Biphasic adaptive evolution of antimicrobial resistance

A considerable body of work exists on the molecular biology of antimicrobial drug resistance, but general population dynamic principles of resistance evolution in microbial populations are still poorly understood. We study an empirically-motivated model of microbial evolution on drug-concentration dependent fitness landscapes. At any concentration a genotype’s fitness depends on two phenotypes, the resistance level and the drug-free growth rate, with a tradeoff between these two. Adaptation on these landscapes at fixed drug concentration occurs in two phases. At first, the system evolves in a smooth region of the landscape with a rapid accumulation of resistance mutations accompanied by a decrease in the drug-free growth rate. In the second phase, there is partial recovery of the drug-free growth rate while the resistance level remains nearly constant. This recovery occurs on the rugged part of the landscape, is slow and involves the exchange of high-cost resistance mutations for low-cost ones. All changes occur via sequential origin-fixation events without any intermediate loss in fitness. We call this process exchange compensation , and note that it is distinct from the well-known phenomenon of fitness compensation that occurs through mutations at secondary sites when a resistant strain is re-cultured in a medium without drugs. The patterns of evolution in our two-dimensional phenotypic space of resistance and drug-less growth rate are independent of model details and therefore of wide relevance to understanding the tempo and mode of compensatory evolution.