Demethylation of H3K27 Regulates Polycomb Recruitment and H2A Ubiquitination

See allHide authors and affiliations

Science  19 Oct 2007:
Vol. 318, Issue 5849, pp. 447-450
DOI: 10.1126/science.1149042


Methylation of histone H3 lysine 27 (H3K27) is a posttranslational modification that is highly correlated with genomic silencing. Here we show that human UTX, a member of the Jumonji C family of proteins, is a di- and trimethyl H3K27 demethylase. UTX occupies the promoters of HOX gene clusters and regulates their transcriptional output by modulating the recruitment of polycomb repressive complex 1 and the monoubiquitination of histone H2A. Moreover, UTX associates with mixed-lineage leukemia (MLL) 2/3 complexes, and during retinoic acid signaling events, the recruitment of the UTX complex to HOX genes results in H3K27 demethylation and a concomitant methylation of H3K4. Our results suggest a concerted mechanism for transcriptional activation in which cycles of H3K4 methylation by MLL2/3 are linked with the demethylation of H3K27 through UTX.

The methylation of lysine residues on histones is often associated with either the activation [methylation of histone H3 lysine 4 (H3K4), H3K36, and H3K79] or repression (methylation of H3K9, H3K27, and H4K20) of transcription (1, 2). Methylation offers an additional level of control by permitting single, double, and triple modification of the same lysine residues (1, 2), resulting in differential regulation of many cellular processes (13). The Methylation of H3K27 is implicated in X chromosome inactivation, imprinting, stem cell maintenance, circadian rhythms, and cancer (46) and is carried out by Enhancer of zeste homolog 2 methyltransferase, a component of the mammalian polycomb repressive complexes (PRCs), including PRC2 (79). Tri- and dimethyl H3K27 are enriched on inactive X chromosomes, as well as at promoter regions of inactive genes (2, 10, 11). Here, we describe the characterization of UTX, a Jumonji C (JmjC)–domain-containing protein capable of demethylating tri- and dimethyl H3K27.

We affinity-purified recombinant human UTX using a baculovirus expression system (Fig. 1A and fig. S1, A and B) and assessed its activity in demethylation assays. UTX has a specific demethylation activity toward tri- and dimethyl H3K27 without affecting methylation on H3K4, H3K9, H3K36, H3K79, and H4K20 (Fig. 1B and fig. S1C). Monomethylated H3K27 was not affected to the same extent when we used comparable concentrations of UTX that diminished di- and trimethylated species (Fig. 1B). The demethylation activity of UTX toward di- and trimethyl H3K27 was confirmed by mass spectrometric analysis of methylated H3K27 peptides corresponding to the H3 tail (fig. S2). Point mutations (H1146A and E1148A) in the catalytic JmjC-domain of UTX abrogated enzymatic activity, substantiating its role as an H3K27 demethylase (Fig. 1C and fig. S1B). Ectopic expression of wild-type UTX, but not its catalytic mutant, resulted in a global decrease in di- and trimethyl H3K27 levels (fig. S3A). Although the tetratricopeptide repeats (TPRs) present in UTX were not essential for the demethylation activity of UTX, the enzyme without TPRs displayed reduced activity (fig. S3, B and C). As is characteristic for the JmjC class of enzymes (1222), demethylation activity by UTX required the addition of Fe(II) and ascorbate (Fig. 1D). However, similar to the case of JARID1d (a trimethyl H3K4 demethylase), we did not find a requirement for exogenously added α-ketoglutarate for UTX activity (Fig. 1D).

Fig. 1.

