OR I G I N A L R E S E A R CH Genome‐wide occupancy of histone H3K27 methyltransferases CURLY LEAF and SWINGER in Arabidopsis seedlings Jie Shu1,2,* | Chen Chen1,2,* | Raj Kumar Thapa1,2 | Shaomin Bian1,3 | Vi Nguyen1 | Kangfu Yu4 | Ze-Chun Yuan1 | Jun Liu5 | Susanne E. Kohalmi2 | Chenlong Li6 | Yuhai Cui1,2 1London Research and Development Centre, Agriculture and Agri-Food Canada, London, Ontario, Canada 2Department of Biology, Western University, London, Ontario, Canada 3College of Plant Science, Jilin University, Changchun, China 4Harrow Research and Development Centre, Agriculture and Agri-Food Canada, Harrow, Ontario, Canada 5Guangdong Academy of Agricultural Sciences, Guangzhou, China 6State Key Laboratory of Biocontrol and Guangdong Key Laboratory of Plant Resources, School of Life Sciences, Sun Yat-sen University, Guangzhou, China Correspondence Chenlong Li, State Key Laboratory of Biocontrol and Guangdong Key Laboratory of Plant Resources, School of Life Sciences, Sun Yat-sen University, Guangzhou, China. Email: lichlong3@mail.sysu.edu.cn and Yuhai Cui, London Research and Development Centre, Agriculture and Agri- Food Canada, London, ON, Canada. Email: yuhai.cui@canada.ca Funding information Natural Science and Engineering Research Council of Canada, Grant/Award Number: RGPIN/04625-2017; Agriculture and Agri- Food Canada; National Natural Science Foundation of China, Grant/Award Number: 31371715; China Scholarship Council Abstract The Polycomb Group (PcG) proteins form two protein complexes, PcG Repressive Complex 1 (PRC1) and PRC2, which are key epigenetic regulators in eukaryotes. PRC2 represses gene expression by catalyzing the trimethylation of histone H3 lysine 27 (H3K27me3). In Arabidopsis (Arabidopsis thaliana), CURLY LEAF (CLF) and SWINGER (SWN) are two major H3K27 methyltransferases and core components of PRC2, playing essential roles in plant growth and development. Despite their importance, genome‐wide binding profiles of CLF and SWN have not been deter- mined and compared yet. In this study, we generated transgenic lines expressing GFP‐tagged CLF/SWN under their respective native promoters and used them for ChIP‐seq analyses to profile the genome‐wide distributions of CLF and SWN in Ara- bidopsis seedlings. We also profiled and compared the global H3K27me3 levels in wild‐type (WT) and PcG mutants (clf, swn, and clf swn). Our data show that CLF and SWN bind to almost the same set of genes, except that SWN has a few hundred more targets. Two short DNA sequences, the GAGA‐like and Telo‐box‐like motifs, were found enriched in the CLF and SWN binding regions. The H3K27me3 levels in clf, but not in swn, were markedly reduced compared with WT; and the mark was undetectable in the clf swn double mutant. Further, we profiled the transcriptomes in clf, swn, and clf swn, and compared that with WT. Thus this work provides a use- ful resource for the plant epigenetics community for dissecting the functions of PRC2 in plant growth and development. K E YWORD S Arabidopsis, CURLY LEAF, SWINGER, genome-wide occupancy, H3K27me3, gene repression *These authors contributed equally to this work. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2019 The Authors. Plant Direct published by American Society of Plant Biologists, Society for Experimental Biology and John Wiley & Sons Ltd. Received: 5 November 2018 | Revised: 14 November 2018 | Accepted: 19 November 2018 DOI: 10.1002/pld3.100 Plant Direct. 2019;1–14. wileyonlinelibrary.com/journal/pld3 | 1 https://orcid.org/0000-0003-0359-6420 https://orcid.org/0000-0003-0359-6420 https://orcid.org/0000-0003-0359-6420 https://orcid.org/0000-0002-3675-3288 https://orcid.org/0000-0002-3675-3288 https://orcid.org/0000-0002-3675-3288 http://creativecommons.org/licenses/by/4.0/ http://www.wileyonlinelibrary.com/journal/PLD3 1 | INTRODUCTION The Polycomb Group (PcG) proteins play crucial roles in epigenetic regulation through maintaining the repressed state of target genes (Dellino et al., 2004). Initially described in Drosophila melanogaster as regulators of gene expression (Lewis, 1978), PcG proteins form two functionally distinct multi‐protein complexes known as Polycomb Repressive Complex 1 and 2 (PRC1 and PRC2) (Margueron & Rein- berg, 2011). PRC2 catalyzes the trimethylation of histone H3 lysine 27 (H3K27me3), a key repressive epigenetic mark in higher organ- isms (Cao et al., 2002; Hansen et al., 2008; Lafos et al., 2011); whereas PRC1 catalyzes the monoubiquitination of histone H2A lysine 119 (H2AK119ub) (Cao & Zhang, 2004). The subunits of PRC2 are highly conserved in multicellular organ- isms, but the number of genes that encode each subunit varies among species (Mozgova & Hennig, 2015). The Drosophila PRC2 has four main components: Enhancer of zeste (E(z)), Suppressor of zeste 12 (Su(z)12), Extra sex comb (Esc), and p55 (Margueron & Reinberg, 2011; Schwartz & Pirrotta, 2007). In human, two copies of E(z)— EZH1 and EZH2 exist (Ciferri et al., 2012). In Arabidopsis, there are three homologous genes for the E(z) subunit, CURLY LEAF (CLF), SWINGER (SWN), and MEDEA (MEA); three for Su(z)12, EMBRYONIC FLOWER 2 (EMF2), VERNALIZATION 2 (VRN2), and FERTILIZATION INDEPENDENT SEED 2 (FIS2); one gene for Esc, FERTILIZATION INDEPENDENT ENDOSPERM (FIE); and five genes for p55, MULTI- COPY SUPRESSOR OF IRA 1-5 (MSI1-5) (Whitcomb, Basu, Allis, & Bernstein, 2007). In Arabidopsis, PRC2 proteins have been shown to play vital roles in many growth and developmental processes (Bouyer et al., 2011; Jiang, Wang, Wang, & He, 2008; Li et al., 2015; Muller‐Xing, Clarenz, Pokorny, Goodrich, & Schubert, 2014; Tang, Lim, et al., 2012). For instance, CLF has long been known to be necessary for the control of leaf and flower morphology, likely through repressing AGAMOUS (AG) and SHOOT MERISTEMLESS (STM) (Goodrich et al., 1997; Schu- bert et al., 2006). CLF is also necessary for WUSCHEL (WUS) repres- sion, contributing to the termination of floral stem cells (Liu et al., 2011), and is required for maintaining root meristem activity by antagonizing the chromatin remodeler PICKLE (PKL) (Aichinger, Vil- lar, Di Mambro, Sabatini, & Köhler, 2011). Unlike the clf mutants that display severe phenotypes including dwarfism, curly leaf, and early flowering, the swn mutants only show weak changes during vegeta- tive phase transition (Xu, Hu, Smith, & Poethig, 2016; Xu, Guo, et al., 2016). However, the clf swn double mutants lose the capacity to dif- ferentiate and form massive somatic embryo‐like structures (Chanvi- vattana et al., 2004; Farrona et al., 2011; Lu, Cui, Zhang, Jenuwein, & Cao, 2011). These prior works have clearly demonstrated the importance of CLF and SWN in plant development, but mostly based on mutant examination and single gene analyses. Meanwhile, although a number of H3K27me3 genome‐wide profiling have been reported (Carter et al., 2018; Cui et al., 2016; Li et al., 2015, 2016; Luo et al., 2013; Wang et al., 2016; Yang et al., 2016), works on genome‐wide occupancy of the CLF and SWN proteins are rather limited. The recently published CLF ChIP‐chip data were based on a transgenic line in which CLF was driven by a constitutive promoter (Xiao et al., 2017). Therefore, there is a pressing need to profile their genome‐wide occupancy for the plant biology community to dissect their functions comprehensively. Here, we generated Arabidopsis transgenic lines expressing tagged CLF or SWN under their native promoters, respectively; and performed ChIP‐seq analyses. We also profiled the genome‐wide changes of H3K27me3 and transcriptomes in the clf, swn, and clf swn mutants. These genomics data would greatly facilitate the deci- phering of the detailed roles of plant PcG proteins in various devel- opmental processes. 2 | EXPERIMENTAL PROCEDURES 2.1 | Plant materials and growth conditions WT and the mutants, clf-29 (SALK_021003) and swn-4 (SALK_109121) of Arabidopsis used in this study were all in the Columbia (Col‐0) background, and have been described previously (Bouveret, Schönrock, Gruissem, & Hennig, 2006; Wang, Tyson, Jackson, & Yadegari, 2006). For all experiments, the seeds were stratified for 3 days at 4°C in darkness before sown on soil or on agar plates containing 4.3 g/L Murashige and Skoog (MS) nutrient mix (Sigma‐Aldrich), 1.5% sucrose (pH 5.75), and 0.8% agar. Plants were grown in the long‐day growth room (16‐hr light/8‐hr dark) at 22°C. Transfer DNA (T‐DNA) insertion mutants were obtained from the Arabidopsis Biological Resource Center (ABRC). The homozygous T‐DNA insertion mutants were identified by PCR‐based genotyping. The 2‐week‐old (counted when the seeds were transferred into the growth room after stratification) plants grown on synthetic media under 16‐hr light/8‐hr dark condition were used for all analyses unless otherwise specified. All primers are listed in Supporting Infor- mation Table S4. 2.2 | Plasmid constructs and transgenic plants Two bacterial artificial chromosome (BAC) clones named F26B6 and T10M13 that harbor the CLF and SWN locus, respectively, were obtained from ABRC. The BAC plasmids were isolated using QIA- GEN Large‐Construct Kit, and then used as the templates to amplify the CLF and SWN genomic sequence including the promoter/regula- tory region. The PCR products were purified using the GenepHlow Gel/PCR Kit, and were then subcloned into the binary vector pMDC107 between the PmeI and AscI sites to obtain pMDC107-gCLF and pMDC107-gSWN. The new constructs were sequenced to con- firm that the CLF (or SWN) and GFP coding sequences were in‐frame. The constructs were introduced into Agrobacterium tumefaciens strain GV3101, which were then used to transform into the clf-29 and swn-4 plants using the floral dip method (Clough & Bent, 1998), respectively. Homozygous transgenic lines with each homozygous genetic background were selected from T3 generation, in which the functional GFP‐tagged proteins were detected. 