Abstract 4647: Identifying drug targets in sarcoma using machine learning and cell phenotype-based compound screening

Abstract

Abstract High-throughput phenotypic screens that incorporate compound biochemical activity annotations are positioned for novel target discovery in cancer. We used machine learning approaches to correlate kinome-wide profiling data of a collection of kinase inhibitors with phenotypic cell proliferation data for the same compounds. We assembled a library of profiled kinase inhibitors with diverse chemotypes and kinome-wide selectivities and determined their anti-proliferative activities in a panel of sarcoma cell lines. Using a previously published machine learning algorithm, we related the compound inhibition profiles across 237 kinases to their abilities to inhibit cell proliferation. This identified Protein Kinase D (PRKD) as a putative novel target kinase selectively in synovial sarcoma cell lines, such that its inhibition leads to a decrease in proliferation in these cells. Next, we reasoned that our approach could be leveraged to identify targets that, when co-inhibited, induce a synergistic phenotype. This would enable rational design of drug combinations for further testing as opposed to the labor-intensive, random, pairwise testing commonly performed. To identify targets synergistic with PRKD, we performed a screen of a synovial sarcoma cell line in the presence of a selective PRKD inhibitor. Several kinases became prioritized as new targets in this “synergy screen.” Combining selective inhibitors for each synergistic target, as defined by Chou-Talalay, (here, CDK and AKT) with PRKD inhibitors synergistically reduced synovial sarcoma cell proliferation. Conversely, combining PRKD inhibitors with a selective inhibitor of a kinase that was deprioritized in the synergy screen (p38 MAPK) resulted in non-synergistic or even antagonistic effects. Overall, our approach provides a promising framework to identify new targets of cancer subtypes and a novel methodology to identify new combinational strategies for treatment. Citation Format: Eric J. Lachacz, Zhi Fen Wu, John L. Bixby, Vance P. Lemmon, Sofia D. Merajver, Hassan Al-Ali, Matthew B. Soellner. Identifying drug targets in sarcoma using machine learning and cell phenotype-based compound screening [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 4647.