cooperation between the h3k27me3 chromatin mark and non … · cooperation between the h3k27me3...

Cooperation between the H3K27me3 Chromatin Mark andNon-CG Methylation in Epigenetic Regulation1

Shaoli Zhou2, Xiaoyun Liu2, Chao Zhou2, Qiangwei Zhou, Yu Zhao, Guoliang Li, and Dao-Xiu Zhou*

National Key Laboratory of Crop Genetic Improvement, Huazhong Agricultural University, 430070 Wuhan,China (S.Z., X.L., C.Z., Q.Z., Y.Z., G.L., D.-X.Z.); Institute for Interdisciplinary Research, Jianghan University,430056 Wuhan, China (X.L.); and Institute of Plant Sciences of Paris-Saclay (IPS2), Université Paris-Saclay,Université Paris-sud 11, 91405 Orsay, France (D.-X.Z.)

ORCID IDs: 0000-0001-9111-1598 (S.Z.); 0000-0002-5740-9777 (X.L.); 0000-0003-4451-9907 (Y.Z.); 0000-0002-1540-0598 (D.-X.Z.).

H3K27me3 is a repressive chromatin mark of genes and is catalyzed by homologs of Enhancer of zeste [E(z)], a component ofPolycomb-repressive complex 2 (PRC2), while DNA methylation that occurs in CG and non-CG (CHG and CHH, where H is A,C, or T) contexts is a hallmark of transposon silencing in plants. However, the relationship between H3K27me3 and DNAmethylation in gene repression remains unclear. In addition, the mechanism of PRC2 recruitment to specific genes is not knownin plants. Here, we show that SDG711, a rice (Oryza sativa) E(z) homolog, is required to maintain H3K27me3 of manydevelopmental genes after shoot meristem to leaf transition and that many H3K27me3-marked developmental genes are alsomethylated at non-CG sites in the body regions. SDG711-binding and SDG711-mediated ectopic H3K27me3 also target genesmethylated at non-CG sites. Conversely, mutation of OsDRM2, a major rice CHH methyltransferase, resulted in loss of SDG711-binding and H3K27me3 from many genes and their de-repression. Furthermore, we show that SDG711 physically interacts withOsDRM2 and a putative CHG methylation-binding protein. These results together suggest that the repression of manydevelopmental genes may involve both DRM2-mediated non-CG methylation and PRC2-mediated H3K27me3 and that thetwo marks are not generally mutually exclusive but may cooperate in repression of developmentally regulated genes in rice.

Histone Lys methylation is an important epigeneticmodification for regulating genome activity and geneexpression. Histone Lys residues can be mono-, di-, ortrimethylated and may have either activating or re-pressive functions in gene expression, depending onthe position and methylation level of lysines. For in-stance, in plants, dimethylation of histone H3 Lys9 (H3K9me2) is almost exclusively associated withheterochromatin regions and is required for repres-sion of repetitive sequences, whereas trimethylationof H3 Lys 27 (H3K27me3) preferentially marks re-pressed or lowly expressed genes (Liu et al., 2010,Chen and Zhou, 2013). H3K27me3 has been impli-cated in gene regulation of many plant developmental

pathways and H3K27me3-marked genes often ex-hibit a high degree of tissue specificity (Zhang et al.,2007; Kohler and Villar, 2008; Zheng and Chen, 2011;Makarevitch et al., 2013), consistent with the func-tion of H3K27me3 in maintaining gene repressionduring growth. Whole genome chromatin immuno-precipitation (ChIP) experiments indicate that about20% of plant genes are marked by H3K27me3 (Wanget al., 2009; He et al., 2010; Lafos et al., 2011; Liuet al., 2015). There is a large tissue-specific variationof H3K27me3, with many genes shown to gain orlose H3K27me3 during cell differentiation, demon-strating a dynamic regulation of this epigeneticmodification in responding to both developmentaland environmental signals in plants (Wang et al.,2009; He et al., 2010; Lafos et al., 2011; Li et al., 2013;Liu et al., 2015).

Polycomb group (PcG) proteins, including Enhancerof Zeste [E(z)], Extra sex combs (Esc), Su(z)12, andNurf55 in Drosophila (Drosophila melanogaster) formthe Polycomb-repressive complex 2 (PRC2), whichcatalyzes H3K27me3 (Schwartz et al., 2006). Planthomologs of the four core components of PRC2 havebeen identified. The rice genome contains two genesencoding homologs of E(z) (i.e. CLF or SDG711 andEZ1 or SDG718), Su(Z)12 (i.e. EMF2a and EMF2b), andESC (i.e. FIE1 and FIE2). The function of rice PRC2homologs have been studied (Luo et al., 2009; Zhanget al., 2012a; Liu et al., 2014, 2015). For instance, down-regulation of the rice E(z) homolog SDG711 results in

1 This research was supported by grants from the National KeyResearch and Development Program of China (2014AA10A600) andthe National Natural Science Foundation of China (31571256,31371241), and supported by Huazhong Agricultural University Sci-entific and Technological Self-Innovation Foundation (Program No.2016RC003).

2 These authors contributed equally to the article.* Address correspondence to [email protected] author responsible for distribution of materials integral to the

findings presented in this article in accordance with the policy de-scribed in the Instructions for Authors (www.plantphysiol.org) is:Dao-Xiu Zhou ([email protected]).

S.Z., X.L., C.Z., and Y.Z. performed the experiments; S.Z., Q.Z.,G.L., and D.-X.Z. analyzed the data; D.-X.Z. and S.Z. designed theexperiments and wrote the paper.

www.plantphysiol.org/cgi/doi/10.1104/pp.16.01238

Plant Physiology�, October 2016, Vol. 172, pp. 1131–1141, www.plantphysiol.org � 2016 American Society of Plant Biologists. All Rights Reserved. 1131 www.plantphysiol.orgon October 19, 2020 - Published by Downloaded from

Copyright © 2016 American Society of Plant Biologists. All rights reserved.

http://orcid.org/0000-0001-9111-1598

http://orcid.org/0000-0002-5740-9777

http://orcid.org/0000-0003-4451-9907

http://orcid.org/0000-0002-1540-0598

mailto:[email protected]

http://www.plantphysiol.org

mailto:[email protected]

http://www.plantphysiol.org/cgi/doi/10.1104/pp.16.01238


genome-wide loss of H3K27me3 and affects manyplant developmental processes (Liu et al., 2014, 2015).It remains unclear how plant PRC2 targets specificgenes to mediate gene repression.

In plants, DNA cytosine methylation that occurs inCG, CHG, and CHH contexts (where H is A, C, or T) is ahallmark of inactivation of transposable elements (TE)and silent sequences (Zhang et al., 2006). CG methyla-tion is common in gene bodies of expressed genes, whilemethylation at all sequence contexts in the promoterregion represses gene transcription (Zemach et al.,2010). In Arabidopsis (Arabidopsis thaliana), CG methyl-ation ismaintainedbyMETHYLTRANSFERASE1 (MET1)and non-CG (CHG and CHH) methylation is main-tained by the redundant activities of additional DNAmethyltransferases such as DOMAINSREARRANGEDMETHYLASE 2 (DRM2), CHROMOMETHYLASES2and 3 (CMT2 and CMT3; Stroud et al., 2013, 2014;Zemach et al., 2013). DRM2 is guided by small inter-fering RNAs (siRNAs) to methylate homologous DNAregions through the process of RNA-dependent DNAmethylation (RdDM; Law et al., 2010). Recent data in-dicated that RdDM is responsible for only a smallfraction of CHH methylation, while CMT2 is the majorCHH methyltransefrase in Arabidopsis (Zemach et al.,2013; Stroud et al., 2014). CMT3 catalyzes CHG meth-ylation in cooperation with H3K9me2 (Du et al., 2012;Stroud et al., 2014). It is established that CHG methyl-ation and H3K9me2 form a positive feedback loop:CHG methylation by CMT3 requires H3K9me2, towhich CMT3 binds via its chromo-and bromo-adjacent homology domains (Law and Jacobsen,2010; Du et al., 2012). H3K9 dimethylation by SUVH(Suppressor of variegation 3-9 homolog) proteinsrequires CHG methylation, to which SUVH pro-teins bind via the SRA (SET and Ring Associated)domain (Johnson et al., 2007; Rajakumara et al.,2011).

