Abstract PR15: Predicting the essentialities of protein-protein interactions in cancer

Abstract

Abstract Many protein-protein interactions (PPIs) are essential to cellular proliferation and survival, and thus represent potential targets for novel compounds with high specificity and therapeutic potential in the treatment of cancer. Despite their therapeutic significance, there are currently no high-throughput experimental platforms to interrogate the essentiality of individual protein-protein interactions. The ability to computationally predict or infer PPI essentiality would help prioritize the development of therapeutic targets and advance understanding of cancer biology, and remains a significant barrier. We have developed a computational method to predict PPI essentiality by combining shRNA studies with network models of protein interaction pathways in an analytic framework. High-throughput single-gene shRNA silencing is a well-established experimental approach to study protein essentiality in genome-wide screens. The silencing of a single gene in an shRNA screen effectively disrupts multiple PPIs, masking their individual contributions to the observed protein essentiality. Our method uses a network model to infer cliques of PPIs that are disrupted as each gene is silenced, then uses these sets to formulate an additive model of how the unknown PPI essentialities combine to form the observed protein essentialities. This process is then deconvoluted to recover the unknown essentialities of each PPI using a regularized solver. We performed this analysis in 108 cell lines characterized for protein essentiality in Project Achilles [1, 2]. A superpathway of 2186 proteins and 11488 protein interactions was constructed from prior knowledge databases [3] and protein interaction screenings. We demonstrate the validity of our approach via prediction of essential PPIs such as TP53-MDM4 in cell lines sensitive to nutlin and CDK4-RB1 interactions in cell lines sensitive to CDK4 inhibitors. [1] Cowley, Weir & Vazquez, et al. Nature Scientific Data 1, Article number: 140035. September 30, 2014 [2] Cheung HW, Cowley GS, Weir BA, et al. Proc Natl Acad Sci. 2011 Jul 26; 108(30):12372-7. [3] Schaefer CF, Anthony K, Krupa S, et al. Nucleic Acids Research. Jan 2009. 37:674-9 Citation Format: Lee AD Cooper, Josue D. Moran, Zenggang Li, Yuhong Du, Sahar Harati, Andrey A. Ivanov, Phillip Webber, Jonathan J. Havel, Margaret A. Johns, Haian Fu, Carlos S. Moreno. Predicting the essentialities of protein-protein interactions in cancer. [abstract]. In: Proceedings of the AACR Special Conference on Computational and Systems Biology of Cancer; Feb 8-11 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 2):Abstract nr PR15.