Abstract
As CRISPR effectors like Cas9 increasingly enter clinical trials for therapeutic gene editing, a future for personalized medicine will require efficient methods to protect individuals from the potential of off-target mutations that may also occur at specific sequences in their genomes that are similar to the therapeutic target. A Cas9 enzyme's ability to recognize their targets (and off-targets) are determined by the sequence of their RNA-cofactors (their guide RNAs or gRNAs). Here, we present a method to screen hundreds of thousands of gRNA variants with short, randomized 5' nucleotide extensions near its DNA-targeting segment-a modification that can increase gene editing specificity by orders of magnitude-to identify extended gRNAs (x-gRNAs) that effectively block any activity at those off-target sites while still maintaining strong activity at their intended targets. X-gRNAs that have been selected for specific target / off-target pairs can significantly out-perform other methods that reduce Cas9 off-target activity overall, like using Cas9 variants engineered for higher specificity in general, and we demonstrate their effectiveness in clinically-relevant gRNAs. Our streamlined approach to efficiently identify highly specific and active x-gRNAs provides a way to move beyond a one-size-fits-all model of high-fidelity CRISPR for safer and more effective personalized gene therapies.