Independence of chromatin conformation and gene regulation during Drosophila dorsoventral patterning

Abstract

The relationship between chromatin organization and gene regulation remains unclear. While disruption of chromatin domains and domain boundaries can lead to mis-expression of developmental genes, acute depletion of regulators of genome organization has a relatively small effect on gene expression. It is therefore uncertain whether gene expression and chromatin state drive chromatin organization, or whether changes in chromatin organization facilitate cell type-specific activation of gene expression. Here, using the dorsoventral patterning of the Drosophila melanogaster embryo as a model system, we provide evidence for the independence of chromatin organization and dorsoventral gene expression. We define tissue-specific enhancers and link them to expression patterns using single-cell RNA-seq. Surprisingly, despite tissue-specific chromatin states and gene expression, chromatin organization is largely maintained across tissues. Our results indicate that tissue-specific chromatin conformation is not necessary for tissue-specific gene expression, but rather acts as a scaffold facilitating gene expression when enhancers become active.

Chromatin is highly organized within discrete chromosome territories, into compartments of active or inactive chromatin, self-interacting domains, and loops between specific loci (reviewed in ^1,2). However, the relationship between the three-dimensional (3D) organization of chromatin and the regulation of gene expression remains unclear. There is considerable evidence that chromatin conformation is important for gene regulation: disruption of domains and domain boundaries can lead to mis-expression of developmental genes, contributing to developmental defects or cancer ^3–8. In addition, the general principles of 3D genome organization are conserved across large evolutionary distances, as well as the chromatin conformation at specific loci ^9–14. Further evidence comes from the identification of interactions between promoters and their regulatory elements ^15–22, and the finding that forced enhancer-promoter looping is sufficient to activate transcription of some genes ^23–26. However, in other cases direct enhancer-promoter contacts may be neither strictly required nor sufficient for gene activation ^27–30. Furthermore, depletion of key regulators of genome organization such as CTCF or cohesin has relatively small effects on gene expression ^31–34, and genomic rearrangements are not always associated with changes in gene expression ^35–38.

While multiple studies have documented differences in chromatin conformation between different cell types or tissues ^39–45, it is not known whether these changes are the cause or consequence of changes in gene expression. Therefore, a fundamental question arises as to whether changes in gene expression and chromatin state drive chromatin reorganization, or whether changes in chromatin organization facilitate cell type-specific activation of genes and their regulatory elements.

Embryonic development requires precise regulation of gene expression, making it an ideal context in which to investigate the relationship between chromatin organization and gene regulation ⁴⁶. In particular, Drosophila melanogaster has long been used as a model organism for the study of development, and the key principles and factors involved in embryonic patterning are well understood ^47,48. Early Drosophila development involves a series of thirteen rapid, synchronous nuclear divisions, before the embryo becomes cellularized and undergoes zygotic genome activation (ZGA) at nuclear cycle (nc) 14 (Fig. 1A)⁴⁹. We and others have previously shown that chromatin organization in Drosophila is established at nc14, coincident with ZGA ^50,51. While a small number of genes are zygotically expressed prior to the major wave of ZGA ⁵², maternally provided cues are responsible for establishing the major anterior-posterior and dorsoventral axes ^53,54. Therefore, by ZGA, cells in different regions of the embryo contain different developmental transcription factors, have different patterns of chromatin accessibility ^55,56, and are primed to express different genes.

Figure 1. Identification of tissue-specific regulatory elements for dorsoventral patterning.

a. Overview of Drosophila early embryonic development. The syncytial blastoderm embryo undergoes 13 cycles of nuclear division before the maternal-to-zygotic transition occurs at nuclear cycle (nc) 14. This involves zygotic genome activation and embryo cellularization, and is followed by gastrulation beginning around 3 hours post fertilization (hpf). ChIP-seq and RNA-seq datasets used in this study were derived from 2-4 hpf embryos, including embryos at the late syncytial blastoderm stage, cellular blastoderm stage, gastrulation, and the beginning of germ band elongation. Single-cell RNA-seq datasets were derived from 2.5-3.5 hpf embryos at the cellular blastoderm and gastrulation stages. Hi-C and Micro-C datasets were derived from hand-sorted cellular blastoderm embryos. b. Dorsoventral patterning of the Drosophila embryo is controlled by the nuclear concentration of Dl. High levels of nuclear Dl specify mesoderm (ME, yellow), intermediate levels specify neuroectoderm (NE, pink), while nuclei without Dl become dorsal ectoderm (DE, blue). The gd⁷, Toll^rm9/rm10, and Toll^10B maternal effect mutations lead to embryos with uniform levels of nuclear Dl, which produce only dorsal ectoderm, neuroectoderm, and mesoderm, respectively. D, dorsal; V, ventral; A, anterior; P, posterior. c. Schematic representation of identification of putative tissue-specific enhancers using csaw ^114,115. We identified H3K27ac-enriched regions (grey shaded areas) and performed pairwise comparisons between genotypes (gd⁷, blue; Toll^rm9/rm10, pink; Toll^10B, yellow) to identify differential H3K27ac between genotypes. Candidate tissue-specific enhancers are enriched for H3K27ac in one genotype compared to both others, and do not overlap a promoter. d. Heatmaps of H3K27ac and H3K27me3 ChIP-seq signal at putative tissue-specific enhancers, normalized counts per million mapped reads in 10-bp bins (CPM) ¹¹⁶. e. Average H3K27ac and H3K27me3 ChIP-seq signal (CPM) at putative tissue-specific enhancers. Shaded area represents +/- one standard deviation from the mean. f. Expression of genes associated with putative tissue-specific enhancers. Top, gd⁷-specific enhancers; middle, Toll^rm9/rm10-specific enhancers; bottom, Toll^10B-specific enhancers. Box plots show median, box spanning first to third quartiles, whiskers extending to smallest/largest values no further than 1.5× the interquartile range (IQR) from the box, and notches extending 1.58 × IQR / sqrt(n) from the median. Outliers are excluded for clarity. Two-sided Wilcoxon rank-sum test; n = 302 gd⁷ enhancer-gene pairs, n = 235 Toll^rm9/rm10 enhancer-gene pairs, n = 337 Toll^10B enhancer-gene pairs. g, h. Examples of chromatin organization at dorsoventral patterning genes (in black; if, g; Doc1, Doc2, Doc3, h). Top, normalized Hi-C contact probability, 2-kb resolution, 3-4 hpf control embryos ⁵⁰. Positive strand genes, orange; negative strand genes, blue. RNA PolII-Ser5p ChIP-seq (CPM) from control embryos, black ¹¹⁷. RNA-seq and H3K27ac ChIP-seq (CPM) are shown in blue (gd⁷), pink (Toll^rm9/rm10, and yellow (Toll^10B) (⁶², this study). Tissue-specific putative enhancers, color-coded bars. Grey shaded area in f, domain with developmentally regulated gene; orange shaded region contains housekeeping genes.

Cell fate along the dorsoventral axis is controlled by the nuclear concentration of the transcription factor Dorsal (Dl) ^57,58, which peaks during nc14 ⁵⁹. Activation of the Toll signaling pathway on the ventral side of the embryo leads to high levels of Dl entering the nucleus, while Dl is excluded from the nucleus on the dorsal side ⁵³. Different levels of Dl concentration are responsible for the specification of different cell fates ^57,58 (Fig. 1B). Maternal effect mutations in the Toll pathway lead to uniform levels of Toll signaling across the whole embryo. Different mutations lead to different concentrations of nuclear Dl, making it possible to obtain females that produce a homogeneous population of embryos that consist entirely of presumptive mesoderm (Toll^10B), neuroectoderm (Toll^rm9/rm10), or dorsal ectoderm (gd⁷) (Fig. 1B). These embryos provide an excellent model system to study tissue-specific regulation during development, which has led to the discovery of key transcription factors, regulatory elements, and processes required for embryo patterning ^58,60–64.

In this study, we use Drosophila dorsoventral patterning as a model system to investigate the relationship between tissue-specific gene regulation and 3D chromatin organization. We focus on the cellular blastoderm stage, approximately 2-3 hours post fertilization (hpf), which is coincident with nc14, establishment of chromatin organization, and the onset of ZGA (Fig. 1A). At this stage, the presumptive dorsal ectoderm, neuroectoderm, and mesoderm have been specified but complex tissues have not been formed. We identify putative regulatory elements involved in dorsoventral patterning, and show that their target genes are developmentally regulated and have a distinct chromatin organization compared to housekeeping genes. We find that while there are clear differences in chromatin state and overall gene expression between gd⁷, Toll^rm9/rm10, and Toll^10B embryos, there is still significant heterogeneity in gene expression at the single-cell level. However, these tissue-specific differences in chromatin state and gene expression are not associated with tissue-specific 3D chromatin organization. Together, these results provide evidence that tissue-specific chromatin conformation is not required for tissue-specific gene expression. Rather, our findings indicate that the organization of the genome into 3D chromatin domains acts as an architectural framework to facilitate correct regulation of gene expression once enhancers become active.

