Super-enhancer-associated transcription factors collaboratively regulate trophoblast-active gene expression programs in human trophoblast stem cells

Abstract

The placenta is an essential organ that supports the growth and development of the fetus during pregnancy. However, cell type-specific enhancers and transcription factors (TFs), and the mechanisms underlying the maintenance and differentiation of trophoblast stem cell (TSC) populations in the human placenta remain elusive. Here, using human TSCs as a model system, we identify 31,362 enhancers that are enriched with the motifs of previously reported TSC-pivotal TFs, including TEAD4, GATA2/3 and TFAP2C. Subsequently, we identify 580 super-enhancers (SEs) and 549 SE-associated genes. These genes are robustly expressed in the human placenta and include numerous TFs, implying that SE-associated TFs (SE-TFs) may play crucial roles in placental development. Additionally, we identify the global binding sites of five TSC-pivotal SE-TFs (FOS, GATA2, MAFK, TEAD4 and TFAP2C), revealing that they preferentially co-occupy enhancers, regulate each other and form a trophoblast-active gene regulatory network. Loss-of-function studies unveil that the five TFs promote self-renewal of TSCs by activating proliferation-associated genes while repressing developmental genes. We further reveal that the five TFs exert conserved and unique functions on placental development between humans and mice. Our study provides important insights into the roles of human TSC-pivotal TFs in regulating placenta-specific gene expression programs.

Figure 1.

ChIP-seq of EP300 in human TSCs revealed global enhancers enriched with the motif of TSC-pivotal TFs. (A) EP300 ChIP-seq (blue) and input (black) signal tracks near TSC marker genes. (B) A pie chart displaying the distribution of EP300-binding sites across the genome. Promoters are within 2 kb from the TSSs, and 'intergenic' is a region excluding a promoter, intron and exon. (C) Enriched GO terms of biological processes in the EP300-binding sites. (D) A boxplot presenting the distribution of EP300 occupancy scores across different genomic loci. P-value was calculated using the Wilcoxon rank sum test. ***P-value <0.0001. (E) Top 10 enriched motifs in the EP300-binding sites.

Figure 2.

SEs control trophoblast-active genes in human TSCs. (A) A line graph showing the number of SEs defined by the ranked EP300 binding signal. (B) A heatmap showing the EP300 occupancy signal around the center of SEs. (C) EP300 ChIP-seq (blue) and input (black) signal tracks near TFs. A red rectangular box indicates an SE. (D) A network map displaying GO terms enriched in molecular functions in SE-associated genes. Node color indicates the false discovery rate (FDR), and the thickness of the line presents the degree of overlapped genes between two terms. (E) KEGG pathways enriched in SE-associated genes. (F and G) Boxplots showing the distribution of expression of SE-TFs in various human tissues (F) and cell types (G). The red line indicates a median expression value among the genes in the placenta tissue. P-value was calculated using the Wilcoxon rank sum test. *P-value <0.01; **P-value <0.001; ns not significant.

Figure 3.

TSC-pivotal TFs collaborate to activate placenta developmental genes. (A) ChIP-seq signal tracks of multiple SE-TFs, EP300 and input near trophoblast-specific gene, KRT8. (B) Distribution of TF-binding sites on the genome. The total number of binding sites is shown on the bar graph. Each color indicates a different genomic region. (C) Heatmaps presenting occupancy signals of various SE-TFs near the center of EP300-binding sites. PC indicates a peak center. (D) A heatmap showing enriched GO terms in the binding sites of various TFs. The color bar indicates a negative log-transformed FDR. (E) A network map illustrating a transcriptional gene regulatory network among various SE-TFs. An arrow points at the target of a TF.

Figure 4.

Co-occupancy of TSC-pivotal TFs is critical to the regulation of TSC-specific and PE-deregulated genes. (A) Heatmaps presenting EP300 occupancy signals at the center of peaks belonging to each group (G1 to G5) that is classified based on the number of TFs co-bound. PC indicates a peak center. (B) Line graphs showing averaged EP300 occupancy (left) and input (right) signals near the center of peaks in the five different groups. (C and D) Boxplots showing the expression distribution of an individual group-associated gene in TSCs (C) and the G1-associated genes in various trophoblast cell types (D). P-value was calculated using the Wilcoxon rank sum test. ***P-value <0.0001. (E) Enriched GO terms of biological processes of G1-associated genes. (F) Percentage of genes in the five groups that are overlapped with PE-deregulated genes. P-values were calculated using a hypergeometric test. PE data were obtained from GSE114691.

Figure 5.

TSC-pivotal TFs directly activate proliferation-associated genes but repress developmental genes. (A) Percentage of up- and down-regulated direct and indirect targets of SE-TFs. (B) Activation and repression activity of TFs predicted by BETA. Red and blue lines indicate the up- and down-regulated genes upon KD of a TF, respectively. The dashed line presents genes that were not differentially expressed as background. The P-value shown in the top left indicates the significance of the Up or Down gene group relative to the Static group as determined by the Kolmogorov–Smirnov test. (C) Enriched motifs in the direct targets of individual TFs. (D) The ratio of repressed to activated genes upon depletion of a TF. Red and blue indicate genes bound by either five TFs or <2 TFs. The x-axis shows the TFs depleted. (E) Gene set enrichment analysis (GSEA) of up- and down-regulated genes upon depletion of TFs between human TSCs and differentiated cells. (F) A heatmap showing enriched GO terms upon KD of a TF. Enrichment scores were obtained from GSEA (KD versus Control).

Figure 6.

TSC-pivotal TFs regulate common and unique target genes between human and mouse TSCs. (A) A heatmap presenting the distribution of TSC-pivotal TF-binding sites across the genome in human and mouse TSCs. (B) A heatmap showing the correlation of the occurrence of TF motifs in the target TFs in human and mouse TSCs. Blue and orange colors indicate TFs in humans and mice, respectively. (C) A heatmap displaying overlaps of TF target genes in human and mouse TSCs. (D) Box plots showing the numbers of TSC-pivotal TFs that are bound on human- and mouse-enriched genes in human and mouse TSCs. H and M indicate humans and mice, respectively. (E) Box plots presenting the relative expression of genes that are classified by the difference in the number of TSC-pivotal TFs bound between human and mouse TSCs. (F) GO terms of biological processes enriched in common targets of six TFs between human and mouse TSCs and unique targets of each species. MMFB and HMFB indicates mouse-multiple-factors-bound and human-multiple-factors-bound, respectively.