phf8 targets histone methylation and rna polymerase ii to...

MOLECULAR AND CELLULAR BIOLOGY, July 2010, p. 3286–3298 Vol. 30, No. 130270-7306/10/$12.00 doi:10.1128/MCB.01520-09Copyright © 2010, American Society for Microbiology. All Rights Reserved.

PHF8 Targets Histone Methylation and RNA PolymeraseII To Activate Transcription�†

Klaus Fortschegger,1§‡ Petra de Graaf,2§ Nikolay S. Outchkourov,2§ Frederik M. A. van Schaik,2H. T. Marc Timmers,2* and Ramin Shiekhattar1,3*

Center for Genomic Regulation, Barcelona Biomedical Research Park, 08003 Barcelona, Spain1; Department of Physiological Chemistry,University Medical Centre, 3584 CG Utrecht, Netherlands2; and Wistar Institute, Philadelphia, Pennsylvania 191043

Received 24 November 2009/Returned for modification 30 December 2009/Accepted 14 April 2010

Mutations in PHF8 are associated with X-linked mental retardation and cleft lip/cleft palate. PHF8 containsa plant homeodomain (PHD) in its N terminus and is a member of a family of JmjC domain-containingproteins. While PHDs can act as methyl lysine recognition motifs, JmjC domains can catalyze lysine de-methylation. Here, we show that PHF8 is a histone demethylase that removes repressive histone H3 dimethyllysine 9 marks. Our biochemical analysis revealed specific association of the PHF8 PHD with histone H3trimethylated at lysine 4 (H3K4me3). Chromatin immunoprecipitation followed by high-throughput sequenc-ing indicated that PHF8 is enriched at the transcription start sites of many active or poised genes, mirroringthe presence of RNA polymerase II (RNAPII) and of H3K4me3-bearing nucleosomes. We show that PHF8 canact as a transcriptional coactivator and that its activation function largely depends on binding of the PHD toH3K4me3. Furthermore, we present evidence for direct interaction of PHF8 with the C-terminal domain ofRNAPII. Importantly, a PHF8 disease mutant was defective in demethylation and in coactivation. This is thefirst demonstration of a chromatin-modifying enzyme that is globally recruited to promoters through itsassociation with H3K4me3 and RNAPII.

Posttranslational modifications of histone tails play an im-portant role in chromatin structure and function (27). Whilethe presence of several modifications correlates with gene ac-tivation and transcription, others have opposing repressivefunctions and are enriched in transcriptionally inactive or het-erochromatic regions. A well-studied type of histone modifi-cation is methylation of lysine residues, which is conferred byhistone methyltransferases (KMTs). There are three possiblestates of lysine methylation, namely, mono-, di-, and trimethyl-ation. As expected for reversible marks involved in dynamicgene regulation, the methyl groups can be removed by histonedemethylases (KDMs) of either the LSD1 or the Jumonji C-terminal-containing (JmjC) family of proteins (43, 48).

The JmjC family of histone demethylases consists of approx-imately 30 members in humans (9, 25). Plant homeodomain(PHD) finger-containing proteins 2 and 8 (PHF2 and PHF8,respectively) and KIAA1718 constitute a subgroup containinga single N-terminal PHD followed by the catalytic JmjC do-main. Several protein domains (including the PHD; the MBT,WD40, and Tudor domains; and chromodomains) have beenshown to bind to peptides containing either an unmodified or

a methylated lysine residue, and some PHDs specifically inter-act with histone H3 trimethylated at lysine 4 (H3K4me3) (42,51, 53). H3K4me3 is considered to be an activating chromatinmark because it is enriched at active RNA polymerase II(RNAPII) transcription start sites (TSSs) (4, 17). In mamma-lian cells, H3K4me3 is mainly established by the SET1/KMT2family, which includes the mixed-lineage leukemia 1 to 4(MLL1 to -4) proteins. These chromatin modifiers can be re-cruited to promoters upon gene activation (16, 52), and certainMLL complexes also contain KDMs that concomitantly re-move the repressive H3K27me3 mark and enhance expression(7, 21, 30).

On the other hand, H3K4me3 can be erased by KDMs of theJARID1/KDM5 family that are part of polycomb-repressivecomplexes (PRC1/2) (29, 35). These complexes also containKMTs that are able to methylate histone H3 at lysine 9 or 27.Methylated H3K9 and H3K27 are repressive marks, which areread by PRC1 via chromodomain-containing proteins, a mech-anism that is thought to be responsible for enforcement andinheritance of silenced chromatin (5, 18). This obvious cou-pling of writers (KMTs), readers (like proteins containingPHDs), and erasers (KDMs) enables cross talk between dif-ferent chromatin modifications (45). Furthermore, many SETand JmjC domain proteins contain additional domains or pu-tative DNA-binding modules, increasing their specificity andaffinity for modified chromatin.

PHF8 is a ubiquitously expressed nuclear protein whosedysfunction is implicated in disease (28). Mutations inthe PHF8 locus on the X chromosome have been linked toSiderius-Hamel syndrome, an X-linked mental retardation(XLMR) that is often accompanied by cleft lip and/or cleftpalate (44). The described mutations result mostly in earlytruncations of the protein before or in the JmjC domain (1,

* Corresponding author. Mailing address for Ramin Shiekhattar:The Wistar Institute, 3601 Spruce Street, Philadelphia, PA 19104.Phone: (215) 898-3896. Fax: (215) 898-3986. E-mail: [email protected]. Mailing address for H. T. Marc Timmers: Department ofPhysiological Chemistry, University Medical Centre, Universiteitsweg100, 3584 CG Utrecht, Netherlands. Phone: (31) 88 756 8981. Fax: (31)88 756 8101. E-mail: [email protected].

‡ Present address: Children’s Cancer Research Institute, St. AnnaKinderkrebsforschung, Zimmermannplatz 10, 1090 Vienna, Austria.

§ K.F., P.D.G., and N.S.O. contributed equally to this work.† Supplemental material for this article can be found at http://mcb

.asm.org/.� Published ahead of print on 26 April 2010.

3286

28). Moreover, a single point mutation (F279S) in the JmjCdomain and a microdeletion of the whole locus were alsoreported to cause the disease phenotype (26, 36).

In the present study, we delineate crucial biochemical prop-erties of PHF8. We report that PHF8 is a demethylase specificfor H3K9me2 and a binder of H3K4me3-marked nucleosomes.Genome localizations show that PHF8, H3K4me3, and RNA-PII cooccupy thousands of promoters. Association studies sug-gest that PHF8 interacts directly with the C-terminal domain(CTD) of the RPB1 subunit of RNAPII. The F279S mutant ofPHF8 displays cellular mislocalization, defective histone de-methylation, and aberrant coactivation function.

MATERIALS AND METHODS

Vectors. Full-length PHF8 (NCBI reference sequences NM_15107.1 formRNA and NP_055922.1 for protein) was cloned into the pFlag-CMV2 vectorfor expression in mammalian cells and into pFastBacHTa-Flag for expression ininsect cells. Truncated versions (containing amino acids 1 to 352 or 1 to 489 ofPHF8) were cloned into pFlag-CMV2 with a short simian virus 40 (SV40)nuclear localization signal (NLS) (PKKKRKVG) added to the C terminus toensure correct nuclear import. The mutations D28A/W29A, H31A/C34A,H247A/D249A, and F279S were generated by site-directed mutagenesis usingthe QuikChange protocol (Stratagene). The cloning primer and mutagenesisoligonucleotide sequences are described in the supplemental material.JARID1A- and JMJD3-HA-pCMV plasmids were kind gifts from Kristian He-lin’s laboratory. A PCR fragment encoding the PHD of human PHF8 (residues2 to 66) (PHF8-PHD) was inserted into the pGEX-2T-derived pRPN265NBplasmid. Glutathione S-transferase (GST)–PHF8-PHD mutant plasmids wereobtained by site-directed mutagenesis using the QuikChange protocol (Strat-agene). The plasmids for GST-Taf3 (amino acids 857 to 924) and GST-UbcH5Bwere described previously (14, 51). The EB2 plasmid encoding GST fused to 27repeats of the consensus CTD (YSPTSPS) was a kind gift of Marc Vigneron(ESBS, Strasbourg, France). All inserts were verified by DNA sequencing.

Recombinant protein expression and purification. Full-length PHF8 contain-ing N-terminal Flag and His6 tags was expressed in Sf9 cells using baculovirus(Invitrogen Bac-to-Bac system). The cells were lysed in whole-cell lysis buffer (20mM Tris-HCl, pH 7.4, 137 mM NaCl, 1 mM EDTA, 1.5 mM MgCl2, 10%glycerol, 1% Triton X-100, 0.2 mM phenylmethylsulfonyl fluoride [PMSF], 1�g/ml aprotinin, 1 �g/ml leupeptin, 1 �g/ml pepstatin) 2 days after infection andincubated with M2 Flag agarose beads (Sigma). The beads were washed fourtimes with BC500 buffer (20 mM Tris-HCl, pH 7.4, 0.2 mM EDTA, 10 mM�-mercaptoethanol, 10% glycerol, 500 mM KCl, 0.1% NP-40, 0.2 mM PMSF, 1�g/ml aprotinin, 1 �g/ml leupeptin, 1 �g/ml pepstatin), and recombinant proteinwas eluted with Flag peptide (400 �g/ml; Sigma) in BC500. After concentrationof the eluates by centrifugation in Microcon YM-10 columns (Millipore), frac-tions were run on 4 to 12% Tris-glycine SDS-polyacrylamide gels and Coomassiestained (Invitrogen), and the protein concentration was estimated by comparisonto bovine serum albumin (BSA) standards.

