Recreation of evolutionary dynamics of Escherichia coli growing in the presence of different erythromycin concentrations during 5 days.

Antibiotic dose responses are used to determine treatments that cure infection and mitigate resistance. Despite an 'inverted-U' dose-response geometry described in the literature whereby slow bacterial adaptation occurs at the highest and lowest dosages with fastest adaptation in between, competing hypotheses claim that both sub- and super-inhibitory dosages hasten resistance. We therefore treated Escherichia coli with the antibiotic erythromycin from zero to high dose to elucidate the genomic and phenotypic basis of the inverted-U and so reconcile different views that relate dose to resistance progression. We observed fastest drug resistance adaptation at the minimal inhibitory concentration (MIC) and genotype-phenotype correlations determined from deep sequencing data reveal the molecular basis of this: simultaneous selection for copy number variation in 3 resistance mechanisms. All drug treatments saw E. coli amplify the antibiotic efflux operon acr at a rate that correlated strongly with population density increases and a second efflux operon, emrE, exhibited an analogous pattern of selection. However, lower dosages led to amplification of drug targets, namely the ribosomal RNA operons and these mechanisms created 3 inverted-U geometries within one evolutionary dataset, one for each mechanism under selection. Importantly, we demonstrate that inverted-U relationships are not a universal feature of dose-resistance relationships: unexpectedly, following an acr knockout, resistance adaptation was fastest for that strain at the very lowest antibiotic dosages.