The Arabidopsis epigenetic regulator ICU11 as an accessory protein of Polycomb Repressive Complex 2

Abstract

Molecular mechanisms enabling the switching and maintenance of epigenetic states are not fully understood. Distinct histone modifications are often associated with ON/OFF epigenetic states, but how these states are stably maintained through DNA replication, yet in certain situations switch from one to another remains unclear. Here, we address this problem through identification of Arabidopsis INCURVATA11 (ICU11) as a Polycomb Repressive Complex 2 accessory protein. ICU11 robustly immunoprecipitated in vivo with PRC2 core components and the accessory proteins, EMBRYONIC FLOWER 1 (EMF1), LIKE HETEROCHROMATIN PROTEIN1 (LHP1), and TELOMERE_REPEAT_BINDING FACTORS (TRBs). ICU11 encodes a 2-oxoglutarate-dependent dioxygenase, an activity associated with histone demethylation in other organisms, and mutant plants show defects in multiple aspects of the Arabidopsis epigenome. To investigate its primary molecular function we identified the Arabidopsis FLOWERING LOCUS C (FLC) as a direct target and found icu11 disrupted the cold-induced, Polycomb-mediated silencing underlying vernalization. icu11 prevented reduction in H3K36me3 levels normally seen during the early cold phase, supporting a role for ICU11 in H3K36me3 demethylation. This was coincident with an attenuation of H3K27me3 at the internal nucleation site in FLC, and reduction in H3K27me3 levels across the body of the gene after plants were returned to the warm. Thus, ICU11 is required for the cold-induced epigenetic switching between the mutually exclusive chromatin states at FLC, from the active H3K36me3 state to the silenced H3K27me3 state. These data support the importance of physical coupling of histone modification activities to promote epigenetic switching between opposing chromatin states.

Keywords: ICU11; Polycomb; chromatin; epigenetic.

Fig. 1.

icu11-3 phenotypes are reminiscent of mutants defective in PRC2. (A) Morphological phenotypes of WT line C12b, the icu11-3 and clf-2 mutants, and synergistic interaction in icu11-3 clf-2 double mutant. (B) Flowering is significantly accelerated under short-day conditions in icu11-3 (P < 0.0001, unpaired Student’s t test). Error bars represent SEM (n = 9 [icu11-3]; n = 27 [WT]). (C–E) Mean gene expression in C12b WT, icu11-3, clf-2, and icu11-3 clf-2 double-mutant lines. Target gene expression is normalized to the housekeeping gene PP2A; error bars represent SEM for three biological replicates. (C) The floral activator FT is overexpressed in early seedling development in icu11-3 compared with WT under short-day conditions. (D) The icu11-3 mutant overexpresses PRC2 target genes during early seedling development. (E) The icu11-3 clf-2 double mutant synergistically overexpresses PRC2 targets compared with either single mutant.

Fig. 2.

icu11-3 affects global histone methylation and DNA methylation at the pericentromeric cen180 repeats. (A) The icu11-3 mutant shows elevation of H3K36me3 and H3K4 methylation, but no change in H3K27me3 compared with WT C12b. Histones were prepared from a nuclear extract from 14-d-old seedlings and serially diluted; Western blots are shown with total histone H3 from the same extraction as loading controls. (B) Southern blot analysis of methylation at the pericentromeric cen180 repeat region. Genomic DNA digested with the methylation-insensitive restriction enzyme MspI (control) and methylation-sensitive enzyme HpaII indicates slightly reduced DNA methylation over centromeric repeats in icu11-3.

Fig. 3.

icu11-3 is defective in vernalization. (A and B) Flowering is significantly more asynchronous at all time points and significantly delayed in the icu11-3 fca-1 genotype after 2 wk vernalization (SI Appendix, Table S4). Days to flowering counted from the end of a pregrowth period of 14 d in warm conditions (NV), or from the end of pregrowth followed by a 2-, 4-, or 6-wk vernalization treatment. Bars represent mean and SEM (n = 20). (B) Flowering plants imaged at 25 d postvernalization. (C–F) Gene expression in the vernalization-requiring fca1 and icu11-3 fca-1 backgrounds. Gene expression is normalized to PP2A; error bars represent SEM for three biological replicates. (C) FT expression at 30 d posttreatment is higher in fca1 than icu11-3 fca-1. (D) Expression of the flowering repressor FLC is higher in icu11-3 fca1 than fca-1 before (NV), during (T0), and 10, 20, or 30 d after vernalization (T10–30), reactivating strongly in icu11-3 fca1 compared with the stable postvernalization repression observed in fca-1. Increasing vernalization time (2, 4, and 6 wk) leads to lower and more stably silenced FLC expression. (E) The antisense RNA COOLAIR is expressed at higher levels in icu11-3 fca1 than fca-1. (F) Expression of VIN3, a cold-induced PRC2 target, is not affected by the icu11-3 mutation.

Fig. 4.

ICU11 influences H3K27me3 and H3K36me3 dynamics at FLC through a vernalization timecourse. (A–E) ChIP analysis of H3K27me3 (A and C) and H3K36me3 (B and D) histone modifications and ICU11 localization (E) at FLC; error bars are SEM of three biological replicates. Abundance, relative to total histone H3 and normalized to STM (H3K27me3) or ACT7 (H3K36me3) was measured in nonvernalized plants, after 4 wk vernalization and vernalization followed by 10 d growth in warm conditions. The icu11-3 fca-1 mutant (C and D) has reduced H3K27me3 and increased H3K36me3 relative to the fca-1 control (A and B). (E) ChIP analysis of ICU11 localization at FLC before and after 4 wk vernalization treatment, normalized to input. Purple lines represent plants containing the GFP-tagged ICU11 construct, while gray lines represent control plants with untagged ICU11. Localization of ICU11-GFP at ACT and STM before (solid bar) and after (empty bar) vernalization shown as positive and negative controls. Underlying A–E, an FLC locus schematic showing UTRs (gray boxes), exons (black boxes), and the Ler background MULE insertion in intron one (white box). (F) Gus staining of FLC expression in FLC^Col-0-GUS, which lacks the MULE in intron one found in Ler background. Expression of this allele is higher in icu11-3, indicating misregulation of the MULE insertion is not responsible for the FLC overexpression phenotype. (G) Expression of FLC-LUC from FLC^Col-0-LUC (no MULE in intron one) is higher in the icu11-3 background than WT before and after 4 wk vernalization and shows reactivation after vernalization followed by postcold growth. Error bars are SEM of three biological replicates.