2 | SHU ET AL. 2.3 | Flowering time measurement WT, mutants, and transgenic plants were grown side by side in soil under the long‐day condition (16‐hr light/8‐hr dark) at 22°C. The num- ber of rosette leaves was counted when the inflorescence stem grew to 1 cm in length. At least 10 plants for each genotype were analyzed and the experiment was repeated three times independently. 2.4 | Western blot Two grams of 2‐week‐old seedlings were collected, and nuclei were isolated according to the ChIP protocol (Gendrel, Lippman, Mar- tienssen, & Colot, 2005) but without the tissue fixation step. The nuclear protein was released by dissolving the nuclei preparation in 300 μl of lysis buffer (50 mM Tris‐HCl, 10 mM EDTA, 1% SDS, and 1× protease inhibitors) and then sonicated. The protein solution was centrifuged at 16,000 g for 10 min at 4°C to remove debris. Proteins were resolved on a 4%–20% Mini‐PROTEAN TGX Precast Protein Gel (Bio‐Rad) by electrophoresis and detected by antibody to GFP (Abcam, ab290; 1:20,000 dilution) and histone H4 (Milli- pore, 07‐108; 1:20,000 dilution). Histone H4 was used as the load- ing control. 2.5 | Chromatin immunoprecipitation (ChIP) assay ChIP experiments were carried out as described (Chen et al., 2017; Gendrel et al., 2005; Li et al., 2016) with minor modifications. For the ChIP with the transgenic plants clf-29 pCLF::CLF-GFP and swn-4 pSWN::SWN-GFP, another transgenic line that carries GFP only in WT genetic background (p35S::GFP) was used as the negative control. Typically, 5 g of plant materials (2‐week‐old seedlings or 2‐week‐old clf swn double mutants) grown on agar plates were harvested, which were then cross‐linked with 1% formaldehyde for 20 min under vacuum and ground into fine powder in liquid nitro- gen. The chromatin was isolated and sheared into 200–800 bp fragments by sonication. The sonicated chromatin was incubated with anti‐GFP (ab290, Abcam) or anti‐H3K27me3 (07–449, Millipore) antibodies overnight at 4°C with gentle rotating. Then the MinElute PCR Purification Kit (Cat# 28004, Qiagen) was used to recover the precipitated DNA according to the manufacturer's instruction. The ChIP DNA was used for Illumina single‐end (1 × 50 bp) sequencing or qPCR. ChIP‐qPCR was performed with three biological replicates, and results were generated as percentage of input ‐DNA according to the Champion ChIP‐qPCR user manual (SABioscience). DNA quan- tity and quality were checked using a Qubit fluorometer (Thermo- Fisher Scientific). 2.6 | ChIP‐seq and data analyses At least 2 ng of each ChIP DNA was used to construct ChIP‐seq library, and two biological replicates for each sample. End repair, adapter ligation and amplification were performed using the Illumina Genomic DNA Sample Prep kit according to the manufacturer's protocol. The HiSeq 2500 platform was used for high‐throughput sequencing of the libraries. Sequence data were analyzed essentially as previously described (Chen et al., 2017; Li et al., 2015, 2016). Briefly, after sequencing, the raw sequence reads were cleaned by removing bases with low quality score and cutting sequencing adap- ter followed by filtering out short reads. Then the cleaned sequence reads were mapped to the Arabidopsis genome (TAIR10) (Lamesch et al., 2012) by Bowtie mapper (Langmead, Trapnell, Pop, & Salzberg, 2009) with default mismatch parameters and retaining only reads that can be mapped uniquely to the genome for further analyses. The number of ChIP‐seq reads for all experiments in this study is listed in Supporting Information Table S5. To identify reads enriched regions (peaks), MACS 1.4 (Zhang et al., 2008) was employed to per- form peak calling with default settings. High‐confidence target regions were defined as strict overlap of the MACS peaks from two biological replicates. Then the data were imported to Integrated Genome Browser (IGB) (Nicol, Helt, Blanchard, Raja, & Loraine, 2009) for visualization. Peak analyses were performed using PeakAn- alyzer (Salmon‐Divon, Dvinge, Tammoja, & Bertone, 2010). The PeakAnnotator was employed to identify functional elements proxi- mal to peak loci though the Nearest Downstream Gene (NDG) sub- routine. The target gene of each peak was defined as the genes closest to a given peak localized around the gene body (from 1 kb upstream of TSSs to TTSs). The gene annotation file was down- loaded from the EnsemblPlants homepage (https://plants.ensembl. org/index.html). The SICER program (window size = 600, gap size = 200, false discovery rate (FDR) ≤ 0.01) (Zang et al., 2009) was used to quantitatively compare the H3K27me3 levels of WT and mutants. Peaks with at least a two‐fold change were kept for further analyses. To identify DNA motifs enriched at CLF‐ and SWN‐bound sites, 300 bp sequence surrounding each peak summit (150 bp upstream and downstream, respectively) was extracted and searched for enriched DNA motifs using the DREME/MEME software suite (http://meme-suite.org/tools/meme-chip) (Machanick & Bailey, 2011) with default settings. The distribution of MACS peaks was per- formed by ChIPseek, a web‐based analysis tool (http://chipseek.c gu.edu.tw) (Chen et al., 2014), with default settings. Briefly, the information of chromosome, start site, and end site for each peak was fed to ChIPseek, based on the annotation file from UCSC tair10 assembly, the location of each peak was further grouped into these categories: promoter‐TSS (1 kb upstream of TSS), intergenic, exon, intron, 5′ UTR, 3′ UTR, and TTS. Heat maps of ChIP‐seq were gener- ated using the computeMatrix and plotHeatmap utilities in deepTools2 (Ramirez et al., 2016). The functional annotation was performed using the online “AgriGo” Gene Ontology (GO) analysis toolkit (http://bioinfo.cau.edu.cn/agriGO/) with default settings. 2.7 | RNA extraction Total RNA was isolated from ~50 mg of 2‐week‐old plants (WT and mutants) using the RNeasy Plant Mini kit (Qiagen) according to sup- plier's instruction. Total RNA was treated with RNase‐free DNaseI kit (Life Techologies). SHU ET AL. | 3 https://plants.ensembl.org/index.html).The https://plants.ensembl.org/index.html).The http://meme-suite.org/tools/meme-chip http://chipseek.cgu.edu.tw http://chipseek.cgu.edu.tw http://bioinfo.cau.edu.cn/agriGO/ 2.8 | RNA‐seq A total amount of 1.5 μg RNA per sample was used as input material for the RNA sample preparations. Sequencing libraries were gener- ated using NEBNext® UltraTM RNA Library Prep Kit for Illumina® (NEB, USA) following manufacturer's recommendations and index codes were added to attribute sequences to each sample. The libraries were sequenced on an Illumina Hiseq 4000 platform and 150 bp paired‐end reads were generated. Three independent biologi- cal replicates of each sample were sequenced separately. The raw sequence reads were cleaned by removing bases with low quality score and cutting sequencing adapter followed by filtering out short reads (Supporting Information Table S6). The cleaned reads were mapped to the TAIR10 Arabidopsis genome using TopHat v2.0.4 with default settings, except that a minimum intron length of 20 bp and a maximum intron length of 4,000 bp were used. Transcript assembly and calculations to identify differentially expressed genes were car- ried out by the Cufflinks package (Trapnell et al., 2012). Briefly, the alignment files after running TopHat were provided to Cufflinks to generate a transcriptome assembly for each sample. Then these assemblies were merged together using the Cuffmerge utility. The merged assembly provided a uniform basis for calculating transcript expression in each sample. The reads and the merged assembly were fed to Cuffdiff, where expression levels were calculated and the statistical significance of changes was tested. Genes with at least 1.5‐fold change in expression (FDR = 5%, p value < 0.05) were con- sidered to be expressed differentially among these samples in this study. 3 | RESULTS 3.1 | Genome‐wide occupancy of CLF and SWN To profile the genome‐wide targets of CLF, we generated a trans- genic line expressing GFP‐tagged CLF (CLF‐GFP) under the control of its own promoter in the clf-29 genetic background (clf-29 pCLF:: CLF-GFP). The transgene could fully rescue the phenotypes of the clf -29 mutant (Figure 1a,b). The GFP signal could be observed by con- focal microscopy (Figure 1c), and the CLF‐GFP protein could be detected by Western blot (Figure 1d). Then ChIP‐seq was performed with 2‐week‐old transgenic plants. A transgenic line expressing GFP only under the tobacco mosaic virus 35S promoter (p35S::GFP) was used as the negative control. Two independent biological replicates of ChIP DNAs were sequenced. We observed a good correlation between the two replicates (the Pearson coefficient is 0.89) (Sup- porting Information Figure S1). A total of 1,041 genomic regions (peaks), corresponding to 1,391 genes, were bound by CLF (Fig- ure 2a; Supporting Information Data S1 and S2). More than half of the CLF binding sites were located in promoter regions and exons, and about 16% of the peaks were in introns (Figure 2b). In general, WT clf-29 clf-29 pCLF::CLF-GFP (a) WT swn-4 swn-4 pSWN::SWN-GFP 0 2 4 6 8 10 12 14 16 18 Ro se tt e le af n um be r c a a a b (b) H4 GF P 126 KDa 121 KDa (d) (c) swn-4 pSWN::SWN-GFPclf-29 pCLF::CLF-GFP F IGURE 1 The transgenic lines expressing GFP‐tagged CLF or SWN. (a) Plant photos showing complementation of the clf-29 and swn-4 flowering phenotypes by the CLF‐ and SWN‐GFP fusion genes driven by their native promoters (pCLF:: CLF-GFP and pSWN::SWN-GFP), respectively. Scale bar: 1.0 cm. (b) Rosette leaf number at bolting of plants in different genetic backgrounds at 22°C under long‐day condition. Error bars indicate standard deviation from at least 30 plants. Lowercase letters indicate significant differences between genetic backgrounds, as determined by Post‐hoc Tukey's HSD test. (c) GFP signals detected by confocal microscopy in 4‐day‐old clf-29 pCLF::CLF-GFP and swn-4 pSWN::SWN-GFP roots. Scale bar: 50.0 μm. (d) Western blot analysis of nuclear extracts