UTX demethylates tri- and dimethyl H3K27 and regulates expression of several HOX genes. (A) Diagrammatic representation of UTX containing six TPRs and one JmjC. (B) Demethylation assay using recombinant (r.) UTX. (C) Comparison of demethylase activities of recombinant UTX and its catalytic mutant [r.UTX (HE/AA)]. (D) Demethylation activities of recombinant UTX (∼800 ng) in the presence (+) or absence (–) of cofactors. α-KG, α-ketoglutarate. For data in (B) to (D), histones were mixed with recombinant proteins, and reaction mixtures were subjected to SDS–polyacrylamide gel electrophoresis, followed by Western blot analysis using antibodies to various methyl marks. 1x, ∼400 ng of r.UTX or its mutant; 3m, trimethyl; 2m, dimethyl; 1m, monomethyl. (E and F) Analysis of mRNA levels of HOXA (E) and HOXC (F) clusters by qRT-PCR after treatment of HEK293 cells with siRNAs (siLuciferase or siUTX). Data are presented as the mean ± SEM (error bars) (n = 3). *(P < 0.05) and **(P < 0.01) indicate statistically significant changes (student's t test).

Recent genome-wide mapping of polycomb target genes revealed that polycomb group proteins and trimethyl H3K27 are enriched in HOXA-D loci (23), whose activity is required for embryonic development and when misregulated may lead to carcinogenesis. To assess the demethylation properties of UTX in vivo, we analyzed expression levels of HOXA and HOXC cluster genes using quantitative reverse transcription polymerase chain reaction (qRT-PCR) after the depletion of UTX with small interfering RNAs (siRNAs) (fig. S4, A and B). Immunostaining analysis using UTX antibodies revealed nuclear staining for endogenous UTX (fig. S5, A and B). Quantitative analysis of mRNA levels indicated that a number of genes in both HOXA and HOXC clusters are repressed after the knockdown of UTX (Fig. 1, E and F). Specifically, HOXA 6, 7, and 13 as well as HOXC4, 5, 6, 8, and 12 displayed greater than 50% repression after UTX depletion.

To demonstrate that UTX mediates such transcriptional effects on HOX genes directly, we examined UTX occupancy on the promoter as well as the body of the HOXA13 and HOXC4 genes with the use of a quantitative chromatin immunoprecipitation (qChIP) assay. We found a nearly 20-fold enrichment of UTX on the HOXA13 and HOXC4 promoters, but only two- to five-fold enrichment on the coding region (Fig. 2, A to D). The depletion of UTX (ChIP analysis revealed about a 50% decrease of UTX occupancy at the HOXA13 and HOXC4 promoters; Fig. 2, I and J) resulted in increased levels of di- and trimethyl H3K27 at the promoters of HOXA13 and HOXC4 but not at the 3′ end of these genes, consistent with the role of UTX as a H3K27 demethylase at the promoter of these genes (Fig. 2, E to H). We also examined the recruitment of PRC1 to HOXA13 and HOXC4, which is targeted to methylated H3K27 sites and possess H2A ubiquitin ligase activity (24, 25). The depletion of UTX leads to an increased occupancy of the Ring finger components of the PRC1 complex (Bmi1 and Ring1A proteins) and a concomitant enhancement of monoubiquitinated H2A at HOXA13 and HOXC4 (Fig. 2, I and J). In contrast, there was no change in the levels of histone H3 or in other histone modifications examined after UTX knockdown (Fig. 2, I and J). Taken together, these results indicate that the demethylation of H3K27 by UTX regulates the occupancy of the PRC1 complex at UTX target genes, leading to the modulation of the transcriptional output.

Fig. 2.

UTX regulates the recruitment of PRC1 and ubiquitination of H2A. (A and B) Diagrammatic representation of the promoter and body of the HOXA13 (A) and HOXC4 (B) genes. a and b indicate regions amplified by PCR. (C and D) Analysis of UTX levels at HOXA13 (C) and HOXC4 (D) genes by a qChIP assay. IgG, immunoglobulin G. (E and F) Analysis of trimethyl H3K27 levels at HOXA13 (E) and HOXC4 (F) genes by qChIP. (G and H) Analysis of dimethyl H3K27 levels at HOXA13 (G) and HOXC4 (H) genes by qChIP. (I and J) Analysis of promoter occupancy of UTX, PRC1 subunits (Bmi1 and Ring1A), and ubiquitinated H2A (ubH2A) at region a of HOXA13 (I) and HOXC4 (J) genes. The relative occupancy represents the fold change in percent input over the control (28). The representative values of percent input set as 1 for HOXA13 and HOXC4 are as follows: UTX, 0.12/0.17; 3mH3K27, 1.2/0.19; 2mH3K27, 0.04/0.03; Bmi1, 0.01/0.02; RING1A, 0.01/0.03; ubH2A, 0.01/0.01; 2mH3K79, 0.43/2.24; and H3, 0.8/0.2. For the ChIP data in (E) to (J), HEK293 cells were treated with siRNAs (siLuciferase or siUTX). Data are presented as the mean ± SEM (error bars) (n ≥ 3). *(P < 0.05) and **(P < 0.01) indicate statistically significant changes (student's t test).