In vitro demethylation assay. Four micrograms of bulk histones from calfthymus (Sigma) was incubated in 30 �l demethylation buffer [20 mM Tris-HCl,pH 7.4, 2 mM ascorbic acid, 1 mM �-ketoglutarate, approximately 100 mM KCl,100 mM NaCl, 50 �M (NH4)2Fe(SO4)2] with or without up to 10 �g recombinantPHF8 for 4 h at 37°C. After denaturation in SDS loading buffer, the reactionmixtures were run on 4 to 20% Tris-glycine SDS-polyacrylamide gels and blottedto polyvinylidene difluoride (PVDF) membranes (Millipore). Western blots wereblocked for �1 h at room temperature with 5% skim milk powder diluted inTris-buffered saline containing 0.1% Tween 20 (TBS-T) before sequential incu-bation for 1 h at room temperature with antibodies. Primary mouse or rabbitantihistone antibodies were usually diluted 1:1,000, and secondary anti-IgG–alkaline phosphatase antibodies were diluted 1:10,000 (Promega) in TBS-T. Sig-nals were developed by incubation with 5-bromo-4-chloro-3-indolylphosphate–nitroblue tetrazolium (BCIP-NBT) substrate, and the blots were dried andscanned. All antibodies used in this study are listed in the supplemental material.

Cell culture and indirect immunofluorescence. 293T, U2OS, Hs68, andHeLa-S3 cells were grown in Dulbecco’s modified Eagle’s medium (DMEM)containing 10% fetal bovine serum (FBS), L-glutamine, and antibiotics. Tran-sient transfection of plasmid DNA was performed 6 h after seeding using Fu-Gene6 transfection reagent according to the manufacturer’s instructions(Roche). For immunofluorescence assays, 293T cells were seeded onto glass

coverslips and transfected with 1 �g PHF8(1-489)NLS-pFlag-CMV2 wild-type ormutant constructs using 3 �l Metafectene Pro transfection reagent (Biontex) percoverslip. Twenty-four to 48 h later, the coverslips were washed with phosphate-buffered saline (PBS) and fixed in fresh 1% formaldehyde-PBS for 10 min. Afterbeing washed, the cells were permeabilized in 0.1% Triton X-100–PBS for �10min and sequentially incubated for 1 h at room temperature with antibodies.Primary mouse or rabbit antihistone antibodies were diluted 1:200 to 1:500, andanti-Flag antibodies were diluted 1:1,000 in 0.1% Triton X-100–PBS. Secondaryanti-mouse IgG–Alexa Fluor-488 and anti-rabbit IgG–Alexa Fluor-568 antibod-ies (Invitrogen) were diluted 1:1,000 in 0.1% Triton X-100–PBS. After DAPI(4�,6-diamidino-2-phenylindole) counterstaining, the coverslips were washedwith PBS and mounted on slides using Mowiol 4-88 containing 2.5% 1,4-diazabicyclo[2.2.2]octane (DABCO) as an antibleach. Detailed information onimage acquisition and statistical analysis is given in the supplemental material.

ChIP. 293T, HeLa-S3, or Hs68 cells were trypsinized from 2 (293T and HeLa)or 16 (Hs68; 4-day serum starved or 4-h serum restimulated) subconfluent 15-cmdishes, washed with PBS, and fixed in 10 ml of 1% formaldehyde-PBS at roomtemperature for 10 min. One milliliter of 1.25 M glycine was added and incu-bated for 5 min at room temperature. The cells were pelleted, washed with coldPBS, and lysed for 15 min on ice in 300 �l chromatin immunoprecipitation(ChIP) lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8, andprotease inhibitors). After the addition of 1.7 ml ChIP dilution buffer (0.01%SDS, 1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl, pH 8, 167 mM NaCl,and protease inhibitors), chromatin was sonicated on ice in a Bioruptor (Diage-node) for 10 min (30 s on, 30 s off; high power load). After dilution with another1 ml of ChIP dilution buffer, solubilized chromatin was cleared and precleared byincubation with protein A-agarose beads (Millipore). Five hundred to 1,000 �l ofthis chromatin solution (25 to 50 �g DNA) was incubated overnight with 2 to 4�g of antibody (listed in the supplemental material). Protein A-bead slurry (30�l) was added, rotated for 2 h at 4°C, and washed 3 times with low-salt buffer(0.1% SDS, 1% Triton X-100, 2 mM EDTA, 20 mM Tris-HCl, pH 8, 150 mMNaCl), one time with high-salt buffer (0.1% SDS, 1% Triton X-100, 2 mMEDTA, 20 mM Tris-HCl, pH 8, 500 mM NaCl), one time with LiCl buffer (250mM LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA, 10 mM Tris-HCl, pH 8),and one time with TE buffer (10 mM Tris-HCl, pH 7.6, 1 mM EDTA). Boundchromatin was eluted for �1 h at room temperature in 100 �l fresh elution buffer(0.1 M NaHCO3, 1% SDS). Cross-links were reversed overnight at 65°C, andDNA was purified using a QIAquick PCR purification kit (Qiagen).

For ChIP followed by high-throughput sequencing (ChIP-seq), approximately10 ng immunoprecipitated DNA per sample was used for end cleanup andadaptor ligation. Size-selected preamplified inserts of 135 � 25 bp were se-quenced by the Illumina/Solexa technique. Total reads were mapped to thehuman genome version 18 using Paolo Ribeca’s GEM mapper (http://www.paoloribeca.net/software/GEM/index.html) (see Table S2 in the supplementalmaterial). Non-strand-specific cluster/peak analyses were performed with theNGS-Analyzer (Genomatix RegionMiner 2.02) using a 100-bp window and aPoisson distribution-derived cutoff value (depending on total reads; at least 8 to10 reads per cluster). Regions overlapping with clusters of irrelevant IgG ChIP(for HeLa and 293T cells) or input DNA control (for Hs68 cells) were filteredout. Overlap analyses were done with Genomatix GenomeInspector (Eldorado12-2008) using our own ChIP-seq data and published HeLa data for H3K4me3(37) and for RNAPII (38). Strand-specific read analyses were performed with theChIP-cor tool (http://ccg.vital-it.ch/chipseq/). Reads were correlated to orientedTSSs (49,392 ENSEMBL50 features). The examined distance from the referencefeature was �1,000 bp, the window width was 20 bp, the cutoff value was 10counts, and the output format was normalized counts. Wiggle files for display inthe UCSC genome browser were generated using the ChIP-center tool (hg18; tagshift, 60 bp; count cutoff, 10; resolution, 20 bp). For correlation studies, pub-lished HeLa expression data were used (10).

For quantitative ChIP (qChIP), purified immunoprecipitated DNA was usedas a template for real-time PCR with 0.2 �M locus-specific primers (sequencesare given in the supplemental material) and Bio-Rad IQ SYBR green mix.Samples and dilution series of input DNA were run in triplicate. The immuno-precipitated percentage of input was calculated using individual standard curves.

Bacterial protein expression and pulldowns. Expression of GST fusion pro-teins was induced, and lysates were prepared essentially as described previously(14). H3 peptides (�0.1 nmol) were coupled to magnetic streptavidin beads(M-280; Dynal) in H3-binding buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl,0.1% NP-40, 10 �M ZnCl2, 1 mM dithiothreitol [DTT], and protease inhibitors),and excess peptide was removed by washing. Crude bacterial lysates were incu-bated with peptide-coated streptavidin beads in binding buffer (50 mM Tris-HCl,pH 8, 150 mM NaCl, 0.1% NP-40, 10 �M ZnCl2, 1 mM DTT, and proteaseinhibitors) for 2 to 3 h at 4°C. Following extensive washing, bound proteins were

VOL. 30, 2010 PHF8 IS A TRANSCRIPTIONAL COACTIVATOR 3287

eluted in SDS sample buffer and analyzed by SDS-PAGE and Coomassie bril-liant blue R-250 (Bio-Rad).

For GST pulldown experiments, glutathione-agarose beads were coated withbacterial lysates containing GST fusion proteins for 60 min at 4°C and subse-quently washed three times with H3-binding buffer. Mononucleosomes fromHeLa cells (prepared as described previously [51]) or transfected 293T celllysates were mixed with 50 �l beads in a total volume of 400 �l and incubatedovernight at 4°C in H3-binding buffer. The beads were washed five times with 500�l of H3-binding buffer. Bound proteins were analyzed by immunoblotting.

Transient reporter coactivation assay. For luciferase assays, U2OS cells weretransfected in triplicate. The firefly reporter luciferase construct 5XGal4MLP-Luc was cotransfected with 50 ng TK promoter-driven Renilla luciferase plasmid(to normalize for differences in transfection efficiency) and PHF8-pFlag-CMV2in combination with Gal4 activator constructs as indicated. Cell lysates wereprepared 24 h after transfection, and the luciferase activity was determined usingthe Dual-Luciferase Reporter Assay System (Promega).

Coimmunoprecipitation (CoIP). 293T cells were lysed 24 to 40 h after trans-fection in IP buffer (50 mM Tris-HCl, pH 8, 150 mM KCl, 5 mM MgCl2, 0.5 mMEDTA, 0.1% NP-40, phosphatase, and protease inhibitor cocktails from SigmaAldrich). Flag-tagged protein was bound to M2 beads (Sigma-Aldrich) for 3 h at4°C, followed by 3 washes with IP buffer. Precipitated proteins were separated bySDS-PAGE and transferred onto PVDF membranes. The membranes weredeveloped with the appropriate antibodies and enhanced chemiluminescence(ECL) (Pierce).

