Figure 4

Specific SRAM methylation patterns define epithelial (E) and mesenchymal (M) cell types. A) Hierarchical clustering of SRAMs for a subset of cell lines (n = 47) with RPPA data for E-cadherin (ECAD) segregates cell lines into high and low ECAD groups. Representative genes that are preferentially unmethylated (AXL) or methylated (CDH1, SPINT1) in the mesenchymal cells are shown. Illumina gene expression levels of EMT markers (ZEB1, Vimentin (VIM), Twist1, Fibronectin (FN1), CDH2, and CDH1) are overlaid to confirm the E vs M identity of the cell lines. B) Wilcoxon rank sum test was used to identify the SRAMs that are differentially methylated between E and M cells, which we call the EMT-SRAMs. The identified 135 probes (111 unique genes) were used for hierarchical clustering of the cell lines. EMT-SRAMs that are predominantly methylated in M cells are called M-SRAMs, and those methylated in E cells are called E-SRAMs. C) EMT-SRAM association with EMT network genes using curated network analysis. Closest second neighbor analysis of genes associated with core EMT factors (“EMT Hub genes”: CDH1, CDH2, FN1, SNAI1, SNAI2, GSC) finds numerous associations with genes within the EMT-SRAM set. Red spheres represent SRAMs that are methylated in M cells; green spheres are those methylated in E cells; white spheres represent linking genes.