Role of remodeling and spacing factor 1 in histone H2A ubiquitination-mediated gene silencing

Abstract

Posttranslational histone modifications play important roles in regulating chromatin-based nuclear processes. Histone H2AK119 ubiquitination (H2Aub) is a prevalent modification and has been primarily linked to gene silencing. However, the underlying mechanism remains largely obscure. Here we report the identification of RSF1 (remodeling and spacing factor 1), a subunit of the RSF complex, as a H2Aub binding protein, which mediates the gene-silencing function of this histone modification. RSF1 associates specifically with H2Aub, but not H2Bub nucleosomes, through a previously uncharacterized and obligatory region designated as ubiquitinated H2A binding domain. In human and mouse cells, genes regulated by RSF1 overlap significantly with those controlled by RNF2/Ring1B, the subunit of Polycomb repressive complex 1 (PRC1) which catalyzes the ubiquitination of H2AK119. About 82% of H2Aub-enriched genes, including the classic PRC1 target Hox genes, are bound by RSF1 around their transcription start sites. Depletion of H2Aub levels by Ring1B knockout results in a significant reduction of RSF1 binding. In contrast, RSF1 knockout does not affect RNF2/Ring1B or H2Aub levels but leads to derepression of H2Aub target genes, accompanied by changes in H2Aub chromatin organization and release of linker histone H1. The action of RSF1 in H2Aub-mediated gene silencing is further demonstrated by chromatin-based in vitro transcription. Finally, RSF1 and Ring1 act cooperatively to regulate mesodermal cell specification and gastrulation during Xenopus early embryonic development. Taken together, these data identify RSF1 as a H2Aub reader that contributes to H2Aub-mediated gene silencing by maintaining a stable nucleosome pattern at promoter regions.

Keywords: H2A ubiquitination; H2Aub binding protein; PRC1; RSF1; transcription repression.

Fig. 1.

RSF1 preferentially associates with H2Aub-enriched nucleosomes in HeLa cells expressing HA-ubiquitin. (A) Silver staining of a protein gel containing equal amounts of H2Aub-enriched or -depleted mononucleosomes reveals polypeptides enriched specifically within each group of nucleosomes (arrowheads). Different staining times were used for core histones and the rest of the proteins. (B) Anti-HA immunoblot of a protein gel containing an aliquot of the above samples shows that a single protein corresponding to the H2Aub position containing HA-ubiquitin (arrowhead). (C) RSF1 is enriched in the H2Aub-enriched nucleosomes in the SILAC assays, shown by log₂-fold enrichment of RSF1 abundance, quantified by the intensity of a RSF1 unique peptide (Fig. S1), in H2Aub-enriched (+ub) or -depleted (−ub) nucleosomes. H and L represent heavy and light lysine labeling, respectively. (D) Immunoblot analyses of proteins contained in the flow-through (Ft) or the eluate (E) of anti-HA IP of H2Aub-containing nucleosomes show that RSF1 is enriched in the H2Aub nucleosomes, along with SNF2H, another component of the RSF complex. In, input. Antibodies used are indicated at the left.

Fig. S1.

Mass spectrometry quantification of a RSF1 unique peptide in H2Aub-enriched and -depleted nucleosomes. (A) Spectrum of a RSF1 fragment (DKDGLmYWYQLDQDHNVR) in H2Aub-enriched and -depleted nucleosomes. In these experiments, H2Aub-enriched nucleosomes were labeled with heavy lysine and H2Aub-depleted nucleosomes were obtained from cells cultured in light medium. (B) Quantification of the labeled RSF1 fragment intensity from A. Both the intensity and log₂-normalized intensity are shown. (C) Spectrum for RSF1 Fragment (DK+6DGLmYWYQLDQDHNVR) in H2Aub-enriched and -depleted nucleosomes. In these experiments, H2Aub-depleted nucleosomes were labeled with heavy lysine and H2Aub-enriched nucleosomes were obtained from cells cultured in light medium. (D) Quantification of the labeled RSF1 fragment intensity from C. Both the intensity and log₂-normalized intensity are shown.