Results

Identification of regulatory elements and genes involved in dorsoventral patterning

In order to understand the relationship between tissue-specific gene regulation and genome organization during embryonic development, we first sought to identify a stringent genome-wide set of candidate tissue-specific regulatory elements involved in dorsoventral patterning. We performed ChIP-seq for H3K27ac, associated with active chromatin, and H3K27me3, associated with repression, in Toll^rm9/rm10 embryos at 2-4 hpf and combined this with ChIP-seq data from 2-4 hpf gd⁷ and Toll^10B embryos from ⁶². Embryos collected at 2-4 hpf will largely consist of embryos in the late cellular blastoderm stage and undergoing gastrulation, thus targeting our timepoint of interest. Note that while H3K27ac is established from nc12 onwards, H3K27me3 is only present from mid nc14 ⁶⁵. Using these data, we carried out genome-wide differential peak identification for H3K27ac (Fig. 1C). We identified 302 regions enriched for H3K27ac in gd⁷ compared to both Toll^rm9/rm10 and Toll^10B, 235 regions specifically enriched in Toll^rm9/rm10, and 337 regions specifically enriched in Toll^10B (Fig. 1D). By requiring significant enrichment in one genotype compared to both other genotypes, we select a highly stringent set of regions with tissue-specific increases in H3K27ac. These putative enhancers overlap with genomic regions that have been shown to drive expression in the expected regions of the embryo (Extended Data Fig. 1D, E). The putative enhancers are also depleted for the repressive chromatin mark H3K27me3 in the embryos in which they are enriched for H3K27ac, compared to the other genotypes (Fig. 1D, 1E), providing further evidence for their tissue specificity. Next, we assigned putative enhancers to target genes using a combination of gene expression data, linear genomic proximity, and chromatin conformation data (see Methods; Supplementary Table 1), and verified that genes assigned to tissue-specific candidate enhancers have significantly higher expression in the tissue where the enhancer is active (Fig. 1F). We conclude that the identified regions represent a stringent set of candidate enhancers associated with regulation of dorsoventral patterning.

Developmentally regulated genes have a distinct regulatory landscape compared to housekeeping genes

We next assessed the chromatin conformation landscape around these tissue-specific regulatory elements and their target genes. Using Hi-C data from 3-4 hpf Drosophila embryos ⁵⁰, we observed that dorsoventral patterning genes are located within self-interacting domains, along with their assigned regulatory elements (e.g. Fig. 1G, grey shaded region; Fig. 1H). These domains are larger than domains not associated with developmentally regulated genes (Extended Data Fig. 1A, B) (mean size 94 kb compared to 66 kb, P < 2.22 × 10^-16). This is in contrast to housekeeping genes, which are enriched at the boundaries between domains and in small domains ^50,66,67 (Fig. 1G, orange shaded region). In addition, domains containing developmentally regulated genes are significantly more likely to overlap with the large regions of high non-coding sequence conservation known as Genomic Regulatory Blocks ^14,68,69 (Extended Data Fig. 1C). These results are robust to different definitions of tissue-specific enhancers (from ⁶², Extended Data Fig. 2) and emphasize the distinct organization of developmentally regulated and housekeeping genes in the Drosophila genome.

Single-cell expression analysis reveals heterogeneity in gene expression within dorsoventral mutant embryos

While the gd⁷, Toll^rm9/rm10, and Toll^10B maternal effect mutants have long been used as models to analyze tissue-specific regulation during dorsoventral patterning ^58,60,61, the extent of cell fate conversion at the single-cell level within these embryos is unknown. Anterior-posterior patterning mechanisms are still active and RNA in situ hybridization experiments suggest that cell fate conversion may be incomplete ⁵⁷. In order to assess the heterogeneity of gene expression and cell identities within these embryos, we carried out single-cell gene expression analysis using the 10X Genomics Chromium platform, analyzing a total of 16,790 high-quality cells across all genotypes. We used embryos at 2.5-3.5 hpf to target the late cellular blastoderm stage when dorsoventral patterning has been established (Fig. 1A) ^59,70. Clustering of single-cell expression profiles from wild type and mutant embryos showed good concordance with bulk RNA-seq (Extended Data Fig. 3A) and identified 15 clusters representing different cell identities within the embryo (Fig. 2A-B, Extended Data Fig. 3C). We identified upregulated marker genes in each cluster and used these to identify the cell identities within these clusters. We identified clusters representing mesoderm (twi, Mlc2, Mdr49), ectoderm (ab, sca, SoxN), in addition to other cell populations such as amnioserosa (Ance, peb), terminal regions of the embryo (fkh), pole cells (pgc), hemocytes (PPO1), and trachea precursors (Osiris gene cluster). Two clusters express cellularization stage genes (bnk, slam) and represent cells from embryos in the earlier stages of cellularization, due to the timed collection. A full list of clusters and cluster marker genes is available in Supplementary Table 2.

Figure 2. Single-cell RNA-seq analysis of gene expression during dorsoventral patterning.

a. Clustering of single-cell gene expression profiles from 2.5-3.5 hpf embryos reveals clusters corresponding to distinct cell populations. b. Expression of cluster marker genes in single cells from control embryos. The top ten marker genes for each cluster based on log₂ fold-change of expression within cluster compared to outside the cluster are shown. Selected marker genes are labelled. c. Single-cell gene expression profiles separated by cell origin. Certain clusters are depleted in the dorsoventral mutant embryos. Blue, area of graph corresponding to ectoderm clusters; pink, neural; yellow, mesoderm. d. Single-cell expression of genes associated with putative tissue-specific enhancers. Color represents Z-score of average expression of genes associated with each group of tissue-specific enhancers.

Visualization of single-cell gene expression profiles from gd⁷, Toll^rm9/rm10, and Toll^10B embryos revealed that specific clusters are depleted in these mutant embryos (Fig. 2C, Extended Data Fig. 3B-C). Cells from clusters representing mesoderm cell fates are almost completely absent in gd⁷ and Toll^rm9/rm10 embryos, and subsets of ectoderm cells are missing in each of the mutants (Fig. 2C, Extended Data Fig. 3B-C). In order to further dissect the ectoderm clusters and identify cells corresponding to dorsal ectoderm and neuroectoderm, we visualized expression of genes assigned to tissue-specific enhancers (Fig. 2D) and known dorsal ectoderm, neuroectoderm, and mesoderm marker genes (Extended Data Fig. 3C, ⁷¹). This revealed that the ectoderm clusters contain distinct subpopulations of cells expressing dorsal ectoderm markers and neuroectoderm markers. These subpopulations correspond to the regions of the cell distribution which are depleted in Toll^rm9/rm10 and gd⁷ respectively (Fig. 2C, compare distributions in the ‘ectoderm’ region). Despite the significant level of cell-to-cell heterogeneity found within the mutant embryos, we observe that certain cell fates are lost. Importantly, the loss of specific cell fates combined with the tissue-specific enhancer usage shown above supports the use of these embryos to model dorsoventral patterning perturbations.

Major features of chromatin organization are maintained across tissues

We next asked how differential usage of regulatory elements and differential gene expression relates to chromatin conformation during dorsoventral patterning. To do so, we generated Hi-C datasets for gd⁷, Toll^rm9/rm10, Toll^10B, and control embryos at the cellular blastoderm stage (late nc14, approximately 2.5-3 hpf), at 2-kb resolution (Supplementary Table 3).

Systematic comparison of the Hi-C datasets across genotypes revealed that on average, characteristic features of chromatin conformation are similar across datasets (Fig. 3). Saddle plots reveal similar strength of compartmentalization in control and gd⁷, Toll^rm9/rm10, and Toll^10B embryos (Fig. 3A, B). We next analyzed overall self-interacting domain strength using domains identified in control 3-4 hpf embryos as a reference (Fig. 3C). While domain strength is weaker in cellular blastoderm embryos than at 3-4 hpf, the strength is similar across all genotypes, suggesting that the vast majority of domains and domain boundaries are present in all tissues. We obtained similar conclusions when we examined chromatin loop strengths, using loops from Kc167 cells ⁷² as a reference (Fig. 3D, E), indicating that loops are maintained across tissues. Finally, we analyzed genome-wide contact probability decay with distance (P(s)). A shallow slope at distances < 100 kb reflects local chromatin compaction into domains, while the flattening of the slope around separation distances of 1 Mb indicates compartment formation (^33,73,74, Fig. 3F). We also examined the derivative of P(s), since this can highlight differences in the strength of domain formation (Fig. 3G) ⁷⁴. These analyses reveal differences in these profiles at distances > 5 Mb, which correspond to genomic rearrangements on balancer chromosomes present in a subset of Toll^rm9/rm10 and Toll^10B embryos (Extended Data Fig. 4). Combined, our results demonstrate that overall genome organization at the level of compartments, domains, and chromatin loops is highly similar across genotypes, suggesting that it is maintained across tissues in cellular blastoderm embryos.

Figure 3. Global chromatin conformation along the dorsoventral axis.

a. Chromatin conformation for a 1.8-Mb region of chromosome 2L in gd⁷, Toll^rm9/rm10, Toll^10B, and control embryos at cellular blastoderm stage, and control embryos at 3-4 hpf ⁵⁰. b. ‘Saddle-plot’ representing genome-wide average chromatin compartmentalization. Active regions interact with other active regions (top left), while inactive regions interact with other inactive regions (bottom right). c. Aggregate analysis of domains identified using Hi-C data from 3-4 hpf embryos at 2-kb resolution. d. Aggregate analysis of chromatin loops identified in ⁷². e. Chromatin conformation for a 300-kb region of chromosome 2L. f. Average contact probability by distance for control (black), gd⁷ (blue), Toll^rm9/rm10 (pink), and Toll^10B (yellow) embryos. g. The derivative of the expected contact probability by distance, highlighting differences between samples at far-cis distances due to the presence of rearranged balancer chromosomes in the Toll^rm9/rm10 and Toll^10B embryos.

