Interactions among beneficial mutations (i.e., epistasis) are often strong enough as to direct adaptation through alternative mutational paths. While alternative solutions should display similar fitness under the primary selective conditions, their properties across secondary environments may differ widely. The extent to which these cryptic differences are to be expected is largely unknown, despite their importance—for example, in identifying exploitable collateral sensitivities among antibiotic resistance mutations. Here we use directed evolution to characterize the diversity of mutational paths through which the prevalent carbapenemase KPC-2 can evolve high activity against the clinically-relevant antibiotic ceftazidime, an initially poor substrate. We identified 40 different substitutions, including many common in clinical settings, spread among 18 different mutational trajectories. Initial mutations determined four major groups into which the trajectories can be classified, a signature of strong epistasis. Despite similar final ceftazidime resistance, groups diverged markedly across multiple phenotypic dimensions, from molecular traits such as in-cell stability and catalytic efficiency to macroscopic traits such as growth rate and activity against other β-lactam antibiotics. Our results indicate that cryptic yet consequential phenotypic differences can accumulate rapidly under strong selection, unpredictably shaping the long-term success of resistance enzymes in their journey across hosts and environments.
Interactions among beneficial mutations (i.e., epistasis) are often strong enough as to direct adaptation through alternative mutational paths. While alternative solutions should display similar fitness under the primary selective conditions, their properties across secondary environments may differ widely. The extent to which these cryptic differences are to be expected is largely unknown, despite their importance—for example, in identifying exploitable collateral sensitivities among antibiotic resistance mutations. Here we use directed evolution to characterize the diversity of mutational paths through which the prevalent carbapenemase KPC-2 can evolve high activity against the clinically-relevant antibiotic ceftazidime, an initially poor substrate. We identified 40 different substitutions, including many common in clinical settings, spread among 18 different mutational trajectories. Initial mutations determined four major groups into which the trajectories can be classified, a signature of strong epistasis. Despite similar final ceftazidime resistance, groups diverged markedly across multiple phenotypic dimensions, from molecular traits such as in-cell stability and catalytic efficiency to macroscopic traits such as growth rate and activity against other β-lactam antibiotics. Our results indicate that cryptic yet consequential phenotypic differences can accumulate rapidly under strong selection, unpredictably shaping the long-term success of resistance enzymes in their journey across hosts and environments. Read More