To assess the recently described UTX interaction with mixed-lineage leukemia (MLL)–containing complexes (26, 27), we developed a human embryonic kidney (HEK) 293–derived stable cell line expressing Flag-UTX. Analysis of Flag-UTX affinity eluate by mass spectrometry revealed the association of UTX with multiple components of MLL2/3-containig complexes (Fig. 3A and table S1). The Flag-UTX complex displayed H3K27 demethylase as well as H3K4 methyltransferase activities (Fig. 3, B to D). Analysis of Flag-UTX affinity eluate by gel filtration further demonstrated the association of Flag-UTX with a ∼2-megadalton (MD) MLL2/3-containing complex (Fig. 3E). To extend this analysis, we developed stable cell lines expressing Flag–WD repeat domain 5 (WDR5), a core component of the MLL-containing complexes. WDR5-containing complexes specifically associate with UTX (fig. S6, A and B). Moreover, the UTX/WDR5-containing complexes display H3K4 methyltransferase activity toward recombinant nucleosomes (fig. S6C). Analysis of WDR5-containing complexes by Superose 6 gel filtration revealed that a fraction of WDR5 associates with UTX in a complex of ∼2 MD displaying H3K4 methyltransfarse activity (fig. S6D).

Fig. 3.

UTX associates with MLL2/3-containing complexes. (A) Analysis of UTX-containing complexes isolated from nuclear extract by silver staining. Nuclear extracts were isolated from HEK293 cells expressing either FLAG-tagged green fluorescent protein (f-GFP) or FLAG-tagged UTX (f-UTX). After affinity chromatography using anti-FLAG resin, eluates of FLAG-UTX and FLAG-GFP (a negative control) were analyzed. GenBank accession numbers for UTX-associated proteins and numbers of peptides identified by mass spectrometric analysis are described in table S1. MLL2 was described as MLL4 in a recent report (26). The asterisks and arrowheads denote breakdowns and nonspecific polypeptides, respectively. IP, immunoprecipitation. (B) Demethylation assay using the FLAG-UTX complex. Histones were used as a substrate. (C) Histone lysine methyltransferase (HKMT) assay using the FLAG-UTX complex. Reconstituted nucleosomes containing wild-type (Wt) recombinant H3 or mutant H3 (Lys4 → Ala4) were used as substrates. CBB, Coomassie brilliant blue. (D) Western blot analysis of an HKMT assay performed as in (C), using antibodies against various methyl marks. Native (Nat) nucleosomes (Nuc) were used as a positive control for Western blot analysis. Rec, reconstituted. (E) Silver staining and Western blot analysis of the UTX complex fractionated by Superose 6 gel filtration.

These observations prompted us to investigate the interplay between the MLL complex, H3K4 methylation, and the UTX protein at the promoter of the HOXA13 and HOXC4 genes. The depletion of UTX did not have a significant effect on promoter occupancy of components of either PRC2 or the MLL2/3 complex as measured by qChIP, nor was there any change in the dior trimethyl H3K4 levels (fig. S7, A and B). These results indicate that although UTX is a component of MLL2/3-containing complexes, the depletion of UTX and a concomitant increase in H3K27 methylation do not trigger changes in either the occupancy of the components of the MLL2/3 complexes or the methylation status of H3K4 at the HOXA13 and HOXC4 promoters.

