Massively Parallel Approaches to Functional Characterization of Natural and Engineered Viruses

Abstract

Next-generation sequencing technology has transformed all areas of molecular biology and fueled a dramatic increase in the availability of viral genomic sequences. Assessing the functional properties of viral genes and genetic elements, however, remains a major bottleneck, and linear investigation of single candidates of interest is increasingly insufficient to keep pace with the demands of basic and translational virology. Assays that parallelize the evaluation of thousands or millions of viral sequences offer an opportunity to scale up characterization of both naturally occurring viruses and engineered viruses with therapeutic potential for gene delivery. In this work I develop and apply high-throughput screens for two purposes: the development of engineered vectors for gene therapy and the identification and characterization of cis-regulatory translation control strategies across naturally occurring viruses. Recombinant adeno-associated virus (AAV) vectors are promising vehicles for gene delivery to treat a wide variety of genetic disorders through gene replacement, gene editing, or gene silencing, but naturally occurring AAV capsids present significant limitations for therapeutic applications. Engineered AAV capsids can expand the repertoire of available vectors possessing desirable properties with respect to tissue tropism and off-target effects. High-throughput approaches to capsid engineering that harness the power of next-generation sequencing to maximize the search space of possible capsid variants hold great promise for this end. In Chapters 2 and 3, I present DELIVER (Directed Evolution of AAV capsids Leveraging In Vivo Expression of transgene RNA), a massively parallel approach to capsid engineering that yields capsids with an enhanced ability to express a transgene in a particular tissue and cell type. In Chapter 2, I describe the properties of the MyoAAV class of capsids that were developed using DELIVER and potently transduce skeletal and cardiac muscle cells of both mice and juvenile macaques. These capsids have immediate potential for clinical applications in the treatment of degenerative musculoskeletal disorders with a genetic basis. I also identify the interaction between MyoAAV and the integrin aVb6 heterodimer as the key determinant of MyoAAV's muscle-specific tropism. In Chapter 3, I apply DELIVER to develop the PAL class of AAV capsids that demonstrate enhanced transduction of the central nervous system (CNS) and reduced off-target transduction of the liver in juvenile macaques. The PAL capsids are among the first reported engineered capsids to show an enhanced ability to transduce the primate CNS compared to naturally occurring capsids. In Chapter 4, I apply a high-throughput technique to a more exploratory problem in molecular biology by using a massively parallel reporter assay to investigate upstream open reading frames (uORFs) in viruses. uORFs—short ORFs initiating 5' of a main coding sequence (mCDS)—are well-known cis-acting negative regulators of translation in eukaryotes. In many cases, uORFs can dynamically respond to changing intracellular conditions and derepress the downstream mCDS in response to specific signals. Despite their importance in key pathways in eukaryotes, uORFs remain understudied in viruses. Our massively parallel approach facilitates systematic characterization of regulatory sequences from hundreds of human-associated viruses, a feat that would be intractable through more traditional experimental means. I provide quantitative evidence that regulatory uORFs are common across a broad range of viruses, are actively translated by ribosomes, and share some fundamental properties with eukaryotic uORFs. Taken together, this dissertation presents a multi-pronged exploration of the utility of massively parallel approaches in the study and translational application of virology.