3D chromatin structures precede genome activation in Drosophila embryogenesis

Abstract

3D chromatin structure is critical for the regulation of gene expression during development. Here we used Micro-C assays at 100-bp resolution to map genome organization in Drosophila melanogaster throughout the first half of embryogenesis. These high-resolution contact maps reveal fine-scale features such as loops and boundaries delineating topologically associating domains. Notably, we observe that 3D chromatin structures form prior to zygotic genome activation and persist during successive mitotic cycles. Integrative analysis with 149 public chromatin immunoprecipitation sequencing (ChIP-seq) datasets identifies four classes of chromatin structuring elements, including a distinct group enriched for GAGA-associated factor (GAF) and Zelda binding, associated with developmental-gene regulation. These elements are mitotically retained and exhibit sequence and structure similarity between D. melanogaster and D. virilis. We propose that 3D chromatin organization in the pre-cellular embryo facilitates deployment of developmentally regulated genes during Drosophila embryogenesis.

Keywords: ChIP-seq; Drosophila; Micro-C; TAD; chromatin boundaries; chromatin loops; clustering; computation; development; embryo; insulators; nucleosome.

Graphical abstract

Figure 1

Micro-C uncovers chromatin structures with unparalleled resolution (A) Micro-C contact-frequency maps surrounding the gene ftz. Contact-frequency maps for nc1–8 (top), nc14 (middle), and s10–12 (bottom) are shown alongside ChIP-seq tracks. (B) Side-by-side comparison of Hi-C and Micro-C contact-frequency maps at the abd-A locus. The Micro-C data from this publication (upper triangle) are from nc14 embryos. The Hi-C data (Ogiyama et al., lower triangle) were collected from late nc14 embryos. The greater number of orange contact points to the right of the diagonal reflect more frequent contacts discovered/measured with Micro-C. (C) The overlap between loop anchors and boundaries. (D) Boxplot summary of boundary strengths in each embryonic stage. ∗p < 0.001, Bonferroni-corrected two-sided Wilcoxon signed-rank test, n = 3,673. (E) Differential Hi-C signal between embryonic stages at loops (INC, increasing; DEC, decreasing; NS, not significant; see STAR Methods for details). The number of loops with significantly increasing, decreasing, or not significantly different strength are specified for each pairwise comparison.

Figure 2

Joint clustering of 149 ChIP-seq datasets at chromatin boundaries and loop anchors defines four distinct classes of CSEs (A) Schematic of the approach. The compendium of ChIP-seq data for various factors across biological contexts was collected and uniformly preprocessed. Loop anchors and boundaries were aggregated into an atlas of CSEs. The CSEs were clustered based on the vector of ChIP-seq signal across the compendium. (B) Column-normalized heatmap of ChIP-seq signal at each CSE in the atlas. We visualized only ChIP-seq datasets that were significantly enriched and with a log fold change of at least two for at least one cluster relative to all other elements. (C) Bar plots showing the structural composition of each CSE cluster. Enrichment was tested with a Bonferroni-corrected two-sided Fisher’s exact test. (D) Bar plots of loop anchor partnering preferences for each cluster of CSEs. Enrichment was tested with a Bonferroni-corrected two-sided Fisher’s exact test. (E) Bar plots showing the cluster-wise percentage of CSEs within 1 kb of a given annotation. Enrichment was tested with a Bonferroni-corrected two-sided Fisher’s exact test. (F) Bar plot of the cluster-wise fraction of genes within 1 kb of a CSE, which have been annotated as having localized patterning during embryogenesis (patterning gene list taken from Levo et al.¹²). Enrichment was tested with a Bonferroni-corrected two-sided Fisher’s exact test.

