Fig. 4: Loss of RNF20 or RNF40 causes reduced H3K4me3 occupancy on IBD risk genes including Vdr.

A Average binding profiles and heatmaps depicting occupancy (RPKM values) of H3K4me3 around the TSS ( ± 5 kb) of genes upregulated, unregulated and downregulated in wild type, Rnf20^fl/fl and Rnf40^fl/fl IECs. Genes were sorted high to low based on H3K4me3 occupancy under wild type conditions. Binding profiles revealed decreased H3K4me3 occupancy specifically in the gene body after Rnf20 or Rnf40 loss leading to peak narrowing. B Genes which lost H3K4me3 occupancy in Rnf20^fl/fl and Rnf40^fl/fl IECs were compared for overlap with IBD susceptibility genes [27, 28] expressed in IECs. H3K4me3 occupancy of 19% of IBD risk genes is dependent on RNF20 and/or RNF40. C Heatmap depicting gene expression patterns of a subset of IBD susceptibility genes in wild type, Rnf20^fl/fl and Rnf40^fl/fl IECs (padj ≤ 0.05, red: upregulated (log2 FC ≥ 1), blue: downregulated (log2 FC ≤ −1)). D Regions that lost H3K4me3 occupancy in Rnf20 and Rnf40 null IECs were compared to previously published ChIP-seq experiments using ChIP Atlas (http://chip-atlas.org) displaying a high overlap between our datasets and VDR-bound regions in the murine small intestine (SI) [29]. E By analyzing publicly available VDR ChIP-seq data, VDR-bound regions in murine SI were identified (75,303 regions) [29] and compared to regions displaying decreased H3K4me3 occupancy in Rnf20 (1,019 regions) and Rnf40 (2,128 regions) knockout IECs. 79.8% (813/1,019 regions) of Rnf20- and 77.6% (1,651/2,128 regions) of Rnf40-dependent H3K4me3 regions were also occupied by VDR. F Binding profiles of VDR in SI [29] and H3K4me3 in IECs revealed H3K4me3 peak narrowing under Rnf20 or Rnf40 knockout conditions. G Reduced H3K4me3 occupancy in the Vdr gene body in Rnf20^fl/fl and Rnf40^fl/fl IECs was confirmed by ChIP-qPCR (n = 4). One-way ANOVA, mean ± SEM.