitioning during cell division can sustain long-lived phenotypic heterogeneity in efflux activity, and whose transcriptional regulation via the mar operon can be maintained as an inducible memory across lineages under repeated selection^28,29,46. By contrast, ΔstfE and ΔstfP followed distinct trajectories and reached MICs of 16 μg/mL. Complementation restored survival and large-colony formation but did not increase the MIC beyond 16 μg/mL, suggesting that the complementation effect was more closely associated with survival or tolerance-associated phenotypes than with resistance gain²⁶. The mechanistic basis for these divergent trajectories remains to be determined, but may be related to the location of stfE and stfP within the e14 prophage network⁴⁵. In WT, the stfP→stfE insertion may contribute to a prophage-linked background in which the efflux-associated marR/acrR/acrB route is favored^28,29,46. In contrast, ΔstfE and ΔstfP lacked this insertional event and followed alternative routes, including a more limited marR-centered trajectory in ΔstfE and bioenergetic/transcriptional variants such as cyoB and rpoD in ΔstfP. Thus, the stfE/stfP background may influence adaptive route accessibility under mutation-induced selection, rather than acting as a direct resistance determinant.

Collectively, this study shows that persister-associated minor variants are not automatic precursors of resistance; rather, their retention and fixation depend on selective context during regrowth and population dynamics during resuscitation. Persister cycling without selection maintained phenotypic susceptibility and showed no detectable genetic fixation by quality-filtered WGS, whereas sub-threshold variants showed stochastic retention or loss rather than stage-specific fixation. In contrast, mutation-induced selection was associated with fixed resistance-associated variants and MIC elevation, and the stfE/stfP prophage background shaped adaptive routes under selection. This study is limited by the use of a single antibiotic and strain, a limited number of cycling rounds, an in vitro setting, and limited single-colony sampling. Future work using broader antibiotic classes, diverse strain backgrounds, longer cycling experiments, and in vivo models will be needed to test the generality of this framework.

Methods

Bacteria and growth conditions

Escherichia coli BW25113 was cultured in LB broth⁴⁷ at 37 °C with shaking at 200 rpm. The mutant strains used in this study were obtained from the Keio and ASKA collection^48,49. The bacterial strains used in this study are listed in Table S6.

Mutation-induced cells

Given that E. coli BW25113 WT cells exhibited robust growth at 4 μg/mL and had an MIC of 8 μg/mL against ampicillin (Amp) (Table S2), an intermediate concentration of Amp (5 µg/mL) was applied to induce mutations^50,51. Exponential-phase cells (9 × 10⁸ CFU/mL) were treated with a sub-MIC concentration of Amp (5 μg /mL), followed by incubation for 30 min (one doubling time) to induce mutations. The Amp-exposed cells (1.1 × 10⁹ CFU/mL at the first stage) were then washed twice with 0.85% NaCl to remove residual Amp, resuspended in fresh LB, and the turbidity Optical density at 600 nm (OD₆₀₀) was adjusted to 0.8 (same as exponential phase). Sub-MIC of Amp was treated for 30 min again. This process was repeated three times using six biological replicates, and the total stages are shown in Fig. 1a. Viable cell counts at each stage were determined by plating serial dilutions on antibiotic-free LB agar and are provided in Table S7. In the mutation induction period, 100 µL of the cultures (approximately 1.1 × 10⁸ CFU/mL at the first stage) were plated onto LB agar supplemented with 8 µg/mL Amp (corresponding to the MIC) and were incubated at 37 °C for 16 h to allow for the growth of resistant colonies.

Large colonies were defined as colonies with a diameter > 2.5 mm and were counted as mutation-induced colonies^26,27. At the final stage of Amp exposure, one large colony from each of three independent parallel cultures was selected and transferred onto LB agar plates containing a higher Amp concentration (10 µg/mL) to verify growth under increased Amp exposure. Colonies that successfully grew on these plates were subsequently cultured in fresh antibiotic-free LB medium and used for further analyses, including MIC assays and DNA extraction for WGS.

Persister-cycling cells

Persister cells were converted from exponential-phase cells (5.3 × 10⁸ CFU/mL) by adding rifampicin (Rif; 100 µg/mL) for 30 min, centrifuging, and treating with Amp (100 µg/mL) for 3 h to remove non-persister cells⁵². This method of Rif-induced persister formation has been validated across multiple independent laboratories, supporting its reproducibility and reliability^53,54.

Persister cells (1.5 × 10⁸ CFU/mL at the first stage) were washed twice with 0.85% NaCl to remove residual Amp and resuscitated by resuspending in fresh LB medium and incubating at 37 °C with shaking until the culture reached an OD₆₀₀ of 0.8. The formation and resuscitation of persister cells were repeated three times, and plating was performed using six biological replicates (Fig. 1a). Viable cell counts at each stage were determined by plating serial dilutions on antibiotic-free LB agar and are provided in Table S7. From each repetition, 100 µL of cells (approximately 6.9 × 10⁷ CFU/mL for the persister-cycling cells of the last stage) were spread on LB agar containing 8 µg/mL Amp, and large-sized colonies (diameter > 2.5 mm) were counted. Two independent bulk cultures of cells resuscitated at the third stage were used for further analyses as persister-cycling cells.

