High-content imaging and network modeling maps drug-induced regulation of cardiac fibroblast phenotype

Abstract

Cardiac fibroblast activation is necessary for wound healing following myocardial infarction. To better understand how candidate fibrosis therapeutics affect fibroblast signaling, we treated human cardiac fibroblasts with combinations of drugs and pathologically relevant cytokines. Using high-content microscopy, we measured 137 image features at the single cell level. Using dimensionality reduction and clustering on these features, we identified fibroblast phenotypic responses that deviate from the classical axis of fibroblast quiescence and activation. We then employed a regression-coupled network modeling approach to predict regulators of cell features that were affected by candidate drugs.