Development of Enhanced Base Editors and Prime Editors through Protein Engineering and Directed Evolution

Abstract

Cytosine base editors (CBEs), adenine base editors (ABEs), and prime editors have already made tremendous impacts in the basic research community and are poised to transform medicine. However, to allow precision genome editing tools to reach their full potential, we must develop new techniques to better understand these complex molecular machines. Similarly, new technologies are needed to manipulate genome editors to make them simultaneously versatile, efficient, and exquisitely precise. This thesis attempts to address some of these problems using directed evolution and protein engineering. In Chapter 1, I review the original reports of base editors and prime editors, as well as subsequent advances to these tools. I also summarize how directed evolution has been employed previously to advance genome editing. I next discuss my work on tuning the on-target and off-target activities of cytosine and adenine base editors. In Chapter 2, I briefly describe a protein engineering campaign to improve base editors through codon optimization, enhanced nuclear localization, and ancestral reconstruction of the base editor deaminase domain. Next, in Chapter 3, I discuss efforts to measure and eliminate Cas9-independent off targets caused by cytosine base editors. This work was prompted by the discovery that the deaminase domain in CBEs can cause random, genome-wide off-target DNA mutations independent of Cas9 targeting. To better understand and measure this phenomenon, we developed four different cellular assays that can be used to assess Cas9-independent off targets in a high-throughput way. We then used these assays to screen different deaminase domains for favorable off-target editing profiles. We found that either catalytically impairing the deaminase domain or transiently delivering the CBE as a protein instead of a plasmid can yield efficient on-target editing while generating near-background levels of Cas9-independent off targets. In the second half of this thesis, I describe my work on enhancing different aspects of prime editing. In Chapter 4, I describe a generalizable, optimized protocol for performing prime editing on a new target of interest in mammalian cells. In Chapter 5, I report the development of a phage-assisted continuous evolution circuit for prime editors (PE-PACE). We use PE-PACE to evolve novel, compact RT domains that reduce prime editor size and improve prime editor activity. We also manipulated the PE-PACE circuit to evolve Cas9 nickases that can improve prime editing efficiencies. By refining PACE-evolved variants using protein engineering, we created next-generation prime editors, termed PE6a-PE6i. These variants offer robust editing activities across a variety of sites, edit types, and cell types, including patient-derived fibroblasts and primary human T-cells. In Chapter 6, I describe additional mechanistic insights and therapeutically relevant delivery strategies enabled by PE6 variants. By comparing PE6 variants to less active editors, we discovered that pegRNA secondary structure inhibits the prime editing activity of less processive RTs, such as the PEmax-deltaRNaseH enzyme that is used for dual-AAV mediated prime editing in vivo. Based on this result, we speculated that certain highly processive PE6 RT variants that are the same size as PEmax-deltaRNaseH could enable new classes of in vivo prime edits. To test this, we constructed dual-AAV delivery systems and performed neonatal intracerebroventricular injections of prime editor viruses. Sequencing genomic DNA from the brains of treated mice revealed that PE6 variants offer order-of-magnitude improvements in editing efficiency compared to the previous state-of-the-art editor PEmax-deltaRNaseH for the insertion of a 40-bp loxP sequence. We also show that PE6 variants enable, for the first time, in vivo twin prime editing. I close by discussing the questions that remain in the field of precision genome editing.