AbstractIn plants and mammals, DNA methylation is a hallmark of transposable element (TE) sequences that contributes to their epigenetic silencing. In contrast, histone H3 lysine 27 trimethylation (H3K27me3), which is deposited by the Polycomb Repressive Complex 2 (PRC2), is a hallmark of repressed genes. Nevertheless, there is a growing body of evidence for a functional interplay between these pathways. In particular, many TE sequences acquire H3K27me3 when they lose DNA methylation and it has been proposed that PRC2 can serve as a back-up silencing system for hypomethylated TEs. Here, we describe in the flowering plant Arabidopsis thaliana the gain of H3K27m3 at hundreds of TEs in the mutant ddm1, which is defective in the maintenance of DNA methylation specifically over TE and other repeat sequences. Importantly, we show that this gain essentially depends on CURLY LEAF (CLF), which is one of two otherwise partially redundant H3K27 methyltransferases active in vegetative tissues. Finally, our results challenge the notion that PRC2 can be a compensatory silencing system for hypomethylated TEs, as the complete loss of H3K27me3 in ddm1 clf double mutant plants was not associated with further reactivation of TE expression nor with a burst of transposition. Instead, and surprisingly, ddm1 clf plants exhibited less activated TEs, and a chromatin recompaction as well as hypermethylation of linker DNA compared to ddm1. Thus, we have described an unexpected genetic interaction between DNA methylation and Polycomb silencing pathways, where a mutation in PRC2 does not aggravate the molecular phenotypes linked to TE hypomethylation in ddm1 but instead partially suppresses them.Author summaryEpigenetic marks are covalent modifications of the DNA or its associated proteins (Histones) that impact gene expression in a heritable manner without changing DNA sequence. In plants and mammals, DNA methylation and trimethylation of Lysine 27 of Histone 3 (H3K27me3) are two conserved, major epigenetic systems that mediate the transcriptional silencing of transposons (invasive mobile genetic elements) and of developmental genes respectively. However, in the absence of DNA methylation, H3K27me3 marks can be recruited to transposons, suggesting that the two silencing systems can be compensatory. To test this hypothesis, we analyzed a compound DNA methylation and H3K27me3 mutant of the plant model Arabidopsis thaliana (importantly, mammals harboring equivalent mutations would not be viable). First, this approach allowed us to gain mechanistic insights into the recruitment of H3K27me3 at transposons. Furthermore, we also showed that transposon silencing release in the DNA methylation mutant was not enhanced, contrary to our initial hypothesis, but, surprisingly, partially suppressed by a mutation in H3K27me3 deposition. Thus, our genomic analysis revealed an unexpected and antagonistic genetic interaction between two major silencing pathways whose interplay is at the heart of many biological processes, including cancer.
Polycomb mutant partially suppresses DNA hypomethylation–associated phenotypes in Arabidopsis