Chromatin conformation at developmentally regulated genes is similar across tissues despite differences in gene expression and chromatin state

In order to systematically assess chromatin conformation across the genome and identify regions with differences, we used CHESS ⁷⁵, an approach for differential chromatin conformation detection based on computer vision techniques. Briefly, Hi-C submatrices are compared genome-wide between pairs of datasets to produce a similarity score and a signal to noise ratio for each pair of genomic windows (see Methods). Using this approach, we compared control and mutant embryos at the cellular blastoderm stage at 5-kb resolution and with a 500-kb window size. As a reference, we compared control cellular blastoderm stage data from this study with Hi-C data from nc14 embryos from ⁵⁰. Subtracting this reference score from the score for each control-mutant comparison allowed us to identify regions with specific differences in genome organization between control and mutant embryos and exclude regions with low similarity scores due to noise. This analysis revealed that most regions across the genome do not display significant differences in 3D chromatin organization between gd⁷, Toll^rm9/rm10, and Toll^10B embryos (Fig. 4A, F, Extended Data Fig. 4 and 5). This agreed with visual examinations of control-mutant difference matrices (Fig. 4D). The subset of regions that do display strong changes in chromatin organization between genotypes can be attributed to genomic rearrangements present on balancer chromosomes in a subset of the Toll^rm9/rm10 and Toll^10B embryos, rather than correlating with the locations of genes that are differentially expressed (Fig. 4A-B,D, Extended Data Fig. 4-6). Of two example loci on chromosome 2 with changes in chromatin organization identified by CHESS in Toll^10B embryos, while one is close to a cluster of Toll^10B-specific enhancers (Extended Data Fig. 5B), the other occurs in a region devoid of H3K27ac, H3K27me3 or RNA-seq signal in any of the mutant genotypes (Extended Data Fig. 5D). To further investigate the relationship between changes in chromatin conformation and gene expression, we analyzed CHESS similarity scores in windows containing genes that are differentially expressed in the mutant embryos ⁶² compared to other genomic windows (Fig. 4C). This revealed a lack of association between differential gene expression and differential chromatin structure genome wide.

Figure 4. Chromatin conformation is not affected by tissue-specific gene expression.

a. CHESS ⁷⁵ similarity scores (ssim) were calculated between mutant and control embryo Hi-C datasets (5-kb resolution, 500-kb window size). As a reference, similarity scores were calculated between control embryo Hi-C data from this study and ⁵⁰. The difference between reference ssim and control/mutant ssim is shown for chromosome 3R (blue, gd⁷; pink, Toll^rm9/rm10; yellow, Toll^10B). Values around zero indicate similar chromatin conformation in control and mutant, while negative values indicate greater differences between control and mutant than between the reference control datasets ⁵⁰. Shaded area, +/- two standard deviations from genome-wide mean. Grey ticks, genes differentially expressed in dorsoventral mutant embryos ⁶². Asterisks, known breakpoint positions for balancer chromosomes present in a subset of Toll^rm9/rm10 and Toll^10B embryos (TM6 and TM3 respectively; ^35,118). b. Example of ssim differences for a 2-Mb region on chromosome 3R. Low ssim differences correlate with the positions of known breakpoints (asterisks) rather than differentially expressed genes. Hi-C data for a subset of this region are shown in d. c. Boxplots of ssim differences for genomic windows with and without genes that are differentially expressed (DE) between the indicated genotype and both other genotypes ⁶². Since adjacent windows overlap, every hundredth window was selected to obtain a non-overlapping set of windows (n = 101 genomic windows with DE genes and 124 without for gd⁷; n = 74 windows with DE genes and 152 without for Toll^rm9/rm10; n = 183 windows with DE genes and 41 without for Toll^10B). P values, two-sided Wilcoxon rank sum test. Box plots show median, box spanning first to third quartiles, whiskers extending to smallest/largest values no further than 1.5 × IQR from the box, and notches extending 1.58 × IQR / sqrt(n) from the median. d. Tissue-specific chromatin data for a 1-Mb subset of the region on chromosome 3R shown in c above. For each genotype, top, normalized Hi-C contact probability and contact probability difference, 5-kb resolution; red, increased contact probability in embryos of the mutant genotype; blue, decreased contact probability. Arrows, changes in chromatin conformation in Toll^rm9/rm10 and Toll^10B embryos at known balancer chromosome breakpoints. Bottom, RNA-seq (CPM) ⁶², H3K27ac and H3K27me3 ChIP-seq data (CPM; ⁶², this study). Tissue-specific putative enhancers, color-coded bars beneath corresponding H3K27ac ChIP-seq track. Lower panel, positive-strand genes, orange; negative strand genes, blue. See also additional example loci in Extended Data Figure 5. e. Tissue-specific chromatin conformation data for a 150-kb region around the Doc1, Doc2 and Doc3. For each genotype, top, normalized Hi-C contact probability, 2-kb resolution; middle, RNA-seq data (CPM) ⁶²; bottom, H3K27ac and H3K27me3 ChIP-seq data (CPM; ⁶², this study). Tissue-specific putative enhancers, color-coded bars beneath corresponding H3K27ac ChIP-seq track. Bottom, “virtual 4C” tracks representing interactions ofa 2-kb region around the promoter of Doc1. Positive-strand genes, orange; negative strand genes, blue. See also additional example loci in Extended Data Figure 6.

To further validate these observations, we visually assessed genome organization in gd⁷, Toll^rm9/rm10, and Toll^10B embryos at known differentially expressed dorsoventral patterning genes (Fig. 4E, Extended Data Fig. 6). Examination of chromatin conformation and chromatin state data at these regions does not indicate any differences in domain organization, boundary formation, or loop formation. For example, the Doc1, Doc2 and Doc3 genes, which encode T-box transcription factors that are specifically expressed in gd⁷ embryos and required for amnioserosa differentiation and dorsolateral ectoderm patterning ⁷⁶, lie in a well-insulated domain that contains multiple gd⁷-specific putative enhancers and is enriched for H3K27me3 in all three mutants at 2-4 hpf, although to a lesser extent in gd⁷ (Fig. 4E). There is no evidence of changes in the insulation of this domain in gd⁷, where the genes are active, compared to the other datasets, nor is there any change in its internal structure such as changes in interactions between enhancers and the target gene promoters. Similar conclusions were obtained from examining additional loci, including the pnr locus (expressed in gd⁷) and the NetA/NetB, if, and sna loci which are active in Toll^10B (Extended Data Fig. 6). To further investigate the effects of tissue-specific enhancer activity and gene expression on chromatin organization, we examined insulation scores ⁷⁷. Average insulation score is low at domain boundaries (Extended Data Fig. 7A), but does not show local decreases at enhancers or the transcription start sites of differentially expressed genes. Importantly, average insulation scores do not change across genotypes (Extended Data Fig. 7), providing further evidence for the maintenance of chromatin conformation. Together, these results suggest that tissue-specific gene expression and enhancer activity do not necessarily involve changes in domain organization.

Chromatin state and 3D organization are still being established at the cellular blastoderm stage ^{50,51,65,78,79}. Therefore, we carried out additional Hi-C experiments in control and gd⁷, Toll^rm9/rm10, and Toll^10B embryos at a later developmental stage (stage 10, approximately 4-5 hpf) to assess whether tissue-specific genome organization develops later in development after the full establishment of histone modifications ⁶⁵. Aggregate analysis of compartments, domains, and loops (Extended Data Fig. 8), revealed that these features are maintained across tissues at stage 10. Inspection of individual loci containing dorsoventral patterning genes (Extended Data Fig. 9) also confirmed that their chromatin organization is similar. A small number of loci showed tissue-specific changes, such as the dpp locus where a small domain encompasses the expressed portion of the gene in gd⁷, but is absent in the other tissues. Overall, these results suggest that most developmentally regulated genes do not develop tissue-specific chromatin organization over this developmental period, further arguing that tissue-specific chromatin organization is not required for tissue-specific expression during this developmental transition.

Micro-C reveals maintenance of fine scale chromatin conformation and enhancer-promoter interactions across tissues

In order to obtain higher resolution data, we carried out Micro-C ⁸⁰ in control and gd⁷ embryos at the cellular blastoderm stage. This allowed us to examine chromatin conformation at dorsoventral patterning genes at 500-bp resolution (Fig. 5A-B). At this resolution, additional structures were revealed. For example, the region adjacent to the if locus containing housekeeping genes appears unstructured in Hi-C data but several small domains (5-10 kb) are visible in Micro-C data (compare Fig. 1G, Fig. 5A). In addition, loops adjacent to the promoters of Doc1, Doc2, and Doc3 that were visible in Hi-C data from 3-4 hpf embryos but not cellular blastoderm embryos were apparent in Micro-C data from these embryos (compare Fig. 1H, Fig. 4E, Fig. 5B). However, some structures that have been identified in Micro-C in mammalian systems were notably absent here: we did not detect ‘stripes’ originating from active promoters or prominent enhancer-promoter loops ^20,21. At this increased level of resolution, there were nevertheless few differences between control and gd⁷ chromatin organization at dorsoventral patterning genes (Fig. 5A-B, Extended Data Fig. 10), further supporting the idea that differential chromatin organization is not required for differential gene expression during dorsoventral patterning.

