Article Figures & Data
Figures
- Figure S1. Generation of Ctr9-GFP knock-in mES cells.
(A) Schematic presentation of the targeting strategy. GFP knock-in to the C terminus of Ctr9 by CRISPR/Cas9–mediated homologous recombination. Cas9 nuclease and an short guide RNA targeted to the C-terminus of Ctr9 were co-transfected to make a double-stranded break in the mouse ESCs genome. The double-strand break is repaired by homologous recombination with a donor template containing GFP flanked by homologous arms for Ctr9, where GFP is designed to be integrated. (B) Cellular localization of the Ctr9-GFP fusion protein. The Ctr9-GFP fusion protein was stained in green by using an anti-GFP antibody. Tubulin was stained in red by an anti-tubulin antibody. The cell nucleus was stained in blue by DAPI. Note the nuclear localization of the Ctr9-GFP fusion protein. (C) esiRNA knockdown of Ctr9. Mouse embryonic stem cells expressing the Ctr9-GFP fusion protein were transfected with esiRNAs targeting Ctr9. The non-targeting esiRNA (Luc esiRNA) served as control. Anti-GFP antibody was used to visualize the fusion protein at around 160 kD (green channel), and anti-tubulin antibody was used to detect tubulin (50 kD, red channel) as the loading control. Specific depletion of Ctr9-GFP fusion protein is seen in the Ctr9 knockdown sample.
- Figure S2. ChIP-seq analysis of Ctr9 binding in mouse embryonic stem cells (mESCs) and NIH3T3 cells.
(A, B) Pie representation of Ctr9 peaks detected at protein-coding genes and intergenic regions in mESCs (A), and NIH3T3 cells (B). (C, D) Representative genome browser tracks of highly expressed genes without Ctr9 binding in mESCs. The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per 1 million (rpm). Each ChIP-Seq experiment was performed with a single sample. (E, F, G, H) Representative genome browser tracks of Ctr9 ChIP-seq in mESCs and NIH3T3 cells. The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Nanog (E) and Oct4 (F) are shown as examples for Ctr9 binding specifically in mESCs. Col1a2 (G) and Col5a1 (H) are shown as examples for Ctr9 binding specifically in NIH3T3 cells.
- Figure 1. NELF, RNAPII Ser5p, and Paf1C occupancy at protein-coding genes.
(A, B) Metagene analyses showing positive correlation between Ctr9 occupancy and gene expression levels in mouse embryonic stem cells (A) and NIH3T3 (B). Based on expression data (RNA-seq), genes were divided into high (top 25%), medium (25–75%), low (bottom 25%), and silent gene expression categories. Ctr9 binding intensities to the upstream, gene body, and downstream part of the gene of each category were calibrated. Each ChIP-Seq experiment was performed with a single sample. (C) Representative genome browser tracks of Ctr9 ChIP-seq in mouse embryonic stem cells (upper) and NIH3T3 cells (lower). The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Esrrb (Top panel) and Col1a1 (middle) are shown for ESC and NIH3T3 cell specific binding, respectively, whereas Ppia (bottom) is shown as example for Ctr9 binding to both cell types. (D) NELFA (blue), RNAPII Ser5p (green), and Ctr9 (red) occupancy at the Actb gene. The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Note the shift of the Ctr9 peak with respect to the NELFA and RNAPII Ser5p peaks. The dotted black line marks the transcription start site (TSS) of the Actb gene. (E) Metagene profiles of ChIP-seq read coverages across 4-kb windows centered around the TSSs of all genes bound by Ctr9. The y-axis shows an average normalized read count scaled to 10 million reads. (F) Box plot of the binding positions of NELFA, RNAPII Ser5p and Ctr9 around the TSS region with peaks detected in ChIP-seq experiments. The y-axis shows distances in base pairs of peaks to the annotated TSS (calculated with the software Homer) (53).
- Figure S3. GFP tagging of NELFA in mES cells.
(A) Cellular localization of LAP (Localization and Affinity Purification) tagged NELFA protein in mouse embryonic stem cells (Poser et al, 2008). NELFA-GFP fusion protein was stained in green by using an anti-GFP antibody. The cell nucleus was stained by DAPI in blue. Note the nuclear localization of the NELFA-GFP fusion protein. (B) esiRNA knockdown of NELFA. Mouse embryonic stem cells expressing NELFA-GFP were transfected with esiRNAs targeting NELFA. Non-targeting esiRNA (Luc esiRNA) served as control. Anti-GFP antibody was used to visualize the fusion protein at around 85 kD (green channel), and anti-tubulin antibody was used to detect tubulin (50 kD, red channel) as the loading control. Specific depletion of NELFA-GFP fusion protein is seen in the NELFA knockdown sample.
