Mathematical comparison of protocols for adapting a bacteriophage to a new host.

Interest in phage therapy-the use of bacterial viruses to treat infections-has increased recently because of the rise of infections with antibiotic-resistant bacteria and the failure to develop new antibiotics to treat those infections. Phages have shown therapeutic promise in recent work, and successful treatment minimally requires giving the patient a phage that will grow on their infecting bacterium. Although nature offers a bountiful and diverse supply of phages, there have been a surprising number of patient infections that could not be treated with phages because no suitable phage was found to kill the patient's bacterium. Here, we develop computational models to analyze an alternative approach to obtaining phages with new host ranges-directed evolution via laboratory propagation of phages to select mutants that can grow on a new host. The models separately explore alternative directed evolution protocols for phage variants that overcome three types of bacterial blocks to phage growth: a block in adsorption, temperate phage immunity to superinfection, and abortive infection. Protocols assume serial transfer to amplify pre-existing, small-effect mutants that are initially rare. Best protocols are sensitive to the nature of the block, and the models provide several insights for enhancing success specific to each case. A common result is that low dilution rates between transfers are beneficial in reducing the mutant growth rate needed to ascend. Selection to overcome an adsorption block is insensitive to many protocol variations but benefits from long selection times between transfers. A temperate phage selected to grow on its lysogens can evolve in any of three phenotypes, but a common protocol favors the desired changes in all three. Abortive infection appears to be the least amenable to evolving phage growth because it is prone to select phages that avoid infection.