Fig 2: LEC, but not SEC, is required for efficient 3'-end processing of RDH genes.a Knockdown of MED26 and ICE1, but not AFF4, increased the ratio of unprocessed transcripts (UTs) to coding transcripts (CTs) of RDH genes, but not c-myc or GAPDH genes. Total RNAs were extracted from cells treated with control, MED26, ICE1, or AFF4 siRNAs, and unproccessed and coding regions were transfected and the ratio of UTs to CTs was measured by real-time qPCR. Data points are the mean of three independent experiments and error bars show the standard deviation. The P values for the indicated comparisons were determined by Student’s t test (*P < 0.05; **P < 0.01). n = 3 biologically independent samples. Source data are provided as a Source Data file. b Heat map showing differences in transcript abundance in HEK293T cells transfected with either control or ZC3H8 siRNAs (log2 siRNA ZC3H8/control), determined by RNA-seq of polyA-selected libraries. c ZC3H8 knockdown increased the ratio of UTs and CTs of RDH genes. Data points are the average of three independent experiments and error bars show standard deviation. The P values for the indicated comparisons were determined by Student’s t test (*P < 0.05; **P < 0.01). n = 3 biologically independent samples. Source data are provided as a Source Data file. d Venn diagram showing overlap of genes upregulated following knockdown of MED26, ICE1, and AFF4. The red box highlights a group of 124 genes increased by MED26 and ICE1 knockdown, but not by knockdown of AFF4. e The top 20 genes in groups affected by MED26 and ICE1 knockdown, but not by knockdown of AFF4, include many RDH genes. n.s., Not significant.

Fig 3: Super-resolution imaging of Mediator and LEC in Cajal bodies.a–d HeLa cells stained with anti-ICE1 (magenta), anti-MED26 (magenta), anti-MED6 (magenta), and anti-coilin (green) antibodies were observed by stimulated emission depletion (STED) microscopy. Scale bars, 0.5 µm. X axis, Y axis, and Z axis are represented by black triangle, white triangle, and red triangle, respectively. e–h Line intensity plots across Cajal bodies (shown in a–d) of HeLa cells stained with anti-ICE1 (magenta), anti-MED26 (magenta), anti-MED6 (magenta), and anti-coilin (green) antibodies.

Fig 4: MED26 depletion leads to aberrant 3'-end processing of RDH transcripts.a Heat map showing differences in transcript abundance in HEK293T cells transfected with either control or MED26 siRNAs (log 2 siRNA MED26/control), determined by RNA-seq of polyA-selected or ribo-depleted libraries. b Organization of RDH genes and amplified regions of coding transcript and unprocessed transcript in quantitative real-time PCR analysis. 3'-End processing sites of RDH transcripts are localized between their stem-loop regions and downstream polyadenylation signals (PASs). c siRNA-mediated MED26 depletion increases the levels of unprocessed RDH transcripts. HEK293T cells were transfected with non-targeting siRNA as a control and each of three different siRNAs (#1, #2, and #3) targeting MED26. Total RNAs were extracted from cells and the ratio of unprocessed transcripts (UTs) to total transcripts (coding transcripts, CTs) from RDH genes was measured by real-time qPCR. Data points are the mean of three independent experiments, and error bars show standard deviation. The P values for the indicated comparisons were determined by Student’s t test (*P < 0.05; **P < 0.01). n = 3 biologically independent samples. Source data are provided as a Source Data file. d Heat map showing the differences in transcript abundance in parental HCT116 cells and HCT116 cells stably expressing MED26-AID in which the AID tag is fused to the C-terminus of MED26 and MED26 is acutely degraded in the presence of auxin. Transcripts derived from the cells were determined by RNA-seq of polyA-selected or ribo-depleted libraries (log2 auxin/non-treat). e Auxin-mediated MED26 degradation increased the levels of unprocessed RDH transcripts. Parental HCT116 cells or HCT116 cells stably expressing MED26-AID were cultured in the presence or absence of auxin. Total RNA was extracted from the cells and the ratio of UTs to total transcripts (CTs) from RDH genes was measured by real-time qPCR. Data points are the mean of three independent experiments and error bars show the standard deviation. The P values for the indicated comparisons were determined by Student’s t test (*P < 0.05; **P < 0.01). n = 3 biologically independent samples. n.s., Not significant. Source data are provided as a Source Data file.

Fig 5: MED26 and LEC are present at RDH and snRNA genes.a, b ChIP-sequence analysis was performed using duplicate ChIP samples. Examples of ChIP-seq genome browser tracks showing the presence of MED26 at RDH genes (a), and snRNA genes (b), are depicted. c Venn diagram showing the overlap of MED26-, ELL-, and ZC3H8-occupied genes. ELL and ZC3H8 ChIP-seq datasets are from are from GEO GSE4793831. d ELL and Pol II occupancy (ChIP/Input) at RDH genes is decreased by MED26 knockdown. Ct values of each ChIP are normalized to that of the input. Each value is the mean of three independent experiments and error bars show the standard deviation. The P values for the indicated comparisons were determined by Student’s t test (*P < 0.05; **P < 0.01). n = 3 biologically independent samples. n.s., Not significant. Source data are provided as a Source Data file.