To assess whether UTX plays a role during cellular differentiation, we used the human pluripotent embryonal carcinoma cell line NT2/D1 that differentiates into neural lineages upon treatment with retinoic acid (RA). We examined the occupancy of UTX at genes known to be activated during differentiation after RA treatment. UTX is present at HOXA1-3 and HOXB1-3 promoters, and the treatment of NT2/D1 cells with RA results in increased occupancy of UTX at these genes (Fig. 4, A to D, and fig. S8, A and E). The increase in UTX occupancy at these HOX genes was paralleled with a decrease in both PRC2 and trimethyl H3K27, leading to the activation of transcription (Fig. 4, E to L, and fig. S8, B, C, F, G, and I to N). Such RA-induced transcriptional activation is accompanied by increased trimethylation of H3K4 and the recruitment of ASH2L, a critical component of the MLL complex (Fig. 4, M to T, and fig. S8, D and H). In addition, whereas the recruitment of ASH2L and the trimethylation of H3K4 occur at 18 hours, the increased occupancy of UTX and the demethylation of H3K27 appear at a later 24-hour time point (Fig. 4). This suggests an ordered recruitment of an H3K4 methyltransferase and the H3K27 demethylase after RA-induced activation. Taken together, these results indicate that components of the MLL complex and UTX could be recruited to RA-responsive genes after RA stimulation to promote the methylation of H3K4, demethylation of H3K27, and a consequent activation of transcription.

Fig. 4.

RA treatment results in recruitment of the UTX/MLL complex, concomitant with decreased levels of trimethyl H3K27 and increased levels of trimethyl H3K4. (A to T) UTX occupancy (A to D), SUZ12 occupancy (E to H), trimethyl H3K27 levels (I to L), trimethyl H3K4 levels (M to P), and ASH2L occupancy (Q to T) at HOXA1, HOXA3, HOXB1, and HOXB2 genes were analyzed by a qChIP assay. Data are presented as the mean ± SEM (error bars) (n ≥ 3 except in the case of ASH2L, where n = 2). The relative occupancy represents the fold change in the percent of input over the control (28). The values of percent input set as 1 in (A) to (T) are as follows: (A), 0.0028%; (B), 0.0012%; (C), 0.003%; (D), 0.004%; (E), 0.046%; (F), 0.031%; (G), 0.099%; (H), 0.035%; (I), 0.039%; (J), 0.16%; (K), 0.02%; (L), 0.022%; (M), 2.47%; (N), 6.3%; (O), 1.11%; (P), 4.96%; (Q), 0.1%; (R), 0.04%; (S), 0.04%; and (T), 0.07%.

In contrast to HOXA1 to A3 and HOXB1 to B3, we did not find either UTX or trimethyl H3K27 at the promoter of the OCT4 gene, a regulator of pluripotency, during the RA-induced repression of OCT4 transcription (fig. S9, A, C, and E). Moreover, whereas UTX is present at the HOXA13 promoter, which is transcriptionally active in undifferentiated cells, RA treatment does not result in further recruitment of UTX or increased trimethyl H3K27 during RA-induced transcriptional repression (fig. S9, B, D, and F). These results indicate that the repression of OCT4 and HOXA13 upon RA treatment in NT2/D1 cells is probably regulated through a distinct mechanism other than that mediated by H3K27 methylation.

In this work, we show that UTX, a component of MLL2/3 complexes, is a JmjC-domain-containing histone demethylase with the ability to specifically demethylate di- and trimethylated H3K27. Our in vivo studies demonstrated a critical role for UTX in the regulation of transcription, as well as levels of H3K27 methylation at HOX gene clusters. We provide evidence that UTX regulates transcription of HOXA and HOXC genes by modulating the recruitment of PRC1 and the monoubiquitination of histone H2A. Our results suggest an ordered cycle of H3K4 trimethylation brought about by the components of the MLL complex, followed by a wave of demethylation of H3K27 mediated by UTX during RA-induced transcriptional activation of HOXA and HOXB cluster genes in NT2/D1 cells. Taken together, these findings reveal that similar to other histone methylation sites, the methylation of H3K27 is a reversible process regulated by cellular signaling events.

Supporting Online Material

Materials and Methods

Figs. S1 to S9

Table S1


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article