Cohesin and polycomb proteins functionally interact to control transcription at silenced and active genes

Abstract

Cohesin is crucial for proper chromosome segregation but also regulates gene transcription and organism development by poorly understood mechanisms. Using genome-wide assays in Drosophila developing wings and cultured cells, we find that cohesin functionally interacts with Polycomb group (PcG) silencing proteins at both silenced and active genes. Cohesin unexpectedly facilitates binding of Polycomb Repressive Complex 1 (PRC1) to many active genes, but their binding is mutually antagonistic at silenced genes. PRC1 depletion decreases phosphorylated RNA polymerase II and mRNA at many active genes but increases them at silenced genes. Depletion of cohesin reduces long-range interactions between Polycomb Response Elements in the invected-engrailed gene complex where it represses transcription. These studies reveal a previously unrecognized role for PRC1 in facilitating productive gene transcription and provide new insights into how cohesin and PRC1 control development.

Figure 1. PRC1 binds both silenced genes and active cohesin-binding genes in wing imaginal discs.

(A) ChIP-chip tracks for Rad21 (cohesin), Polycomb (PRC1; Pc-RJ), and H3K27me3 (PRC2) of the invected-engrailed gene complex in anterior (A, black) and posterior (P, red) wing discs. Bars underneath ChIP tracks indicate significance at p≤10⁻³. A wing disc expressing red fluorescent protein (RFP) under control of en2.4-GAL4 is shown to the right. (B) RNA polymerase II (Pol II, black, 8WG16 antibody), Rad21 and Nipped-B (blue), PRC1 subunits (Psc, Pc, Ph) and PRC2-generated H3K27me3 (green) ChIP at constitutively active genes in whole wing discs. A higher resolution view of the dm/myc gene is in Figure S4. (C) Rad21, PRC1 (Ph, Psc, Pc-VP, Pc-RJ) and H3K27me3 ChIP at invected-engrailed using whole wing discs. (D) Venn diagram of the overlap of Rad21 (cohesin), Polycomb (Pc, PRC1), and H3K27me3 (PRC2) binding to all annotated genes in posterior wing imaginal discs. Binding was called at threshold of p≤10⁻³. There are 4,830 genes that bind both Rad21 and Pc, 1,004 genes that bind Pc but not Rad21, 185 genes that bind Pc and H3K27me3, and 187 genes that bind Pc, H3K27me3 and Rad21. In the latter group, only a few such as invected and engrailed show the large overlapping domains of cohesin and H3K27me3 extending over several kilobases that we define as the cohesin-H3K27me3 restrained state . (E) Association of PRC1 subunits with active genes in whole wing discs. The mRNA levels for 13,132 genes measured by microarray are sorted from lowest to highest according the heat map. The aligned heat maps show the total ChIP signal (Figure S7, method 1) for the indicated proteins for each gene.

Figure 2. Cohesin and PRC1 affect each other's binding to active and PcG-silenced genes in cultured BG3 cells.

(A) Overlap of Rad21 (cohesin), Polycomb (Pc, PRC1) and H3K27me3 (PRC2) binding at all annotated genes. Binding was called at a statistical threshold of p≤10⁻². (B) The top graphs show the mRNA levels (log2 scale) for 13,132 genes measured by microarray in control BG3 cells. The aligned bar plots just below show the change in total Pc ChIP signal for each gene after Rad21 depletion and change in total Rad21 ChIP signal after Ph depletion (Figure S7, method 1). (C) Percent of cohesin-binding (red) and non-binding genes (blue) that show an increase (UP) or decrease (DOWN) in Pc ChIP signal after Rad21 depletion, and genes that show an increase or decrease in Rad21 after Ph depletion. The genes are subdivided into active genes (∼7,000), and the genes marked by H3K27me3 that bind Pc (∼600 genes because not all inactive genes are PcG-silenced). Cohesin binding was called at a statistical threshold of p≤10⁻³ and increases and decreases were called by differences in ChIP signal that were at least 2 standard deviations from the median genome-wide difference that extend for at least 105 bp (Figure S7, method 2).

Figure 3. Examples of effects of cohesin and PRC1 on each other's binding, and effects of PRC1 depletion on Pol II occupancy in BG3 cells.

ChIP-chip tracks of Rad21, Polycomb (Pc, PRC1), Rpb3 (total Pol II), or Ser2P Pol II (elongating Pol II), from control (mock-treated) BG3 cells and Rad21 or Polyhomeotic (Ph, PRC1) depleted cells are shown. Bars underneath the ChIP tracks indicate binding at the p≤10⁻³ threshold. The Δ tracks show the differences in the ChIP enrichment (MAT score) between the RNAi depleted and control cells. Bars underneath the Δ tracks indicate significant increases (+) or decreases (−) in ChIP enrichment after RNAi treatment, using method 2 in Figure S7. The myc (diminutive, dm) gene, is an example of an active, cohesin-PRC1 binding gene. Sex combs reduced (Scr) and Antennapedia (Antp) in the Antennapedia complex (ANT-C) are examples of PcG-silenced genes.