Fig. 2.

RSF1 binds H2Aub nucleosomes through a previously uncharacterized UAB domain. (A) Schematic representation of the RSF1 domain organization and fragments used in nucleosome pull-down assay. (B) The full-length and fragment 7 (amino acids 770–807) of RSF1 bind to H2Aub nucleosomes. (Top) CBB-stained protein gel containing purified full-length RSF1 and its serial fragments. F1: 1–97 aa; F2, 98–182 aa; F3, 183–397 aa; F4, 398–608 aa; F5, 609–690 aa; F6, 691–785 aa; F7, 770–807 aa; F8, 808–890 aa; F9, 891–941 aa; F10, 942–1,441 aa. (Middle) Pull-down assays show that only the full-length RSF1 and fragment 7 (designated as the UAB domain) bind to H2Aub nucleosomes. (Bottom) Treatment with the H2Aub-specific deubiquitinase USP16 abolishes binding of RSF1 and fragment 7 to the nucleosomes. (C) The UAB domain specifically pulls down reconstituted nucleosomes subjected to in vitro ubiquitination by PRC1 (lane 2), but not unmodified nucleosomes (lane 1). (D) The UAB domain specifically pulls down H2Aub (lane 1), but not H2Bub (lane 2) nucleosomes. (E) The UAB domain is required for RSF1 to interact with H2Aub nucleosomes in 293T cells. Cells were transfected with an empty vector or vectors expressing Flag-RSF1 or mutated Flag-RSF1 (MT) in which the UAB domain was in-frame deleted. Mononucleosomes were prepared and subjected to anti-Flag IP followed by immunoblots. (F) Metaplot of Flag-RSF1 in control (blue line) and Ring1B KO (red line) mouse ESCs. For all data, the minimum value is set to 0. The signal plot is normalized by subtracting IgG ChIP-seq signal from the same cell lines. Transcription start site (TSS) plus 10-kb upstream and transcription termination site (TTS) plus 10-kb downstream are shown. (G) Venn diagram shows the overlap between RSF1 bound genes and H2Aub marked genes in mouse ESCs. About 82% of H2Aub sites are bound by RSF1.

Fig. S2.

The UAB domain of RSF1 binds to H2Aub nucleosomes specifically. (A) Immunoblots of H2Aub-containing nucleosomes before (lane 1) and after (lane 2) treatment of USP16, a histone H2A specific deubiquitinase (25). Antibodies used are indicated on the left side of the panel. (B) The UAB domain pull-down assays with histone octamer or individual histones as input. The Top panel is a CBB stained gel image containing histone octamer (lane 1, prepared from H2Aub-containing nucleosomes), individual histones (lanes 2–5, purified from E. coli) and an unrelated protein Im7 (lane 6, purified from E. coli). The Bottom three panels show the pull-down proteins, as analyzed by immunoblots (third panel) and CBB staining (second and fourth panels).

Fig. 3.

RSF1 and RNF2/Ring1B modulate the expression of a large cohort of common targets in human and mouse cells. (A) Venn diagram shows the overlap of the genes with significant changes in expression in RNF2 or RSF1 KD HeLa cells. (B) Venn diagram shows the overlap of the genes with significant changes in expression in Ring1B or RSF1 KO mouse ESCs. (C) Representative images of the HoxB8 gene locus showing data obtained from the RNA-seq (Top two panels), RSF1 ChIP-seq (third panel), and H2Aub ChIP-seq (fourth panel) assays in mouse ESCs containing RSF1-Flag-HA knockin. Parallel H2A and Flag ChIP in parental ESCs were shown as controls. Gene diagrams are shown (Bottom). (D) Heatmaps shows fold-changes of the expression of the Hox genes in RSF1 KO mouse ESCs compared with control wild-type ESCs from RNA-seq data (first column), tag counts for H2Aub ChIP-seq data (second column), and RSF1 ChIP-seq data (third column) in control wild-type ESCs within 5 kb of gene TSS. The tag counts are RPM-normalized and background-subtracted. (E) Heatmaps shows fold-changes in the expression of all genes in RSF1 KO mouse ESCs compared with wild-type control ESCs from RNA-seq data (first column), tag counts for H2Aub signal (second column), and RSF1 signal (third column) in control wild-type ESCs within 5 kb of gene TSS. Genes are ranked by their expression levels and only the top and bottom 1,000 genes are shown. The tag counts are RPM-normalized and background subtracted.

