|Home | About | Journals | Submit | Contact Us | Français|
Jarid2 was recently identified as an important component of the mammalian Polycomb repressive complex 2 (PRC2), where it has a major effect on PRC2 recruitment in mouse embryonic stem cells. Although Jarid2 is conserved in Drosophila, it has not previously been implicated in Polycomb (Pc) regulation. Therefore, we purified Drosophila Jarid2 and its associated proteins and found that Jarid2 associates with all of the known canonical PRC2 components, demonstrating a conserved physical interaction with PRC2 in flies and mammals. Furthermore, in vivo studies with Jarid2 mutants in flies demonstrate that among several histone modifications tested, only methylation of histone 3 at K27 (H3K27), the mark implemented by PRC2, was affected. Genome-wide profiling of Jarid2, Su(z)12 (Suppressor of zeste 12), and H3K27me3 occupancy by chromatin immunoprecipitation with sequencing (ChIP-seq) indicates that Jarid2 and Su(z)12 have very similar distribution patterns on chromatin. However, Jarid2 and Su(z)12 occupancy levels at some genes are significantly different, with Jarid2 being present at relatively low levels at many Pc response elements (PREs) of certain Homeobox (Hox) genes, providing a rationale for why Jarid2 was never identified in Pc screens. Gene expression analyses show that Jarid2 and E(z) (Enhancer of zeste, a canonical PRC2 component) are not only required for transcriptional repression but might also function in active transcription. Identification of Jarid2 as a conserved PRC2 interactor in flies provides an opportunity to begin to probe some of its novel functions in Drosophila development.
Different and distinct gene expression patterns are established during development, which need to be maintained and regulated. This is important to allow for the integrity of cell identity and thus the functional preservation of tissues and organs. However, at the same time, transcribed loci must be equipped with an intrinsic flexibility to regulate these expression patterns and initiate changes if necessary. The core components that are required for the maintenance of gene expression or gene repression have been characterized quite extensively to date (25, 40). Trithorax and Polycomb group genes play antagonistic roles in determining whether a gene is transcriptionally turned on or off, respectively (17, 34). In Drosophila, so far four distinct complexes, pleiohomeotic repressive complex (PhoRC), Polycomb repressive complex 2 (PRC2), Polycomb repressive complex 1 (PRC1), and recently Polycomb repressive deubiquitinase (PR-DUB) have been described to play a role in Polycomb group-mediated gene repression (1, 39–41, 44). However, little is known about the factors involved in controlling recruitment and activity of these complexes on chromatin or about the mechanisms that drive such changes (38). It should be expected that quite a significant number of proteins would convey Polycomb group-mediated transcriptional changes in order to allow an uncoupling of individual gene activity from that of a group of Polycomb group-controlled genes. Functional redundancy might account for part of the problem to discover such candidates. Furthermore, biochemical approaches might be hindered by the fact that such context-specific and more gene-specific recruiters are contained in only a minor fraction of Polycomb repressive complexes.
Recently, Jarid2 the founding member of the JmjC domain-containing protein family (16) which plays important developmental roles in mice and Drosophila (14, 18, 37), has been characterized as a component of PRC2 in embryonic stem (ES) cells (19, 22, 32, 33, 42, 53). The consensus indicates that in ES cells, PRC2 recruitment to many of its targets requires Jarid2. However, levels of bulk histone 3 trimethylated at K27 (H3K27me3) in ES cells depleted of Jarid2 were reported to be only slightly changed at best. This also holds true when individual PRC2 target genes are analyzed. Even though core components of the PRC2 complex were lost from chromatin in the absence of Jarid2, H3K27me3 was not reproducibly affected to a similar degree. Additionally, gene expression analyses in Jarid2−/− ES cells did not confirm a genome-wide derepression of PRC2 target genes as would be expected for any core component of PRC2 (19).
To further address whether Jarid2 constitutes a core PRC2 component, is involved in recruitment of PRC2 to chromatin, and regulates H3K27 methylation in Drosophila, we have purified a Jarid2 complex from flies and performed a global in vivo analysis of Suppressor of zeste 12 [Su(z)12] and H3K27me3 occupancy in Jarid2 mutant animals. Our data confirm that Drosophila Jarid2 purifies with the core members of the PRC2 complex. In imaginal discs, global H3K27me3 levels are only weakly but reproducibly affected under Jarid2 mutant and Jarid2-overexpressing conditions. Our genome-wide studies suggest that in Drosophila, under physiological conditions, Jarid2 does not appear to be a canonical component of the PRC2 complex as PRC2 recruitment is not altered on most target genes in Jarid2 mutant animals. Interestingly, overexpression of Jarid2 results in reduced Su(z)12 binding and changed chromatin compaction on polytene chromosomes, highlighting a possible role for Jarid2 in altering chromatin architecture. Genome-wide, Jarid2 and Su(z)12 binding correlate very well. However, certain loci, such as Homeobox (Hox) genes, differ significantly from this pattern. Here, Jarid2 occupancy on Polycomb response elements (PREs) is often very low where usually the highest enrichment for Su(z)12 can be observed. Gene expression analyses suggest a PRC2-dependent and -independent role for Jarid2 in transcriptional regulation. Jarid2 appears to be involved in the regulation of a certain number of PRC2 target genes and also transcriptionally controls a subset of genes independently of PRC2. Our data not only imply a function for Jarid2 and PRC2 in transcriptional repression but also support a possible role for both Jarid2 and PRC2 in active transcription on genes that are occupied by these factors.
Genomic DNA encompassing the jarid2 locus contained in an attB-P(acman)-CR-BW plasmid (clone number CH322-118D12) served as a platform for N-terminally FLAG tagging Jarid2. First, the galK-positive and counterselection scheme was used to N-terminally insert a galK cassette in frame of the Jarid2 open reading frame (ORF) according to a previously published protocol (48).
