Abstract 1349: Systematic deconvolution of kinase inhibitor profiles identifies synthetic lethal targets in ERBB2-mutant and BRD4-NUT rearranged cancer

The development of targeted therapies that efficiently inhibit cancer signaling pathways is one of the main goals of modern precision cancer medicine. Consequently, genetic and biological phenotypic data of in vitro screens are increasingly utilized to develop compounds directed against distinct oncogenic alterations. However, current targeted therapies are often limited to small genetically defined patient cohorts due to the very finite number of proteins amenable to direct chemical inhibition. An alternative approach is the exploitation of synthetic lethality, i.e. inhibition of an unaltered protein required for cell viability in a certain genetic background. Systematic chemo-genomic analyses of cancer cell lines have been shown to be suitable tools for the identification of novel synthetic lethal dependencies in cancer (Chan et al. Sci Trans Med, 2011; Sos et al. PNAS, 2012; Kim et al. Cell 2013). To systematically extend this strategy to non-small cell lung cancer (NSCLC) we characterized the efficacy of 1505 chemical compounds based on a variety of kinase inhibitor motifs in a high-throughput screen against 80 NSCLC cell lines. We extracted patterns of biological activity based on chemical and genetic information and found that potency and selectivity of compounds are strongly related to their molecular scaffold, but independent of their overall chemical complexity. We thereby discovered a sunitinib derivative that exhibited exquisite activity against ERBB2-mutant cell lines but was devoid of ERBB2 kinase activity. Instead a kinome scan and an shRNA screen suggested a mechanism of synthetic lethality by activity against NTRK family members. Moreover a CDK9 inhibitor was identified as selective and potent against a midline carcinoma cell line - a tumor entity characterized by recurrent BRD4-NUT gene fusions. Using additional cell lines we validated the upregulation of c-Fos and selective induction of apoptosis in BRD4-NUT positive midline carcinoma compared to control cell lines following CDK9 inhibition. This can augment existing therapeutic approaches, which have primarily focused on directly targeting the fusion product with bromodomain inhibitors, and offers a novel target in this entity. In conclusion, by systematically screening a large number of compounds against a panel of genetically well characterized NSCLC cell lines and incorporating chemical information we were able to derive structure activity relationships and to identify potential synth