Histone deacetylases maintain expression of the pluripotent gene network via recruitment of RNA polymerase II to coding and noncoding loci

Abstract

Histone acetylation is a dynamic modification regulated by the opposing actions of histone acetyltransferases (HATs) and histone deacetylases (HDACs). Deacetylation of histone tails results in chromatin tightening, and therefore, HDACs are generally regarded as transcriptional repressors. Counterintuitively, simultaneous deletion of Hdac1 and Hdac2 in embryonic stem cells (ESCs) reduces expression of the pluripotency-associated transcription factors Pou5f1, Sox2, and Nanog (PSN). By shaping global histone acetylation patterns, HDACs indirectly regulate the activity of acetyl-lysine readers, such as the transcriptional activator BRD4. Here, we use inhibitors of HDACs and BRD4 (LBH589 and JQ1, respectively) in combination with precision nuclear run-on and sequencing (PRO-seq) to examine their roles in defining the ESC transcriptome. Both LBH589 and JQ1 cause a marked reduction in the pluripotent gene network. However, although JQ1 treatment induces widespread transcriptional pausing, HDAC inhibition causes a reduction in both paused and elongating polymerase, suggesting an overall reduction in polymerase recruitment. Using enhancer RNA (eRNA) expression to measure enhancer activity, we find that LBH589-sensitive eRNAs are preferentially associated with superenhancers and PSN binding sites. These findings suggest that HDAC activity is required to maintain pluripotency by regulating the PSN enhancer network via the recruitment of RNA polymerase II.

Figure 1.

HDAC1 and 2 (HDAC1/2) maintain ESC pluripotency in a gene dosage–dependent manner. (A) Microarray data from n = 3 biological replicates showing relative expression (log₂ fold-change) of indicated pluripotency-associated genes in Hdac1/2-KO (DKO) and Hdac1-KO; Hdac2-Het cells at 0 (Ctrl) and 3 d after deletion (OHT). (B) Gene set enrichment analysis (GSEA) plot of microarray data showing Ctrl samples are enriched for the Muller PluriNet (ES = 0.47; P < 0.0001) and pluripotency (ES = 0.70; P < 0.0001) gene sets compared with Hdac1-KO; Hdac2-Het KO cells. (C) Western blot data for HDAC1, HDAC2, NANOG, and POU5F1 proteins in Hdac1/2-KO (DKO) and Hdac1-KO; Hdac2-Het cells at 0 (Ctrl) and 3 d after deletion (OHT). Tubulin, alpha 4A (TUBA4A) was used as a protein loading control.

Figure 2.

HDAC inhibition attenuates pluripotency-associated gene expression without inducing differentiation. (A) Microarray analysis: the number of genes differentially expressed (1.5-fold; FDR ≤ 0.05) at the indicated times following LBH589 treatment. (B) GSEA plot showing enrichment for the defined pluripotency gene set (Kim_Core_Module) in control versus LBH589-treated ESCs. LBH589 2 h (ES = 0.527; P < 0.0001), 6 h (ES = 0.789; P < 0.0001), and 18 h (ES = 0.83; P < 0.0001). (C) GSEA for Ctrl and LBH589 samples showed equal enrichment for the formation of the primary germ layer (GO:0001704; Germ Layer). (D) Microarray data from n = 3 biological replicates showing relative expression of indicated pluripotency-associated genes in ESCs treated with LBH589 for 2, 6, and 18 h. The statistical difference was calculated using Benjamini and Hochberg's false-discovery rate (FDR). Asterisk denotes significant changes in expression (≤1.5-fold; FDR ≤ 0.05) relative to untreated control ESCs.

Figure 3.

LBH589 and JQ1 target the same subset of pluripotency-associated genes and reduce BRD4 recruitment. (A) RT-qPCR analysis of pluripotency-associated gene expression following the treatment of ESCs with LBH589 or JQ1 for the indicated time points. Box plots represent the max–min expression range and mean from three biological replicates. (B) RT-qPCR analysis for Nanog expression in ESCs treated with LBH589 and/or overexpressing BRD4, as indicated. Values represent mean (±SEM) relative expression compared with the wild-type for n = 6 biological replicates. Statistical differences were calculated using a paired Student's t-test.

Figure 4.

HDAC inhibition causes increased expression of promoter upstream transcripts (PROMPTs) and a loss of promotor-proximal polymerase. (A) Heatmaps displaying a log₂-fold change of PRO-seq read counts in 200-bp bins (TSS ± 5 kb) at RefSeq genes displaying increased or decreased pause index changes following JQ1 (2 h) or LBH589 (2 h and 6 h) treatment. Genes were ranked based on the log₂-transformed fold-change of RNA polymerase in the promoter-proximal region. (B) A number of gene clusters (A–H) with similar transcriptional changes are shown. For each treatment, genes were clustered based on change (±1.5-fold; FDR ≤ 0.05) in gene body or paused index changes. (C) Metagene plots (TSS ± 5 kb) of RNAP II densities at genes showing increased gene body levels following the treatment with JQ1 or LBH589. Insets show increased antisense transcripts (PROMPTs) relative to TSS in greater detail. (D) PRO-seq Integrative Genomics Viewer (IGV; Robinson et al. 2011) bedGraph screenshots of Fos and Tpst2 TSS with control, JQ1, or LBH589 treatments for the indicated times. Shaded areas indicate increased antisense transcription in LBH589-treated samples.

Figure 5.

HDACs are required for expression of the pluripotent gene network via recruitment of RNA polymerase II (RNAP II). (A) PRO-seq analysis displaying the effects of JQ1 or LBH589 treatment on the level of gene body transcripts for pluripotent gene network members. Heatmap shows the log₂ fold-change relative to untreated controls. (B) Heatmap shows the effects of JQ1 and LBH589 treatment on the pausing index (the ratio of promoter-proximal to gene body transcript levels) changes relative to untreated controls. (C) Metagene plots (TSS ± 2 kb) of pluripotency-associated genes showing reduced initiation and elongation in LBH589-treated samples. (D) RNAP II density at Nanog, Sox2, and Pou5f1 TSS in either control or JQ1- or LBH589-treated ESCs. (E) ChIP-qPCR for RNAP II at the indicated promoters in either control or JQ1- or LBH589-treated (2 h) ESCs. Bars represent the mean (±SEM) of n = 3 biological replicates. (*) P = <0.05, (**) P = <0.01, (***) P = <0.001, (n.s.) not significant.

Figure 6.

HDACs positively regulate a subset of eRNAs associated with superenhancers (SEs). (A) Volcano plot showing changes in eRNA transcription following treatment with JQ1 or LBH589 for the indicated time points. Statistically enriched (±1.5-fold; FDR ≤ 0.05) eRNA transcripts are indicated in orange, blue, and green. (B) IGV genomic tracks of PRO-seq RNAP II densities at intergenic POU5F1/SOX2/NANOG (PSN) binding sites associated with the indicated SE regions. Shaded areas indicate regions of decreased transcription following LBH589 treatment. (C) Metagene plots of RNAP II density at intergenic PSN binding sites following the indicated treatments.