The FLAG-HA sequence with galK homology arms was amplified with the primer pair Jarid2 FLAG A forward and Jarid2 FLAG A reverse and the pair Jarid2 FLAG B forward and Jarid2 FLAG B reverse (see sequences below in “Primers”) using clone CH322-118D12 as a template. The galK cassette was exchanged with the FLAG-HA PCR product with galK homology arms as described previously (48).
Six grams of 6- to 18-h-old embryos were homogenized with a tissue tearer (Tissue-Tearor, model 985370-395; BioSpec Products, Inc.) in 20 ml of buffer A (15 mM HEPES, pH 7.5, 10 mM KCl, 5 mM MgCl2, 0.1 mM EDTA, 0.5 mM EGTA, 350 mM sucrose, 0.5 mM dithiothreitol [DTT], protease inhibitor cocktail [item 05056489001; Roche]) at low speed followed by homogenization in a Wheaton 40-ml homogenizer for 20 strokes with a loose pestle. The homogenate was passed through a 70-μm-pore-size cell strainer (2350; Falcon) and carefully overlaid on a cushion of 15 ml of buffer B (15 mM HEPES, pH 7.5, 10 mM KCl, 5 mM MgCl2, 0.1 mM EDTA, 0.5 mM EGTA, 850 mM sucrose, 0.5 mM DTT, protease inhibitor cocktail [05056489001; Roche]). After centrifugation for 10 min at 5,000 × g at 4°C, the nuclear pellet was resuspended in 5 ml of buffer C (20 mM HEPES, pH 7.5, 420 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 10% glycerol, 0.5 mM DTT, protease inhibitor cocktail [05056489001; Roche]) and incubated on a rotator for 30 min at 4°C. The nuclear extract was centrifuged for 30 min at 14,000 rpm at 4°C. After removal, the salt concentration of the supernatant was adjusted to 300 mM NaCl with 0.4 volumes of buffer C without NaCl. The nuclear extract was incubated on a rotator at 4°C overnight with 200 μl of FLAG-agarose beads (A2220; Sigma) and 10 U/ml Benzonase. FLAG-agarose beads were washed with 25 ml of wash buffer (20 mM HEPES, pH 7.5, 300 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 10% glycerol, 0.1% Triton X-100, 0.5 mM DTT, protease inhibitor cocktail [05056489001; Roche]), incubated for 5 min at 4°C on a rotator and centrifuged for 5 min at 500 rpm at 4°C. The wash step was repeated, and the FLAG-agarose beads where transferred with 1 ml of wash buffer to a 1.5-ml tube and centrifuged for 2 min at 1,000 rpm at 4°C. Elution was performed for 30 min at 4°C with 200 μl of wash buffer containing 200 μg/ml FLAG-peptide following centrifugation for 2 min at 1,000 rpm at 4°C. The elution step was repeated another time, and the elution fractions were pooled and passed over a filter column (732-60008; Bio-Rad). Pooled elution fractions at dilutions of 1/20 and 1/40 were analyzed by silver staining and Western blotting, respectively.
FLAG affinity-purified samples were precipitated with trichloroacetic acid and subjected to Multidimensional protein identification technology (MudPIT) analysis as previously described (50). The tandem mass spectrometry (MS/MS) data set was searched against a protein sequence database containing 18,425 nonredundant Drosophila melanogaster sequences (NCBI, 11 April 2011 release), 177 sequences for usual contaminants, and, to estimate false-discovery rates, 18,602 randomized sequences for each protein entry.
Imaginal discs were dissected in ice-cold phosphate-buffered saline (PBS), pH 7.4, fixed for 20 min at room temperature (RT) in PBS, pH 7.4, with 4% paraformaldehyde by gentle shaking, briefly washed with PBS, pH 7.4, and then blocked with blocking buffer (50 mM Tris, pH 6.8, 150 mM NaCl, 0.5% NP-40, 5% bovine serum albumin [BSA]) for 30 min at RT on a shaker. Primary antibodies were added in wash buffer (50 mM Tris, pH 6.8, 150 mM NaCl, 0.5% NP-40, 1% BSA), and samples were incubated at 4°C overnight. Imaginal discs were washed three times in wash buffer for 20 min at RT by gentle shaking. Secondary antibodies were applied in wash buffer, and samples were incubated for 4 h at RT by gently shaking. Samples were washed three times in wash buffer for 20 min at RT with gentle shaking and fixed in PBS, pH 7.4, with 4% paraformaldehyde with gentle shaking before being mounted in Vectashield (H-1200).
Polytene chromosome squashes were performed as described before (27).
The following antibodies were used: rabbit (rb) anti-Jarid2; Masamitsu Yamaguchi) at 1:1,000 for Western analysis, 1:200 for polytenes, and 1:400 for imaginal discs; rb anti-E(z) [where E(z) is Enhancer of zeste) full-length (Jürg Müller) at 1:2,000 for Western analysis, 1:200 for polytenes, 1:400 for imaginal discs, and 24 μl for chromatin immunoprecipitation with high-throughput sequencing (ChIP-seq); rb anti-Su(z)12 C terminus (Jürg Müller) at 1:200 for polytenes and 40 μl serum for ChIP-seq; rb anti-Su(z)12 full-length (Jürg Müller) for ChIP-seq; rb anti-Pc (25762; Santa Cruz Biotechnology) at 1:50 for polytenes; mouse (m) anti-α-tubulin (T9026, clone DM1A; Sigma,) at 1:2,000 for Western analysis; m antihemagglutinin (anti-HA; MMS-101P; Covance) at 1:2,000 for Western analysis and at 1:100 for polytenes; rb anti-H3K27me1 (07-448; Upstate) at 1:50 for imaginal discs; rb anti-H3K27me2 (24684; Abcam) at 1:50 for imaginal discs; rb anti-H3K27me3 (39155; Active Motif) at 1:100 for polytenes and imaginal discs and 40 μg for ChIP-seq. Secondary antibodies were obtained from Molecular Probes.
