Accessible chromatin reveals regulatory mechanisms underlying cell fate decisions during early embryogenesis

Abstract

This study was conducted to investigate epigenetic landscape across multiple species and identify transcription factors (TFs) and their roles in controlling cell fate decision events during early embryogenesis. We made a comprehensively joint-research of chromatin accessibility of five species during embryogenesis by integration of ATAC-seq and RNA-seq datasets. Regulatory roles of candidate early embryonic TFs were investigated. Widespread accessible chromatin in early embryos overlapped with putative cis-regulatory sequences. Sets of cell-fate-determining TFs were identified. YOX1, a key cell cycle regulator, were found to homologous to clusters of TFs that are involved in neuron and epidermal cell-fate determination. Our research provides an intriguing insight into evolution of cell-fate decision during early embryogenesis among organisms.

Figure 1

Accessible chromatin demonstrates the epigenetic dynamics across different developmental stages. (A) Pairwise analysis of peaks presented in ATAC-seq samples. Left, heatmap demonstrates the overlapping rate between peaks in each sample, right, histogram showing peak number identified in each sample. (B) Called peak counts for 28 human ATAC-seq datasets as a function of the number of uniquely mapped reads used for peak calling. (C) The average ATAC-seq enrichment of active genes around TSS region. The center of accessible regions was used to produce the distribution plots. The upstream and downstream regions (2 kb) of TSS are mappable. Only part of human samples was presented. Full of the enrichment plots were presented in Fig. S7. (D) Distribution of reads mapped to genome of human samples. (E) Average plots and heatmaps of ATAC-seq signals at ATAC-seq transposase hypersensitive sites (THSs). The regions in the heatmaps are ranked from highest ATAC-seq signal (top) to lowest (bottom). The cluster manually set to 4. (F) Distribution of peaks and DNA methylation marks in chromosomes. Peak density was calculated by average peak counts divided by peak length (kb). Only human plot was showed, the plots of other species were presented in Fig. S8 (G) The IGV views showing the ATAC-seq enrichment near key cell-fate-determined TFs during early embryogenesis.

Figure 2

Genomic and functional annotation of accessible regions. (A) Genomic distributions of enriched accessible regions identified in ATAC-seq samples. THS peaks within TSS ± 3 kb are considered as promoter THS, and those not located in promoters, exons, introns, or UTRs are labeled as distal intergenic. (B) GO functional enrichment analysis of overlapped peaks. Upper part integrates Upset and Venn method to identify overlapping peaks across all human samples. Bottom part is a bar plot of GO enrichment of overlapping peaks. (C) Heatmaps showing the ATAC-seq enrichment (RPKM) (left) and the comparison of ATAC-seq signal within consensus ATAC-seq peaks by Pearson’s Coefficient Correlation algorithm. The colored bubbles represent different samples. The ATAC-seq enrichment signals were normalized by log2(FPKM + 1).

Figure 3

Classification and functional annotation of TFs identified from ATAC-seq samples. (A) Classification of TFs that identified from all ATAC-seq samples. (B) Gene family classification of the identified TFs in human. The gene family information were collected from JASPAR (http://jaspar.genereg.net/). The information of other species were listed in Fig. S12. (C) Functional and expression pattern analysis of TFs identified from ATAC-seq samples. Right part, annotation of all TFs and randomly selected 10 for motif analysis. Top left, heatmaps showing the top 20 enriched GO terms of all TFs using Metascape enrichment. Bottom left, expression patterns of all TFs across different tissues. The expression profiles were obtained from ENCODE database (https://www.encodeproject.org/). And the raw expression matrices were normalized by log2(FPKM + 1).

Figure 4

Expression profiles of TFs involved in early cell fate determination. (A) The expression patterns of stat-of-the-art early cell fate determined TFs during early embryonic development. (B) Exhibiting the expression profiles of cell-fate TFs during early embryogenesis. Bottom, showing the correlations of expression pattern between public-accepted TFs of early cell-fate determination and some that we identified in this research. The normalized expression matrices were collected from NCBI database from GSE101571 accession.

Figure 5

Transcriptional regulatory networks underlying early embryonic cell fate determinations. (A) TFs mainly play roles in Arabidopsis root epidermis, seedling, leaf, and QC (quiescent center) development, orange, blue, and red lines indicated interactions that control seedling, leaf, and QC cell fates, respectively. Green dashed lines represent homology relationship between Arabidopsis TFs and human TFs. (B) Expression profiles of these Arabidopsis TFs during early embryogenesis. The x-axis represents different stages of Arabidopsis early embryonic development, which were annotated in Figure S16. (C) TFs determine the cell fate of C. elegans epidermis. Orange dashed lines represent homology relationship between nematode TFs and human TFs. (D) Expression patterns of these five TFs during C. elegans early embryogenesis. (E) TFs involved in pattern formation of stress responses in yeast. Purple dashed lines represent homology relationship between yeast TFs and human TFs. (F) Expression patterns of these eight TFs during S. cerevisiae cell cycle. (G) TFs participate in cell fate determination of eye, gland, and nerve system in fruit fly. Orange lines demonstrate TF interactions that control nerve system, red lines demonstrate TF interactions that control gland development, green lines demonstrate TF interactions that control eye cell fate. The red dashed lines represent homology relationship between fruit fly TFs and human TFs. (H) Expression patterns of these 21 TFs during D. melanogaster cell cycle. (I) Regulatory circuits determine mouse and human early embryonic cell fate. (J) Expression patterns of these TFs during early embryogenesis of mouse and human. The interaction relationships were all predicted by STRING database. The normalized expression profiles were collected from NCBI database accessions of GSE101571 (human), GSE66582 (mouse), GSE25180 (fruit fly), GSE77944 (nematode), GSE123010 (Arabidopsis), and GSE104904 (yeast).

Figure 6

Evolution basis of members of homeobox TFs and their roles in cell fate decisions. (A) Key cell cycle transition TFs, YOX1, and its orthologs in five other species play roles in cell fate decision. Homologous TFs of YOX1 in human, mouse, fruit fly and nematode are involved in neuron cell fate determination, while in Arabidopsis, control epidermis cell fate. All of the TFs are members of homeobox family. (B) Heat maps showing the expression patterns of the cell-fate TFs during early embryogenesis. E2–24 representing 2–24 embryo stages. The normalized expression matrices were collected from NCBI GSE101571 (human), GSE66582 (mouse), GSE77944 (nematode), GSE123010 (Arabidopsis), modENCODE (fruitfly), and GSE104904 (yeast). (C) Phylogenetic trees of the homeobox TFs across species. Neighbor joining and 500 bootstrap runs were carried out using the protein sequence. (D) Multiple alignment of homeobox TFs across species. The left bar showing the adjusted p value of motif corresponding to each TFs. The right part representing the consensus sequences alignment of the corresponding TFs. The standard numbering of a typical HD (homeodomain) with 60 residues starting from 10 (the upstream 9 residues were not shown) is given at the bottom, and the blue-lined box denote the conserved regions from 20 to 59 of HDs.

Figure 7

An integrated regulatory network model for the control of cell fate decision events during early stage of embryogenesis/cell cycle. This diagram depicts the regulatory interactions of cell fate determination occurring early embryogenesis/cell cycle. networks in black lined boxes represent regulatory interactions corresponding to each species. Transcription factors are denoted by ovals. Dashed grey lines represent orthologs. Green solid lines represent positive regulation and red lines represent negative regulation. Genes in colored boxes represent gene complexes/interactions. The question mark represents the regulatory relationship was unknown. AS-C achaete–scute complex.