Compared to Arabidopsis, the rice genome displaysa different global DNAmethylation landscape and hasa much higher level of genome-wide DNA methyla-tion (Li et al., 2012; Tan et al., 2016). In addition, CHHmethylation is enriched in euchromatic regions of therice genome and plays an important role in gene reg-ulation (Tan et al., 2016). However, the relationshipbetween H3K27me3 and DNA methylation and themechanism by which PRC2 targets to specific loci inrice remain unclear. In this work, we show that manyfunctional genes methylated at non-CG sites are tar-geted by SDG711-mediated H3K27me3. Further anal-ysis suggests that SDG711 may be recruited to subsetof non-CG methylated genes by interaction withOsDMR2 and/or SDG703, a member of SUVH pro-teins that recognize methylated CHG (Qin et al., 2010;Zhou andHu, 2010; Zhao and Zhou, 2012). Our resultsindicate that non-CG methylation and H3K27me3are not generally exclusive for epigenetic modifica-tion of rice genes and that DNA methylation andH3K27me3 may cooperate to repress a set of devel-opmental genes in rice.

RESULTS

SDG711 Is Involved in H3K27me3 Reprogramming duringSAM to Leaf Transition

We have previously determined genome-wideH3K27me3 in rice shoot apical meristems (SAMs; Liuet al., 2015). To study changes of H3K27me3 after SAMto leaf transition, we analyzed the genomic distribu-tions of the mark in 12-d-old seedling leaves of wild-type and SDG711 over-expression (711OX) plants byChIP-seq (Fig. S1A; Supplemental Table S1) (Liu et al.,2014). To minimize effect of callus culture duringtransgenic plant production, callus-regenerated wild-type seedling leaves were used. About 20–40 millionclean sequence reads, after filtered by the Trimmomaticsoftware (version 0.32), were aligned to the rice genome(RGAP version 7.0). Threshold of read coverage definedby randomization (Q value , 0.001) was used toidentify H3K27me3-marked genes. About 78.1% (7,113out of 9,106) of the H3K27me3-marked genes over-lapped with those identified in rice leaves in a previousstudy (Guo et al., 2015; Supplemental Fig. S1B). In ad-dition, contour plots of TPKM of the two ChIP-seqdatasets revealed a correlation coefficient equal to 0.669(Supplemental Fig. S1C), indicating the reliability of theChIP-seq data. When compared with shoot apical mer-istem (Liu et al., 2015), we observed that 3,479 genes lostH3K27me3 while 1,361 genes gained the mark in leaves(Fig. 1A). By contrast, in 711OX leaves, 2,428 genesgained H3K27me3 and 1,049 genes lost the mark com-pared to wild-type leaves (Fig. 1A). Most (1,549/2,428)of the genes that gained H3K27me3 in 711OX leavescorresponded to those that lost the mark after SAM toleaf transition in wild type (Fig. 1A; Supplemental Fig.S2), suggesting that loss of H3K27me3 in leavesmight bedue to reduced expression of SDG711 observed in leavescompared to SAM (Liu et al., 2015).

We found that many genes encoding transcriptionfactors and growth hormone (cytokinin and auxin) sig-naling proteins gained or lost H3K27me3 after SAM toleaf transition (Fig. 1B; Supplemental Table S2). Amongthose transcription factor genes that gained H3K27me3in leaves, some have been reported to be expressed inSAM and to regulate meristem functions (e.g. OSH6,OsKN2, OsMADS22, OsMDP1; Postma-Haarsma et al.,1999; Sentoku et al., 1999, 2000, 2005; Duan et al., 2006;Lee et al., 2008). In addition, we found that most of thetranscription factors (59/81), cytokinin (8/9), and auxin(5/5)-related genes that gained themark in 711OX leavescorresponded to those that lost H3K27me3 after SAM toleaf transition in wild type (Fig. 1B; Supplemental TableS2). Quantitative ChIP-PCR and RT-PCR analysis of atotal of 26 genes validated the ChIP-seq data and indi-cated that higher levels of H3K27me3 in SAM or in711OX leaves repressed the genes (Fig. 1, C and D). Theanalysis revealed that the expression level of SDG711was important for reprogramming of H3K27me3 anddevelopmental gene expression during SAM to leaftransition.

1132 Plant Physiol. Vol. 172, 2016

Zhou et al.

www.plantphysiol.orgon October 19, 2020 - Published by Downloaded from Copyright © 2016 American Society of Plant Biologists. All rights reserved.

http://www.plantphysiol.org/cgi/content/full/pp.16.01238/DC1










H3K27me3 Marks Genes Methylated at Non-CG Sites

To study genic DNA methylation in rice leaves, weanalyzed the genome-wide cytosine methylation ofthe same leaf materials as for H3K27me3 analysisby BS-Seq. We obtained about 50 million clean reads(125 bp per read) per sample. About 47%–60% of theread tags were uniquely mapped to the genome usingthe BatMeth software. After removing PCR-amplifiedredundancy, 65% and 82% of total reads in 711OX andwild type, respectively, were kept for further analysis(Supplemental Table S1). The genome coverage was,respectively, 16.4 and 11.04 for wild type and 711OXfrom the redundancy-removed reads (SupplementalTable S1), indicating a deep sequencing and a highquality of the methylome. The error conversion rates ofthe BS sequences were below 1.75% for CG, 1.23% forCHG, and 0.73 for CHH, comparable to previous data(Stroud et al., 2013).We found that about 25% of the riceprotein-coding genes were methylated at CHG or CHHsites in the body regions. The non-CGmethylated geneshad almost no transcript detected in leaves. We couldnot detect any difference with those marked by both

non-CG and H3K27 methylation (Fig. 2A). When weexamined the expression of the methylated genes indifferent rice organs, we found that the percentage ofthe double-marked genes was lower than those markedby either non-CG methylation or H3K27me3 (Fig. 2B).Importantly, many transcription factors (267) andgrowth hormone signaling protein (99) genes weremethylated at non-CG sites in the body regions(Supplemental Table S3). In wild type, about 37%(3,365/9,016) of H3K27me3-marked genes were meth-ylated at non-CG sites (Fig. 2C). Analysis of the high-throughput data indicated that genes with highH3K27me3 levels displayed significantly higher CHGand CHH methylation in the transcribed or body re-gions (P values , 0.001, t tests; Supplemental Fig. S3).Many TF and hormone-related genes were enrichedfrom the double-marked genes (Fig. 2D), suggestingthat the double marking may be particularly involvedin repression of developmental genes in rice leaves.Analysis of SDG711 overexpression plants revealedthat about 30% of the genes with ectopic or higherH3K27me3 in 711OX leaves showed methylation atnon-CG sites in the wild type (Fig. 3, A and B). The