Figure 4. PRC1 affects Pol II occupancy and function at both silenced and active genes.

(A) The change in total Pol II ChIP signal (ΔRpb3, Figure S7 method 1) in the bodies of the 13,132 genes measured for mRNA levels (X axis) plotted versus the change in elongating phosphorylated Pol II (ΔSer2P Pol II) (Y axis) caused by Ph (PRC1) depletion. Each gene is color-coded by a heat map to show the total H3K27me3 level in control cells. Many silenced genes show increases in both total and Ser2P Pol II, while the majority of active genes show increased total Pol II and reduced Ser2P Pol II. (B) Same plot as in A, except the genes are color-coded to show the change in mRNA level. Many silenced genes increase in expression, consistent with an increase in Ser2P Pol II, and many active genes decrease in expression, consistent with reduced Ser2P Pol II. (C) Change in total Pol II plotted versus the change in mRNA level after Ph depletion. Many genes show modest to large increases in total Pol II, but can show either reduced or increased mRNA levels. (D) Change in Ser2P Pol II ChIP signal plotted versus change in mRNA level (log2) for 13,132 genes after Ph depletion. Changes in Ser2P Pol II correlate with changes in mRNA levels. (E) Box plots showing the Pause Index (total Pol II density in the 400 bp region surrounding the transcription start site divided by the total Pol II density in the gene body, Figure S7 method 1) for all annotated genes before and after Ph depletion. (F) Box plots showing the ratio of Ser2P Pol II to total Pol II (Rpb3) in the bodies of all annotated genes before and after Ph depletion.

Figure 5. Cohesin depletion reduces long-range interactions between PRE-containing regions in the invected-engrailed locus in BG3 cells.

Looping interactions were measured by chromosome conformation capture (3C). (A) The top is a diagram of the inv-en locus with ChIP-chip tracks for H3K27me3 and cohesin, which is aligned with the 3C panels below to show the anchor locations. Anchors a and d, shown in the top and bottom panels, are outside of the complex and show high local interactions but do not interact with points inside the complex. Anchor b (second panel) in the PRE-containing region upstream of inv, interacts with the regulatory region between inv and en, the en PRE/promoter region, and the end of the regulatory region near the 3′ end of the tou gene. Anchor c (third panel) near the en PRE, interacts with the PRE region upstream of inv, the regulatory region between inv and en, and with the end of the regulatory region near tou. In Rad21 depleted cells (red), mRNA levels increased 20 to 40-fold , consistent with several independent experiments , and the looping between the inv PRE and the en PRE and regulatory region between the genes is decreased (arrows). Each interaction was measured with at least two biological replicates and two RT-PCR reactions per replicate. Error bars are the standard error of all RT-PCR replicates. (B) 3C analysis in Sg4 cells (yellow lines), in which inv-en is PcG-silenced, compared to the looping in BG3 cells (black lines).

Figure 6. Three cohesin-PcG states and their interactions.

In the PcG-silenced state (left), PRC1 (green rectangle) and PRC2 (tan rectangle) are both present, and the target genes are marked by the H3K27me3 histone modification made by PRC2. There is little detectable cohesin outside of Polycomb Response Elements (PREs). Depletion of cohesin increases PRC1 binding, and vice versa, and PRC1 depletion usually increases total Pol II, phosphorylated Pol II, and mRNA. The rare cohesin-PcG restrained state (bottom center) in which cohesin, PRC1 and PRC2 are all present may be a transition state between the silenced state and the cohesin-PRC1 active state (right). In the restrained state, cohesin and PRC1 both repress transcription, but the genes are not fully silenced. At the restrained inv-en gene complex in BG3 cells, cohesin depletion reduces long-range interactions between two Polycomb Response Elements (PREs), suggesting that cohesin represses by facilitating PRE-PRE looping. The cohesin-PRC1 active state (right) has promoter-proximal paused RNA Pol II with a short nascent RNA (red), PRC2 and H3K27me3 are absent, and cohesin aids PRC1 binding. Cohesin is the multi-subunit ring (orange, blue, purple) around the DNA. DSIF (tan circle) and NELF (green circle) are pausing factors, which along with the Pol II tail are phosphorylated by P-TEFb to induce transition to elongation. PRC1 depletion increases total Pol II but decreases elongating phosphorylated Pol II (Ser2P Pol II), with a corresponding decrease in mRNA. We thus posit that PRC1 prevents paused Pol II from entering into elongation until it is fully phosphorylated by P-TEFb. We theorize that binding of PRC1 to active genes limits the amount of unbound PRC1 (top) available to bind to PcG-silenced genes. This explains the in vivo genetic antagonism between cohesin and PRC1 in gene silencing, and how cohesin depletion simultaneously decreases PRC1 levels at active genes and increases them at silenced genes.