(567t) Optimization of High-Risk Adeno-Associated Virus (AAV) Libraries for Directed Evolution of Novel Gene Vectors

Adeno-associated virus (AAV) has demonstrated marked success as a gene therapy vector, due to its unique safety profile. Directed evolution approaches to AAV capsid design have generated targeted AAV variants without a priori knowledge of respective targets. Techniques for generating virus diversity have included error-prone PCR, DNA shuffling, and peptide display, leaving random insertion a largely unexplored arena for creating useful virus diversity. High-risk approaches like random insertion are complicated by the intricate assembly of the virion, which is composed of 60 intercalating capsid subunits. Recent pioneering reports have demonstrated the severely compromised packaging fitness of random deletion/duplication libraries of AAV5 generated using random insertion by DNase I. This scenario presents a steep, but interesting design challenge in creating high-risk AAV libraries for directed evolution, where wild-type (wt) exhibits a stark advantage during virus production. Consequentially, virus library preps are dominated by wt even though the plasmid libraries used to produce them may contain very little wt genes. Here, we characterize the mutant-wt population dynamics of a random domain insertion library using a simple model that describes their behavior during virus production. We then explore design approaches that maximize the fitness of AAV random insertion libraries during production and selection. These studies provide critical insight into high-risk AAV library design and provide a methodical framework for addressing wt competition during virus library production.