For recombineering, the following primers were used: Jarid2 GALK forward, GCGAAAGAAAAGAGGCGAGAGAAAAAGAGGCGCGAAAAGTTCGCAACAAACCTGTTGACAATTAATCATCGGCA; Jarid2 GALK reverse, GTCTCTGTCTCCTTGCGGCGACGCTTGTTGTTGTCTTTGGAGACAACCATTCAGCACTGTCCTGCTCCTT; Jarid2 FLAG A forward, GCAAATCTATACATAATAAAGAAGTGTCCGGCG; Jarid2 FLAG A reverse, CGGGAACATCGTAGGGGTAGCCCTTGTCATCATCGTCCTTGTAGTCCATTTTGTTGCGAACTTTTCGCGCCTCTT; Jarid2 FLAG B forward, CAAGGACGATGATGACAAGGGCTACCCCTACGATGTTCCCGATTACGCTATGGTTGTCTCCAAAGACAACAACAAGCG; Jarid2 FLAG B reverse, GTTGAGCATATGTGGACTGGCGGACT. For quantitative reverse transcription-PCR (RT-PCR), the following primers were used: Jarid2 forward, GTCAAAAGGCCAAGAAGCAG; Jarid2 reverse, TTGCAGGCAGTAAATGTTGC; E(z) forward, CAGCAAGGAACTGGAGGAAG; E(z) reverse, ATCATCTTCGCCCTGTTTTG; drp49 forward, CCAGTCGGATCGATATGCTAA; drp49 reverse, GTTCGATCCGTAACCGATGT; βTub56D forward, TAAAATTCTCGGCGGCTACAA; βTub56D reverse, CGCACCTGGTGGTACATCAG.
The following fly lines were used: E(z)61 (Julia Zeitlinger), FLAG-HA-Jarid2 (this study), Jarid2LA00681 (Bloomington Drosophila Stock Center [Bloomington] 22187), Jarid2e03131/TM6B (Masamitsu Yamaguchi), Jarid2MB00996/TM3 (Bloomington 22958), Jarid2MB00996 FRT80B/TM3 (this study), UAS-E(z)-RNAi (Bloomington 27993; RNAi is RNA interference), act5c-GAL4/TM6B (Bloomington 3954), en-GAL4 UAS-GFP (Andreas Bergmann), and ey-FLP; ubi-GFP FRT80B (Andreas Bergmann).
One gram of larvae was homogenized (Tissue-Tearor, model 985370-395; BioSpec Products, Inc.) on low speed for 2 min in 10 ml of buffer A1 (15 mM HEPES, pH 7.5, 15 mM NaCl, 60 mM KCl, 4 mM MgCl2, 0.5% Triton X-100, 0.5 mM DTT, and complete, EDTA-free protease inhibitors [05056489001 Roche]) plus 1.3 ml of 16% paraformaldehyde (1.8% final) following further homogenization with a Dounce tissue grinder 15 times with a loose pestle (15-ml Dounce; Wheaton). The sample was incubated on a Nutator at room temperature until 15 min after the first homogenization step and then quenched with 1.13 ml of 2.5 M glycine (225 mM glycine final) for 5 min at room temperature. The homogenate was passed through a 70-μm-pore-size cell strainer (2350; Falcon) and centrifuged for 5 min at 2,000 × g at 4°C. The supernatant was aspirated, resuspended in 5 ml of buffer A1, and centrifuged for 5 min at 2,000 × g at 4°C. The supernatant was aspirated again, resuspended in 5 ml of buffer A2 (10 mM Tris HCl pH 7.5, 10 mM EDTA, 0.5 mM EGTA, 0.25% Triton X-100, 0.5 mM DTT, and complete, EDTA-free protease inhibitors [05056489001; Roche]) and centrifuged for 5 min at 2,000 × g at 4°C. The previous step was repeated one more time, and the pellet was resuspended in 4 ml of radioimmunoprecipitation assay (RIPA) buffer (25 mM Tris, pH 7.5, 140 mM NaCl, 1% Triton X-100, 1 mM EDTA, 0.1% SDS, 0.1% Na-deoxycholate, 0.5 mM DTT, and complete, EDTA-free protease inhibitors [05056489001; Roche]) with 0.5% N-lauroylsarcosine. Samples (1.3 ml) were sonicated (Bioruptor; Diagenode) two times for 15 min (50% on/off) in 15-ml hard plastic tubes (430055; Corning) following centrifugation for 20 min at 14,000 rpm at 4°C. The supernatant was kept, and 10 μl of sonicated chromatin was reserved for gel analysis (to check sizing), and 150 μl was used as an input control. The sizing sample was reverse cross-linked at 65°C overnight with 90 μl of Tris-EDTA (TE) buffer and 3 μl of proteinase K (30 mg/μl), and 5 μl of proteinase K was used for the input sample. The remaining chromatin was diluted 5-fold with RIPA buffer and incubated overnight with the respective antibody on a Nutator at 4°C. Sixty microliters of protein A-agarose (15918-014; Invitrogen) was washed in 5 ml of RIPA buffer and centrifuged for 2 min at 1,000 rpm at 4°C. The supernatant was aspirated, and the chromatin sample was added and incubated for 2 h on a Nutator at 4°C. After centrifugation for 2 min at 1,000 rpm at 4°C, the protein A-agarose was transferred into a 1.5-ml tube, washed with 1 ml of RIPA buffer, incubated for 5 min on a Nutator at room temperature, and centrifuged for 2 min at 2,500 rpm at 4°C. The supernatant was aspirated, and the previous washing steps were repeated another seven times. Elution was performed on a Nutator at room temperature for 20 min with 300 μl of elution buffer (0.1 M NaHCO3, 1% SDS) containing proteinase K (1 ml of elution buffer and 5 μl of 30 mg/μl proteinase K), and the sample was centrifuged for 2 min at 2,500 rpm at room temperature. The supernatant was kept, and the elution step was repeated. Elution fractions were pooled and reverse cross-linked at 65°C overnight. DNA was isolated with a Qiagen PCR purification kit and eluted in 50 μl of H2O, and DNA concentration was determined by a Pico Green Assay. Twenty nanograms of DNA was amplified (single-end primers; Illumina) and analyzed on a 2100 Bioanalyzer (Agilent Technologies) before being submitted to sequencing (Genome Analyzer; Illumina). ChIP-seq of eye-antenna imaginal discs was performed according to the method of Papp and Müller (30). Samples were sonicated and processed as described above.
