Enhanced antibiotic resistance development from fluoroquinolone persisters after a single exposure to antibiotic

Abstract

Bacterial persisters are able to tolerate high levels of antibiotics and give rise to new populations. Persister tolerance is generally attributed to minimally active cellular processes that prevent antibiotic-induced damage, which has led to the supposition that persister offspring give rise to antibiotic-resistant mutants at comparable rates to normal cells. Using time-lapse microscopy to monitor Escherichia coli populations following ofloxacin treatment, we find that persisters filament extensively and induce impressive SOS responses before returning to a normal appearance. Further, populations derived from fluoroquinolone persisters contain significantly greater quantities of antibiotic-resistant mutants than those from untreated controls. We confirm that resistance is heritable and that the enhancement requires RecA, SOS induction, an opportunity to recover from treatment, and the involvement of error-prone DNA polymerase V (UmuDC). These findings show that fluoroquinolones damage DNA in persisters and that the ensuing SOS response accelerates the development of antibiotic resistance from these survivors.

Fig. 1

Time-lapse microscopy with E. coli MG1655. a Representative phase contrast and b GFP fluorescence images of recovering OFL-treated wild-type MG1655 E. coli bearing P_recA-gfp. Yellow arrow indicates persister. Red circles indicate division events. The far-right panel shows the incipient colony formation from a persister. Images are representative of six biological replicates. c Growth and d fluorescence of OFL-treated (n = 30) and e, f untreated cells (n = 20) recovering from treatment were analyzed. Of the 30 treated cells, five were persisters (black data points) and 25 were non-persisters (blue markers). 95% of the cell length and fluorescence measurements lie below the respective red dashed lines in c–f. The scale bars represent 10 μm. Refer to Supplementary Movies 1–3 for time-lapse videos

Fig. 2

Time-lapse microscopy of E. coli MG1655 ΔrecA and MG1655 lexA3. a Representative phase contrast and b GFP fluorescence images of recovering OFL-treated MG1655 ΔrecA harboring P_recA-gfp. Growth and fluorescence of OFL-treated (c, d) and untreated (e, f) MG1655 ΔrecA recovering from treatment were analyzed. g Phase contrast and h GFP fluorescence images of OFL-treated MG1655 lexA3 harboring P_recA-gfp. Growth and fluorescence of OFL-treated (i, j) and untreated (k, l) MG1655 lexA3 recovering from treatment were analyzed. 95% of the cell length and fluorescence measurements lie below the respective red dashed lines. Appreciable filamentation and fluorescence enhancement, and persister formation were not observed in either mutant. Fifteen cells from each treatment condition were quantified. The scale bars represent 10 μm. Refer to Supplementary Movies 5–8 for time-lapse videos

Fig. 3

High frequencies of resistant mutants in cultures from OFL persisters. a The fold-changes between the number of resistant cells per 10⁹ culturable CFUs (RPB: resistant colonies per billion) to 175 ng/mL OFL (n = 9), 500 μg/mL RIF (n = 13), 30 μg/mL CARB (n = 6), 100 μg/mL DCS (n = 12), or 20 μg/mL FOS (n = 3) that arise from OFL persisters (RPB_persisters) compared to those that arise from untreated cells (RPB_untreated). Raw values for each independent replicate are provided in Supplementary Fig. 6a. b Wild-type (n = 8) and pre-existing populations of mutants resistant to 175 ng/mL OFL, 500 μg/mL RIF, 30 μg/mL CARB, 100 μg/mL DCS, or 20 μg/mL FOS (all n = 5) were cultured to stationary phase in LB and subjected to OFL persistence assays. RIF, CARB, and FOS-resistant mutants did not have a higher tolerance to 5 μg/mL OFL compared to wild-type, whereas OFL- and DCS-resistant mutants exhibited higher tolerances. c Without recovery in an antibiotic-free environment, enhancement of RIF-resistant mutants from OFL persisters is not observed. Fold-change represents RIF-resistant mutants derived from OFL-treated populations (RPB_OFL) and untreated populations (RPB_untreated) before antibiotic addition (“Before Treatment”, n = 3) and after OFL removal at the end of 5 h of treatment (“After treatment, no recovery”, n = 4). d Time course analysis of RIF-resistance development during recovery reveals that an enhancement in resistance can be observed as early as 8 h of recovery from OFL treatment (n = 7). We note that 10⁹ culturable CFUs could not be obtained at 2 and 4 h of recovery in LB, and thus resistance could not be quantified at those earlier time points. Error bars portray S.E.M. For a, c, and d, a value of 1 is indicative of equivalent abundances (no change). Asterisks in these panels denote a log-transformed fold-change that is significantly different compared to a log-transformed value of 1 (p ≤ 0.05, using two-tailed t-tests with unequal variances). For b, asterisks denote statistically significant difference in survival fraction compared with the wild-type