Figure 1. Function of SDG711 in H3K27-me3 reprogramming during SAM-leaftransition. A, Numbers of genes differ-entially marked by H3K27me3 in wildtype and 711OX seedling leaves and inSAMs. B, Enrichment of hormone andtranscription factors genes for differ-ential H3K27me3 between SAMs andleaves (upper part) and between wild-type and 711OX leaves (lower part). Foldchanges are differences relative to theobserved frequencies of specific sub-sets within the genome. C, Validation oftranscript levels (top) and ChIP-seq data(bottom) of 16 randomly selected genes(14 with and 2 without changes ofH3K27me3 and expression in leafcompared to SAM). Significance of dif-ferences between the samples detectedby PCR (from 3 repeats) is indicated by**, P value, 0.01 and *, P value, 0.05.Bars are means 6 SD from three biolog-ical repeats. D, Validation of transcriptlevels (top) and ChIP-seq data (bottom)of 10 randomly selected genes (8 withand 2 without changes of H3K27me3and expression in 711OX compared towild type). Significance of differencesbetween the samples detected by PCR(from 3 repeats) is indicated by **,P value , 0.01 and *, P value , 0.05.Bars are means 6 SD from three biolog-ical repeats.

Plant Physiol. Vol. 172, 2016 1133

Interplay between H3K27me3 and Non-CG Methylation








ectopically methylated genes displayed much higherthan genome-wide levels of non-CG methylation (Fig.3, C and D). The data suggested that non-CG meth-ylation was not inhibitory for SDG711-mediatedH3K27me3 and that the two marks were not mutuallyexclusive in epigenetic modification of genes in rice.

SDG711 Binding Sites Are Enriched forNon-CG Methylation

To study genome-wide SDG711-binding sites andtheir relationship with DNA methylation, we per-formed ChIP-Seq analysis of rice wild-type 12-d-oldseedling leaves by using anti-SDG711 antibodies (Liu

et al., 2014, 2015). More than 20million clean readswereobtained, most of which (96%) were aligned to the ricereference genome (Supplemental Table S1). More than11 thousands of peaks (P value, 1e-9) were identified,which corresponded to 9,640 protein genes. Analysisof the ChIP-Seq reads revealed that, like H3K27me3,SDG711-binding was enriched in gene bodies (Fig. 4A).In addition, SDG711-associated genes displayed higherlevels of H3K27me3 than the genome-wide average(Fig. 4A). Pairwise scatter plots indicated H3K27me3was positively, although not perfectly, correlated withSDG711-binding (R2=0.665; Fig. 4B). This nonperfectcorrelation was similar to what was observed in Ara-bidopsis plants expressing FIE-HA fusion protein byanti-HA ChIP and in Drosophila (Schwartz et al., 2006;

Figure 2. Many genes are comarked byH3K27e3 and non-CG methylation. A, Left,numbers of protein-coding genes methylatedin the body region at CG, CHG, and CHHsites and their transcript levels in seedlings.Right, expression levels of genes marked byeither H3K27me3 or non-CG methylation, orby both marks. B, Expression rates of thegenes marked by either H3K27me3 ornon-CG methylation or both in different riceorgans. C, Overlapping of genes marked byH3K27me3 with those methylated at CG andnon-CG sites in the body regions. D, Enrich-ment of hormone-related and transcriptionfactor genes comarked by H3K27me3 andnon-CG methylation. Fold changes are dif-ferences relative to the observed frequenciesof specific subsets within the genome.Asterisks indicate significant enrichments(P value , 0.05, Fisher’s tests).


Zhou et al.




Deng et al., 2013), suggesting that PRC2-binding wasnot always correlated with H3K27me3. About 41.3%(3,985 out of 9,640) of genes bound by SDG711 weremethylated at non-CG sites (Fig. 4, C and E). The datasupported the above observations and suggested thatSDG711-mediated H3K27me3 targeted many genesmethylated at non-CG sites.To study if there was any consensus sequences

among SDG711-binding sites, we used MEME soft-ware and identified a CG-rich and a GAGA-like motifs(P value = 6.5e-59 and 1.3e-9, respectively; Fig. 4D).The GAGAmotif is also enriched in Arabidopsis PRC2binding sites (Deng et al., 2013).Many transcription factors genes and several cy-

tokinin and auxin-related genes were targeted bySDG711 (Supplemental Fig. S4A; Supplemental TableS2). Among these genes, some are key developmentalregulators (e.g. OsCKX1, OsCKX2/Gn1a, OsPIN3t,Crl-5, RFL, and LC1; Ashikari et al., 2005; Rao et al.,2008; Kitomi et al., 2011; Zhang et al., 2012b; Zhaoet al., 2013; Supplemental Fig. S4B). These geneswere found to be enriched for non-CG methylation(Supplemental Fig. S4C). SDG711-binding andH3K27me3of these genes were validated by ChIP-PCR (Sup-plemental Fig. S4D). The data indicated that importantdevelopmental genes were epigenetically targeted byboth SDG711-mediated H3K27me3 and non-CG meth-ylation in rice.

Loss of CHH Methylation in the osdrm2 Mutant ReducedH3K27me3 of Developmental Genes

The osdrm2 mutation resulted in a loss of morethan 85% of CHH and 23% of CHG methylation

genome-wide (Tan et al., 2016). In total, there were9,175 genes that were differentially methylated (DMG)at CHH sites (Tan et al., 2016). To study whether re-duced non-CG methylation affected H3K27me3 in therice genome, we analyzed H3K27me3 in osdrm2 byChIP-seq and found a general decrease of H3K27me3and that 1,865 genes lost the mark in the mutant (Fig.5A), among which 432 lost CHH methylation in themutant (Fig. 5B). In addition, we observed that theosdrm2 mutation also reduced H3K27me3 from manygenes that were not methylated at CHH sites. Thismight be, at least partially, due to the activation of alarge number of genes in the mutant (Tan et al., 2016),as it is shown that gene up-regulation is associatedwith reduced H3K27me3 and increased H3K4me3 inrice (Liu et al., 2015).

Among the 432 genes that lost both non-CG meth-ylation and H3K27me3, some of which were found tobe involved in SAM function (e.g. APG, TDR; Heangand Sassa, 2012; Niu et al., 2013). We selected eight ofthe genes (including OsSAUR11 and OsWRKY45) thatmet the following criteria: with loss of both non-CGmethylation and H3K27me3 in osdrm2, up-regulated inosdrm2, and bound by SDG711 in wild type for exper-imental analysis. Analysis by McrBC-digestion genesconfirmed loss of DNA methylation from the loci inthe mutant (Fig. 5C). ChIP-PCR analysis with anti-H3K27me3 of the genes confirmed the ChIP-seq dataof osdrm2 (Fig. 5D). RT-PCR analysis indicated thatthese genes were de-repressed in the mutant (Fig. 5D).Interestingly, SDG711-binding to the genes was alsoreduced in the mutant (Fig. 5D), suggesting thatOsDRM2 might be involved in SDG711-mediatedH3K27me3 at these loci.

Figure 3. SDG711-mediated H3K27me3targets genes methylated at non-CG sitesin the body regions. A, Venn map be-tween genes with ectopic H3K27me3 in711OX and non-CG methylation in wildtype plant. B, Genome browser screenshots of BS-seq and anti-H3K27me3ChIP-seq of genes with (upper four genes)or without (lower two genes) ectopicH3K27me3. C, Wild-type levels and dis-tribution of CG, CHG, and CHH methyl-ation in the 59 end (2 kb), gene body (thickline), and 39 end (2 kb) of the genes thatshowed ectopic H3K27me3 in 711OXcompared with genome-wide level. D,Box plots of CG, CHG, and CHH meth-ylation levels in the gene bodies withectopic H3K27me3 in 711OX. Signifi-cances of differences (Student’s t tests) areindicated (**, P value , 0.01).












