Engineering Gene Targeting Reagents Through Computational Design and Directed Evolution of Protein-DNA Interactions

Homing endonucleases have great potential as tools for targeted gene therapy and gene correction. These DNA cleavage enzymes are capable of inducing gene repair or disruption by stimulating DNA repair pathways at a specific location in a genome. However, identifying variants of these enzymes capable of cleaving specific DNA targets of interest is necessary before the widespread use of such technologies is possible. Enzyme engineering should be informed by detailed analyses of the wild-type system. In my early work redesigning the target site specificity of the homing endonuclease I-AniI, it was discovered that, despite the approximate two-fold symmetry of the enzyme-DNA complex, there is almost complete segregation of the interactions responsible for substrate binding and transition state stabilization. This separation of the roles of these domains was revealed by doing kinetic studies on target sites differing by a single base-pair from the wild-type substrate. Computationally redesigned variants of I-AniI that achieved new specificities on one side did so by modulating binding, while redesigns with altered specificities on the other side modulated catalysis. While some computational designs showed successfully altered specificity, many designs targeting other single base-pair substitutions were unsuccessful. The next step of my work involved developing better computational methods in order to improve recapitulation of the experimental data. I made improvements the ROSETTA program, giving an energy bonus to native-like interactions collected from the RCSB protein databank and developing protocols for sampling diverse design sequences. Despite advancements in both directed evolution and computational methods, protein engineering is challenging and exploring alternative methods for altering endonuclease specificity is necessary. New endonuclease ORFs can be identified due to substantial increases in available sequence data. I have additionally shown that enzyme hybrids can be built using homologue information, by transferring amino acids from homologues onto the I-AniI scaffold, improving activity and generating new specificities. This wealth of information from millions of years of endonuclease evolution will be used to guide and improve current rational engineering methods.