Fig. 4

Resistance enhancement involves RecA, SOS induction, and UmuDC. a OFL persisters derived from ΔrecA (n = 3), which is defective in SOS induction and recombinatorial repair, do not give rise to a population with increased RIF resistance. Complementation of ΔrecA restores the enhancement of RIF-resistant mutants from OFL persisters (n = 6), whereas persisters from ΔrecA mutants bearing the pBAD33 control do not give rise to a population with enhanced RIF resistance (n = 8). b Increase in RIF-resistant mutants was not observed in OFL persisters derived from lexA3 mutants (n = 4), which cannot induce SOS. c Upon deleting umuDC, which encodes error-prone, SOS-induced DNA polymerase V, a significant increase in RIF-resistant mutants in OFL persister-derived populations was observed (n = 5), but the magnitude of that increase was significantly lower than that of wild-type (n = 13). Mutants lacking error-prone polymerase II (ΔpolB, n = 6) or error-prone polymerase IV (ΔdinB, n = 6) had significant increases in RIF-resistant mutants in OFL persister-derived populations that were of a comparable magnitude as wild-type. d Complementation of UmuDC on pBAD33 in ΔumuDC mutants restores increase in RIF-resistant colonies from OFL persisters (n = 7), whereas the presence of pBAD33 control fails to restore the phenotype (n = 9). Error bars portray S.E.M. * indicates significance (p ≤ 0.05) between log-transformed fold-changes and log-transformed value of 1 (fold-change of 1 would indicate no difference in resistant mutants). # indicates significance (p ≤ 0.05, using two-tailed t-tests with unequal variances) between log-transformed fold-changes and those of wild-type. Zero values for ΔrecA+pBAD33 were not included in the statistical analysis because they could not be log-transformed

Fig. 5

Fidelity of DNA repair in OFL persisters. a WGS was performed from DNA extracted from 10 persisters, 10 untreated controls, and 4 parental cultures, and number of nucleotides with reference allele frequencies (RAF) less than or equal to the indicated values are depicted in the histogram. RAF of 1 indicates that all sequencing reads of that nucleotide were of the reference nucleotide, whereas RAF of 0 indicates that all sequence reads had a non-reference nucleotide. Values between 0 and 1 indicate that some reads were of the reference nucleotide. The RAFs at different thresholds were indistinguishable among the persisters, untreated controls, and parental cultures, suggesting that OFL persisters generally repair DNA damage correctly. b Chromosomal content of stationary-phase cultures prior to OFL treatment was determined using PicoGreen stains (n = 3; see Supplementary Fig. 8i for controls used to determine fluorescence intensities corresponding to chromosome number). 30.8 ± 1.6% of cells contained one chromosome, whereas 68.4 ± 1.6% of cells contained two or more chromosomes, indicating that the majority of cells had the chromosomal content for repair by homologous recombination. Histogram shown is representative of three biological replicates. c Mutants defective in homologous recombination, which requires the resolution of Holliday junctions by RuvA and/or RecG, demonstrate reduced survival to OFL treatment in stationary phase (n = 3). d Enhancement in RIF resistance was observed in progeny of CIPRO persisters (n = 4), similar to offspring of OFL persisters, demonstrating that the phenomenon of resistance enhancement following treatment and recovery is shared between distinct fluoroquinolones. Error bars depict S.E.M. For c, * indicates significance between log-transformed survival fractions and those of wild-type (p ≤ 0.05, using two-tailed t-tests with unequal variances). For d, a value of 1 is indicative of equivalent abundances (no change), and * denotes a log-transformed fold-change that is significantly different compared to a log-transformed value of 1 (p ≤ 0.05, using two-tailed t-tests with unequal variance)