The kinase ERK plays a conserved dominant role in the heterogeneity of epithelial-mesenchymal transition in pancreatic cancer cells.

Epithelial-mesenchymal transition (EMT) occurs heterogeneously among carcinoma cells to promote chemoresistance. Identifying the signaling pathways involved will nominate drug combinations to promote chemoresponse, but cell population-level studies can be misleading, and single-cell transcriptomics are limited to indirect ontology-based inferences. To understand EMT heterogeneity at a signaling protein level, we combined iterative indirect immunofluorescence imaging of pancreas cancer cells and tumors and mutual information (MI) analysis. Focusing first on mitogen-activated protein kinase pathways, MI indicated that cell-to-cell variation in ERK activity determined EMT heterogeneity in response to different growth factors and chemotherapeutics but that JNK compensated when MEK was inhibited. Population-level models could not capture these experimentally validated MI inferences. The dominant role of ERK was consistently indicated by MI even when the analysis was expanded to include seven potential EMT-regulating signaling nodes. More generally, this work provides an approach for studying multivariate signaling-phenotype relationships based on protein measurements in any setting.