The chromatin Remodeler CHD8 is required for activation of progesterone receptor-dependent enhancers

Abstract

While the importance of gene enhancers in transcriptional regulation is well established, the mechanisms and the protein factors that determine enhancers activity have only recently begun to be unravelled. Recent studies have shown that progesterone receptor (PR) binds regions that display typical features of gene enhancers. Here, we show by ChIP-seq experiments that the chromatin remodeler CHD8 mostly binds promoters under proliferation conditions. However, upon progestin stimulation, CHD8 re-localizes to PR enhancers also enriched in p300 and H3K4me1. Consistently, CHD8 depletion severely impairs progestin-dependent gene regulation. CHD8 binding is PR-dependent but independent of the pioneering factor FOXA1. The SWI/SNF chromatin-remodelling complex is required for PR-dependent gene activation. Interestingly, we show that CHD8 interacts with the SWI/SNF complex and that depletion of BRG1 and BRM, the ATPases of SWI/SNF complex, impairs CHD8 recruitment. We also show that CHD8 is not required for H3K27 acetylation, but contributes to increase accessibility of the enhancer to DNaseI. Furthermore, CHD8 was required for RNAPII recruiting to the enhancers and for transcription of enhancer-derived RNAs (eRNAs). Taken together our data demonstrate that CHD8 is involved in late stages of PR enhancers activation.

Fig 1. Genome-wide analysis of CHD8 binding sites under normal proliferation conditions.

(A) Distribution of CHD8 peaks at low (P < 10^–10), middle (P < 10^–15), and high (P < 10^–20) confidence thresholds, in proliferating T47D-MTVL cells relative to known RefSeq genes. Promoters: ± 2 kb around transcription start site (TSS); Downstream extremities: ± 2 kb around transcription end site; Exons: exonic regions; Introns: intronic regions; Intergenic > 2 kb away from RefSeq TSS. (B) Meta-gene representation of CHD8 ChIP-seq signal at the low confidence threshold. Log2 normalized ratios versus IgG signal are represented. (C) Genome Browser view of IgG, CHD8, H3K4me3 and RNAPII occupancy in a region of chromosome 8. Numbers in the y-axis are reads per million mapped reads. (D) CHD8 occupancy around the centre of the CHD8 binding sites at TSS (red), introns (black) or intergenic regions (blue). (E) Overlapping between CHD8 identified peaks at the low confidence threshold (12655), RNAPII peaks (46586) and H3K4me3 peaks (25210).

Fig 2. Hormone-dependent CHD8 recruitment to PR binding sites.

(A) Overlapping between CHD8 identified peaks in proliferating cells (blue) or in cells stimulated with R5020 for 5 min (red) or 45 min (green). (B) Distribution of CHD8 peaks in cells stimulated with R5020 for 5 or 45 min. Categories as in Fig 1A. (C) Enrichment of CHD8 binding in response to R5020 or vehicle, upon treatment for 5 min (R5020 5 min, EtOH 5 min, red) or 45 min (R5020 45 min, EtOH 5 min, green) or in proliferating un-induced conditions (proliferation, blue), in four regions containing progesterone-responsive genes: HSD11B2, FKBP5, NFE2L3 and IL6ST. (D) Most significant de novo motif (P-value: 4.7x10^-75) identified using ChIPseeqerFIRE and MEME suite [72,73], in the CHD8-binding regions of T47D-MTVL cells stimulated with R5020 for 45 min. (E) Overlapping between progesterone-dependent CHD8 binding sites (green) and PRbs (red) [25] in T47D-MTVL cells stimulated with R5020. (F-J) CHD8 occupancy after 5 (red) or 45 (green) min of R5020 treatment, plotted as the average density of reads counted around the centre of all PRbs (F), around PRbs showing a high (G) or a low (H) nucleosome remodelling index (NRI), around p300 binding sites after R5020 (I) and around PRbs that show H3K4me1 enrichment (J). CHD8 occupancy is expressed as normalized tag density. PRbs, PR binding sites.

Fig 3. CHD8 is required for progesterone-dependent gene regulation.

(A) Flow diagram depicting the knockdown strategy for CHD8 and the hormone treatment in gene expression and ChIP studies. T47D-MTVL cells were transfected with control siRNA (siCt) or siRNA against CHD8 (siCHD8), subjected to serum deprivation during 48 h and then stimulated with 10 nM R5020 (R5020) or vehicle (EtOH) for 45 min or 6 h, depending on the experiment. (B) Western blot analysis of CHD8 expression upon transfection of T47D-MTVL cells with control siRNA (siCt) or siRNA against CHD8 (siCHD8). (C) Box-and-whisker plots of the change in gene expression of CHD8-dependent genes (see Materials and Methods) after 6 h of R5020 treatment. siCHD8: cells depleted of CHD8; siCt, control cells. (D) Overlapping between R5020-dependent CHD8-target genes (green) and genes that are differentially regulated in response to R5020 in CHD8-silenced cells with respect to control cells (see Materials and Methods) (purple, siCHD8-affected genes). Chip-seq CHD8 peaks were assigned to the closer gene. (E) Effect of CHD8 depletion in progestin-dependent expression of the following genes: HSD11B2, MMTV-Luc, DUSP1, FKBP5, NFE2L3 and IL6ST. Level of CHD8 expression was determined as control of silencing (upper panel). mRNA levels were determined by RT-qPCR after 45 min or 6 h of stimulation. Data are the mean of at least n = 6 qPCR reactions from three independent experiments. Error bars represent ± SD values. * p < 0.001; ** p < 0.0001 with respect to siCt, using Student’s t-test. (F) ChIP analysis of CHD8 occupancy at the MMTV regulatory region upon stimulation with R5020 during 5 or 45 min. Data are the mean of at least n = 6 qPCR reactions from three independent experiments.

