Integration of Functional Dependencies and Molecular Alterations of Myeloma Cells

During the last two decades, multiple myeloma (MM) cell lines and patient-derived samples have been extensively profiled for alterations in their genome, transcriptome, epigenome, and proteome, with the anticipation that molecules with the most recurrent or pronounced dysregulation (e.g. compared to nonmalignant plasma cells) could represent attractive novel therapeutic targets. Functional genomics platforms, including CRISPR/Cas9 systems, provide direct quantitative information on the impact of gene perturbation on tumor cell survival and proliferation, and we evaluated the information provided from such genome-wide studies on MM cell lines in vitro to examine how much overlap exists in the landscapes of vulnerabilities vs. molecular alterations in MM cells. The comparison of CRISPR essentiality screens in MM cell lines (n=10) vs. 300+ non-MM cell lines from solid tumors and hematologic malignancies allowed us to identify 40+ genes representing preferential or more pronounced/recurrent dependencies for MM cells vs. non-MM cells. These genes prominently featured transcription factors (e.g. IRF4, PRDM1, MAF, IKZF1, IKZF3, NFKB pathway members ), epigenetic regulators (e.g. EP300), and kinases (e.g. PIM2, IKBKB). Our integrated molecular evaluation of these genes examined a series of molecular alterations, including transcript overexpression in MM cell lines (or patient derived samples) vs. normal plasma cells, or in relapsed/refractory vs. earlier stages of disease; mutational status; transcript levels correlating with shorter progression free or overall survival across multiple molecularly-annotated clinical trial data sets in MM; and extent of open chromatin (based on H3K27Ac chromatin marks) within or proximal to each gene in MM cell lines. While some of the candidate MM-preferential essential genes exhibited at least one of the aforementioned molecular alterations, the large majority of these genes either harbored no such alterations or were not ranked in the top 50-100 genes in terms of magnitude or frequency of these alterations. Notably <10% of MM newly diagnosed MM patients harbor mutations in any of these genes; and only 4 genes had mutations in >1 MM cell line examined in our screen. While some MM-preferential dependencies genes (e.g. IRF4, PRDM1, MAF) have higher transcript levels in MM vs. non-MM cell lines, most of the 300+ genes that fall in the latter category are not selectively essential for MM cells. We extended our study to genes reported to have recurrent non-synonymous mutations in MM: among the 50+ genes with >3% mutation rate in newly diagnosed MM patient-derived cells, only 5 genes appeared as a dependency in 2 or more MM cell lines tested and only 1 gene had CRISPR data compatible with a candidate role as negative regulator of survival/proliferation. Focusing the analysis on specific categories of mutations did not alter this conclusion. While we observed several examples of MM cells lines in which known oncogenic events (e.g. mutations in KRAS or NRAS) were associated with dependency on the respective genes, we also observed other examples in which dependency on such genes was identified even in their wild-type state and without identifiable loss-of-function events in known negative regulators. These observations collectively support the notion that functional genomics approaches can identify promising candidate therapeutic targets, which may not be readily identifiable based on established annotations of molecular alterations in the genome, transcriptome, or epigenome of tumor cells.