Phf19 links methylated Lys36 of histone H3 to regulation of Polycomb activity

Abstract

Polycomb-group proteins are transcriptional repressors with essential roles in embryonic development. Polycomb repressive complex 2 (PRC2) contains the methyltransferase activity for Lys27. However, the role of other histone modifications in regulating PRC2 activity is just beginning to be understood. Here we show that direct recognition of methylated histone H3 Lys36 (H3K36me), a mark associated with activation, by the PRC2 subunit Phf19 is required for the full enzymatic activity of the PRC2 complex. Using NMR spectroscopy, we provide structural evidence for this interaction. Furthermore, we show that Phf19 binds to a subset of PRC2 targets in mouse embryonic stem cells and that this is required for their repression and for H3K27me3 deposition. These findings show that the interaction of Phf19 with H3K36me2 and H3K36me3 is essential for PRC2 complex activity and for proper regulation of gene repression in embryonic stem cells.

Figure 1

Phf19 is a component of the PRC2 complex. (a) SDS-PAGE followed by silver staining of TAP-purified PHF19 (TAP-PHF19) in 293T cells. TAP−, TAP-empty. (b) Summary of peptides identified by MS in eluates from TAP-PHF19 purification. emPAI, experimentally modified protein abundance. (c) Western blot analysis of eluates from TAP-PHF19 purification confirming the specific association with the PRC2. Note that the PRC1 subunit RING1b does not interacts with PHF19. (d) Endogenous co-immunoprecipitation (IP) in mES cells of Phf19 with PRC2 components. Note that Jarid2 is not present in Phf19–PRC2 complexes. (e) Size-exclusion chromatography of 293T nuclear extracts followed by western blotting. Note that PHF19 and JARID2 co-migrate with components of the PRC2 complex but elute at different size fractions—around 790 kDa and 1.4 mDa, respectively. (f) Schematic representation of the domain architecture of human PHF19, depicting the Tudor domain, two plant homeodomain (PHD) fingers, the extended homology (EH) domain, and a chromo-like domain. (g) The chromo-like domain is necessary and sufficient to co-immunoprecipitate the endogenous PRC2 core component SUZ12. (h) Predicted secondary structures of the chromo-like domains compared to the first chromodomain of the HP1 proteins.

Figure 2

Binding of Phf19-Tudor to methylated H3K36. (a) Histone peptide array showing specific binding of GST-labeled Phf19-Tudor to H3K36me2 and H3K36me3 peptides. aa, amino acids. (b) Histone peptide pull-down assay using recombinant Phf19, Phf1, MTF2 and Pcl domains with methylated histone peptides. Phf19-Tudor, Phf1-Tudor and MTF2-Tudor all bind to H3K36me2 and H3K36me3 peptides. D. melanogaster Pcl-Tudor, which lacks an otherwise conserved aromatic amino acid, requires the addition of the PHD1 for binding.

Figure 3

NMR-based structural analysis of the complex between Phf19-Tudor and an H3 peptide methylated at Lys36. (a) NMR 2D ¹⁵N-¹H correlation spectrum showing the chemical shift changes of the H_N resonances of Phf19-Tudor upon addition of the 11-mer H3K36me3 peptide (31-ATGGVKme3KPHRY-41; from blue to red). The protein concentration was 0.2 mM. Amino acid single-letter code is used for brevity. (b) Plots of the changes in chemical shifts induced upon complex formation as a function of the concentration of the H3K36me3 wild-type (amino acids 31–41) and mutant peptides for the amide resonance of Asp76. The values of K_d are calculated as average values for 8 to 14 well-resolved H_N resonances of the complexes. (c) Ribbon representation of the lowest-energy structure of the Phf19-Tudor complex with the H3K36me3 peptide. Phf19-Tudor side chains in contact with the H3K36me3 peptide are shown as sticks. Phf19-Tudor: C, violet; O, red; N, blue. H3K36me3 peptide: C, green; O, red; N, blue. Amino acids of Phf19-Tudor are labeled in black and those of the H3K36me3 peptide in green (PDB 4BD3). (d) Surface representation of the Phf19-Tudor domain colored by electrostatic charge. White, neutral; blue, positive; red, negative. Side chains of the protein in contact with the peptide are shown in violet. Only residues 35–41 are shown for the histone H3K36me3 peptide. Phf19-Tudor: C, violet; O, red; N, blue; H3K36me3 peptide: C, green; O, red; N, blue.

