The ancestral chromatin landscape of land plants

Abstract

Recent studies have shown that correlations between chromatin modifications and transcription vary among eukaryotes. This is the case for marked differences between the chromatin of the moss Physcomitrium patens and the liverwort Marchantia polymorpha. Mosses and liverworts diverged from hornworts, altogether forming the lineage of bryophytes that shared a common ancestor with land plants. We aimed to describe chromatin in hornworts to establish synapomorphies across bryophytes and approach a definition of the ancestral chromatin organization of land plants. We used genomic methods to define the 3D organization of chromatin and map the chromatin landscape of the model hornwort Anthoceros agrestis. We report that nearly half of the hornwort transposons were associated with facultative heterochromatin and euchromatin and formed the center of topologically associated domains delimited by protein coding genes. Transposons were scattered across autosomes, which contrasted with the dense compartments of constitutive heterochromatin surrounding the centromeres in flowering plants. Most of the features observed in hornworts are also present in liverworts or in mosses but are distinct from flowering plants. Hence, the ancestral genome of bryophytes was likely a patchwork of units of euchromatin interspersed within facultative and constitutive heterochromatin. We propose this genome organization was ancestral to land plants.

Keywords: Anthoceros agrestis; Marchantia polymorpha; bryophytes; chromatin; epigenetic; evolution.

Fig. 1

Association of chromatin marks with genomic features. (a) Distribution of histone posttranslational modifications (PTMs) over genomic features. The total length of PTMs overlapping specified genomic features was divided by the total length of PTM peaks to determine each proportion. Features that cover < 0.5% of PTM peaks are not shown. H3K9me1, H3K27me1 are expected as PTMs of constitutive heterochromatin and H3K27me3 is a PTM of facultative heterochromatin with a distribution contrasting with the active PTMs H3K4me3 and H3K36me3. (b) Profile plot of CG, CHG, and CHH methylation levels over PTM peaks. Sequences 1 kb upstream and downstream of the peak center are included. Average methylation over 10 bp bins is plotted. (c) Genomic features of the Anthoceros agrestis genome. The Circos plot illustrates the genomic features of the A. agrestis genome using rings to display different information with a window size of 100 kb. The rings represent the following features: (a) gene density, (b) transposable element (TE) density, (c) ribosomal DNA (rDNA) density, (d) DNA methylation density, (e) H3K9me1 peak density, (f) H3K27me1 peak density, (g) H3K27me3 peak density, (h) H3K27me3 peak density, (i) H3K36me3 peak density. (d, e) Integrative Genomics Viewer (IGV) browser screenshot demonstrating vicinity of TEs covered by H3K9me1 and expressed genes (d) and TEs covered by H3K27me3 (e). The regions shown are 30 kb in length from the scaffold AnagrOXF.S1. Posttranslational modifications tracks are bigwig files scaled by H3 coverage in 10‐bp windows. DNA methylation tracks are bigwig files showing methylation levels of each cytosine site covered by at least 10 reads. ‘TEs’ and ‘Genes’ tracks are annotation files for TEs and genes, respectively. ‘RNA‐seq’ tracks are bigwigs of mapped RNA‐seq reads from gametophyte tissue and sporophyte tissue (Li et al., 2020). Scales are noted in square brackets in each track.

Fig. 2

Higher order structure of Anthoceros agrestis chromosomes. (a) Hi‐C contact heatmap of chromosomes. The blocks were utilized to symbolize the signal linked with the two contact positions. The color depth corresponds to the intensity of the interaction between DNA molecules. The darker shades indicate a stronger interaction between them. (b) Density plots showing distributions of epigenetic marks per chromosome. The density of DNA methylation or histone posttranslational modifications (PTMs) were calculated as the number of DNA methylation sites or numbers of peaks of histone PTMs in each 40 kb window, divided by window size (40 kb), and plotted for each chromosome. The median value for each chromosome is represented by a solid vertical line. (c) Violin plot showing expression level of protein coding genes (PCGs) per chromosome. Expression levels are indicated by Transcript per Million (TPM) values transformed by using asinh function. Width is relative to PCG density. Red dots indicate median expression values. (d) Violin plots showing density distribution of epigenetic marks in topologically associating domains (TADs) and TAD boundaries. The density of epigenetic marks was calculated as the number of histone PTM peaks or DNA methylation sites in each 40 kb window, divided by window length, and plotted for TADs and TAD boundaries. The median value is represented by a solid horizontal line. P‐values from the Wilcoxon test are indicated on each plot. (e) Violin plots showing density distribution of epigenetic marks in different compartments. The density of epigenetic marks calculated as above is plotted against A or B compartment. The median value is represented by a solid horizontal line.

