Figure 3: Global changes in the chromatin landscape between proliferative and invasive states.
From: Decoding the regulatory landscape of melanoma reveals TEADS as regulators of the invasive cell state
(a) Multidimensional scaling using RNA-seq, H3K27ac and FAIRE-seq data reveals a clear separation of the invasive samples, MM047 and MM099, from the samples in the proliferative state. (b) Mosaic plots obtained by clustering 55,919 regulatory regions show very similar chromatin profiles for MM047 and MM099, while the proliferative samples are characterized by higher heterogeneity. (c) Gene expression changes between invasive and proliferative samples correlate with changes in H3K27ac (for each gene from our signatures, the H3K27Ac differential peak called by MACS2 with the largest fold change in 20 kb around the TSS is selected). Spearman’s rank correlation of coefficient is shown. (d) Inverse correlation between H3K27me3 peaks and gene expression changes. (e) Aggregation plots of the read coverage (y axis) indicate that the TSS (x axis) of genes that are expressed higher in invasive samples (643 genes, left column) show higher FAIRE and H3K27ac signal but lower H3K27me3 signal in the same invasive (orange) samples than in the proliferative samples. Vice versa, the TSS of genes that are more expressed in the proliferative samples (772 genes, right column) show higher activating signals in the proliferative samples, and higher repressive signal in the two invasive samples. (f) Concordance between chromatin landscape in vitro and in vivo, where in vivo hypermethylated regions in the invasive samples from TCGA data show high activity (H3K27ac) in proliferative in vitro cultures (nine blue curves), but no activity in the invasive cultures (two orange curves).
