Genome-scale measurement of off-target activity using Cas9 toxicity in high-throughput screens

Abstract

CRISPR-Cas9 screens are powerful tools for high-throughput interrogation of genome function, but can be confounded by nuclease-induced toxicity at both on- and off-target sites, likely due to DNA damage. Here, to test potential solutions to this issue, we design and analyse a CRISPR-Cas9 library with 10 variable-length guides per gene and thousands of negative controls targeting non-functional, non-genic regions (termed safe-targeting guides), in addition to non-targeting controls. We find this library has excellent performance in identifying genes affecting growth and sensitivity to the ricin toxin. The safe-targeting guides allow for proper control of toxicity from on-target DNA damage. Using this toxicity as a proxy to measure off-target cutting, we demonstrate with tens of thousands of guides both the nucleotide position-dependent sensitivity to single mismatches and the reduction of off-target cutting using truncated guides. Our results demonstrate a simple strategy for high-throughput evaluation of target specificity and nuclease toxicity in Cas9 screens.

Figure 1. Design and performance of Cas9 library.

(a) Schematic of library design indicating design of gene-targeting guides, safe-targeting guides, and non-targeting guides. (b) Growth screens were performed in duplicate in three cell lines, and a ricin screen was performed in duplicate in a single cell line. (c) Precision/recall curve for performance on gold-standard essential genes. These curves graphically display the trade-off between the fraction of genes correctly identified as essential (precision) and the fraction of all essential genes identified (recall). Untreated conditions were compared to plasmid library composition and replicates were combined and analysed with casTLE. Previous 4-guide duplicate screen in K562 included as reference. (d) Schematic of nucleotide sugar and n-glycan synthesis genes in ricin screen results. Ricin treated conditions were compared to untreated conditions in K562, and replicates combined and analysed using casTLE. Blue boxes indicate the gene knockout protected the cell from ricin while red boxes indicate the gene knockout sensitized the cell to ricin. NS indicates non-significance (q>0.1). White boxes indicate that these genes are known to be on-pathway but were not identified as ricin regulators. (e) Quantile-quantile plot showing altered distribution of P values using safe-targeting or non-targeting control in growth screens. P values are calculated from both replicates using casTLE.

Figure 2. Off-target detection using nuclease-induced toxicity.

(a) Median and quartile range of full-length 19 and 20 bp guides in three cell lines by number of 1-mismatch off-targets. P values are from a Mann–Whitney test compared to guides with 0 off-targets. (b) Single mismatches closer to the PAM are less tolerated. Guides with no perfect off-targets and exactly one 1-mismatch off-targets were stratified by the location of the mismatch. Those with the location of the mismatch farther from the PAM site display greater disenrichment, demonstrating greater cutting and toxicity at these off-target sites. Error bars represent range of median results from two replicates. (c) Median and quartile range of truncated 17 and 18 bp guides in three cell lines by number of 1-mismatch off-targets. Guides with any 0-mismatch off-targets were excluded. P values are from a single-tailed Mann–Whitney test compared to guides with 0 off-targets. (d,e) Box plots of enrichment scores for guides targeting hit genes at 10% FDR for (d) growth screens and (e) a ricin screen. Signs have been flipped for ricin resistant genes for comparison. Box is length of quartile, whiskers represent 1.5 × quartile, and dots indicate enrichment of outlier guides. P values calculated using a single-tailed Mann–Whitney test. NS indicates non-significance (P>0.01).