Repeated Exposure of Escherichia coli to High Ciprofloxacin Concentrations Selects gyrB Mutants That Show Fluoroquinolone-Specific Hyperpersistence

Abstract

Recent studies have shown that not only resistance, but also tolerance/persistence levels can evolve rapidly in bacteria exposed to repeated antibiotic treatments. We used in vitro evolution to assess whether tolerant/hyperpersistent Escherichia coli ATCC25922 mutants could be selected under repeated exposure to a high ciprofloxacin concentration. Among two out of three independent evolution lines, we observed the emergence of gyrB mutants showing an hyperpersistence phenotype specific to fluoroquinolones, but no significant MIC increase. The identified mutation gives rise to a L422P substitution in GyrB, that is, outside of the canonical GyrB QRDR. Our results indicate that mutations in overlooked regions of quinolone target genes may impair the efficacy of treatments via an increase of persistence rather than resistance level, and support the idea that, in addition to resistance, phenotypes of tolerance/persistence of infectious bacterial strains should receive considerations in the choice of antibiotic therapies.

Keywords: DNA gyrase; experimental evolution; persisters; quinolones; tolerance.

Figure 1

Repeated exposures to ciprofloxacin lead to the emergence of hyperpersistence and resistance phenotypes. (A) Chronological set up of the different antibiotic treatments and bacterial population samples evolved from the H1 parental strain. (B) Set up of the evolutionary experience with repeated ciprofloxacin treatments. Bacteria were exposed to 10 ciprofloxacin treatments as follows: bacteria were grown overnight, stationary phase cultures were sampled and diluted by a factor of 1:30 before addition of ciprofloxacin at 1 μg/ml for 5 h. After antibiotic removal by centrifugation and washing, all bacteria were grown again overnight in fresh medium for the next cycle. (C) Time-kill assays performed on bacterial populations obtained after 10 ciprofloxacin treatments. (D) Evolution of ciprofloxacin MIC of bacterial populations along treatments. (E) Evolved bacteria from the three evolution lines showed increased MDK₉₉.

Figure 2

The L422P substitution in GyrB is associated with an hyperpersistence phenotype specific to fluoroquinolones. (A) MICs of the parental strain (H1), of the H11-I1 mutant clone from line I and of a spontaneous resistant mutant (R) were determined for three fluoroquinolones (ciprofloxacin, norfloxacin, and enrofloxacin) and two older generation quinolones (nalidixic acid and flumequine). (B–D) Bactericidal assays in the presence of fluoroquinolones at 256-fold the MIC of each strain (ciprofloxacin, enrofloxacin, and norfloxacin) showed an altered bactericidal effect for the H11-I1 mutant, with a fraction of persisters increased by a factor 100 to 1,000. (E,F) Bactericidal essays in the presence of old generation quinolones (nalidixic acid and flumequine) showed no significant alteration of bactericidal effect for the H11-I1 mutant. Significant p-values compared with the H1 strain are noted (^**p ≤ 0.01).

Figure 3

The hyperpersistent phenotype associated with the GyrB L422P substitution is strain-dependent. (A) Levels of resistance to ciprofloxacin of reconstructed mutants of Escherichia coli ATCC25922, BL21, and MG1655 carrying the GyrB L422P or S464Y substitutions. SCAR strains are susceptible controls with no introduced mutation in gyrB, but containing the scar left by removal of the chloramphenicol resistance gene used for selection of recombinant clones. For each mutation, when possible, two independent recombinant clones were analyzed and then named (1) and (2). MICs of the three strains of E. coli with the L422P substitution were similar to those of the susceptible strains (SCAR) with only a 2-fold factor difference. (B) Bactericidal assays performed on the reconstructed E. coli ATCC25922 strains showed an alteration in bactericidal activity in the presence of ciprofloxacin for E. coli ATCC25922 (L422P) clones. (C,D) Bactericidal assays conducted on strains E. coli MG1655 and BL21 (L422P), (S464Y), and (SCAR) showed no alteration in bactericidal activity in the presence of ciprofloxacin. Significant p-values compared to SCAR1 and SCAR2 control clones are noted (^**p ≤ 0.01).