Directed Evolution and High Throughput Evaluation of the CRISPR sgRNA

Abstract

CRISPR editing is inefficient at over two thirds of genetic targets. A primary cause is misfolding between the spacer and scaffold regions of the gRNA, which hinders the formation of functional ribonucleoprotein complexes (RNPs) with Cas9. While current efforts to overcome low efficiency have focused on either ortholog discovery, chemical sgRNA modifications, and Cas9 protein mutagenesis. Exploring the vast sequence space (1E48+) within the single guide RNA scaffold (sgRNA) remains in its infancy, posing important questions of how these bacterial systems can be optimized for use in eukaryotic settings. There is an abundance of CRISPR orthologs, with thousands discovered to date and many developed by artificial intelligence, however, their optimization and characterization remains time and resource intensive, with methods that accelerate these processes representing an area of significant potential. While Cas9 protein mutagenesis has effectively reduced off target effects and expanded protospacer adjacent motif (PAM) recognition activities, new sgRNAs have not been optimized to the novel protein variants. With the arrival of large data artificial intelligence approaches structural biology, the lack of large datasets outside of naturally occurring CRISPR orthologs greatly limits its progress. Herein, we apply systematic evolution by Exponential enrichment to explore the sgRNA sequence space of the leading gene therapy gene editing variant saCAS9. We develop hundreds of highly efficient gRNA variant scaffolds utilizing a novel saBLADE (saCAS9 binding and ligand directed enrichment) methodology, which leverages asymmetric editing target dissociation over positive and negative rounds of directed evolution. This work furthers the original BLADE system developed in the Sullenger lab by successfully utilizing variant gRNAs to edit genomic targets with activity exceeding that of the wildtype gRNA scaffold. Evolved gRNA scaffolds contain 7 to 42% mutations relative to the wildtype canonical scaffold and improve cellular gene editing efficiency over wildtype gRNA at all targets tested. We find our gRNA scaffold variants obtain the highest percent improvement at difficult to edit targets, maximizing activity rescue where it is most necessary.Next, we address the gene Crispr field’s bottleneck in evaluating gRNA efficiency, by repurposing saBLADE directed evolution to measure CAS9 enzymatic activity over a single round of enrichment in a process we term the saBLADE single round SELEX activity assay. Our activity assay integrates the selection methods from saBLADE multiple round evolution into a single round. SaBLADE offers a significantly lower cost compared to current cellular screens, while maintaining strong correlation with cellular RNP activity (R² = 0.778). Our arsenal of saBLADE sgRNAs, and saBLADE activity assay showcases the power and flexibility of directed evolution to accelerate the development of gene editors, enabling efficient editing at previously unattainable sequences.