Experimental data for Pseudomonas aeruginosa from experimental evolution under different bottleneck sizes and antibiotic selection pressures

We here combined evolution experiments with genomic and genetic analyses to assess whether bottleneck size and antibiotic-induced selection influences the evolutionary path to resistance in pathogenic Pseudomonas aeruginosa, one of the most problematic opportunistic human pathogens. Two sets of evolution experiments were performed across 16 transfers, in which either the aminoglycoside gentamicin or the fluoroquinolone ciprofloxacin were used as antibiotic. The evolutionary response was studied using counts of bacterial cells at the end of each transfer period (i.e., yield) or by calculating the growth rate from regular optical density measurement during the respective transfer periods. For the populations at the end of the evolution experiments, we also determined their antibiotic resistance with the help of standardized dose response curves. Moreover, we performed whole genome sequencing to assess the frequency of variants, which emerged and spread during the evolution experiments. We further focused on variants in two specific genes, which were selectively favoured in the gentamicin evolution experiments, and assessed their relative fitness using competition experiments. We found that antibiotic resistance is favoured under high antibiotic selection and weak bottlenecks, but also under low antibiotic selection and severe bottlenecks. We found that the absence of high resistance under low selection and weak bottlenecks is caused by the spread of low-resistance variants with high competitive fitness under these conditions. We conclude that bottlenecks in combination with drug-induced selection are currently neglected key determinants of pathogen evolution and antibiotic treatment outcome.Funding provided by: Deutsche ForschungsgemeinschaftCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100001659Award Number: JA 2342/2-1Funding provided by: Deutsche ForschungsgemeinschaftCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100001659Award Number: SCHU 1415/12-2Funding provided by: Deutsche ForschungsgemeinschaftCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100001659Award Number: EXC 2167–390884018Funding provided by: Leibniz-GemeinschaftCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100001664Award Number: Leibniz Science Campus EvoLUNGFunding provided by: Max-Planck-GesellschaftCrossref Funder Registry ID: http://dx.doi.org/10.13039/501100004189Award Number: FellowshipMaterial: All experiments were performed with Pseudomonas aeruginosa strain PA14 and mutants thereof, which arose during the evolution experiments. Bacteria were grown in M9 minimal medium, consisting of 7 g/l K2HPO4, 2 g/l KH2PO4, 0.588 g/l Na3C6H5O7, 1 g/l (NH4)2SO4, 0.1 g/l MgSO4, and supplemented with 2 g/l glucose and 1 g/l casamino acids. Different single colonies from M9 agar plates (M9 supplemented with 15 g/l agar) were picked to initiate the independent replicate populations of the evolution experiments. Exponential phase cultures with an optical density (OD) of 0.1 (equivalent to 104-105 CFU/ml) were used as inoculum for resistance assays. Flow cytometry: A Guava easyCyte flow cytometer was used to assess cell counts of bacterial cultures during the evolution and competition experiments. Cells were suspended at an appropriate concentration (< 1500 cells/µl) in phosphate buffered saline (PBS) and subjected to flow cytometry using a flow rate of 0.236 µl/s for either 30 seconds or until a total cell count of 5000 cells per sample was reached. 1.9 mM propidium iodide was used to stain dead cells and thus to determine the number of viable cells during flow cytometry. For each culture, the total number of cells was calculated from the measured cell concentration, adjusted by the number of dead cells, sampling volume, and dilution factor. These calculations were used to determine the transfer volume during the evolution experiment and to infer final yield at the end of each season as a proxy for bacterial fitness. Dose-response curves and IC determination for the ancestral populations: We used dose-response curves in broth cultures and OD measurements as a proxy for bacterial growth in consistency with standard diagnostic approaches for antibiotic resistance measurements (e.g., the Vitek 2 approach, bioMérieux Ltd). Bacterial growth was assessed in 100 µl volumes at ten different antibiotic concentrations in a 96-well plate using a fully randomized design, including eight technical replicates per concentration, eight no-drug controls and eight medium-only controls. Growth was measured as OD after 12 hours at 37 °C under constant shaking (double-orbital, 900 rpm) in Tecan plate readers. The lowest antibiotic concentration for which no visible growth was optically measurable was taken as the drug's MIC. Design and general setup of evolution experiments: We developed a novel protocol for experimental evolution that includes cell counting by flow cytometry, to ensure that an exact number of cells are transferred across growth periods. The evolution experiment was run in 96-well plates over 16 growth periods and included two distinct transfer bottleneck sizes (BN of 50,000 cells and 5,000,000 cells) and three inhibitory antibiotic concentrations (IC0, IC20, IC80), combined in a full factorial design. The evolution treatments were randomized across the 96-well plates using a block design and each included eight fully independent replicates. Experimental evolution was performed independently with two distinct antibiotics, the aminoglycoside gentamicin and the fluoroquinolone ciprofloxacin. In detail, after each growth period, a 4 µl subsample of each bacterial culture was diluted 1:1000 in PBS in two steps in a new 96-well plate and then used to determine the exact cell count by flow cytometry (see above protocol). In parallel, the remainder of the bacterial culture was centrifuged at 5000 rpm for 75 minutes to remove the old growth medium and resuspended in fresh M9 medium. The resuspension volumes were calculated individually, based on the flow cytometry results. Resuspension