Fig 4. PR is necessary for hormone-dependent CHD8 recruitment to PR enhancers.

(A) ChIP-seq binding profile for PR, P300 and CHD8 upon R5020 treatment at the indicated enhancer regions. Numbers in the y axis indicates number of reads, while numbers in the x axis indicate chromosomal position. (B) PR and CHD8 expression in T47-YV or T47D-MTVL was analyzed by Western blotting with anti-PR and anti-CHD8 antibodies. (C) ChIP analysis of PR and CHD8 occupancy at the indicated enhancers in T47-YV or T47D-MTVL cells stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min. (D) ChIP analysis of CHD8 and PR at the indicated enhancers in T47D-MTVL cells transfected with control siRNA (siCt) or siRNA against CHD8 (siCHD8), and then stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min. (C, D) Data are the mean of at least n = 6 qPCR reactions from three independent experiments. Error bars represent ± SD values. * p < 0.02; ** p < 0.01; *** p < 0.001 using Student’s t-test.

Fig 5. Depletion of FOXA1 stimulates PR and CHD8 recruitment to PRbs.

(A) Overlapping between progesterone-dependent CHD8 binding sites (green), progesterone-dependent PRbs (red) [25] and FOXA1 binding sites (blue) in T47D cells (ENCODE dataset GSM803409). (B) ChIP analysis of FOXA1 occupancy at the indicated enhancers in T47-YV or T47D-MTVL cells stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min. (C) Western blot analysis of FOXA1 expression upon transfection of T47D-MTVL cells with control siRNA (siCt) or siRNA against FOXA1 (siFOXA1). (D-E) ChIP analysis of CHD8 (D) and PR (E) at the indicated enhancer in T47D-MTVL cells transfected with control siRNA (siCt) or siRNA against FOXA1 (siFOXA1), and then stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min. (B, D-E) Data are the mean of at least n = 6 qPCR reactions from three independent experiments. Error bars represent ± SD values. * p < 0.01; ** p < 0.001; *** p < 0.0001 using Student’s t-test; n.s., not significant.

Fig 6. SWI/SNF complexes interact with CHD8 and are involved in CHD8 recruitment.

(A) SWI/SNF subunits co-immunoprecipitate with CHD8. Extract from T47D-MTVL cells were subjected to immunoprecipitation using anti-CHD8 antibody. Precipitated proteins were then revealed by western blotting using the indicated antibodies against BAF or PBAF subunits. (B) ChIP analysis of BAF155 at the indicated regions in T47D-MTVL cells stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min. (C) ChIP analysis of CHD8 at the indicated regions in T47D-MTVL cells transfected with control siRNA (siCt) or siRNA against BRM and BRG1 (siBRM+siBRG1), and then stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min. (B, C) Data are the mean of at least n = 6 qPCR reactions from three independent experiments. Error bars represent ± SD values. * p < 0.05; ** p < 0.01; *** p < 0.001 using Student’s t-test.

Fig 7. CHD8 involvement in modification of enhancer chromatin.

(A) CHD8 is not required for H3K27 acetylation. ChIP analysis of H3K27Ac enrichment at the indicated enhancers in T47D-MTVL cells transfected with control siRNA (siCt) or siRNA against CHD8 (siCHD8), and then stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min. Data are the mean of at least n = 6 qPCR reactions from three independent experiments. Error bars represent ± SD values. * p < 0.01; ** p < 0.001; *** p < 0.0001 using Student’s t-test. (B) CHD8 contributes to open chromatin at PR enhancers. DNase I sensitivity at the indicated regions in T47D-MTVL cells transfected with control siRNA or siRNA against CHD8, and then stimulated with R5020 or vehicle for 45 minutes. Error bars represent ± SD (n = 3).

Fig 8. CHD8 is required for RNAPII recruitment and eRNA synthesis at PR enhancers.

(A) Time course analysis of PR, CHD8 and RNAPII recruitment to FKBP5e, NFE2L3e and IL6STe enhancers by ChIP. T47D-MTVL cells were stimulated with R5020 (R5020) or vehicle (EtOH) for the indicated times and then processed for ChIP by using antibodies against PR, CHD8 and RNAPII. (B) ChIP analysis of RNAPII enrichment at the indicated enhancers in T47D-MTVL cells transfected with control siRNA (siCt) or siRNA against CHD8 (siCHD8), and then stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min. (C) Expression of enhancer RNAs (eRNA) from the indicated enhancers in T47D-MTVL cells transfected with control siRNA (siCt) or siRNA against CHD8 (siCHD8), and then stimulated with R5020 (R5020) or vehicle (EtOH) for 45 min or 6 h. (A-C) Data are expressed as fold induction relative to the level in ethanol treated cells. Data are the mean of at least n = 6 qPCR reactions from three independent experiments. Error bars represent ± SD values. * p < 0.01; ** p < 0.001; *** p < 0.0001 using Student’s t-test.