Diseases of genetic origin can be treated by correcting underlying errors at the nucleic acid level.Genome editing has become a widely used approach, although there are limitations to current methods. The potential for off-target edits, delivery barriers, immune stimulation, and concerns over the feasibility of redosing limit the effectiveness of genome editing technologies. More recently, interest has grown in employing endogenous human enzymes for the correction of pathogenic mutations. The Adenosine Deaminase Acting on RNA (ADAR) family of human enzymes offer therapeutic potential due to their ability to convert adenosine to inosine in double stranded RNA. Human ADARs can be directed to predetermined target sites in the transcriptome by complementary guide strands, allowing for the correction of disease-causing mutations at the RNA level. A particular advantage of RNA editing as opposed to DNA editing is that any potential off-target transcript editing does not result in a permanent change to the genome. These studies aim to develop a mechanistic understanding of the protein-nucleic acid contacts that enable ADAR to edit RNA, and to exploit this knowledge in the design of optimized guide strands for site-directed RNA editing. Chapter 1 gives a broad introduction to ADAR enzymes and their use in directed RNA editing. In Chapter 2, we explore a specific ADAR-RNA contact that has a large influence on the rate of reaction. This understanding helps us to rationalize pathophysiologic conditions that are associated with dysregulated RNA editing. Moreover, understanding the mechanistic basis of this important interaction allows for the design of modifications to the guide strand in this position in Chapter 3. The single nucleobase modification we identified leads to over a 3-fold increase in the directed editing yield via endogenous ADARs. This result advances the approach of recruiting endogenous ADARs for site-directed RNA editing. It is typical for directed editing with ADARs to focus on sites within a 5’-UAG-3’ sequence context, as this is the natural substrate preference of ADARs. However, in Chapter 4 we expand upon previous work to support ADAR activity within editing-resistant sequence contexts containing a 5’ guanosine. We found that pairing this 5’ guanosine across a purine in the guide strand led to between an 8- and 60-fold rate enhancement. In addition, through in vitro studies with modified purine analogs we identified positions on the purine base that have a strong influence on the ADAR reaction, to allow for the design of chemically modified nucleosides that promote even greater ADAR activity. This expands the scope of disease-causing mutations that can be effectively targeted by ADARs. The crystal structure of the ADAR2 deaminase domain bound to dsRNA inspired the work done in Chapters 2 through 5. However, there is no available high-resolution crystal structure of the other catalytically active enzyme in the family: ADAR1. Therefore, we generated Rosetta-based molecular models with constraints from biochemical data to define structural features unique to ADAR1. These models support the discovery of a novel zinc binding site present on the surface of the ADAR1 deaminase domain but absent in ADAR2. Furthermore, the models explain previously observed properties of the ADAR1 deaminase domain and suggest roles for specific residues present in a binding loop that is partially responsible for substrate selectivity. Given the success of the ADAR1 deaminase domain model in making predictions about the roles of specific residues in the protein, a model of the ADAR3 deaminase domain was also generated. Lastly, Chapter 6 details ongoing efforts to improve ADAR editing of non-ideal targets of disease relevance, while expanding our understanding of ADAR’s tolerance for chemical modifications within the guide strand that are critical for cellular editing.