dKDM2 couples histone H2A ubiquitylation to histone H3 demethylation during Polycomb group silencing

Abstract

Transcription regulation involves enzyme-mediated changes in chromatin structure. Here, we describe a novel mode of histone crosstalk during gene silencing, in which histone H2A monoubiquitylation is coupled to the removal of histone H3 Lys 36 dimethylation (H3K36me2). This pathway was uncovered through the identification of dRING-associated factors (dRAF), a novel Polycomb group (PcG) silencing complex harboring the histone H2A ubiquitin ligase dRING, PSC and the F-box protein, and demethylase dKDM2. In vivo, dKDM2 shares many transcriptional targets with Polycomb and counteracts the histone methyltransferases TRX and ASH1. Importantly, cellular depletion and in vitro reconstitution assays revealed that dKDM2 not only mediates H3K36me2 demethylation but is also required for efficient H2A ubiquitylation by dRING/PSC. Thus, dRAF removes an active mark from histone H3 and adds a repressive one to H2A. These findings reveal coordinate trans-histone regulation by a PcG complex to mediate gene repression.

Figure 1.

Identification of dRAF. (A) Outline of the chromatographic scheme used to purify dRAF. (B) dRING, PH, and PSC also exist outside PRC1. Protein A Sepharose beads coated with either control anti-GST antibodies (Mock IP) or affinity-purified antibodies directed against PC (α-PC IP) were used to immunopurify PC and associated factors from partially purified and concentrated embryo nuclear extracts (H0.4). Bound and unbound material (FT α-PC IP) was analyzed by Western immunoblotting using antibodies directed against the indicated proteins. (C) dRING and PSC form part of a complex distinct from PRC1 and devoid of PC and PH. The unbound fraction after 2 subsequent PC immunodepletions (FT α-PC IP) was incubated with protein A Sepharose beads coated with affinity-purified antibodies directed against dRING. Bound (α-dRING IP) and unbound (FT α-dRING IP) material was analyzed followed by Western immunobloting. (D) Identification of dRAF. H0.4, twice PC-immunodepleted (FT α-PC IP) input of the mock-, or dRING immunopurified (α-dRING IP) fractions were resolved by SDS-PAGE and visualized by silver staining. Proteins present in bands excised from a gel run in parallel were identified by nanoflow LC-MS/MS mass spectrometry. Identified proteins are indicated. Their mascot score and number of unique peptides identified are Mtor: 4741 (Mascot score), 55 (number of unique peptides); PSC: 2226, 31; dKDM2 (dRAF1, CG11033): 1920, 23; Ulp1: 1566, 20; dRAF2 (CG4877): 1312, 17; dRING (SCE): 1160, 12. Asterisks indicate background proteins we routinely observe in our immunopurifications: (*1) Heat shock cognate 4; (*2) γ-tubulin; (*3) RanGap; (*4) β-tubulin; (*5) yolk protein 1 and 2; (*6) RFC.

Figure 2.

dRAF and PRC1 are separate PcG complexes that share dRING and PSC. (A) dRING, PSC, dKDM2, dRAF2, Mtor, and Ulp1 associate in crude Drosophila embryo nuclear extracts (NE). Nuclear extract was incubated with Protein A Sepharose beads coated with either control anti-GST antibodies (mock) or affinity-purified α-dRING, α-PC, α-PH, or α-PSC antibodies. After extensive washes with a buffer containing 600 mM KCl and 0.1% NP-40, bound proteins were resolved by SDS-PAGE and analyzed by immunoblotting. Lane 1 represents 10% of the input material used in the binding reactions. (B) dRING and PSC, but not PC or PH, coimmunoprecipitate with dRAF2, Mtor, and dKDM2. Coimmunoprecipitations were performed and analyzed as described above. (C) Outline of the chromatographic scheme used to characterize the dRAF complex further. (D) The POROS 20 Heparin 400 mM eluate fractionated by Sephacryl S-300 size-exclusion chromatography. The indicated fractions were combined and resolved by SDS-PAGE, followed by immunoblotting with antibodies directed against dKDM2, PSC, dRING, dRAF2, Mtor, Ulp1, Pc, PH, and ISWI. The core dRAF subunits and interactors peaked in fractions corresponding to apparent molecular masses ranging from ∼600 to 900 kDa. The elution of the voided volume (void) and the elution of the markers ferritin (440 kDa) and aldolase (158 kDa) are indicated. (E) The dRAF peak fraction #14 from the Sephacryl S-300 column was centrifuged through a 10%–25% glycerol gradient, and collected fractions were examined for the presence of dRAF or PRC1 subunits by immunoblotting.

