Identification of Molecular Markers of Tumor Cell Sensitivity and Resistance to Natural Killer Cells through Genome-Wide CRISPR Activation and CRISPR Editing Screens in Multiple Myeloma Cell Lines: Implications for Anti-Myeloma Immunotherapy

Natural killer (NK) cells represent a promising immunotherapeutic approach as they can potently kill tumor cells without triggering graft-versus-host reactions. Indeed, infusion of high numbers of NK cells, either autologous or allogeneic, after their ex vivo expansion and activation, has been feasible and safe in clinical studies. However, prior studies and early clinical trials indicate that tumor cells can exhibit decreased response to NK due to the protective effect of nonmalignant mesenchymal stromal cells; and depending on the genetic background of the tumor cells. To our knowledge, since earlier subgenome-scale RNAi-based studies, there have been no genome-wide CRISPR-based screens to identify candidate markers conferring tumor cell resistance or sensitivity to NK cells in multiple myeloma (MM). To address this void, and building on a recent loss-of-function (LOF) study by our group on solid tumors, we sought to identify genes regulating the response of MM cells to the cytotoxic activity of NK cells by conducting a genome-wide CRISPR/Cas9-based gene editing (Brunello library of sgRNA) and gene activation (Calabrese library of sgRNA) screens in MM.1S cells co-cultured with primary NK (pNK) cells (effector-to-target [E:T] ratio of 3.75:1) derived from healthy donor peripheral blood mononuclear cells (PBMCs) cultured in vitro in GMP SCGM medium with IL-2. Briefly, MM.1S cells engineered to stably express the nuclease SpCas9 (Brunello) or a catalytically inactive programmable RNA-dependent DNA-binding protein (dCas9)-VP64 (Calabrese) were also transduced with lentiviral particles for a pool of ~70,000 (Brunello library) or ~120,000 (Calabrese) sgRNAs, targeting exons of ~20,000 genes (plus non-targeting control sgRNAs), under conditions of transduction which allow for an average of no more than 1 sgRNA to be incorporated in a given cell. This allowed us to convert the initial population of MM.1S cells into heterogeneous pools in which each gene is subject to individual LOF or gain-of-function (GOF), due to Cas9-mediated editing, by only 1 sgRNA. Flow cytometry was performed to verify pNK viability, purity (CD56 and CD3), and expression of p46 receptor, surrogate marker of NK cell activity. These screens identified genes whose knock-out (Brunello sgRNA library) or activation (Calabrese sgRNA library) led to NK cells resistance or potential sensitivity. The hits observed in the current MM-oriented study exhibited, compared to our similar studies in solid tumor model, substantial gene level differences, but notable overlap at the pathway level (including death receptor pathways, NK activating pathways), which suggests that mechanisms determining tumor cell response vs. resistance to NK cells operate through modules consistent across tumors, but manifested through potentially different members of the respective pathways in different neoplasms. For instance, in this MM-oriented study, we identified that NK cell sensitivity of tumor cells is modulating by activation of several metabolic and homeostatic genes, receptor kinases, and interestingly membrane-bound proteins of the mucin family, e.g. MUC1, and MUC4, which have been reported to play a role in NK-mediated tumor killing in other types of cancer. MUC1 in particular has a clinical relevance as a small molecule inhibitor with prior preclinical studies in MM is available. Interestingly, our GOF screen identified as potential NK cell sensitizers TNFRSF10B, a death receptor related to TNFRSF10A (a hit identified in our studies in solid tumors), the putative death receptor adaptor TRADD, and the NK ligands PVR and ULBP1. Interestingly, genes such as PTEN and TP53, commonly associated with high-risk MM, didn't affect the response to NK cell, suggesting that NK cell-based therapies may potentially have a role in treatment of MM patients with high-risk clinical or biological features. In conclusion, this is the first study applying both LOF and GOF genome-wide screens to NK cell response in MM. The combination of such screens performed in parallel provide complementary and orthogonal information that allows us to identify genes that might not have been appreciated if only either LOF or GOF alone screens had been performed. We envision that the methodology and results presented herein will provide a framework towards validation of molecular markers which can help to optimize and individualize the use of NK cell-based therapy in MM.