Heat stress-induced transposon activation correlates with 3D chromatin organization rearrangement in Arabidopsis

Abstract

In higher eukaryotes, heterochromatin is mainly composed of transposable elements (TEs) silenced by epigenetic mechanisms. But, the silencing of certain heterochromatin-associated TEs is disrupted by heat stress. By comparing genome-wide high-resolution chromatin packing patterns under normal or heat conditions obtained through Hi-C analysis, we show here that heat stress causes global rearrangement of the 3D genome in Arabidopsis thaliana. Contacts between pericentromeric regions and distal chromosome arms, as well as proximal intra-chromosomal interactions along the chromosomes, are enhanced. However, interactions within pericentromeres and those between distal intra-chromosomal regions are decreased. Many inter-chromosomal interactions, including those within the KNOT, are also reduced. Furthermore, heat activation of TEs exhibits a high correlation with the reduction of chromosomal interactions involving pericentromeres, the KNOT, the knob, and the upstream and downstream flanking regions of the activated TEs. Together, our results provide insights into the relationship between TE activation and 3D genome reorganization.

Fig. 1. Heat induces TE activation.

a Percentages of up- and downregulated protein coding genes (PCGs) and transposable elements (TEs) under heat stress. Red bars: upregulation; Blue bars: downregulation. Numbers on the bars indicate the absolute number of TEs and PCGs. b Scatter plot of TE CPMs in heat vs. control conditions. CPM counts per million. Red dots: upregulated TEs under heat stress. Blue dots: downregulated TEs under heat stress. c The percentages of DNA transposons and retrotransposons for all (total) and for heat-activated TEs. Numbers marked on the bars indicate the absolute numbers of heat-activated TEs and total TEs from different TE classes. d The enrichment scores of TE superfamilies of heat-activated TEs. Each enrichment score was calculated as the ratio of the percentage of a certain TE superfamily in heat-activated TEs to that in all TEs. Numbers above the bars indicate the absolute numbers of heat-activated TEs in the corresponding superfamily. The blue dotted line marks the point of overrepresentation (enrichment score >1). e The lengths, GC contents, and H3K9me2 levels of TEs. The differences between heat-activated TEs and total TEs in the Arabidopsis genome was compared by a two-sided Mann–Whitney U test for each feature. For each box plot, center lines indicate the medians; boxes show the 25th and 75th percentiles; whiskers extend to the minimum and maximum. f Chromosomal distributions of heat-activated TEs in chromosomes 1 and 2. The uppermost track shows the density of Copia (blue) and Gyspy (red) retrotransposon superfamilies and DNA transposons (green). The two tracks in the middle show the distribution of retrotransposons and DNA Transposons. Expression differences were calculated by log₁₀ (Heat_CPM − Control_CPM). TEs belonging to the Copia and Gyspy retrotransposon superfamilies and DNA transposons are colored in blue, red, and green, respectively. Other retrotransposons are colored in gray. The bottom track shows the positions of centromeres (black) and pericentromeres (gray).

Fig. 2. Characterization of two groups of heat-activated TEs.

a Heatmap of heat-activated TEs in Control, Heat, and Recovery. The chromatin track shows the genomic locations of heat-activated TEs. The class track shows the three classes of the heat-activated TEs. The expression track shows the expression levels of heat-activated TEs in log₂ (CPM + 1). The bottom three tracks show relative expression levels (Z-scores) for Control, Heat, and Recovery and range from −2 (blue) to 2 (red). Most heat-activated TEs, classified into Group 1, are silenced again during recovery. Some heat-activated TEs, classified into Group 2, exhibited persistent activation or were not completely restored during recovery. b Expression dynamics of randomly selected Group 1 and Group 2 TEs as determined by RT-qPCR. The expression value of each replicate was normalized to the expression level of Col-0. Circles denote relative expression values. Bars indicate mean values of three technical replicates. c The enrichment (red) and depletion (blue) patterns of epigenetic features in Group 1 and Group 2 TEs compared with total TEs in Arabidopsis. Relative density ranges from −2 to 2. Gray indicates that the epigenetic feature is neither enriched nor depleted (two-sided Mann–Whitney U test, p > 0.05). Source data underlying  b and c are provided as a Source Data file.

Fig. 3. Heat induces rearrangement of chromatin organization.

a Genome-wide chromatin organization revealed by a Hi-C interaction frequency heatmap at 100 kb resolution for all Arabidopsis chromosomes in Control and Heat. Each pixel denotes all interactions between any two 100 kb genomic loci from the linear genome. Intensity represents log₂ normalized contact frequencies. The top track shows the positions of centromeres (black) and pericentromeres (gray). b Genome-wide heatmap of relative interaction differences between Heat and Control. The relative difference is calculated as the difference of two interaction frequencies divided by the mean of two interaction frequencies. In the plot, Hi-C interactions that become stronger (chromosome condensation) in Heat compared with Control are in red to orange, while Hi-C interactions that become weaker (chromosome loosening) in Heat compared with Control are in blue to green. Hi-C interactions that do not change in Heat compared with Control are in white. The genomic bin size is 100 kb and the upper- and lower-triangular matrices are symmetric. The top track shows the positions of centromeres (black) and pericentromeres (gray). c Averaged scaling plot of interaction frequencies against increasing genomic distance for all Arabidopsis chromosomes. The genomic bin size is 100 kb. d Averaged scaling plot of interaction frequencies against increasing genomic distance for all Arabidopsis chromosome arms. The genomic bin size is 100 kb. e Averaged scaling plot of interaction frequencies against increasing genomic distance for all Arabidopsis pericentromeric regions. The genomic bin size is 100 kb.

