Quantitative Approaches to Bacteria Evolving in the Presence of Antibiotics
Some problems in antibiotic resistance can be tackled using quantitative approaches where mathematical modelling takes a central role. One of these is the question of whether mixing and cycling is best for antibiotic stewardship. Prof. Beardmore will discuss how this can be framed in the theory of optimisation and how that relates to clinical trials. Modelling can be helpful in the lab too, so he will show how to apply diffusion theory to disk diffusion assays and use quantitative imaging to derive nonlinear dose responses for the sizes of zones of inhibition.
In a similar vein, Prof. Beardmore will discuss a quantitative fluorescence imaging approach applied to videos of CFP/RFP/GFP-tagged strains of E.coli that produce data on the rapid evolution of efflux pump (AcrAB) expression in antibiotic gradients. Now, tetracycline antibiotics are a substrate of this pump and a so-called “hangover experiment” whereby a PhD student forgot to turn off a spectrophotometer until 48h of a growth-inhibition assay had passed, led to the following surprising result. Doxycycline didn’t reduce the growth rate of lab E.coli while the student nursed the hangover at home, instead it increased both lag phase and population density at 48h well beyond the untreated, no-drug control.
Prof. Beardmore’s group studied this apparent benefit from ribosome-binding antibiotics by extending the hangover experiment to a growth-death assay lasting weeks. This showed doxycycline and erythromycin interact with ROS to help mitigate population decline among E.coli populations, consistent with a “grow fast, die young tradeoff” described by others. It turns out a resistance mechanism can remove the population benefits these antibiotics bring. Finally, if there’s time, he will mention the group’s AI approach to the Pfizer dataset, ATLAS, that probes why CLSI in the US and EUCAST in Europe publish very different clinical breakpoints for the same antibiotic-pathogen pairs.