Sequencing reads were acquired through the primary Solexa image analysis pipeline. Here, bases were called, and reads were filtered for quality according to default Solexa standards. Filtered reads were then aligned to the fly genome (University of California, Santa Cruz, D. melanogaster genome version 3 [UCSC dm3]) using the Bowtie (20) alignment tool, version 0.12.7. Only those sequences that matched uniquely to the genome with up to two mismatches were retained for subsequent analysis. Wild-type sequence reads were pooled. Enriched regions of ChIP-seq signal for Jarid2 and Su(z)12 were determined by the MACS (52) peak-finding program, version 1.4.0rc2. Sequence reads for each ChIP-seq data set and their associated whole-cell extract controls were used for the input and control files, respectively. The effective genome size was configured appropriately for the fly data sets, and the P value cutoff was set to 1.00e−08 or an FDR of <1% and a fold change greater than 4. All other MACS parameters were left at default levels. Genes were called bound for Jarid2 and Su(z)12 if an enriched region occurred within 1 kb of an annotated transcript start site for any isoform of a gene from Ensembl, version 63. Based on the correlation of Jarid2 and Su(z)12 enrichment (see Fig. 4B) and random sampling of genes to verify Jarid2 and Su(z)12 occupancy, the list of Jarid2-enriched transcripts was used as the basis to define Jarid2/Su(z)12 cooccupancy.
Read coverage information in the track figures was created using R by extending the reads 150 bases toward the interior of the sequenced fragment and then by computing the number of extended reads in 25-bp windows as the count of extended reads per million reads sequenced ([RPM] counts/million). The resulting coverage object was exported and visualized using the UCSC genome browser (15). Pie charts depicting locations of enriched regions are based on the transcript annotations from Ensembl, version 63, where each peak region was annotated relative to the nearest transcript. Enrichment calculations were based on the ratio of normalized read counts for the ChIP of interest divided by the normalized read counts for the associated control. Su(z)12 enrichments were computed for each transcript in Ensembl, version 63, using 100 bp on both sides of the annotated start site using the whole-cell extract levels. The ratio for H3K27me3 enrichment was computed over the entire ORF length of each transcript using the H3 normalized read counts as the control. Affymetrix probe mappings were mapped to gene identifiers. Enrichment plots consist of all annotated transcripts for each gene mapped or described. Changes in Su(z)12 were determined using calculations of enrichment ratios separately for two Jarid2 mutants and the wild type with their associated whole-cell extracts. The geometric mean of enrichment ratios was computed for the two mutants, and this value was divided by the wild-type enrichment ratio, which results in a ratio that measures gain or loss. Genes were determined to have gained Su(z)12 if the resulting ratio showed a 2-fold increase of enrichment and if the gene was bound in the Jarid2 mutant condition by Su(z)12 using ChIP-seq analysis. Genes were determined to have lost Su(z)12 if the resulting ratio showed a 2-fold decrease of enrichment and if the gene was bound in the wild-type condition by Su(z)12. Changes in H3K27me3 were also determined using calculations of enrichment ratios separately for two Jarid2 mutants and the wild type with associated H3 samples. The geometric mean of enrichment ratios was computed for the two mutants, and this value was divided by the wild-type enrichment value. The average enrichment ratio for each of the two mutants and the wild-type ratio were determined by taking the geometric mean of the wild-type and mutant enrichment ratios. Genes determined to have gained or lost H3K27me3 showed a 2-fold increase or decrease, respectively, of enrichment, and the average enrichment ratio was greater than 0.25.
Eye imaginal discs (~500) were dissected as quickly as possible in cold SFX insect medium (SH30278.02; HyClone) and transferred into a microcentrifuge tube containing 500 μl of SFX medium, followed by centrifugation for 3 min at 300 × g at 4°C in a microcentrifuge. The supernatant was carefully removed, and 0.5 ml of 0.25% trypsin solution with EDTA (solution needs to be at 37°C) was added. After discs were preincubated for 1 min at 37°C, they were mechanically disrupted with a needle (size 25G1) by passage of the solution up and down nine times without creating bubbles. The resulting cell clumps were then incubated for another 1.5 min (maximum, 4 min) at 37°C, and the trypsinization was stopped by the addition of 1 ml of cold SFX medium. The cells were centrifuged for 5 min at 300 × g at 4°C in a microcentrifuge, and the supernatant was carefully removed. The resulting cell pellet was resuspended in 1 ml of Schneider medium with 10% fetal bovine serum (FBS) and submitted for cytometry.
Dissociated cells (eye-antenna imaginal discs) were separated based on their green fluorescent protein (GFP) fluorescing properties and sorted directly into TRIzol (100,000 cells per 1 ml of TRIzol). RNA (from dissociated cells) was purified based on a standard protocol from the Drosophila Genomics Resource Center ([DGRC] Indiana University). The RNA pellet was resuspended in 30 μl of RNase-free water. RNA from third-instar larvae was purified by grinding 20 larvae with a plastic pestle in RLT buffer (RNeasy Kit; Qiagen). After the suspension was passed over a QIAshredder column (Qiagen), samples were processed as described by the manufacturer. The RNA was analyzed with an RNA nano-chip from Agilent.
