Directed evolution is a powerful method used to create novel biomolecules. Evolving new proteins involves diversification through random mutations and selection for variants with desired traits. However, the protein-sequence space is so vast, even the most extensive libraries only cover a minute fraction of all possible sequences. Directed and natural evolution relies on a parent protein that already displays the desired or a similar function to some extent to be selected. However, some proteins evolve more readily than others. What determines evolvability of proteins in directed, and natural evolution is not understood and our ability to predict how certain mutations influence protein function remains crude at best.
We developed a novel method to simultaneously characterise phenotype and genotype for individual protein variants within large libraries at the single-molecule level. This will allow us to study the evolution of proteins with single-molecule sensitivity and uncover details previously obscured by insensitive screening methods.
Using a cell-free expression system, we generate protein variants with covalently attached unique DNA-based barcodes. We identify variants using our novel sequencing-by-hybridisation approach that allows single-molecule genotyping of 103–104 variants. The DNA-based barcodes consist of multiple single-stranded DNA indices. Sets of fluorescent hybridisation probes, complementary to the different barcode indices can be used to read out these barcodes.
We validate our method to study a small library of phylum Planctomycetes ClpS-variants. ClpS is an N-end adapter family protein that has previously been used for single-molecule protein sequencing. This ClpS binds N-terminal leucine, isoleucine, and valine. We study up to 20 variants in a single experiment, measuring molecular association and dissociation constants. We identify novel ClpS variants with different binding kinetics and identify residues responsible for N-terminal peptide binding and release on the single-molecule level.
Our method can be scaled up to screen the phenotype of thousands of protein variants in one single-molecule experiment. This will allow us to create evolutionary landscapes, study evolvability, and generate new and improved proteins for applications in biotechnology.