DNA polymerase epsilon is required for heterochromatin maintenance in Arabidopsis

Abstract

Background: Chromatin organizes DNA and regulates its transcriptional activity through epigenetic modifications. Heterochromatic regions of the genome are generally transcriptionally silent, while euchromatin is more prone to transcription. During DNA replication, both genetic information and chromatin modifications must be faithfully passed on to daughter strands. There is evidence that DNA polymerases play a role in transcriptional silencing, but the extent of their contribution and how it relates to heterochromatin maintenance is unclear.

Results: We isolate a strong hypomorphic Arabidopsis thaliana mutant of the POL2A catalytic subunit of DNA polymerase epsilon and show that POL2A is required to stabilize heterochromatin silencing genome-wide, likely by preventing replicative stress. We reveal that POL2A inhibits DNA methylation and histone H3 lysine 9 methylation. Hence, the release of heterochromatin silencing in POL2A-deficient mutants paradoxically occurs in a chromatin context of increased levels of these two repressive epigenetic marks. At the nuclear level, the POL2A defect is associated with fragmentation of heterochromatin.

Conclusion: These results indicate that POL2A is critical to heterochromatin structure and function, and that unhindered replisome progression is required for the faithful propagation of DNA methylation throughout the cell cycle.

Keywords: DNA methylation; DNA polymerase epsilon; Heterochromatin; Replication stress; Silencing.

Fig. 1

Genome-wide release of silencing in a new pol2a-12 mutant allele. a L5-GUS transgene activity detected by X-Gluc histochemical staining in 3-week-old L5 plants and anx2/pol2a-12 mutants. b Representative pictures of 16-day-old L5 and anx2 plants. Scale bar: 1 cm. c (top) Gene model for POL2A showing point mutations. The new nomenclature of pol2a mutant alleles used in this study is detailed in Table S1. The til1-4/pol2a-4 mutant allele contains two point mutations in the POL2A gene: one in exon 12 and a second one in intron 14. abo4-1/pol2a-8 is a G to A mutation at position 4171 changing Gly 522 to Asp. The same mutation was identified in anx3/pol2a-13. esd7-1/pol2a-10 is a G to A mutation at position 8074 changing Gly 992 to Arg. anx4/pol2a-14 is a C to T transition at position 5023 changing Leu 579 to Phe. (bottom) POL2A protein model showing point mutations and their corresponding amino acid changes (positions are indicated relative to the start ATG). Functional domains are indicated according to the conserved domain database from NCBI. DNA PolB ε: DNA polymerase type-B epsilon subfamily catalytic domain. Exo: DNA polymerase family B 3′ to 5′ exonuclease domain. DUF1714: domain of unknown function. d Transcript accumulation at four silent loci detected by RT-qPCR, normalized to the ACTIN2 gene with pol2a-8 set to 1. Asterisks mark statistically significant differences (unpaired two-sided Student’s t test, P < 0.05). Error bars represent standard error of the mean across three biological replicates. e Number of PCGs and TEs detected as differentially expressed in pol2a-12. f Proportion of TE superfamilies of all TEs in the Arabidopsis genome and of TEs upregulated in pol2a-12. g Changes in transcript accumulation in pol2a-12 relative to L5 control plants represented along chromosome 3 by log₂ ratios of average reads per kilobase per million mapped reads (RPKM) over non-overlapping 100 kb bins (black, left y-axis). Total TE density is the proportion of TE annotations per 100 kb bins, indicating the pericentromeric region (gray, right y-axis)

Fig. 2

POL2A ensures silencing independently of ATXR5/6-mediated H3K27me1. a Overlap between TEs upregulated in atxr5/6 (data from Ikeda et al. [35]) and pol2a-12. b Metaplots showing average H3K27me1 enrichment (log₂ signal over input) at H3K27me1 peaks. Shaded areas show standard deviation. One biological replicate is shown. c Metaplots showing H3K27me1 changes (log₂ mutant / WT) pol2a-12 and atxr5/6 (data from Ma et al. [43]) at H3K27me1 peaks overlapping TEs upregulated in both atxr5/6 and pol2a-12, represented as in b. Average of two replicates is shown. d Transcript accumulation in reads per kilobase per million mapped reads (RPKM) in indicated genotypes. The effect of genotype was verified with a Kruskal-Wallis rank-sum test. Significant differences between groups, evaluated by a Dwass-Steel-Crichtlow-Fligner test, are indicated by lowercase letters (P < 0.05). Two biological replicates are shown for each genotype, except for pol2a atxr5/6 where only one sample was analyzed

