Rapid creation of a novel protein function by in vitro coevolution.

We have developed a simple and efficient method for creation of novel protein functions in an existing protein scaffold. The in vitro coevolution method involves design of a hypothetical pathway for the target function followed by stepwise directed evolution of the corresponding protein along the pathway. As a test case, this strategy was used to engineer variants of human estrogen receptor alpha ligand-binding domain (hERalphaLBD) with novel corticosterone activity. Two steroids, testosterone and progesterone, that provide a progressive structural bridge between 17beta-estradiol and corticosterone, were chosen to assist the directed evolution of hERalphaLBD. A total of approximately 10(6) variants were screened in four rounds of random mutagenesis, resulting in two hERalphaLBD variants that respond to corticosterone. Creation of this new ligand activity required the presence of four simultaneous mutations. In addition, several required mutations were located outside the ligand binding pocket and yet exerted important action on ligand binding. Our results demonstrate the ability of in vitro coevolution to create novel protein function that is difficult or impossible to achieve by existing protein engineering approaches and also shed light on the natural evolution of nuclear hormone receptors. This in vitro coevolution approach should provide a powerful, broadly applicable tool for engineering biological molecules and systems with novel functions.