SDG711 Physically Interacts with OsDRM2 and SDG703

To further study the mechanism by which non-CG methylation was involved in SDG711-mediatedH3K27me3, using yeast (Saccharomyces cerevisiae)two-hybrid system, we tested whether SDG711 couldinteract with rice proteins that are involved in DNAmethylation or bind to methylated cytosines. In ourmini-screening, we detected two potential interactingcandidates: OsDRM2 and SDG703 (Fig. 6A). SDG703belongs to the SUVH group proteins that containthe SRA domain capable to bind to methylated CHGsites (Johnson et al., 2007; Fig. 6A). Deletion analysisindicated that the five conserved cysteines (C5) do-main of SDG711 was necessary and sufficient for theinteraction with OsDRM2 and SDG703 (Liu et al.,2014; Fig. 6A). The N-terminal region of OsDRM2(containing the first UBA, ubiquitin-associated do-main; Dangwal et al., 2013), was involved in the

interaction with SDG711 (Fig. 6A). The interactionswere confirmed by in vitro pull-down (Fig. 6B).Coimmunoprecipitation assays of rice callus nuclearprotein extracts with anti-OsDRM2 followed bywestern blot analysis with anti-SDG711 indicatedthat the interaction of the proteins occurred in vivo(Fig. 6C). In vivo interaction between SDG711 andSDG703 was demonstrated by coimmunoprecipita-tion experiments using transfected tobacco (Nicotianabenthamiana) protoplasts (Fig. 6D). These data sug-gested that SDG711 could be recruited to non-CGmethylated loci by interacting with DRM2 and/orSDG703.

DISCUSSION

The relationship between H3K27me3 and DNAmethylation in gene silencing is unclear in plants,

Figure 4. SDG711 protein preferentiallytargets gene bodies with high non-CGmethylation. A, SDG711 binding isenriched in gene body regions. Top,Comparison of anti-SDG711 and anti-H3K27me3 ChIP-Seq reads in gene bod-ies (thick line) and 59 and 39 ends (2 kb);Bottom, comparison of H3K27me3 levelsof SDG711-binding genes with thegenome-wide averages in genic regions.B, Contour plots of anti-SDG711 and anti-H3K27me3 ChIP-Seq reads. C, Overlap ofSDG711-binding genes with those meth-ylated at non-CG sites. D, Consensus se-quences of SDG711-binding motifs. E,Genome browser screen shots of BS-seqand anti-SDG711 ChIP-seq of genes with(upper four genes) or without (lower twogenes) SDG711-binding.


Zhou et al.



although antagonistic function between the two markshas been detected in imprinted loci in Arabidopsis(Lindroth et al., 2008; Weinhofer et al., 2010). Thepresent data showing that H3K27me3 and non-CGmethylation co-occur in body region of many genessuggest that the two marks are not exclusive, but arelikely to be mutually reinforced at a set of func-tional genes in rice. Although not always perfectlyaligned at some regions of the gene bodies, the twomarks co-occurred at the gene or locus level. Theco-occurrence might reflect specific features of therice genome-wide DNA methylation. For instance,rice non-CG methylation is not peaked close to thecentromere regions as found in Arabidopsis and thatnon-CG, especially CHH, methylation is mainlyfound in genic regions and a large number of ricegenes are marked and repressed by DNA methyla-tion (Feng et al., 2010; Tan et al., 2016). The currentstudy indicates that many genes with non-CGmethylation at the body region are also marked byH3K27me3. The even lower expression levels ofnon-CG methylated genes compared to those onlymarked by H3K27me3 suggest that DNA methyla-tion plays a more prominent role in gene repressionin rice. At this stage, it is not clear whether thecomarking by H3K27me3 and non-CGmethylation isto reinforce gene repression or to transit to a differentchromatin state for chromatin remodeling during the

developmental processes, although relatively fewerdouble marked genes are expressed in differentorgans (Fig. 2B).

In mammals that have only CG methylation, thereexists a close relationship between DNA methylationand H3K27me3. Notably, in cancer, the codepen-dency of marks is largely redistributed with an in-crease of the dual repressive marks at CpG islandsand transcription start sites of silent genes (Stathamet al., 2012). In mouse embryonic stem cells, PRC2and H3K27me3 localize mainly to inactive regionsrich in CG dinucleotides or CG islands and almost97% of PRC2 target genes are associated with anno-tated CpG islands or similar CG-rich regions (Kuet al., 2008). Recently, it is shown that a cellular signalfor recruiting the polycomb machinery is encodedwithin the target DNA sequences and that CG fre-quency and positioning are sufficient to recruit PRC2and establish an H3K27me3 domain in mammals(Jermann et al., 2014).

In Drosophila, which lacks DNA methylation,polycomb can be recruited to polycomb-responsiveelements (PRE) (Ringrose et al., 2003). Binding sitesof several transcription factors including GAGA fac-tors are enriched with PRE and might contribute toPolycomb recruitment (Ringrose et al., 2003; Mullerand Kassis, 2006). Several mechanisms have been iden-tified to be involved in PRC2 recruitment to specific

Figure 5. osdrm2 mutation results in adecrease of genome-wide H3K27me3.A, Left, Venn diagram of H3K27me3-marked genes of wild type and osdrm2.Right, Pairwise scatter plots of H3K27me3levels between wild type and osdrm2.B, Venn diagram of genes that lostH3K27me3 and non-CG methylation.C, McrBC-digestion and PCR tests ofmethylation of 8 genes that lost methyl-ation in osdrm2. Three replicates areshown. D, Validations of H3K27me3modification and SDG711-binding andrelative transcript levels of the eightgenes by quantitative ChIP-PCR andRT-PCR. Significances of differences areindicated by **, P value , 0.01.





loci. In addition to interaction with sequence-specifictranscription factors or chromatin proteins, PRC2 canbe recruited by long noncoding RNAs (lncRNA) tospecific loci and mammalian PRC2 binds thousandsof RNAs in vivo (Davidovich and Cech, 2015). Inplants, the mechanisms by which PcG proteins arerecruited to specific genomic loci remain elusive, al-though it is shown that the lncRNA COLDAIR re-cruits a component of PRC2 and targets PRC2 tosilence the FLC gene (Heo and Sung, 2011). Identifi-cation of CG-rich and GAGA-like sequences asSDG711-binding motifs suggests a partial conserva-tion of PcG recruitment mechanism in differentorganisms. In addition, CG-rich and GAGA-like se-quences are potential cytosine methylation sites. Theenrichment of non-CG methylation in the body re-gion of genes targeted by SDG711-binding andH3K27me3 (Fig. 5), and physical interactions be-tween SDG711 and the CHH methyltransferaseOsDRM2 and between SDG711 and the putativeCHG-binding protein SDG703 (Fig. 6), support thehypothesis that non-CG methylation and/or OsDRM2may be involved in PRC2 recruitment andH3K27me3 ata subset of genes in rice.