Fig. 3

Heterochromatin fragmentation in pol2a-12. a DAPI-stained nuclei extracted from WT and pol2a-12 plants. Scale bar: 5 μm. b Number of chromocenter per nucleus (left), area of individual chromocenter normalized to the entire nucleus area (middle) and relative heterochromatic fraction (right) in WT and pol2a-12 quantified on 56 and 59 DAPI-stained nuclei, respectively. P-values from an unpaired two-sided Student’s t-test are indicated. c Metaplots showing H2A.W enrichment (log₂ signal over input) in L5 and pol2a-12 at H2A.W peaks (top) and at peaks overlapping TEs upregulated in pol2a-12 (bottom). Shaded areas show standard deviation. TE annotations were extended 1 kb upstream

Fig. 4

CHG and H3K9 hypermethylation in pol2a mutants. a Average methylation rates in CG, CHG and CHH contexts in L5 and pol2a-12. b Methylation rates in CG, CHG, and CHH contexts in L5, pol2a-12, Col-0 (WT), pol2a-8, and pol2a-10, averaged over non-overlapping 100 kb bins on chromosome 4. c Metaplots showing methylation levels at TEs upregulated in pol2a-12. Annotations were aligned to their 5′ or 3′ end and average methylation was calculated for each 100-bp bin from 3 kb upstream to 3 kb downstream. d Differentially methylated regions (DMRs) identified in pol2a-12 (see the “Methods” section). DMRs were further sorted between euchromatin and heterochromatin based on their genomic location. e Metaplots showing H3K9me2 enrichment (log₂ signal over input) in L5 and pol2a-12 at H3K9me2 peaks. Shaded areas show standard deviation. One replicate is shown

Fig. 5

CMT3-dependent CHG methylation compensates pol2a-12 molecular defects. a Metaplots showing TE methylation rates in CHG context in the indicated genotypes. Annotations were aligned to their 5′ or 3′ end and average methylation was calculated for each 100-bp bin from 3 kb upstream to 3 kb downstream. b Transcript accumulation in reads per kilobase per million mapped reads (RPKM) at the indicated genes. Two replicates per samples are shown. c TE methylation changes in CHG context in pol2a-12 and pol2a-12 cmt3 normalized to WT and cmt3, respectively. d Venn diagrams showing the overlap between TEs upregulated in pol2a-12, cmt3 and pol2a-12 cmt3. e (left) Relative heterochromatic fraction (RHF) evaluated from DAPI-stained pictures of nuclei from the indicated genotypes. The effect of genotype was verified with a Kruskal-Wallis rank-sum test. Significant differences between groups were evaluated by a Dwass-Steel-Crichtlow-Fligner test and are indicated by lowercase letters (P < 0.05). The number of analyzed nuclei per genotype is indicated below boxplots. DAPI-stained nuclei extracted from rosette leaves of the indicated genotypes (right). Scale bar: 5 μm

Fig. 6

DNA replication hindrance provokes CHG hypermethylation and release of silencing. A, B Metaplots showing TE methylation rates in CHG context in the indicated conditions. Annotations were aligned to their 5′ or 3′ end and average methylation was calculated for each 100-bp bin from 3 kb upstream to 3 kb downstream. For B, we used published datasets for rpa2a (Stroud et al. [22]) and pold2 (Zhang et al. [16, 17]). C, D Average methylation rates in CHG context in L5 and pol2a-12 in the absence (0 mm) or presence (2 mM) of hydroxyurea (HU) calculated at all TEs (C) or LTR/Gypsy TEs (D). E CHG methylation rates over non-overlapping 100 kb bins on chromosome 4. F Proportion of γ-H2A.X labeled nuclei in L5 and pol2a-12 plants treated or not with HU. A two-way ANOVA showed the significant effect (P < 2e−16) of HU treatment. Lowercase letters indicate significant differences between groups using Tukey’s post hoc tests (P < 0.05). Error bars represent standard error of the mean across three or four biological replicates, as indicated above bars. G Number of γ-H2A.X foci per nucleus in L5 and pol2a-12 plants treated or not with HU, excluding nuclei without γ-H2A.X signal. P-values from a two-sided unpaired Wilcoxon rank-sum test are indicated. H Transcript accumulation at three silent loci analyzed by RT-qPCR in L5 and pol2a-12 seedlings treated with various concentrations of HU, normalized to the ACTIN2 gene with L5 0mM HU set to 1. Asterisks mark statistically significant differences (two-sided unpaired Student’s t test, *P < 0.05, **P < 0.005). Error bars represent standard error of the mean across three biological replicates