Source data are available for this figure.
Source Data for Figure S3[LSA-2020-00792_SdataFS3.tif]
- Figure 2. Paf1C is enriched at enhancers, and correlates with enhancer activities.
(A) Genome browser tracks of ChIP-seq results obtained for Ctr9, H3K27ac, and H3K4me1, and RNA-seq results in the vicinity of the Pou5f1 (Oct4) gene in mES cells. The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Black boxes indicate annotated Pou5f1 enhancers. Each ChIP-Seq experiment was performed with a single sample. (B) Box plots of Ctr9 binding densities on typical enhancers (TEs) and super enhancers (SEs). Significantly higher levels of Ctr9 binding was measured comparing SEs with TEs. P-value < 2.2 × 10−16 according to Wilcoxon rank sum test (SEs versus TEs). (C) Venn diagram analysis of H3K27ac, H3K4me1, Ctr9 occupancy on SEs. The numbers represent the percentages of SEs with corresponding histone modifications, and/or Ctr9 binding. (D) Venn diagram analysis of H3K27ac, H3K4me1, Ctr9 occupancy on TEs. The numbers represent the percentages of TEs with corresponding histone modifications, and/or Ctr9 binding. (E, F) Representative genome browser tracks for Ctr9, H3K27ac, and H3K4me1 at SEs without Ctr9 binding (E), or with Ctr9 binding (F). The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Black boxes indicate the annotated SEs. (G) Experimental evaluation of SE activities. Indicated DNA elements were tested to drive expression of the firefly luciferase gene. The y-axis shows the luciferase measurements in A.U. The values are normalized to samples transfected with the empty reporter vector. The pRL-SV40 plasmid was used as transfection efficiency control. Data are presented as the mean ± SD from three independent experiments. Error bars depict SD.
- Figure 3. Occupancy of NELFA and Ctr9 at enhancers.
(A) Heat maps showing ChIP-Seq occupancy patterns of NELFA, Ctr9, RNAPII Ser2P, RNAPII Ser5P, H3K27ac, and H3K4me1 at enhancers that are bound by NELFA. Color-coding is based on quantile-normalized read coverage signals, with yellow indicating stronger binding. Black dots covering NELFA and Ctr9 heat maps are placed at summits of predicted ChIP-Seq peaks. All heat maps are showing 2-kb windows centered around NELFA peak summits, oriented towards the nearest Ctr9 peak summit. Each ChIP-Seq experiment was performed with a single sample. (B) Box plots showing distance in base pairs between NELFA and Ctr9 binding sites on typical enhancers and super enhancers (SEs). The average numbers in base pairs for the two classes are presented within the plots. Distances of peaks were measured between predicted peak summits, as reported by the peak calling software Homer (53). (C, D) Representative genome browser tracks for Ctr9 and NELFA ChIP-seq at SEs showing unilateral (C) or bilateral (D) Ctr9 binding on SEs in relation to NELFA. The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Black boxes indicate annotated SEs.
- Figure S4. Ctr9, RNAPII Ser2p, and Ser5p occupancies at protein-coding genes.
Representative genome browser tracks of ChIP-seq for Ctr9 (red), RNAPII Ser2p (brown), and RNAPII Ser5p (green) for a protein-coding gene in mES cells. Each ChIP-Seq experiment was performed with a single sample. The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. The Pabpc1 gene is shown as an example.
- Figure 4. Paf1C and RNAPII Ser2p correlate with enhancer RNA (eRNA) transcripts.
(A, B) Representative genome browser tracks showing RNAPII Ser2p, RNAPII Ser5p, H3K27ac and eRNA on super enhancers (SEs) with Ctr9 binding (A) or without Ctr9 binding (B). (A, B) RNAPII Ser5p and H3K27ac are detectable in both (A) and (B), whereas RNAPII Ser2p and eRNA transcripts are detected only on SE with Ctr9 binding (A). The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Black boxes indicate annotated SEs. Each ChIP-Seq experiment was performed with a single sample. (C) Venn diagram analysis of RNAPII Ser2p, RNAPII Ser5p, and Ctr9 occupancy on SEs. The numbers indicate SEs carrying either histone modifications or Ctr9 binding. (D) Analysis of Ctr9 occupancy and eRNA transcripts on SEs. The numbers indicate SEs with Ctr9 binding and eRNA transcripts. The plot shows that all 126 eRNAs were transcribed from SEs with Ctr9 binding.
