Regulatory interactions between RNA and polycomb repressive complex 2

Abstract

Polycomb repressive complex 2 (PRC2) is a histone methyltransferase that is localized to thousands of mammalian genes. Though important to human disease and as a drug target, how PRC2 is recruited remains unclear. One model invokes cis-regulatory RNA. Herein, we biochemically and functionally probe PRC2's recognition of RNA using the X-inactivation model. We observe surprisingly high discriminatory capabilities. While SUZ12 and JARID2 subunits can bind RNA, EZH2 has highest affinity and is somewhat promiscuous. EED regulates the affinity of EZH2 for RNA, lending greater specificity to PRC2-RNA interactions. Intriguingly, while RNA is crucial for targeting, RNA inhibits EZH2's catalytic activity. JARID2 weakens PRC2's binding to RNA and relieves catalytic inhibition. We propose that RNA guides PRC2 to its target but inhibits its enzymatic activity until PRC2 associates with JARID2 on chromatin. Our study provides a molecular view of regulatory interactions between RNA and PRC2 at the chromatin interface.

Figure 1. PRC2 discriminates between cognate and nonspecific RNA

(A,B) Coomassie staining (A) and Western blotting (B) of purified mouse PRC2 subunits. (C) HMTase assay (top) and Coomassie staining of H3 (bottom) show that purified PRC2 is active. (D) RepA stem-loop structures I-IV (Maenner et al., 2010). (E,F) EMSAs with long (≥300nt) (E) and shorter (≤200nt) (F) RNA probes. Tsix I-IV, antisense of RepA I-IV. Tsix I-II, antisense of RepA I-II. Tsix III-IV, antisense of RepA II-IV. Probe used at 2nM unless otherwise indicated. PRC2 concentrations between 40–1000nM as shown. GST, 1000nM. (G) Two-probe competition experiments. Left: Competition between RepA I-IV and P4-P6 shows PRC2’s specific preference for RepA I-IV over P4-P6. RNA and PRC2 concentrations as indicated in nM. Quantification of RepA I-IV bound and free fractions, as well as P4-P6 input are indicated. Right: Competition between hHOTAIR 1-300 and P4-P6 for the same PRC2 pool shows preference for hHOTAIR 1-300 over P4-P6 further showing PRC2 specific RNA recognition capability. Quantification of hHOTAIR 1-300 bound and free fractions, as well as P4-P6 input are indicated. (H) EMSA with 2nM RepA I-IV probe and cold competitor (RepA I-IV, tRNA) at indicated concentrations. Please also refer to Figures S1 and S2.

Figure 2. A range of PRC2-RNA binding affinities

(A) Example of double-filter binding assay used to obtain binding isotherms and K_d values. Top (bound), nitrocellulose membrane. Bottom (free), Hybond-N+ membrane. (B) Binding isotherms of PRC2 to 2 nM RNA of species indicated. Equilibrium dissociation constants (K_d) and R² values shown in table. N/A, not applicable because curves were too flat. “≫1000 nM” denotes a K_d in excess of that which could be measured under our assay conditions. (C) Reciprocal binding experiments in which 1–1000nM RNA is titrated against constant 50nM PRC2, with K_d and R² values shown. A labeled corresponding RNA (<1%) was used as tracer. (D) EMSA using 2nM RepA I-IV probe in the presence of cold competitor RNA species (RepA I-IV, Tsix I-IV, RepA I-IV: Tsix I-IV pre-annealed duplex, MBP) at indicated concentrations.

Figure 3. RNA binding potential of PRC2 subunits and subcomplexes

(A) EMSA shows high affinity of EZH2 (concentrations shown) for RNA. (B) EMSA shows that SUZ12 also binds RNA but at higher concentrations. (C) EMSA shows that EED binds poorly to all tested RNAs. (D) Binding isotherms of PRC2 single subunits EZH2 and SUZ12, and subcomplexes: EZH2+EED (EZ+EE), SUZ12+EED (S+EE), EZH2+SUZ12 (EZ+S) and EZH2+SUZ12+EED (EZ+S+EE). Isotherms generated from data obtained from double-filter binding experiments. (E) EMSAs showing PRC2 subcomplexes interactions with long (≥300nt) RNAs. (First panel) shows that EED (EE) attenuates EZH2 (EZ) binding to RNA. (Second panel) EMSA shows that EED has modest effects on SUZ12 binding to RNA. (Third panel) EMSA shows that the EZH2-SUZ12 subcomplex (EZ+S) also binds RNA, but with lower affinity that EZH2 alone. (Bottom) EMSA of the EZH2-SUZ12-EED subcomplex (EZ+S+EE) showing reduced RNA binding. (F) Table of K_d and R² values for PRC2 single subunits- and PRC2 subcomplexes-RNA interactions. Please also refer to Figure S2.

Figure 4. RNA suppresses the histone methyltransferase activity of EZH2

(A) Effect of RepA I-IV on the activity of purified PRC2. HMTase assay (top) and Coomassie staining (bottom) of in vitro-assembled recombinant nucleosomes (bottom) showing active PRC2. PRC2 (150nM). RepA concentrations are indicated. (B) Kinetics of histone H3 methylation by PRC2 showing an inhibitory effect of 50nM RepA I-IV. The incorporation of ³H-methyl groups into histone H3 is plotted as a function of time. (C) Kinetics of histone H3 methylation by PRC2 showing changes in inhibition at increasing concentrations of various RNA species. Legend indicates [RNA] in nM. (D) Table depicting the steady-state rates of histone H3 methylation by PRC2 in the presence of different RNAs from (C). (E) Histone H3 methylation rate as a function of RNA concentration at 150nM PRC2. The plot shows the percentage of the methylation rate exhibited at a given [RNA] compared to the rate in the absence of RNA (100%). The RNA concentration is plotted on a log scale, with the corresponding linear values indicated on the bottom. Corresponding table shows the IC₅₀ and R² values for tested RNAs. (F) Comparison of the data in (E) among RNA species shows that the dose-response behavior of the methylation rate in response to different RNAs exhibits significant differences. (G) Plot showing percent of Histone H3 methylation activity as a function of RNA concentration at 20nM PRC2. Table shows the IC₅₀ and R² values for RepA I-IV and MBP 1-300. (B–G) Data are presented as mean ± SD.

