Drug effect prediction based on phosphoprotein interactions

Abstract

Drug effect prediction requires understanding of protein functions and interactions, and protein interactions are affected by cell types and post-translational modification, namely phosphorylation. A specific post-translationally modified form of protein product is called a proteoform and should be distinguished from other proteoforms arising from the same protein. A cell line-specific and retinoic acid treatment-specific protein-protein interaction network was constructed by integrating the traditional protein-protein interaction information and proteoform information and experimental transcriptome and phosphoproteome profiles for two breast cancer cell lines MCF-7 (retinoic acid-sensitive) and BT-474 (retinoic acid-resistant) each before and after treatment. The effect of retinoic acid was computed from measuring the shortest path length from the drug targets to the sample-specific network comparing the length before and after the treatment. Despite the lack of statistical significance, the decrease of the average shortest path length to MCF-7-specific proteins may reflect the sample-specific proteins to be affected in favor of retinoic acid treatment compared to that of BT-474-specific proteins.