Engineering and application of modular chimeric tRNA synthetases in mRNA display

Abstract Macrocyclic peptide therapeutics (MPTs) have risen as a powerful tool for targeting “undruggable” proteins of interest. This class of compounds can mimic protein-protein interactions, disrupting shallow interfaces that would be otherwise difficult to target with small molecule drugs. Many naturally occurring MPTs contain non-canonical amino acids (ncAAs) in their core scaffolds, such as modified sidechains, backbone alkylations, β-linkages, etc. Including diverse building blocks in MPTs can confer greater specificity and binding affinity towards a target. Two primary techniques are used for including ncAAs into peptides, flexizyme and orthogonal tRNA synthetases (ORSs). While flexizyme can charge any tRNA with any pre-activated ncAA, this non-specificity prohibits utilization in situ, necessitating purification of charged tRNAs and limiting the maximum number of replaced codons to ~10. Conversely, ORSs are capable of catalytically recharging tRNAs with ncAAs in situ, but are typically limited to a single tRNA and a narrow scope of structurally similar ncAAs, requiring lengthy engineering campaigns to modify tRNA and/or ncAA specificity. Recently, a new class of chimeric ORSs has been reported, where a tRNA recognition domain (RD) is tethered to a catalytic domain (CD) by a flexible linker, allowing independent engineering of both domains. Here, I propose to combine the best aspects of both ncAA incorporation systems by engineering a system of modular chimeric tRNA synthetases (MoChi-RS) to allow “plug-and-play” swapping of tRNA recognition and catalytic function. Candidate tRNA-RD pairs were sampled from nature using a custom bioinformatics pipeline leveraging sequence similarity network analysis, AlphaFold, and genome-wide tRNA searching. These candidates will be assayed for co-orthogonality and engineered for diversity to allow tandem multi-codon replacement. In parallel, catalytic domains from reported ORSs, such as those capable of charging β-amino and α-hydroxy acids, will be screened and engineered for enhanced activity and modularity. Additionally, generation of a new flexizyme-mimetic catalytic domain will be targeted using in vivo selection, deep sequencing, and machine learning guided engineering. Finally, these developed tools will be used in an mRNA display campaign to search for novel and stable binders to the SH2 domains of Spleen-associated tyrosine kinase (SYK), a cytosolic kinase implicated in neuroinflammation, Aβ accumulation, and gliosis in Alzheimer’s disease.