CRISPR-Based Dual-Knockout Screens Reveals Network of Interactions between Key Dependencies for Myeloma Cells and Highlights IRF4 As a Central Regulator of Myeloma Cell Behavior

As multiple myeloma (MM) remains incurable, development of more effective novel therapies will require a deeper understanding of which genes, pathways and molecular networks thereof that govern the biological behavior of MM cells. The emergence of new CRISPR-based functional genomics approaches, including gene editing for loss-of-function (LOF) studies, allowed our group to obtain insight into the role of individual genes for MM survival and cell proliferation in vitro, in vivo and in the context of treatment resistance (e.g. De Matos Simoes et al., Shirasaki et al., Gandolfi et al. ASH 2017). Through these studies, we identified 50+ genes with more pronounced sgRNA depletion in MM cells vs. non-MM lines, which we consider MM-preferential essential genes. These genes include many transcription factors (TFs) such as IRF4, PRDM1, NF-κB, and MAF. We hypothesized that, while these transcription factors are individually important for MM cell survival and proliferation, they cooperate to regulate MM cell behavior in a manner that cannot be fully captured by 1st-generation, single knockout, CRISPR studies. To address this hypothesis, we performed CRISPR/Cas9 dual knockout (DKO) screens against a collection of ~100 genes, which included ~50 MM preferential dependencies; additional genes with broad-spectrum as dependency in in MM and other neoplasias (e.g. KRAS, BRD4, MCL1, BCL2); tumor suppressors (TP53, PTEN); genes which are frequently expressed in MM cells, but not are major dependencies for them in single knockout studies with CRISPR (e.g. ZBP1); as well as control sgRNAs). For these studies, we applied an "orthogonal Cas9" system: MM.1S cells expressing 2 different Cas9 nucleases (from S. pyogenes and S. aureus, respectively) received a pooled lentiviral library of constructs containing 2 sgRNAs per construct, with each sgRNA operating under one of these 2 Cas9 versions, to avoid possible imbalance in editing due to 2 sgRNAs competing for a single Cas9. Our custom library contained all the 5,000 pairwise combinations of dual KOs of the 100 selected genes and their respective single-gene KOs; with at least 4 sgRNAs per gene for each type of cas9, and for a total of ~54,000 single or double KOs represented in the screen. The study contained multiple different types of controls, including the ability to compare the single gene KO data contained within this DKO study vs. our single gene KO genome-scale screens, which we observed to be very similar. In our DKO study, IRF4 was identified as the most common and strongest synergistic partner with synergy scores (SynDKO score) <-1.0 in >50% of its pairwise interactions with other genes of our study: the synergistic partners of IRF4 included MM-preferentially essential genes, broad-spectrum dependencies and non-essential genes for MM cells, indicating functional interactions of IRF4 with a broad spectrum of genes and its importance as a master regulator of MM cells. Other highly recurrent partners for synergistic interactions included MM-preferential dependencies such as TCF3, ZBTB38, PIM2, IKZF1 and EP300; other chromatin remodeling regulators such as CREBBP and ARID1A; or the anti-apoptotic Bcl-2 family members BCL2L1and MCL1. Interestingly, we observed that dual CRISPR knockout for both IKZF1 and IKZF3 did not produce a stronger anti-MM effect than the dual KO of either gene with IRF4 or single KO IRF4 alone. These results suggest that dual genetic depletion of IKZF1 and IKZF3, which simulates the pharmacological depletion of these transcription factors by thalidomide derivatives, induces an anti-MM effect that is quantitatively less pronounced than the knock-down of IRF4, alone or in combination with LOF of either IKZF1 or IKZF3, suggesting the value of further efforts to develop novel therapies to potently and comprehensively suppress the activity of IRF4. Interestingly, we noted that LOF of TP53 or PTEN attenuates the effect of LOF for several essential genes examined in our study, but not IRF4. To our knowledge this is the first study that leverages the power of CRISPR editing to systematically examine the functional interactions between pairs of MM genes. Our results point to several transcription factors, chromatin remodeling genes and anti-apoptotic regulators as "nodes" for recurrent synergistic pairwise-interactions with other genes and importantly identify IRF4 as central regulator in the hierarchy of these interactions in MM cells.