Network modeling of kinase inhibitor polypharmacology reveals pathways targeted in chemical screens
We measured changes in gene expression upon treatment of PANC1 cells with gemcitabine by RNA-seq. To understand the gene regulatory changes in response to gemcitabine treatment, we profiled DNaseI hypersensitivity for gemcitabine and vehicle-treated PANC1 cells. A) GO enrichment analysis for genes changing expression in response to gemcitabine (at the top in orange are genes up-regulated upon drug treatment, below in black are the genes down-regulated). B) Overlap between the genes that change expression when treated with gemcitabine (differentially expressed genes), genetic modifiers of the gemcitabine resistance (genetic hits) and targets of our hit kinase inhibitors that sensitize cells to gemcitabine (kinase hits). We find a modest overlap between the three sets, a trend observed before when comparing complementary high-throughput profiling approaches . C) Construction of a TF-DNA regulatory network using DNaseI hypersensitivity data collected for cells treated and untreated with gemcitabine. We called peaks on the DNaseI data (combined with existing data for the same cell line from the ENCODE project), and then scanned each peak for TF binding sites using Transfac motif matrices. We assign a TF-gene regulatory interaction if we find a TF motif in a DNaseI peak that is within 5kb of the gene’s transcription start site. D) Top 50 transcription factors enriched in the promoters of differentially expressed genes. For each transcription factor, we performed a Fisher’s Exact Test to ask whether we see an overrepresentation of the transcription factor’s associated motifs in the promoters of genes changing expression in response to gemcitabine, compared to its presence in all promoters harboring a DNaseI peak. Note: we show here only those TFs with motifs in promoters of more than 100 differentially expressed genes.