Figure 5. Enhancer-promoter interactions do not correlate with tissue-specific enhancer activity and gene expression.

a, b. High resolution chromatin organization at dorsoventral patterning genes (in black; if, a; Doc1, Doc2, Doc3, b). Top, normalized Micro-C contact probability, 500-bp resolution. RNA-seq (CPM), and H3K27ac and H3K27me3 ChIP-seq (CPM), in blue (gd⁷), pink (Toll^rm9/rm10, and yellow (Toll^10B) (⁶², this study). Tissue-specific putative enhancers, color-coded bars. “Virtual 4C” tracks representing interactions of a 2-kb region around the promoter of if (a) or Doc1 (b). Positive-strand genes, orange; negative strand genes, blue. c. Schematic representation of construction of enhancer-promoter interaction aggregates shown in d and e. Enhancers (grey bars) were assigned to putative target genes (see Methods). Enhancers within 5 kb of their assigned promoter are excluded. Regions of the interaction matrix corresponding to enhancer-promoter interactions (circles) are extracted and averaged across sets of tissue-specific enhancers. d, e. Aggregate contact analysis for putative tissue-specific enhancers (E) and the promoters (P) of their assigned genes. The average observed/expected contact probability is shown for Hi-C data at 2-kb resolution in a window of 60 kb around putative enhancer-promoter interactions (d), or for Micro-C data at 1-kb resolution in a window of 30 kb (e). Rows represent enhancer sets; columns represent genotypes. f. Quantification of contact probability between putative enhancers and their assigned target promoters. Panels represent enhancer sets; x-axis represents Hi-C and Micro-C data from different genotypes. There are no significant differences in interaction strength between control and mutant datasets (two-sided Wilcoxon rank-sum test; n = 302 gd⁷ enhancer-gene pairs, n = 235 Toll^rm9/rm10 enhancer-gene pairs, n = 337 Toll^10B enhancer-gene pairs). Box plots show median, box spanning first to third quartiles, whiskers extending to smallest/largest values no further than 1.5 × IQR from the box, and notches extending 1.58 × IQR / sqrt(n) from the median.

Finally, traditional models of gene regulation by enhancers predict that interactions between regulatory elements and their target promoters increase upon tissue-specific gene expression ⁸¹. While punctate enhancer-promoter interactions were not visible in Hi-C or Micro-C data and changes in enhancer-promoter interaction strength were not apparent upon visual examination (Fig. 4E, Fig. 5A-B, Extended Data Fig. 6 and 10), we performed aggregate analysis to systematically determine whether subtle changes in interaction strength manifest at these loci (Fig. 5C). This revealed that there is no significant increase in Hi-C or Micro-C interaction frequency between enhancers and their target promoters in the tissue in which the enhancer is active (Fig. 5D-F). This suggests that increased enhancer-promoter interaction frequency is not required for the tissue-specific expression of dorsoventral patterning genes. Taken together, our results provide evidence for the independence of tissue-specific gene expression and chromatin conformation during dorsoventral patterning.

Discussion

Previous studies have produced conflicting results regarding the relationship between gene expression, chromatin state, and 3D chromatin organization. Here, we set out to understand this relationship in the context of embryonic development in Drosophila. Using the well-studied dorsoventral patterning system, we have shown that despite significant differences in chromatin state and gene expression between tissues along the dorsoventral axis of the embryo, chromatin conformation is largely maintained across tissues. This suggests that cell type-specific gene regulation does not require cell type-specific chromatin organization in this context. Nevertheless, developmentally regulated genes and enhancers are organized into chromatin domains. We suggest that this organization plays a permissive role to facilitate the precise regulation of developmental genes.

We made use of maternal effect mutations in the Toll signaling pathway which lead to embryos that lack the usual patterning of the dorsoventral axis ⁵³ and have long been used as a system to study the specification of mesoderm (Toll^10B), neuroectoderm (Toll^rm9/rm10), and dorsal ectoderm (gd⁷) cell fates as well as the regulation of tissue-specific gene expression ^60–64. However, these embryos are still under the influence of anterior-posterior patterning signals and do not show completely uniform cell identities ⁶⁰. We sought to investigate heterogeneity of cell identity at the single-cell level by using single-cell gene expression profiling. This revealed that certain cell types are indeed maintained in all three Toll pathway mutants, including pole cells and other terminal region cell identities, hemocytes, and trachea precursor cells (Fig. 2). However, heterogeneity of gene expression is reduced in the mutants, as shown by the loss of cells assigned to mesoderm clusters in gd⁷ and Toll^rm9/rm10 embryos, and the depletion of ectoderm subsets in each of the mutants. These datasets showcase the advantages of measuring cellular heterogeneity at the single-cell level and provide a useful resource for further characterization of these embryos and investigation of the regulation of dorsoventral patterning.

Although the gd⁷, Toll^rm9/rm10, and Toll^10B embryos still have heterogeneous gene expression profiles, nevertheless there are clear differences in chromatin state and overall gene expression between them ^60–63. We expanded on previous studies by identifying putative enhancers specific to neuroectoderm in addition to dorsal ectoderm and mesoderm. This allowed the identification of tissue-specific putative enhancer-gene pairs, which correspond well with known dorsoventral patterning enhancers and genes that are differentially expressed across the dorsoventral axis. These regulatory elements and their target genes are located inside chromatin domains, distinct from the enrichment of housekeeping genes at domain boundaries ^{50,66,67,72,82,83}. This is in line with previous results that suggest that 3D chromatin domains act as regulatory domains ^14,84–87.

We find that this domain organization is maintained across tissues, even in cases where there are significant changes in the local chromatin state and gene expression (Fig. 3 and Extended Data Fig. 4). This is consistent with earlier results from Hi-C experiments carried out in anterior and posterior embryo halves, which also showed no differences ⁸⁸ and with previous studies in Drosophila cell lines and other systems, which suggested that domains are widely conserved across different tissues and even different species ^13,43,67,89. In order to explain this maintenance of organization across cell lines, it has been proposed that active chromatin, especially at broadly expressed genes, is responsible for partitioning the genome into domains ^9,67. Rowley et al. ⁹ proposed that compartmentalization of active and inactive chromatin, at the level of individual genes, underlies the formation of insulated chromatin domains. This model predicts that when a developmentally regulated gene is active, its domain would merge with, or have increased interactions with, neighboring domains containing active genes such as broadly expressed housekeeping genes. Our results do not support this model, as we find no evidence that new domains are driven by changes in chromatin state or by active expression of developmentally-regulated genes. In contrast this supports the idea that, similarly to mammalian domain architecture, additional factors such as insulator proteins modulate domain organization in Drosophila ^2,90. Therefore, based on current data, we do not believe that active transcription is the key determinant of 3D chromatin organization in this system.

While overall and locus-specific chromatin organization are maintained across tissues, our Hi-C and Micro-C analyses identify a small number of examples of regions that do have changes in organization (Extended Data Figs. 5, 8, 9). However, at these loci there is no clear relationship between changes in organization and changes in chromatin state or expression, and the vast majority of developmentally regulated loci in this system do not have changes. It will be important for future studies to further investigate these loci to understand what drives these rare changes.

We also investigated chromatin organization at the level of enhancer-promoter interactions. Previous studies have produced conflicting results about whether these interactions are correlated with tissue-specific activation of gene expression. We found no evidence for widespread enrichment of interactions between enhancers and their target promoters, including in tissues where they are active. This is in contrast with previous studies using 3C approaches that have found evidence of enriched enhancer-promoter interactions ^{15,16,18,20,21}, which may precede ^19,22 or correlate with ^40,45 transcriptional activation. Notably, Ghavi-Helm et al. 2014 found that a subset of Drosophila long-range enhancer-promoter pairs do form stable interactions that are enriched above local background ¹⁹. While these loops are visible in our dataset, our results suggest that such loops are not likely to be the primary mechanism of promoter regulation during Drosophila development, perhaps because most enhancers are close to their target promoters. Many stable loops in the Drosophila genome are instead associated with Polycomb-mediated repression ^51,91.

Hi-C provides information about the average conformation across a population of hundreds of thousands of nuclei, which contain dynamic ensembles of different 3D conformations ^90,92–100. While our scRNA-seq results indicate that the mutant embryos contain a range of different cell types, we believe that our results indicate that the 3D chromatin structures in these cell types are drawn from the same population of possible conformations. This is supported by results from a recent study ¹⁰¹ that analyzes the structure of the Doc and sna loci in Drosophila embryos using Hi-M, a high-resolution single-cell imaging approach. Strikingly, this orthogonal technique also reveals chromatin organization that is consistent across different tissues in the embryo, despite differential expression of these genes. Imaging-based approaches directly measure spatial proximity between genomic loci, whereas Hi-C and Micro-C rely on crosslinking to detect chromatin interactions. Therefore, care must be taken when comparing these approaches. Nevertheless, both approaches indicate that genome organization is maintained across different tissues in this system.