- Figure S5. Zoom out view occupancy of NELFA and Ctr9 at enhancers.
(A) Heat maps showing ChIP-Seq occupancy patterns of NELFA, Ctr9, RNAPII Ser2P, RNAPII Ser5P, H3K27ac, and H3K4me1 at enhancers that are bound by NELFA. Color-coding is based on quantile-normalized read coverage signals, with yellow indicating stronger binding. Black dots covering NELFA and Ctr9 heat maps are placed at summits of predicted ChIP-Seq peaks. (B) Heat maps showing ChIP-Seq occupancy patterns of NELFA, Ctr9, RNAPII Ser2P, RNAPII Ser5P, H3K27ac, and H3K4me1 at super enhancers that are not bound by Ctr9. Each ChIP-Seq experiment was performed with a single sample. Color-coding is based on quantile-normalized read coverage signals, with yellow indicating stronger binding. All heat maps are showing 50-kb windows centered around NELFA peak summits, oriented towards the nearest Ctr9 peak summit.
- Figure 5. Paf1C regulates gene expression by modulating enhancer activity.
(A) Enhancer activity of indicated super enhancer (SE) sequences not bound by Ctr9 after transfection with a non-targeting control silencing trigger (black) or Ctr9 knockdown (grey) are shown. (B) Enhancer activity of indicated SE sequences that are bound by Ctr9 after transfection with a non-targeting control silencing trigger (black) and Ctr9 knockdown (grey) are shown. The y-axis shows the luciferase measurements in A.U. The values are normalized to samples transfected with the empty reporter vector control. The pRL-SV40 plasmid was used as transfection efficiency control. Data are presented as the mean ± SD from three independent experiments. Error bars depict SD. (A, B) Note that depletion of Ctr9 strongly reduced enhancer activity of SEs with Ctr9 binding (panel B), whereas the same treatment had no significant effects on SEs without Ctr9 binding (panel A). (C) MA plot showing global gene expression changes upon Ctr9 knockdown, with each point representing one transcript. White-filled and black-filled circles indicate transcripts associated with typical enhancers (TEs) or SEs, respectively. Grey dots indicate all other transcripts that are not associated with SEs or TEs. A dashed red line shows a boundary at which gene expression is not altered. The y-axis shows a log2 fold change of transcript per million (tpm) values measured in cells treated by Ctr9 RNAi versus control. Only genes considered as expressed (tpm ≥ 1) are shown. (D) Box plots showing distributions of gene expression changes upon RNAi-based Ctr9 knockdown. Genes associated with different types of enhancers are split into separate groups (labelled as TE and SE) and the third group, labelled “Other genes,” contains genes not associated with any annotated enhancer. The y-axis shows a log2 fold change of tpm values measured in cells treated by Ctr9 RNAi versus control. P-value < 2.58 × 10−05 (SEs versus TEs) and P-value < 1.7 × 10−09 (SEs versus “Other genes”) were calculated according to Mann–Whitney U test. (E) Box plots showing general reductions of Pol II Ser2p occupancy at enhancers upon Ctr9 knockdown. The first two panels show total read counts across 8,563 typical enhancers (TEs) and 231 super enhancers (SEs). The third panel, labelled “Shuffled,” serves as a control showing data from 8,794 randomized genomic intervals of the same length as in both sets of enhancers. Numbers above horizontal bars are sample medians. (F) Enhancer activity of Oct4 SE after transfection with a non-targeting silencing trigger control (black) or Ctr9 knockdown (grey) is shown. The y-axis shows the luciferase measurements in A.U. (G) RNAPII Ser2p occupancy on Oct4 SEs after transfection with a non-targeting silencing trigger control (blue) or Ctr9 knockdown (red) is shown. The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Black boxes indicate the annotated SEs. Each ChIP-Seq experiment was performed with a single sample. (H) qRT-PCR quantification of Oct4 expression after transfection with a non-targeting silencing trigger control (black) or Ctr9 knockdown (grey). Oct4 expression was normalized to the expression of the housekeeping gene GAPDH. Values and shown as fold changes to the sample transfected with non-targeting silencing trigger control. Data are presented as the mean ± SD from three independent experiments. Error bars depict SD. (I) Oct4 expression determined by RNA-seq after transfection with a non-targeting silencing trigger control (blue) or Ctr9 knockdown (red). The y-axis represents normalized read density in reads per 10 million. Statistically significant differences were determined by a two-tailed t test (** indicates P < 0.01 and *** indicates P < 0.001).