from 2‐week‐ old clf-29 pCLF::CLF-GFP and swn-4 pSWN:: SWN-GFP seedlings. Antibodies used: GFP (anti‐GFP; top) and H4 (anti‐histone H4; bottom). WT: wild type. 4 | SHU ET AL. the CLF binding signals were found to be the strongest at transcrip- tion start sites (TSSs), and gradually declined toward transcription termination sites (TTSs) (Figure 2c). Similarly, we also generated a SWN‐GFP transgenic line in the swn-4 background (swn-4 pSWN::SWN-GFP) and the transgene could fully rescue the late flowering phenotype of swn-4 (Figure 1a,b). We could observe the GFP signal by confocal microscopy (Figure 1c). The SWN‐GFP protein could be detected by Western blot (Figure 1d). Two independent biological ChIP‐seq experiments were carried out using 2‐week‐old swn-4 pSWN::SWN-GFP seedlings. The two replicates were well correlated with a high Pearson coefficient (0.92) (Supporting Information Figure S1). We found that SWN binds to 1,298 genomic sites, corresponding to 1,877 genes (Figure 2a; Supporting Informa- tion Data S1 and S2). About 43% of the SWN binding sites were found in exons, followed by 24% in promoter regions and 16% in introns (Figure 2b). The SWN ChIP‐seq signals were generally enriched at TSSs, and gradually decreased toward TTSs (Figure 2c). Then we compared the binding profiles of CLF and SWN. The two H3K27 methyltransferases exhibited highly similar chromatin‐ associated profiles (Figure 2c; Supporting Information Table S1), which are consistent with previously observed genetic redundancy between the two genes (Chanvivattana et al., 2004; Farrona et al., 2011). Furthermore, we analyzed the distribution of CLF and SWN binding along each of the chromosomes. As shown in Supporting Information Figure S2, CLF and SWN predominantly bind to euchro- matic regions. Almost all of the CLF‐bound genes were also bound by SWN (1,385 out of the 1,391 CLF targets); only 6 genes were bound by CLF uniquely, while nearly 500 more genes bound by SWN solely (Figure 2d; Supporting Information Data S2). ChIP sig- nals at some selected loci representing the CLF and SWN co‐targets Samples Binding Sites Total Target Genes CLF 1,041 1,391 SWN 1,298 1,877 (a) (b) (d) 1,385492 61,385492 6 SWN-bound genes (1,877) CLF-bound genes (1,391) Telo-box-like mo�f E-value = 4.3e-128 256 peaks/24.6% E-value = 5.9e-089 96 peaks/9.2% E-value = 8.5e-113 332 peaks/25.6% E-value = 7.1e-067 84 peaks/6.5% GAGA-like mo�f CLF SWN (g) EMF1 LFYWOX1 1 200 1 200 1 200 CLF SWN p35S::GFP (e) 1 200 1 200 1 200 CLF SWN p35S::GFP CLE9 NSP2 PDLP6 (f) 0 10 20 30 35S:GFP SWN-GFP TSS TTS–2 Kb 2 Kb SWN-bound Genes M ea n of n or m al ize d re ad s p er 5 0 bp 0 10 20 30 35S:GFP CLF-GFP CLF-bound Genes TSS TTS–2 Kb 2 Kb M ea n of n or m al ize d re ad s p er 5 0 bp (c) Exon Promoter-TSS Intron TTS Intergenic 5' UTR 3' UTR 43% 24% 16% 10% 5% 1% 1% SWN sites (1,298) 39% 24% 16% 12% 7% 1% 1% CLF sites (1,041) F IGURE 2 Genome‐wide occupancy of CLF and SWN. (a) Table showing the numbers of CLF and SWN binding sites and target genes. (b) Pie charts showing the distribution of CLF and SWN at annotated genic and intergenic regions in the genome. (c) Mean density of CLF/SWN occupancy at all target genes. Plotting regions were scaled to the same length as follows: 5′ ends (−2 kb to transcription starting site (TSS)) and 3′ ends (transcription stop site (TTS) to downstream 2 kb) were not scaled, and the gene body was scaled to 3 kb. (d) Venn diagram showing the overlap between the genes occupied by CLF and those by SWN. (e, f) ChIP‐seq genome browser views of CLF and SWN co‐ occupancy (e) and SWN unique occupancy at selected genes (f). Gene structures are shown underneath each panel. (g) Two motifs enriched in CLF and SWN peaks. The number and percent of peaks containing the motifs are shown. SHU ET AL. | 5 and SWN unique targets were shown in Figure 2e,f, respectively. The ChIP‐seq results were validated by ChIP‐quantitative PCR (ChIP‐ qPCR) analyses on several randomly selected target genes (Support- ing Information Figure S3). Furthermore, we also compared our CLF/ SWN target genes with FIE target genes by taking advantage of the published datasets (Deng et al., 2013; Xiao et al., 2017). As shown in Supporting Information Figure S4, CLF/SWN and FIE share a large fraction of targets (Supporting Information Data S3).This result is understandable, given the fact that these experiments were carried out with plants of different ages. It might also imply that CLF/SWN and FIE probably have distinct functions, although the clf swn double mutants share similar phenotypes with the fie mutants (Bouyer et al., 2011; Chanvivattana et al., 2004). In mammals, similarly, a partial overlap between the occupancy of H3K27 methyltransferases EZH1 and EED has been reported recently (Bodega et al., 2017). 3.2 | Identification of two DNA motifs from CLF and SWN binding sites To examine whether any particular DNA motifs are enriched in the identified CLF and SWN binding regions, stringent motif searches were performed and two DNA motifs were identified for both CLF and SWN, a GAGA‐like motif and a Telo‐box‐like motif (Figure 2g; Supporting Information Table S2). The GAGA‐like is the most enriched motif (24.6% of CLF peaks; 25.6% of SWN peaks) with a high significance (E‐value ≤8.5e−113). In Drosophila, there are many PcG response elements (PREs) including GA‐repeats that are recog- nized by GAGA factors (GAFs) (Farkas et al., 1994; Hodgson, Argiro- poulos, & Brock, 2001; Katsani, Hajibagheri, & Verrijzer, 1999). The Telo‐box motif is a short sequence identical to plant telomere‐repeat units (AAACCCTA)n, which was originally observed in the 5′ flanking regions of genes encoding ribosomal proteins and the translation elongation factor EF1α (Gaspin, Rami, & Lescure, 2010). Although not highly enriched in the CLF and SWN peaks (9.2% of CLF peaks; 6.5% of SWN peaks), this motif was discovered with a high signifi- cance (E‐value ≤7.1e−067). The Telo‐box motif has been shown very recently to be recognized by the telomere‐repeat factors (TRBs) which then recruit CLF/SWN through direct physical interaction with CLF/SWN (Zhou et al., 2018). Of note, the two motifs were also pre- viously identified in the genomic binding regions of other PcG pro- teins, FIE and LHP1 (Deng et al., 2013; Molitor et al., 2016; Xiao et al., 2017; Zhou, Hartwig, James, Schneeberger, & Turck, 2016). 3.3 | Functions of CLF and SWN target genes in plant growth and development To reveal the potential functions of CLF and SWN target genes, we performed a Gene Ontology (GO) analysis and found that CLF and SWN preferentially bind to genes involved in developmental path- ways and abiotic and biotic stress responses (Figure 3a; Supporting Information Data S4). Particularly, genes encoding members of sev- eral transcription factor (TF) families that are involved in develop- mental processes and stress responses are markedly overrepresented in the CLF and SWN co‐targets, such as the Homeobox, WRKY, and MADS families (Figure 3b; Supporting Information Table S3). In con- trast, the SWN unique targets are enriched in those involved in lipid localization/storage, cell wall modification, and post‐embryonic devel- opment (Figure 3c; Supporting Information Data S4). Shown in Figure 3d‐g are ChIP‐seq signals at selected gene loci that play important roles in several growth and developmental pro- cesses. The major seed regulatory genes including LEAFY COTYLE- DON2 (LEC2), FUSCA3 (FUS3), and ABSCISIC ACID INSENSITIVE3 (ABI3) (Tang et al., 2008; Tang, Bian, et al., 2012; Tang, Lim, et al., 2012; To et al., 2006) and seed maturation genes, such as AT3G17520 which belongs to the late embryogenesis abundant pro- tein (LEA) family (Lin, Pajak, Marsolais, McCourt, & Riggs, 2013) were co‐bound by CLF and SWN (Figure 3d). These findings suggest that CLF and SWN may be involved in the repression of the seed maturation program at seedling stage by directly targeting to the key maturation genes. CLF/SWN binds to several major floral transition genes including SHORT VEGETATIVE PHASE (SVP) and AGAMOUS- LIKE 15 (AGL15), which belong to type II MADS box gene family (Chiang, Barua, Kramer, Amasino, & Donohue, 2009; Harding, Tang, Nichols, Fernandez, & Perry, 2003; Hartmann et al., 2000; Li et al., 2015; Michaels et al., 2003; Smaczniak, Immink, Angenent, & Kauf- mann, 2012), and flower organ identity genes like AGAMOUS (AG) and APETALA1 (AP1) (Figure 3e), indicating that CLF and SWN play an important role in silencing the flower development program dur- ing vegetative growth. It is well documented that the antagonistic activities of microRNAs miR156 and miR172 coordinate the juvenile‐ to‐adult leaf transition; miR319 represses the onset of senescence by targeting the TEOSINTE BRANCHED1, CYCLOIDEA, and PCF (TCP) transcription factors; and miR167 is involved in the regulation of auxin homeostasis and adventitious rooting (Chen, 2004; Rubio‐ Somoza & Weigel, 2011; Schommer et al., 2008; Wu et al., 2009). The gene loci encoding these microRNAs are all co‐targets of CLF and SWN (Figure 3f). Plant hormones such as auxin mediate many growth and developmental processes (Bari & Jones, 2009). We noticed that CLF and SWN bind to genes that are involved in hor- monal signaling pathways, for example, genes that encode key play- ers in auxin signaling pathways, the Aux/IAA transcriptional repressors, GH3 family, and SMALL AUXIN UPREGULATED RNA (SAUR) family (Woodward & Bartel, 2005) (Figure 3g). 