Our results are consistent with several recent studies in mammals as well as Drosophila, which provide evidence that stable enhancer-promoter contacts are not always required for gene activation ^27–30,102. This is in line with models in which transient or indirect contacts with a regulatory element are sufficient to activate transcription ^102–106, such as through the formation of nuclear microenvironments or phase-separated condensates ^107–109.

Together, our results indicate that differential chromatin organization is not a necessary feature of cell-type specific gene expression. We propose that chromatin organization into domains instead provides a scaffold or framework for the regulation of developmental genes during and after the activation of zygotic gene expression (Fig. 6, left and middle). This helps render developmental enhancers “poised” for timely regulation of target genes upon receipt of appropriate cellular signals (Fig. 6, right). Other mechanisms of priming have been described, including paused Pol II at promoters ^110,111 and pioneer factors bound to poised enhancers ^64,112,113. Feedback effects such as downstream modification of chromatin state and additional mechanisms including looping between Polycomb-bound elements and segregation of active and inactive chromatin may then act as layers on top of the initially established domain structure.

Figure 6. Model of the relationship between chromatin conformation and developmentally regulated gene expression.

Before ZGA, the genome is unstructured (left), with domain boundaries appearing at a subset of regions associated with binding of RNA PolII and Zelda. Chromatin domains are established at ZGA, and domain structure is the same across tissues with different gene expression and transcription factor binding (middle). Differential activity of regulatory elements in the context of the same chromatin conformation leads to different patterns of gene expression in the developing embryo (right). Thick and thin blue bars represent high and low levels of H3K27me3 respectively; dashed lines represent inactive genes while solid lines represent actively transcribed genes.

Methods

Drosophila stock maintenance

yw; eGFP-PCNA flies used as controls for Hi-C and the first scRNA-seq control experiment were kindly provided by S.A. Blythe and E. Wieschaus ⁷⁸ and maintained on standard cornmeal-agar food. The w¹¹¹⁸ flies used for the second scRNA-seq control experiment and the Toll mutant fly stocks gd⁷/winscy hs-hid, Toll^10B/TM3 e Sb Ser/OR60 and Toll ^rm9/rm10/TM6 e Tb Sb were grown on potato mash-agar food. All fly stocks were incubated at 25 °C with a 12-hour light/dark cycle.

The embryos representing presumptive dorsal ectoderm were collected from gd⁷ homozygous flies. One-day old larvae laid by gd⁷/winscy hs-hid were heat shocked for 1.5 h at 37 °C twice with 24 h interval to eliminate gd⁷ heterozygous animals. Embryos from Toll^10B/TM3 e Sb Ser or Toll^10B/OR60 heterozygous females represented presumptive mesoderm. Toll ^rm9 /Toll ^rm10 trans-heterozygous females were used for collecting presumptive neuroectoderm embryos.

ChIP-seq

Toll ^rm9 /Toll ^rm10 2-4 h old embryos for ChIP sequencing were collected and fixed as described above for Hi-C. Fixed embryos were flash-frozen in liquid nitrogen and stored at -80 °C until further use. Frozen embryos were homogenized in sonication buffer (50 mM HEPES, 140 mM NaCl, 1 mM EDTA, 1% Triton, 0.1% sodium deoxycholate, 0.1% SDS and protease inhibitor) using a Dounce homogenizer. The sample were spun at 4,000 g for 5 min and the pellet containing the intact nuclei was resuspended in same buffer supplemented with 0.5% N-Lauroylsarcosine and SDS to final concentration of 0.5%. The chromatin was sheared to fragment size in the range of 200–500 bp using a Bioruptor (Diagenode). The solubilized chromatin fraction was cleared by centrifugation and used for immunoprecipitation after diluting 5 times with sonication buffer. Immunoprecipitation with either 2 μg of H3K27ac (Abcam, ab4729) or 5 μg of H3K27me3 (Abcam, ab6002) antibody was carried out on chromatin corresponding to 20-25 μl of embryos at 4 °C overnight. Chromatin-antibody complexes were captured for at least 3 h using a mix of Protein A and G Dynabeads (Invitrogen). The captured immunoprecipitated complex were washed 10 min each with sonication buffer (50 mM HEPES, 140 mM NaCl, 1 mM EDTA, 1% Triton, 0.1% sodium deoxycholate, 0.1% SDS), WashA (as sonication buffer, but with 500 mM NaCl), WashB (20 mM Tris pH 8, 1 mM EDTA, 250 mM LiCl, 0.5% NP-40, 0.5% sodium deoxycholate) and TE. After the washes Dynabeads with bound chromatin-antibody complexes were resuspended in 100 μl TE supplemented with 20 mg/ml RNase A and incubated at 50 °C for 30 min. Cross-linking was reversed by adding Tris pH 8.0 and SDS to a final concentration of 50 mM and 0.1% respectively and heating at 68 °C for at least 4 h. Protein digestion was carried out by Proteinase K treatment at 55 °C for 2 h, followed by purifying ChIP DNA using ChIP DNA Clean & Concentrator kit (Zymo research #D5205). ChIP-seq libraries were prepared on the ChIP DNA eluted in 60 μl of DNA elution buffer, using the NEBNext Ultra II DNA Library Prep Kit (NEB). ChIP samples were single-end (1 × 75 bp) sequenced on Illumina NextSeq platform at BEA core facility, Stockholm.

scRNA-seq

We adapted the collection and methanol fixation procedures described in ^71,119. Following a pre-collection period of at least one hour, fly embryos were collected on yeasted apple juice plates at 25 °C. After 1 hour of collection, the embryos on the plate were incubated at 25 °C for 2.25 hours. Embryos were dechorionated for 2 min in 2.6% sodium hypochlorite, rinsed with water, and suspended in PBS, 0.5% Triton X-100. Embryos were rinsed with cell culture grade DPBS without Ca²⁺ and Mg²⁺ to remove residual detergent, and placed on ice at precisely 3.5 hours after the start of collection. Embryos were resuspended in 500 μl ice-cold dissociation buffer (cell culture grade DPBS without Ca²⁺ and Mg²⁺, 0.04% BSA) and dissociated with a clean metal pestle. Cells and tissue fragments were pelleted at 500 x g for 5 min at 4 °C, then gently resuspended in 100 μl Trypsin-EDTA 0.25% and incubated for 3 minutes. After 3 minutes Trypsin was quenched by adding 1 ml cell culture grade DPBS without Ca²⁺ and Mg²⁺, 10% FCS. Cells were pelleted at 1,000 g for 5 min at 4 °C, then resuspended in 500 μl dissociation buffer, pelleted again and resuspended in 100 μl dissociation buffer. A 10 μl aliquot of cells was kept and counted using an improved Neubauer chamber or a Luna2 cell counter. To fix cells, 4 volumes of 100% methanol, pre-chilled at -20 °C, were slowly added to the cells. Fixed cells were stored at -80 °C and used within 3 days.

We performed scRNA-seq using the 10X Genomics Chromium Single Cell 3’ Reagents v3, according to the manufacturer’s instructions (Rev B). Methanol-fixed cells were spun at 3,000 g at 4 °C for 5 min and resuspended in 500 μl DPBS + 0.04% BSA to rehydrate. Rehydrated cells were counted using a Luna2 cell counter, and the volume used for library preparation was chosen for a targeted recovery of 5,000 cells. Libraries were sequenced on an Illumina NextSeq 500, using paired-end sequencing with read 1 length 28 cycles, index read length 8 cycles, and read 2 length 91 cycles.

Hi-C

We adapted the fixation and sorting procedure described in ¹¹⁷ for in situ Hi-C ^120,121. Following a pre-collection period of at least one hour, fly embryos were collected on yeasted 0.4% acetic acid agar plates or apple juice plates at 25 °C. After 1 hour of collection, the embryos on the plate were incubated at 25 °C for 2 hours for collection of cellular blastoderm embryos and 4 hours for collection of stage 10 embryos. Embryos were dechorionated for 2 min in 2.6% sodium hypochlorite, rinsed with water, and transferred to vials containing 2 ml of PBS, 0.5% Triton X-100 and 6 ml of heptane. Cross-linking was initiated by adding 100 μl of 37% formaldehyde, followed by vigorous shaking. After 10 min, samples were spun at 500 g for 1 min and the upper heptane layer was removed. 15 min after the start of fixation, 5 ml of PBS, 0.5% Triton X-100, 125 mM glycine was added to the embryos, followed by vigorous shaking for 1 min. The embryos were rinsed 3 times with PBS, 0.5% Triton X-100. Embryos were sorted in small batches under a light microscope, based on morphology, to select embryos of the appropriate developmental stage and remove damaged embryos or embryos with abnormal morphology. Embryos from gd⁷, Toll^rm9/rm10, and Toll^10B mutant mothers do not gastrulate correctly and do not have normal morphology at stage 10; therefore sorting of these embryos was largely based on removal of embryos that were clearly dead, dying, or from earlier or later stages. Sorted embryos were aliquoted so that a single tube contained enough embryos for one experiment, then flash-frozen in liquid nitrogen and stored at -80 °C. We used 30-60 embryos for each in situ Hi-C experiment.