Affymetrix Drosophila 2 arrays were analyzed in R, version 2.11.1, using the packages affy (10), version 1.26.1, and limma (46), version 3.4.3. Normalization was done using robust multiarray averaging (RMA). Annotation information for the probes was taken from Ensembl, version 63. Differentially expressed genes were called with an unadjusted P value of <0.05 and fold change of at least 1.5 in either direction. Bound probes for MA (where M is the intensity ratio and A is the average intensity for a dot in the plot) plots from fly larvae expression values display bound gene lists for either Su(z)12 or Jarid2 from larval ChIP-seq analysis; eye disc MA plots display bound probes from the Su(z)12 gene list from eye disc ChIP-seq.
Sequencing and expression data sets described in this study have been deposited at the National Center for Biotechnology Information Gene Expression Omnibus under accession number GSE36039.
In order to purify endogenous Jarid2 complexes, we used recombineering technology to construct a fly line that contains a FLAG-HA sequence at the N terminus of the Jarid2 ORF (see Materials and Methods). The resulting FLAG-HA-Jarid2 line was able to rescue lethality of the two strongest available Jarid2 mutants, Jarid2MB00996 and Jarid2e03131 (data not shown; see also Fig. S1A to C in the supplemental material). Also, colocalization studies on polytene chromosomes confirm that FLAG-HA-Jarid2 shows a distribution pattern similar to that of untagged endogenous Jarid2 (Fig. 1A), demonstrating that in the tagged line Jarid2 is not mislocalized. FLAG immunopurification was performed on extracts from 6- to 18-h-old Drosophila embryos to isolate Jarid2 and any associated proteins (Fig. 1B to D). However, full-length FLAG-HA-Jarid2 appears to be highly unstable as only a truncated form of FLAG-HA-Jarid2 could reproducibly be detected on a silver-stained SDS-PAGE gel (Fig. 1B, lane 2, asterisk) or in the purified fraction after FLAG peptide elution on a Western blot (Fig. 1C, lane 4, upper panel). In contrast, nuclear extracts from Drosophila embryos contain the full-length form of FLAG-HA-Jarid2 (~250 kDa) as well as the truncated version that runs around 130 kDa (Fig. 1C, lane 2, upper panel).
Multidimensional protein identification technology (MudPIT) analysis readily detects the PRC2 core components Enhancer of zeste [E(z)], Suppressor of zeste 12 [Su(z)12], Extra sex combs (Esc), and chromatin assembly factor 1 (Caf1, also known as p55 or Nurf55) as factors that copurify with Jarid2 (Fig. 1D). The relative quantities of PRC2 core components appear to be slightly higher than those of FLAG-HA-Jarid2 because the truncated version of FLAG-HA-Jarid2 might possibly result in fewer detectable peptides, and the relative abundance calculation assumes full-length Jarid2. The interaction of FLAG-HA-Jarid2 and E(z) was further confirmed by Western blotting (Fig. 1C, lane 4, lower panel). However, Jarid2 colocalizes only to some degree with E(z) on polytene chromosomes (Fig. 1E), and their banding patterns differ significantly (compare Fig. 1E′ with E″). In addition to these core components of PRC2, another protein, Jing, which is the Drosophila homolog of mammalian AEBP2, was copurified with FLAG-HA-Jarid2 (Fig. 1D) (3, 6, 23). In summary, these studies demonstrate that Jarid2 is a conserved member of PRC2 complexes in Drosophila but differs significantly in its distribution pattern from canonical PRC2 core components such as E(z) on polytene chromosomes.
To determine the role of Jarid2 in PRC2 function in an in vivo system, we studied the effect of Jarid2 overexpression (Fig. 2A to D) and of Jarid2 mutations (Fig. 2E) on H3K27 methylation, the mark implemented by PRC2. Overexpression of Jarid2 was confirmed by a Jarid2 antibody (Fig. 2A). The white arrow points out the posterior compartment where Jarid2 is overexpressed (Fig. 2A′). Jarid2 overexpression results in reduction of H3K27me3 levels (compare Fig. 2B′, arrow, with C′, arrow), but no changes for H3K27me1 could be observed (Fig. 2D′, arrow). To assess H3K27me3 in Jarid2 mutant tissue, we created Jarid2 eye-antenna imaginal disc mosaics consisting of patches of Jarid2 mutant tissue (Fig. 2E, no GFP expression and therefore negatively marked) and wild-type tissue (Fig. 2E, expresses GFP and therefore marked in green) within the same disc. This allowed us to directly compare H3K27me3 levels between Jarid2 mutant tissue and wild-type tissue side by side under exactly the same experimental conditions. Our studies show that H3K27me3 is increased in Jarid2 mutant clones in the posterior region of eye-antenna imaginal discs where photoreceptor differentiation is in progress (Fig. 2E′; two large Jarid2 mutant clones are outlined with dashed lines and highlighted by arrows). No modifications in Jarid2 mutant clones were observed for other methyl marks tested (see Table S1 in the supplemental material). Quantitative RT-PCR experiments confirmed a very strong decrease of Jarid2 mRNA in Jarid2 mutant animals that were used to create mutant clones in eye-antenna imaginal discs (see Fig. S1A and B in the supplemental material) compared to wild-type controls (see Fig. S1A and B). This is also evident on the protein level as assessed by Western blotting (see Fig. S1C). Even though a global increase and reduction in H3K27me3 could be reproducibly detected under both Jarid2 mutant (Fig. 2E′, arrows) and Jarid2-overexpressing conditions, respectively (Fig. 2B′, white arrow), they are relatively weak and do not compare to the effects observed by RNA interference (RNAi) knockdown of E(z) in the posterior half of wing imaginal discs (Fig. 2F), which essentially eliminates all H3K27me3 (Fig. 2F′, arrow; see also Fig. S1D and E). However, these studies support the biochemical data in that Jarid2 only alters the levels of H3K27 methylation, as expected of a modulator of the PRC2 complex.