3.4 | Genome‐wide profiling of H3K27me3 in clf-29, swn-4, and clf-29 swn-4 To further examine the functional redundancy between CLF and SWN in Arabidopsis seedlings, the genome‐wide distributions of H3K27me3 in WT, clf-29, swn-4, and clf-29 swn-4 were examined. In WT, we were able to identify 6,854 genes marked by H3K27me3 (Supporting Information Data S5 and S6), which largely overlap with previously published genome‐wide profiles (Supporting Information Figure S5a) (Li et al., 2015; Wang et al., 2016). By analysing the dis- tribution pattern of H3K27me3 peaks over annotated genomic fea- tures, we found that the marker has an even distribution from TSSs 6 | SHU ET AL. to TTSs (Figure 4a). Nearly 63% of the peaks were localized in genic regions (Supporting Information Figure S5b). The peaks of H3K27me3 and CLF/SWN are very similar in terms of their width range and are generally much broader than those of REF6, a DNA binding factor that shows typical narrow binding peaks (Figure 4b) (Li et al., 2016). When comparing occupancy profile of H3K27me3 with that of CLF/SWN, it is obvious that the vast majority of the CLF and SWN binding genes were marked by H3K27me3 (1,253 out of 1,391 for CLF; 1,643 out of 1,877 for SWN) (Figure 4c; Supporting Information Data S6), consistent with the notion that both enzymes directly bind to their target loci to deposit H3K27me3. To our surprise, three quar- ters of the H3K27me3‐marked genes were not occupied by either CLF or SWN in WT seedlings (Figure 4c). The global level of H3K27me3 in clf-29 was significantly lower than that in WT, while in swn-4, it was almost identical to that in WT (Figure 4d). In the clf-29 swn-4 double mutant, nearly no H3K27me3 could be detected (Fig- ure 4d,e), which is consistent with previous reports (Lafos et al., 2011; Zhou, Romero‐Campero, Gómez‐Zambrano, Turck, & Calonje, 2017). To eliminate the possibility that the loss of H3K27me3 was due to the change of nucleosome distribution, we checked the H3 dis- tribution in the clf-29 swn-4 double mutant. The H3 distribution levels in clf-29 swn-4 were very similar to those in WT, with slight reduction at some loci examined (Supporting Information Figure S6). Taken together, these results suggest that CLF and SWN are probably the only enzymes that catalyze H3K27me3 at this developmental stage. Comparing the H3K27me3 levels in clf-29 and WT, we identified 552 genes at which the levels of H3K27me3 were reduced at least (a) (c) Biological process(b) (d) LEC2 1 200 1 200 1 200 FUS3 ABI3 AT3G17250 CLF SWN p35S::GFP (f) MIR156C MIR167A MIR172D MIR319C 1 200 1 200 1 200 CLF SWN p35S::GFP (e) AG 1 200 1 200 1 200 SVP AGL15 AP1 CLF SWN p35S::GFP (g) IAA1 IAA2 GH3.2 SAUR18 200 200 1 1 1 200 CLF SWN p35S::GFP TF families Number of genes in the genome Number/percent of genes co-targeted by CLF/SWN Total TFs 1,700 307/18.1% Homeobox 91 31/34.1% WRKY 72 20/27.8% TCP 26 7/26.9% MADS 109 28/25.7% bHLH 161 36/22.4% AP2-EREBP 138 28/20.3% NAC 96 19/19.8% – Lo g 1 0 (p v al ue ) F IGURE 3 Functional categorization of CLF and SWN target genes. (a) Gene Ontology (GO) analysis of CLF and SWN co‐target genes. (b) Table showing the transcription factor gene families that are co‐targeted by CLF and SWN. The transcription factor gene families can be found at the Arabidopsis genome resources website (http://arabidopsis.med.ohio-sta te.edu/AtTFDB/). (c) GO analysis of the SWN unique target genes. (d‐g) ChIP‐seq genome browser views of CLF and SWN occupancy at selected loci; p35S::GFP transgenic plants as the negative control. Gene structures are shown underneath each panel. SHU ET AL. | 7 http://arabidopsis.med.ohio-state.edu/AtTFDB/ http://arabidopsis.med.ohio-state.edu/AtTFDB/ 0 10 20 30 40 H3K27me3 Input –2 Kb TSS TTS 2 Kb H3K27me3-marked Genes sdaer dezila mr onfo nae M pe r 5 0 bp (d)(c) WT_K27 CLF SWN 39398 5,2081,250136 3 3 (e) CLF SWN H3K27me3 htgnerts gnidnib FLC –2 Kb +2 Kb0 Low High (a) (f) (b) 200 200 200 GIK AT2G18150 AT2G18140 WT_K27 clf-29 swn-4_K27 swn-4_K27 clf-29_K27 Type I (13 peaks/2.3%) Type II (326 peaks/59.1%) Type III (213 peaks/38.6%) 1 1 1 1 200 (g) 1 150 1 150 1 150 1 150 1 150 WT_K27 CLF SWN BRM BRM input SVP WRKY40 BEL1 (h) Pe rc en t o f p ea ks 0 10 20 30 40 50 60 1 Kb 1. 5 Kb 2 Kb 2. 5 Kb 3 Kb 3. 5 Kb 4 Kb 4. 5 Kb 5 Kb 5. 5 Kb 6 Kb 7 Kb 8 Kb 9 Kb 10 K b CLF SWN WT_K27 REF6 1,336 37 515 18 73 CLF clf_k27 down clf_k27 up 1,868 3 29 6 30 swn_k27 up swn_k27 down SWN 0 20 40 M ea n de ns ity (t ag s/ 50 b p) -2 kb 0 +2 kb WT clf-29 swn-4 clf-29 swn-4 swn-4 input WT input clf-29 input clf-29 swn-4 input H3K27me3 F IGURE 4 Genome‐wide profiling of H3K27me3 in clf-29, swn-4, and clf-29 swn-4. (a) Mean density of H3K27me3 occupancy at all target genes in WT. Plotting regions were scaled to the same length as follows: 5′ ends (−2 kb to transcription starting site (TSS)) and 3′ ends (transcription stop site (TTS) to downstream 2 kb) were not scaled, and the gene body was scaled to 3 kb. (b) The width ranges for CLF, SWN, and WT_K27 (WT_H3K27me3) peaks. The x axis shows the width of peaks within the ranges (e.g. 1 kb: width <1 kb; 1.5 kb: 1 kb ≤ width ≤ 1.5 kb). The y axis represents the percent of peaks in each range. The REF6 peaks were used as representative “narrow peaks” (Li et al., 2016). (c) Venn diagram showing the overlap among the genes marked with H3K27me3 in WT (WT_K27) and those occupied by CLF or SWN. (d) Mean density of H3K27me3 levels in WT, clf-29, swn-4, and clf-29 swn-4. Inputs from all backgrounds are shown in gray. The average signal within 2 kb genomic regions flanking the center of the H3K27me3 peaks in WT is shown. (e) Heat maps representing the co‐ occupancy of CLF and SWN in the genome (blue, left), and the H3K27me3 levels in WT, clf-29, swn-4, and clf-29 swn-4 (green, right). Each horizontal line represents a CLF/SWN binding peak or H3K27me3 peak. Columns show the genomic region surrounding each peak summit. Signal intensities are indicated by the shade of blue or green. (f) Venn diagrams showing the overlaps between the genes occupied by CLF and the genes with decreased (clf_K27 down) and increased H3K27me3 levels (clf_K27 up) in clf-29 (left); and between the genes occupied by SWN and the genes with decreased (swn_K27 down) and increased H3K27me3 levels (swn_K27 up) in swn-4 (right). (g) ChIP‐seq signals at representative genomic loci showing three distinct types of H3K27me3 reduction pattern in clf-29 compared to WT. Gene structures are shown underneath the panel. (h) ChIP‐seq genome browser views of selected genes showing the distribution of H3K27me3 in WT (WT_K27), and the occupancy of CLF, SWN, and BRM (Li et al., 2016). BRM ChIP‐seq input signals at these genes are also shown as the negative control. Gene structures are shown underneath the panel. WT: wild type. 8 | SHU ET AL. two‐fold in clf-29, and 91 genes that showed at least two‐fold increase (Supporting Information Data S7). However, only 37 out of the 552 genes were CLF direct targets (Figure 4f). In swn-4, the changes of H3K27me3 level were only detected at 68 genes (32 increase and 36 decrease; two‐fold cut off) (Figure 4f; Supporting Information Data S7), which were consistent with the findings shown in Figure 4d that the global level of H3K27me3 in swn-4 seemed to be unchanged relative to that of WT. After close exami- nation of the regions with reduced H3K27me3 in clf-29, we identi- fied three types of reduction pattern (Figure 4g). In type I, loss of CLF had only weak or no effect on the H3K27me3 peak summit, but caused drastic reduction of H3K27me3 in flanking regions (type I, Figure 4g). The remaining H3K27me3 signals at the summit com- pletely disappeared in the clf-29 swn-4 double mutant, as shown in the ChIP‐seq described as above and confirmed by ChIP‐qPCR (Sup- porting Information Figure S7). In type II, a partial reduction of H3K27me3 across the entire H3K27me3‐marked region was observed in clf-29, and the remaining signal was completely abol- ished in the clf-29 swn-4 double mutant (Supporting Information Fig- ure S7). In type III, we saw almost complete loss of H3K27me3 in clf-29 (Supporting Information Figure S7), suggesting a non‐redun- dant role of CLF in catalyzing H3K27me3 at these genes. Noticeably, there is a small group of genes that are CLF/SWN targets, but showed no or very low levels of H3K27me3 (Figure 4h). This is reminiscent of our previous findings at some genomic loci such as the floral repressor locus SVP (Li et al., 2015). In that work we demonstrated that the chromatin remodeler BRM could promote gene expression by limiting the deposition of H3K27me3 through suppressing the recruitment and/or the activity of CLF/SWN. There- fore, to find out whether the lack of H3K27me3 deposition is due to the presence of BRM, we checked the BRM occupancy at those CLF/SWN targets (Archacki et al., 2017; Li et al., 2016). Indeed, we saw nearly half of these genes (63 out of 136) were bound by BRM as well (Supporting Information Data S8). ChIP‐seq signals showing the H3K27me3 levels and the occupancy of CLF, SWN, and BRM at several representative loci are presented in Figure 4h. 3.5 | Transcriptome profiling in clf-29, swn-4, and clf-29 swn-4 To further understand the action of CLF and SWN, we performed RNA‐seq experiments to profile transcriptome changes in the clf-29, swn-4, and clf-29 swn-4 mutants and compared to those in WT. In clf-29, we found that 591 genes were transcriptionally up‐regulated compared to WT, whereas the expression of 426 genes was down‐ regulated (Figure 5a; Supporting Information Data S9), by at least 1.5‐fold (absolute fold change [FC] ≥ 1.5) at a false discovery rate (FDR) <0.05. In swn-4, a total of 374 genes showed increased expression and 266 genes showed decreased expression compared to WT (Figure 5a; Supporting Information Data S9). Overall, fewer genes showing altered expression were detected in swn-4 compared to those in clf-29, which may explain why the swn-4 mutants display less severe morphological defects compared to the clf-29 mutants. In clf-29 swn-4, more than 10,000 genes (6,058 up‐regulated; 4,601 down‐regulated) showed differential expression compared to WT (Figure 5a; Supporting Information Data S9). In the WT plants, it is clear that the global expression levels of all the H3K27me3‐marked genes and CLF‐/SWN‐bound genes were lower than the rest of genes in the genome (Supporting Information Figure S8). This obser- vation is expected given the recognized role of PRC2 in repressing gene expression (Wang et al., 2016). Next, we compared the binding profile of CLF/SWN with the transcriptome data. Generally, the expression of those genes bound by CLF/SWN is similar between the clf-29 and swn-4 single mutants and WT, but showing significant up‐regulation in the clf-29 swn-4 double mutant (Figure 5b), indicating that CLF and SWN play partially redundant roles at these loci. Furthermore, nearly one‐third of the CLF/SWN target genes showed up‐regulation in clf-29 swn-4 (Fig- ure 5c). To examine the biological functions of CLF/SWN targets that were up‐regulated in the double mutant, GO analysis was per- formed. The up‐regulated genes were functionally enriched in biosynthetic process, developmental process, and response to stimu- lus (Supporting Information Figure S9; Data S10). We also examined whether there is a correlation between the loss of H3K27me3 and the expression levels of transcripts in the clf swn double mutants. As shown in Figure 5d, we observed up‐regulation of many genes in clf- 29 swn-4, particularly those that were heavily marked by H3K27me3 in WT plants, but overall there was no clear correlation between the loss of H3K27me3 and the expression levels in clf-29 swn-4. 