PNNL

Abstract

Protein kinases drive carcinogenesis; their genetic alterations, including activating mutations and copy-number amplification, can guide targeted therapies. Yet, kinase inhibitors target proteins, and proteomic analyses are required to validate genomic observations and reveal new treatment opportunities. Herein, we developed and applied a new algorithm, OPPTI, to elucidate the landscape of overexpressed kinases across 10 cancer types using curated global mass spectrometry proteomics data of 1,071 cases. ERBB2 and EGFR protein overexpression correlated with their respective genomic amplifications. Meanwhile, overexpression of CDK4/6 and MET proteins frequently occurred in cancer types with few corresponding DNA- and RNA-level alterations. Overlapping overexpressed kinases with CRISPR screen data, we identified 23 currently- druggable kinases and nominated 11 new targets, including NEK8 for breast cancer; PRKRA, HK2, and NRBP2 for lung cancer, RIOK3 for gastric cancer, and MAP4K1 for endometrial cancer whose expressions were validated by immunochemistry. Finally, we demonstrated that in vitro and in vivo models of hepatocellular carcinoma with higher PDGFRB protein, a kinase overexpressed in selected patient tumors, exhibited improved survival upon targeted treatment. Overall, our results reveal that protein-level alterations may not be readily observed by genomics, represent cancer vulnerabilities, and need to be carefully considered as potential targets in precision oncology.