Figure 3.

dKDM2 and PRC1 control overlapping transcriptomes. (A) S2 cells were either mock-treated or incubated with dsRNAs directed against selective dRAF or PRC1 subunits dKDM2, dRING, PSC, PC, and PH. Whole-cell extracts were prepared and analyzed by Western immunoblotting. (B) Representation of 12 expression profiles in a three-dimensional transcriptome space, derived after PCA. RNA was isolated, labeled, and hybridized on Affymetrix Drosophila Genome 2 arrays. Expression indexes were calculated using the Robust Multichip Average (RMA) algorithm (Irizarry et al. 2003). The Minimum Covariance Determinant algorithm (Rousseeuw and van Driessen 1999) was used to remove genes that were expressed at very low levels. Next, we applied one-way ANOVA on each probe set to identify genes that changed significantly (P < 0.05) upon RNAi treatment. For these ∼5500 genes, we determined gene expression profiles by taking the ratios between average gene expression indexes obtained from specific RNAi- and mock-treated cells. Expression profiling of subunits of the trxG BAP and PBAP complexes, BRM, MOR, SNR1, OSA, Polybromo, BAP170, and OSA, as well as ISWI has been described (Moshkin et al. 2007). The expression profiles are shown as a projection on the first three PCs after varimax rotation. Each transcriptome represent significant gene targets after three independent biological replicate experiments. (C) Venn diagrams depicting the overlap and differences between the transcriptional targets of the dRAF-signature subunit dKDM2; dRING and PSC, shared by dRAF and PRC1; and the PRC1-selective subunits PC and PH. Numbers indicate significant target genes affected by depletion of the indicated factor(s).

Figure 4.

dkdm2 is an enhancer of Pc but a suppressor of trx and ash1. (A) Representative examples of homeotic transformations that were scored in the transheterozygous progeny of a series of crosses in which each of the dkdm2^KG04325 dkdm2^EY01336 or dkdm2^DG12810 mutant alleles or a wild-type (wt) allele were combined with either Pc¹ or Pc³ mutations. Sex combs are a row of dark, thick bristles, which normally only occur on the first pair of legs of male flies. In flies with defective Pc silencing, sex combs also appear on the second (L2–L1) or third (L3–L1) leg. Transformation of the fourth abdominal into the semblance of the posterior fifth (A4-A5) can be detected by the increased pigmentation of A4. Some flies display a defective wing development indicative of a wing-to-haltere (W–H) transformation. (B) Graphical representation of the frequencies of homeotic transformations in flies heterozygous for Pc¹ or Pc³ mutations but wild type for dkdm2 (black bars), or transheterozygous animals carrying Pc¹ or Pc³ combined with either dkdm2^KG04325 (yellow bars), dkdm2^EY01336 (red bars), or dkdm2^DG12810 (orange bars) mutations. The frequency of homeotic transformations is significantly higher in flies transheterozygous for dkdm2 and Pc mutations compared with flies heterozygous for only Pc mutations, as determined by Student’s t-test (P < 0.05). Two exceptions are indicated. (C) Representative examples of homeotic transformations observed in the transheterozygous progeny of homozygous trx¹ or ash1¹⁰ females crossed with either wild-type or dkdm2 mutant males. Abbreviations of the homeotic transformations: (H–W) haltere-to-wing; (A5–A4) transformation of the fifth abdominal segment into the semblance of the fourth. (D) Graphical representation of the frequencies of homeotic transformations in flies heterozygous for either Trx¹ or ash1¹⁰ (black bars) or transheterozygous animals carrying either trx¹ or ash1¹⁰ in combination with the indicated dkdm2 mutations. The frequency of homeotic transformations is significantly lower in flies transheterozygous for dkdm2 and either trx¹ or ash1¹⁰ compared with trx¹ or ash1¹⁰ heterozygotes wild type for dKDM2, as determined by Student’s t-test (P < 0.05).

Figure 5.

Depletion of endogenous dKDM2 causes increased histone H3K36me2 and loss of H2Aub. (A) Reduction of dKDM2 levels leads to a selective increase in H3K36me2. S2 cells were either mock-treated or incubated with dsRNAs directed against selective dRAF or PRC1 subunits dKDM2, dRING, PSC, PC, and PH (see Fig. 3A). Histones were purified by acid extraction and resolved by SDS-PAGE followed by Western blotting using the indicated antibodies directed against selective methyl marks. (B) dKDM2, dRING, and PSC are required for histone H2A ubiquitylation but do not affect H2Bub. The ubiquitylation status of histones purified from RNAi-treated S2 cells were analyzed by Western blotting using antibodies directed against either H2A, H2Aub, ubiquitin (ub), or H2B. The nonubiquitylated histones as well as H2Aub and H2Bub are indicated.