4 | DISCUSSION Epigenetic regulation of gene expression through histone modifica- tions is fundamental for plant growth and development. In particular, the repressive histone modification H3K27me3 marks a large frac- tion of the Arabidopsis genome dynamically and plays critical roles in plant development (Derkacheva & Hennig, 2014; Gan, Xu, & Ito, 2015; Li et al., 2015; Wiles & Selker, 2017; Zhang et al., 2007). However, the global binding profiles of the two major H3K27me3 methyltransferases, CLF and SWN, have not been determined and compared. Thus, our CLF and SWN ChIP‐seq data, together with the H3K27me3 ChIP‐seq and the transcriptome data will be valuable resources for the plant epigenetics research community. Further, the CLF- and SWN-GFP fusion transgenic lines (driven by their corre- sponding native promoters) could be used to study the function of PcG proteins in many other developmental processes and stress response pathways in Arabidopsis. In the following, we would like to highlight and discuss several interesting findings from this study. These findings provide new bio- logical insights into the mechanisms underlying PRC2 activities, par- ticularly the functional interplays between CLF and SWN. Our results also shed fresh lights on the roles of PRC2 in two develop- mental transitions, i.e., flowering and the repression of seed genes during vegetative growth. Further, future investigations are war- ranted to address many of the new questions arising from this study, SHU ET AL. | 9 which would undoubtedly enhance our understanding of PcG activi- ties and their regulation during plant growth and development. Our data indicate that CLF and SWN are highly redundant in depositing H3K27me3, and probably they are the only two pro- teins that are able to catalyze H3K27me3 at the seedling stage. Strikingly, the majority of H3K27me3‐marked regions were not bound by either CLF or SWN. Since CLF and SWN are the enzymes required to catalyze H3K27me3 during seedling stage, therefore, H3K27me3‐marked regions are expected to be occupied by CLF and/or SWN. This apparent discrepancy/mystery could pos- sibly be explained by technical and/or biological reasons. Techni- cally, it would be very hard to detect binding signals by ChIP‐seq if CLF and SWN bind to some regions very weakly. Similar findings were reported in Drosophila: genome‐wide comparison of PRC2 and H3K27me3 profiles identified very “weak” PcG binding sites (as the authors called), where H3K27me3 but not PRC2 was detected (Schwartz et al., 2006). Biologically, explanations for the smaller number of CLF and SWN peaks compared to the H3K27me3 peaks could be that CLF and SWN bind to chromatins transiently, they might leave target sites as long as the H3K27me3 is deposited. Supporting this possibility, a fluorescence recovery after photobleaching study (FRAP) in Drosophila found that PcG members are exchangeable at specific loci in the genome (Ficz, Heintzmann, & Arndt‐Jovin, 2005). We also need to bear in mind that ChIP‐seq profiling is just a snapshot of the occupancy, but the H3K27me3 marks observed are the culmination of PRC2 Log2 fold change –L og 10 )eulav p( clf-29/WT Down 426 Up 591 swn-4/WT –L og 10 (p v al ue ) Down 266 Up 374 clf-29 swn-4/WT –L og 10 (p v al ue ) Down 4,601 Up 6,058 Log2 fold change Log2 fold change (a) level noisserpxE (F PK M ) a a a b (b) CLF-/SWN-bound genes (c) Gene�c background Expression change Overlap with CLF/SWN Targets CLF SWN clf-29 Up (591) 82/5.9% 97/5.2% Down (426) 42/3.0% 58/3.1% swn-4 Up (374) 28/2.0% 38/2.0% Down (266) 9/0.6% 16/0.9% clf-29 swn-4 Up (6,058) 476/34.2% 651/34.7% Down (4,601) 180/12.9% 247/13.2% (d) WT clf-29 swn-4 clf-29 swn-4 –2 Kb +2 Kb0 Low High clf-29 swn-4 clf-29 swn-4 H3K27me3 Gene expression (mutant/WT) Log10(FC) F IGURE 5 Transcriptome profiling of clf-29, swn-4, and clf-29 swn-4. (a) Volcano plots displaying significantly up‐regulated and down‐ regulated genes in clf-29, swn-4, and clf-29 swn-4 compared to WT (red dots, p < 0.05, fold change > 1.5), respectively. The x axis represents the Log2 value of fragments per kilobase per million (FPKM) mapped reads in each mutant/WT, and the y axis is the −Log10 of the p value for the significance of differential expression. (b) Box plots representing the average expression level (FPKM) of CLF and SWN co‐targets in clf-29, swn-4, and clf-29 swn-4. Lowercase letters indicate significant differences between genetic backgrounds, one‐way ANOVA. (c) Table showing the percent of overlaps for genes bound by CLF/SWN with differential expression in clf-29, swn-4, and clf-29 swn-4 compared to these in WT. (d) Heat maps illustrating the ChIP‐seq density in WT and mutants (clf-29, swn-4, and clf-29 swn-4), ranked by H3K27me3 read intensity within ±2 kb of peak summits in WT (green, left), and the RNA‐seq intensity for the H3K27me3‐marked loci in WT with the same order. Each horizontal line represents an H3K27me3 peak. Columns show the genomic region surrounding each peak summit. Signal intensities are indicated by the shade of green. The expression intensity is measured by Log10 (FC), FC = fold change (mutant vs. WT FPKM). WT: wild type. 10 | SHU ET AL. activity at all earlier stages, as events that cannot be fully recorded at just one time point. Another potentially significant finding is about the differential roles of CLF and SWN in depositing the H3K27me3 mark at the same loci. As shown in Figure 4g, we saw genomic sites where CLF and SWN play partially redundant or non‐redundant roles in deposit- ing H3K27me3. Perhaps more interestingly, we also found genomic sites where loss of CLF led to a marked reduction of H3K27me3 in flanking regions, but not the peak summit, suggesting that CLF might help the spreading of H3K27me3 to flanking regions from a “core” area. Similar H3K27me3 pattern changes were also reported in pre- vious studies (Li & Cui, 2016; Wang et al., 2016; Yang et al., 2017; Yuan et al., 2016). The remaining H3K27me3 signal at the summit in the clf-29 mutants might be catalyzed by SWN. Indeed, the H3K27me3 signal at the summit completely disappeared in the clf- 29 swn-4 double mutant (Figure 4g). Thus, SWN might play redun- dant roles in catalyzing H3K27me3 only at the peak summit, but not the flanking regions. These observations present us questions that warrant future investigation, such as the locus‐specific recruitment of CLF/SWN and their functional interplay in depositing H3K27me3. In terms of locus‐specific recruitment of PcG proteins, this and other recent studies have identified DNA motifs that are enriched in PcG binding sites (Figure 2g) (Deng et al., 2013; Molitor et al., 2016; Xiao et al., 2017; Zhou et al., 2018). Future studies are needed to deci- pher their roles in mediating the genomic targeting of PcG proteins. One possible way to examine the role of these DNA motifs in recruiting PcG proteins is by using the CRISPR/Cas9 system to dis- rupt the motif in vivo and then check the enrichment levels of PcG proteins (Li et al., 2018). It is also crucial to identify the DNA sequence‐specific players that mediate the locus‐specific recruitment of PcG proteins (Zhou et al., 2018). Our transcriptome analyses revealed that the expression of more genes was altered in the clf swn double mutant than either the clf or swn single mutants (Figure 5). This observation was consistent with the genetic redundancy of CLF and SWN and the recognized func- tions of PcG proteins in repressing gene expression. Although thou- sands of genes were clearly up‐regulated in the double mutants, we were surprised to find that the expression of many other genes remained unchanged and thus there was no overall correlation between the loss of H3K27me3 and up‐regulation of gene expres- sion. Similar observation was made in our previous transcriptome analysis in brm and ref6 mutants—many BRM/REF6 targets did not show alteration in gene expression (Li et al., 2015). This may be due to the possibility that many other factors also function at these sites, and the gene expression outcome may not be significantly affected by only one of them. In addition, there was almost equal number of genes that showed down‐regulation in the double mutants, which were not likely caused directly by the loss of H3K27me3, but rather indirectly via some intermediate effectors. It is well known that PcG proteins play important roles in devel- opmental transitions, but the underlying mechanisms are incom- pletely defined. Here, by integrating CLF/SWN and H3K27me3 ChIP‐seq and RNA‐seq data (Supporting Information Figure S10), we attempted to show how CLF/SWN may act to regulate two devel- opmental transitions, that is, repression of seed genes during vege- tative growth and control