Persister-Amp cycling cells

For persister-Amp cycling cells, persister cells were converted from exponential-phase cells and resuscitated to an OD₆₀₀ of 0.8 using the same procedure described above. The resuscitated persister cells were then treated with Amp (5 µg/mL) for 30 min under the same conditions used for mutation induction. After washing twice with 0.85% NaCl, the cells were converted to persister cells again by treatment with Rif and Amp. The cycle of persister formation, resuscitation, and Amp treatment (5 µg/mL) was repeated three times, and plating was performed using three biological replicates. Viable cell counts at each stage were determined by plating serial dilutions on antibiotic-free LB agar and are provided in Table S7. As a result, 2.00 × 10⁸ CFU/mL of persister-Amp cycling cells were obtained at the final stage (Fig. 1a). From each repetition, 100 µL of cells (approximately 2.0 × 10⁷ CFU/mL for the persister cells of the first stage) were spread on LB agar containing 8 µg/mL Amp, and large-sized colonies (diameter > 2.5 mm) were counted. Three independent bulk cultures of cells resuscitated and exposed to Amp at the third stage were used for further analyses as persister-Amp cycling cells.

Stationary-phase cells and NaCl-stored persister cells

Started at exponential phase (7.3 × 10⁸ CFU/mL), cells were incubated at 37 °C with shaking for 4 days to reach stationary phase, and stationary-phase cells (1.7 × 10⁷ CFU/mL) were harvested (Fig. 1a). At the same time, persister cells (2.8 × 10⁸ CFU/mL) were generated as described above, stored in 0.85% NaCl at 37 °C with shaking for 4 days, and NaCl-stored persister cells (1.3 × 10⁷ CFU/mL) were harvested. To confirm that GASP-associated resistance enrichment did not occur during this period, MIC was measured at days 0, 2, and 4, and time-kill curves were performed on day-4 cells (Table S2b; Fig. S1). Plating was performed using four biological replicates. Viable cell counts during the 4-day incubation or storage period were determined by plating serial dilutions on antibiotic-free LB agar and are provided in Table S7. Each 100 µL of these cultures (1.7 × 10⁶ CFU/mL for stationary-phase cells and 1.3 × 10⁶ CFU/mL for NaCl-stored persister cells) was plated on LB agar containing 8 µg/mL Amp, and large-sized colonies (diameter > 2.5 mm) were counted. Two independent bulk cultures of stationary-phase cells and three independent bulk cultures of NaCl-stored persister cells were used for further analyses.

Minimum inhibitory concentration (MIC) assay

To determine the MIC of Amp, exponential-phase cells (7.0 × 10⁸ CFU/mL) were diluted (1:1000) with LB broth containing different concentrations of Amp and grown for 24 h. In the case of cells in different physiological states, the inoculum was adjusted to the same OD₆₀₀ with exponential-phase cells. The MIC was determined as the lowest concentration that prevented an increase in growth, as evidenced by the lack of change in turbidity⁵⁵.

Ampicillin susceptibility assay

Cells in different physiological states were adjusted to an OD₆₀₀ of 0.8 (approximately 10⁸ CFU/mL). Each culture was then inoculated into 96-well plates at 1% (v/v) in LB medium. After incubation until reaching the exponential phase, Amp (100 µg/mL) was added. Growth and Death was monitored at regular time intervals throughout the experiment by measuring OD₆₀₀ using BioTek Synergy HTX microplate reader.

Growth and survival assay of specific gene-deletion mutants

WT and mutant strains obtained from the KEIO library⁴⁸ were cultured in LB or LB supplemented with kanamycin (50 µg/mL) until reaching the exponential phase (OD₆₀₀ reaches 0.8). For growth analysis, exponential-phase WT and mutant cells were inoculated into 96-well plates at 1% (v/v) in LB or M9 minimal medium (42 mM Na₂HPO₄, 22 mM KH₂PO₄, 8.6 mM NaCl, 18.7 mM NH₄Cl) supplemented with 0.4% (w/v) glucose. For ampicillin killing analysis, exponential-phase cells were harvested by centrifugation and resuspended in LB or M9/0.4% glucose media containing 100 µg/mL ampicillin. For resuscitation analysis, natural persisters (ampicillin-induced persisters)^31,56 were generated by treating cells with ampicillin (100 µg/mL) for 3 h, followed by washing twice with 0.85% NaCl. The OD₆₀₀ of natural persisters adjusted to the same OD₆₀₀ with exponential phase and inoculated into 96-well plates at 1% (v/v) in LB or M9/0.4% glucose media. Cell growth, death, and resuscitation was monitored by measuring OD₆₀₀ over time using a BioTek Synergy HTX microplate reader.

