Phage-assisted evolution of degron tags and botulinum neurotoxins for targeted protein degradation and delivery, alongside base editing for Rett syndrome

Abstract

The work described in this thesis spans three research projects centered on directed evolution and precision genome editing. The first two projects develop phage-assisted continuous evolution (PACE)-based hybrid circuits to evolve novel degron tags and botulinum neurotoxin (BoNT) receptor-binding domains (RBDs) for targeted protein degradation and cell type–selective delivery. The third project focuses on adenine base editing strategies to precisely correct Rett-causing mutations in vitro and in vivo. Chapter 1 is organized into four sections. The first provides background on the ubiquitin- proteasome system, targeted protein degradation, and chemically inducible degron tags. The second introduces BoNT biology and summarizes prior efforts to reprogram BoNT specificity and repurpose them as delivery vehicles. The third section introduces PACE, with an emphasis on hybrid circuit architectures relevant to the work described here. Finally, the fourth section reviews the development of adenine and cytosine base editors, beginning with the discovery and adaptation of CRISPR-Cas9. Proteolysis-targeting chimeras (PROTACs) and molecular glue degraders have greatly expanded the scope of druggable proteins. However, extending these modalities to new proteins of interest (POIs) often requires costly and time-consuming discovery campaigns to identify selective and potent ligands. To bypass bespoke ligand discovery, researchers have developed degron tags, protein domains that can be genetically fused to a POI and render it susceptible to molecule-induced degradation. While existing degron tag technologies are powerful, many fall short of ideal properties, including compact tag size, favorable ligand properties, and reliance on endogenous E3 ligase machinery. Chapter 2 describes our use of PACE to develop SD40, a 36- amino acid degron tag that responds to an otherwise inert CRBN-based molecular glue. SD40 functions as both an N- and C-terminal tag and enables rapid and robust degradation of exogenous and endogenous fusion proteins in the presence of PT-179, a highly selective degrader. We also developed SD56, a mouse-compatible degron tag that supports robust depletion of fusion proteins with performance similar to SD40. Together, this work establishes chemically inducible degron tags that address key limitations of existing systems and introduces a PACE circuit architecture compatible with molecular glue interactions. Chapter 3 applies PACE-based hybrid circuits to evolve BoNT RBDs for targeted cytosolic delivery. BoNTs are exceptionally efficient and selective at delivering their native protease cargos into motor neurons. Furthermore, multiple reports have demonstrated that BoNTs are highly amenable to engineering, enabling both reprogramming of protease substrate specificity and delivery of diverse protein cargos. Nonetheless, native BoNT delivery remains largely restricted to neuronal cell types, limiting broader applications of engineered proteases and alternative payloads. Here, we use PACE to directly evolve the BoNT/B RBD to bind alternative human receptors with high affinity. We implement a peptide-based evolution strategy that circumvents the challenges of expressing mammalian ectodomains in the E. coli cytosol, enabling selection against receptor-derived epitopes while preserving translation to full-length ectodomain binding. The evolved RBDs bind their intended targets with high affinity, exhibit loss of affinity for their native receptor (human synaptotagmin I), and support receptor- dependent delivery both in exogenous and endogenous contexts. We also describe an orthogonal retargeting in which the RBD is replaced with de novo designed minibinders, enabling similarly receptor-dependent, cell-type-specific delivery. Although in vivo validation is needed, this work establishes a general strategy for constructing PACE-compatible circuits to evolve binders against mammalian receptors and provides novel methods for developing BoNTS with reprogrammed tropism. Finally, chapter 4 describes base editing strategies for precise correction of pathogenic mutations implicated in Rett syndrome. Rett syndrome is a severe neurodevelopmental disorder caused by mutations in MECP2 that disrupt its native function. Therapeutic options remain limited; Trofinetide is the only FDA-approved drug and, although effective, provides patients with modest clinical improvement. Emerging approaches, including gene therapies, seek to address Rett syndrome at its root cause, but MECP2 dosage must be tightly controlled, as overexpression can result in a similar disease manifestation, as observed in MECP2 duplication syndrome. This dosage sensitivity complicates gene replacement strategies, where controlling expression levels in vivo is challenging. Precision genome editing is well suited for treating Rett because it can correct the endogenous gene, effectively restoring WT-like levels. We developed highly efficient adenine base editing strategies for three common Rett-causing mutations, with minimal bystander editing. These strategies achieved highly efficient correction in engineered cell lines and patient-derived fibroblasts, and select candidates were advanced to a humanized MECP2R270X mouse model where treatment extended lifespan. Collectively, this work establishes a set of candidate strategies for correcting recurrent pathogenic MECP2 mutations and provides initial in vivo evidence supporting the therapeutic potential of base editing for Rett syndrome.