INT-Hi-C reveals distinct chromatin architecture in endosperm and leaf tissues of Arabidopsis

Abstract

Higher-order chromatin structure undergoes striking changes in response to various developmental and environmental signals, causing distinct cell types to adopt specific chromatin organization. High throughput chromatin conformation capture (Hi-C) allows studying higher-order chromatin structure; however, this technique requires substantial amounts of starting material, which has limited the establishment of cell type-specific higher-order chromatin structure in plants. To overcome this limitation, we established a protocol that is applicable to a limited amount of nuclei by combining the INTACT (isolation of nuclei tagged in specific cell types) method and Hi-C (INT-Hi-C). Using this INT-Hi-C protocol, we generated Hi-C data from INTACT purified endosperm and leaf nuclei. Our INT-Hi-C data from leaf accurately reiterated chromatin interaction patterns derived from conventional leaf Hi-C data. We found that the higher-order chromatin organization of mixed leaf tissues and endosperm differs and that DNA methylation and repressive histone marks positively correlate with the chromatin compaction level. We furthermore found that self-looped interacting genes have increased expression in leaves and endosperm and that interacting intergenic regions negatively impact on gene expression in the endosperm. Last, we identified several imprinted genes involved in long-range and trans interactions exclusively in endosperm. Our study provides evidence that the endosperm adopts a distinct higher-order chromatin structure that differs from other cell types in plants and that chromatin interactions influence transcriptional activity.

Figure 1.

Genome-wide high-resolution INT-Hi-C approach. (A) INT-Hi-C workflow to generate tissue-specific libraries in plants. (B) Scatter plots showing the comparison of chromatin interaction frequencies among the biological replicates and (C) between INT-Hi-C and publically available conventional Hi-C data (13,15) in leaf tissue. Pearson correlation coefficients are also shown. (D) 2D interaction map of Arabidopsis leaf and endosperm at 100 kb resolution. The intense yellow diagonal reflects the enrichment of interacting reads in close proximity. Each pixel represents interactions between a 100 kb locus and another 100 kb locus. Chromosomes are represented through black bars from left to right and top to bottom. The telomeres of each chromosome is represented by black circle. Color scale bar ranging from black to yellow represents lower to higher enrichment of interacting reads, respectively.

Figure 2.

Distinct chromatin architecture in endosperm and leaf. (A) Distribution of structural domains (SDs) on chromosome arms in leaf and endosperm. Red color denotes compact structural domains (CSDs), while blue represents loose structural domains (LSDs). Dotted black lines represent the pericentromeric regions of each chromosomes were defined previously [36]. (B) Comparison of contact map of endosperm and leaf using HiCPlotter. Upper and lower panels show normalized contact maps of chromosome 2 for leaf and endosperm, respectively. Color scale bar ranging from white to red represents the lower to higher enrichment of interacting reads. (C) Differential interactions matrix of endosperm versus leaf for chromosome 2. Red and blue colors show enrichment or depletion of interactions reads in the endosperm, respectively, while white color indicates no change between endosperm and leaf. (D) 3D structure of Arabidopsis endosperm (left panel) and leaf (right panel) genome derived from normalized genome-wide interaction matrix (color code; purple, yellow, blue, green, red and black represent chr1, chr2, chr3, chr4, chr5 and centromere, respectively).

Figure 3.

Distribution of epigenetic marks in CSDs and LSDs. Box plots showing the level of various (A) cytosine methylation (CG, CHG and CHH) and (B) median values of z-score-normalized histone modifications (H3K9me2, H3K27me3 and H3K27me1) in CSDs and LSDs of leaf and endosperm tissue. Red color denotes compact structural domains (CSDs), while blue represents loose structural domains (LSDs) in leaf and endosperm respectively. Significance was determined using a Mann–Whitney U test (*P< 0.005, ** P< 0.0001). (C) Box plots showing the distribution of genes (left panel) and TEs (right panel) in CSDs and LSDs of leaf and endosperm. Significance was determined using a Mann–Whitney U test (NS = Not significant, * P< 0.005, ** P< 0.0005).

Figure 4.

Relationship of SDs with epigenetic marks. Box plots showing the level of various (A) cytosine methylation (CG, CHG, and CHH) and (B) median values of z-score-normalized histone modifications (H3K9me2, H3K27me3, and H3K27me1) in constitutive (C) and interchangeable (I) CSDs and LSDs in leaf and endosperm. Domains which maintain its compact or loose status in both endosperm and leaf referred as constitutive SDs (C-CSD and C-LSD, respectively), while domains that switch from compact to loose and vice versa are referred to as interchangeable SDs (I-CSD and I-LSD). Red color denotes level of various cytosine and histone methylation in C-CSDs; pink represents level of various cytosine and histone methylation in I-CSDs; blue represents level of various cytosine and histone methylation in C-LSDs, while light blue represents level of various cytosine and histone methylation in I-LSD in leaf and endosperm. Significance was determined using a Mann–Whitney U test (NS = not significant, * P< 0.005, ** P< 0.005).

Figure 5.

Chromatin interactions and gene expression. (A) Venn diagram showing common and unique interactions identified in pooled leaf and endosperm libraries at 10 kb resolution. For common interactions significance was determined using a hypergeometric test. (B) Comparison of RNA-seq-based expression counts (RPKM- reads per kilobase per million) of genes with self-looped structures compared to control genes without self-looped structures in leaf and endosperm. Significance was determined using a Mann–Whitney U test. (C) Comparison of RNA-seq-based expression counts of genes interacting with intergenic regions in endosperm and leaf with non-interacting genes in the respective tissue Significance was determined using a Mann–Whitney U test.

Figure 6.

Genomic imprinting and chromatin interactions. (A) Circos plot showing cis (>10kb distance) and trans interactions of imprinted genes along the chromosomes in endosperm (upper panel) and leaf (lower panel). Small and large highlighted regions on each chromosome represent the centromeres and pericentromeres, respectively. (B) Comparison of RNA-seq-based expression counts of interacting PEGs, MEGs, interacting non-imprinted and non-interacting genes in endosperm and leaf tissue. Significance was determined using a Mann-Whitney U test.