Single-colony tracking of persister cycling and resuscitation

Two independent exponential-phase cultures (8.1 × 10⁸ CFU/mL) were harvested and subjected to Rif (100 µg/mL, 30 min) followed by Amp (100 µg/mL, 3 h) to eliminate non-persister cells. Cells were washed twice with 0.85% NaCl, diluted to 10⁻³ (8.1 × 10⁵ CFU/mL), plated on LB agar, and incubated at 37 °C for 16 h to resuscitate. From each plate, two single colonies were randomly picked and grown in LB to 7.33 × 10⁸ CFU/mL; these constituted the first-cycle resuscitated persister cultures. The same Rif-Amp treatment, washing, dilution, plating, and resuscitation steps were repeated once more to generate second-cycle persister and resuscitated cultures. Genomic DNA was extracted from two independent bulk persister preparations and from two independent resuscitated cultures (each represented by two replicate single colonies). The corresponding procedures are shown in Fig. S4a.

DNA preparation

All DNA was extracted using an Aprep Total DNA Mini kit (APBio, Korea), and more than two independent replicates were performed. DNA samples were quantified using a BioTek Synergy HTX TAKE3 (Agilent Technologies, Santa Clara, CA, USA).

WGS

The quality and quantity of purified DNA were assessed using fluorometry (Qubit; Invitrogen, Carlsbad, CA, USA) and gel electrophoresis. Briefly, 100 ng of genomic DNA from each sample was fragmented through acoustic shearing using a Qsonica 800 R2 instrument (Qsonica LLC, Newtown, CT, USA). Fragments of 350 bp were ligated to Illumina adapters and then amplified using PCR. An appropriate size for the final library is 500–600 bp. Libraries were quantified using a TapeStation 4200 instrument (Agilent Technologies) and a KAPA Library Quantification Kit (KK4824; Kapa Biosystems, Wilmington, MA, USA). Pair-end libraries of 150 bp were generated and sequenced on an Illumina Novaseq 6000 platform (Illumina, San Diego, CA, USA). Reads were mapped to the E. coli BW25113 reference genome (NZ_CP009273.1) using BWA 0.7.17⁵⁷. Duplicate reads were removed using GATK Mark Duplicates⁵⁸. The raw fastq sequence was deposited in the Sequence Read Archive (accession no: PRJNA1092394).

Variant calling and analysis

Variants were called using GATK HaplotypeCaller⁵⁸ and annotated using SnpEff³³. Variant allele frequency (AF) was calculated for all called variants based on allele depth. Variant calls were then filtered according to GATK hard-filtering recommendations with minor modifications: QD ≥ 2, FS < 60, SOR < 3 for SNPs, and SOR < 10 for indels⁵⁸. Filtered variants with AF = 1 were defined as fixed mutations, whereas filtered variants with AF ≤ 0.5 were defined as minor variants. Unfiltered variants with AF ≤ 0.5 were not used to define fixed mutations but were considered as exploratory low-confidence minor signals. Gene functions were annotated using EcoCyc³⁴ and eggNOG³⁵, antibiotic-resistance genes were identified using the Comprehensive Antibiotic Resistance Database³⁶, and tolerance-associated genes were classified based on previously reported tolerance mechanisms²⁷.

Statistical analysis

All data are the mean ± standard deviation. Statistical analysis was performed by two-way analysis of variance (ANOVA) in GraphPad Prism software version 10.6.0 (San Diego, CA, USA). a represents a significant difference between viable cell counts with exponential cells and mutation-induced cells or persister cycling cells (P < 0.05). b represents significant differences between viable cell counts of WT and mutants (P < 0.05).

Supplementary information

Supporting information_053126 final^{(4.4MB, docx)}

Acknowledgements

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (NRF-2020R1F1A1072397; RS-2023-00210305) for SYS. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author contributions

S.S. conceived and designed the study, supervised the project, analyzed and interpreted the data, and wrote the manuscript. G.P. and H.K. performed the experiments, contributed to data analysis and interpretation, and co-wrote the manuscript. All authors read and approved the final manuscript.

Data Availability

Raw FASTQ sequencing data have been deposited in the NCBI Sequence Read Archive under BioProject PRJNA1092394. All other data supporting the findings of this study are included in the article and its Supplementary Information files.

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s44259-026-00261-0.

