Abstract PR08: Combinatorial CRISPR-Cas9 reveals many cancer synthetic lethal interactions are private to cell type

• Published in: Molecular cancer therapeutics 2017-10, Vol.16 (10_Supplement), p.PR08-PR08
• Main Authors: Shen, John Paul, Zhao, Dongxin, Sasik, Roman, Luebeck, Jens, Birmingham, Amanda, Bojorquez-Gomez, Ana, Kreisberg, Jason, Ideker, Trey, Mali, Prashant
• Format: Article
• Language: English
• ISSN: 1535-7163; 1538-8514
• DOI: 10.1158/1538-8514.SYNTHLETH-PR08

We developed a systematic approach to map human genetic networks by combinatorial CRISPR-Cas9 perturbation coupled to robust analysis of growth kinetics. We targeted all pairs of 73 cancer genes with multiplexed guide RNAs in two cell lines, testing 23,652 combinations. Numerous therapeutically relevant interactions were identified, most private to one cell line. These patterns replicated with combinatorial drugs at 80% precision. Thus, cellular context will be critical to synthetic-lethal therapies. Here, we combined multiplex targeting with array-based oligonucleotide synthesis to create dual-gRNA libraries covering up to 105 defined gene pairs. We conducted genetic interaction screens by transducing the dual-gRNA lentiviral library into a population of cells stably expressing Cas9, maintaining these cells in exponential growth over the course of four weeks, then sampling the relative changes in gRNAs at days 3, 14, 21 and 28 post-transduction. To robustly quantify gene fitness and genetic interactions, we developed a computational analysis framework that integrates all samples across the multiple days of the experiment. Using this method we evaluated all pairwise gene knockout combinations among a panel of 73 genes divided between tumor-suppressor genes (TSG) and cancer-relevant drug targets (DT), a subset of which were also verified oncogenes. Experiments were performed in two cancer cell lines: HeLa, a cervical cancer cell line driven by Human Papilloma Virus (HPV); and A549, a lung cancer cell line driven by KRAS G12S mutation. With nine gRNA pairs per combination, the library comprised 23,652 double gene knockout constructs and 657 single gene constructs; testing two replicates in each cell line yielded a total of 94,608 unique tests of interaction. Measurements of gene fitness (fg) were well correlated between biological replicates in the same cell line; Pearson r=0.96, p = 4.2×(10)^(-40), as were the π scores for significant genetic interactions; r = 0.75 p = 7.7×(10)^(-12). Moreover, we observed a significant correlation between the number of genetic interactions identified for a gene and its single gene fitness (HeLa: r = 0.77, p = 3.4×(10)^(-10); A549: r = 0.45, p = 0.0018), suggesting that network hubs may have increased functional importance to cancer cells relative to genes with fewer interactions; such a relationship has been previously observed in model organisms but not before in humans5. Interestingly, we found that the genetic interacti