- Figure S6. Paf1C regulates gene expression by modulating enhancer activity.
(A) Enhancer activity of Tbx3 super enhancer (SE) after transfection with a non-targeting control silencing trigger (black) and Ctr9 knockdown (grey). The y-axis shows the luciferase measurements in A.U. The values are normalized to sample transfected with the empty reporter vector. The pRL-SV40 plasmid was used as transfection efficiency control. Data are presented as the mean ± SD from three independent experiments. Error bars depict SD. (B) RNAPII Ser2p occupancy on Tbx3 gene after transfection with a non-targeting control silencing trigger (blue) and Ctr9 knockdown (orange). The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Black boxes indicate the annotated Tbx3 SE. Each ChIP-Seq experiment was performed with a single sample. (C) qRT-PCR quantification of Tbx3 expression after transfection with a non-targeting silencing trigger control (black) and Ctr9 knockdown (grey). Tbx3 expression was normalized to the expression of the housekeeping gene GAPDH, and shown as fold changes to the sample transfected with the non-targeting silencing trigger. Data are presented as the mean ± SD from three independent experiments. Error bars depict SD. (D) Tbx3 expression determined by RNA-seq after transfection with a non-targeting silencing trigger (blue) and Ctr9 knockdown (orange). The y-axis represents normalized read density in reads per 10 million. (E) Enhancer activity of Zfp638 SE, which is not bound by Ctr9, after transfection with a non-targeting control silencing trigger (black) and Ctr9 knockdown (grey). The y-axis shows the luciferase measurements in A.U. The values are normalized to sample transfected with the empty reporter vector. The pRL-SV40 plasmid was used as transfection efficiency control. Data are presented as the mean ± SD from three independent experiments. Error bars depict SD. (F) RNAPII Ser2p occupancy on Zfp638 gene after transfection with a non-targeting control silencing trigger (blue) and Ctr9 knockdown (orange). The x-axis indicates the chromosome position, and the y-axis represents normalized read density in reads per million. Black boxes indicate the annotated Tbx3 SE. (G) Quantification of Zfp638 expression after transfection with a non-targeting silencing trigger control (black) and Ctr9 knockdown (grey). Zfp638 expression was determined by reads from RNAseqs and shown as fold changes to the sample transfected with the non-targeting silencing trigger. (H) Zfp638 expression determined by RNA-seq after transfection with a non-targeting silencing trigger (blue) and Ctr9 knockdown (orange). The y-axis represents normalized read density in reads per 10 million. Statistically significant differences were determined by a two-tailed t test (** indicates P < 0.01 and *** indicates P < 0.001).
- Figure 6. The combination of H3K27ac and Ctr9 DNA-occupancy improves the prediction of active enhancers.
(A) Enhancer activities of indicated regions with H3K27ac+/Ctr9+ or H3K27ac+/Ctr9− marks in NIH3T3 cells (grey) and mouse embryonic stem cells (black). The y-axis shows the luciferase measurements in A.U. The values are normalized to samples transfected with the empty reporter vector. The pRL-SV40 plasmid was used as transfection efficiency control. Data are presented as the mean ± SD from three independent experiments. Error bars depict SD. Chromosome positions of each indicated region are listed in Table S9. (A, B) Representative genome browser views of regions that were analyzed for enhancer activity in panel (A). Genome sites with H3K27ac+/Ctr9+ mark in NIH3T3 cells (upper panel), H3K27ac+/Ctr9− marks in NIH3T3 cells (middle panel), and H3K27ac+/Ctr9+ marks in both NIH3T3 and mES cells (lower panel) are shown. (A) The x-axis indicates the chromosome position of the corresponding fragments analyzed in panel (A). The y-axis represents normalized read density in reads per million. Each ChIP-Seq experiment was performed with a single sample. (C) Model of indicated states of enhancer activity. Proposed roles of indicated protein complexes and chromatin modifications are depicted. Polycomb repressive complex 2, PRC2.
Supplementary Materials
Table S8 Predicted enhancers in NiH3T3 cells.
