Identification of transcription factors for drug-associated gene modules and biomedical implications

Abstract

One of the major findings in systems biomedicine is that both pathogenesis of diseases and drug mode of action have a module basis. However, the transcription factors (TFs) regulating the modules remain largely unknown. In this study, by using biclustering approach FABIA (factor analysis for bicluster acquisition), we generate 49 modules for gene expression profiles on 1309 agent treatments. These modules are of biological relevance in terms of functional enrichment, drug-drug interactions and 3D proximity in chromatins. By using the information of drug targets (some of which are TFs) and biological regulation, the links between 28 modules and 12 specific TFs, such as estrogen receptors (ERs), nuclear factor-like 2 and peroxisome proliferator-activated receptor gamma, can be established. Some of the links are supported by 3D transcriptional regulation data [derived from ChIA-PET (chromatin interaction analysis using paired-end tags) experiments] and drug mode of action as well. The relationships between modules and TFs provide new clues to interpreting biological regulation mechanisms, in particular, the lipid metabolism regulation by ERα. In addition, the links between natural products (e.g. polyphenols) and their associated modules and TFs are helpful to elucidate their polypharmacological effects in terms of activating specific TFs, such as ERs, nuclear factor-like 2 and peroxisome proliferator-activated receptor gamma.