Genomics of rapid adaptation to antibiotics: Convergent evolution and scalable sequence amplification.

Evolution may lead to extremely fast trait changes, especially in response to high selection intensities. A prime example is the surge of antibiotic resistance in bacteria. The genomic underpinnings of such rapid changes may provide information on the genetic processes that enhance fast responses and the particular trait functions under selection. Here, we combine our previously published data with newly sequenced genomes of experimentally evolved \textit{Escherichia coli} for a refined dissection of the genomics of rapid adaptation to antibiotics. Our new results identified repeated patterns of convergent evolution. Firstly, we confirmed our previous finding that amplification of a 316 kb genomic region, containing several known antibiotic resistance genes, is enriched under high antibiotic stress such as that exerted by drug combination. Importantly, our novel analyses revealed sequence amplification to be scalable yet costly: Higher dosage of antibiotic combinations coincided with higher levels of sequence amplification, which significantly dropped within 24\,h in the absence of drugs. Secondly, convergent evolution was also found in single drug treatments, for which mutational changes accumulated in genes connected to the AcrA-AcrB-TolC drug efflux pump and those possibly mediating DNA integrity. Convergent evolution was observed across organizational levels, ranging from sequence amplification, prevalence of individual mutations, high variant frequencies in specific genes, to the unusual repeated and independent occurrence of a particular silent Glycine codon mutation. We conclude that constrained evolutionary trajectories underlie rapid adaptation to antibiotics, whereby sequence amplification appears to represent the most potent, albeit costly genomic response mechanism to high antibiotic stress.