References

1.Rossolini, G. M., Arena, F., Pecile, P. & Pollini, S. Update on the antibiotic resistance crisis. Curr. Opin. Pharm.18, 56–60 (2014). [DOI] [PubMed] [Google Scholar]
2.Christaki, E., Marcou, M. & Tofarides, A. Antimicrobial Resistance in Bacteria: Mechanisms, Evolution, and Persistence. J. Mol. Evol.88, 26–40 (2020). [DOI] [PubMed] [Google Scholar]
3.Shah, D. et al. Persisters: a distinct physiological state of E. coli. BMC Microbiol.6, 53 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Lewis, K. Multidrug tolerance of biofilms and persister cells. Curr. Top. Microbiol. Immunol.322, 107–131 (2008). [DOI] [PubMed] [Google Scholar]
5.Hofsteenge, N., van Nimwegen, E. & Silander, O. K. Quantitative analysis of persister fractions suggests different mechanisms of formation among environmental isolates of E. coli. BMC Microbiol.13, 25 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Wakamoto, Y. et al. Dynamic persistence of antibiotic-stressed mycobacteria. Science339, 91–95 (2013). [DOI] [PubMed] [Google Scholar]
7.Balaban, N. Q. et al. Definitions and guidelines for research on antibiotic persistence. Nat. Rev. Microbiol.17, 441–448 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Korch, S. B., Henderson, T. A. & Hill, T. M. Characterization of the hipA7 allele of Escherichia coli and evidence that high persistence is governed by (p)ppGpp synthesis. Mol. Microbiol50, 1199–1213 (2003). [DOI] [PubMed] [Google Scholar]
9.Korch, S. B. & Hill, T. M. Ectopic overexpression of wild-type and mutant hipA genes in Escherichia coli: effects on macromolecular synthesis and persister formation. J. Bacteriol.188, 3826–3836 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Cohen, N. R., Lobritz, M. A. & Collins, J. J. Microbial persistence and the road to drug resistance. Cell Host Microbe13, 632–642 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Windels, E. M. et al. Bacterial persistence promotes the evolution of antibiotic resistance by increasing survival and mutation rates. ISME J.13, 1239–1251 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Sebastian, J. et al. De novo emergence of genetically resistant mutants of mycobacterium tuberculosis from the persistence phase cells formed against antituberculosis drugs in vitro. Antimicrob. Agents Chemother. 61, 10.1128/aac.01343-16 (2017). [DOI] [PMC free article] [PubMed]
13.Wilmaerts, D., Windels, E. M., Verstraeten, N. & Michiels, J. General mechanisms leading to persister formation and awakening. Trends Genet.35, 401–411 (2019). [DOI] [PubMed] [Google Scholar]
14.Salini, S. et al. The error-prone polymerase DnaE2 mediates the evolution of antibiotic resistance in persister Mycobacterial cells. Antimicrob. Agents Chemother.66, e0177321 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Swaminath, S. et al. Mycobacterium smegmatis moxifloxacin persister cells produce high levels of hydroxyl radical, generating genetic resisters selectable not only with moxifloxacin, but also with ethambutol and isoniazid. Microbiol. (Read.)166, 180–198 (2020). [DOI] [PubMed] [Google Scholar]
16.Van den Bergh, B. et al. Frequency of antibiotic application drives rapid evolutionary adaptation of Escherichia coli persistence. Nat. Microbiol.1, 16020 (2016). [DOI] [PubMed] [Google Scholar]
17.Vogwill, T., Comfort, A. C., Furió, V. & MacLean, R. C. Persistence and resistance as complementary bacterial adaptations to antibiotics. J. Evol. Biol.29, 1223–1233 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Bakkeren, E., Diard, M. & Hardt, W. D. Evolutionary causes and consequences of bacterial antibiotic persistence. Nat. Rev. Microbiol18, 479–490 (2020). [DOI] [PubMed] [Google Scholar]
19.Ueno, H., Kato, Y., Tabata, K. V. & Noji, H. Revealing the metabolic activity of persisters in Mycobacteria by Single-Cell D(2)O Raman imaging spectroscopy. Anal. Chem.91, 15171–15178 (2019). [DOI] [PubMed] [Google Scholar]
20.Wang, X. et al. Cryptic prophages help bacteria cope with adverse environments. Nat. Commun.1, 147 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Bergum, O. E. T. et al. SOS genes are rapidly induced while translesion synthesis polymerase activity is temporally regulated. Front. Microbiol.15, 2024 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Hoeksema, M., Jonker, M. J., Bel, K., Brul, S. & ter Kuile, B. H. Genome rearrangements in Escherichia coli during de novo acquisition of resistance to a single antibiotic or two antibiotics successively. BMC Genomics19, 973 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Pal, A. et al. Clinically relevant mutations in regulatory regions of metabolic genes facilitate early adaptation to ciprofloxacin in Escherichia coli. Nucleic Acids Res.52, 10385–10399 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Uppalapati, S. et al. Prophage terminase with tRNase activity sensitizes Salmonella enterica to oxidative stress. Science384, 100–105 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Song, S. et al. Escherichia coli cryptic prophages sense nutrients to influence persister cell resuscitation. Environ. Microbiol.23, 7245–7254 (2021). [DOI] [PubMed] [Google Scholar]
26.Brauner, A., Fridman, O., Gefen, O. & Balaban, N. Q. Distinguishing between resistance, tolerance and persistence to antibiotic treatment. Nat. Rev. Microbiol.14, 320–330 (2016). [DOI] [PubMed] [Google Scholar]
27.Levin-Reisman, I. et al. Antibiotic tolerance facilitates the evolution of resistance. Science355, 826–830 (2017). [DOI] [PubMed] [Google Scholar]
28.Okusu, H., Ma, D. & Nikaido, H. AcrAB efflux pump plays a major role in the antibiotic resistance phenotype of Escherichia coli multiple-antibiotic-resistance (Mar) mutants. J. Bacteriol.178, 306–308 (1996). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Bergmiller, T. et al. Biased partitioning of the multidrug efflux pump AcrAB-TolC underlies long-lived phenotypic heterogeneity. Science356, 311–315 (2017). [DOI] [PubMed] [Google Scholar]