of flowering time. The seed development regulatory genes LEC1, LEC2, FUS3, ABI3, and the four 2S and the three 12S genes (Lin et al., 2013; Tang et al., 2008; Tang, Bian, et al., 2012; Tang, Lim, et al., 2012; To et al., 2006) were all marked by H3K27me3. The levels of H3K27me3 showed no significant changes in either the clf-29 or swn-4 single mutants, but all com- pletely abolished in the clf-29 swn-4 double mutant, consistent with the depression of these seed genes in the double mutants (Chanvi- vattana et al., 2004; Farrona et al., 2011; Lu et al., 2011). Although these seed genes were all clearly marked by H3K27me3, the occu- pancy of CLF/SWN varied considerably. The 2S and 12S genes appeared to be mainly bound by SWN, not by CLF. The LEC2, FUS3, and ABI3 genes were bound by both CLF and SWN, but the binding intensities varied drastically, with the strongest binding being found at FUS3, greatly reduced at LEC2 and ABI3. The LEC1 gene was bound by SWN only (Supporting Information Figure S10a). These findings suggest that CLF and SWN play redundant, yet dif- ferential, roles in repressing the seed maturation program during the seedling stage. Our data also provide some clues for the early flowering pheno- type of the clf mutants. As shown in Supporting Information Fig- ure S10b, several genes that encode positive regulators of floral transition such as AGLs 17, 19, 24, and 71, SOC1, and FT, showed up‐regulation in clf-29, but not in swn-4. The two flowering repres- sor‐encoding genes FLC and SVP showed differential expression (up‐ and down‐regulated, respectively). The increase of FLC expression is consistent with the reduction of H3K27me3; and the decrease of SVP could be (at least partially) due to the slight increase of H3K27me3 level at this locus in clf-29. As FLC and SVP may actually work together by forming a repressor complex (Li et al., 2008; Mateos et al., 2015), it is reasonable to speculate that the reduction of SVP in clf-29 would likely result in less amount of the complex and consequently more FT. This is likely the case, as we and others have shown that SVP represses flowering in a dosage‐sensitive man- ner (Hartmann et al., 2000; Li et al., 2015). Of note, we and others have consistently found the up‐regulation of FT in the clf-29 mutants (Farrona et al., 2011; Jiang et al., 2008; Liu et al., 2018). But, unlike the previous studies, we did not find significant CLF/SWN enrich- ment at this locus (Supporting Information Figure S10b). This appar- ent inconsistency could be due to the different plant materials used in these independent studies, that is, the CLF-GFP fusion gene used in our work was driven by the native CLF promoter, while it was dri- ven by constitutive promoters in other studies. ACKNOWLEDGMENTS We thank the Arabidopsis Biological Resource Centre for providing the mutant seeds used in this study; and Alex Molnar for help with preparing the figures. This work was supported by grants from the Natural Science and Engineering Research Council of Canada (RGPIN/04625‐2017, to Y.C.); Agriculture and Agri‐Food Canada (to SHU ET AL. | 11 Y.C.); and National Natural Science Foundation of China (31371715, to J.L.). J.S. was supported by a graduate fellowship from the China Scholarship Council. CONFLICT OF INTEREST The authors declare no conflict of interest. AUTHOR 'S CONTRIBUTION J.S., C.C., C.L., S.E.K., and Y.C. conceived and designed the experi- ments; V.N. and Z.C.Y. performed all the Sanger DNA sequencing; J.S. performed all the rest of the experiments; J.S, C.C., R.K.T., S.B., K.Y., and J.L. analyzed ChIP‐seq and RNA‐seq data; J.S., C.C., C.L., and Y.C. wrote the paper. ACCESSION NUMBERS The gene accession numbers that were used in this study are as fol- lows: CLF (AT2G23380) and SWN (AT4G02020). ChIP‐seq and RNA‐seq data have been deposited in the National Center for Biotechnology Information GEO database under accession number GSE108960. ORCID Jie Shu https://orcid.org/0000-0003-0359-6420 Yuhai Cui https://orcid.org/0000-0002-3675-3288 REFERENCES Aichinger, E., Villar, C. B. R., Di Mambro, R., Sabatini, S., & Köhler, C. (2011). The CHD3 chromatin remodeler PICKLE and Polycomb group proteins antagonistically regulate meristem activity in the Arabidopsis Root. The Plant Cell, 23, 1047–1060. https://doi.org/10.1105/tpc. 111.083352 Archacki, R., Yatusevich, R., Buszewicz, D., Krzyczmonik, K., Patryn, J., Iwanicka-Nowicka, R., … Swiezewski, S. (2017). Arabidopsis SWI/ SNF chromatin remodeling complex binds both promoters and ter- minators to regulate gene expression. Nucleic Acids Research, 45, 3116–3129. https://doi.org/10.1093/nar/gkw1273 Bari, R., & Jones, J. D. G. (2009). Role of plant hormones in plant defence responses. Plant Molecular Biology, 69, 473–488. https://doi.org/10. 1007/s11103-008-9435-0 Bodega, B., Marasca, F., Ranzani, V., Cherubini, A., Della Valle, F., Neguembor, M. V., … Orlando, V. (2017). A cytosolic Ezh1 isoform modulates a PRC2–Ezh1 epigenetic adaptive response in postmitotic cells. Nature Structural & Molecular Biology, 24, 444. https://doi.org/ 10.1038/nsmb.3392 Bouveret, R., Schönrock, N., Gruissem, W., & Hennig, L. (2006). Regula- tion of flowering time by Arabidopsis MSI1. Development, 133, 1693– 1702. https://doi.org/10.1242/dev.02340 Bouyer, D., Roudier, F., Heese, M., Andersen, E. D., Gey, D., Nowack, M. K., … Schnittger, A. (2011). Polycomb repressive complex 2 controls the embryo‐to‐seedling phase transition. PLoS Genetics, 7, e1002014. https://doi.org/10.1371/journal.pgen.1002014 Cao, R., Wang, L., Wang, H., Xia, L., Erdjument-Bromage, H., Tempst, P., … Zhang, Y. (2002). Role of histone H3 lysine 27 methylation in Polycomb‐group silencing. Science, 298, 1039–1043. https://doi.org/ 10.1126/science.1076997 Cao, R., & Zhang, Y. (2004). The functions of E(Z)/EZH2‐mediated methy- lation of lysine 27 in histone H3. Current Opinion in Genetics & Devel- opment, 14, 155–164. https://doi.org/10.1016/j.gde.2004.02.001 Carter, B., Bishop, B., Ho, K. K., Huang, R., Jia, W., Zhang, H., … Ogas, J. (2018). The chromatin remodelers PKL and PIE1 act in an epigenetic pathway that determines H3K27me3 homeostasis in Arabidopsis. The Plant Cell, 30, 1337–1352. https://doi.org/10.1105/tpc.17.00867 Chanvivattana, Y., Bishopp, A., Schubert, D., Stock, C., Moon, Y.-H., Sung, Z. R., & Goodrich, J. (2004). Interaction of Polycomb‐group proteins controlling flowering in Arabidopsis. Development, 131, 5263–5276. https://doi.org/10.1242/dev.01400 Chen, X. (2004). A microRNA as a translational repressor of APETALA2 in Arabidopsis flower development. Science, 303, 2022–2025. https://d oi.org/10.1126/science.1088060 Chen, T. W., Li, H. P., Lee, C. C., Gan, R. C., Huang, P. J., Wu, T. H., … Tang, P. (2014). ChIPseek, a web‐based analysis tool for ChIP data. BMC Genomics, 15, 539. https://doi.org/10.1186/1471-2164-15-539 Chen, C., Li, C., Wang, Y., Renaud, J., Tian, G., Kambhampati, S., … Cui, Y. (2017). Cytosolic acetyl‐CoA promotes histone acetylation pre- dominantly at H3K27 in Arabidopsis. Nature Plants, 3, 814–824. https://doi.org/10.1038/s41477-017-0023-7 Chiang, G. C. K., Barua, D., Kramer, E. M., Amasino, R. M., & Donohue, K. (2009). Major flowering time gene, FLOWERING LOCUS C, regu- lates seed germination in Arabidopsis thaliana. Proceedings of the National Academy of Sciences of the United States of America, 106, 11661–11666. https://doi.org/10.1073/pnas.0901367106 Ciferri, C., Lander, G. C., Maiolica, A., Herzog, F., Aebersold, R., & Nogales, E. (2012). Molecular architecture of human Polycomb repressive com- plex 2. eLife, 1, e00005. http://dx.doi.org/10.7554/eLife.00005.002 Clough, S. J., & Bent, A. F. (1998). Floral dip: A simplified method for Agrobacterium‐mediated transformation of Arabidopsis thaliana. The Plant Journal, 16, 735–743. https://doi.org/10.1046/j.1365-313x. 1998.00343.x Cui, X., Lu, F., Qiu, Q., Zhou, B., Gu, L., Zhang, S., … Cao, X. (2016). REF6 recognizes a specific DNA sequence to demethylate H3K27me3 and regulate organ boundary formation in Arabidopsis. Nature Genetics, 48, 694–699. https://doi.org/10.1038/ng.3556 Dellino, G. I., Schwartz, Y. B., Farkas, G., McCabe, D., Elgin, S. C. R., & Pirrotta, V. (2004). Polycomb silencing blocks transcription initiation. Molecular Cell, 13, 887–893. https://doi.org/10.1016/S1097-2765(04)00128-5 Deng, W., Buzas, D. M., Ying, H., Robertson, M., Taylor, J., Peacock, W. J., … Helliwell, C. (2013). Arabidopsis Polycomb repressive complex 2 binding sites contain putative GAGA factor binding motifs within coding regions of genes. BMC Genomics, 14, 593. https://doi.org/10. 1186/1471-2164-14-593 Derkacheva, M., & Hennig, L. (2014). Variations on a theme: Polycomb group proteins in plants. Journal of Experimental Botany, 65, 2769– 2784. https://doi.org/10.1093/jxb/ert410 Farkas, G., Gausz, J., Galloni, M., Reuter, G., Gyurkovics, H., & Karch, F. (1994). The Trithorax‐like gene encodes the Drosophila GAGA factor. Nature, 371, 806. https://doi.org/10.1038/371806a0 Farrona, S., Thorpe, F. L., Engelhorn, J., Adrian, J., Dong, X., Sarid-Krebs, L., … Turck, F. (2011). Tissue‐specific expression of FLOWERING LOCUS T in Arabidopsis is maintained independently of Polycomb group protein repression. The Plant Cell, 23, 3204–3214. https://doi. org/10.1105/tpc.111.087809 Ficz, G., Heintzmann, R., & Arndt-Jovin, D. J. (2005). Polycomb group protein complexes exchange rapidly in living Drosophila. Development, 132, 3963–3976. https://doi.org/10.1242/dev.01950 Gan, E.