Figure 4. FBXL19 is required for appropriate CDK8 occupancy at a subset of CpG island promoters.

(A) A schematic illustrating how addition of OHT (4-hydroxytamoxifen) leads to the generation of Fbxl19^ΔCXXC ES cells. (B) Western blot analysis of an OHT-treatment time course in the Fbxl19^fl/fl ES cell line. Hours of treatment are shown. FBXL19 protein was C-terminally tagged with a T7 epitope (Figure 2—figure supplement 1B, C) to allow for Western blot detection with an anti-T7 antibody. FBXL19 runs as a doublet that shifts in size following deletion of the CxxC domain. TBP was probed as a loading control. (C) Metaplots showing CDK8 enrichment in Fbxl19^fl/fl ES cells (WT) and Fbxl19^ΔCXXC ES cells (OHT) at all CDK8 peaks (left), peaks showing decreased CDK8 occupancy (↑, middle) and peaks showing increased CDK8 occupancy (↓, right). The number of total peaks in each group is indicated. p-values denote statistical significance calculated by Wilcoxon rank sum test comparing ChIP-seq read counts between WT and OHT samples across a 1.5-kbp interval flanking the center of CDK8 peaks. (D) Screen shots showing ChIP-seq traces for CDK8 in wild type Fbxl19^fl/fl ES cells (WT) and Fbxl19^ΔCXXC ES cells (OHT). BioCAP and FS2-FBXL19 tracks are given for comparison. Left – CDK8 peaks showing reduced CDK8 binding in Fbxl19^ΔCXXC ES cells; Right – CDK8 peaks showing increased CDK8 binding in Fbxl19^ΔCXXC ES cells (indicated with rectangles). (E) Boxplots showing log2 fold change (log2FC) of CDK8 ChIP-seq signal (RPKM) at all CDK8 peaks (n = 24273), and those with reduced CDK8 (n = 783, ↓) and increased CDK8 (n = 379, ↑). p-values were calculated using a Wilcoxon rank sum test. (F) Boxplots showing the size of CDK8 peaks as in (E). p-values were calculated using a Wilcoxon rank sum test. (G) Venn diagrams representing the overlap between CDK8 peaks and super enhancers (SE) at CDK8 peaks as in (E). Percent overlap of all SEs is indicated. (H) A metaplot (left) showing BioCAP enrichment at CDK8 peaks as in (E) and boxplot quantification (right) of BioCAP RPKM levels. p-Values calculated using Wilcoxon rank sum test are indicated. (I) A metalplot (left) showing FS2-FBXL19 enrichment at CDK8 peaks as in (E) and boxplot quantification (right) of FS2-FBXL19 RPKM levels. p-values calculated using Wilcoxon rank sum test are indicated.

Figure 4—figure supplement 1. FBXL19 is required for appropriate CDK8 occupancy at a subset of CpG island promoters.

(A) A Venn diagram showing the overlap between CDK8 peaks (n = 24273) and NMIs (n = 27698). (B) A scatter plot showing Spearman correlation between CDK8 signal and BioCAP at NMIs. (C) Heatmaps showing enrichment of CDK8, FS2-FBXL19 and BioCAP at FBXL19 peaks (n = 11322) sorted by decreasing FBXL19 signal. (D) A Venn diagram showing the overlap between FBXL19 peaks (n = 11322), CDK8 peaks (n = 24273), and NMIs (n = 27698). Percent overlap of all FBXL19 peaks is shown. (E) A schematic of the Fbxl19^fl/fl allele showing the location of the loxP sites. Treatment with tamoxifen (OHT) results in Cre-mediated deletion of the second exon, encoding for the ZF-CxxC domain but leaving the rest of the gene and protein intact. (F) RT-qPCR of FBXL19 expression in Fbxl19^fl/fl ES cells before (WT) and following tamoxifen treatment (OHT). Primers specific for the ZF-CxxC and LRR domains were used. Expression is relative to expression in WT ES cells. Error bars show SEM of three biological experiments. (G) A Western blot analysis of the expression of Mediator subunits in Fbxl19^fl/fl (WT) and Fbxl19^ΔCXXC (OHT) ES cells. HDAC1 and TBP were probed as loading controls. (H) Heatmaps showing enrichment of CDK8 in Fbxl19^fl/fl (WT) and Fbxl19^ΔCXXC (OHT), FS2-FBXL19 and BioCAP signal at ATAC peaks divided based on change in CDK8 binding in Fbxl19^ΔCXXC ES cells as in Figure 4E. The mean of the enrichment for each group is shown above the heatmaps. Left: zoomed heatmaps for ATAC peaks associated with a decrease (↓) or an increase (↑) in CDK8 binding and a differential heatmap representing log2 fold change of CDK8 signal between OHT and WT samples at the differential peaks. (I) Percent overlap between CDK8 peaks (as in Figure 4E) and transcription start sites of genes (TSS, left) or FBXL19 peaks (right). (J) Metaplots showing enrichment of FS2-FBXL19, BioCAP and CDK8 ChIPseq signal at all CDK8 peaks (left), FBXL19-bound CDK8 peaks (FBXL19+, middle) and nonFBXL19 CDK8 peaks (FBXL19-, right). Peaks were divided based on change in CDK8 binding in Fbxl19^ΔCXXC ES cells as in Figure 4E.