A very high coincidence between H3K9me2and CHG methyaltion has been observed mainly

in TE and repetitive sequences in Arabidopsis(Bernatavichute et al., 2008). H3K9me2 and CHGmethylation form a mutually reinforced loop in-volving the H3K9 methyltransferase KRYPTONITE(KYP, or SUVH4) and the CHG methyltransferaseCMT3 (Johnson et al., 2007). The low frequency ofH3K9me2 and CHG methylation occurrence in genebody regions is attributed to activity of the Jumonji-Chistone demethylase IBM/JMJ25 that removesH3K9me2 from gene bodies (Saze et al., 2008). Recentresults showed that H3K9me2 also recruits CMT2that methylates both CHG and CHH sites in Arabi-dopsis (Stroud et al., 2014). The present data showinga high correlation between H3K27me3 and non-CGmethylation in body region of many genes raises thepossibility that H3K27me3 may replace H3K9me2 tomediate gene repression, which may perhaps preventactivation of genes by an IBM1-like activity. Thishypothesis is reminiscent of the results in Arabi-dopsis met1 mutants showing that H3K27me3 andH3K9me2/CHG methylation can be mutually exclu-sive and replace one another for transcriptional re-pression in a locus-specific manner (Deleris et al.,2012). The inhibitory effect of ectopic or high levels ofH3K27me3 on CHG methylation is in favor of thehypothesis.

Figure 6. SDG711 physically interactswith OsDRM2 and SDG703. A, Yeasttwo-hybrid assays of SDG711 andOsDRM2 or SDG703. B, Pull-downassay of interaction of SDG711 (C5domain) and the OsDRM2 protein (left)and the SDG703 protein (right). C,Co-IP assay of interaction of SDG711and OsDRM2 in vivo. Rice nuclearprotein extracts were immunoprecipi-tated by anti-OsDRM2 or IgG andanalyzed by western blot using anti-SDG711. D, Co-IP assay of interactionof SDG711 and SDG703 in vivo. Pro-tein extracts of tobacco protoplaststransfected with SDG703-HA alone,SDG703-HA together with GFP orSDG711-GFP, and SDG711-GFP alonewere checked first by western blots(WB) with anti-GFP, then immunopre-cipitated with anti-HA, followed bywestern blotting with anti-GFPand anti-HA.


Zhou et al.



MATERIAL AND METHODS

Plant Materials and Growing Conditions

Rice (Oryza sativa spp. japonica) plants used in this study for ChIP-Seq andBS-Seq analysis were from the ‘DongJin’ (DJ) background callus culture-regenerated wild type, SDG711 overexpression (lines 2, 4, and 5 combined;711OX) and osdmr2 mutant plants as described previously (Liu et al., 2014).

Seedling leaves from all genotypes grown in one-half-strength Murashigeand Skoog medium under a 16-h light/8-h dark cycle at 30°C for 12 DAG (daysafter germination) were harvested and fixed in formaldehyde for chromatincross-linking. Seedling leaves of callus culture-regenerated wild type SDG711over-expression lines grown in one-half-strength Murashige and Skoog me-dium under a 16-h light/8-h dark cycle at 30°C for 12 DAGwere harvested andfrozen in liquid nitrogen for DNA, RNA, or chromatin extraction.

Chromatin Immunoprecipitation

About 2 g of seedling leaves were crosslinked in 1% formaldehyde undervacuum. Chromatinwas extracted and fragmented to 200–750 bp by sonication,and ChIP was performed as previously described (Sun and Zhou, 2008; Huet al., 2012). Antibodies of H3K27me3 were purchased from ABclonal (a2363),and anti-SDG711 were prepared previously (Liu et al., 2014). The precipitatedand input DNA samples were analyzed either by high throughput sequencingor by real-time PCRwith gene-specific primers listed in Supplemental Table S4.All assays were performed at least three times from two biological replicates.

ChIP-Seq and Data Analysis

DNA from ChIP was used to construct sequencing libraries following theprotocol provided by Illumina TruSeq ChIP Sample Prep Set A described in(http://support.illumina.com/content/dam/illumina-support/documents/documentation/chemistry_documentation/samplepreps_truseq/truseqchip/truseq-chip-sample-prep-guide-15023092-b.pdf). Raw reads were generatedby HiSEquation 2000. Trimmomatic (version 0.32) was used to filter out low-quality reads and crop all reads to a uniform length (45 bp; Bolger et al., 2014).Clean reads were mapped to the rice genome (RGAP version 7) by defaultallowing up to two mismatches using Bowtie2 (version 2.1.0; Trapnell et al.,2012). Samtools (version 0.1.17) is used to remove potential PCR duplication.Peaks of H3K27me3 were called out by MACS software with the defaultP value (1e-5), and peaks of anti-SDG711 using a P value of 1e-9 because of thelargest percentage overlap of H3K27me3 marked genes and SDG711 boundgenes, and the output wig files were used for viewing the data by Gbrowse 2.0.

For identification of H3K27me3 marked genes, a threshold of read coveragedefined by randomization (Q value , 0.001) is used as described in Guo et al.(2015). For identification of SDG711 binding genes, genes with peaks in its genebody or in the 2Kb flanking regions were defined as SDG711-binding genes.

For quantitative analysis of histone modification, read count for a gene wasnormalized to TPKM (tags per kilobase of gene model per million mappedreads), which divided the read number of each gene by total read number,multiplied by 106, and normalization to eliminate interference of gene length bydivided by gene length andmultiplied by 1,000 (1 kb). For detecting differential(or ectopic) histone modification by quantitative comparison, the followingformula, described in (He et al., 2010), was used, in which a Bayesian approachwas applied to calculate the P value of differentially marked genes according toobserving a given number of tags. Genes with P value , 1e-4 and TPKM foldchange . 2 were considered as ectopic marked genes.

pðyjxÞ ¼�N2

N1

�y ðxþ yÞ!

x!y!�1þ N2

N1

�ðxþyþ1Þ

P ¼ min

(∑k# y

k¼0pðkjxÞ; ∑

∞

k¼ypðkjxÞ

)

BS-Seq and Data Analysis

For genomic bisulfite sequencing, the total genomic DNA of callus culture-regenerated wild type, SDG711 over-expression lines were prepared using theDNeasyplantmini kit (Qiagen).GenomicDNA(5g)wasused togenerateBS-Seqlibraries as previously described (Feng et al., 2011). Sequencing was performed

at the Novogene Bioinformatics Technology Co. Ltd, China. Trimmomatic(version 0.32) was used to filter out low quality reads and BatMeth (Lim et al.,2012) was used to align clean tags to the rice genome (RGAP version 7) bydefault parameters followed by removing PCR-amplified redundancy. Meth-ylation levels were calculated by no. C/(no. C + no. T).

For DMR and DMG, the MethylSig software was used to identify differentmethylated regions or genes by using q value (adjusted P value by the BHmethod), 0.05 andmethylation difference larger than 25, 6, and 2, respectively,for CG, CHG, and CHH (Park et al., 2014). For identification of CG-, CHG-, andCHH-methylated genes, thresholds of average methylation level of genome-wide CG (53.27%), CHG (8.32%), and CHH (2.21%) sites were used, and geneswith a sequenced cytosine number less than 50 were excluded.

Expression Analysis

For rice-seedling expression profiles, RNA-Seq data (ID: SRR576931), whichcome from the similar sample or material with us, were downloaded from SRAresources (Cui et al., 2013). Clean tags filtered by Trimmomatic (version 0.32)were aligned to rice genome (RGAP version 7) by Tophat (version 2.0.13). Thesuite of Cufflink software (version 2.2.1) was used for splicing transcripts andanalyzing expression scores (FPKM, reads per kilobase of exon model permillion mapped reads) of each gene.

Motif Search

All sequences of SDG711 binding sites were extracted for analysis byMEMEto identify consensus sequence motifs (Bailey and Elkan, 1994).