We performed in situ Hi-C according to the protocol in ^50,120,121, using MboI as the restriction enzyme, with minor modifications for optimization for low input according to ¹²².

Micro-C

Embryos were collected as in Hi-C with minor modifications. Following the first crosslinking with formaldehyde, the reaction was quenched with 2 M Tris-HCl pH 7.5 (0.75 M final). Embryos were washed with PBST, and a second crosslinking step was carried out using long crosslinkers DSG and EGS (Sigma) at 3 mM final in PBST, for 45 mins at room temperature with passive mixing. The reaction was quenched again with Tris-HCl pH 7.5 at 0.75 M final for 5 mins. Embryos were washed, sorted under a microscope, snap frozen with liquid nitrogen, and stored at -80 C. Micro-C libraries were constructed according to ²¹, with modifications. At least 300 nc14 embryos were used per library. Embryos were crushed in the Eppendorf tube with liquid nitrogen cooled plastic pestles using 500 μl buffer MB1 (50 mM NaCl, 10 mM Tris, 5 mM MgCl₂, 1 mM CaCl₂, 0.2% NP-40, 1× PIC). Chromatin was digested with a pre-determined amount of Micrococcal Nuclease (Worthington Biochem) to yield 90% monomer vs. 10% dimer given the appropriate number of embryos. When embryos were limited, as in gd⁷ libraries, size selection of dinucleosomal DNA was not carried out via gel extraction. Instead, total DNA was carried over into the final library construction phase, and finally size selected for the appropriate di-nucleosomal band (350-500 bp) on a 3.5% NuSieve agarose gel after PCR. Libraries were pair-end sequenced on an Illumina Novaseq S1 100 nt Flowcell, with read 1 length 50 cycles, index read length 6 cycles, and read 2 length 50 cycles.

ChIP-seq analysis

ChIP-seq reads were mapped to the dm6 genome using Bowtie2 (version 2.3.3.1 ¹²³). Mapped reads were filtered to remove alignments with quality scores less than 30, as well as secondary and supplementary alignments. PCR duplicates were marked using sambamba (version 0.6.8, ¹²⁴). Coverage tracks were generated using the bamCoverage tool from deepTools (version 3.2.0, ¹¹⁶) with the following parameters: “-of bigwig --binSize 10 --normalizeUsing CPM --extendReads 200 --ignoreDuplicates --minMappingQuality 30” and keeping only reads from chromosomes X, 2L, 2R, 3L, 4, and Y. ChIP-seq peaks were called using MACS2 (version 2.2.6, ¹²⁵) with the following parameters: “--nomodel --extsize 147 -g dm ” or “--nomodel --extsize 147 -g dm --broad --min-length 500 --max-gap 200” for broad peaks. We used merged input samples for each genotype as the controls for all peak calling, due to a lack of sample-matching information for the published datasets that were re-analyzed.

RNA-seq analysis

RNA-seq reads were quantified using Salmon (1.1.0, ¹²⁶) and the Flybase r6.30 transcripts. Salmon was used in mapping-based mode, with the following parameters: “-l A --validateMappings –seqBias”. For visualization purposes, RNA-seq reads were also aligned to the dm6 genome, using Hisat2 (version 2.1.0, ¹²⁷). Mapped reads were filtered to remove alignments with quality scores less than 30, as well as secondary and supplementary alignments. PCR duplicates were marked using sambamba (version 0.6.8, ¹²⁴). Coverage tracks were generated using the bamCoverage tool from deepTools (version 3.2.0, ¹¹⁶) with the following parameters: “-of bigwig --binSize 10 --normalizeUsing CPM --extendReads 200 --ignoreDuplicates --minMappingQuality 30” and keeping only reads from chromosomes X, 2L, 2R, 3L, 4, and Y.

We used tximport (version 1.14.2, ¹²⁸) to import quantifications from Salmon into R (3.6.3) and estimate transcripts per million (TPM) values. We carried out pairwise differential expression analysis between gd⁷, Toll^rm9/rm10, and Toll^10B using DESeq2 (version 1.26.0, ¹²⁹) with default parameters.

Identification of candidate tissue-specific enhancers

In order to identify tissue-specific enhancers, we first carried out pairwise differential H3K27ac signal analysis using csaw (version 1.20.0, ^114,115) and edgeR (version 3.28.1, ¹³⁰). We used 2,000-bp windows for the background calculations and selected 150-bp windows with a 2.5-fold enrichment over the background. Windows were merged using the parameters “tol = 100” and “max.width = 5000”. Merged regions with a false discovery rate (FDR) < 0.05 and with a consistent direction of change across all windows were selected for downstream analysis. Candidate tissue-specific enhancers were defined by taking the intersection of regions identified as enriched for H3K27ac in each genotype compared to both others.

We validated the putative enhancers by comparing to enhancers identified in previous studies. 6 of 22 dorsal ectoderm enhancers identified from a literature search ⁶² overlap with our gd⁷-specific enhancers, while 10 of 37 mesoderm enhancers overlap our Toll^10B-specific enhancers (Fig. S1C). The relatively low overlap can be explained by the fact that many literature enhancers have H3K27ac signal in Toll^rm9/rm10 as well as either gd⁷ or Toll^10B. Putative enhancers were also overlapped with regions tested for enhancer activity in Drosophila embryos by ¹³¹. Regions (“tiles”) tested by Kvon et al. that were active in at least one tissue and timepoint were lifted over to dm6 from dm3. 165 putative enhancers overlapped a total of 183 tiles by at least 1 bp. Out of 27 gd⁷ enhancers, which overlap tiles that are active in stage 4-6 or stage 7-8, 19 are active in either dorsal ectoderm or amnioserosa precursors/subsets. Out of 35 Toll^rm9/rm10 enhancers, which overlap tiles that are active in stage 4-6 or stage 7-8, 26 are active in brain or ventral nerve cord precursors, procephalic ectoderm, or ventral ectoderm. Out of the 35 Toll^10B enhancers, which overlap tiles that are active in stage 4-6 or stage 7-8, 27 are active in mesoderm precursors/subsets.

Enhancer heatmaps were made using the plotHeatmap tool from deepTools (version 3.2.0, ¹¹⁶). Overlaps between different enhancer sets were visualized using UpSetR (version 1.4.0, ^132,133).

Assignment of candidate enhancers to target genes

We defined “housekeeping genes” as genes that have at least ‘low’ expression in all stages and tissues according to Flybase RNA-seq data (1,867 genes). We filtered the set of genes from the Flybase 6.30 transcripts to remove these housekeeping genes, as well as any genes with an average TPM < 1 in the gd⁷, Toll^rm9/rm10, and Toll^10B bulk RNA-seq. Candidate tissue-specific enhancers were assigned to target genes using the following rules: first, we assigned any enhancers that overlapped a single transcript to that gene. Next, we assigned enhancers to the closest promoter that was not separated from the enhancer by a domain boundary (using consensus boundaries from 3-4 hpf embryos, see below). The remaining enhancers were assigned to the closest promoter within the same domain, or, if they were not inside a domain, to the closest promoter.

scRNA-seq analysis

We used CellRanger (version 3.1.0) to produce fastq files for the scRNA-seq data and to align, filter, and quantify reads based on the BDGP6.22 genome release (Ensembl 98) to produce feature-barcode matrices. We imported the filtered matrices into R using DropletUtils (version 1.6.1, ¹³⁴), and performed additional quality control analysis using scater (version 1.14.6, ¹³⁵). Doublets were identified using scDblFinder (version 1.1.8, ¹³⁶), with an estimated doublet rate of 3.9%, and removed. Normalization for library size across cells was performed using scater and scran (version 1.14.6, ¹³⁷) using the “deconvolution” approach described in ¹³⁸, in which cells are pre-clustered and size factors estimated using the calculateSumFactors() function.

Downstream analysis was carried out using Seurat (version 3.1.4, ^139,140). The VST method was used to select the top 3,000 variable features for each sample, then all datasets were integrated using the control dataset with the highest number of cells (replicate 1) as the reference dataset, and the first 30 dimensions. We performed clustering using the Shared Nearest Neighbor approach implemented in the Seurat functions FindNeighbors and FindClusters, using the first 12 dimensions from PCA, a k.param value of 60, and a clustering resolution of 0.5. These parameters were chosen because they produced clusters that were stable to small variations in the parameter values.

We carried out differential expression analysis using the Seurat function FindMarkers in order to identify genes with higher expression in a cluster compared to all other cells, and in pairwise comparisons. We carried out Gene Ontology enrichment analysis on the results marker gene sets using the enrichGO function from clusterProfiler (version 3.14.3, ¹⁴¹), and simplified the results to remove semantically similar terms using the simplify function from clusterProfiler with the Wang method and a similarity threshold of 0.7. These marker gene sets and enriched GO terms, along with expression of known markers for embryonic cell populations, were used to assign putative cluster identities.

In order to quantify the average expression of particular gene sets in Figure 2C and Extended Data Figure 3C, we calculated the sum of expression of those genes per cell, and then expressed this as a Z-score across all cells.

Pooled scRNA-seq reads from all barcodes were analyzed using Salmon as described above.

