Plasticity of fibroblast transcriptional response to physical and biochemical cues revealed by dynamic network analysis

Abstract

SummaryData on human skin fibroblast transcriptional responses to external cues were used to reconstruct dynamic gene regulatory networks. The goal of the reconstruction was to determine dynamic network interactions (quantitative predictive relationships of mutual regulatory influences of and on transcription factor expression) from time course data on 56 transcript expression levels obtained following different external cues. The inherently under-determined nature of this problem was addressed in part by excluding putative regulatory motifs that did not appear to be functional in multiple independent experiments from different independent external perturbations. Data were obtained from a previously published experiment in which the 56 transcripts were assayed by bioluminescence in live cells cultured on substrates of varying levels of stiffness and exposed to different levels of arginylglycylaspartic acid (RGD) peptide. The inferred dynamical networks were validated via comparison of predictions to a priori known interactions from gene databases. We discovered that exposures to different substrate stiffnesses and to RGD stimulate responses that are mediated through GATA4, SMAD3/4, ETS-1, and STAT5 and other genes, which can initiate hypertrophic, fibrotic, and inflammatory responses. The developed dynamical system identification method for discovering new mechanotransduction pathways is applicable to the identification of gene regulatory networks in numerous emerging applications where time-series data on multiple state variables and from multiple external perturbations are available.