Yeast Two-Hybrid Assay

Constructs for yeast two-hybrid analysis were generated using the Match-maker Gold Yeast Two-Hybrid System (Clontech) vectors pGBKT7 and pGADT7,which express protein fusions to theGAL4DNA-binding domain or transcriptional-activation domain, respectively. Full length of cDNA inserts encoding SDG711 anddifferent domains of cDNA inserts respectively containing EZD1 domain, fiveconserved cysteines (C5) domains, and SANT, Cys-rich (CXC), and SET domain ofSDG711were introduced inpGADT7using the restriction enzymesBamHIandNdeIand the Matchmaker Gold Yeast Two-Hybrid System as described in the usermanuals. Full length of cDNA inserts encoding SDG711 andDMT706 and SDG703,different domain of cDNA inserts containing one UBA domain, two UBA domains,and a methytransferase domain of DMT706 were introduced in pGBKT7 using therestriction enzymes BamHI and SalI and the Matchmaker Gold Yeast Two-HybridSystem as described in the user manuals. The analysis was performed in strainAH109 carrying HIS3 and MEL1 reporters for reconstituted GAL4 activity.

OsDRM2 Polyclonal Antibody Preparation

For preparation of OsDRM2 polyclonal antibody, a 675 bp DNA fragmentencoding a 225-amino acid peptide of OsDRM2 (residues 1 to 225) was clonedinto pET32a vector (Novagen). The recombinant protein was expressed in E. coliDE3 (Transgen) and purified using nickel nitrilotriacetic acid agarose (Qiagen)to produce rabbit polyclonal antibodies (prepared by Abclonal).

Pull-Down Assay

Full cDNA coding sequences of OsDRM2 and SDG703 were cloned intoPET28a,while C5 domain coding sequence of SDG711was cloned into pGEX4T-1. All these recombinant vectors and empty vectors were transformed intoEscherichia coli BL21 (DE3) individually, and 0.2 mM isopropyl b-D-thiogalactoside(IPTG) was added to induce the expression of those proteins. The cell lysatecontaining 25–50 mg His-fused OsDRM2 and SDG703 or His tag were incu-bated with MagneHis Protein Purification System (V8550, GE Healthcare) at4° for 2 h under gentle agitation. After that, the coated His-fused protein beadswere washed three times in 13 phosphate-buffer saline (PBS) solution. Then,the cell lysates containing 25–50 mg GST-fused proteins or GST tag wereadded into the washed His-fused protein beads. After 1 h of incubation at 4°under gentle agitation, the bound protein-bead complexes were sedimentedby centrifugation and washed five times in 13 PBS solution. Then, the beadswere resuspended in 50mL SDS-PAGE loading buffer and heated for 10 min inthe boiled water. Protein samples were resolved on 10% SDS-PAGE gels forfurther blotting analysis using an anti-GST antibody.





http://support.illumina.com/content/dam/illumina-support/documents/documentation/chemistry_documentation/samplepreps_truseq/truseqchip/truseq-chip-sample-prep-guide-15023092-b.pdf




Co-IP Assay

For Co-IP assays, nucleoprotein extracts were prepared from the leaves oftransgenic plants with 2 volumes of immunoprecipitation buffer (25 mM Tris–HCl, pH 8.0, 1 mM EDTA, 150 mM NaCl, 0.1% NP-40, 1% Triton X-100, 0.2 mM

NaVO3, 0.2 mM PMSF and 10% [v/v] glycerol). The extracts were centrifuged at16,000 rcf for 20 min at 4°C, and the supernatant was incubated with anti-OsDRM2 and IgG antibody independently incubated protein A (Invitrogen)beads overnight at 4°C. The resin was washed three times with immunopre-cipitation buffer and then eluted in the SDS sample buffer. After separation on12% SDS-PAGE gels, the proteins were transferred onto nitrocellulose mem-branes and probed with anti-SDG711 in a dilution of 1:1000.

Accession Numbers

Sequence data from this article can be found in the Rice Genome AnnotationProject Web site (http://rice.plantbiology.msu.edu/) under the following ac-cession numbers: SDG711, Os06g16390; OsCKX1, Os01g09260; OsCKX2,Os01g10110; OsPIN3t, Os01g45550; Crl-5, 07g03250; RFL, Os04g51000; LC1,Os01g57610. The ChIP-seq and BS-seq data described in this paper have beendeposited into the GEO database (GSE71640, GSE81436).

Supplemental Data

The following supplemental materials are available.

Supplemental Figure S1. Analysis of H3K27me3-marked genes in wildtype (callus regenerated), SDG711 over-expression plants by ChIP-seqassays.

Supplemental Figure S2. Genome browser screenshots of randomly se-lected regions of H3K27me3 in wild type SAM and leaves and 711OXleaves.

Supplemental Figure S3. H3K27me3 levels correlate with non-CG meth-ylation in the rice genome.

Supplemental Figure S4. Key developmental genes that are marked byH3K27me3 and bound by SDG711 show higher levels of non-CG meth-ylation.

Supplemental Table S1. ChIP-seq data of anti-H3K27me3 and anti-SDG711 and BS-seq data.

Supplemental Table S2. Developmental genes with H3K27me3 changes inSAM-leaf transition, 711OX, osdmr2, and bound by SDG711.

Supplemental Table S3. Rice developmental genes with CHG and CHHmethylation in seedling leaves.

Supplemental Table S4. Primers used for quantitative RT-PCR, ChIP-PCR,and McrBC digestion-PCR analysiS.

ACKNOWLEDGMENTS

We thank Dr Qinghua Zhang for high throughput sequencing and QingluZhang and Prof. Xianghua Li for help in field experiments and management.

Received August 5, 2016; accepted August 16, 2016; published August 17, 2016.

LITERATURE CITED

Ashikari M, Sakakibara H, Lin S, Yamamoto T, Takashi T, Nishimura A,Angeles ER, Qian Q, Kitano H, Matsuoka M (2005) Cytokinin oxidaseregulates rice grain production. Science 309: 741–745

Bailey TL, Elkan C (1994) Fitting a mixture model by expectation maximizationto discover motifs in biopolymers. Proc Int Conf Intell Syst Mol Biol 2: 28–36

Bernatavichute YV, Zhang X, Cokus S, Pellegrini M, Jacobsen SE (2008)Genome-wide association of histone H3 lysine nine methylation withCHG DNA methylation in Arabidopsis thaliana. PLoS One 3: e3156

Bolger AM, Lohse M, Usadel B (2014) Trimmomatic: a flexible trimmer forIllumina sequence data. Bioinformatics 30: 2114–2120

Chen X, Zhou DX (2013) Rice epigenomics and epigenetics: opportunitiesand challenges. Curr Opin Plant Biol 16: 164–169

Cui X, Jin P, Cui X, Gu L, Lu Z, Xue Y, Wei L, Qi J, Song X, Luo M, An G,Cao X (2013) Control of transposon activity by a histone H3K4 de-methylase in rice. Proc Natl Acad Sci USA 110: 1953–1958

Dangwal M, Malik G, Kapoor S, Kapoor M (2013) De novo methyltrans-ferase, OsDRM2, interacts with the ATP-dependent RNA helicase,OseIF4A, in rice. J Mol Biol 425: 2853–2866

Davidovich C, Cech TR (2015) The recruitment of chromatin modifiers bylong noncoding RNAs: lessons from PRC2. RNA 21: 2007–2022