In order to assess a possible role of Jarid2 in recruiting PRC2 complexes to chromatin, we compared Jarid2 (see Fig. S2A to D in the supplemental material), Su(z)12 (Fig. 3A to D) and H3K27me3 (Fig. 3E to H) and Polycomb (Pc) (Fig. 3I to L) localization patterns on polytene chromosomes in the wild type (Fig. 3A, E, and I; see also Fig. S2A), Jarid2 mutant (Fig. 3B, F, and J; see also Fig. S2B), and under Jarid2-overexpressing (Fig. 3C, G, and K; see also Fig. S2C) and E(z) mutant (Fig. 3D, H, and L; see also Fig. S2D) conditions. Jarid2 is widely distributed in wild-type animals (see Fig. S2A′) and is completely lost from chromatin in Jarid2 mutants (see Fig. S2B′). Surprisingly, Su(z)12 binding to chromatin (compare Fig. 3A′ and B′) and H3K27me3 levels (compare Fig. 3E′ and F′) are not detectably affected in Jarid2 mutants. Similarly, the pattern of Pc (compare Fig. 3I′ and J′), a core component of Polycomb repressive complex 1 (PRC1), is not considerably altered in Jarid2 mutants. This seems to suggest that, at least in Drosophila, Jarid2 is not globally required for recruitment of the PRC2 or PRC1 complex to Polycomb response elements (PREs).
Upon overexpression, Jarid2 binding to chromatin is strongly enhanced (see Fig. S2C′ in the supplemental material), and in some cases, the polytene chromosomes appear shortened, bloated, and poorly banded, indicating defects in higher-order chromosome compaction (Fig. 3C; see also Fig. S2C). In other cases this bloating phenotype is weaker (Fig. 3K) or not distinctly observable (Fig. 3G). Similar bloating phenotypes have been described for chromatin remodelers such as Imitation SWI (Iswi) (5), and it has been shown that mutants of another chromatin remodeling factor, the trithorax group protein Kismet (Kis), globally affect H3K27 methylation (47). This phenotype is interesting in the light of a recent report which suggests a genetic interaction between yet another chromatin remodeler, brahma (brm), and Jarid2 (7), providing further evidence for a possible role of Jarid2 in regulating higher-order chromatin structure. Interestingly, in Jarid2-overexpressing larvae, both Su(z)12 recruitment (compare Fig. 3A′ and C′) and H3K27me3 levels (compare Fig. 3E′ and G′) are negatively affected. Pc localization on polytene chromosomes was marginally reduced when Jarid2 was overexpressed (compare Fig. 3I′ and K′).
The concept that PRC2 is generally not required for recruitment of PRC1 was further confirmed when E(z)61 mutants were analyzed (4). Upon a temperature shift to 29°C for 24 h, E(z)61 mutant third-instar larvae show almost a complete loss of Su(z)12 from polytene chromosomes (compare Fig. 3A′ and D′) and consequently H3K27me3 is lost (Fig. 3H′). Nevertheless, Pc recruitment is not appreciably altered (compare Fig. 3I′ and L′). This strongly supports a PRC2-independent recruitment of PRC1 to chromatin (see also references 29 and 30) and furthermore suggests that at least in vivo the H3K27me3 mark is not the primary determinant for Pc binding, in contrast to previous proposals based largely on in vitro data (9, 26). The absence of E(z) does also not impair Jarid2 localization on polytene chromosomes (compare Fig. S2A′ and D′ in the supplemental material), suggesting that Jarid2 can be recruited to chromatin independently of PRC2.
To address whether Jarid2 affects recruitment of PRC2 at a molecular level, ChIP-seq experiments for Jarid2, Su(z)12, and H3K27me3 in the wild type and Jarid2 mutants were performed. Most Jarid2 and Su(z)12 peaks are found at transcription start sites (TSSs) (Fig. 4A; see also Fig. S3A in the supplemental material). Of the peaks that were detected upstream, downstream, or inside genes, some could define TSSs of novel transcription units, as recently proposed for Polycomb repressive complex 1 (PRC1) (8). In wild-type larvae the binding patterns of Jarid2 and Su(z)12 correlate very well with each other (Pearson coefficient, 0.76) (Fig. 4B) except for a small group of genes, which display high Su(z)12 occupancy and low Jarid2 occupancy (Fig. 4B, highlighted in red). Interestingly, in this class, 7 out of 30 genes are Hox genes (see Table S2 in the supplemental material). The binding patterns of Jarid2 and Su(z)12 on the Bithorax complex illustrate this finding (Fig. 4C). Occupancy of canonical PRC2 components such as Su(z)12 is usually very high on this locus and was distributed in several peaks (Fig. 4C, red tracks). At many of these highly occupied Su(z)12 sites, Jarid2 occupancy is relatively low (Fig. 4C, compare red with green tracks). This indicates a fundamentally different Jarid2 binding behavior than that of canonical PRC2 members for some of the major PRC2 target genes such as Hox genes.
