The large-scale genetic characterizations of human cancer have given us a detailed description of the genetic alterations across a broad spectrum of cancer. In many instances gain-of-function alterations of a certain frequency emerge immediately as candidate drug targets. These data, by themselves, however, do not directly provide us with a detailed functional understanding of those genes whose protein products are necessary for the maintenance of cancer viability. In order to more deeply understand the genes and pathways critical to the survival of cancer cells previous and ongoing efforts have provided functional annotation of the Cancer Cell Line Encyclopedia through the use of deep pooled shRNA or sgRNA libraries. For example, Project DRIVE (Deep RNAi-screening for Viability Effects in cancer) characterized the viability effects of knockdown of 7,837 genes using 20 shRNAs per gene across 398 cancer cell lines. The DepMap project targeted 18,333 genes in 625 cell lines using CRISPR/CAS9. The data from these efforts are highly robust identifying all known oncogenes mutated in the set of interrogated cell lines, along with multiple new cancer dependent features not previously described. Emergent notable targets from these efforts include PRMT5 and the WRN helicase both showing strong preferential dependence linked to mutational events.
Notably, few synthetic lethal nodes have emerged from these analyses downstream or within the critical oncogenic pathways that dominate the landscape of human cancer. We hypothesize that this is due to functionally redundant paralogous genes that, while potentially targetable by single therapeutics, are not targeted in current knockdown/knockout screens. This is exemplified by discordance between certain therapeutics and shRNA or sgRNA gene inactivation. For example, the cell line sensitivity profiles of MEK inhibitors which target MEK1 and MEK2, or CDK4/6 inhibitors do not robustly correlate with single gene knockout data.
Based on prior work using dual S. pyogenes and S. aureus CAS9 systems, we have built a pairwise sgRNA library targeting roughly 3000 genes and the resulting 5,000 paralog pairs. Initial screening data has demonstrated robustness of the methods. More importantly, new paralog dependencies relevant to NRAS mutant cancers have emerged.