-S., Xu, Y., & Ito, T. (2015). Dynamics of H3K27me3 methyla- tion and demethylation in plant development. Plant Signaling & 12 | SHU ET AL. http://www.ncbi.nlm.nih.gov/nuccore/GSE108960 https://orcid.org/0000-0003-0359-6420 https://orcid.org/0000-0003-0359-6420 https://orcid.org/0000-0003-0359-6420 https://orcid.org/0000-0002-3675-3288 https://orcid.org/0000-0002-3675-3288 https://orcid.org/0000-0002-3675-3288 https://doi.org/10.1105/tpc.111.083352 https://doi.org/10.1105/tpc.111.083352 https://doi.org/10.1093/nar/gkw1273 https://doi.org/10.1007/s11103-008-9435-0 https://doi.org/10.1007/s11103-008-9435-0 https://doi.org/10.1038/nsmb.3392 https://doi.org/10.1038/nsmb.3392 https://doi.org/10.1242/dev.02340 https://doi.org/10.1371/journal.pgen.1002014 https://doi.org/10.1126/science.1076997 https://doi.org/10.1126/science.1076997 https://doi.org/10.1016/j.gde.2004.02.001 https://doi.org/10.1105/tpc.17.00867 https://doi.org/10.1242/dev.01400 https://doi.org/10.1126/science.1088060 https://doi.org/10.1126/science.1088060 https://doi.org/10.1186/1471-2164-15-539 https://doi.org/10.1038/s41477-017-0023-7 https://doi.org/10.1073/pnas.0901367106 http://dx.doi.org/10.7554/eLife.00005.002 https://doi.org/10.1046/j.1365-313x.1998.00343.x https://doi.org/10.1046/j.1365-313x.1998.00343.x https://doi.org/10.1038/ng.3556 https://doi.org/10.1016/S1097-2765(04)00128-5 https://doi.org/10.1186/1471-2164-14-593 https://doi.org/10.1186/1471-2164-14-593 https://doi.org/10.1093/jxb/ert410 https://doi.org/10.1038/371806a0 https://doi.org/10.1105/tpc.111.087809 https://doi.org/10.1105/tpc.111.087809 https://doi.org/10.1242/dev.01950 Behavior, 10, e1027851. https://doi.org/10.1080/15592324.2015. 1027851 Gaspin, C., Rami, J.-F., & Lescure, B. (2010). Distribution of short intersti- tial telomere motifs in two plant genomes: Putative origin and function. BMC Plant Biology, 10, 283. https://doi.org/10.1186/1471- 2229-10-283 Gendrel, A.-V., Lippman, Z., Martienssen, R., & Colot, V. (2005). Profiling his- tone modification patterns in plants using genomic tiling microarrays. Nature Methods, 2, 213–218. https://doi.org/10.1038/nmeth0305-213 Goodrich, J., Puangsomlee, P., Martin, M., Long, D., Meyerowitz, E. M., & Coupland, G. (1997). A Polycomb‐group gene regulates homeotic gene expression in Arabidopsis. Nature, 386, 44–51. https://doi.org/ 10.1038/386044a0 Hansen, K. H., Bracken, A. P., Pasini, D., Dietrich, N., Gehani, S. S., Mon- rad, A., … Helin, K. (2008). A model for transmission of the H3K27me3 epigenetic mark. Nature Cell Biology, 10, 1291–1300. https://doi.org/10.1038/ncb1787 Harding, E. W., Tang, W., Nichols, K. W., Fernandez, D. E., & Perry, S. E. (2003). Expression and maintenance of embryogenic potential is enhanced through constitutive expression of AGAMOUS-like 15. Plant Physiology, 133, 653–663. https://doi.org/10.1104/pp.103.023499 Hartmann, U., Höhmann, S., Nettesheim, K., Wisman, E., Saedler, H., & Huijser, P. (2000). Molecular cloning of SVP: A negative regulator of the floral transition in Arabidopsis. The Plant Journal, 21, 351–360. https://doi.org/10.1046/j.1365-313x.2000.00682.x Hodgson, J. W., Argiropoulos, B., & Brock, H. W. (2001). Site‐specific recognition of a 70‐Base‐Pair Element containing d(GA)(n) repeats mediates bithoraxoid polycomb group response element‐dependent silencing. Molecular and Cellular Biology, 21, 4528–4543. https://doi. org/10.1128/MCB.21.14.4528-4543.2001 Jiang, D., Wang, Y., Wang, Y., & He, Y. (2008). Repression of FLOWERING LOCUS C and FLOWERING LOCUS T by the Arabidopsis Polycomb repressive complex 2 components. PLoS ONE, 3, e3404. https://doi. org/10.1371/journal.pone.0003404 Katsani, K. R., Hajibagheri, M. A., & Verrijzer, C. P. (1999). Co‐operative DNA binding by GAGA transcription factor requires the conserved BTB/POZ domain and reorganizes promoter topology. The EMBO Journal, 18, 698–708. https://doi.org/10.1093/emboj/18.3.698 Lafos, M., Kroll, P., Hohenstatt, M. L., Thorpe, F. L., Clarenz, O., & Schu- bert, D. (2011). Dynamic regulation of H3K27 trimethylation during Arabidopsis differentiation. PLoS Genetics, 7, e1002040. https://doi. org/10.1371/journal.pgen.1002040 Lamesch, P., Berardini, T. Z., Li, D., Swarbreck, D., Wilks, C., Sasidharan, R., … Huala, E. (2012). The Arabidopsis Information Resource (TAIR): Improved gene annotation and new tools. Nucleic Acids Research, 40, D1202–D1210. https://doi.org/10.1093/nar/gkr1090 Langmead, B., Trapnell, C., Pop, M., & Salzberg, S. L. (2009). Ultrafast and memory‐efficient alignment of short DNA sequences to the human genome. Genome Biology, 10, R25. https://doi.org/10.1186/gb-2009- 10-3-r25 Lewis, E. B. (1978). A gene complex controlling segmentation in Droso- phila. Nature, 276, 565–570. https://doi.org/10.1038/276565a0 Li, C., Chen, C., Chen, H., Wang, S., Chen, X., & Cui, Y. (2018). Verifica- tion of DNA motifs in Arabidopsis using CRISPR/Cas9‐mediated muta- genesis. Plant Biotechnology Journal, 16, 1446–1451. https://doi.org/ 10.1111/pbi.12886 Li, C., Chen, C., Gao, L., Yang, S., Nguyen, V., Shi, X., … Cui, Y. (2015). The Arabidopsis SWI2/SNF2 chromatin remodeler BRAHMA regulates Polycomb function during vegetative development and directly acti- vates the flowering repressor gene SVP. PLoS Genetics, 11, e1004944. https://doi.org/10.1371/journal.pgen.1004944 Li, C., & Cui, Y. (2016). A DNA element that remembers winter. Nature Genetics, 48, 1451. https://doi.org/10.1038/ng.3727 Li, C., Gu, L., Gao, L., Chen, C., Wei, C. Q., Qiu, Q., … Cui, Y. (2016). Concerted genomic targeting of H3K27 demethylase REF6 and chromatin‐remodeling ATPase BRM in Arabidopsis. Nature Genetics, 48, 687–693. https://doi.org/10.1038/ng.3555 Li, D., Liu, C., Shen, L., Wu, Y., Chen, H., Robertson, M., … Yu, H. (2008). A repressor complex governs the integration of flowering signals in Arabidopsis. Developmental Cell, 15, 110–120. https://doi.org/10. 1016/j.devcel.2008.05.002 Lin, Y., Pajak, A., Marsolais, F., McCourt, P., & Riggs, C. D. (2013). Char- acterization of a cruciferin deficient mutant of Arabidopsis and its util- ity for overexpression of foreign proteins in plants. PLoS ONE, 8, e64980. https://doi.org/10.1371/journal.pone.0064980 Liu, X., Kim, Y. J., Muller, R., Yumul, R. E., Liu, C., Pan, Y., … Chen, X. (2011). AGAMOUS terminates floral stem cell maintenance in Ara- bidopsis by directly repressing WUSCHEL through recruitment of Polycomb group proteins. The Plant Cell, 23, 3654–3670. https://doi. org/10.1105/tpc.111.091538 Liu, X., Yang, Y., Hu, Y., Zhou, L., Li, Y., & Hou, X. (2018). Temporal‐speci- fic interaction of NF‐YC and CURLY LEAF during the floral transition regulates flowering. Plant Physiology, 177, 105–114. https://doi.org/ 10.1104/pp.18.00296 Lu, F., Cui, X., Zhang, S., Jenuwein, T., & Cao, X. (2011). Arabidopsis REF6 is a histone H3 lysine 27 demethylase. Nature Genetics, 43, 715. https://doi.org/10.1038/ng.854 Luo, C., Sidote, D. J., Zhang, Y., Kerstetter, R. A., Michael, T. P., & Lam, E. (2013). Integrative analysis of chromatin states in Arabidopsis identified potential regulatory mechanisms for natural antisense transcript pro- duction. Plant Journal, 73, 77–90. https://doi.org/10.1111/tpj.12017 Machanick, P., & Bailey, T. L. (2011). MEME‐ChIP: Motif analysis of large DNA datasets. Bioinformatics (Oxford, England), 27, 1696–1697. https://doi.org/10.1093/bioinformatics/btr189 Margueron, R., & Reinberg, D. (2011). The Polycomb complex PRC2 and its mark in life. Nature, 469, 343–349. https://doi.org/10.1038/nat ure09784 Mateos, J. L., Madrigal, P., Tsuda, K., Rawat, V., Richter, R., Romera-Bran- chat, M., … Coupland, G. (2015). Combinatorial activities of SHORT VEGETATIVE PHASE and FLOWERING LOCUS C define distinct modes of flowering regulation in Arabidopsis. Genome Biology, 16, 31. https://doi.org/10.1186/s13059-015-0597-1 Michaels, S. D., Ditta, G., Gustafson-Brown, C., Pelaz, S., Yanofsky, M., & Amasino, R. M. (2003). AGL24 acts as a promoter of flowering in Arabidopsis and is positively regulated by vernalization. The Plant Jour- nal, 33, 867–874. https://doi.org/10.1046/j.1365-313X.2003.01671.x Molitor, A., Latrasse, D., Zytnicki, M., Andrey, P., Houba-Herin, N., Hachet, M., … Gaudin, V. (2016). The Arabidopsis hnRNP‐Q Protein LIF2 and the PRC1 subunit LHP1 function in concert to regulate the transcription of stress‐responsive genes. The Plant Cell, 28, 2197– 2211. https://doi.org/10.1105/tpc.16.00244 Mozgova, I., & Hennig, L. (2015). The Polycomb group protein regulatory network. Annual Review of Plant Biology, 66, 269–296. https://doi. org/10.1146/annurev-arplant-043014-115627 Muller-Xing, R., Clarenz, O., Pokorny, L., Goodrich, J., & Schubert, D. (2014). Polycomb‐group proteins and FLOWERING LOCUS T maintain commitment to flowering in Arabidopsis thaliana. The Plant Cell, 26, 2457–2471. https://doi.org/10.1105/tpc.114.123323 Nicol, J. W., Helt, G. A., Blanchard, J. S. G., Raja, A., & Loraine, A. E. (2009). The Integrated Genome Browser: Free software for distribution and exploration of genome‐scale datasets. Bioinformatics (Oxford, England), 25, 2730–2731. https://doi.org/10.1093/bioinformatics/btp472 Ramirez, F., Ryan, D. P., Gruning, B., Bhardwaj, V., Kilpert, F., Richter, A. S., … Manke, T. (2016). deepTools2: A next generation web server for deep‐sequencing data analysis. Nucleic Acids Research, 44, W160– W165. https://doi.org/10.1093/nar/gkw257 Rubio-Somoza, I., & Weigel, D. (2011). MicroRNA networks and develop- mental plasticity in plants. Trends in Plant Science, 16, 258–264. https://doi.org/10.1016/j.tplants.2011.03.001 SHU ET AL. | 13 https://doi.org/10.1080/15592324.2015.1027851 https://doi.org/10.1080/15592324.2015.1027851 https://doi.org/10.1186/1471-2229-10-283 https://doi.org/10.1186/1471-2229-10-283 https://doi.org/10.1038/nmeth0305-213 https://doi.org/10.1038/386044a0 https://doi.org/10.1038/386044a0 https://doi.org/10.1038/ncb1787 https://doi.org/10.1104/pp.103.023499 https://doi.org/10.1046/j.1365-313x.2000.00682.x https://doi.org/10.1128/MCB.21.14.4528-4543.2001 https://doi.org/10.1128/MCB.21.14.4528-4543.2001 https://doi.org/10.1371/journal.pone.0003404 https://doi.org/10.1371/journal.pone.0003404 https://doi.org/10.1093/emboj/18.3.698 https://doi.org/10.1371/journal.pgen.1002040 https://doi.org/10.1371/journal.pgen.1002040 https://doi.org/10.1093/nar/gkr1090 https://doi.org/10.1186/gb-2009-10-3-r25 https://doi.org/10.1186/gb-2009-10-3-r25 https://doi.org/10.1038/276565a0 https://doi.org/10.1111/pbi.12886 https://doi.org/10.1111/pbi.12886 https://doi.org/10.1371/journal.pgen.1004944 https://doi.org/10.1038/ng.3727 https://doi.org/10.1038/ng.3555 https://doi.org/10.1016/j.devcel.2008.05.002 https://doi.org/10.1016/j.devcel.2008.05.002 https://doi.org/10.1371/journal.pone.0064980 https://doi.org/10.1105/tpc.111.091538 https://doi.org/10.1105/tpc.111.091538 https://doi.org/10.1104/pp.18.00296 https://doi.org/10.1104/pp.18.00296 https://doi.org/10.1038/ng.854 https://doi.org/10.1111/tpj.12017 https://doi.org/10.1093/bioinformatics/btr189 https://doi.org/10.1038/nature09784 https://doi.org/10.1038/nature09784 https://doi.org/10.1186/s13059-015-0597-1 https://doi.org/10.1046/j.1365-313X.2003.01671.x https://doi.org/10.1105/tpc.16.00244 https://doi.org/10.1146/annurev-arplant-043014-115627 https://doi.org/10.1146/annurev-arplant-043014-115627 https://doi.org/10.1105/tpc.114.123323 https://doi.org/10.1093/bioinformatics/btp472 https://doi.org/10.1093/nar/gkw257 https://doi.org/10.1016/j.tplants.2011.03.001 Salmon-Divon, M., Dvinge, H., Tammoja, K., & Bertone, P. (2010). PeakAnalyzer: Genome‐wide annotation of chromatin binding and modification loci. BMC Bioinformatics, 11, 415. https://doi.org/10. 