Deleris A, Stroud H, Bernatavichute Y, Johnson E, Klein G, Schubert D,Jacobsen SE (2012) Loss of the DNA methyltransferase MET1 InducesH3K9 hypermethylation at PcG target genes and redistribution ofH3K27 trimethylation to transposons in Arabidopsis thaliana. PLoSGenet 8: e1003062

Deng W, Buzas DM, Ying H, Robertson M, Taylor J, Peacock WJ, DennisES, Helliwell C (2013) Arabidopsis Polycomb Repressive Complex2 binding sites contain putative GAGA factor binding motifs withincoding regions of genes. BMC Genomics 14: 593

Du J, Zhong X, Bernatavichute YV, Stroud H, Feng S, Caro E, VashishtAA, Terragni J, Chin HG, Tu A, Hetzel J, Wohlschlegel JA, et al (2012)Dual binding of chromomethylase domains to H3K9me2-containingnucleosomes directs DNA methylation in plants. Cell 151: 167–180

Duan K, Li L, Hu P, Xu SP, Xu ZH, Xue HW (2006) A brassinolide-suppressed rice MADS-box transcription factor, OsMDP1, has a nega-tive regulatory role in BR signaling. Plant J 47: 519–531

Feng S, Jacobsen SE, Reik W (2010) Epigenetic reprogramming in plantand animal development. Science 330: 622–627

Feng S, Rubbi L, Jacobsen SE, Pellegrini M (2011) Determining DNAmethylation profiles using sequencing. Methods Mol Biol 733: 223–238

Guo Z, Song G, Liu Z, Qu X, Chen R, Jiang D, Sun Y, Liu C, Zhu Y, YangD (2015) Global epigenomic analysis indicates that epialleles contributeto Allele-specific expression via Allele-specific histone modifications inhybrid rice. BMC Genomics 16: 232

He G, Zhu X, Elling AA, Chen L, Wang X, Guo L, Liang M, He H, ZhangH, Chen F, Qi Y, Chen R, et al (2010) Global epigenetic and transcrip-tional trends among two rice subspecies and their reciprocal hybrids.Plant Cell 22: 17–33

Heang D, Sassa H (2012) Antagonistic actions of HLH/bHLH proteins areinvolved in grain length and weight in rice. PLoS One 7: e31325

Heo JB, Sung S (2011) Vernalization-mediated epigenetic silencing by along intronic noncoding RNA. Science 331: 76–79

Hu Y, Liu D, Zhong X, Zhang C, Zhang Q, Zhou DX (2012) CHD3 proteinrecognizes and regulates methylated histone H3 lysines 4 and 27 over asubset of targets in the rice genome. Proc Natl Acad Sci USA 109: 5773–5778

Jermann P, Hoerner L, Burger L, Schübeler D (2014) Short sequences canefficiently recruit histone H3 lysine 27 trimethylation in the absence ofenhancer activity and DNA methylation. Proc Natl Acad Sci USA 111:E3415–E3421

Johnson LM, Bostick M, Zhang X, Kraft E, Henderson I, Callis J, JacobsenSE (2007) The SRA methyl-cytosine-binding domain links DNA andhistone methylation. Curr Biol 17: 379–384

Kitomi Y, Ito H, Hobo T, Aya K, Kitano H, Inukai Y (2011) The auxinresponsive AP2/ERF transcription factor CROWN ROOTLESS5 is in-volved in crown root initiation in rice through the induction of OsRR1, atype-A response regulator of cytokinin signaling. Plant J 67: 472–484

Köhler C, Villar CB (2008) Programming of gene expression by Polycombgroup proteins. Trends Cell Biol 18: 236–243

Ku M, Koche RP, Rheinbay E, Mendenhall EM, Endoh M, Mikkelsen TS,Presser A, Nusbaum C, Xie X, Chi AS, Adli M, Kasif S, et al (2008)Genomewide analysis of PRC1 and PRC2 occupancy identifies twoclasses of bivalent domains. PLoS Genet 4: e1000242

Lafos M, Kroll P, Hohenstatt ML, Thorpe FL, Clarenz O, Schubert D(2011) Dynamic regulation of H3K27 trimethylation during Arabidopsisdifferentiation. PLoS Genet 7: e1002040

Law JA, Ausin I, Johnson LM, Vashisht AA, Zhu JK, Wohlschlegel JA,Jacobsen SE (2010) A protein complex required for polymerase Vtranscripts and RNA- directed DNA methylation in Arabidopsis. CurrBiol 20: 951–956

Law JA, Jacobsen SE (2010) Establishing, maintaining and modifying DNAmethylation patterns in plants and animals. Nat Rev Genet 11: 204–220

Lee S, Choi SC, An G (2008) Rice SVP-group MADS-box proteins, OsMADS22and OsMADS55, are negative regulators of brassinosteroid responses. Plant J54: 93–105


Zhou et al.


http://rice.plantbiology.msu.edu/










Li T, Chen X, Zhong X, Zhao Y, Liu X, Zhou S, Cheng S, Zhou DX (2013)Jumonji C domain protein JMJ705-mediated removal of histone H3 ly-sine 27 trimethylation is involved in defense-related gene activation inrice. Plant Cell 25: 4725–4736

Li X, Zhu J, Hu F, Ge S, Ye M, Xiang H, Zhang G, Zheng X, Zhang H,Zhang S, Li Q, Luo R, et al (2012) Single-base resolution maps of cul-tivated and wild rice methylomes and regulatory roles of DNA meth-ylation in plant gene expression. BMC Genomics 13: 300

Lim JQ, Tennakoon C, Li G, Wong E, Ruan Y, Wei CL, Sung WK (2012)BatMeth: improved mapper for bisulfite sequencing reads on DNAmethylation. Genome Biol 13: R82

Lindroth AM, Park YJ, McLean CM, Dokshin GA, Persson JM, HermanH, Pasini D, Miró X, Donohoe ME, Lee JT, Helin K, Soloway PD (2008)Antagonism between DNA and H3K27 methylation at the imprintedRasgrf1 locus. PLoS Genet 4: e1000145

Liu C, Lu F, Cui X, Cao X (2010) Histone methylation in higher plants.Annu Rev Plant Biol 61: 395–420

Liu X, Zhou C, Zhao Y, Zhou S, Wang W, Zhou DX (2014) The rice en-hancer of zeste [E(z)] genes SDG711 and SDG718 are respectively in-volved in long day and short day signaling to mediate the accuratephotoperiod control of flowering time. Front Plant Sci 5: 591

Liu X, Zhou S, Wang W, Ye Y, Zhao Y, Xu Q, Zhou C, Tan F, Cheng S,Zhou D-X (2015) Regulation of Histone Methylation and Reprogram-ming of Gene Expression in the Rice Inflorescence Meristem. The PlantCell: tpc.15.00201

Liu X, Zhou S, Wang W, Ye Y, Zhao Y, Xu Q, Zhou C, Tan F, Cheng S,Zhou DX (2015) Regulation of histone methylation and reprogrammingof gene expression in the rice inflorescence meristem. Plant Cell 27:1428–1444

Luo M, Platten D, Chaudhury A, Peacock WJ, Dennis ES (2009) Expres-sion, imprinting, and evolution of rice homologs of the polycomb groupgenes. Mol Plant 2: 711–723

Makarevitch I, Eichten SR, Briskine R, Waters AJ, Danilevskaya ON,Meeley RB, Myers CL, Vaughn MW, Springer NM (2013) Genomicdistribution of maize facultative heterochromatin marked by trimethy-lation of H3K27. Plant Cell 25: 780–793