Fig. S3.

RSF1 binding overlaps with RNF2/Ring1B and H2Aub binding. (A) Metaplot of Flag-RSF1 in control (blue line) and Ring1B KO (red line) mouse ESCs. For all data, the minimum value is set to 0. The signal plot is normalized by subtracting Flag ChIP-seq signal from the parental ESC lines. TSS plus 10-kb upstream and TTS plus 10-kb downstream are shown. (B) Venn diagram showing the overlap of binding of RSF1 (blue), Ring1B (teal), and PRC2 (pink) in mouse ESCs. (C) Representative images of HoxB7 and HoxC6 genes with aligned tags from RNA-seq (Top two panels), RSF1 ChIP-seq (third panel), and H2Aub ChIP-seq (fourth panels) in mouse ESCs. Parallel H2A and Flag ChIP were shown as controls. Gene diagrams are shown at the Bottom.

Fig. S4.

Significant correspondence between RSF1 and RNF2/Ring1B regulated genes. (A) Immunoblot analyses of HeLa cells with KD of RSF1 or RNF2 confirmed the reduction of corresponding target proteins. Antibodies used are indicated on the right of the panels. (B) Waterfall plot showing gene expression changes in RNF2 KD and RSF1 KD HeLa cells. The majority of these genes show changes in the same direction. (C) Immunoblot analyses of control, RSF1, RNF2, and SNF2H KD HeLa cells confirmed the reduction of corresponding target proteins. Antibodies used are indicated on the right of the panels. (D) RT-qPCR analysis of the expression of representative genes in control HeLa cells and in HeLa cells with KD of RSF1, RNF2, and SNF2H. (E) Immunoblots of wild-type, RSF1 KO, or Ring1B KO mouse ESCs confirmed the reduction of corresponding target proteins. Antibodies used are indicated on the right of the panels. (F) Waterfall plot showing gene expression changes in Ring1B KO and in RSF1 KO mouse ESCs. The majority of these genes show changes in the same direction. (G) Venn diagram shows the overlap of H2Aub bound genes with significant changes in expression in Ring1B or RSF1 KO mouse ESCs.

Fig. 4.

RSF1 represses transcription activation from H2Aub nucleosome-containing chromatin in vitro. (A) Supercoiling assay of reconstituted chromatin template. The purified supercoiled plasmids (lane 1) were relaxed to closed circular templates using topoisomerase I (lane 2). After protease digestion and phenol-chloroform extraction, the relaxed plasmids (lane 2) were mixed with histone/ACF/Nap1 in the presence of topoisomerase I to reconstitute chromatin template that restore DNA supercoiling. DNA was then purified and changes in linking number were demonstrated (lanes 3 and 4). (B) In vitro transcription assay on naked DNA and on chromatin templates reconstituted with H2A or semisynthetic H2Aub. GAL4-VP16 and p300/AcCoA activate transcription. RSF1 inhibits transcription activation on H2Aub chromatin template, but has no effect on transcription activation on H2A chromatin.

Fig. 5.