Knockdown experiments and microarray analysis. HeLa cells were trans-fected with pRetroSuperPuro plasmids containing short hairpin RNAs (shRNAs)targeting PHF8 (PHF8-shA or -shB) or nontargeting control (NT-sh) (sequencesare given in the supplementary material). One day after transfection, selectionwith 2.5 �g/ml puromycin was performed for two more days. Cells were har-vested with Trizol reagent (Invitrogen) for RNA isolation or with nondenaturinglysis buffer for protein isolation or were fixed and sonicated for qChIP analysis.PHF8 knockdown was confirmed by quantitative reverse transcription (qRT)-PCR and immunoblotting.

Three independent knockdown experiments were performed in HeLa cellswith PHF8-shA and control NT-sh. Total RNA (500 ng) was reverse transcribed,labeled by in vitro transcription with Cy3 and Cy5, and hybridized to wholehuman genome arrays in dye swap (Agilent G4112F). After scanning of thearrays (Agilent G2565BA), features were extracted with GenePix Pro 6.0 andanalyzed with AFM 4.0 (6). Probes with average changes higher than 2-fold wereconsidered regulated. Nonannotated transcripts, PHF8, and three ambiguoustranscripts (CCDC6, SCD, and CNOT6, which had different probes up- anddownregulated) were excluded from further analysis. For 11 transcripts, regula-tion was confirmed by qRT-PCR using cDNAs obtained by reverse transcriptionof total RNA with Moloney murine leukemia virus (MMLV) enzyme andoligo(dT)18 primer. The sequences of qRT-PCR and qChIP primers are listed inthe supplemental material.

Microarray and ChIP-seq data accession number. Data from the ChIP-seqand microarray experiments were deposited as a SuperSeries under accessionnumber GSE20753 in the GEO database (http://www.ncbi.nlm.nih.gov/geo/).

RESULTS

PHF8 is a demethylase specific for dimethylated histone H3lysine 9. Several JmjC-containing proteins have been shown tospecifically demethylate histone H3 methylated at lysine 4, 9,27, or 36. These enzymes remove methyl moieties by an oxi-dative reaction that requires an Fe2� ion and �-ketoglutarateas cofactors. Since the JmjC domain of PHF8 carries all theamino acids needed for cofactor binding (Fig. 1A), PHF8 wassuggested to be a histone demethylase (25) and was very re-cently demonstrated to be active in vitro (31). We expressedfull-length PHF8 in Sf9 insect cells using the baculovirus sys-tem and isolated the recombinant protein by Flag affinity pu-rification (Fig. 1B). Bulk histones from calf thymus were usedas PHF8 substrates and analyzed by immunoblotting using an-tibodies specific for different methyl lysine modifications of histones.We observed a profound reduction of H3K9me2 after incubationwith PHF8. In contrast, we did not detect any change for theother histone methyl marks, including, H3K9me3, H3K27me3/2/1,

H3K4me3/2, H3K36me3/2, H3K79me3/2, and H4K20me3/2/1 (Fig.1C). Demethylation was proportional to the amount of enzyme(see Fig. S1A in the supplemental material) and the reactiontime (see Fig. S1B in the supplemental material) and wasdependent on the cofactors Fe2�, �-ketoglutarate, and ascor-bic acid (see Fig. S1C in the supplemental material). LikeLoenarz and colleagues (31), we also observed weak activity onH3K9me1 (see Fig. S1B in the supplemental material), but incontrast, we could not detect significant reduction ofH3K27me2 or H3K36me2 (Fig. 1C). This difference may havebeen due to different recombinant proteins used in the twostudies: ours was a full-length PHF8, while theirs was devoid ofthe PHD and the C-terminal half. While our work was underreview, other studies were published that also showedH3K9me2-specific demethylase activity for PHF8 (15, 19, 55).

To verify its demethylase activity in vivo, we expressed aFlag-tagged truncated version of PHF8 (containing amino ac-ids 1 to 489 and a C-terminal nuclear localization signal) in293T cells. After 24 to 48 h, the cells were fixed and subjectedto indirect immunofluorescence with antibodies against Flagand histone H3 methylated at different lysine residues. Themean fluorescence signals of 10 of each untransfected andtransfected nucleus were quantified and compared using Stu-dent’s t test (see Table S1 in the supplemental material).H3K9me2 was markedly reduced in cells that overexpressedwild-type Flag-PHF8(1-489) (Fig. 1D). The JmjC domain con-ferred the H3K9me2 demethylase activity. This was indicatedby the lost activity of an H247A/D249A mutant predicted to bedefective in cofactor binding (Fig. 1E). Interestingly, theXLMR point mutant F279S was devoid of activity (Fig. 1F).This disease mutant displayed aberrant localization to the cy-toplasm, formed aggregates in vivo (Fig. 1F), and had reducedsolubility at low salt concentrations in vitro (data not shown).Our observations with this XLMR mutant are consistent withrecent work by Loenarz et al. (31). Intriguingly, we also ob-served dramatically impaired demethylase activity when wemutated the putative H3K4me3-binding residues D28 andW29 to alanine residues in the PHD (Fig. 1G). However, wecould observe a small demethylation activity with this mutantfollowing prolonged expression (see Table S1 in the supple-mental material).

While ectopic expression of PHF8(1-489) resulted in a de-crease in H3K9me1 (Fig. 1H), there were no changes forH3K9me3 (Fig. 1I), H3K4me2 (Fig. 1J), H3K27me2 (Fig. 1K),or H3K36me2 (see Table S1 in the supplemental material).The demethylation of H3K9me1 was deemed to be less robustboth in vivo (Fig. 1H) and in vitro (Fig. S1B). The increasedamounts of H3K9me1 following the demethylation ofH3K9me2 may mask the extent of H3K9me1 activity of PHF8.

Further truncation of the PHF8 protein leading to a proteinspanning amino acids 1 to 352 containing the PHD and JmjCdomain was devoid of demethylation activity in vivo (see Fig.S2A in the supplemental material). Surprisingly, the full-lengthPHF8 was also deemed to be very low in demethylation activitywhen expressed in vivo (see Fig. S2B in the supplementalmaterial). These results suggest that the approximately 100amino acids following the consensus JmjC domain are impor-tant for demethylation activity in vivo. Moreover, it is likelythat the C-terminal half of PHF8 exerts a negative effect on thedemethylation activity in vivo, as full-length PHF8 displayed

3288 FORTSCHEGGER ET AL. MOL. CELL. BIOL.

FIG. 1. PHF8 is an H3K9me2 demethylase. (A) Schematic representation of PHF8 with its PHD (red), JmjC domain (cyan), and nuclearlocalization signals (magenta). Below, conserved amino acids crucial for metal ion chelation (red) and for binding to H3K4me3 (green) and�-ketoglutarate (cyan) are depicted. F279 (blue) is mutated to serine in some patients with XLMR. (B) Recombinant full-length PHF8 wasproduced in insect cells and quantified by comparison to a BSA standard after Coomassie staining. The individual lanes were spliced from one gel.(C) Bulk histones were incubated without () or with (�) recombinant PHF8 and analyzed by immunoblotting using the given specific histoneantibodies. (D to K) Indirect-immunofluorescence analyses of 293T cells transfected for 24 h with Flag-PHF8(1-489)NLS wild type or theH247A/D249A, F279S, or D28A/W29A mutant. Flag and the indicated histone H3 methyl lysine antibodies were used. Transfected Flag-positivecells are indicated by the arrowheads. The image field dimension is 68 �m. Quantifications of mean fluorescence signals are given in Table S1 inthe supplemental material.


diminished activity. Finally, consistent with a published report(31), overexpression of PHF8(1-489) in HeLa cells did notyield a change in H3K9me2 (see Fig. S2C in the supplementalmaterial), suggesting an important role for a cell type specific-ity factor for demethylation in vivo. Taken together, our resultsindicate that PHF8 displays demethylation activity towardmono- and dimethyl H3K9. However, we found evidence forcell-type-specific factors regulating the demethylation activityof PHF8 in vivo. Importantly, we showed the contribution ofthe PHD to the enzymatic activity of PHF8 in 293T cells invivo, while the domain was shown not to be crucial for itsactivity in vitro (31).

The PHD of PHF8 binds histone H3 methylated at lysine 4.Comparison of the PHF8 PHD with H3K4me3-specific PHDsof the bromodomain PHD finger transcription factor (BPTF)and TBP-associated factor 3 (TAF3) (39, 51) revealed conser-vation of crucial Zn2�- and H3K4me3-binding residues (indi-cated in red and green in Fig. 1A). In peptide pulldown assays,we found that bacterially expressed GST fusion with the PHDof PHF8 (amino acids 2 to 66) was specifically retained bybiotinylated H3K4me3/2 peptides (amino acids 1 to 17), butnot by H3K4me1, H3K9me3/2/1, and H3K36me3/2/1 (aminoacids 22 to 39) or unmethylated H3 peptides (Fig. 2A). ThePHD of Taf3 was used as a positive control in these assays.Interaction of the PHF8-PHD with H3K4me3 was not com-promised by modifications at neighboring residues, such asmethylation of H3R2/H3K9 or phosphorylation of H3T6/H3S10. In contrast, concomitant phosphorylation of H3T3blocked K4me3 binding (Fig. 2B). H3T3 phosphorylation ismediated by the mitotic haspin kinase (11), and this correlateswith the observation that PHF8 dissociates from mitotic chro-mosomes despite the presence of H3K4me3 (see Fig. S3A andB in the supplemental material). As expected, mutation ofZn2�-binding (H31A/C34A) (data not shown) or aromatic-cage (D28A/W29A) residues in the PHF8-PHD abolishedH3K4me3 binding (Fig. 2B). Similar results were obtained witha mammalian expression construct containing both the PHDand JmjC domain of PHF8 (amino acids 1 to 352). Whilewild-type PHF8, the JmjC H247A/D249A mutant, and thedisease mutant (F279S) bound H3K4me3/2 peptides, the PHDD28A/W29A mutant did not (Fig. 2C).