Despite substantial colocalization of Jarid2 and Su(z)12 genome-wide, only comparatively few genes show significant changes in Su(z)12 binding (Fig. 4C, red tracks) in Jarid2 mutants, confirming our observations on polytene chromosomes. No general effects on Su(z)12 binding were apparent when wild-type Jarid2 mutant tissues in eye imaginal discs were compared (data not shown). Nonetheless, in a few cases, a significant increase or reduction (≥2 or ≤0.5) of Su(z)12 occupancy in Jarid2 mutants can be observed (Fig. 4C, red tracks, black arrows). In contrast to the mammalian findings, Su(z)12 occupancy can also be increased on certain genes in Jarid2 mutants (Fig. 4C, red tracks, left arrow). Neither genes with reduced Su(z)12 occupancy (Fig. 4C, red tracks, right arrow; see also Fig. S3B in the supplemental material) nor genes with increased Su(z)12 occupancy (Fig. 4C, red tracks, left arrow; see also Fig. S3C) displayed any significant changes (≥2 or ≤0.5) in H3K27me3 enrichment in Jarid2 mutants (Fig. 4C, blue tracks) or represented genes with no or low H3K27me3 enrichment (see Fig. S3B and C, upper two tracks). Although genes with changes in Su(z)12 occupancy and the majority of other genes [Su(z)12/Jarid2 bound and unbound] do not generally display significant changes (≥2 or ≤0.5) in H3K27me3 enrichment, a relatively big group of genes shows a very weak but statistically observable increase in H3K27me3 in Jarid2 mutant versus wild-type larvae (see Fig. S3D, dashed box). Likewise, the sum of these small increases in H3K27me3 enrichment of a large group of genes might be the reason why H3K27me3 is globally increased in Jarid2 mutant eye-antenna imaginal disc clones (Fig. 2E).
To gain further insight as to how Jarid2 regulates the expression of its target genes, we performed gene expression analyses from wild-type, Jarid2 mutant, and E(z)-RNAi larvae (Fig. 5; see Fig. S4 and S5 in the supplemental material), A total of 204 Jarid2/Su(z)12 cobound genes showed at least a 1.5-fold gain (Fig. 5A, highlighted in blue), and 136 Jarid2/Su(z)12 cobound genes had a more than 1.5-fold loss (Fig. 5A, highlighted in red) in expression in Jarid2 mutant larvae. Genes with a significant increase/reduction in Su(z)12 localization and/or increased/reduced H3K27me3 enrichment in Jarid2 mutants (≥2 or ≤0.5) did not generally correlate with gene expression changes (data not shown). Not all Jarid2/Su(z)12 cobound genes that are transcriptionally changed in Jarid2 mutants are also modified in E(z)-RNAi larvae (Fig. 5B) and vice versa see Fig. S4A and B). We find evidence for coregulation of PRC2 target genes by Jarid2 and E(z) (Fig. 5C and D show individual examples), confirming a role for Jarid2 in PRC2-mediated regulation of transcription. Jarid2 and E(z) only not appear to function in maintenance of transcriptional repression (Fig. 5C) but are also possibly involved in activation of common target genes (Fig. 5D). However, the possibility exists that a gene might be cobound by Jarid2/Su(z)12 and yet would not directly be transcriptionally controlled by these factors. Despite the binding of Jarid2/Su(z)12, a change in transcription in the corresponding mutant/RNAi animal could be a secondary consequence of other genes that are under direct control of Jarid2/Su(z)12. Generally, Jarid2/Su(z)12 cobound genes that show transcriptional changes in Jarid2 mutants and/or E(z)-RNAi larvae do not correlate with changes (≥2 or ≤0.5) in Su(z)12 occupancy (Fig. 5E) and/or changes in H3K27me3 enrichment (Fig. 5F) in Jarid2 mutants. We also identified Jarid2/Su(z)12 cobound genes that are regulated by Jarid2 independently of canonical PRC2 members [E(z)] (see Fig. S4C and D) and target genes solely controlled by canonical PRC2 members [E(z)] independently of Jarid2 (see Fig. S4E and F).
It needs to be stressed that both ChIP-seq experiments and gene expression analyses have been carried out in larvae which represent a mixed population of cells. This could prevent the identification of genes that are regulated by Jarid2 in a tissue-specific manner. Alterations in Su(z)12 occupancy and H3K27me3 enrichment might be masked because only a subpopulation of cells is controlled by Jarid2. Therefore, we also performed ChIP-seq studies for Su(z)12 occupancy and gene expression analyses in eye-antenna imaginal discs (see Fig. S5A and B in the supplemental material). Similar to the results obtained in larvae, gene expression changes in Jarid2 mutant eye-antenna imaginal disc clones usually do not correlate with altered Su(z)12 occupancy (see Fig. S5B). Taken together, these results imply a role for Jarid2 in PRC2-dependent and -independent regulation of transcription. Our data support a role for Jarid2 and PRC2 in maintenance of transcriptional repression (Fig. 5C) and suggest the possibility for both Jarid2 and PRC2 to function in transcriptional activation or elongation (Fig. 5D). For most of its target genes, Jarid2 seems to carry out its role in transcriptional regulation independently of changes in H3K27me3 or PRC2 recruitment. However, the general tendency implies that most Jarid2/PRC2-coregulated genes have relatively low levels of H3K27me3 (Fig. 5F).
In this study, we describe the purification of a Jarid2 complex in Drosophila. Consistent with previous results in mammalian systems, we find that Jarid2 is a component of PRC2 (Fig. 1) (19, 22, 32, 33, 42, 53). We provide evidence that in imaginal discs (Fig. 2) and on polytene chromosomes (Fig. 3), Jarid2 is required to fine-tune global H3K27me3 levels. Jarid2 might accomplish this by modulating the activity of the core complex [E(z), Su(z)12, Esc, and Caf1]. Our data indicate that Jarid2 could play an inhibitory role in the implementation of H3K27me3 as Jarid2 mutant imaginal disc clones display a global increase (Fig. 2E) and as overexpression of Jarid2 results in a reduction in H3K27me3 (Fig. 2B and and3G).3G). Despite having a JmjC domain, Jarid2 has been predicted (16) and reported (22, 42) to be catalytically inactive as a histone demethylase. Therefore, it is unlikely but not impossible that it could function in this manner toward H3K27me3, thereby counteracting PRC2 activity. Even if Jarid2 would be inactive as a histone demethylase, it might still be able to bind to chromatin and prevent spreading of the H3K27me3 mark, such as opposing a possible positive spreading effect of Esc (EED in mammals) (24).