30.Martínková, M., Kitanishi, K. & Shimizu, T. Heme-based globin-coupled oxygen sensors: linking oxygen binding to functional regulation of diguanylate cyclase, histidine kinase, and methyl-accepting chemotaxis. J. Biol. Chem.288, 27702–27711 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Fang, X. & Allison, K. R. Resuscitation dynamics reveal persister partitioning after antibiotic treatment. Mol. Syst. Biol.19, e11320 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Sulaiman, J. E. & Lam, H. Proteomic investigation of tolerant Escherichia coli populations from cyclic antibiotic treatment. J. Proteome Res.19, 900–913 (2020). [DOI] [PubMed] [Google Scholar]
33.Cingolani, P. et al. 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. Fly6, 80–92 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Karp, P. et al. The EcoCyc database. EcoSal13, eesp–0019–eesp–2024 (2025). [Google Scholar]
35.Huerta-Cepas, J. et al. eggNOG 5.0: a hierarchical, functionally and phylogenetically annotated orthology resource based on 5090 organisms and 2502 viruses. Nucleic Acids Res.47, D309–d314 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Alcock, B. P. et al. CARD 2023: expanded curation, support for machine learning, and resistome prediction at the comprehensive antibiotic resistance database. Nucleic Acids Res.51, D690–d699 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Wein, T. & Dagan, T. The Effect of Population Bottleneck Size and Selective Regime on Genetic Diversity and Evolvability in Bacteria. Genome Biol. Evol.11, 3283–3290 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Izutsu, M., Lake, D. M., Matson, Z. W. D., Dodson, J. P. & Lenski, R. E. Effects of periodic bottlenecks on the dynamics of adaptive evolution in microbial populations. Microbiology (Reading)170, 10.1099/mic.0.001494 (2024). [DOI] [PMC free article] [PubMed]
39.Feng, J. et al. Evolution of resistant mutants in Pseudomonas aeruginosa persister cells under meropenem treatment. Microorganisms13, 1672 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Pereira, C., Larsson, J., Hjort, K., Elf, J. & Andersson, D. I. The highly dynamic nature of bacterial heteroresistance impairs its clinical detection. Commun. Biol.4, 521 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Sergiev, P. V., Serebryakova, M. V., Bogdanov, A. A. & Dontsova, O. A. The ybiN Gene of Escherichia coli Encodes Adenine-N6 Methyltransferase Specific for Modification of A1618 of 23 S Ribosomal RNA, a Methylated Residue Located Close to the Ribosomal Exit Tunnel. J. Mol. Biol.375, 291–300 (2008). [DOI] [PubMed] [Google Scholar]
42.Joanny, G. et al. Polyadenylation of a functional mRNA controls gene expression in Escherichia coli. Nucleic Acids Res.35, 2494–2502 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Maes, A. et al. Landscape of RNA polyadenylation in E. coli. Nucleic Acids Res.45, 2746–2756 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Schmitt, M. W. et al. Detection of ultra-rare mutations by next-generation sequencing. Proc. Natl. Acad. Sci. USA.109, 14508–14513 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Mehta, P., Casjens, S. & Krishnaswamy, S. Analysis of the lambdoid prophage element e14 in the E. coli K-12 genome. BMC Microbiol4, 4 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Guet, C. C., Bruneaux, L., Oikonomou, P., Aldana, M. & Cluzel, P. Monitoring lineages of growing and dividing bacteria reveals an inducible memory of mar operon expression. Front. Microbiol.14, 2023 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Bertani, G. Studies on lysogenesis I: the mode of phage liberation by lysogenic Escherichia coli. J. Bacteriol.62, 293–300 (1951). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Baba, T. et al. Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Mol. Syst. Biol. 2, 2006.0008 (2006). [DOI] [PMC free article] [PubMed]
49.Kitagawa, M. et al. Complete set of ORF clones of Escherichia coli ASKA library (A Complete Set of E. coli K -12 ORF Archive): Unique Resources for Biological Research. DNA Res.12, 291–299 (2005). [DOI] [PubMed] [Google Scholar]
50.Gross, R., Yelin, I., Lázár, V., Datta, M. S. & Kishony, R. Beta-lactamase dependent and independent evolutionary paths to high-level ampicillin resistance. Nat. Commun.15, 5383 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Türkyılmaz, O. & Darcan, C. Resistance mechanism of Escherichia coli strains with different ampicillin resistance levels. Appl. Microbiol. Biotechnol.108, 5 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Kwan, B. W., Valenta, J. A., Benedik, M. J. & Wood, T. K. Arrested protein synthesis increases persister-like cell formation. Antimicrob. Agents Chemother.57, 1468–1473 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Pu, Y. et al. ATP-dependent dynamic protein aggregation regulates bacterial dormancy depth critical for antibiotic tolerance. Mol. Cell73, 143–156 (2019). [DOI] [PubMed]
54.Narayanaswamy, V. P. et al. Novel Glycopolymer Eradicates Antibiotic- and CCCP-Induced Persister Cells in Pseudomonas aeruginosa. Front. Microbiol.9, 2018 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Wiegand, I., Hilpert, K. & Hancock, R. E. Agar and broth dilution methods to determine the minimal inhibitory concentration (MIC) of antimicrobial substances. Nat. Protoc.3, 163–175 (2008). [DOI] [PubMed] [Google Scholar]
56.Lewis, K. Persister cells. Annu. Rev. Microbiol.64, 357–372 (2010). [DOI] [PubMed] [Google Scholar]
57.Li, H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv preprintarXiv:1303.3997 10.48550/arXiv.1303.3997 3(2013). [DOI]
58.DePristo, M. A. et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat. Genet.43, 491–498 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supporting information_053126 final^{(4.4MB, docx)}