We next assessed whether the PHD of PHF8 can also inter-act with nucleosomal H3K4me2/3. Using GST–PHF8-PHD,but not the D28A/W29A mutant, we could retain mononucleo-somes derived from HeLa cells (confirmed by immunoblottingagainst H2A, H3, and H4), which were enriched in H3K4me3/2but not H3K9me2 (Fig. 2D). Again GST-Taf3-PHD served asa positive control. Analysis of PHF8 in vivo using antibodiesagainst PHF8 revealed colocalization with H3K4me3 andRNAPII (see Fig. S3A and B in the supplemental material) inputative euchromatic regions of interphase 293T cells, while itwas depleted at the nuclear periphery and in perinucleolarregions, where H3K9me2 was enriched (see Fig. S3C in thesupplemental material).

Having established that the PHF8-PHD can bind theH3K4me3 mark, we examined the observation that PHD in-tegrity was important for in vivo demethylase activity (Fig. 1G).To test K4me3 involvement in this, we cotransfected 293T cellswith wild-type Flag-PHF8(1-489)NLS and the H3K4me3-spe-cific demethylase JARID1A/KDM5A (8). The ability of PHF8

to demethylate H3K9me2 was significantly reduced by con-comitant H3K4me3 removal (see Fig. S4A in the supplementalmaterial), while control cotransfections of PHF8 with JMJD3/KDM6B, an H3K27me3 demethylase (13), did not result inimpaired PHF8 demethylase activity (see Fig. S4B in the sup-plemental material). The mean fluorescence signals ofH3K9me2/H3K4me3 for PHF8/JARID1A and H3K9me2/H3K27me3 for PHF8/JMJD3 cotransfections were quantified

FIG. 2. The PHD of PHF8 binds H3K4me3. (A) Bacterial lysatescontaining GST-PHF8(2-66) (top) or GST-Taf3(857-924) (bottom)were incubated with streptavidin beads coated with H3 peptides car-rying the indicated modifications and a C-terminal lysyl biotin. Boundproteins were analyzed by SDS-PAGE and Coomassie staining. Thearrowheads indicate the positions of the GST fusion protein. (B) Wild-type (top) and D28A/W29A mutant (bottom) GST–PHF8-PHD wereanalyzed as described for panel A for binding to methylated andphosphorylated H3 peptides (1 to 17) as indicated above. (C) Lysatesof transfected 293T cells expressing wild-type or mutant Flag-PHF8(1-352) proteins were incubated with streptavidin beads coated with theindicated H3 peptides. Bound proteins were analyzed by immunoblot-ting using Flag antibodies. The arrowheads show the positions of theindicated proteins, and the asterisks show background binding to H3peptides. (D) Glutathione beads were coated with GST, GST–PHF8-PHD, GST–PHF8-PHD mutant (D28A/W29A), and GST-Taf3-PHDas a positive control and incubated with native HeLa mononucleo-somes. The bound material was analyzed by immunoblotting using thegiven antibodies or Ponceau S staining as indicated.


and compared for transfected and untransfected cells (see Fig.S4C in the supplemental material). This is in agreement withthe stimulation of H3K9me2 demethylation activity on pep-tides by the presence of H3K4me3 (19). Taken together, theseexperiments suggest that the PHD of PHF8 can bind specifi-cally to H3K4me3-marked nucleosomes and that this binding isimportant for its histone demethylase activity in vivo.

PHF8 is a global reader of H3K4me3. In order to determinethe genomic location of PHF8, we performed chromatin im-munoprecipitation followed by high-throughput sequencingusing the Illumina/Solexa technology (ChIP-seq). Sonicatedchromatin from fixed 293T, HeLa-S3, and serum-starved andrestimulated Hs68 cells was immunoprecipitated with rabbitanti-PHF8 antibody and normal rabbit IgG as a negative con-trol. In Hs68 fibroblasts, we used input DNA as a control andperformed ChIP with rabbit PHF8 or H3K4me3 antibody. Onaverage, 14 million total reads were obtained per flow cell, andapproximately 63% of these could be unambiguously mappedto the human genome. Cluster analysis of uniquely alignedreads revealed more than 10,000 significant peaks for PHF8 inall three cell lines. The ChIP-seq breakdown and cluster iden-tities are given in Table S2 in the supplemental material. Thepeaks were almost exclusively located in close proximity toTSSs, and we observed strong overlap between PHF8,H3K4me3, and RNAPII clusters using our own and publisheddata (37, 38) (Table 1), suggesting that PHF8 preferentiallyoccupied thousands of active or poised promoters. ChIP-seqdata were validated at several c-myc target loci by quantitativereal-time PCR of chromatin-immunoprecipitated DNA ob-tained with rabbit PHF8, c-myc, or normal IgG. PHF8 exhib-ited high (for NCL [see Fig. S5A in the supplemental mate-rial]), medium (for CDK4, CCNB1, and ITGB1 [see Fig. S5B toD in the supplemental material]), or no (for TERT [see Fig.S5E in the supplemental material]) enrichment in these re-gions, which also corresponded to H3K4me3 and RNAPIIoccupancies.

A more detailed strand-specific analysis of sense (5�) orantisense (3�) reads in respect to oriented TSSs revealed asimilar distribution and a phase shift of 120 bp for PHF8 in293T (Fig. 3A) and HeLa-S3 (Fig. 3B) cells and for PHF8 (Fig.3C) and H3K4me3 (Fig. 3D) in Hs68 cells. No enrichment ofreads for controls, like IgG ChIP of HeLa cells (Fig. 3E), or forinput DNA of serum-stimulated Hs68 cells (Fig. 3F) was ob-served at TSSs. Active and poised TSSs exhibit defined posi-tioning of nucleosomes, which tend to be methylated at H3K4(4, 41). Similar to previously reported experiments using nativeChIP of mononucleosomes (40, 41) or sonicated cross-linkedchromatin (37), we also observed several phased H3K4me3peaks at the expected nucleosome positions in Hs68 (Fig. 3D).

The phase shift in our experiment was with 120 bp; however, itwas smaller than the 150 bp obtained with mononucleosomes,possibly because of harsher fragmentation conditions duringthe sonication process. Analysis of PHF8 reads indicated onepeak immediately before and one after the TSS (Fig. 3A to C).Therefore, we concluded that PHF8 bound preferentially tonucleosomes at the 2 and �1 positions. A short region at theTSS corresponding to nucleosome position 1 was depleted ofH3K4me3 and PHF8 reads. In active or poised promoters, thisnucleosome position is occupied either by RNAPII and thebasal transcription machinery (41) or by unstable H2A.Z/H3.3-containing nucleosomes (23). Association of PHF8, H3K4me3,and PolII in vivo was further substantiated by their colocaliza-tion in immunofluorescence analysis (see Fig. S3A and B in thesupplemental material).

Next, we determined the overlap of PHF8-occupied genes inthe 3 examined cell lines. Out of 10,876 genes in HeLa, 9,210in 293T, and 8,014 in Hs68 cells, 6,632 shared PHF8 enrich-ment at their promoters, while others were occupied in onlyone or two of the cell lines (Fig. 3G).

Subsequently, we correlated PHF8 occupancy with expres-sion and H3K4me3 levels. Basically, all annotated genes occu-pied by PHF8 in Hs68 cells also carried H3K4me3 peaks, butwe detected about twice as many clusters and occupied genesfor H3K4me3 as for PHF8. We sorted the genes by decreasingcounts of H3K4me3 reads at their promoters, grouped theminto bins of 500, and determined the percentage of PHF8cooccupancy per bin. While 70% of the promoters with highand medium H3K4me3 levels were cooccupied by PHF8, lessthan 10% of promoters with low H3K4me3 also carried PHF8(Fig. 3H). Furthermore, the average PHF8 read count at pro-moters with high H3K4me3 levels was about twice as high as atthose with low levels (data not shown). In a similar analysis, wetook advantage of Affymetrix data for HeLa from a previousstudy (10), sorted genes by decreasing expression levels, clas-sified them into bins of 500, and for each bin determined thepercentage of PHF8 occupancy at their promoters in HeLacells. While 90% of genes with high transcript levels displayedPHF8 occupancy, only 10% of genes with low expression levelswere occupied by PHF8 (Fig. 3I). These results suggest apositive correlation between genes that are occupied by PHF8and those that display transcriptional activity. However, about30% of the genes with high H3K4me3 levels in Hs68 were notcooccupied by PHF8.

PHF8 is a transcriptional coactivator. PHF8 may activatetranscription, for example, by binding to an activating chroma-tin mark (H3K4me3) and/or by removing a repressive chroma-tin modification (H3K9me2). We used a reporter assay to testthese possibilities (51). Cotransfection of human osteosarcoma

TABLE 1. Summary of ChIP-seq cluster analysis

Cell line No. ofPHF8 clusters

% TSSoverlap (�1,000 bp)

No. ofoccupied promoters

% H3K4me3overlap (�1,000 bp)

% RNAPIIoverlap (�1,000 bp)

HeLa-S3 19,447 96 10,876 94a 88b

293T 13,778 97 9,210 NDc NDHs68 � FBS 16,411 83 8,014 81 ND

a Data from Robertson et al. (37).b Data from Rozowsky et al. (38).c ND, not determined.


