Long-term evolution of antibiotic tolerance in Pseudomonas aeruginosa lung infections

Abstract Pathogenic bacteria respond to antibiotic pressure with the evolution of resistance but survival can also depend on their ability to tolerate antibiotic treatment, known as tolerance. While a variety of resistance mechanisms and underlying genetics are well characterized in vitro and in vivo, an understanding of the evolution of tolerance, and how it interacts with resistance in situ is lacking. We assayed for tolerance and resistance in isolates of Pseudomonas aeruginosa from chronic cystic fibrosis lung infections spanning up to 40 years of evolution, with 3 clinically relevant antibiotics: meropenem, ciprofloxacin, and tobramycin. We present evidence that tolerance is under positive selection in the lung and that it can act as an evolutionary stepping stone to resistance. However, by examining evolutionary patterns across multiple patients in different clone types, a key result is that the potential for an association between the evolution of resistance and tolerance is not inevitable, and difficult to predict.


Supplementary table legends (tables in excel file)
Table S1: An overview of isolates of the transmissible clone types DK1 and DK2, with the sampling year, patient ID and time since first sampling of clone type.The tolerance and resistance data is divided by antibiotic, cipro = ciprofloxacin, mero = meropenem, tobra = tobramycin.For each isolate and drug is the overall type given (L = low tolerance, H = high tolerance, S = sensitive, R = resistant) and average CFU as undiluted (CFU1), diluted 10 fold (CFU10) and diluted 100 fold (CFU100), and average OD for each experiment, and the overall average OD per isolate.The MIC measured after 24h is given, and the resistance profile as sensitive, intermediate or resistant following the clinical cut-offs listed in the methods.
Table S2: Results from GLMs for antibiotic tolerance, testing the effect of time and max OD, and the interaction between the two.In blue is highlighted p-values < 0.05 and the best fit model with the lowest Akaike information criterion (AIC).Analysis for DK1 and DK2 separately.
Table S3: Results from GLMs for antibiotic resistance, testing the effect of time and max OD, and the interaction between the two.In blue is highlighted p-values < 0.05 and the best fit model with the lowest Akaike information criterion (AIC).Analysis for DK1 and DK2 separately.
Table S4: Results from 2-way ANOVA, testing the effect of max OD on tolerance and resistance, and the interaction between the two.Analysis for DK1 and DK2 separately.
Table S5: Results from 2-way ANOVA, testing the effect of length of infection on tolerance and resistance, and the interaction between the two.Analysis for DK1 and DK2 separately.et al. 2013 (DK2).Each isolate is denoted by the sample year, patient ID and sample ID.The heat map shows the presence (red square) or absence (white square) of tolerance (T) and resistance (R) to the three antibiotics, in the order ciprofloxacin, meropenem, and tobramycin.Isolates with missing phenotypic data are shown with a strike-through.To ease interpretation, branch lengths are not drawn to scale.Isolates that only show high tolerance when the culture is diluted 10 times are shown in pink (only found in DK1).1994P33F3 1992P33F3 1992P33F3 1991P33F3 1997P33F3 2003P33F3 2003P33F3 2004P33F3 2003P33F3 2007P33F3 2007P33F3 2007P33F3 2005P33F3 2004P33F3 2004P33F3 2002P11M3 2002P23M2 2002P77F1 NN 2002P40F2 2001P22F2 1987P24M2 2002P24M2 2002P80F1 Nna 1984P73F1 1991P73F1 1984P73F1 1984P73F1 2005P73F1 2002P73F1 1972P84M0 1973P05M1 1973P43F0 1973 P14F1 Reference ci pr oT ci pr oR m er oT m er oR to br aT to br aR Fig. S2 Graphs show the difference in persistence measured as CFU counts from undiluted culture or culture diluted 10 times for isolates classified as low tolerance, high tolerance, and isolates that "revive" and become high tolerance only when antibiotic is diluted.Only the "revival" category has a mean difference above 50.Boxplots show median CFU counts ± 25 percentiles and values as dots.

Supplementary text
We tested the correlation between the evolution of resistance and tolerance using the fitPagel function in phytools in R, using the fitMk method, and an equal rates model.We performed the analyses for DK1 and DK2 separately.
We found that for DK2, the evolution of resistance significantly depended on tolerance for ciprofloxacin and meropenem, as the transition from susceptible, nontolerant to resistant nontolerant, and vice versa was unlikely to occur.For DK1, tolerance to meropenem depended on significantly resistance, the transition from nontolerant resistant to tolerant resistant, and vice versa was unlikely to occur.
Between antibiotics, tolerance and resistance to ciprofloxacin and meropenem, and meropenem and tobramycin was correlated for DK1.For both DK1 and DK2 resistance to ciprofloxacin and tobramycin was correlated.

Fig. S3
Fig. S3 Correlation between the mean CFU count at 10 times dilution, and the standard deviation