Deep Mutational Scan of a DNA Polymerase via Compartmentalized Self-Replication

We present a novel platform for high-resolution mapping of DNA polymerase activity and stability under the effects of harsh chemistries that are incompatible with most other mutational scanning methods. This approach pairs compartmentalized self-replication (CSR), a high-throughput method for polymerase directed evolution, with deep mutational scanning (DMS), a method of quantifying variant effect via next-generation sequencing of libraries that are subjected to a functional selection. We demonstrate the validity of this “CSR-DMS” platform by showing that it identifies loss-of-function variants at sites with known DNA binding or catalytic activity in the wild-type protein. We further explore the efficacy of this method by imposing denaturing selective pressures (heat or guanidinium thiocyanate) during screening and showing that variants with high positive enrichment under these selective pressures possess higher resistance to denaturation in activity assays. These variants may be useful for “direct” diagnostic workflows that detect biomarkers from crude sample matrices with little to no sample purification. Furthermore, the mechanisms of stabilizing mutations can be inferred from trends in the scores of similar mutations in sequence-to-function heatmaps and corroborated by the behavior of residues of interest in molecular dynamics simulations of the wild-type protein. These mechanisms, uncovered by CSR-DMS, inform future approaches to rational design of extremely stable DNA polymerases. Overall, we propose that CSR-DMS could be used both for the study of the biophysical mechanisms of selective pressures and for the engineering of polymerases with novel capabilities. A barcoded single amino acid variant library of DNA polymerase spanning its entire sequence was expressed in E. coli. Cells were encapsulated in oil-aqueous emulsion droplets and subjected to PCR cycling either alone, in conjunction with excessive heat treatment, or in conjunction with added guanidinium. The PCR cycling forced variant polymerases to amplify barcode sequences from their own coding plasmids under a "compartmentalized self-replication" selection scheme. The emulsion was broken after 20 cycles of PCR and barcoded DNA fragments were isolated and sequenced with Illumina sequencing. Barcodes were associated back to corresponding DNA polymerase variants by comparison with a barcode-variant map created by sequencing the variant library plasmid using PacBio HiFi long-read sequencing.