FIG. 3. PHF8 reads H3K4me3 at TSSs, and occupancy correlates with transcript levels. ChIP-seq reads of PHF8 in 293T (A), HeLa (B), andserum-stimulated Hs68 (C) cells; of H3K4me3 in serum-starved Hs68 cells (D); of negative-control IgG in HeLa cells (E); and of input in serum-stimulated Hs68 cells (F) were strand specifically correlated with annotated TSSs in a range from bp 1000 to �1000. Normalized count densities of 5�(red), 3� (green), and combined raw (black) reads are shown. The putative positions of phased nucleosomes are numbered and depicted as blue ovals.(G) Venn diagram depicting the overlap of PHF8-occupied genes in HeLa (blue), 293T (red), and Hs68 (green) cells. (H) Correlation of H3K4me3 andPHF8 occupancies in Hs68 cells. Genes with H3K4me3 peaks were sorted by read counts and grouped into bins of 500, and the percentage of genes inthe bin that were also occupied by PHF8 was determined. Bins with decreasing H3K4me3 levels are shown from left to right; the correspondingpercentages of PHF8 cooccupancy are given on the y axis. (I) Correlation of PHF8 and transcript levels in HeLa cells. Genes were sorted by normalizedexpression levels and grouped into bins of 500, and the percentage of genes in the bin that were occupied by PHF8 was determined. Bins with decreasingtranscript levels are shown from left to right; the corresponding percentages of PHF8 occupancy are shown on the y axis.


U2OS cells with a firefly luciferase reporter construct and aconstruct encoding full-length Flag-tagged PHF8 indicated a6-fold increase in promoter activity (Fig. 4A). Moreover, ec-topic expression of PHF8 revealed a coactivator function to-ward Gal4-tagged Ash2, p53, c-myc, and E2F (Fig. 4A).Shorter versions of PHF8 (1-352 and 1-489) displayed im-paired coactivation, although they were expressed at similarlevels, as shown by Western blotting (Fig. 4A). To extend this,we initially assessed the functional consequences of PHD andJmjC mutations in the context of truncated PHF8(1-489).While PHF8(1-489) displayed impaired coactivation capacitycompared with full-length PHF8, it was still able to enhanceE2F activity by about 2.5-fold (Fig. 4B). Interestingly, while thePHD (D28A/W29A) and XLMR (F279S) mutants displayeddiminished coactivation functions, the JmjC H247A/D249Amutant showed only a small decrease in coactivator function.

We next tested whether these observations extended to thefull-length PHF8. Similar to PHF8(1-489), in the context offull-length PHF8, the PHD D28A/W29A mutant or the F279Smutant was impaired in coactivation of Gal4-p53 or Gal4-Ash2(Fig. 4C). Again, the catalytic JmjC H247A/D249A mutantdisplayed a minor reduction in coactivation. Taken together,these results indicate that a functional PHD is an importantcomponent of the coactivator function of PHF8 and that theactivity of the JmjC domain may not be an essential determi-nant, at least in the context of luciferase assays.

PHF8 interacts with the CTD of RNA polymerase II.Genomic localization analysis of PHF8 revealed a preferencefor transcription start sites that are H3K4 methylated andoccupied by RNAPII. We hypothesized that PHF8 could in-teract with preinitiation complex components like the generaltranscription factor TFIID or RNAPII. To examine this, weperformed CoIP experiments with full-length Flag-PHF8 in293T cells. Immunoblot analysis revealed coprecipitation of theRNAPII subunits RPB1 (POLR2A) and RPB3 (POLR2C), but notof the TAF1 subunit of TFIID (Fig. 5A). When we comparedfull-length Flag-PHF8 with truncated versions in CoIPs, theN-terminal half of PHF8 (1-489) displayed a weak associationwith RNAPII, and such an interaction could not be detectedwith the 1-352 version (Fig. 5B). These results indicate that theC-terminal half of PHF8 plays an important role in RNAPIIassociation. This contention is consistent with coimmunoprecipi-tation results of the full-length PHD D28A/W29A mutant and theJmjC H247A/D249A mutant, which interacted with RNAPII sim-ilarly to the wild type (Fig. 5C). Even though larger DNAamounts were transfected, the XLMR F279S mutant showedlower levels of expression and consequently reduced amounts ofprecipitated RNAPII (Fig. 5C), probably due to mislocalizationand aggregation.

The CTD of the largest RNAPII subunit, RPB1, has beenshown to be a binding platform for many proteins (33). To testPHF8 interaction with the CTD, we performed GST pulldownexperiments using GST fused to the CTD (consisting of 27heptad repeats) and lysates from Flag-PHF8-transfected cells.GST-CTD bound full-length PHF8 and truncated 1-489, butnot 1-352, while GST-Taf3 was used as a negative control forthese experiments (Fig. 5D). Interestingly, none of thetested PHF8 mutations (D28A/W29A, H247A/D249A, orF279S) affected association of full-length PHF8 with GST-CTD (Fig. 5E).

FIG. 4. PHF8 coactivates reporter gene expression. (A) U2OS cells weretransfected in triplicate with the 5XGal4MLP-Luc reporter plasmid andpcDNA3 empty vector or pFlag-CMV2 expression plasmid for full-lengthPHF8 or its truncated forms, 1-352 and 1-489, in the presence or absence ofGal4-Ash2, Gal4-myc, Gal4-p53, or Gal4-E2F as indicated. The experimentshown is representative of three biological replicates, with error bars indicat-ing standard deviations (SD) within the experiment. The graphs represent thefold activation relative to transfection with the 5XGal4MLP-Luc reporterplasmid alone. Below, Flag-PHF8 expression was examined by immunoblot-ting. wt, wild type. (B) U2OS cells were transfected as described for panel Awith the pcDNA3 empty vector or pFlag-CMV2 expression plasmid ofPHF8(1-489) or its D28A/W29A (PHDmut), H247A/D249A (JmjCmut), orF279S mutant in the presence or absence of Gal4-E2F as indicated.(C) U2OS cells were transfected as described for panel A with pcDNA3empty vector or pFlag-CMV2 expression plasmid of full-length PHF8 or itsD28A/W29A (PHDmut), H247A/D249A (JmjCmut), or F279S mutant inthe presence or absence of Gal4-Ash2 or Gal4-p53 as indicated.


FIG. 5. PHF8 interacts with the CTD of RNAPII. (A) 293T cells were either mock transfected with pcDNA3 or transfected with pFlag-CMV2containing full-length PHF8. Cell lysates were collected and subjected to immunoprecipitation using anti-Flag beads (IP M2). Precipitated proteinswere analyzed by immunoblotting with the indicated antibodies. 5%inp., 5% input. (B) 293T cells were either mock transfected with pcDNA3 ortransfected with full-length, 1-489, or 1-352 PHF8. Immunoprecipitation and detection of precipitated proteins was performed as described forpanel A. The RPB1 antibodies 8WG16 and H-224 were directed against the CTD and an N-terminal epitope, respectively. (C) 293T cells wereeither mock transfected with pcDNA3 or transfected with pFlag-CMV2 containing full-length PHF8, PHDmut (D28A/W29A), JmjCmut (H247A/D249A), or F279S mutant. Immunoprecipitation and detection of precipitated proteins were performed as described for panel A. (D) Bacteriallysates containing GST fusions with the CTD of the RPB1 subunit of RNAPII or with the PHD of murine Taf3 (as a control) were bound toglutathione-agarose beads. “Empty” indicates that no bacterial lysate was added. The beads were washed and incubated with 293T cell lysatestransfected with full-length PHF8, PHF8(1-489), or PHF8(1-352). Bound proteins were analyzed by immunoblotting using Flag antibodies. Thearrowheads indicate the positions of the PHF8 proteins. (E) GST pulldown experiment with GST fused to the CTD repeats or to UbcH5B (as acontrol) using 293T cell lysates transfected with PHF8-pFlag-CMV2 or the PHD (D28A/W29A), JmjC (H247A/D249A), and F279S mutants asindicated above the lanes. Binding reactions were processed as for panel D. (F) Bacterial lysates containing GST fusions with the CTD repeatsof the RPB1 subunit of RNAPII were bound to glutathione-agarose beads. After being washed, the GST-CTD-coated beads were incubated withor without CDK7/cyclin H/Mat1 or CDK9/cyclin T as indicated above the blots and ATP as indicated below. The beads were washed and incubatedwith 293T cell lysates transfected with PHF8-pFlag-CMV2. Bound proteins were analyzed by immunoblotting using Flag (top) or GST (bottom)antibodies. The positions of PHF8, GST-CTD, and its phosphorylated form are indicated.


Next, we tested whether phosphorylation of the CTD influ-ences its interaction with PHF8. Phosphorylation of serine 5 inthe heptad repeat is conferred by CDK7 and takes place in theinitiation phase near the transcription start site, while phos-phorylation of serine 2 by CDK9 is thought to occur in thetranscribed region during processive elongation (33). Follow-ing preincubation of GST-CTD with the CDK7 or CDK9 ki-nase, we observed the expected mobility shift in CTD depen-dent on ATP addition (Fig. 5F). However, phosphorylation ofthe CTD using either kinase did not alter the interaction withPHF8 (Fig. 5F). These results suggest that PHF8 interactsmainly through its C-terminal half with the CTD of RNAPIIand that this association is not influenced by the phosphoryla-tion status of the CTD. Taken together, the interaction prop-erty of PHF8 with RNAPII is consistent with the genomiclocalization of these proteins and the transcription coactivatorfunction of PHF8.

