PGE2 alters chromatin through H2A.Z-variant enhancer nucleosome modification to promote hematopoietic stem cell fate

Abstract

Prostaglandin E2 (PGE₂) and 16,16-dimethyl-PGE₂ (dmPGE₂) are important regulators of hematopoietic stem and progenitor cell (HSPC) fate and offer potential to enhance stem cell therapies [C. Cutler et al. Blood 122, 3074-3081(2013); W. Goessling et al. Cell Stem Cell 8, 445-458 (2011); W. Goessling et al. Cell 136, 1136-1147 (2009)]. Here, we report that PGE₂-induced changes in chromatin at enhancer regions through histone-variant H2A.Z permit acute inflammatory gene induction to promote HSPC fate. We found that dmPGE₂-inducible enhancers retain MNase-accessible, H2A.Z-variant nucleosomes permissive of CREB transcription factor (TF) binding. CREB binding to enhancer nucleosomes following dmPGE₂ stimulation is concomitant with deposition of histone acetyltransferases p300 and Tip60 on chromatin. Subsequent H2A.Z acetylation improves chromatin accessibility at stimuli-responsive enhancers. Our findings support a model where histone-variant nucleosomes retained within inducible enhancers facilitate TF binding. Histone-variant acetylation by TF-associated nucleosome remodelers creates the accessible nucleosome landscape required for immediate enhancer activation and gene induction. Our work provides a mechanism through which inflammatory mediators, such as dmPGE₂, lead to acute transcriptional changes and modify HSPC behavior to improve stem cell transplantation.

Keywords: chromatin; hematopoietic stem cell; prostaglandin.

Fig. 1.

Phospho-CREB regulates dmPGE₂-induced gene expression changes through binding at distal regulatory elements. (A) Schematic representation of the experimental approach used in this study. Cells are stimulated for 2 h with dmPGE₂ or vehicle control (DMSO) after which transcriptome and epigenome profiling was performed. (B) Venn diagram showing the number of genes up-regulated in green (535) and down-regulated in red (152) in CD34⁺ HSPCs after 2 h of dmPGE₂ stimulation in comparison to control-treated cells, as determined by RNA-Seq analysis. DEG criteria: FPKM ≥1 after treatment; fold change ≥1.5 or ≤0.67 (n = 3 biologically independent experiments). (C) Examples of genes identified as differentially expressed by RNA-Seq (n = 3 biologically independent experiments; mean values ± SEM). (D) Number of genes containing at least one pCREB peak in the proximity after dmPGE₂ stimulation. pCREB peaks assigned to a gene when located within a window from −5 kb upstream of the (TSS) to +5 kb downstream of the TTS were considered (n = 2 biologically independent ChIP-Seq experiments). (E) Correlation between pCREB binding and gene expression in response to dmPGE₂. pCREB density was calculated by dividing the total number of pCREB peaks associated to each gene category (up−, down−, and nonregulated genes) by the total amount of base pairs that this category occupies in the genome. pCREB peaks were assigned to a gene when located from +5 kb upstream of the TSS to +5 kb downstream of the TTS. Peak density in the genome was calculated by considering random distribution of pCREB sites in the whole genome. (F) Genomic distribution of unique pCREB peaks (inducible; present only after dmPGE₂ stimulation) versus ubiquitous pCREB peaks. (G) Enrichment of pCREB binding at four representative dmPGE₂ response genes: CXCL2 (promoter, intergenic), CXCL8 (promoter, intergenic), PDE4B (intronic), and FOSL2 (promoter, intronic). Gray bars indicate intronic and intergenic pCREB peaks. Genomic location of presented window is indicated at the bottom of the panels.

Fig. 2.