Data Availability Statement

[CR1] 1.Rossolini, G. M., Arena, F., Pecile, P. & Pollini, S. Update on the antibiotic resistance crisis. Curr. Opin. Pharm.18, 56–60 (2014). [DOI] [PubMed] [Google Scholar]

[CR2] 2.Christaki, E., Marcou, M. & Tofarides, A. Antimicrobial Resistance in Bacteria: Mechanisms, Evolution, and Persistence. J. Mol. Evol.88, 26–40 (2020). [DOI] [PubMed] [Google Scholar]

[CR3] 3.Shah, D. et al. Persisters: a distinct physiological state of E. coli. BMC Microbiol.6, 53 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR4] 4.Lewis, K. Multidrug tolerance of biofilms and persister cells. Curr. Top. Microbiol. Immunol.322, 107–131 (2008). [DOI] [PubMed] [Google Scholar]

[CR5] 5.Hofsteenge, N., van Nimwegen, E. & Silander, O. K. Quantitative analysis of persister fractions suggests different mechanisms of formation among environmental isolates of E. coli. BMC Microbiol.13, 25 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR6] 6.Wakamoto, Y. et al. Dynamic persistence of antibiotic-stressed mycobacteria. Science339, 91–95 (2013). [DOI] [PubMed] [Google Scholar]

[CR7] 7.Balaban, N. Q. et al. Definitions and guidelines for research on antibiotic persistence. Nat. Rev. Microbiol.17, 441–448 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Korch, S. B., Henderson, T. A. & Hill, T. M. Characterization of the hipA7 allele of Escherichia coli and evidence that high persistence is governed by (p)ppGpp synthesis. Mol. Microbiol50, 1199–1213 (2003). [DOI] [PubMed] [Google Scholar]

[CR9] 9.Korch, S. B. & Hill, T. M. Ectopic overexpression of wild-type and mutant hipA genes in Escherichia coli: effects on macromolecular synthesis and persister formation. J. Bacteriol.188, 3826–3836 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR10] 10.Cohen, N. R., Lobritz, M. A. & Collins, J. J. Microbial persistence and the road to drug resistance. Cell Host Microbe13, 632–642 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR11] 11.Windels, E. M. et al. Bacterial persistence promotes the evolution of antibiotic resistance by increasing survival and mutation rates. ISME J.13, 1239–1251 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR12] 12.Sebastian, J. et al. De novo emergence of genetically resistant mutants of mycobacterium tuberculosis from the persistence phase cells formed against antituberculosis drugs in vitro. Antimicrob. Agents Chemother. 61, 10.1128/aac.01343-16 (2017). [DOI] [PMC free article] [PubMed]

[CR13] 13.Wilmaerts, D., Windels, E. M., Verstraeten, N. & Michiels, J. General mechanisms leading to persister formation and awakening. Trends Genet.35, 401–411 (2019). [DOI] [PubMed] [Google Scholar]

[CR14] 14.Salini, S. et al. The error-prone polymerase DnaE2 mediates the evolution of antibiotic resistance in persister Mycobacterial cells. Antimicrob. Agents Chemother.66, e0177321 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR15] 15.Swaminath, S. et al. Mycobacterium smegmatis moxifloxacin persister cells produce high levels of hydroxyl radical, generating genetic resisters selectable not only with moxifloxacin, but also with ethambutol and isoniazid. Microbiol. (Read.)166, 180–198 (2020). [DOI] [PubMed] [Google Scholar]

[CR16] 16.Van den Bergh, B. et al. Frequency of antibiotic application drives rapid evolutionary adaptation of Escherichia coli persistence. Nat. Microbiol.1, 16020 (2016). [DOI] [PubMed] [Google Scholar]

[CR17] 17.Vogwill, T., Comfort, A. C., Furió, V. & MacLean, R. C. Persistence and resistance as complementary bacterial adaptations to antibiotics. J. Evol. Biol.29, 1223–1233 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR18] 18.Bakkeren, E., Diard, M. & Hardt, W. D. Evolutionary causes and consequences of bacterial antibiotic persistence. Nat. Rev. Microbiol18, 479–490 (2020). [DOI] [PubMed] [Google Scholar]