Furthermore, even though Jarid2 could be purified with the PRC2 core members (Fig. 1) and its occupancy generally correlates very well with canonical PRC2 components such as Su(z)12 (Fig. 4B), it does not appear to play a significant role in regulating PRC2 recruitment in a physiological context, as assessed by our Jarid2 mutant animal studies (Fig. 3 to to4).4). Apparent differences with published mammalian studies, which imply a major role for Jarid2 in recruitment of PRC2, could be explained by variation in the mechanisms employed or by the fact that the recruitment of PRC2 in ES cells generally differs from that in differentiated tissues. For example, PREs have been known to be highly effective in recruiting PRC2 to target sites in Drosophila (36, 43). In mammals, attempts have been made to identify functionally analogous sequences but with only limited success (45, 49). Indeed, it seems more likely that the recruitment of PRC2 in mammals not only requires specific sequences but is also more dependent on additional factors (proteins and RNA), which might explain why PRC2 recruitment is more strongly affected in Jarid2-depleted cells and why PRC1 recruitment in some instances appears to be dependent on PRC2 (H3K27me3) (2). However, our data in Drosophila salivary glands and that of other groups suggest that recruitment of PRC2 (and methylation of H3K27) is not a prerequisite for targeting of PRC1 (Fig. 3L) (29, 30), and the generality of this mechanism is also increasingly questioned in the mammalian system (35). Nonetheless, when Jarid2 is overexpressed in Drosophila, changes in chromosome compaction can be observed (Fig. 3C; see also Fig. S2C in the supplemental material). Under these conditions, Jarid2 extensively occupies the chromosomes (see Fig. S2C), and Su(z)12 localization and H3K27me3 are negatively affected (Fig. 3C and G). It is possible that increasing Jarid2 levels beyond a certain physiological level might interfere with PRC2 integrity. Larger amounts of Jarid2 might alter the stoichiometry of the PRC2 subunits, resulting in destabilization of the PRC2 complex on chromatin.
Jarid2 also behaves differently from other canonical PRC2 members in Drosophila, as is evident from its binding pattern on certain Hox genes (see Table S2 in the supplemental material). Here, occupancy of PRE sites by canonical PRC2 members is one of the highest in the whole genome (Fig. 4C, red tracks). In contrast, Jarid2 displays relatively low occupancy on many of these loci (Fig. 4C, compare red with green tracks), implying a minor or different function for Jarid2 in controlling transcription of these well-described PRC2 targets. It is also possible that at these loci Jarid2 has a more transient association or even that it is less accessible to interact with the antibodies that we have generated. However, our findings are also in agreement with modifier screens that have been performed in Drosophila to identify major regulators of Polycomb group-mediated phenotypes but that were unable to capture Jarid2 (11).
Additionally, our data suggest that Jarid2 appears to control PRC2-dependent transcription (Fig. 5; see also Fig. S4 in the supplemental material), although not necessarily in the same way as expected for canonical PRC2 members. For example, in contrast to the mammalian findings (19, 22, 32, 33, 42, 53), we observe that PRC2-mediated transcriptional regulation by Jarid2 in Drosophila is generally independent of changes in Su(z)12 occupancy (Fig. 5E; see also Fig. S5B) and does not correlate with changes in H3K27me3 enrichment (Fig. 4C and and5F).5F). However, it needs to be stressed that most Jarid2/PRC2 cobound genes with altered expression patterns in Jarid2 mutants and E(z)-RNAi larvae contain no or low levels of H3K27me3 (Fig. 5F), which is in contrast to the mammalian system where PRC2 components are usually found only at genes with high H3K27me3 enrichment (2, 21). Nonetheless, in Drosophila, genes with high H3K27me3 enrichment exist that change in transcription in Jarid2 mutants and E(z)-RNAi animals (Fig. 5F), demonstrating that H3K27me3 is not necessarily instructive for transcriptional repression per se. To date most of the evidence ascribing to H3K27me3 the role of a repressive mark is based on correlation from the observation that PRC2 components colocalize with H3K27me3 and that the respective genes seem to be transcriptionally silenced. Our data imply that this might generally be the case but that there are also exceptions to the rule. That certain H3K27me3 patterns can also be connected to transcriptionally active genes in mammals has just recently been reported (51).
Finally, our results imply that Jarid2 and PRC2 are not only involved in maintenance of gene repression (Fig. 5C) but could also function in active transcriptional processes such as transcriptional activation of elongation (Fig. 5D). This is in agreement with previous reports (13, 28, 29, 31) and demonstrates that PRC2 has cellular functions that extend beyond what we have learned from its role at Hox genes. Importantly, our studies also suggest that despite a very good correlation of Jarid2 and Su(z)12 occupancies (Fig. 4B), Jarid2 might function in transcriptional repression and activation independently of the canonical PRC2 complex [E(z)] (see Fig. S4C and D in the supplemental material) and vice versa (see Fig. S4E and F). This distinction in target genes between Jarid2 and canonical PRC2 components [E(z)] provides additional confirmation that Jarid2 in some respects behaves fundamentally differently than the canonical PRC2 complex. Together with the varied functions proposed for Jarid2 in mammals (12), our studies highlight the diverse aspects of Jarid2 function in PRC2-mediated gene regulation.
We are very thankful to Jürg Müller for providing E(z) and Su(z)12 antibodies, to Julia Zeitlinger for the E(z)61 fly line, to Andreas Bergmann for the ey-FLP; ubi-GFP FRT80B and en-GAL4 UAS-GFP stocks, and to the Bloomington Stock Center for various fly lines (see “Fly lines” above). We also thank Kathryn Wagner, Rhonda Egidy, and Allison Peak for their help with the gene expression analyses and Dale Dorsett and Edwin Smith for helpful comments and suggestions.
H.-M.H. is a fellow of The Jane Coffin Childs Memorial Fund for Medical Research. Studies in the manuscript were supported in part by JSPS grants to M.Y. and NIH R01CA150265 to A.S.
Published ahead of print 21 February 2012
Supplemental material for this article may be found at http://mcb.asm.org/.