We also examined the possibility that PHF8 may physicallyinteract with transcriptional activators and coactivators thatwere used for functional-coactivation assays (Fig. 4A). To thisend, we performed CoIPs with lysates from 293T cells thatwere cotransfected with Flag-PHF8 and Gal4-Ash2, Gal4-p53,or c-myc. We were unable to detect efficient association ofPHF8 with any of these proteins (see Fig. S6A and B in thesupplemental material). These results indicate that the coacti-vator function of PHF8 is not mediated by direct associationwith transcription factors, but more likely via its interactionwith RNAPII.

Knockdown of PHF8 has specific but not global effects ongene expression. Next, we investigated the effects of PHF8knockdown on gene transcription. A specific shRNA (PHF8-shA) was able to significantly decrease the PHF8 level in HeLacells 3 days posttransfection compared to a nontargeting con-trol (NT-sh), as shown on the RNA level by quantitative RT-PCR (Fig. 6A) and on the protein level by immunoblotting(Fig. 6B). Total RNAs from three independent knockdown

FIG. 6. Knockdown of PHF8 in HeLa cells using shRNA affectsspecific transcripts. (A) PHF8-shA reduced PHF8 mRNA levels byabout 90% compared with control NT-sh, as determined by quantitative

RT-PCR. PHF8 transcript levels were normalized to ACTB andGAPDH (glyceraldehyde-3-phosphate dehydrogenase) housekeepinggenes using the 2CT method. The values correspond to means �SD (n � 3). (B) PHF8-shA reduced PHF8 protein levels comparedwith NT-sh. HeLa whole-cell lysates were used for immunoblots usinggoat (g) and rabbit (r) PHF8 and mouse (m) �-actin antibodies asindicated. (C) Overview of microarray results. Log2 values of averagednormalized fluorescence ratios (PHF8-shA to NT-sh) were plottedaccording to their ranks for all 41,000 probes. Probes that were morethan 2-fold changed are indicated by arrows (combined, 1,344 probes,or 3.3% of the total). (D) Box-and-whisker plot depicting PHF8 readcounts at promoters in untransfected HeLa cells for upregulated anddownregulated gene sets. The numbers of genes that were PHF8 oc-cupied and their percentages of the annotated genes in the set aregiven below. (E) Box-and-whisker plot depicting transcript levels ofuntransfected HeLa cells (normalized robust multiarray average[RMA] values derived from the work of Cuddapah et al. [10]) for thegene sets that were up- and downregulated upon PHF8 knockdown.The numbers of genes for which expression data were available andtheir percentages of the annotated genes in the set are given below. Inpanels C and D, the central lines represent medians, the hinges rep-resent quartiles, and the whiskers represent minimum and maximumvalues. N. S., no significant difference (according to Student’s t test).Experiments were performed 72 h after shRNA vector transfection.


experiments were compared in two-color microarray dyeswaps; 960 probes were more than 2-fold downregulated, while384 had over 2-fold-increased levels after knockdown (Fig.6C). The down- and upregulated probes corresponded to 780and 270 annotated transcripts, respectively (see Table S3 in thesupplemental material). The observation that almost threetimes more transcripts were down- than upregulated is in linewith a coactivator role of the knockdown target. However,knockdown did not result in a global defect compromisingtranscription of all PHF8-occupied genes. Furthermore, theup- and downregulated genes showed similar percentages ofPHF8 occupancy (about 64%), similar levels of PHF8 reads attheir promoters (Fig. 6D), and comparable transcript levels inuntransfected HeLa cells (Fig. 6E). Functional-annotationanalysis of downregulated transcripts using DAVID (12) re-vealed enrichment for pleckstrin homology domain-containingproteins and proteins that are involved in the focal-adhesionpathway (see Table S3 in the supplemental material). Takentogether, these results indicate that PHF8 does not play a rolein basal transcriptional regulation, as the number of genesdisplaying transcriptional responsiveness to PHF8 depletion isonly a fraction (about 5%) of those occupied by PHF8. It ismore likely that PHF8 is required for the fine tuning of thetranscriptional output, which is governed by the signaling path-ways that regulate the individual PHF8-responsive genes. Thisscenario is consistent with our coactivation function for PHF8,indicating that it could further enhance the activities of diversetranscriptional activators.

Next, we analyzed the genes that displayed downregulationfollowing PHF8 knockdown and that have been reported to belinked to mental retardation. These comprise 11 candidategenes, i.e., MAOA, DPSYL2, CLN5, AP1S2, SMC1A, IDS,SLC9A6, CUL4B, OPHN1, OCRL, and AMMECR1. Usingquantitative RT-PCR, we could validate the downregulation ofthe candidate genes (see Fig. S7A in the supplemental mate-rial). While the less effective short hairpin (PHF8-shB) alsoresulted in downregulation of these genes, the extent of thiseffect was smaller than that seen for the more effective PHF8-shA (see Fig. S7A in the supplemental material). Moreover,while we observed a reduction of PHF8 occupancy at thecandidate gene promoters upon PHF8 knockdown, we couldnot see a consistent change in H3K9me2 or RNAPII levels (seeFig. S7B in the supplemental material). Since the family mem-ber KIAA1718 (also called JHDM1D or KDM7A), which isable to demethylate H3K9me2 and H3K27me2 (19, 20, 49),was induced 2.1-fold upon PHF8 knockdown, it might be com-pensating for the loss of PHF8.

DISCUSSION

The key findings of this study are as follows. First, PHF8 wasdelineated as a novel histone demethylase of the JmjC familyspecific for dimethyl H3K9 and, to a lesser extent, for themonomethyl state. Second, the PHD of PHF8 was identified asan important determinant in recognition of trimethyl H3K4.Third, a critical function for the PHD in histone demethylationand transcriptional coactivation was revealed. Fourth, it wasshown that a single point mutant of PHF8 associated withX-linked mental retardation in humans is defective for histonedemethylation and transcriptional coactivation. Fifth, by ge-

nome-wide location analysis of PHF8, a pattern of promoteroccupancy similar to that of trimethyl H3K4 was revealed.Finally, a direct association between PHF8 and the CTD of thelarge subunit of RNA polymerase II was demonstrated.

We showed that PHF8 is an active H3K9me2 demethylasesimilar to the JHDM2/KDM3 family (54). Loenarz et al.recently additionally reported in vitro demethylation ofH3K27me2 and H3K36me2 using a construct lacking thePHD and the C-terminal domain of PHF8 (31). However, inour study, full-length PHF8 was unable to demethylateH3K27me2 or H3K36me2 and was quite specific for dem-ethylation of di- and monomethyl H3K9. Moreover, overex-pression of PHF8(1-489) did not result in a decrease of anyof the investigated histone modifications but H3K9me2 andH3K9me1 (see Table S1 in the supplemental material). Veryrecently, Feng et al. also reported that PHF8(1-690) has onlymarginal activity on H3K27- or H3K36-methylated peptides(15). H3K9me2 is an important repressive chromatin modifi-cation whose removal signals for coactivation of gene expres-sion. Indeed the importance of the murine H3K9me2 demeth-ylase Jhdm2a for proper activation of target genes by removingthis repressive mark at the respective promoters has alreadybeen demonstrated (34, 47). This demethylation may result inthe release of repressive factors, or it may facilitate the recruit-ment of transcription factors and coactivators.

We found that the PHD of PHF8 plays a critical role in itsdemethylation activity in vivo, as well as its transcriptionalcoactivation function. Our results indicate that the PHD ofPHF8 serves as a specific reader of H3K4 tri- and dimethylmodifications and may act as an important determinant inanchoring PHF8 at transcription start sites of PHF8-occupiedpromoters. Moreover, PHF8 occupies a wide spectrum ofH3K4-trimethylated promoters, suggesting that it might con-tribute to the activating effect of this histone modification ontranscription. This contention is further supported by its abilityto function as a coactivator for a number of transcriptionalactivators that we have examined, including Ash2, p53, c-myc,and E2F. These activators have been linked to recruitment ofthe MLL histone methyltransferase complex to mediate H3K4methylation at promoters (32, 50). Therefore, we propose amodel by which increased H3K4me3 levels, through the actionof an MLL-related family of enzymes, would lead to furtherrecruitment of PHF8 to the promoter of responsive genes.Once at the promoter, PHF8, through its association withRNAPII, may stabilize the preinitiation complex formation,leading to enhanced transcription (Fig. 7).

We found that the disease-causing mutant of PHF8 (F279S)displays aberrant cellular localization and is devoid of de-methylation activity. Furthermore, this mutant exhibited re-duced activity in our coactivation assays. While we did notobserve a general defect in transcription or increasedH3K9me2 levels upon knockdown of PHF8, a specific tran-scriptional defect in neuronal cells may underlie the diseasephenotype of this mutant PHF8. Indeed, XLMR patients dis-play specific defects in the development of neurons and themidline, arguing against a global role for PHF8 in transcrip-tion. Possible reasons for this lack of global effects on methyl-ation levels or transcription following PHF8 mutations couldbe that in most tissues, the loss of PHF8 is compensated for byredundant proteins like JHDM2, PHF2, or KIAA1718, the last


of which is induced by PHF8 knockdown, or that the activityof endogenous PHF8 is strictly regulated by signaling path-ways and switched on in a tissue-, time-, and/or locus-spe-cific manner.