Hi-C analysis

Hi-C data were analyzed using FAN-C (version 0.8.28) ¹⁴². Paired-end reads were scanned to identify ligation junctions, split at ligation junctions if any were present, and then aligned independently to the dm6 genome using BWA-MEM (version 0.7.17-r1188) ¹⁴³. Aligned reads were filtered to retain only uniquely-aligned reads with a mapping quality of at least 3. Reads were then paired based on read names and assigned to restriction fragments. “Inward” and “outward” reads separated by less than 1 kb, representing likely unligated fragments and self-ligated fragments respectively, were removed. In addition, we removed PCR duplicates, reads mapping more than 500 bp from a restriction site, and self-ligations where both reads map to the same fragment.

We generated two biological replicate datasets for each genotype, which showed high similarity. Therefore, we pooled biological replicates to reach 2-kb resolution. Matrices created from merged biological replicates were binned and filtered using FAN-C default parameters to remove bins with coverage less than 10% of the median coverage. Normalization was performed using Knight-Ruiz matrix balancing ¹⁴⁴. Expected contacts are calculated as the average contacts at each genomic distance separation. Hi-C data were visualized using plotting tools from FAN-C and using HiGlass ¹⁴⁵.

Micro-C analysis

Micro-C analysis was performed using FAN-C in the same way as the Hi-C analysis, except that reads were assigned to 100-bp genomic bins rather than restriction enzyme fragments, and “inward” and “outward” reads assigned to the same or adjacent bins (separated by less than 50 bp) were removed. Normalization was performed using iterative correction ¹⁴⁶, as Knight-Ruiz matrix balancing had prohibitive memory requirements at high resolution.

Domain and boundary identification

The insulation score was calculated as described in ⁷⁷, using FAN-C, for 2-kb and 5-kb resolution matrices, each with window sizes of 4, 6, 8, and 10 bins. Domain boundaries were calculated from the insulation score with a delta parameter of 3 and filtered to keep only boundaries with a boundary score of at least 0.7. Consensus boundaries for each sample were created by overlapping boundaries called at the two different resolutions and four different window sizes, and keeping those boundaries which were identified using at least four out of the total eight parameter combinations. Domains were created by pairing boundaries, and domains less than 10 kb or more than 500 kb in size were removed.

Hi-C aggregate analysis

Aggregate compartment, domain, and loop plots were created using FAN-C. Compartment analysis was carried out using Hi-C matrices that had been masked to remove pericentromeric heterochromatin. Pericentromeric heterochromatin was identified using H3K9me3 ChIP-seq data from 0-4 hpf and 4-8 hpf embryos from modENCODE ¹⁴⁷. H3K9me3 ChIP-seq data were processed as described above, binned at 10-kb resolution, and bins with enrichment for H3K9me3 compared to input in both 0-4 hpf and 4-8 hpf embryos were selected. Bins closer than 25 kb were merged, regions smaller than 20 kb were removed, and remaining large regions within 100 kb were merged. This produced a small number of regions of which one per chromosome clearly corresponded to the pericentromeric heterochromatin. Compartments were identified using the first eigenvector of the correlation matrix of the normalized Hi-C data, using GC content to orient the eigenvector. The compartment eigenvector for the 3-4 hpf Hi-C data from Hug et al. 2017 was used as the reference for the aggregate compartment plots (“saddle plots”) ^{74,98,142,146}. Compartment aggregates are plotted with all regions with an eigenvector value of zero collapsed so that they represent a single row/column in the aggregate matrix. Domain aggregates were also created using the domains identified in the Hug et al. 3-4 hpf Hi-C data. Loop aggregates were created using the loops identified in Kc167 cells by ⁷². Similar results were obtained using loops from ^88,91 (not shown).

We constructed a BEDPE file of putative enhancer-promoter interactions, considering all unique transcript start sites for assigned target genes. Interactions with a separation of at least 10 kb were used to create enhancer-promoter aggregate plots.

Hi-C similarity score analysis with CHESS

We used CHESS (Comparison of Hi-C Experiments using Structural Similarity; version 0.2.0 ⁷⁵) to compare Hi-C data from embryos of different genotypes. Briefly, CHESS treats Hi-C interaction matrices as images and applies the concept of the structural similarity index (SSIM), which is widely used in image analysis. We applied CHESS to 5-kb resolution Hi-C matrices using windows of 500 kb and a step size of 5 kb to produce similarity scores for pairwise Hi-C comparisons. Hi-C data from stage 5 control embryos were used as the reference dataset, and compared to data from stage 5 dorsoventral mutant embryos and nc14 control embryos from Hug et al. 2017. In order to identify regions of the genome with significant changes between the reference and query datasets, we selected regions with a SSIM Z-score less than -2, and a signal-to-noise ratio Z-score of at least 1. For the boxplots in Figure 4C, windows were classified as containing DE genes if they contained at least one gene that is significantly upregulated or downregulated in the query genotype compared to both other genotypes, and as having no DE genes otherwise. Only every hundredth window was considered, in order to use only non-overlapping windows. The final numbers of windows considered are: gd⁷ – 101 DE, 124 non-DE; Toll^rm9/rm10 – 74 DE, 152 non-DE; Toll^10B – 183 DE, 41 non-DE. CHESS is available at https://github.com/vaquerizaslab/chess.

Statistics and visualization

Statistical tests were carried out in R (version 3.6.3) and visualization was performed using the ggplot2 package ¹⁴⁸. Box plots are defined with boxes spanning the first to third quartiles. The whiskers extend from the box to the smallest/largest values no further than 1.5× the interquartile range (IQR) away from the box. The notches extend 1.58 × IQR / sqrt(n) from the median. All statistical tests are two-sided.

Extended Data

Extended Data Fig. 1. Properties of domains containing tissue-specific putative enhancers and validation of tissue-specific enhancer activity.

a, b. Putative enhancers active in gd⁷, Toll^rm9/rm10, and Toll^10B embryos overlap with enhancers from ¹³⁰ and are enriched for enhancers that drive expression in relevant regions of the embryo at stages 4-6 (a) and 7-8 (b). The two largest categories are labelled for each stacked bar. c. Size of domains containing putative tissue-specific enhancers (n = 394 domains, black) and those without enhancers (n = 1097 domains, grey). Two-sided Wilcoxon rank sum test, p < 2x10^-16. d. Genes per kilobase inside domains containing putative tissue-specific enhancers (n = 394 domains) and those without enhancers (n = 1097 domains). Two-sided Wilcoxon rank sum test, p = 2.1 x 10^-12. Box plots show median, box spanning first to third quartiles, whiskers extending to smallest/largest values no further than 1.5 * IQR from the box, and notches extending 1.58 * IQR / sqrt(n) from the median. Outliers are excluded for clarity. e. Overlap of chromatin domains (n = 1491 domains) with Genomic Regulatory Blocks (GRBs) from ¹⁴⁹. Chi-squared test p = 1.8 x 10^-7. f. Schematic illustration of assignment of enhancers to target genes. Broadly expressed and non-expressed genes are omitted for clarity (see Methods). Enhancers are assigned to genes as follows: 1) enhancers that overlap a transcript are assigned to that gene; 2) remaining unassigned enhancers are assigned to the closest promoter, unless that promoter is in a different domain; 3) remaining enhancers are assigned to the closest promoter in the same domain; 4) remaining enhancers are assigned to the closest promoter, even if it is in a different domain.

Extended Data Fig. 2. Properties of domains containing tissue-specific putative enhancers are robust to different enhancer definitions.

a. UpSet plot ^131,132 showing overlap between the putative tissue specific enhancers identified in this study, in ref. ⁶², and mesoderm and dorsal ectoderm enhancers identified from a literature search by ⁶². Lower panel y axis represents enhancer sets; intersections between these sets are shown by joined dots. Bars in upper panel represent intersection sizes. For example, 6 putative dorsal ectoderm enhancers are identified in this study, identified in ⁶², and also identified from a literature search (rightmost column). b. Expression of genes associated with putative enhancers identified by ⁶² in gd⁷ and Toll^10B mutant embryos. Top, genes associated with gd⁷-specific putative enhancers (n = 380 enhancer-gene pairs); bottom, genes associated with Toll^10B-specific putative enhancers (n = 416 enhancer-gene pairs). Box plots show median, box spanning first to third quartiles, whiskers extending to smallest/largest values no further than 1.5 * IQR from the box, and notches extending 1.58 * IQR / sqrt(n) from the median. Outliers are excluded for clarity. c. Size of domains containing putative tissue-specific enhancers identified by ⁶² (n = 313 domains, black) and those without enhancers (n = 1178 domains, grey). Two-sided Wilcoxon rank sum test, p = 1.5 x 10^-12.d. Genes per kilobase inside domains containing putative tissue-specific enhancers (n = 313 domains) identified by ⁶² and those without enhancers (n = 1178 domains). Two-sided Wilcoxon rank sum test p = 5.5 x 10^-14. Box plots show median, box spanning first to third quartiles, whiskers extending to smallest/largest values no further than 1.5 * IQR from the box, and notches extending 1.58 * IQR / sqrt(n) from the median. e. Overlap of chromatin domains (n = 1491 domains) with Genomic Regulatory Blocks (GRBs) from ¹⁴⁹. Chi-squared test p = 7.7 x 10^-15.