Fig. 4. Heat weakens chromatin compartmentation.

a Pie chart showing the percentages of chromatin compartment switching induced by heat. A compartment: active compartment; B compartment: repressive compartment. b Saddle plots of chromatin compartmentalization: mean cis observed interaction frequencies divided by expected interaction frequency between 20 kb bins. The plots are ordered by PC1 values from Control. Interactions between A compartments are in the top right, and interactions between B compartments are in the bottom left. c Compartment strengths in Control and Heat for each chromosome arm split by centromere. Compartment strength is defined as natural logarithm of interactions of AA + BB normalized by interactions of AB. For the box plot, center lines indicate the medians; boxes show the 25th and 75th percentiles; whiskers extend to the minimum and maximum. d Heatmap showing contact frequencies between and within chromosomes 3 and 4 in Control and Heat. The upper- and lower-triangular matrices correspond to Control and Heat, respectively. The values on the diagonal line (interactions between nearby bins) are assigned to zero. The KEEs are highlighted by dashed, red circles and interconnected by dashed lines to indicate the relationships with each other and the positions of KEEs in the linear genome. The track below shows the positions of centromeres (black), pericentromeres (gray), the knob hk4s (blue), and KEEs (red). The lower panel shows the positions and differential expression of heat-activated TEs within chromosomes 3 and 4, PC1 values in Control condition, PC1 values in Heat condition, and ∆PC1 values (PC1_Heat − PC1_Control) horizontally. The positive PC1 values indicate A compartments. The negative PC1 values indicate B compartments. The positions of pericentromeres and the knob hk4s with increased PC1 values are indicated. Source data underlying a and d are provided as a Source Data file.

Fig. 5. Reduced interactions among KEEs under heat stress.

a Pairwise interaction scores of all ten KEEs in Control and Heat, defined as the interactions per hundred square kilobases per billion mapped reads (ihskb). KEE3/KEE4 and KEE7/KEE8 are highlighted. For each box plot, center lines indicate the medians; boxes show the 25th and 75th percentiles; whiskers extend to the minimum and maximum. b Perspective plots of the interactions between KEE7 and KEE8 in Control and Heat. The X-axis and Y-axis show the genomic positions of KEE7 (Chr4: 11,050,000–11,206,537) and KEE8 (Chr4: 15,441,465–15,537,500). The Z-axis shows the interaction scores. c Triangle heatmaps of the relative differences between Heat and Control in chromosomes 3 and 4. Hi-C interactions that become stronger (chromosome condensation) in Heat than those in Control are in red to orange, while Hi-C interactions that become weaker (chromosome loosening) in Heat than those in Control are in blue to green. Hi-C interactions that do not change in Heat compared with Control are in white. The genomic bin size is 100 kb. The interactions of KEE3/KEE4 and KEE7/KEE8 are highlighted with circles. The bottom track shows the positions of centromeres (black), pericentromeres (gray), the knob hk4s (blue), and KEEs (red). d 3C results showing the significantly decreased interactions between KEE3 and KEE4 in Heat compared with that in Control, and the increases in Recovery compared with Heat. Circles denote relative interaction frequency. Bars indicate mean values of three technical replicates. e FISH results showing the association rates of KEE3/KEE4 and of KEE7/KEE8. Bars indicate the association rates (“Methods”). Significant differences between two groups at a time were determined by the two-sided Fisher’s exact test, with p values adjusted by the false discovery rate method for multiple comparisons. Exact p values are shown above the bars. Source data of d and e are provided as a Source Data file.

Fig. 6. Chromatin organization rearrangement revealed by high-resolution Hi-C correlates with heat-induced TE expression.

a, b Changes of local chromatin organization and TE expression at KEE3 and KEE4 loci after heat treatment. Two 100 kb regions (depth: 50 kb) containing KEE3 and KEE4, respectively, are shown. The top panels are heatmaps showing natural logarithm of 1 kb Knight and Ruiz (KR) normalized contact frequency in Control and Heat. Below are heatmaps (O/E) showing natural logarithm of observed/expected contact frequency in Control and Heat at 1 kb resolution. Below are heatmaps showing TAD-separation scores at different window sizes for each Hi-C genomic bin in Control and Heat. Loops were collected and transformed from a previous study (“Methods”). TE expression track shows the expression levels of TEs in Control and Heat. Their expression levels in Control and Heat are under the same scale and overlayered together. Only TE expression is shown. The expression of PCGs is not shown (“Methods”). The expanded panels show TE expression and the corresponding chromatin loops over KEE3 and KEE4. c Metagene plot of TAD-separation scores over bodies of heat-activated TEs and total TEs and flanking regions. Only TEs with length over 1,000 bp are considered. TSS transcription start site, TTS transcription termination site. d Heatmap of TAD-separation scores over bodies of 263 heat-activated TEs and flanking regions in Control. e Local rescaled pileup of chromatin organization over bodies of heat-activated TEs (upper panel) and total TEs (lower panel) in Control, Heat, and different biological replicates. The values are normalized by a local background (“Methods”). The diagonal bars are the locations of scaled heat-activated TEs (upper panel) and total TEs (lower panel). f Metagene plot of RNA-Seq read coverage over bodies of heat-activated TEs in Control, Heat, and Recovery. Heat stress can activate heterochromatin-associated transposon elements (TEs). Here, the authors show that heat stress leads to global rearrangement of 3D genome and TEs activation closely correlates with 3D chromatin organization rearrangement in Arabidopsis.