[CR19] 19.Ueno, H., Kato, Y., Tabata, K. V. & Noji, H. Revealing the metabolic activity of persisters in Mycobacteria by Single-Cell D(2)O Raman imaging spectroscopy. Anal. Chem.91, 15171–15178 (2019). [DOI] [PubMed] [Google Scholar]

[CR20] 20.Wang, X. et al. Cryptic prophages help bacteria cope with adverse environments. Nat. Commun.1, 147 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR21] 21.Bergum, O. E. T. et al. SOS genes are rapidly induced while translesion synthesis polymerase activity is temporally regulated. Front. Microbiol.15, 2024 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Hoeksema, M., Jonker, M. J., Bel, K., Brul, S. & ter Kuile, B. H. Genome rearrangements in Escherichia coli during de novo acquisition of resistance to a single antibiotic or two antibiotics successively. BMC Genomics19, 973 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Pal, A. et al. Clinically relevant mutations in regulatory regions of metabolic genes facilitate early adaptation to ciprofloxacin in Escherichia coli. Nucleic Acids Res.52, 10385–10399 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Uppalapati, S. et al. Prophage terminase with tRNase activity sensitizes Salmonella enterica to oxidative stress. Science384, 100–105 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR25] 25.Song, S. et al. Escherichia coli cryptic prophages sense nutrients to influence persister cell resuscitation. Environ. Microbiol.23, 7245–7254 (2021). [DOI] [PubMed] [Google Scholar]

[CR26] 26.Brauner, A., Fridman, O., Gefen, O. & Balaban, N. Q. Distinguishing between resistance, tolerance and persistence to antibiotic treatment. Nat. Rev. Microbiol.14, 320–330 (2016). [DOI] [PubMed] [Google Scholar]

[CR27] 27.Levin-Reisman, I. et al. Antibiotic tolerance facilitates the evolution of resistance. Science355, 826–830 (2017). [DOI] [PubMed] [Google Scholar]

[CR28] 28.Okusu, H., Ma, D. & Nikaido, H. AcrAB efflux pump plays a major role in the antibiotic resistance phenotype of Escherichia coli multiple-antibiotic-resistance (Mar) mutants. J. Bacteriol.178, 306–308 (1996). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR29] 29.Bergmiller, T. et al. Biased partitioning of the multidrug efflux pump AcrAB-TolC underlies long-lived phenotypic heterogeneity. Science356, 311–315 (2017). [DOI] [PubMed] [Google Scholar]

[CR30] 30.Martínková, M., Kitanishi, K. & Shimizu, T. Heme-based globin-coupled oxygen sensors: linking oxygen binding to functional regulation of diguanylate cyclase, histidine kinase, and methyl-accepting chemotaxis. J. Biol. Chem.288, 27702–27711 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR31] 31.Fang, X. & Allison, K. R. Resuscitation dynamics reveal persister partitioning after antibiotic treatment. Mol. Syst. Biol.19, e11320 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Sulaiman, J. E. & Lam, H. Proteomic investigation of tolerant Escherichia coli populations from cyclic antibiotic treatment. J. Proteome Res.19, 900–913 (2020). [DOI] [PubMed] [Google Scholar]

[CR33] 33.Cingolani, P. et al. 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. Fly6, 80–92 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Karp, P. et al. The EcoCyc database. EcoSal13, eesp–0019–eesp–2024 (2025). [Google Scholar]

[CR35] 35.Huerta-Cepas, J. et al. eggNOG 5.0: a hierarchical, functionally and phylogenetically annotated orthology resource based on 5090 organisms and 2502 viruses. Nucleic Acids Res.47, D309–d314 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR36] 36.Alcock, B. P. et al. CARD 2023: expanded curation, support for machine learning, and resistome prediction at the comprehensive antibiotic resistance database. Nucleic Acids Res.51, D690–d699 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR37] 37.Wein, T. & Dagan, T. The Effect of Population Bottleneck Size and Selective Regime on Genetic Diversity and Evolvability in Bacteria. Genome Biol. Evol.11, 3283–3290 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR38] 38.Izutsu, M., Lake, D. M., Matson, Z. W. D., Dodson, J. P. & Lenski, R. E. Effects of periodic bottlenecks on the dynamics of adaptive evolution in microbial populations. Microbiology (Reading)170, 10.1099/mic.0.001494 (2024). [DOI] [PMC free article] [PubMed]

[CR39] 39.Feng, J. et al. Evolution of resistant mutants in Pseudomonas aeruginosa persister cells under meropenem treatment. Microorganisms13, 1672 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR40] 40.Pereira, C., Larsson, J., Hjort, K., Elf, J. & Andersson, D. I. The highly dynamic nature of bacterial heteroresistance impairs its clinical detection. Commun. Biol.4, 521 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR41] 41.Sergiev, P. V., Serebryakova, M. V., Bogdanov, A. A. & Dontsova, O. A. The ybiN Gene of Escherichia coli Encodes Adenine-N6 Methyltransferase Specific for Modification of A1618 of 23 S Ribosomal RNA, a Methylated Residue Located Close to the Ribosomal Exit Tunnel. J. Mol. Biol.375, 291–300 (2008). [DOI] [PubMed] [Google Scholar]

