A PHD-polycomb repressive complex 2 triggers the epigenetic silencing of FLC during vernalization

Abstract

Vernalization, the acceleration of flowering by winter, involves cold-induced epigenetic silencing of Arabidopsis FLC. This process has been shown to require conserved Polycomb Repressive Complex 2 (PRC2) components including the Su(z)12 homologue, VRN2, and two plant homeodomain (PHD) finger proteins, VRN5 and VIN3. However, the sequence of events leading to FLC repression was unclear. Here we show that, contrary to expectations, VRN2 associates throughout the FLC locus independently of cold. The vernalization-induced silencing is triggered by the cold-dependent association of the PHD finger protein VRN5 to a specific domain in FLC intron 1, and this association is dependent on the cold-induced PHD protein VIN3. In plants returned to warm conditions, VRN5 distribution changes, and it associates more broadly over FLC, coincident with significant increases in H3K27me3. Biochemical purification of a VRN5 complex showed that during prolonged cold a PHD-PRC2 complex forms composed of core PRC2 components (VRN2, SWINGER [an E(Z) HMTase homologue], FIE [an ESC homologue], MSI1 [p55 homologue]), and three related PHD finger proteins, VRN5, VIN3, and VEL1. The PHD-PRC2 activity increases H3K27me3 throughout the locus to levels sufficient for stable silencing. Arabidopsis PHD-PRC2 thus seems to act similarly to Pcl-PRC2 of Drosophila and PHF1-PRC2 of mammals. These data show FLC silencing involves changed composition and dynamic redistribution of Polycomb complexes at different stages of the vernalization process, a mechanism with greater parallels to Polycomb silencing of certain mammalian loci than the classic Drosophila Polycomb targets.

Fig. 1.

VRN2 is constitutively associated with FLC. (A) Schematic representation of the FLC locus. (B and C) ChIP showing association of a VRN2-TAP protein fusion expressed under the control of the VRN2 or the 35S promoter with different FLC regions. VRN2-TAP is a carboxyl-terminal fusion of the tandem affinity purification (TAP) tag (38) to the VRN2 protein. The construct was transformed into the vrn2–1 fca-1 mutant; both fusions rescue the vrn2 phenotype. Q-PCR results are normalized relative to an fca-1 control (not containing a TAP-tagged protein), represented as a dashed line. Each histogram shows the results from each fusion protein for one complete experiment; replicate experiments (Fig. S1) all showed enrichment relative to the control over the entire locus.

Fig. 2.

VRN5 association with FLC and PHD-PRC2 complex composition. (A) ChIP showing association of 35S-VRN5-TAP protein with different FLC regions. 35S-VRN5-TAP is a carboxyl-terminal fusion to the TAP-tag that rescues a vrn5–1 fca-1 mutant phenotype. Q-PCR results are normalized relative to the fca-1 control (not containing a TAP-tagged protein), represented as a dashed line. The graph shows the average enrichment ± SE obtained from three independent ChIP experiments. Semiquantitative PCR analysis revealed no association of VRN5 in FLC regions B, G, and UTR in T0 plants (data not shown). (B) Size exclusion chromatography analysis of VRN5-TAP protein. VRN5-TAP protein eluted as part of a multiprotein complex and free protein of ≈ 88 kDa. Protein extracts were prepared from seedlings before vernalization (NV) or after cold treatment (T0) and separated through Superdex 200. Size markers (kDa) are indicated. Vo, void volume. (C) VRN5-TAP–associated proteins. Plants carrying a fully functional VRN5-TAP fusion expressed under the control of the endogenous promoter were harvested after 8 weeks cold and used in TAP-tag affinity purification. Protein components were identified using LC-MS/MS (Dataset S1). The list includes the proteins previously reported to be involved in vernalization and Polycomb function by genetic or molecular analysis. A protein was judged to be associated with the VRN5-TAP protein when it was absent from the vernalized fca-1 control material taken through the same purification procedure. Full details on the peptides and scores from the TAP purification and control sample are included in SI Text available on line as an Excel spreadsheet.

Fig. 3.

Histone modifications at the FLC locus. Immunoprecipitated DNA was analyzed by real-time qPCR, and enrichment was represented as percentage of INPUT (% INPUT). All ChIP were normalized for histone H3 occupancy. Data in the graphs are the average of four qPCR assays from two independent ChIP experiments; the bars represent standard error. (A) ChIP analysis of histone H3 acetylation. (B) ChIP analysis of histone H3K27 trimethylation. (C) qPCR results of H3K27me3 ChIP for region FLC-C in plants carrying a vrn2, vrn5, or vin3 mutation. Region C is shown, but the same profile was also observed for all other FLC regions.

Fig. 4.

Possible sequence of events in the assembly and association of specific Polycomb complexes at different stages of epigenetic silencing of FLC during vernalization. In nonvernalized plants the PRC2 core complex is associated with the whole FLC locus. Prolonged cold leads to VRN5 localization specifically at region C as part of the PHD-PRC2 complex with VIN3 and VEL1 and to a decrease in histone acetylation. In plants returned to warm conditions, VIN3 is no longer expressed, and VRN5 associates more widely throughout the locus, concomitant with a significant increase in H3K27 trimethylation.