Figure 3

A distinct class of boundaries associated with early activation of housekeeping and cell-cycle genes (A) Heatmaps showing H3K4me3 (0–4 h) and RNA Polymerase II (2–3 h) ChIP-seq signal at boundaries in each CSE cluster. Each row of the heatmap represents ChIP-seq signal centered at a CSE. Rows in each cluster are sorted by mean signal in the center 3.2 kb. (B) Fraction of all housekeeping genes within 1 kb of a CSE of each cluster. (∗∗∗p < 1e−4, Fisher’s exact test). (C) Genome Browser view of the SppL locus. Micro-C contact-frequency maps from nc1–8 (top), nc14 (middle), and s10–12 (bottom) are shown alongside ChIP-seq data. SppL and Lnk are linked to housekeeping functions such as proteolysis and core metabolic processes.^, (D) Non-redundant biological process GO terms that were significantly enriched among the set of genes flanking (≤5 kb) CSEs. (E) Temporal trends in the normalized expression levels (±SEM) of genes flanking (≤500 bp) CSEs in each cluster. The mean normalized expression is plotted.

Figure 4

Regulatory elements associated with zygotic expression exhibit similarity of both sequence and structure (A) Heatmaps showing Zld (nc12–14) and GAF (2–4 h) ChIP-seq signal at all CSEs in each cluster. Each row of the heatmap represents ChIP-seq signal centered at a CSE. Rows in each cluster are sorted by mean signal in the center 3.2 kb. (B) Chromatin accessibility at CSEs as measured by ATAC-seq. Accessibility is recorded in 3-min intervals from the beginning of nc11 to the end of nc13. (C) Bar plots showing the cluster-wise similarity of CSEs in D. virilis. Enrichment was tested with a Bonferroni-corrected two-sided Fisher’s exact test. (D) Aligned multi-genome view of Micro-C contact-frequency maps for a locus D. melanogaster (top), which exhibits conserved structure in D. virilis (bottom). Note that LOC6622731 and LOC6622717 are orthologs of trh and klar, respectively. The light gray bands between the two views indicate conserved sequences identified with lastz (STAR Methods). (E) Scatterplot between motif looping frequency (fraction of motif instances in CSEs that overlap looping CSEs) and motif conservation (average PhyloP27 over all motif instances in ATAC-seq peaks in CSEs). r, Pearson correlation.

Figure 5

Nucleosomal analysis reveals distinct patterns of TF regulation (A) MNase coverage at nc14 for all CSE clusters. (B) MNase coverage for all TSSs, sorted by Pol2 (left), BEAF-32 (middle), or GAF (right) binding strength at the TSS. Clusters represent top 5%, 5%–10%, 10%–40%, and bottom 60% TSSs as ordered by the binding strength. (C) MNase coverage or ChIP-seq data plotted around the TSSs with the top 5% of BEAF-32 or GAF binding. Outline represents ±1 SEM.

Figure 6

Developmental chromosomal structural elements are stably maintained throughout early mitoses (A) Experimental method used to isolate mitotic embryos for the nc12 mitotic sample and the new nc14 sample. (B) Boxplot summary of mitotic stability for CSEs separated by cluster. We excluded boundaries missing in nc1–8, nc12M, s10–12, or either of the nc14 samples. All pairwise comparisons of clusters were significant except where noted otherwise (Bonferroni-corrected two-sided Mann-Whitney U test; see supplemental table). (C) Genome Browser view of the robo2 locus. Micro-C contact-frequency maps from the nc12M (top) and nc14v2 (bottom) samples are shown alongside ChIP-seq data. (D) Spearman’s rank correlation between mitotic stability of boundaries and their signal in embryonic ChIP-seq datasets. Significant negative and positive correlations are colored blue and red, respectively (Table S2). (E) Insulation in nc14v2 or nc12M at mitotically retained GAF binding sites (left) or interphase-only GAF binding sites (right) taken from Bellec et al.. Outline represents ±1 SEM. Statistical significance (∗∗∗p < 1e−4; ∗p < 0.05) was tested using a Wilcoxon signed-rank test for insulation at the central value (corresponding to the CSE).