RSF1 is required for maintaining the pattern and stability of H2Aub nucleosome arrays at gene promoters. (A) Boxplot of the distance between nucleosomes around TSS in control and two RSF1 KO mouse ESC lines. RSF1 KO affects the organization of H2Aub chromatin around TSS. C, control; 1 and 2, two RSF1 knockout mouse ESCs. (B) Intensity of H2Aub nucleosomes in control (black line) and RSF1 KO (red line) mouse ESCs. (C) H2Aub nucleosomes isolated from control or RSF1 KO HA-ubiquitin-expressing HeLa cells were subjected to increasing concentrations of salt wash. Immunoblot was used to detect the released histones. Gly, glycine elution of the anti-HA beads after 1.0 M salt treatment; HF29, parental HA-ubiquitin overexpressing HeLa cells. Antibodies used are indicated on the left of the panels. (D) RSF1 is required for the binding of linker histone H1 to H2Aub nucleosomes. Immunoblots of H2Aub nucleosomes isolated from control and two independent RSF1 KO HeLa cell lines overexpressing HA-ubiquitin. Antibodies used are indicated on the left of the panels. (E) A proposed model of RSF1 in H2Aub target gene silencing. RSF1 binds to H2Aub nucleosomes to establish and maintain the stable H2Aub nucleosome pattern at promoter regions. The stable nucleosome array leads to a chromatin architecture which is refractory to further remodeling required for H2Aub target gene activation. When RSF1 is knocked out, H2Aub nucleosome patterns are disturbed and nucleosomes become less stable despite the presence of H2Aub. These H2Aub nucleosomes are subjected to chromatin remodeling for gene activation.

Fig. S5.

RSF1 is required for maintaining stable promoter nucleosome arrays. (A) Binding intensity of H2Aub nucleosomes (blue line) and all nucleosomes (black line) in mouse ESCs. (B) Binding intensity of all nucleosomes in control (black line) and RSF1 KO (red line) mouse ESCs.

Fig. 6.

RSF1 collaborates with Ring1 to regulate mesodermal specification during early Xenopus embryogenesis. (A) Both RSF1 and Ring1 regulate Xenopus gastrulation. Injection of RSF1-MO (20 ng) and Ring1-MO (50 ng) induced similar gastrulation defects in Xenopus tadpoles, with embryos displaying short axis and open blastopore. In contrast, RNF2-MO induced a milder phenotype of bent axis and head defects. (B) RSF1-MO and Ring1-MO act cooperatively to induce gastrulation defects. A combination of low doses of Ring1-MO and RSF1-MO resulted in an embryonic phenotype that mimicked or surpassed that with the high doses of individual MOs. (C) At late gastrula stages, RSF1-MO and Ring1-MO cause delay in blastopore closure. Combination of low doses of RSF1-MO and Ring1-MO induced similar gastrulation defects as that when higher doses of individual MOs were used. RNF2-MO did not have obvious effects on blastopore closure. (D) In situ hybridization demonstrated that KD of RSF1 or Ring1 reduced mesodermal markers in a similar fashion. The pan-mesodermal gene Brachyury (Bra), the dorsal mesodermal marker Chordin (Chd), and the ventrolateral mesodermal marker Wnt8 were all reduced upon RSF1 or Ring1 KD. KD of RNF2 did not alter expression of these mesodermal markers. The embryos were injected with the MOs and the lineage tracer encoding nuclear β-Gal into the marginal zone of one cell at the two-cell stage, and the injected region was revealed by staining with the β-Gal substrate Red-Gal.

Fig. S6.

RSF1, Ring1, and RNF2, are expressed highly and widely during Xenopus embryonic development. Semiquantitative RT-PCR analyses of RSF1, Ring1, and RNF2 expression in Xenopus larvis at different stages of embryo development.

Fig. S7.

The UAB motif harbors features for interaction with H2AK119ub nucleosomes. (A) Sequence alignment of the UAB motif across species. Conserved amino acids are heighted in black or gray. The arginine residues likely involved in binding to the nucleosome acidic patch are marked with asterisks. The two clusters of conserved aliphatic residues potentially binding to the ubiquitin hydrophobic patch are indicated by bracket. (B) A structural modeling of the UAB motif complexed with H2AK119ub nucleosome. The UAB motif is shown in pink. The electrostatic potential of nucleosome core octamer is shown, where blue stands for positive charge and red for negative charge. The side chains of the residues in the ubiquitin hydrophobic patch are shown in spheres, and H2A/H2B ubiquitination sites are indicated with arrows. The distance of the acidic patch to H2BK120 is shorter than that to H2AK119, indicating that the spacing between arginine fingers and aliphatic clusters is important for the preferred binding to H2AK119ub nucleosomes.