Characterization and evolution of artificial RNA ligases

Abstract

Enzymes enable biocatalysis with minimal by-products, high regio- and enantioselectivity, and can operate under mild conditions. These properties facilitate numerous applications of enzymes in both industry and research. Great progress has been made in protein engineering to modify properties such as stability and catalytic activity of an enzyme to suit specific processes. On the contrary, the generation of artificial enzymes de novo is still challenging, and only few examples have been reported. The study and characterization of artificial enzymes will not only expand our knowledge of protein chemistry and catalysis, but ultimately improve our ability to generate novel biocatalysts and engineer those found in nature. My thesis focused on the characterization of an artificial RNA ligase previously selected from a library of polypeptide variants based on a non-catalytic protein scaffold. The selection employed mRNA display, a technique to isolate de novo enzymes in vitro from large libraries of 1013 protein variants. The artificial RNA ligase catalyzes the formation of a phosphodiester bond between two RNA substrates by joining a 5'-triphospate to a 3'-hydroxyl, with the release of pyrophosphate. This activity has not been observed in nature. An initial selection carried out at 23°C yielded variants that were poorly suitable for biochemical and biophysical characterization due do their low solubility and poor folding. We hence focused our studies on a particular improved ligase variant called ligase 10C, isolated from a subsequent selection performed at 65°C. Here we report the structural and biochemical characterization of ligase 10C. We solved the three-dimensional structure of this enzyme by NMR. Unexpectedly, the original structure of the parent scaffold used for building the original library was abandoned. The enzyme instead adopted a novel dynamic fold, not previously observed in nature. The structure was stabilized by metal coordination, yet lacked secondary structural motifs entirely. We also compared the catalytic and thermodynamic properties of ligase 10C to enzyme variants previously selected at lower temperature (23°C). Ligase 10C displayed a remarkable increase in melting temperature of 35°C compared to its mesophilic counterpart. In addition, its activity at 23°C was about 10-fold higher compared to the mesophilic variants. This work was the first mRNA display selection for catalytic activity at high temperature, and further highlighted the capacity of the technique to select for proteins with rare properties. To facilitate detailed mechanistic studies of this unnatural enzyme, a crystal structure would be essential. Unfortunately, ligase 10C did not form crystals likely due to its highly dynamic regions. With the goal of identifying a truncated less flexible version of the enzyme that would be more suited for crystallization, we generated a library of random deletion variants of ligase 10C and performed an mRNA display selection to identify shorter active variants. Finally, we describe the attempted selection of an enzyme for the same RNA ligation reaction from a completely random polypeptide library. The long-term goal of the overarching project in the Seelig lab is to elucidate and compare the structure and mechanism of enzymes generated from different starting points, yet catalyzing the same reaction, to obtain insights into potential evolutionary pathways. In summary, our work revealed the unusual structural and biophysical properties of the artificial ligase 10C, and thereby demonstrated the power and flexibility of mRNA display as a technique for the selection of de novo enzymes.