volumes were set to achieve a concentration of 5 x 105 cells/?l per culture. The plate for the next growth period was prepared accordingly (100 ?l per well minus the calculated transfer volume) and the respective volumes were then transferred to the new plate: 0.1 ?l for 50,000 cells and 10 ?l for 5,000,000 cells. The freshly inoculated plate was sealed with transparent foil and incubated for 9.5 h in a Tecan plate reader at 37 °C under constant shaking (double-orbital, 900 rpm). OD measurements were taken in 15-minutes intervals throughout each growth period of the evolution experiments and later used for inference of growth rates. Assessment of bacterial fitness: We calculated two proxies of bacterial fitness during the evolution experiment. On the one hand, we used the absolute cell counts, determined with flow cytometry at the end of each growth period, to infer final yield for each growth period and replicate population, followed by calculation of relative yield by dividing the cell counts for each replicate population with that for the corresponding no-drug control (either IC0-k50 for strong bottleneck treatments or IC0-M5 for weak bottleneck treatments). This data is provided in Mahrt_etal_Data_1.csv. On the other hand, we also used the continuous OD measurements to calculate growth rate for each growth period and each replicate population. This data is provided in Mahrt_etal_Data_2.csv. Antibiotic resistance of evolved populations: Populations of the last transfer period were challenged against different antibiotic concentrations of the treatment drug, to obtain dose-response curves for the evolved populations (see above protocol). For this assessment, all cultures were standardized to the same population size of 500,000 cells. All bacterial populations were consistently subjected to eight distinct antibiotic concentrations: two below (IC50, IC80) and six above the MIC (2xMIC, 4xMIC, 8xMIC and 16xMIC) of the ancestral population. We specifically chose this design to ensure comparability of the experimental procedures for the evolved populations. Control measurements were done in the absence of antibiotics. After 12 h growth, endpoint ODs were obtained for all concentrations. This data is provided in the file Mahrt_etal_Data_3.csv. Variant frequencies during the evolution experiment: We performed whole genome sequencing of entire populations. DNA was isolated from experimentally evolved populations and the original starting cultures. WGS of DNA samples of the last transfer period was performed at the Competence Centre for Genomic Analysis (CCGA) Kiel, Germany. Sequencing libraries were generated with the Nextera DNA Flex library preparation kit and sequencing was performed on the Illumina HiSeq 4000 platform using the Illumina paired-end technology with read lengths of 150 bp and an average base coverage of >100. In addition, DNA extracted from transfers 3, 5, 7, 9, 11 and 13 were sequenced on the Illumina NextSeq platform at the Max Planck Institute for Evolutionary Biology (MPI-EB) in Plön, Germany, with read lengths of 150 bp and an average base coverage of >40. Sequence reads were provided in the fastq format. We excluded populations IC20-M5 H2 and IC80-M5 G7 from the GEN experiment, because material from some transfers could not be recovered. Quality and quantity of reads were checked with FastQC. Trimmomatic was used to remove sequencing adapters from the Nextera library and to filter out low quality reads. High quality reads were mapped to the UCBPP_PA14 reference genome with the software bwa. The generated .bam files were scanned for SNPs, insertions and deletions using the variant calling programs FreeBayes, PinDel and VarScan. The resulting output files were filtered for duplicates, ancestral variants, and variants found in the evolved controls using R and additionally checked by visually inspecting the called genome positions provided by the .bam file in the IGV genome browser. The detected variants were annotated with the help of SnpEff and the Pseudomonas database (available at http://pseudomonas.com). The variant frequencies for transfer periods 3 and 16 is provided in the file Mahrt_etal_Data_4.csv and the variant frequencies from across the evolution experiments in the file Mahrt_etal_Data_5.csv. Antibiotic resistance of selected variants: We used the antibiotic resistance data obtained above for selected populations, which had a high abundance of variants in either the gene pmrB or the gene ptsP, in order to compare variation in resistance associated with the two genes. The antibiotic resistance data for this comparison is provided in the file Mahrt_etal_Data_6.csv. Competition assays and clone frequency analysis using amplicon sequencing: In total, three pairs of different pmrB and ptsP mutants were competed against one another and against the PA14 reference in 96-well plates. As a control, each single strain was incubated in individual wells under the same conditions on a separate 96-well plate. The used clones were picked as a single CFU from M9 agar plates, which had been inoculated with a cryo-preserved culture of an evolved bacterial population, which should contain only a single variant according to the performed WGS of bacterial populations. The picked clones were cultivated in 5 ml M9-medium at 37 °C. Cultures of competing strains were set to an OD of 0.1 mixed at a 1:1 ratio before inoculation of the competition cultures. The cultures were then transferred to 96-well plates. Competitions ran with culture volumes of 100 µl at different gentamicin concentrations for 12 hours at 37 °C in a Tecan plate reader. The relative frequency of the competing strains was determined using amplification of diagnostic genome regions and their subsequent sequencing. Bacterial pellets from the end of the competition experiments were resuspended in 50 ?l nuclease-free water and boiled for 15 minutes. A two-step PCR was subsequently performed to amplify the region of interest and to ligate barcodes to the amplicons, followed by sequencing on the Illumina NextSeq platform of the MPI-EB. The mean frequencies of every replicate were calculated and is provided in the file Mahrt_etal_Data_7.csv