Protein engineering often requires detailed mechanistic knowledge of the essential structure/function relationships for a given protein. Furthermore, obtaining detailed structure/function knowledge for proteins requires extensive structural analysis and biochemical characterization. Here, we have developed a high-throughput platform for the rapid identification of key functional regions of a viral protein shell or capsid. Adeno-associated viral (AAV) vectors have demonstrated considerable potential as gene delivery vectors to treat a broad range of diseases, including hemophilia and Parkinson's. However, both cellular (e.g. attachment and trafficking) and systemic (e.g. neutralizing antibody responses and tissue transport) mass transfer barriers limit efficient gene delivery of AAV vectors. A greater understanding of viral structure/function relationships would benefit both rational design and directed evolution approaches to engineer enhanced vectors. To date, more than 100 different natural variants of AAV have been isolated from both human and non-human tissues. The sequence variation in the viral protein capsid of these variants confers a broad range of gene delivery properties including binding to a variety of cell surface receptors such as heparan sulfate proteoglycan (HSPG), sialic acid, fibroblast growth factor receptor (FGFR), and platelet derived growth factor receptor (PDGFR). Furthermore, exhaustive rational mutagenesis techniques have successfully mapped regions of the AAV2 capsid which are involved in HSPG binding and putatively, FGFR binding. However, the lack of extensive structure/function knowledge of other AAV variants has hindered our ability to engineer more effective variants and understanding of the basic biology of AAV infection.

Accordingly, we have developed a novel, rapid, high-throughput platform to identify regions of the capsid of any AAV variant that are involved in a specific viral function. Specifically, we have generated highly diverse AAV libraries based on AAV5 and AAV6 through random mutagenesis of the cap gene and selected for mutants that are defective in specific steps of viral infection. For example, to map key residues involved in receptor binding, we selected the libraries for variants with decreased cell binding in iterative rounds of binding to CHO cells. Characterization of clones from the selected libraries yielded >15 novel variants of both AAV5 and AAV6 that exhibit decreased affinity for CHO cells. Interesting, a majority of these residues cluster to distinct regions of the capsid for AAV5 as compared to AAV6. Furthermore, several of these variants demonstrate altered gene delivery efficiencies to mutant CHO cell lines lacking sialic acid or HSPG, suggesting some of these residues are necessary for viral binding to sialic acid and other cellular receptors. This work demonstrates that our novel forward genetics platform is an efficient and effective approach to map functional regions of the AAV capsid and potentially other proteins, further our knowledge of basic AAV biology, and enhance efforts to engineer viruses and other proteins with customized properties.