Systematic Characterization of the "Resistome" of Tumor Cells Against Panels of Small Molecule Inhibitors — Application in the Context of Inhibitors of Transcriptional Cyclin-Dependent Kinases (tCDKs) and Implications for Future "Individualized" Clinical Applications

The development of CRISPR/Cas9-based functional genomic approaches have greatly accelerated the pace of research on characterizing the molecular mechanisms of tumor cell resistance to diverse classes of therapeutics. We reasoned that this technology can be applied to systematically and prospectively identify candidate genes regulating tumor cell sensitivity vs. resistance against a collection of inhibitors with overlapping vs. distinct specificities against different structurally- and functionally-related molecular targets; and that the signature of genes emerging from such genome-wide resistance screens could also exhibit overlapping vs. distinct features consistent with the profile targets of the respective compounds. We applied this hypothesis in the context of a collection of small molecule inhibitors against transcriptional cyclin-dependent kinases (tCDKs). This choice was specifically motivated by our recent observations that a collection of transcription factors selectively drive the biology of multiple myeloma (MM) cells (compared to other neoplasias): while direct and selective small molecule inhibitors against these myeloma-selective transcription factors are not currently available, the expression of some them or their target genes is regulated by tCDKs, such as CDK7. We evaluated the CDK7/12/13 inhibitor THZ-01, the CDK7-selective inhibitor YKL05-124 and the CDK12/13 dual inhibitor MFH2-90-1 (referred herein as YKL and MFH respectively); and confirmed that these inhibitors exhibit in vitro activity against MM cells, with variable potency across cell lines. For our CRISPR/Cas9-based screens, we utilized MM.1S cells engineered to stably express the Cas9 nuclease and lentivirally transduced with the pooled genome-wide sgRNA library Brunello (~70,000 sgRNAs, 4 sgRNAs per gene; plus additional non-targeting control sgRNAs); and exposed the MM.1S-Cas9+ sgRNA+ pools of MM cells to each of these compounds (2 separate screens for THZ-01; 1 for YKL and 1 for MFH). Treatments consistent of successive rounds of treatment (at each compounds IC₂₅ dose for anti-MM activity) for several weeks of culture until treatment-resistant populations of tumor cells emerged. PCR and next-generation sequencing we performed, as in other studies of our group and others, to quantify the relative enrichment or depletion of sgRNAs against different genes in treatment-resistant cells vs. sensitive treatment-naïve cells. We documented concordance of results between the 2 THZ-01 resistance screens; and pronounced overlap, but also distinct differences between the genes associated with resistance or sensitization to THZ vs YKL (consistent with their common effect on CKD7, but the lack of CDK12/13 inhibition by YKL) or between THZ-01 vs. MFH; while more limited overlap was observed between MFH and YKL, consistent with their different specificities. The collection of genes associated with resistance to CDK7 or CDK7/12/13 inhibition contain several recognizable, but previously understudied, members or regulators of the CDK family; as well as a diverse group of other genes with known or putative roles in regulation of cell cycle, cell survival, autophagy, proteostasis and intracellular trafficking. Importantly, both the overlapping and distinct "resistome" signatures identified by CRISPR/Cas9 screens for these compounds were different from "resistome" signatures which we have obtained from similar genomewide CRISPR studies for resistance against other classes of investigational or established therapeutics for MM. Our study exemplifies the feasibility of applying CRISPR/Cas9 gene-editing based treatment resistance screens in order to not only define the molecular networks regulating the sensitivity vs. resistance of tumor cells to the respective treatments, but also to provide insights into known or previously underappreciated aspects of the specificity of these compounds against diverse structurally related targets. We envision that similar applications of this approach can be applied to prospectively identify which molecules mediate the antitumor activity of pharmacological agents for which their targets are not currently defined.