Müller J, Kassis JA (2006) Polycomb response elements and targeting of Poly-comb group proteins in Drosophila. Curr Opin Genet Dev 16: 476–484

Niu N, Liang W, Yang X, Jin W, Wilson ZA, Hu J, Zhang D (2013) EAT1promotes tapetal cell death by regulating aspartic proteases during malereproductive development in rice. Nat Commun 4: 1445

Park Y, Figueroa ME, Rozek LS, Sartor MA (2014) MethylSig: a whole genomeDNA methylation analysis pipeline. Bioinformatics 30: 2414–2422

Postma-Haarsma AD, Verwoert II, Stronk OP, Koster J, Lamers GE, Hoge JH,Meijer AH (1999) Characterization of the KNOX class homeobox genesOskn2 and Oskn3 identified in a collection of cDNA libraries covering theearly stages of rice embryogenesis. Plant Mol Biol 39: 257–271

Qin FJ, Sun QW, Huang LM, Chen XS, Zhou DX (2010) Rice SUVH histonemethyltransferase genes display specific functions in chromatin modi-fication and retrotransposon repression. Mol Plant 3: 773–782

Rajakumara E, Law JA, Simanshu DK, Voigt P, Johnson LM, Reinberg D,Patel DJ, Jacobsen SE (2011) A dual flip-out mechanism for 5mC recognitionby the Arabidopsis SUVH5 SRA domain and its impact on DNAmethylationand H3K9 dimethylation in vivo. Genes Dev 25: 137–152

Rao NN, Prasad K, Kumar PR, Vijayraghavan U (2008) Distinct regulatoryrole for RFL, the rice LFY homolog, in determining flowering time andplant architecture. Proc Natl Acad Sci USA 105: 3646–3651

Ringrose L, Rehmsmeier M, Dura JM, Paro R (2003) Genome-wide pre-diction of Polycomb/Trithorax response elements in Drosophila mela-nogaster. Dev Cell 5: 759–771

Saze H, Shiraishi A, Miura A, Kakutani T (2008) Control of genic DNAmethylation by a jmjC domain-containing protein in Arabidopsis thali-ana. Science 319: 462–465

Schwartz YB, Kahn TG, Nix DA, Li XY, Bourgon R, Biggin M, Pirrotta V(2006) Genome-wide analysis of Polycomb targets in Drosophila mela-nogaster. Nat Genet 38: 700–705

Sentoku N, Kato H, Kitano H, Imai R (2005) OsMADS22, an STMADS11-like MADS-box gene of rice, is expressed in non-vegetative tissues andits ectopic expression induces spikelet meristem indeterminacy. MolGenet Genomics 273: 1–9

Sentoku N, Sato Y, Kurata N, Ito Y, Kitano H, Matsuoka M (1999) Re-gional expression of the rice KN1-type homeobox gene family duringembryo, shoot, and flower development. Plant Cell 11: 1651–1664

Sentoku N, Sato Y, Matsuoka M (2000) Overexpression of rice OSH genesinduces ectopic shoots on leaf sheaths of transgenic rice plants. Dev Biol220: 358–364

Statham AL, Robinson MD, Song JZ, Coolen MW, Stirzaker C, Clark SJ(2012) Bisulfite sequencing of chromatin immunoprecipitated DNA(BisChIP-seq) directly informs methylation status of histone-modifiedDNA. Genome Res 22: 1120–1127

Stroud H, Do T, Du J, Zhong X, Feng S, Johnson L, Patel DJ, Jacobsen SE(2014) Non-CG methylation patterns shape the epigenetic landscape inArabidopsis. Nat Struct Mol Biol 21: 64–72

Stroud H, Greenberg MV, Feng S, Bernatavichute YV, Jacobsen SE (2013)Comprehensive analysis of silencing mutants reveals complex regula-tion of the Arabidopsis methylome. Cell 152: 352–364

Sun Q, Zhou DX (2008) Rice jmjC domain-containing gene JMJ706 encodesH3K9 demethylase required for floral organ development. Proc NatlAcad Sci USA 105: 13679–13684

Tan F, Zhou C, Zhou Q, Zhou S, Yang W, Zhao Y, Li G, Zhou D-X (2016)Rice DNA methylation pathways display specific features of rice DNAmethylation pathways. Plant Physiol 171: 2041–2054

Trapnell C, Roberts A, Goff L, Pertea G, Kim D, Kelley DR, Pimentel H,Salzberg SL, Rinn JL, Pachter L (2012) Differential gene and transcriptexpression analysis of RNA-seq experiments with TopHat and Cuf-flinks. Nat Protoc 7: 562–578

Wang X, Elling AA, Li X, Li N, Peng Z, He G, Sun H, Qi Y, Liu XS, DengXW (2009) Genome-wide and organ-specific landscapes of epigeneticmodifications and their relationships to mRNA and small RNA tran-scriptomes in maize. Plant Cell 21: 1053–1069

Weinhofer I, Hehenberger E, Roszak P, Hennig L, Köhler C (2010)H3K27me3 profiling of the endosperm implies exclusion of polycombgroup protein targeting by DNA methylation. PLoS Genet 6: 6

Zemach A, Kim MY, Hsieh PH, Coleman-Derr D, Eshed-Williams L, ThaoK, Harmer SL, Zilberman D (2013) The Arabidopsis nucleosome re-modeler DDM1 allows DNA methyltransferases to access H1-containingheterochromatin. Cell 153: 193–205

Zemach A, McDaniel IE, Silva P, Zilberman D (2010) Genome-wide evolu-tionary analysis of eukaryotic DNA methylation. Science 328: 916–919

Zhang L, Cheng Z, Qin R, Qiu Y, Wang JL, Cui X, Gu L, Zhang X, Guo X,Wang D, Jiang L, Wu CY, et al (2012a) Identification and characteri-zation of an epi-allele of FIE1 reveals a regulatory linkage between twoepigenetic marks in rice. Plant Cell 24: 4407–4421

Zhang Q, Li J, Zhang W, Yan S, Wang R, Zhao J, Li Y, Qi Z, Sun Z, Zhu Z(2012b) The putative auxin efflux carrier OsPIN3t is involved in thedrought stress response and drought tolerance. Plant J 72: 805–816

Zhang X, Clarenz O, Cokus S, Bernatavichute YV, Pellegrini M, GoodrichJ, Jacobsen SE (2007) Whole-genome analysis of histone H3 lysine27 trimethylation in Arabidopsis. PLoS Biol 5: e129

Zhang X, Yazaki J, Sundaresan A, Cokus S, Chan SW, Chen H, Henderson IR,Shinn P, Pellegrini M, Jacobsen SE, Ecker JR (2006) Genome-wide high-resolution mapping and functional analysis of DNA methylation in arabi-dopsis. Cell 126: 1189–1201

Zhao SQ, Xiang JJ, Xue HW (2013) Studies on the rice LEAF INCLINATION1(LC1), an IAA-amido synthetase, reveal the effects of auxin in leaf inclinationcontrol. Mol Plant 6: 174–187

Zhao Y, Zhou DX (2012) Epigenomic modification and epigenetic regula-tion in rice. J Genet Genomics 39: 307–315

Zheng B, Chen X (2011) Dynamics of histone H3 lysine 27 trimethylation inplant development. Curr Opin Plant Biol 14: 123–129

Zhou DX, Hu Y (2010) Regulatory Function of Histone Modifications in Con-trolling Rice Gene Expression and Plant Growth. Rice (N Y) 3: 103–111





cooperation between the h3k27me3 chromatin mark and non … · cooperation between the h3k27me3...

Documents