Computational correction of copy number effect improves specificity of CRISPR–Cas9 essentiality screens in cancer cells

Abstract

The CRISPR-Cas9 system has revolutionized gene editing both on single genes and in multiplexed loss-of-function screens, enabling precise genome-scale identification of genes essential to proliferation and survival of cancer cells1,2. However, previous studies reported that a gene-independent anti-proliferative effect of Cas9-mediated DNA cleavage confounds such measurement of genetic dependency, leading to false positive results in copy number amplified regions3,4. We developed CERES, a computational method to estimate gene dependency levels from CRISPR-Cas9 essentiality screens while accounting for the copy-number-specific effect. As part of our efforts to define a cancer dependency map, we performed genome-scale CRISPR-Cas9 essentiality screens across 342 cancer cell lines and applied CERES to this dataset. We found that CERES reduced false positive results and estimated sgRNA activity for both this dataset and previously published screens performed with different sgRNA libraries. Here, we demonstrate the utility of this collection of screens, upon CERES correction, in revealing cancer-type-specific vulnerabilities.

Novel functional insights revealed by distinct protein-protein interactions of the residual SWI/SNF complex in SMARCA4-deficient small cell carcinoma of the ovary, hypercalcemic type

“…Fileset. doi:10.6084/m9.figshare.7655150) 46 . The CERES score for each of these genes in COV434 and BIN67 cell lines were compared to the average effect of each gene knock-out across all 625 cell lines in the database using a Z score calculated as Z=(x -μ)/σ where x is the CERES score for the gene of interest in either COV434 or BIN67, μ is the average CERES score for the gene of interest across all cell lines in the database, and σ is the standard deviation of the CERES score for the gene of interest across all cell lines in the database.…”

“…In order to determine synthetic lethal dependencies of protein interactors exclusive to the residual SWI/SNF in SCCOHT cells, we analyzed how CRISPR knock-out of genes encoding SCCOHTexclusive SWI/SNF-interacting proteins affected growth of SCCOHT cell lines compared to the average effect across all cell lines in the database using publicly available data 46 . Depletion of 6 genes, highlighted in Figure 9, selectively reduced growth of both BIN67 and COV434.…”

Precision Targeting of BFL-1/A1 and an ATM Co-dependency in Human Cancer

“…First, we conducted LDH release assays upon treating U937 cells with the corresponding serial dilution of D-NA-NOXA SAHB-15 R31E and observed no non-specific membrane-lytic activity, as measured at 30 min and 4 hr (Figures S2F and S2G). We also conducted the above-described cytotoxicity studies on additional cancer cell lines that exhibit dependency on BFL-1, such as OCI-AML-3 AML and Jurkat T cell leukemia cells (Kim et al, 2005; Meyers et al, 2017), and observed pharmacologic synergy similar to that seen in U937 cells (Figures S3A–S3D). Next, we tested two additional ATM inhibitors, Ku-60019 (Golding et al, 2009) and Ku-55933 (Hickson et al, 2004), and noted the same pattern of synergy in U937 cells (Figures S3E–S3H) but not in MV4;11 cells (Figures S3I–S3L) as observed for AZD0156 (Figure 4D–4I).…”

“…BFL-1 dependency analyses were performed using the reported genome-scale CRISPR-Cas9 loss-of-function screening dataset (Meyers et al, 2017). We defined 2,027 genes as likely pan-essential genes, whose dependency scores fall in the bottom 23% of gene scores in at least 90% of the cell lines evaluated.…”

“…Here, we harnessed a series of design principles to transform cysteine-reactive NOXA SAHBs into cell-permeable constructs with precision selectivity for BFL-1, revealing a powerful and reliable workflow for optimizing stapled peptides. Given the importance of combination therapies in overcoming the apoptotic blockades of human cancer, we further explored BFL-1 co-dependencies (defined as those gene dependencies that are correlated with a dependency on BFL-1 ) by mining the publicly available CRISPR-Cas9 dependency map database (Meyers et al, 2017). The genetic analyses informed the discovery of an unanticipated synergy between our selective BFL-1 inhibitor and molecular targeting of the ataxia-telangiectasia-mutated (ATM) kinase in BFL-1-dependent acute myeloid leukemia (AML).…”