Genetic-epigenetic interplay in the determination of plant 3D genome organization

Abstract

The 3D chromatin organization plays a major role in the control of gene expression. However, our comprehension of the governing principles behind nuclear organization remains incomplete. Particularly, the spatial segregation of loci with similar repressive transcriptional states in plants poses a significant yet poorly understood puzzle. In this study, employing a combination of genetics and advanced 3D genomics approaches, we demonstrated that a redistribution of facultative heterochromatin marks in regions usually occupied by constitutive heterochromatin marks disrupts the 3D genome compartmentalisation. This disturbance, in turn, triggers novel chromatin interactions between genic and transposable element (TE) regions. Interestingly, our results imply that epigenetic features, constrained by genetic factors, intricately mold the landscape of 3D genome organisation. This study sheds light on the profound genetic-epigenetic interplay that underlies the regulation of gene expression within the intricate framework of the 3D genome. Our findings highlight the complexity of the relationships between genetic determinants and epigenetic features in shaping the dynamic configuration of the 3D genome.

Graphical Abstract

Figure 1.

ddm1 displays a reconfiguration of its epigenome. (A) Screenshot of chromosome 1 from JBrowse2 genome browser representing H3K9me2 (blue), H3K27me1 (brown) and H3K27me3 (green) ChIP-seq data in wild type and ddm1 mutant. Gene density is displayed at the top with a higher density of TEs and genes respectively in blue and red. (B) Immunofluorescence detection of H3K9me2 (pink) and H3K27me3 (green) histone modifications and DAPI staining (grey) in isolated tomato nucleus from wild type and ddm1. (C–E) On the left panels, heat-maps representing ChIP-seq profiles for H3K27me3 (green), H3K9me2 (blue), and H3K27me1 (brown) within H3K27me3 hypomethylated regions in ddm1 compared to the wild type (C); within H3K27me3 hypermethylated regions and H3K9me2 hypomethylated regions (D); within H3K27me3 hypermethylated regions and no change in H3K9me2 regions (E). The color gradient from blue to red characterize the counts from low to high in the respective regions. On the right panels, TEs superfamilies repartition in percentage for the whole tomato genome (grey) and for regions of each class (pink). Pie charts represent the proportion of genes (blue) and TEs (orange) in each class. *** corresponds to significant differences (pvalue < 5% with a Student test) observed between two groups indicated by black bars. (F) Scatter plot of TEs in gene-rich regions (red), TEs in gene-poor regions (blue) with trend line according to the fold change of H3K27me3 on x-axis and H3K27me1 on y-axis. (G) Heatmaps of ChIP-seq signals (H3K27me3 (green), H3K9me2 (blue), H3K27me1 (brown)) and DNA methylation (CHH (black)) from differentially methylated regions (DMRs) in ddm1 compared to WT and selected in Corem et al. (2018).

Figure 2.

The reconfiguration of ddm1 mutant epigenome impacts its transcriptome. (A) Gene ontology enrichment of up-regulated genes. (B) Gene ontology enrichment of down-regulated genes. Counts correspond to the number of genes associated with the corresponding GO enrichment term. P.adj were calculated using Fisher's exact test. (C) Scatter plot of differentially expressed genes in chromosome 1. Up-regulated (red) and down-regulated genes (blue) with associated and adjusted P-value. ChIP-seq data (H3K27me3, H3K27me1 and H3K9me2), DNA methylation data (CG, CHG and CHH), gene and TE density are linked with the plot at the bottom of the subfigure, wild type (brown) and ddm1 (light blue). (D, E) Heatmaps of up-regulated (D) and down-regulated (E) genes in the wild type and ddm1. H3K27me3 (green), H3K27me1 (brown), H3K9me2 (blue); and DNA methylation data in all contexts (black). Each histone mark has a heatmap (right) showing the difference between ddm1 and the wild type. All DNA methylation heatmaps are represented as differences between ddm1 and WT. Each row corresponds to one gene and were ranked according to the H3K27me3 signal.

Figure 3.

Chromatin architecture is affected in the ddm1 mutant with a weakened compartmentalization. (A) Chromosome 10 Hi-C interaction map at increasing levels of resolution. Gene and TE density are pictured at the top of the first map and a quantification of the compartmentalization, represented by a binary segmentation of the eigenvector (obtained with cooltools software) is shown on the top of the last map. (B) Analysis of compartment dynamics according to the interactions between compartments for the wild type and ddm1 (obtained with Pentad software). (C) Boxplot of compartment strength calculated from inter- and intra-compartment interactions for the wild type and ddm1. (D) Aggregate TAD analysis in wild type and ddm1 (left), and differential analysis is presented (right). Dark blue denotes a loss of interaction in ddm1 (obtained with GENOVA software). (E) Insulation heatmap (left) with a main insulation profile represented as the insulation score versus the relative position in kb (right) in wild type (grey) and ddm1 (red). The right panel shows an average score and each row corresponds to a TAD-border in the left panel.

Figure 4.

Reconfiguration of the epigenome in ddm1 mutant induces new interactions across the genome. (A) Pearson correlations of distance-normalized Hi-C interaction frequency maps of chromosome 10. The ChIP-seq signal for H3K9me2 (blue), H3K27me1 (brown), H3K27me3 (green) and PC1 component from principal component analysis (black) were aligned to the map, for wild type and ddm1 (left). Maps with reorganized bins according to their H3K27me3 (middle) and PC1 value (right). (B-C) Aggregate plots quantifying the mean of aggregated WT and ddm1 Hi-C contacts matrices between regions marked with H3K27me3 in the wild type (B) and between new regions marked with H3K27me3 in ddm1 versus the regions previously mentioned in the wild type (C). (D) Screenshot of a region in the chromosome 10 from WashU genome browser for H3K27me3 ChIPseq data, depicted in light blue, and Hi-C differential interactions arcs (pink). Differential interaction arc intervals were chosen to illustrate the most significant Zscores corresponding to WT-specific interactions at the top, and ddm1-specific interactions at the bottom. Gene density (with white corresponding to a low density and red, a high density of genes) of the chromosome and the portion of the region can be seen at the top of the screenshot.

Figure 5.

Model summarizing how epigenetic modifications orchestrate the interplay between genic and TEs regions. There is a clear segregation between the different chromatin constituents in the wild type: constitutive heterochromatin domains (red), Polycomb domains (blue) and active domains (green). The distribution of H3K27me3 (blue), H3K9me2 (red) and H3K27me1 (brown) marks coincides with the establishment of separate compartments. In contrast, the compartmental delineation is considerably more diffuse in ddm1. The ectopic depositions of H3K27me3 and the general reshuffling of the other histone marks induce new interactions, particularly between genic and TEs regions (violet). Created with Biorender.