Our data suggested that the interaction with H3K4me3 wascrucial for PHF8 function, since mutation of aromatic-cageresidues in the PHD, as well as removal of H3K4me3 byJARID1A, significantly reduced demethylation and coactiva-tion. It has been shown that activation of neuron-specific genesoccurs by recruitment of the H3K4 methyltransferase MLL1during in vitro differentiation of P19 cells (52). On the otherhand, deletions in the H3K9 methyltransferase EHMT1 causethe 9q34.3 subtelomeric deletion syndrome, one feature ofwhich is mental retardation (24). Additionally, mutations inother PHD protein-coding genes, like ATRX and PHF6, areimplicated in X-linked neurological disorders (3). SeveralH3K4me3 binders, like the inhibitors of growth protein family(ING1 to -5), the Taf3 subunit of TFIID, or the NURF subunitBPTF, have been demonstrated to be involved in gene tran-scription and chromatin remodeling (42, 51, 53). These pro-teins display different affinities for the H3K4me3 mark, andtheir recruitment to promoters could be stimulated via inter-action with other transcription factors, DNA, or chromatin.Another JmjC protein, SMCX/KDM5C, is also implicated inXLMR, but in contrast to PHF8, one of its two PHDs binds

H3K9me3. SMCX has been shown to act as a demethylasespecific for H3K4me3 and a transcriptional repressor (22, 46).This argues for the importance of a balanced H3K4 and H3K9methylation and readout in neuronal development and brainfunction (2). Writers, erasers, or readers exhibit synergistic oropposing roles by interaction at or competition for genomicbinding sites in order to produce the desired transcriptionaloutcome. Disturbances in the fine tuning of this delicate equi-librium brought about by mutations in genes like PHF8 canresult in diseases such as hereditary disorders or cancer.

ACKNOWLEDGMENTS

We thank Kristian Helin, Min Gyu Lee, and Clement Carre forproviding reagents and the Ultrasequencing and Microarrays Units atthe CRG for performing Solexa and microarray analyses. We areparticularly grateful to Marc Vigneron (ESBS, Strasbourg, France) forproviding unpublished RNAPII antibodies and the GST-CTD con-struct. We thank Harm-Jan Vos for help in peptide synthesis, Hettyvan Teeffelen for technical assistance, and Michiel Vermeulen forsharing unpublished results.

K.F. received an Erwin-Schrodinger fellowship from the AustrianScience Fund FWF (J2728-B12). P.D.G., N.S.O., and H.T.M.T.were supported by grants from the Netherlands Organization forScientific Research (NWO-CW TOP 700.57.302), the EuropeanUnion (EUTRACC LSHG-CT-2006-037445), and the NetherlandsProteomics Center. R.S. was supported by a grant from the NIH(CA090758).

We declare that we have no competing financial interest.

REFERENCES

1. Abidi, F. E., M. G. Miano, J. C. Murray, and C. E. Schwartz. 2007. A novelmutation in the PHF8 gene is associated with X-linked mental retardationwith cleft lip/cleft palate. Clin. Genet. 72:19–22.

2. Akbarian, S., and H. S. Huang. 2009. Epigenetic regulation in humanbrain—focus on histone lysine methylation. Biol. Psychiatry 65:198–203.

3. Baker, L. A., C. D. Allis, and G. G. Wang. 2008. PHD fingers in humandiseases: disorders arising from misinterpreting epigenetic marks. Mutat.Res. 647:3–12.

4. Barski, A., S. Cuddapah, K. Cui, T. Y. Roh, D. E. Schones, Z. Wang, G. Wei,I. Chepelev, and K. Zhao. 2007. High-resolution profiling of histone meth-ylations in the human genome. Cell 129:823–837.

5. Bernstein, E., E. M. Duncan, O. Masui, J. Gil, E. Heard, and C. D. Allis.2006. Mouse polycomb proteins bind differentially to methylated histone H3and RNA and are enriched in facultative heterochromatin. Mol. Cell. Biol.26:2560–2569.

6. Breitkreutz, B. J., P. Jorgensen, A. Breitkreutz, and M. Tyers. 2001. AFM4.0: a toolbox for DNA microarray analysis. Genome Biol. 2:SOFT-WARE0001.

7. Cho, Y. W., T. Hong, S. Hong, H. Guo, H. Yu, D. Kim, T. Guszczynski, G. R.Dressler, T. D. Copeland, M. Kalkum, and K. Ge. 2007. PTIP associates withMLL3- and MLL4-containing histone H3 lysine 4 methyltransferase com-plex. J. Biol. Chem. 282:20395–20406.

8. Christensen, J., K. Agger, P. A. Cloos, D. Pasini, S. Rose, L. Sennels, J.Rappsilber, K. H. Hansen, A. E. Salcini, and K. Helin. 2007. RBP2 belongsto a family of demethylases, specific for tri- and dimethylated lysine 4 onhistone 3. Cell 128:1063–1076.

9. Cloos, P. A., J. Christensen, K. Agger, and K. Helin. 2008. Erasing themethyl mark: histone demethylases at the center of cellular differentiationand disease. Genes Dev. 22:1115–1140.

10. Cuddapah, S., R. Jothi, D. E. Schones, T. Y. Roh, K. Cui, and K. Zhao. 2009.Global analysis of the insulator binding protein CTCF in chromatin barrierregions reveals demarcation of active and repressive domains. Genome Res.19:24–32.

11. Dai, J., S. Sultan, S. S. Taylor, and J. M. Higgins. 2005. The kinase haspinis required for mitotic histone H3 Thr 3 phosphorylation and normal meta-phase chromosome alignment. Genes Dev. 19:472–488.

12. Dennis, G., Jr., B. T. Sherman, D. A. Hosack, J. Yang, W. Gao, H. C. Lane,and R. A. Lempicki. 2003. DAVID: Database for Annotation, Visualization,and Integrated Discovery. Genome Biol. 4:P3.

13. De Santa, F., M. G. Totaro, E. Prosperini, S. Notarbartolo, G. Testa, and G.Natoli. 2007. The histone H3 lysine-27 demethylase Jmjd3 links inflamma-tion to inhibition of polycomb-mediated gene silencing. Cell 130:1083–1094.

14. Dominguez, C., A. M. Bonvin, G. S. Winkler, F. M. van Schaik, H. T.Timmers, and R. Boelens. 2004. Structural model of the UbcH5B/CNOT4

FIG. 7. Model for PHF8 coactivator function. (A) Inactive chro-matin bears repressive chromatin marks, like methylated H3K9 orH3K27 (nucleosomes are represented by gray boxes and H3K9me2 byred symbols). (B) Upon induction, transcription factors (TF) bind totarget sites close to the transcription start site and bring in H3K4methylation complexes (MLL; H3K4me3 is indicated by green sym-bols). (C) PHF8 binds to emerging H3K4me3 marks, removes nearbyH3K9me2, helps basal transcription factors to recruit polymerase(RNAPII), and consequently coactivates transcription. However, otherbinders compete for H3K4me3-binding sites, and the activity of PHF8itself may be regulated.


complex revealed by combining NMR, mutagenesis, and docking ap-proaches. Structure 12:633–644.

15. Feng, W., M. Yonezawa, J. Ye, T. Jenuwein, and I. Grummt. 2010. PHF8activates transcription of rRNA genes through H3K4me3 binding andH3K9me1/2 demethylation. Nat. Struct. Mol. Biol. 17:445–450.

16. Goo, Y. H., Y. C. Sohn, D. H. Kim, S. W. Kim, M. J. Kang, D. J. Jung, E.Kwak, N. A. Barlev, S. L. Berger, V. T. Chow, R. G. Roeder, D. O. Azorsa,P. S. Meltzer, P. G. Suh, E. J. Song, K. J. Lee, Y. C. Lee, and J. W. Lee. 2003.Activating signal cointegrator 2 belongs to a novel steady-state complex thatcontains a subset of trithorax group proteins. Mol. Cell. Biol. 23:140–149.

17. Guenther, M. G., S. S. Levine, L. A. Boyer, R. Jaenisch, and R. A. Young.2007. A chromatin landmark and transcription initiation at most promotersin human cells. Cell 130:77–88.

18. Hansen, K. H., A. P. Bracken, D. Pasini, N. Dietrich, S. S. Gehani, A.Monrad, J. Rappsilber, M. Lerdrup, and K. Helin. 2008. A model fortransmission of the H3K27me3 epigenetic mark. Nat. Cell Biol. 10:1291–1300.

19. Horton, J. R., A. K. Upadhyay, H. H. Qi, X. Zhang, Y. Shi, and X. Cheng.2010. Enzymatic and structural insights for substrate specificity of a family ofjumonji histone lysine demethylases. Nat. Struct. Mol. Biol. 17:38–43.

20. Huang, C., Y. Xiang, Y. Wang, X. Li, L. Xu, Z. Zhu, T. Zhang, Q. Zhu, K.Zhang, N. Jing, and C. D. Chen. 2010. Dual-specificity histone demethylaseKIAA1718 (KDM7A) regulates neural differentiation through FGF4. CellRes. 20:154–165.

21. Issaeva, I., Y. Zonis, T. Rozovskaia, K. Orlovsky, C. M. Croce, T. Nakamura,A. Mazo, L. Eisenbach, and E. Canaani. 2007. Knockdown of ALR (MLL2)reveals ALR target genes and leads to alterations in cell adhesion andgrowth. Mol. Cell. Biol. 27:1889–1903.

22. Iwase, S., F. Lan, P. Bayliss, L. de la Torre-Ubieta, M. Huarte, H. H. Qi, J. R.Whetstine, A. Bonni, T. M. Roberts, and Y. Shi. 2007. The X-linked mentalretardation gene SMCX/JARID1C defines a family of histone H3 lysine 4demethylases. Cell 128:1077–1088.

23. Jin, C., C. Zang, G. Wei, K. Cui, W. Peng, K. Zhao, and G. Felsenfeld. 2009.H3.3/H2A.Z double variant-containing nucleosomes mark ‘nucleosome-freeregions’ of active promoters and other regulatory regions. Nat. Genet. 41:941–945.

