In vitro evolution of artificial enzymes: Method development and applications

Artificial enzymes have the potential to aid in the production of pharmaceuticals and facilitate basic biomedical research. There are two methods for making artificial enzymes: rational design and de novo selection. Rational design utilizes detailed knowledge of enzyme catalysis to design an enzyme active site, and then introduces this active site into a protein. However, due to the limited understanding of protein folding and structure-function relationships this approach is still extremely challenging and far from routine. In contrast, we utilize a directed evolution approach to isolate de novo artificial enzymes from a large library of protein variants by in vitro selection. Each of the trillions of proteins in a library are tested in a single experiment to determine if any have the desired activity. The artificial enzymes are created when the library is made so a high quality library is important for success. My thesis research focuses on two goals: (1) Construct a library built on the robust (alpha/beta)8 barrel enzyme scaffold for future enzyme selections and (2) Characterize a thermostable artificial RNA ligase and develop an application for this enzyme. The (alpha/beta)8 fold is used to catalyze a wide range of chemical reactions in nature. We used this fold to create a library containing > 1014 unique proteins by replacing loops of the catalytic face with randomized codons via PCR. Small sub-libraries were subjected to a protease-based folding selection to improve library quality by enriching for folded sequences. The final folding-enriched library contained > 1012 folded proteins representing an up to 50-fold improvement relative to a control library. These libraries will provide a valuable source of new enzymes for future in vitro selections. The previously generated artificial RNA ligases join 5'-triphosphate RNA to the 3'-hydroxyl of a second RNA substrate; a reaction not observed in nature. However the enzymes were also highly dynamic, which prevented the solving of the protein structure by NMR or X-ray crystallography. A more structured enzyme, called ligase 10C, was isolated by performing the ligase selection at 65°C and its structure was solved revealing a novel primordial fold. Here, we describe the detailed biochemical characterization of ligase 10C. Using a variety of RNA substrates, we also determined how ligation rates change with sequence composition revealing an enzyme with broad sequence specificity. We developed a method for the specific ligation and sequencing of 5'-triphosphorylated RNA. These results highlight ligase 10C as an attractive tool for the selective isolation of 5'-triphosphate RNA from a complex mixture, something which is difficult with current methods.