Modular binder technology by NGS-aided, high-resolution selection in yeast of designed armadillo modules

• Published in: Proceedings of the National Academy of Sciences - PNAS 2024-07, Vol.121 (27), p.e2318198121
• Main Authors: Stark, Yvonne, Menard, Faye, Jeliazkov, Jeliazko R, Ernst, Patrick, Chembath, Anupama, Ashraf, Mohammed, Hine, Anna V, Plückthun, Andreas
• Format: Article
• Language: English
• Subjects: Affinity; Amino acid sequence; Amino acids; Armadillo Domain Proteins - genetics; Armadillo Domain Proteins - metabolism; Binders; Biological Sciences; Epitopes; Epitopes - genetics; High-Throughput Nucleotide Sequencing - methods; Modules; Next-generation sequencing; Peptide Library; Peptides; Peptides - chemistry; Peptides - genetics; Peptides - metabolism; Phenotypes; Protein Binding; Saccharomyces cerevisiae - genetics; Saccharomyces cerevisiae - metabolism; Yeast; Yeasts
• ISSN: 0027-8424; 1091-6490; 1091-6490
• DOI: 10.1073/pnas.2318198121

Establishing modular binders as diagnostic detection agents represents a cost- and time-efficient alternative to the commonly used binders that are generated one molecule at a time. In contrast to these conventional approaches, a modular binder can be designed in silico from individual modules to, in principle, recognize any desired linear epitope without going through a selection and hit-validation process, given a set of preexisting, amino acid-specific modules. Designed armadillo repeat proteins (dArmRP) have been developed as modular binder scaffolds, and we report here the generation of highly specific dArmRP modules by yeast surface display selection, performed on a rationally designed dArmRP library. A selection strategy was developed to distinguish the binding difference resulting from a single amino acid mutation in the target peptide. Our reverse-competitor strategy introduced here employs the designated target as a competitor to increase the sensitivity when separating specific from cross-reactive binders that show similar affinities for the target peptide. With this switch in selection focus from affinity to specificity, we found that the enrichment during this specificity sort is indicative of the desired phenotype, regardless of the binder abundance. Hence, deep sequencing of the selection pools allows retrieval of phenotypic hits with only 0.1% abundance in the selectivity sort pool from the next-generation sequencing data alone. In a proof-of-principle study, a binder was created by replacing all corresponding wild-type modules with a newly selected module, yielding a binder with very high affinity for the designated target that has been successfully validated as a detection agent in western blot analysis.