Microbial evolution in drug gradients

Most experiments on the evolution of bacterial resistance to antibiotics are done in well-mixed "test tube" conditions. However, in real infections, the bacterial cells are often arranged in spatially structured biofilms, or are embedded in human tissues which restrict diffusion of drugs and nutrients. Because of this, spatial gradients of antibiotics are very likely to arise when treating real infections.
 
Inspired by experimental work from Bob Austin's group in Princeton, Philip Greulich, Bartek Waclaw and I developed a theoretical model that can explain how evolution of drug resistance can be greatly accelerated by a drug gradient. Basically, this happens because the population of bacteria spreads into the gradient as a series of waves, each one more resistant to the drug than the last. At the edge of a population wave, resistant mutants that emerge have a large advantage because the population density is very low and so they do not have to compete with neighbouring cells.
 
We also found that this mechanism only holds if each step in the "mutational pathway to resistance" involves increasing drug resistance; for pathways that involve a fitness valley, a drug gradient can actually slow down evolution of resistance. We are currently working on experimental realisations of this model, to test the relevance of its predictions for real drugs and real bacteria.