Stimuli-responsive enhancers gain chromatin accessibility and TF binding after dmPGE₂ stimulation. (A and B) Heat maps (A) and average enrichment profiles (B) of histone marks, ATAC accessibility, and TFs at enhancers before and after dmPGE₂ treatment. H3K27ac-enriched regions identified by ChIP-Seq are classified as de novo, enhanced, or background enhancers according to the change in H3K27ac levels observed following dmPGE₂ stimulation (n = 2 biologically independent experiments). A randomly sampled, comparable number of background enhancers (486) is shown. (C) Enrichment of histone mark, ATAC accessibility, and TF binding in response to dmPGE₂ at four representative stimuli-response enhancers. Genomic location of presented window and nearest gene is indicated at the bottom of the panel.

Fig. 3.

Stimuli-responsive enhancers mediate dmPGE₂-induced gene expression changes. (A) Gene expression changes of genes associated with stimuli-responsive and background enhancers. Enhancers were assigned to an individual nearest gene. Only genes with a mapped TSS within 15 kb of an enhancer were considered. Box plots shows median, 25th and 75th percentiles, whiskers are from 5th and 95th percentiles. Dots indicate outliers. (B) Percentages of enhancer nearest genes with fold changes in expression ≥1.5-fold or ≤0.67-fold for each enhancer category. (C) Up-regulated genes with a fold change in expression ≥1.5 (535) and their associated enhancers. (D) Heat maps of H3K27ac, ATAC accessibility, and TFs around enhancers before and after dmPGE₂ treatment. De novo, enhanced, or background enhancers were subset based on the presence or absence of pCREB after dmPGE₂. A randomly sampled, comparable number of background enhancers (486) is shown. Numbers of enhancers within each subset is indicated on the right of the heatmap. (E) Gene expression changes of genes associated with pCREB-positive and pCREB-negative enhancers. Enhancers were assigned to an individual nearest gene. Only genes with a mapped TSS within 15 kb of an enhancer were considered. Box plots shows median, 25th and 75th percentiles, whiskers are from 10th and 90th percentiles. Dots indicate outliers. For all analyses presented here, a randomly sampled set of background enhancers (486) was used.

Fig. 4.

DmPGE₂-responsive enhancers retain accessible nucleosomes after stimulation. (A) Average nucleosome occupancy profiles at stimuli-responsive and background enhancers from 4 MNase titration points (n = 3 biologically independent MNase-Seq experiments). (B) Nucleosome profiles of low and high MNase-Seq at stimuli-responsive and background enhancers (n = 3 biologically independent experiments). P-values by Wilcoxon rank-sum test. (C) Nucleosome fragment frequency and enrichment of H3K27ac, H2B, and pCREB at three representative stimuli-responsive and background enhancers. Genomic location of presented window and nearest gene is indicated at the bottom of the panel. Genomic location of presented window and enhancer nearest gene is indicated at the bottom of the panel. For all analyses presented here, a randomly sampled set of background enhancers (486) was used.

Fig. 5.

Modification of H2A.Z-variant-accessible nucleosomes at stimuli-responsive enhancers by HATs p300 and Tip60. (A) Heat maps of H3.3, H2A.Z, H2A.Zac, p300, and Tip60 binding at enhancers before and after dmPGE₂ treatment. H3K27ac-enriched regions identified by ChIP-Seq are classified as de novo, enhanced, or background enhancers according to the change in H3K27ac levels observed following dmPGE₂ stimulation (n = 2 biologically independent experiments). (B) Average enrichment profiles of histone variants and HATs before and after dmPGE₂ treatment in de novo, enhanced, or background enhancers. (C) Enrichment of chromatin binding factors and TFs in response to dmPGE₂ at three representative stimuli-response enhancers. Genomic location of presented window and enhancer nearest gene is indicated at the Bottom of the panel. For all analyses presented here, a randomly sampled set of background enhancers (486) was used.

Fig. 6.

H2A.Z-variant nucleosomes cooccupy pCREB binding sites at stimuli-responsive enhancers. (A) Enrichment of histone variants at nucleosome positions surrounding pCREB peaks within enhancers before and after dmPGE₂ stimulation. Position 0 indicates the nucleosome overlapping with pCREB peak centers. (B) Co-IP showing that pCREB associated with H2A.Z in U937 myeloid leukemia cells (n = 3 biologically independent experiments).