Figure 1. Profiling chromatin accessibility dynamics during early cardiopharyngeal cell development.

(A) Embryos, larvae and lineage diagram showing B7.5 blastomeres, their cardiopharyngeal progeny, and the main stages sampled for ATAC-seq. Anterior tail muscle (ATM, gray), trunk ventral cell (TVC, green), secondary TVC (STVC, yellow), first heart precursor (FHP, red), second heart precursor (SHP, orange), atrial siphon precursor cells (ASMF, blue). Stages (St.) according to Hotta et al. (2007) with hours post fertilization (hpf). (B) Spearman correlation of RPKM (reads per kb per million mapped reads) values in 14,178 regions changing accessibility over time or between B7.5 and B-line mesenchyme lineages. (C) Temporal changes in chromatin accessibility for 5,450 regions. ‘B7.5 6 > 10’: 3,691 regions more accessible at Mesp>LacZ 6 hpf than Mesp>LacZ 10 hpf. ‘B7.5 6 < 10’: 1,759 regions more accessible at Mesp>LacZ 10 than Mesp>LacZ 6 hpf. The accessibility of these regions is shown for Mesp>LacZ 6 hpf, Mesp>LacZ 10 hpf, and Hand-r>LacZ 18 hpf vs. the average (avg) accessibility in the control cells. Cell-type-specific chromatin accessibility is shown in the comparison of Mesp>LacZ and MyoD905>GFP at 10 and Hand-r>LacZ and MyoD905>GFP 18 hpf. (D) Gene Set Enrichment Analysis (GSEA) normalized enrichment score of defined gene sets in regions ranked by difference in accessibility between time points as indicated (see Materials and methods).

Figure 1—figure supplement 1. General characterization of the accessome.

(A) Fragment size for B7.5 control samples at developmental time indicated (first three panels from the left) showing consistent approximately 147 bp periodicity, likely corresponding to nucleosome-protected fragments. This periodicity is not present in the purified genomic DNA (gDNA) input control (far right panel). Dotted lines show the cumulative fraction of reads larger than a given size. (B) Proportion of genomic features covered by the accessome and proportion of the accessome covered by genomic features (in bp). (C) Two-tailed binomial test for enrichment of accessible elements overlapping a genomic feature. Accessible elements associated with any DE gene from any scRNA-seq or bulk RNA-seq experiment were considered DE peaks. Bars show the predicted (based on bp coverage of the genome by a feature) and observed probabilities that an accessible element will overlap a genomic feature. Error bars show the 99% confidence interval. (D) Comparison of GC content of genomic features and accessible regions overlapping these features. (E) A 6 kb region displaying MACS2-called peaks in seven ATAC-seq libraries (upper panel), gene model (black) and accessome (gray). (F) Accessible element size distribution.

Figure 1—figure supplement 2. Characterization of promoter regions.

(A) Read density within 2 kb upstream or downstream of TSS shows global closing of TSS over time. (B) Accessibility of the TSS is maintained in response to Fgfr^DN perturbation at 10 hpf. (C) Intersections of overlapping promoter elements. ‘ATAC-seq promoter peaks’ are accessible elements overlapping the putative promoter (−1,107 to +107 bp from the TSS). ‘DE promoters’ are differentially expressed putative promoters in any bulk RNA-seq condition (see Materials and methods). (D) Two-tailed binomial test for enrichment of accessible elements overlapping TSS-seq sites by genomic feature. Bars show the observed probability that an accessible element overlapping a genomic feature will also overlap a TSS-seq element. Error bars show the 99% confidence interval. The expected probability was calculated from the percent of the accessome overlapping TSS-seq elements (3.8%). (E) One-tailed hypergeometric test for enrichment of promoter motifs in genomic features. Eukaryotic core promoter motifs were taken from Haberle and Stark (2018). (F) Spearman correlation of inferred motif enrichment (see Materials and methods). Motifs in each class of element were ranked based on a one-tailed hypergeometric test.

Figure 1—figure supplement 3. Annotation of the accessome.

(A) Example diagram of accessible elements surrounding a gene locus. (B) Distribution of number of genes associated with each peak. (C) Distribution of number of peaks associated with each gene. (D) Gene size vs. number of associated peaks. Red and blue dots show all transcription factor (TF) and signaling molecule (SM) genes, respectively. (E) Density of peaks per kb of each gene. (F) Two-tailed binomial test for enrichment of accessible elements associated with TF or signaling molecule genes. Predicted probability of an element associating to a TF or signaling molecule is given by the proportion of bps in all gene bodies ± 10 kb covering TFs or signaling molecule gene bodies ± 10 kb. Error bars show the 99% confidence interval of the binomial test. Only genes that were differentially expressed in any scRNA-seq or bulkRNA-seq comparison (Wang et al., 2019) were considered.