Extended Data Fig. 3. Validation of scRNA-seq data.

a. PCA of pooled single-cell RNA-seq data and RNA-seq data. The first principal component separates techniques while the second principal component separates genotypes. Replicate control single-cell RNA-seq experiments cluster together, demonstrating robustness. b. Clustering of single-cell gene expression profiles from 2.5-3.5 hpf embryos reveals clusters corresponding to distinct cell populations, as in Fig. 2A but separated by genotype of origin. c. Expression of tissue-specific marker genes across single cells from different genotype origins. Marker genes for dorsal ectoderm (Ance, CG2162, Doc1, Doc2, egr, peb, tok, ush, zen), neuroectoderm (ac, brk, CG8312, l(1)sc, mfas, Ptp4E, sog, SoxN, vnd), mesoderm (CG9005, Cyp310a1, GEFmeso, ltl, Mdr49, Mes2, NetA, ry, sna, stumps, twi, wgn, zfh1), and pole cells (pgc) were obtained from ⁷¹.

Extended Data Fig. 4. Balancer chromosomes contribute to differences between Hi-C matrices in embryos from different genotypes.

a. Whole genome contact probability maps for control, gd⁷, Toll^rm9rm/10, and Toll^10B embryos at the cellular blastoderm stage. Arrows mark artefacts in the Hi-C data that indicate rearrangements on balancer chromosomes present in a subset of Toll^rm9rm/10 (TM6) and Toll^10B (TM3 and OR60) embryos. b. CHESS ⁷⁵ similarity scores were calculated between mutant and control embryo Hi-C datasets (5kb resolution, 500 kb window size). As a reference, similarity scores were calculated between control embryo Hi-C data from this study and Hug et al. 2017. The difference between the reference similarity score and the control/mutant similarity score is shown for all chromosomes (blue, gd⁷; pink, Toll^rm9/rm10; yellow, Toll^10B). Values around zero represent regions where chromatin conformation is similar between control and mutant, while negative values represent regions where there are greater differences between control and mutant than between the reference control datasets. Shaded area, +/- two standard deviations from genome-wide mean. Grey ticks, positions of genes that are differentially expressed between dorsoventral mutant embryos ⁶². Asterisks mark positions of known rearrangement breakpoints on the TM3 and TM6 balancer chromosomes ^35,148.

Extended Data Fig. 5. A small number of regions have changes in chromatin conformation detected by CHESS that are not associated with known genomic rearrangements.

a, c. CHESS ⁷⁵ similarity scores were calculated between mutant and control embryo Hi-C datasets (using 5kb resolution, 500 kb window size). As a reference, similarity scores were calculated between control embryo Hi-C data from this study and Hug et al. 2017. The difference between this reference similarity score and the control/mutant similarity score is shown (blue, gd⁷; pink, Toll^rm9/rm10; yellow, Toll^10B). Values around zero represent regions where chromatin conformation is similar between control and mutant, while negative values represent regions where there are greater differences between control and mutant than between the reference control datasets. Shaded area, +/- two standard deviations from genome-wide mean. Grey ticks, positions of genes that are differentially expressed between dorsoventral mutant embryos ⁶². b, d. For each genotype, top, normalized Hi-C contact probability and contact probability difference (5kb resolution); red, regions with increased contact probability in embryos of the mutant genotype; blue, decreased contact probability. Arrows, regions with a change in contact probability in Toll^10B. Bottom, RNA-seq (CPM) ⁶², H3K27ac and H3K27me3 ChIP-seq data (CPM; ⁶², this study). Tissue-specific putative enhancers, color-coded bars beneath corresponding H3K27ac ChIP-seq track. Lower panel, gene annotations. Positive-strand genes, orange; negative strand genes, blue. a. Example of differences in CHESS similarity scores for a 2 Mb region on chromosome 2L. b. Hi-C data for a 500 kb subset of the region shown in a. c. Example of differences in CHESS similarity scores for a 2 Mb region on chromosome 2R. d. Hi-C data for a 500 kb subset of the region shown in c.

Extended Data Fig. 6. Chromatin conformation is not affected by tissue-specific gene expression.

Tissue-specific chromatin data for regions containing dorsoventral patterning genes. For each genotype, top, normalized Hi-C contact probability (2 kb resolution); middle, RNA-seq data (CPM) ⁶²; bottom, H3K27ac and H3K27me3 ChIP-seq data (CPM; ⁶², this study). Tissue-specific putative enhancers, color-coded bars beneath corresponding H3K27ac ChIP-seq track. Lower panels, “virtual 4C” tracks for each genotype representing interactions of a 2 kb region around the promoters of genes of interest, as highlighted by the grey rectangle; gene annotations. Positive-strand genes, orange; negative strand genes, blue. Dorsoventral patterning genes of interest, black. a. NetA and NetB. b. pnr. c. if. d. sna.

Extended Data Fig. 7. Insulation does not change across genotypes at tissue-specific enhancers or differentially expressed genes.

a. Average insulation scores around consensus boundaries identified in 3-4 hpf Hi-C ⁵⁰. Insulation scores from control, gd⁷, Toll^rm9/rm10, and Toll^10B embryos at the cellular blastoderm stage were calculated at 2 kb resolution with a 16 kb window size. Shaded areas represent +/- one standard deviation from the mean. b. Average insulation scores around gd⁷, Toll^rm9/rm10, and Toll^10B -specific enhancers. c. Average insulation scores around the transcription start sites (TSSs) of genes that are up- or down-regulated in the given genotype compared to both other genotypes.

Extended Data Fig. 8. Global chromatin conformation along the dorsoventral axis by Micro-C at the cellular blastoderm stage and Hi-C at stage 10.

Chromatin conformation for a 1.8 Mb region of chromosome 2L by Micro-C in control and gd⁷ embryos at the cellular blastoderm stage and gd⁷, Toll^rm9/rm10, Toll^10B, and by Hi-C in control embryos at stage 10. b. ‘Saddle-plot’ representing genome-wide average chromatin compartmentalisation. Active regions tend to interact with other active regions (top left), while inactive regions interact with other inactive regions (bottom right). c. Aggregate analysis of domains identified using Hi-C data from 3-4 hpf embryos at 2kb resolution. d. Aggregate analysis of chromatin loops identified in ⁷². e. Chromatin conformation for a 300 kb region of chromosome 2L, showing domains and a loop. f. Average Micro-C contact probability at different distances for control (black) and gd⁷ (blue) embryos at the cellular blastoderm stage. g. The derivative of the expected Micro-C contact probability by distance. h. Average Hi-C contact probability at different distances for control (black), gd⁷ (blue), Toll^rm9/rm10 (pink), and Toll^10B (yellow) embryos at stage 10. i. The derivative of the expected Hi-C contact probability by distance for stage 10 embryos, highlighting differences between samples at far-cis distances due to the presence of balancer chromosomes in the Toll^rm9/rm10 and Toll^10B embryos.

Extended Data Fig. 9. Chromatin conformation remains similar across tissues at a later developmental stage.

Tissue-specific Hi-C data from stage 10 embryos (approximately 4-5 hpf) for regions containing dorsoventral patterning genes. For each genotype, top, normalized Hi-C contact probability, 2 kb resolution. Middle, “virtual 4C” tracks for each genotype representing interactions of a 2 kb region around the promoters of genes of interest (grey rectangle). Bottom, RNA-seq (CPM) for embryos at 0-4 hpf and 4-8 hpf ¹⁴⁶; gene annotations. Positive-strand genes, orange; negative strand genes, blue. Dorsoventral patterning genes of interest, black. a. Doc1, Doc2, and Doc3. b. dpp. c. sog. d. if. e. NetA and NetB. f. sna.

Extended Data Fig. 10. Micro-C confirms that high-resolution chromatin conformation is largely unaffected by tissue-specific gene expression and enhancer activity.

Micro-C data for regions containing dorsoventral patterning genes. For control and gd⁷ embryos, top, normalized Micro-C contact probability, 500 bp resolution; middle, RNA-seq data (CPM) ⁶²; bottom, H3K27ac and H3K27me3 ChIP-seq data (CPM; ⁶², this study). Tissue-specific putative enhancers, color-coded bars beneath corresponding H3K27ac ChIP-seq track. Lower panels, “virtual 4C” tracks representing interactions of a 2 kb region around the promoters of genes of interest (grey rectangle); gene annotations; positive-strand genes, orange; negative strand genes, blue. Dorsoventral patterning genes of interest, black. Arrow highlights a region in the pnr locus that gains insulation in gd⁷ embryos. a. C15 . b. pnr. c. NetA and NetB. d. sna.

Code availability

All computational analysis is described in the Methods, and code is also available at github.com/vaquerizaslab/Ing-Simmons_et_al_dorsoventral_3D_genome (archived on Zenodo, DOI: 10.5281/zenodo.4272002).

Data availability

The Hi-C, Micro-C, scRNA-seq, and ChIP-seq data produced in this study have been submitted to ArrayExpress and are available with the following accession numbers: E-MTAB-9306, E-MTAB-9784, E-MTAB-9304, and E-MTAB-9303, respectively. In addition, we analyzed data from the following publicly available datasets: GEO accessions GSE68983, GSE18068, and GSE16013 and ArrayExpress accession E-MTAB-4918. Datasets are listed in full in Supplementary Table 4. Genome sequences and gene annotations were obtained from Flybase r6.30 (flybase.org) and Ensembl version 98 (www.ensembl.org).