Antibiotic resistance spread via plasmids is a serious threat to successfully fight infections and makes understanding plasmid transfer in nature crucial to prevent the rise of antibiotic resistance. Studies addressing the dynamics of plasmid conjugation have yet neglected one omnipresent factor: prophages (viruses integrated into bacterial genomes), whose activation can kill host and surrounding bacterial cells. To investigate the impact of prophages on conjugation, we combined experiments and mathematical modelling. Using E. coli, prophage lambda and the multidrug-resistant plasmid RP4 we find that prophages can substantially limit the spread of conjugative plasmids. This inhibitory effect was strongly dependent on environmental conditions and bacterial genetic background. Our empirically parameterized model reproduced experimental dynamics of cells acquiring either the prophage or the plasmid well but failed to predict the number of cells acquiring both elements. This suggests more complex interactions between conjugative plasmids and prophages in sequential infections. Varying phage and plasmid infection parameters over empirically realistic ranges revealed that plasmids can overcome the negative impact of prophages through high conjugation rates. Overall, the presence of prophages introduces an additional death rate for plasmid carriers, the magnitude of which is determined in non-trivial ways by the environment, the phage and the plasmid.

Data have been collected from experimental work including conjugation and lysogenization assays, competition assays, 24h growth rate measurements.

Detection limit: To prevent overgrowing of plates we had to adjust the dilutions that we chose to plate out to be able to count CFUs for the total population density (which was several orders of magnitude higher than of individual fluorescent subpopulations). By doing so, we introduced a detection limit for several subpopulations, which, depending on the total population size, varied for the different time-points. In order to account for this detection limit, we divided the lowest dilution (that we plated and that still had countable colonies for each timepoint and plate-type) by two and assigned this value to the respective subpopulations.