Figure 5. JARID2 binds RNA with modest affinity but strongly attenuates PRC2 binding to RNA

(A) EMSA shows that JARID2 binds specifically but with moderate affinity to different size RNAs. (B) Binding isotherms of JARID2 to indicated RNAs. Equilibrium dissociation constants (K_d) and R² values shown in table. (C) Average of three double-filter binding experiments showing a significant decrease on RNA binding in PRC2 complexes containing EED and JARID2. Data normalized to PRC2 binding. Minimal protein binding to MBP is observed at 200nM protein. (D) Significant decrease on RNA binding to PRC2 and EZH2 is also observed when purified JARID2 is added to PRC2 complexes in double-filter binding experiments. RNA[2nM] and protein concentrations are indicated. Data normalized to PRC2 binding. (E) Binding isotherms for 25nM protein complexes that include PRC2 and/or JARID2. In PRC2:JARID2 reactions, JARID2 was added stoichiometrically to PRC2. RNAs of indicated species are titrated. K_d and R² values shown in table. (B–G) Data are presented as mean ± SD. Please also see Figure S3 for the reciprocal titrations, which yielded similar results.

Figure 6. Regulatory interactions of PRC2, JARID2, and RNA at the chromatin interface

(A) Effect of JARID2 on the H3 methylation activity of purified PRC2. HMTase assay (top) and Coomassie staining (bottom) of in vitro assembled recombinant nucleosomes. Histones are indicated. JARID2 (150nM) and PRC2 (150nM). (B) Kinetics of histone H3 methylation by PRC2. Incorporation of ³H-methyl groups into histone H3 is plotted as a function of time. (Left) Time-course of PRC2 histone H3 methylation showing increased methylation rate in the presence of 150nM JARID2. (Middle) Inhibitory effect of 50nM RepA I-IV. (Right) Addition of JARID2 (150nM) reduces the inhibitory effect of RepA on the rate of histone H3 methylation by PRC2 (150nM). The incorporation of ³H-methyl groups into histone H3 is plotted as a function of time. (C) Kinetics of histone H3 methylation by PRC2 in the presence of JARID2 and increasing concentrations of various RNA species. (D) Table depicting the steady-state methylation rates in the presence of JARID2 (150nM) from (D). (E) Histone H3 methylation rate as a function of RNA concentration in the presence of JARID2. The plot shows the percentage of the methylation rate exhibited at a given [RNA] compared to the rate in the absence of RNA (100%). The RNA concentration is plotted on a log scale, with the corresponding values indicated on the bottom. The attached table shows the IC₅₀ and R² values for the tested RNAs. (F) Comparison of the data in (E) among RNA species shows that the dose-response behavior of the methylation rate in response to different RNA exhibits significant differences. (G) Effect of JARID2 on the incorporation of ³H-methyl groups into Histone H3 at 20nM PRC2. (H) Plot showing percent of Histone H3 methylation activity as a function of RNA concentration in the presence of 20nM PRC2 and JARID2. The RNA concentration is plotted on a log scale, with the corresponding linear values indicated on the bottom. Corresponding table shows the IC₅₀ and R² values for RepA I-IV and MBP 1-300. (B–H) Data are presented as mean ± SD.

Figure 7. Summary, synthesis, and model

(A) Summary and synthesis of our data in the context of published works, taking into account the cumulative in vivo and in vitro observations. (B) Proposed steps of PRC2 recruitment, chromatin loading, and H3K27 trimethylation, using RepA as a model. [1] High level expression of Tsix RNA titrates the RepA-PRC2 interaction, either by directly binding PRC2 or RepA (Ogawa et al., 2008; Zhao et al., 2008). Duplex RNA is inert for PRC2 binding (Fig. 1). [2] When Tsix is developmentally downregulated, RepA co-transcriptionally recruits PRC2 (Zhao et al., 2008). EZH2 binds RNA strongly; SUZ12 binds more weakly (Fig. 3). [3] After Tsix RNA is depleted, PRC2 loads onto chromatin, but its HMTase activity is held in check by RepA (Fig. 4). In the absence of JARID2 [4A], PRC2’s HMTase activity is limited, but low-level H3K27me3 may be detected. [5A] PRC2 dissociates from chromatin without JARID2 on ChrX (da Rocha et al., 2014). When JARID2 is present [4B], RNA-EZH2 interactions are weakened and EZH2’s HMTase activity is promoted, resulting in high levels of H3K27me3. JARID2 can act substoichiometrically but it is not known whether JARID2 is free or bound to chromatin. [5B] The PRC2:JARID2 complex efficiently spreads H3K27me3 along ChrX. (C) Schematic representation of positive and negative regulatory relationships among PRC2 subunits, JARID2, RNA, and the chromatin target. Green arrows, positive regulation. Red arrows, negative regulation. Note that JARID2’s relationship to PRC2’s HMTase activity is “positive” via two sequential negative events.