Skip to main content
Fig. 5 | Genome Biology

Fig. 5

From: Loss of MLL3/4 decouples enhancer H3K4 monomethylation, H3K27 acetylation, and gene activation during embryonic stem cell differentiation

Fig. 5

Distal H3K4me1 and H3K27ac are not functionally coupled with formative transcriptional activation. A Naive ChromHMM emissions for 16 states yields all combinations of H3K4me1 and H3K27ac in WT and DKO. Heatmaps are clustered by each chromatin state. Black arrows denote states with MLL3/4-dependent H3K27ac. B Gene expression changes of nearest TSS for all sites in each given state. Monte Carlo permutation sampling of “all genes control” to perform Mann-Whitney U test for each state. Benjamini-Hochberg corrected. Selected statistics shown, additional statistics provided in Additional file 2. C Same as A with Formative samples. D Same as B with Formative RNA-seq. All heatmap values and range are in CPM. For metagene analysis, the range in CPM is the same as shown in heatmap for each factor. ChromHMM states without any H3K4me1 or H3K27ac emission probability are excluded. E For all H3K4me1/H3K27ac+ sites in either Naive (top panel) or Formative (bottom panel), the total number of MLL3/4-dependent enhancers for a gene compared with the fold change of RNA levels DKO/WT in Log2CPM. Each dot represents one gene. Blue line represents generalized linear model, gray 95% confidence interval. F Boxplots of relative expression DKO/WT of RNA levels for all genes associated with any H3K4me1/H3K27ac+ peak in either the naive (top panel) or formative state (bottom panel). Each gene is binned by the percentage of their associated enhancer loss in DKOs. *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001

Back to article page