Figure 4

Phf19 is in integral part of the PRC2 complex in mES cells. (a) Histogram showing the distribution of Phf19 ChIP-seq peaks relative to the TSS. Almost all (93%) of Phf19 peaks were found within ±10 kb of a known TSS. (b) Venn diagrams showing the overlap of Phf19, Suz12 and H3K27me3 target genes in mES cells. (c) Suz12, Phf19 and H3K27me3 occupancy in Phf19 binding sites are plotted as the average profile of ChIP-seq reads (read density per base pair) around the summit of Phf19 peaks. (d) Gene ontology analysis of Phf19 target genes in mES cells. (e–g) ChIP-qPCR assays using an antibody anti-Phf19 and IgG as negative control. Results are presented as a percentage of the input material. Values represent the average and s.d. of three independent experiments. (e) Validation of Phf19 target genes. Phf19 knockdown cells showed reduced binding, demonstrating the specificity of the antibody. Klf4 and Tbx3 served as negative controls. (f) ChIP-qPCR analysis shows a reduced Phf19 binding to target genes after differentiation induction by all-trans retinoic acid (RA) supplemented at 1 × 10⁻⁶ M for 72 h. (g) ChIP-qPCR analysis in cells deleted for Eed, a core subunit of the PRC2, indicates that Phf19 binding is dependent on a functional PRC2. n = 3 in e–g.

Figure 5

Phf19 is required for PRC2 binding to target genes. (a) Changes in H3K27me3 and Suz12 levels evaluated by ChIP-seq analysis in Phf19, H3K27me3 and Suz12 co-targets following Phf19 knockdown (KD). (b) ChIP-qPCR analysis shows that binding of the PRC2 component Suz12 is reduced in Phf19 knockdown cells. Data represent average and s.d. of three independent experiments. (c) Western blot analysis of histone modifications in Phf19 knockdown cells shows a global reduction of H3K27me3 and a concomitant increase in H3K27me1 and H3K27ac. Two different Phf19-depleted cell lines were analyzed (shPhf19–3 and shPhf19–4) (left). Expression levels of Phf19 were determined by RT-qPCR in the wild type and in the cell lines depleted of shPhf19–3 and shPhf19–4. Data represent the average and s.d. of three independent experiments (right).

Figure 6

Role of Phf19 in pluripotency. (a) Microarray expression analysis of Phf19 knockdown cells. In cells maintained under pluripotency conditions (−RA), 734 genes were upregulated and 861 were downregulated (fold change >1.5 and adjusted, P < 0.01). After treatment with all-trans RA (10⁻⁶ M) for 72 h, 441 genes were upregulated and 965 were downregulated (fold change >1.5 and adjusted, P < 0.01). (b) Expression levels of pluripotency genes and several Phf19 target genes were determined by RT-qPCR under the experimental conditions described in a. Two different Phf19-depleted cell lines were analyzed (shPhf19–3 and shPhf19–4). Expression levels were normalized to the those of Rpo housekeeping gene. Data represent the average of three independent experiments (n = 3). (c) Control ES cells (shRd) or cells depleted of Phf19 (shPhf19) were injected into Swiss Nude mice at two flanks (1 × 10⁶ cells per flank). Two weeks after injection, mice were killed and tumors were collected. The tumor weights for both sets of teratomas are shown. The differences in size were statistically significant (n = 8, Student’s t-test, P < 0.05).

Figure 7

Molecular mechanism of Phf19-mediated gene repression. (a) ChIP-qPCR analyses were performed in mES cells maintained under pluripotency conditions (−RA) or after treatment with all-trans RA (+RA), supplemented at 1 × 10⁻⁶ M for 5 d. Data represent the average and s.d. of three independent experiments (n = 3). (b) ChIP-qPCR experiments in control and Phf19 knockdown cells, and in knockdown cells stably transfected with the Flag-tagged human wild-type Phf19 (WT.hPhf19) or with Phf19 mutated (mut.hPhf19) in the Tudor domain (mutation W50A, which renders it unable to bind H3K36me3). Amplification of the Oct4 promoter was used as a negative control. Data represent the average and s.d. from triplicate experiments. (c) Expression levels of Phf19 target genes were determined by RT-qPCR under the experimental conditions described in b. Data represent the average and s.d. from triplicates. (d) Phf19 facilitates and/or stabilizes binding of the PRC2 complex at H3K36-methylated promoters. Binding of Phf19-Tudor to H3K36me2 and H3K36me3 mediates the recruitment of Polycomb proteins and of the H3K36 demethylase KDM2b, which in turn causes gene silencing.