Integrative Analysis and Refined Design of CRISPR Knockout Screens

Abstract

Genome-wide CRISPR-Cas9 screen has been widely used to interrogate gene functions. However, the analysis remains challenging and rules to design better libraries beg further refinement. Here we present MAGeCK-NEST, which integrates protein-protein interaction (PPI) and improves the inference accuracy when fewer guide-RNAs (sgRNAs) are available. MAGeCK-NEST also adopts a maximum-likelihood approach to remove sgRNA outliers, which are characterized with higher G-nucleotide counts, especially in regions distal from the PAM motif. Noticing that various replication cycles affect knockout effects, we further normalized MAGeCK-NEST output considering cell replication cycles. Normalized CRISPR-Cas9 screens using different libraries can thus be integrated as a ‘reference’, from which condition-specific hits could be derived. Moreover, we found that choosing non-targeting sgRNAs as negative controls lead to strong bias, which can be mitigated by sgRNAs targeting “safe harbor”, a region of the genome that is considered to be both transcriptionally active and its disruption does not lead to discernable phenotypic effects. Custom-designed screens confirmed our findings, and further revealed that 19nt sgRNAs consistently gave the best signal-to-noise separation. These methods and characterizations enabled development of an improved genome-wide CRISPR screen library and application in dissecting the mechanism of methyltransferase EZH2 inhibitors. Pharmacological inhibition of EZH2 preferentially suppresses the growth of lymphoma cells with activating mutations in EZH2 that augment PRC2-dependent silencing. However, it remains unknown whether these EZH2-targeting compounds have inhibitory effects in solid tumors that generally do not carry EZH2 mutations. In a panel of human prostate cell lines, we found those with competent androgen receptor (AR) signaling are sensitive to EZH2 inhibitors. However, in both sensitive and insensitive prostate cancer cells, inhibitor treatment significantly reduced global H3K27 trimethylation (H3K27me3) levels, suggesting a PRC2-independent mechanism. In sensitive CRPC cells, however, EZH2 inhibitors induce a specific gene signature that is highly associated with AR signaling. Compound treatment disrupted the interaction between EZH2 and AR, and impaired AR recruitment to its target gene loci. To further explore EZH2 function, we performed CRISPR-Cas9 screens in EZH2 inhibitor-treated and un-treated conditions. Modeling and pathway analysis suggested that EZH2 collaborates with base excision repair pathway, whose gene expression is down regulated by EZH2 inhibitors and promoters are enriched with PRC2-independent EZH2 bindings. Collectively, we used CRISPR screens to identify a novel function of EZH2 in prostate cancers.