Integrons are mobile genetic elements that have played an important role in the dissemination of antibiotic resistance. As shown previously (Souque et al, 2021), the integron can generate under stress combinatorial variation in resistance cassette expression by cassette re-shuffling, accelerating the evolution of resistance. However, the flexibility of the integron integrase site recognition motif hints at potential off-target effects of the integrase on the rest of the genome that may have important evolutionary consequences. Here we test this hypothesis by selecting for increased piperacillin resistance populations of P.aeruginosa with a mobile integron containing a hard-to-mobilise beta-lactamase cassette to minimize the potential for adaptive cassette re-shuffling. We found that integron activity can both decrease overall survival rate but also improve the fitness of the surviving populations. Off-target inversions mediated by the integron accelerated plasmid adaptation by disrupting costly conjugative genes otherwise mutated in control populations lacking a functional integrase. Plasmids containing integron-mediated inversions were associated with lower plasmid costs and higher stability than plasmids carrying mutations, albeit at a cost of reduced conjugative ability. These findings highlight the potential for integrons to create structural variation that can drive bacterial evolution, and they provide an interesting example showing how antibiotic pressure can drive the loss of conjugative genes.

Antibiotic minimum inhibitory concentration (MIC) - MICs of integrons arrays / MICs of evolved populations / MICs of evolved plasmids in ancestral background

Minimum inhibitory concentrations (MIC) for each antibiotic were determined following the broth microdilution method from the Clinical and Laboratory Standards Institute (CLSI) guidelines. 5x105 c.f.u bacteria inocula were prepared in cation-adjusted Mueller-

Hinton Broth 2 (MH2) using individual colonies grown on selective agar and incubated in doubling concentrations of antibiotics for 20h in three technical replicates. Cultures optical density (OD600) was then read using a Biotek Synergy plate reader and wells were considered empty when the overall OD600 was under 0.1. The MIC for each assay was defined as the minimal concentration in which growth was inhibited in all three technical replicates. The final MICs values are the average of two to three replicate assays (from separately prepared inocula, on different days).

Growth curve assays (Fitness of evolved populations / fitness of evolved plasmids in ancestral background)

Growth rates in the absence of antibiotics were determined using a Biotek Synergy plate reader. Populations were inoculated in similar conditions to the MIC assay (5x105 c.f.u bacteria inocula prepared in cation-adjusted (MH2) from individual colonies) and followed for 24h at 37*C with periodic shaking, with OD600 measurements taken every 15 minutes. Intrinstic growth rate were determined using the Growthcurver package  (Sprouffske and Wagner, 2016). Each measurement was done in three technical replicates (separate wells from the same inocula) and replicated three times over different days (biological replicates).

Evolutionary ramp experiment (survival data)

90 populations of each strain (PA01:WTA4 and PA01:ΔintA4) were initially inoculated from individual colonies grown on selective agar in MH2 at a piperacillin concentration of 1/8 MIC. The outer wells of every plate were kept as negative controls and strain distribution among the plates was kept balanced to limit plate effects. Plates were kept at 37°C with shaking at 225 rpm. Every day the populations were diluted 1/10 000 in a doubling concentration of MIC until reaching x8 MIC. As controls, 15 populations of each strain were transferred either without antibiotic or at a constant piperacillin concentration of 1/8 MIC. The OD of each population was measured each day and a population was considered extinct when its OD fell below 0.1 after 20h of incubation. 

Conjugation rate

Determination of the conjugation rates of the ancestral and evolved plasmids was performed using filter mating with a rifampicin resistant PA01 acceptor (D512G rpoB mutant from (Qi et al., 2014)). Populations of donor and recipients were grown from overnight for 3h30 with (donors) and without (recipient) antibiotics, washed and resuspended in NaCl (0.9%). Donor and recipient populations were mixed at a 1:1 ratio, deposited on filter on agar plate without antibiotic alongside pure donor/recipients control and incubated for 1h. Populations were then resuspended in NaCl 0.9%. Concentrations of donors, recipients and transconjugants were determined at both T0 and Tf using appropriate antibiotics (ceftazidime 100mg/L, rifampicin 64mg/L and both combined). Conjugation rate was calculated using the Simonsen method (Simonsen et al., 1990).

Plasmid persistence

Populations with ancestral and evolved plasmids were inoculated from single colonies in the presence of antibiotic and passaged for 6 days in the absence of antibiotic with a similar dilution (1:10 000) as the experimental evolution experiment. After 6 days the ratio of plasmid containing cells was determined by serial dilution plating with (for the cfu of the plasmid containing cells) and without ceftazidime (for the cfu of the entire population). 

Transcription data

All RNA samples were treated with the TURBO DNA-free Kit (ThermoFisher) to eliminate genomic DNA. Concentration of the RNA samples was quantified using the Quantifluor RNA system (Promega). cDNA was synthesized from 100ng of RNA templates using random primers from the GoScript Reverse Transcription Mix (Promega). qPCR was carried out on the StepOnePlus Real-time PCR platform (Applied Biosystems) using the iTaq Universal SYBR Green Supermix. The cassette, as well as two reference genes (actpA and acp), were amplified using the primers described in Supplementary Table S2 in two technical replicates for each extraction. Standard curves for the pair of cassette primers was included in each PCR using restriction enzyme-digested gDNA as template, and used to quantify the amount of target cDNA in each sample to control for inter-run variations. For each biological replicate the cassette transcript levels were normalized based on the geometric means of the two internal reference genes, using the first array A1 as a reference, before division by its plasmid copy number. The biological replicates were then averaged for each array and normalised by the expression of the best array for each gene. Transcription data for the aadB gene is reproduced from the following dataset: https://doi.org/10.5061/dryad.rv15dv469