Figure 6.

dKDM2 stimulates selective H2A ubiquitylation by dRING-PSC. (A) We (co)expressed and purified dRAF and PRC1 core subunits as various multiprotein assemblies or by themselves using the baculovirus system. Purified factors and complexes include Flag-tagged dKDM2 (F-dKDM2)/dRING/PSC, PCC comprising F-PH/PSC/PC/dRING, F-dKDM2, F-dRING/PSC, F-PSC, F-dRING, and F-PH/PC. Following extract preparation, immunopurification, and elution under native conditions using Flag-peptides, proteins were resolved by SDS-PAGE and visualized by Coomassie staining. Asterisks indicate nonspecific background proteins. (B) dRAF core complex demethylates H3K36me2. Oligonucleosomes were incubated with either buffer control, dKDM2/dRING/PSC, or PCC. Approximately equimolar amounts (∼30 nM) of each protein complex were added, as judged by Coomassie staining (shown in A). Reaction mixtures were resolved by SDS-PAGE followed by Western blotting using antibodies directed against the indicated methyl marks or the core domain of histone H3. The bottom panel shows the core histones present in the reaction visualized by Coomassie staining. (C) dKDM2 alone is sufficient to demethylate H3K36me2. Oligonucleosomes were incubated with either a buffer control or increasing amounts of dKDM2 (∼10, 20, 40, or 80 nM). Analysis was as described above. (D) dRAF core complex (dKDM2/dRING/PSC) ubiquitylates histone H2A, but not H2B. Oligonucleosomes or a buffer control were incubated in the presence of approximately equimolar amounts (∼30 nM) of either dKDM2/dRING/PSC or PCC. Reaction mixtures were resolved by SDS-PAGE followed by Western blotting using antibodies directed against H2A, H2Aub, ubiquitin (ub), and H2B. Note that the amount of histones loaded did not allow detection of endogenous H2Aub or H2Bub. (E) dKDM2 stimulates histone H2A ubiquitylation by dRING and dRING/PSC. Oligonucleosomes were incubated with increasing amounts of dRING, PSC, or dKDM2 (∼20, 40, or 80 nM), dKDM2 was also titrated in the presence of ∼40 nM (++) PSC, and dKDM2 and/or PSC were titrated in reactions containing ∼20 nM (+) dRING. Analysis was as described above. (F) PC/PH does not stimulate H2A ubiquitylation by dRING/PSC. Oligonucleosomes were incubated in the presence of either 20 (+) or 40 nM (++) dRING/PSC and increasing amounts of PC-PH (∼10, 20, 40, or 80 nM). As a control, dKDM2 (∼10, 20, 40, or 80 nM) was added to reactions containing 20 nM (+) dRING/PSC. Analysis was as described above.

Figure 7.

H2A ubiquitylation and H3K36me2 demethylation can occur concomitantly but are not interdependent. (A) H2A ubiquitylation and H3K36me2 demethylation can occur concomitantly in the same reaction. Oligonucleosomes or a buffer control were incubated in the presence of dRING/PSC (∼30 nM), dKDM2 (∼30 nM), or with increasing amounts of dKDM2 (∼10, 20, 40, or 80 nM) in the presence of ∼30 nM dRING/PSC. Reaction products were resolved by SDS-PAGE followed by Western blotting using the indicated antibodies. (B) H3K36me2 and H2Aub are mutually exclusive nucleosomal marks in bulk chromatin. Approximately 2 mg of purified endogenous mononucleosomes were immunopurified using Protein A Sepharose beads coated with antibodies directed against H3K36me2. Input, bound, and unbound FT fractions were collected, resolved by 18% SDS-PAGE, and analyzed by Western blotting using antibodies directed against H3K36me2, H2Aub, and H2A. (C) Stimulation of H2A ubiquitylation by dKDM2 is independent of H3K36me2. Monononucleosomes (left panel) or H3K36me2-depleted mononucleosomes (right panel) were incubated in the presence of either ∼30 nM dRING/ PSC alone or together with ∼40 or ∼80 nM dKDM2 or ∼80 nM dKDM2. Analysis was as described above. (D) Wild-type Flag-tagged dKDM2 (F-dKDM2) or mutants F-dKDM2(T241A) and F-dKDM2(H244A) were expressed in Sf9 cells using the baculovirus system. Following extract preparation, immunopurification, and elution under native conditions using Flag-peptides, proteins were resolved by SDS-PAGE and visualized by Coomassie staining. (E) H3K36me2 demethylation defective dKDM2 mutants remain fully able to stimulate H2A ubiquitylation by dRING/PSC. Oligonucleosomes were incubated with ∼20, 40, or 80 nM of dKDM2, dKDM2 (T241A), or dKDM2 (H244A), respectively. Reaction products were resolved by SDS-PAGE followed by Western blotting.