UC Berkeley

Gene therapy – the introduction of genetic material into cells and tissues of interest for a therapeutic purpose – has emerged as a very promising treatment for many diseases. Recent advances in genomics and proteomics, coupled with the advent of genome editing technologies, have generated an immense pool of potential nucleic acid cargoes that could be delivered as therapies for a wide array of diseases, ranging from monogenic disorders to cancer. However, before such therapies can be successful, a major hurdle must be overcome: the development of gene-carrying vehicles – also referred to as vectors – that can safely, efficiently, and specifically deliver those therapeutic payloads to the desired cells. The goal of this dissertation was therefore to address a major need in the field: the development of improved gene delivery vectors.

To date, more than 2,000 clinical trials employing gene transfer have taken place, establishing the safety of a number of vectors. Non-viral vectors can be easily produced at a large-scale and are amenable to the engineering of their chemical and physical properties via chemical modifications, but they suffer from a low delivery efficiency and cell toxicity. On the other hand, viral vectors harness the highly evolved mechanisms that viruses have developed to efficiently recognize and infect cells and offer several advantages that make them suitable candidates for use in gene delivery, both for therapeutic application and as tools for biological studies. In fact, gene therapy has enjoyed increasing success in clinical trials for numerous disease targets in large part due to the gene delivery capabilities viral vectors. Vectors derived from viruses have been used in the majority (over 68%) of gene therapy clinical trials to date, and the most frequently used have been based on adenovirus, retrovirus, vaccinia virus, herpesvirus, and adeno-associated virus (AAV).

AAV vectors are non-pathogenic and can transduce numerous dividing and non-dividing cell types. Because of these characteristics, AAV vectors have been utilized for gene therapy in various tissues. The amino acid composition of the viral capsid affects tropism (tissue specificity), cell receptor usage, and susceptibility to anti-AAV neutralizing antibodies – properties that influence efficacy in therapeutic gene delivery. However, AAV vectors can still encounter formidable impediments to efficacious gene delivery, including poor transduction (infection and expression of delivered gene) of some cell types, off-target transduction, difficulties with biological transport barriers, and potential risks associated with the integration of their genetic load. Extensive engineering of the AAV capsid promises to overcome these delivery challenges and improve numerous clinically relevant properties. To this end, the overarching goal of my work in the Schaffer Laboratory, which is presented in this thesis dissertation, was to advance current gene delivery methods through the engineering and characterization of novel adeno-associated virus vectors for gene therapy and research applications.

To access new viral capsid sequences with potentially enhanced infectious properties and to gain insights into AAV’s evolutionary history, we computationally designed and experimentally constructed an ancestral AAV capsid library. We performed selection for infectivity on the library, studied the resulting amino acid distribution, and characterized the selected variants, which yielded viral particles that were broadly infectious across multiple cell types. Ancestral variants displayed higher thermostability than modern (extant) natural AAV serotypes, a property that makes them promising templates for protein engineering applications, including directed evolution. Additionally, some variants displayed high in vivo infectivity on a mouse model, highlighting their potential for gene therapy.

Motivated by the success of directed evolution in the engineering of proteins with novel or enhanced properties, I worked in the engineering of AAV vectors for gene delivery to glioblastoma multiforme (GBM), a highly aggressive type of brain cancer. For this, I conducted directed evolution to select AAV variants with selective localization to and infectivity on GBM tumor cells and tumor initiating cells (TICs). Using an accurate GBM mouse model, I performed in vitro and in vivo selection, recovering viral particles that successfully trafficked to tumor cells and TICs in the brain after systemic administration to tumor-bearing animals. Following three rounds of in vivo selection, convergence was achieved upon several variants, the most abundant of which emerged from the ancestral reconstruction library. The selected variants are currently being characterized and assessed for their ability to deliver reporter and therapeutic genes, hopefully resulting in improved suppression of tumor progression compared to delivery with existing AAV serotypes. These novel vectors could enable new, potent therapies to treat GBM tumors and pave the way for engineering AAV vectors for other cancer targets.

In summary, this dissertation presents work on the development and characterization of a novel AAV capsid library, as well as on the implementation of this and of other libraries towards the engineering of novel AAV variants with selective gene delivery properties for brain tumors. The work herein presented aims to advance both the field of AAV vector engineering as a whole and the specific application of AAV vectors towards next generation cancer therapies.