Fig. 3

Association of chromatin marks on protein coding genes. (a) Violin plot showing expression level of protein coding genes (PCGs) associated with histone posttranslational modifications (PTMs). Expression levels are indicated by Transcript per Million (TPM) values transformed by using asinh function. Width is relative to PCG density. Red dots indicate median expression values. (b) Heatmap of k‐means clustering of genes based on PTMs. Prevalence of each mark (columns) based on its score normalized against H3 signals per 10 bp bins. Sequences 2 kb upstream and downstream of the start codon are included. Red stands for enrichment and blue for depletion. Each row corresponds to one gene, with multiple genes grouped into blocks that have been defined as clusters P1 through P5. (c) Pie chart showing proportions of PCG clusters in all PCGs. (d) Violin plot showing expression level of PCGs per PCG cluster. Expression levels are indicated by TPM values transformed by using asinh function. Width is relative to the density of PCGs. Red dots indicate median expression values. (e) Violin plot showing length of PCG per PCG cluster. The width is relative to the density of PCGs. Red dots indicate median values. Clusters not sharing the same letter are significantly different (Tukey–kramer test, P < 0.05). (f) Profile plot of CG, CHG, and CHH methylation levels over PCGs per PCG cluster. Gene body of each PCG is scaled to 2 kb and sequences 1 kb upstream and downstream are included. Average methylation over 10 bp bins is plotted. (g) Stacked bar chart showing numbers of PCGs overlapped by transposable elements (TEs) at least 25% of their length per PCG cluster. Different colors indicate TE clusters defined in Fig. (4a).

Fig. 4

Association of chromatin marks on transposable elements (TE). (a) Heatmap of k‐means clustering of TEs based on histone posttranslational modifications (PTMs). Prevalence of each PTM (columns) based on its score normalized against H3 signals per 10 bp bins. Each TE annotation is scaled to 1 kb and sequences 1 kb upstream and downstream are included. Red color stands for enrichment and blue for depletion. Each row corresponds to one TE, with multiple TEs grouped into blocks that have been defined as clusters T1 through T8. (b) Pie chart showing proportions of TE clusters in all TEs. (c) Violin plot showing length of TEs per TE cluster. Width is relative to the density of TEs. Red dots indicate median values. Clusters not sharing the same letter are significantly different (Tukey–kramer test, P < 0.05). (d) Profile plot of CG, CHG, and CHH methylation levels over TEs per TE cluster. Each TE annotation is scaled to 1 kb and sequences 1 kb upstream and downstream are included. Average methylation over 10 bp bins is plotted. (e) Stacked bar chart indicating proportions of TE families in each TE cluster (T1–T8) in comparison with TE family proportion in the entire genome (All). (f) Stacked bar chart showing numbers of TEs overlapped by protein coding genes (PCGs) at least 25% of their length per TE cluster. Different colors indicate PCG clusters defined in Fig. (3b).

Fig. 5

Positional relationships between protein coding genes (PCGs) and transposable elements (TEs). (a) Stacked bar chart showing the proportion of closest PCGs or TEs upstream or downstream of PCGs and TEs in comparison with the proportion in the entire genome (All). *, P < 2.2 × 10⁻¹⁶ (Fisher's exact test). (b, c) Stacked bar chart showing the proportion of closest PCGs or TEs upstream or downstream of PCGs (P1–P4 in (c)) and TEs (T1–T8 in (d)) per cluster in comparison with the observed proportions in the entire PCGs or TEs (All). *, P = 0 (permutation test). (d–f) Histograms showing distance from PCGs in P3(d) or TEs in T4 and T5 (e, f, respectively) to the closest PCGs (left) or TEs (right) per cluster.

Fig. 6

Chromatin organization in the common ancestor of bryophytes. Ancestral chromatin organization in bryophytes is inferred by chromatin synapomorphies shared by three bryophyte species. In contrast to the large pericentromeric heterochromatin observed in flowering plants (colored in orange), the constitutive heterochromatin of bryophytes forms small units and is scattered over chromosomes in bryophytes (orange bars). These constitutive heterochromatin marked by H3K9me (red circles) and DNA methylation (brown circles) primarily consist of TEs (orange box). In addition, facultative heterochromatin marked by H3K27me3 (blue circles) contains not only PCGs (blue box) but also TEs.