24. Kleefstra, T., H. G. Brunner, J. Amiel, A. R. Oudakker, W. M. Nillesen, A.Magee, D. Genevieve, V. Cormier-Daire, H. van Esch, J. P. Fryns, B. C.Hamel, E. A. Sistermans, B. B. de Vries, and H. van Bokhoven. 2006.Loss-of-function mutations in euchromatin histone methyl transferase 1(EHMT1) cause the 9q34 subtelomeric deletion syndrome. Am. J. Hum.Genet. 79:370–377.

25. Klose, R. J., E. M. Kallin, and Y. Zhang. 2006. JmjC-domain-containingproteins and histone demethylation. Nat. Rev. Genet. 7:715–727.

26. Koivisto, A. M., S. Ala-Mello, S. Lemmela, H. A. Komu, J. Rautio, and I.Jarvela. 2007. Screening of mutations in the PHF8 gene and identification ofa novel mutation in a Finnish family with XLMR and cleft lip/cleft palate.Clin. Genet. 72:145–149.

27. Kouzarides, T. 2007. Chromatin modifications and their function. Cell 128:693–705.

28. Laumonnier, F., S. Holbert, N. Ronce, F. Faravelli, S. Lenzner, C. E.Schwartz, J. Lespinasse, H. Van Esch, D. Lacombe, C. Goizet, F. Phan-DinhTuy, H. van Bokhoven, J. P. Fryns, J. Chelly, H. H. Ropers, C. Moraine, B. C.Hamel, and S. Briault. 2005. Mutations in PHF8 are associated with X linkedmental retardation and cleft lip/cleft palate. J. Med. Genet. 42:780–786.

29. Lee, M. G., J. Norman, A. Shilatifard, and R. Shiekhattar. 2007. Physical andfunctional association of a trimethyl H3K4 demethylase and Ring6a/MBLR,a polycomb-like protein. Cell 128:877–887.

30. Lee, M. G., R. Villa, P. Trojer, J. Norman, K. P. Yan, D. Reinberg, L. DiCroce, and R. Shiekhattar. 2007. Demethylation of H3K27 regulates poly-comb recruitment and H2A ubiquitination. Science 318:447–450.

31. Loenarz, C., W. Ge, M. L. Coleman, N. R. Rose, C. D. Cooper, R. J. Klose,P. J. Ratcliffe, and C. J. Schofield. 2010. PHF8, a gene associated with cleftlip/palate and mental retardation, encodes for an N epsilon-dimethyl lysinedemethylase. Hum. Mol. Genet. 19:217–222.

32. Luscher-Firzlaff, J., I. Gawlista, J. Vervoorts, K. Kapelle, T. Braunschweig,G. Walsemann, C. Rodgarkia-Schamberger, H. Schuchlautz, S. Dreschers,E. Kremmer, R. Lilischkis, C. Cerni, A. Wellmann, and B. Luscher. 2008.The human trithorax protein hASH2 functions as an oncoprotein. CancerRes. 68:749–758.

33. Meinhart, A., T. Kamenski, S. Hoeppner, S. Baumli, and P. Cramer. 2005.A structural perspective of CTD function. Genes Dev. 19:1401–1415.

34. Okada, Y., G. Scott, M. K. Ray, Y. Mishina, and Y. Zhang. 2007. Histone

demethylase JHDM2A is critical for Tnp1 and Prm1 transcription and sper-matogenesis. Nature 450:119–123.

35. Pasini, D., K. H. Hansen, J. Christensen, K. Agger, P. A. Cloos, and K.Helin. 2008. Coordinated regulation of transcriptional repression by theRBP2 H3K4 demethylase and Polycomb-Repressive Complex 2. Genes Dev.22:1345–1355.

36. Qiao, Y., X. Liu, C. Harvard, M. J. Hildebrand, E. Rajcan-Separovic, J. J.Holden, and M. E. Lewis. 2008. Autism-associated familial microdeletion ofXp11.22. Clin. Genet. 74:134–144.

37. Robertson, A. G., M. Bilenky, A. Tam, Y. Zhao, T. Zeng, N. Thiessen, T.Cezard, A. P. Fejes, E. D. Wederell, R. Cullum, G. Euskirchen, M. Krzywin-ski, I. Birol, M. Snyder, P. A. Hoodless, M. Hirst, M. A. Marra, and S. J.Jones. 2008. Genome-wide relationship between histone H3 lysine 4 mono-and tri-methylation and transcription factor binding. Genome Res. 18:1906–1917.

38. Rozowsky, J., G. Euskirchen, R. K. Auerbach, Z. D. Zhang, T. Gibson, R.Bjornson, N. Carriero, M. Snyder, and M. B. Gerstein. 2009. PeakSeqenables systematic scoring of ChIP-seq experiments relative to controls. Nat.Biotechnol. 27:66–75.

39. Ruthenburg, A. J., C. D. Allis, and J. Wysocka. 2007. Methylation of lysine4 on histone H3: intricacy of writing and reading a single epigenetic mark.Mol. Cell 25:15–30.

40. Schmid, C. D., and P. Bucher. 2007. ChIP-Seq data reveal nucleosomearchitecture of human promoters. Cell 131:831–833.

41. Schones, D. E., K. Cui, S. Cuddapah, T. Y. Roh, A. Barski, Z. Wang, G. Wei,and K. Zhao. 2008. Dynamic regulation of nucleosome positioning in thehuman genome. Cell 132:887–898.

42. Shi, X., T. Hong, K. L. Walter, M. Ewalt, E. Michishita, T. Hung, D. Carney,P. Pena, F. Lan, M. R. Kaadige, N. Lacoste, C. Cayrou, F. Davrazou, A.Saha, B. R. Cairns, D. E. Ayer, T. G. Kutateladze, Y. Shi, J. Cote, K. F. Chua,and O. Gozani. 2006. ING2 PHD domain links histone H3 lysine 4 methyl-ation to active gene repression. Nature 442:96–99.

43. Shi, Y., F. Lan, C. Matson, P. Mulligan, J. R. Whetstine, P. A. Cole, and R. A.Casero. 2004. Histone demethylation mediated by the nuclear amine oxidasehomolog LSD1. Cell 119:941–953.

44. Siderius, L. E., B. C. Hamel, H. van Bokhoven, F. de Jager, B. van den Helm,H. Kremer, J. A. Heineman-de Boer, H. H. Ropers, and E. C. Mariman.1999. X-linked mental retardation associated with cleft lip/palate maps toXp11.3-q21.3. Am. J. Med. Genet. 85:216–220.

45. Suganuma, T., and J. L. Workman. 2008. Crosstalk among histone modifi-cations. Cell 135:604–607.

46. Tahiliani, M., P. Mei, R. Fang, T. Leonor, M. Rutenberg, F. Shimizu, J. Li,A. Rao, and Y. Shi. 2007. The histone H3K4 demethylase SMCX links RESTtarget genes to X-linked mental retardation. Nature 447:601–605.

47. Tateishi, K., Y. Okada, E. M. Kallin, and Y. Zhang. 2009. Role of Jhdm2ain regulating metabolic gene expression and obesity resistance. Nature 458:757–761.

48. Tsukada, Y., J. Fang, H. Erdjument-Bromage, M. E. Warren, C. H. Borch-ers, P. Tempst, and Y. Zhang. 2006. Histone demethylation by a family ofJmjC domain-containing proteins. Nature 439:811–816.

49. Tsukada, Y., T. Ishitani, and K. I. Nakayama. 2010. KDM7 is a dual de-methylase for histone H3 Lys 9 and Lys 27 and functions in brain develop-ment. Genes Dev. 24:432–437.

50. Tyagi, S., A. L. Chabes, J. Wysocka, and W. Herr. 2007. E2F activation of Sphase promoters via association with HCF-1 and the MLL family of histoneH3K4 methyltransferases. Mol. Cell 27:107–119.

51. Vermeulen, M., K. W. Mulder, S. Denissov, W. W. Pijnappel, F. M. vanSchaik, R. A. Varier, M. P. Baltissen, H. G. Stunnenberg, M. Mann, andH. T. Timmers. 2007. Selective anchoring of TFIID to nucleosomes bytrimethylation of histone H3 lysine 4. Cell 131:58–69.

52. Wynder, C., M. A. Hakimi, J. A. Epstein, A. Shilatifard, and R. Shiekhattar.2005. Recruitment of MLL by HMG-domain protein iBRAF promotes neu-ral differentiation. Nat. Cell Biol. 7:1113–1117.

53. Wysocka, J., T. Swigut, H. Xiao, T. A. Milne, S. Y. Kwon, J. Landry, M.Kauer, A. J. Tackett, B. T. Chait, P. Badenhorst, C. Wu, and C. D. Allis.2006. A PHD finger of NURF couples histone H3 lysine 4 trimethylationwith chromatin remodelling. Nature 442:86–90.

54. Yamane, K., C. Toumazou, Y. Tsukada, H. Erdjument-Bromage, P. Tempst,J. Wong, and Y. Zhang. 2006. JHDM2A, a JmjC-containing H3K9 demeth-ylase, facilitates transcription activation by androgen receptor. Cell 125:483–495.

55. Yu, L., Y. Wang, S. Huang, J. Wang, Z. Deng, Q. Zhang, W. Wu, X. Zhang,Z. Liu, W. Gong, and Z. Chen. 2010. Structural insights into a novel histonedemethylase PHF8. Cell Res. 20:166–173.


phf8 targets histone methylation and rna polymerase ii to...

Documents