A systems pharmacology approach to determine the mechanisms of action of pleiotropic natural products in breast cancer from transcriptome data

Plant-derived compounds as natural products have attracted a lot of attention in the treatment of complex diseases, especially cancers, primarily due to their poly-pharmacologic mechanisms of action. However, methodological limitations have impeded gaining complete knowledge of their molecular targets. While most of the current understanding of these compounds is based on reductive methods, it is increasingly becoming clear that holistic techniques, leveraging current improvements in omic data collection and bioinformatics methods, are better suited for elucidating their systemic effects. Here, to provide an explanation to the mechanisms of action of plant-derived natural products in breast cancer, we applied a data integration approach to comprehensively study oncogenic signaling pathways targeted by withaferin A, actein, compound kushen injection and indole-3-carbinol. Specifically, we mapped the transcriptome-level response of cancer cell lines to these molecules on a human protein-protein interaction network and constructed the underlying active subnetworks. We used these subnetworks to define the perturbed signaling pathways and validated their relevance in carcinogenesis. The similarity of each identified oncogenic signaling pathway in terms of overlapping genes was subsequently used to construct pathway-pathway interaction networks, which were used to reduce pathway redundancy and to identify pathway crosstalk. Filtered pathways were then mapped on three major carcinogenesis processes. The results showed that the pleiotropic effects of plant-derived drugs at the gene expression level can be used to predict targeted pathways. Thus, from such pathways, it is possible to infer a systemic mechanism of action of such natural products.