Mutations causing low level antibiotic resistance ensure bacterial survival in antibiotic-treated hosts

Abstract

In 474 genome sequenced Pseudomonas aeruginosa isolates from 34 cystic fibrosis (CF) patients, 40% of these harbor mutations in the mexZ gene encoding a negative regulator of the MexXY-OprM efflux pump associated with aminoglycoside and fluoroquinolone resistance. Surprisingly, resistance to aminoglycosides and fluoroquinolones of mexZ mutants was far below the breakpoint of clinical resistance. However, the fitness increase of the mutant bacteria in presence of the relevant antibiotics, as demonstrated in competition experiments between mutant and ancestor bacteria, showed that 1) very small phenotypic changes cause significant fitness increase with severe adaptive consequences, and 2) standardized phenotypic tests fail to detect such low-level variations. The frequent appearance of P. aeruginosa mexZ mutants in CF patients is directly connected to the intense use of the target antibiotics, and low-level antibiotic resistance, if left unnoticed, can result in accumulation of additional genetic changes leading to high-level resistance.

Figure 1

Mapping of clinical obtained mutations within the mexZ gene in P. aeruginosa. Here, mexZ mutations were mapped to one of the three functional mexZ domains; the DNA binding domain (α-helix 1 to α-helix 3), where α-helix 3 is the DNA recognition helix, and a C-terminal domain (α-helix 4 to α-helix 9). For each domain, the number of mexZ mutations consisting of a SNP or INDEL is noted with the accompanying MIC determination for the known MexXY-OprM substrates tobramycin, ciprofloxacin, and gentamicin. The clinical break-off points are presented according to EUCAST guidelines.

Figure 2

MexZ and mexY expression for selected clinical isolates. Relative expression of mexZ and mexY determined by RT-qPCR on RNA extracted from selected clinical isolates all of which contained a mexZ mutation apart from sample #1 from patient 9 (see Table S3 for details). All data were normalized to 16S RNA and ΔCt values between the genes of interest and 16S RNA were set at 1 for PAO1 wild-type strain. Experiments were performed in triplicate.

Figure 3

mexY and oprM expression. Comparative expression of the MexZ, MexY and OprM mRNA by RT–qPCR of PAO1 and PAO1 ΔmexZ, bacteria grown to exponential phase (OD₆₀₀ = 0.5) in LB medium. Every experiment were related to PAO1 (wild-type) as indicated by the dash line. If amikacin was supplemented it was added at OD₆₀₀ 0.1 at sub-MIC concentration (1 µg/mL). All data were normalized to the endogenous reference gene rpsL. No mexZ expression was detected in experiments with PAO1 ΔmexZ and no bar is shown. Experiments were performed in triplicate.

Figure 4

mexZ deficient bacteria has a faster recovery rate when challenged with MexXY-OprM substrates. In panel (A–C) we see antibiotic tolerance measured over time for PAO1, PAO1 ΔmexZ, and PAO1 ΔmexY to; (A) sub-MIC concentrations of colistin (1 µg/mL), (B) sub-MIC concentrations of ciprofloxacin (0.064 µg/mL), and (C) sub-MIC concentrations of amikacin (0.5 µg/mL). In (A,B), and (C) PAO1 was grown in the absence of colistin, amikacin, or ciprofloxacin, serving as a control. Panel (D) to (G) shows direct competition experiments between PAO1 Tn7:gfp (PAO1) and PAO1 ΔmexZ; (D) competition in LB in the absence of antibiotic, (E) competition in LB supplemented with sub-MIC concentrations of colistin (1 µg/mL), (F) competition in LB supplemented with sub-MIC concentrations of ciprofloxacin (0.125 µg/mL), (G) competition in LB supplemented with sub-MIC concentrations of amikacin (1 µg/mL). Bars represent the standard error of the log10 mean number of CFU per ml.

Figure 5

Survival of PAO1 wild-type and PAO1 ΔmexZ during treatment of biofilms with ciprofloxacin and meropenem. (A) Representative images of PAO1 tagged with yellow fluorescent protein and PAO1 ΔmexZ tagged with cyan fluorescent protein after live/dead staining (dead cells were stained for visualization using propidium iodide). Biofilms were either untreated, treated with one bolus (4 mg/l) Ciprofloxacin treatment for 24 h, or one bolus (107 mg/l) Meropenem for 24 h. Biofilms were grown on glucose minimal medium and treated with antibioitcs at day 3 (see Material and Methods). (B) COMSTAT quantification of the biomass of live and dead cells after live/dead staining of a PAO1 wild-type biofilm and PAO1 ΔmexZ biofilm from the same treatment regime as described in (A). In at least two independent experiments 3–4 random images were taken from each flow channel. A t-test (two-tailed) was performed between PAO1 and PAO1 ΔmexZ treated with ciprofloxacin showing a highly significant difference in survival (p < 0.001) as indicated by the three ***.