A dimeric state for PRC2

Abstract

Polycomb repressive complex-2 (PRC2) is a histone methyltransferase required for epigenetic silencing during development and cancer. Long non-coding RNAs (lncRNAs) can recruit PRC2 to chromatin. Previous studies identified PRC2 subunits in a complex with the apparent molecular weight of a dimer, which might be accounted for by the incorporation of additional protein subunits or RNA rather than PRC2 dimerization. Here we show that reconstituted human PRC2 is in fact a dimer, using multiple independent approaches including analytical size exclusion chromatography (SEC), SEC combined with multi-angle light scattering and co-immunoprecipitation of differentially tagged subunits. Even though it contains at least two RNA-binding subunits, each PRC2 dimer binds only one RNA molecule. Yet, multiple PRC2 dimers bind a single RNA molecule cooperatively. These observations suggest a model in which the first RNA binding event promotes the recruitment of multiple PRC2 complexes to chromatin, thereby nucleating repression.

Figure 1.

Expression and purification of PRC2 4m and PRC2 5m as MBP fusions. (A) Fractionation of PRC2 4m over Sephacryl S-500 HR column after cleavage of MBP tags. (B) Identical process performed for PRC2 5m. (C) SDS-PAGE confirms that all PRC2 subunits appear with the expected molecular weights, before and after cleaving MBP tags.

Figure 2.

SEC confirming that reconstituted PRC2 5m is a dimer. (A) SEC using Superdex 200 10/300 column for PRC2 5m, before and after cleaving MBP tags. (B) SEC using analytical Superose 6 PC 3.2/30 column for PRC2 5m, before and after cleaving MBP tags.

Figure 3.

SEC-MALS confirming that reconstituted PRC2 5m is a dimer. SEC-MALS performed using Shodex Protein KW-803 column for PRC2 5m, after removing MBP tags. Retention volume range used to calculate molecular mass is shown for each peak.

Figure 4.

Co-immunoprecipitation indicates the existence of more than one EZH2 within reconstituted PRC2. (A) Experiment outline: nucleic acid-free human PRC2 was reconstituted by co-expressing and co-purifying EZH2 as both MBP-tagged and untagged variants, together with all four other subunits as untagged proteins. Complexes containing at least one copy of tagged EZH2 were bound to amylose beads, washed and eluted. The expected fraction of each variant in each step is indicated for the case of dimeric PRC2. (B) Input, sample prior to MBP pull down. MBP pull down, complexes containing MBP-EZH2 were recovered using amylose beads. Immunoblotting using an anti-EZH2 antibody confirmed the presence of both MBP-tagged and untagged EZH2 in the complex. Controls included the same complex reconstituted either without untagged EZH2 or without MBP-tagged EZH2, as indicated. Immunoblotting using an anti-MBP antibody confirmed the presence of the MBP tag only on designated samples and the absence of C-terminal truncated MBP-EZH2 with molecular weight similar to EZH2. Immunoblotting was used to confirm the presence of other core PRC2 subunits SUZ12, EED and RBBP4 (see Supplementary Figure S3 for complete blots). (C) Semi-quantitative immunoblotting using anti-EZH2 antibodies revealed a ratio of ∼3:1 tagged:untagged EZH2 in the MBP pull-down sample.

Figure 5.

Each PRC2 dimer interacts with a single RNA molecule. PRC2 5m or 4m was incubated with a low concentration of radiolabeled HOTAIR 1–200 and various concentrations of unlabeled HOTAIR 1–2148 and then subjected to EMSA. The electrophoretic mobility of PRC2-HOTAIR 1–200 was identical in the presence or absence of the unlabeled HOTAIR 1–2148, indicating that HOTAIR 1–2148 is not incorporated into the PRC2-HOTAIR 1–200 complex. A notable retardation of PRC2 was obtained when bound to HOTAIR 1–2148 alone, in the absence of HOTAIR 1–200, demonstrating the ability of these EMSA conditions to detect such a supershift, if it were present.

Figure 6.

A single RNA molecule can bind multiple PRC2s. (A) Different concentrations of PRC2 5m were incubated with 10 nM fluorescent PRC2 and 500 nM radiolabeled HOTAIR 1–200 RNA. Retardation of PRC2-RNA complex was observed upon increasing the PRC2 concentration, indicating multiple PRC2 binding events on the RNA. (B) Different concentrations of PRC2 5m were incubated with HOTAIR 1–2148, before performing EMSA. In addition to the expected increment in the fraction of bound RNA, gradual retardation of HOTAIR 1–2148 occurred with increasing PRC2 concentration, indicating multiple binding events of PRC2 on the RNA.

Figure 7.

Molecular model for the recruitment of multiple PRC2s following a single RNA binding event. (A) Short nascent transcripts are unlikely to bind PRC2, as the affinity of PRC2 to short RNAs is low (23). (B) When a nascent transcript reaches a length of a few tens to a few hundreds of bases, sufficient to span both RNA-binding sites on a PRC2 dimer, then PRC2 can bind the RNA promiscuously with submicromolar affinity (23). Binding involves multiple PRC2 dimers that bind the RNA in a cooperative manner, thus providing a potential nucleation point for recruitment to chromatin. (C) If PRC2 encounters previously deposited H3K27me3 marks, its affinity to nucleosomes will increase (5,6,44) and it will deposit to chromatin as shown. On the other hand, if PRC2 senses H3K4me3 or H3K36me3 active marks, its affinity to nucleosomes will be reduced (7,8), preventing recruitment to chromatin. In the absence of recruitment to chromatin, the RNA may serve as a decoy that will strip PRC2 away from active genes (23,39).