1186/1471-2105-11-415 Schommer, C., Palatnik, J. F., Aggarwal, P., Chételat, A., Cubas, P., Farmer, E. E., … Weigel, D. (2008). Control of jasmonate biosynthesis and senescence by miR319 targets. PLoS Biology, 6, e230. https://doi.org/ 10.1371/journal.pbio.0060230 Schubert, D., Primavesi, L., Bishopp, A., Roberts, G., Doonan, J., Jenuwein, T., & Goodrich, J. (2006). Silencing by plant Polycomb‐group genes requires dispersed trimethylation of histone H3 at lysine 27. The EMBO Journal, 25, 4638–4649. https://doi.org/10.1038/sj.emboj.7601311 Schwartz, Y. B., Kahn, T. G., Nix, D. A., Li, X.-Y., Bourgon, R., Biggin, M., & Pirrotta, V. (2006). Genome‐wide analysis of Polycomb targets in Drosophila melanogaster. Nature Genetics, 38, 700–705. https://doi. org/10.1038/ng1817 Schwartz, Y. B., & Pirrotta, V. (2007). Polycomb silencing mechanisms and the management of genomic programmes. Nature Reviews Genet- ics, 8, 9–22. https://doi.org/10.1038/nrg1981 Smaczniak, C., Immink, R. G. H., Angenent, G. C., & Kaufmann, K. (2012). Developmental and evolutionary diversity of plant MADS‐domain factors: Insights from recent studies. Development, 139, 3081–3098. https://doi.org/10.1242/dev.074674 Tang, X., Bian, S., Tang, M., Lu, Q., Li, S., Liu, X., … Cui, Y. (2012). Micro- RNA‐mediated repression of the seed maturation program during vegetative development in Arabidopsis. PLoS Genetics, 8, e1003091. https://doi.org/10.1371/journal.pgen.1003091 Tang, X., Hou, A., Babu, M., Nguyen, V., Hurtado, L., Lu, Q., … Cui, Y. (2008). The Arabidopsis BRAHMA chromatin‐remodeling ATPase is involved in repression of seed maturation genes in leaves. Plant Physiology, 147, 1143–1157. https://doi.org/10.1104/pp.108.121996 Tang, X., Lim, M.-H., Pelletier, J., Tang, M., Nguyen, V., Keller, W. A., … Cui, Y. (2012). Synergistic repression of the embryonic programme by SET DOMAIN GROUP 8 and EMBRYONIC FLOWER 2 in Arabidopsis seedlings. Journal of Experimental Botany, 63, 1391–1404. https://doi. org/10.1093/jxb/err383 To, A., Valon, C., Savino, G., Guilleminot, J., Devic, M., Giraudat, J., & Parcy, F. (2006). A network of local and redundant gene regulation governs Arabidopsis seed maturation. The Plant Cell, 18, 1642–1651. https://doi.org/10.1105/tpc.105.039925 Trapnell, C., Roberts, A., Goff, L., Pertea, G., Kim, D., Kelley, D. R., … Pachter, L. (2012). Differential gene and transcript expression analysis of RNA‐seq experiments with TopHat and Cufflinks. Nature Protocols, 7, 562–578. https://doi.org/10.1038/nprot.2012.016 Wang, H., Liu, C., Cheng, J., Liu, J., Zhang, L., He, C., … Zhang, Y. (2016). Arabidopsis flower and embryo developmental genes are repressed in seedlings by different combinations of polycomb group proteins in association with distinct sets of cis‐regulatory elements. PLoS Genetics, 12, e1005771. https://doi.org/10.1371/journal.pgen.1005771 Wang, D., Tyson, M. D., Jackson, S. S., & Yadegari, R. (2006). Partially redundant functions of two SET‐domain polycomb‐group proteins in controlling initiation of seed development in Arabidopsis. Proceedings of the National Academy of Sciences of the United States of America, 103, 13244–13249. https://doi.org/10.1073/pnas.0605551103 Whitcomb, S. J., Basu, A., Allis, C. D., & Bernstein, E. (2007). Polycomb Group proteins: An evolutionary perspective. Trends in Genetics, 23, 494–502. https://doi.org/10.1016/j.tig.2007.08.006 Wiles, E. T., & Selker, E. U. (2017). H3K27 methylation: A promiscuous repressive chromatin mark. Current Opinion in Genetics & Develop- ment, 43, 31–37. https://doi.org/10.1016/j.gde.2016.11.001 Woodward, A. W., & Bartel, B. (2005). Auxin: Regulation, action, and interaction. Annals of Botany, 95, 707–735. https://doi.org/10.1093/ aob/mci083 Wu, G., Park, M. Y., Conway, S. R., Wang, J.-W., Weigel, D., & Poethig, R. S. (2009). The sequential action of miR156 and miR172 regulates developmental timing in Arabidopsis. Cell, 138, 750–759. https://doi. org/10.1016/j.cell.2009.06.031 Xiao, J., Jin, R., Yu, X., Shen, M., Wagner, J. D., Pai, A., … Wagner, D. (2017). Cis and trans determinants of epigenetic silencing by Poly- comb repressive complex 2 in Arabidopsis. Nature Genetics, 49, 1546– 1552. https://doi.org/10.1038/ng.3937 Xu, Y., Guo, C., Zhou, B., Li, C., Wang, H., Zheng, B., … Wu, G. (2016). Regulation of vegetative phase change by SWI2/SNF2 chromatin remodeling ATPase BRAHMA. Plant Physiology, 172, 2416–2428. https://doi.org/10.1104/pp.16.01588 Xu, M., Hu, T., Smith, M. R., & Poethig, R. S. (2016). Epigenetic regulation of vegetative phase Change in Arabidopsis. The Plant Cell, 28, 28–41. https://doi.org/10.1105/tpc.15.00854 Yang, H., Berry, S., Olsson, T. S. G., Hartley, M., Howard, M., & Dean, C. (2017). Distinct phases of Polycomb silencing to hold epigenetic memory of cold in Arabidopsis. Science, 357, 1142–1145. https://doi. org/10.1126/science.aan1121 Yang, M., Wang, X., Huang, H., Ren, D., Su, Y., Zhu, P., … Deng, X. W. (2016). Natural variation of H3K27me3 modification in two Arabidop- sis accessions and their hybrid. Journal of Integrative Plant Biology, 58, 466–474. https://doi.org/10.1111/jipb.12443 Yuan, W., Luo, X., Li, Z., Yang, W., Wang, Y., Liu, R., … He, Y. (2016). A cis cold memory element and a trans epigenome reader mediate Polycomb silencing of FLC by vernalization in Arabidopsis. Nature Genetics, 48, 1527. https://doi.org/10.1038/ng.3712 Zang, C., Schones, D. E., Zeng, C., Cui, K., Zhao, K., & Peng, W. (2009). A clustering approach for identification of enriched domains from his- tone modification ChIP‐Seq data. Bioinformatics (Oxford, England), 25, 1952–1958. https://doi.org/10.1093/bioinformatics/btp340 Zhang, X., Clarenz, O., Cokus, S., Bernatavichute, Y. V., Pellegrini, M., Goodrich, J., & Jacobsen, S. E. (2007). Whole‐genome analysis of his- tone H3 lysine 27 trimethylation in Arabidopsis. PLoS Biology, 5, e129. https://doi.org/10.1371/journal.pbio.0050129 Zhang, Y., Liu, T., Meyer, C. A., Eeckhoute, J., Johnson, D. S., Bernstein, B. E., … Liu, X. S. (2008). Model‐based analysis of ChIP‐seq (MACS). Gen- ome Biology, 9, R137. https://doi.org/10.1186/gb-2008-9-9-r 137 Zhou, Y., Hartwig, B., James, G. V., Schneeberger, K., & Turck, F. (2016). Complementary activities of TELOMERE REPEAT BINDING proteins and Polycomb group complexes in transcriptional regulation of target genes. The Plant Cell, 28, 87–101. https://doi.org/10.1105/tpc.15.00787 Zhou, Y., Romero-Campero, F. J., Gómez-Zambrano, Á., Turck, F., & Calonje, M. (2017). H2A monoubiquitination in Arabidopsis thaliana is generally independent of LHP1 and PRC2 activity. Genome Biology, 18, 69. https://doi.org/10.1186/s13059-017-1197-z Zhou, Y., Wang, Y., Krause, K., Yang, T., Dongus, J. A., Zhang, Y., & Turck, F. (2018). Telobox motifs recruit CLF/SWN–PRC2 for H3K27me3 deposition via TRB factors in Arabidopsis. Nature Genetics, 50, 638– 644. https://doi.org/10.1038/s41588-018-0109-9 SUPPORTING INFORMATION Additional supporting information may be found online in the Supporting Information section at the end of the article. How to cite this article: Shu J, Chen C, Thapa RK, et al. Genome‐wide occupancy of histone H3K27 methyltransferases CURLY LEAF and SWINGER in Arabidopsis seedlings. Plant Direct. 2019;3:1–14. https://doi.org/ 10.1002/pld3.100 14 | SHU ET AL. https://doi.org/10.1186/1471-2105-11-415 https://doi.org/10.1186/1471-2105-11-415 https://doi.org/10.1371/journal.pbio.0060230 https://doi.org/10.1371/journal.pbio.0060230 https://doi.org/10.1038/sj.emboj.7601311 https://doi.org/10.1038/ng1817 https://doi.org/10.1038/ng1817 https://doi.org/10.1038/nrg1981 https://doi.org/10.1242/dev.074674 https://doi.org/10.1371/journal.pgen.1003091 https://doi.org/10.1104/pp.108.121996 https://doi.org/10.1093/jxb/err383 https://doi.org/10.1093/jxb/err383 https://doi.org/10.1105/tpc.105.039925 https://doi.org/10.1038/nprot.2012.016 https://doi.org/10.1371/journal.pgen.1005771 https://doi.org/10.1073/pnas.0605551103 https://doi.org/10.1016/j.tig.2007.08.006 https://doi.org/10.1016/j.gde.2016.11.001 https://doi.org/10.1093/aob/mci083 https://doi.org/10.1093/aob/mci083 https://doi.org/10.1016/j.cell.2009.06.031 https://doi.org/10.1016/j.cell.2009.06.031 https://doi.org/10.1038/ng.3937 https://doi.org/10.1104/pp.16.01588 https://doi.org/10.1105/tpc.15.00854 https://doi.org/10.1126/science.aan1121 https://doi.org/10.1126/science.aan1121 https://doi.org/10.1111/jipb.12443 https://doi.org/10.1038/ng.3712 https://doi.org/10.1093/bioinformatics/btp340 https://doi.org/10.1371/journal.pbio.0050129 https://doi.org/10.1186/gb-2008-9-9-r137 https://doi.org/10.1105/tpc.15.00787 https://doi.org/10.1186/s13059-017-1197-z https://doi.org/10.1038/s41588-018-0109-9 https://doi.org/10.1002/pld3.100 https://doi.org/10.1002/pld3.100