[CR42] 42.Joanny, G. et al. Polyadenylation of a functional mRNA controls gene expression in Escherichia coli. Nucleic Acids Res.35, 2494–2502 (2007). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.Maes, A. et al. Landscape of RNA polyadenylation in E. coli. Nucleic Acids Res.45, 2746–2756 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR44] 44.Schmitt, M. W. et al. Detection of ultra-rare mutations by next-generation sequencing. Proc. Natl. Acad. Sci. USA.109, 14508–14513 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR45] 45.Mehta, P., Casjens, S. & Krishnaswamy, S. Analysis of the lambdoid prophage element e14 in the E. coli K-12 genome. BMC Microbiol4, 4 (2004). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR46] 46.Guet, C. C., Bruneaux, L., Oikonomou, P., Aldana, M. & Cluzel, P. Monitoring lineages of growing and dividing bacteria reveals an inducible memory of mar operon expression. Front. Microbiol.14, 2023 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Bertani, G. Studies on lysogenesis I: the mode of phage liberation by lysogenic Escherichia coli. J. Bacteriol.62, 293–300 (1951). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Baba, T. et al. Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection. Mol. Syst. Biol. 2, 2006.0008 (2006). [DOI] [PMC free article] [PubMed]

[CR49] 49.Kitagawa, M. et al. Complete set of ORF clones of Escherichia coli ASKA library (A Complete Set of E. coli K -12 ORF Archive): Unique Resources for Biological Research. DNA Res.12, 291–299 (2005). [DOI] [PubMed] [Google Scholar]

[CR50] 50.Gross, R., Yelin, I., Lázár, V., Datta, M. S. & Kishony, R. Beta-lactamase dependent and independent evolutionary paths to high-level ampicillin resistance. Nat. Commun.15, 5383 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR51] 51.Türkyılmaz, O. & Darcan, C. Resistance mechanism of Escherichia coli strains with different ampicillin resistance levels. Appl. Microbiol. Biotechnol.108, 5 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR52] 52.Kwan, B. W., Valenta, J. A., Benedik, M. J. & Wood, T. K. Arrested protein synthesis increases persister-like cell formation. Antimicrob. Agents Chemother.57, 1468–1473 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR53] 53.Pu, Y. et al. ATP-dependent dynamic protein aggregation regulates bacterial dormancy depth critical for antibiotic tolerance. Mol. Cell73, 143–156 (2019). [DOI] [PubMed]

[CR54] 54.Narayanaswamy, V. P. et al. Novel Glycopolymer Eradicates Antibiotic- and CCCP-Induced Persister Cells in Pseudomonas aeruginosa. Front. Microbiol.9, 2018 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR55] 55.Wiegand, I., Hilpert, K. & Hancock, R. E. Agar and broth dilution methods to determine the minimal inhibitory concentration (MIC) of antimicrobial substances. Nat. Protoc.3, 163–175 (2008). [DOI] [PubMed] [Google Scholar]

[CR56] 56.Lewis, K. Persister cells. Annu. Rev. Microbiol.64, 357–372 (2010). [DOI] [PubMed] [Google Scholar]

[CR57] 57.Li, H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv preprintarXiv:1303.3997 10.48550/arXiv.1303.3997 3(2013). [DOI]

[CR58] 58.DePristo, M. A. et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat. Genet.43, 491–498 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Selective context, rather than persister cycling alone, drives resistance fixation in Escherichia coli

Garin Park

Hyein Kim

Sooyeon Song

Abstract

Introduction

Results

Persister cycling without selection maintains baseline MIC

Fig. 1. Experimental design and colony phenotypes under antibiotic stress.

Table 1.

Quality-filtered WGS detects fixed variants only under mutation-induced selection

Fig. 2. Genome-resolved mutation profiles and variant distributions across cellular states.

Prophage-linked insertions are associated with stress-exposed states

Fig. 3. Prophage-linked insertions at stfE and BW25113_RS25865 under antibiotic stress.

The stfE/stfP background alters survival, MIC outcomes, and mutational trajectories under mutation-induced selection

Fig. 4. Effects of ΔstfE and ΔstfP on colony phenotypes and survival under antibiotic stress.

Fig. 5. Divergent mutational trajectories in ΔstfE and ΔstfP mutants under antibiotic stress.

Persister-associated low-frequency variants are stochastically retained or lost during cycling

Discussion

Methods

Bacteria and growth conditions

Mutation-induced cells

Persister-cycling cells

Persister-Amp cycling cells

Stationary-phase cells and NaCl-stored persister cells

Minimum inhibitory concentration (MIC) assay

Ampicillin susceptibility assay

Growth and survival assay of specific gene-deletion mutants

Single-colony tracking of persister cycling and resuscitation

DNA preparation

WGS

Variant calling and analysis

Statistical analysis

Supplementary information

Acknowledgements

Author contributions

Data Availability

Competing interests

Footnotes

Supplementary information

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases