Arabidopsis REF6 is a histone H3 lysine 27 demethylase - Nature

12 downloads 0 Views 970KB Size Report
Polycomb group (PcG)-mediated histone H3 lysine 27 trimethylation (H3K27me3) has a key role in gene repression and developmental regulation1–4. There is ...
letters

Arabidopsis REF6 is a histone H3 lysine 27 demethylase Polycomb group (PcG)-mediated histone H3 lysine 27 trimethylation (H3K27me3) has a key role in gene repression and developmental regulation1–4. There is evidence that H3K27me3 is actively removed in plants5–8, but it is not known how this occurs. Here we show that RELATIVE OF EARLY FLOWERING 6   (REF6), also known as Jumonji domain–containing protein 12 (JMJ12), specifically demethylates H3K27me3 and H3K27me2, whereas its metazoan counterparts, the KDM4 proteins, are H3K9 and H3K36 demethylases9,10. Plants overexpressing REF6 resembled mutants defective in H3K27me3-mediated gene silencing. Genetic interaction tests indicated that REF6 acts downstream of H3K27me3 methyltransferases. Mutations in REF6 caused ectopic and increased H3K27me3 level and decreased mRNA expression of hundreds of genes involved in regulating developmental patterning and responses to various stimuli. Our work shows that plants and metazoans use conserved mechanisms to regulate H3K27me3 dynamics but use distinct subfamilies of enzymes.

a

b

DAPI

REF6-YFP-HA

The repression of regulatory genes in development by H3K27me3 is a conserved mechanism in both plants and metazoans, and ­several thousand genes (more than 15% of all transcribed genes) in Arabidopsis are marked by such modifications1–3. In mammals, H3K27me3 is dynamically deposited by the Enhancer of Zeste homolog 2 (EZH2) and EZH1 subunits of the Polycomb repressive complex 2 (PRC2) and is removed by KDM6a and KDM6b, two Jumonji domain–containing histone demethylases10. Sequence similarity and genetic experiments have shown that Arabidopsis EZH2 homologs, including CURLY LEAF (CLF), SWINGER (SWN) and MEDEA (MEA), are potential H3K27me3 methyltransferases3,11. However, unlike histone H3K4 demethylase (JMJ14) and H3K9 demethylase (INCREASE IN BONSAI METHYLATION 1, IBM1), which could be predicted based on homology to their metazoan counterparts9,11–15, phylogenetic analysis did not reveal any homologs of KDM6a or KDM6b in Arabidopsis or rice9. Thus, although the existence of an active H3K27me3 demethylase activity is strongly implicated5–8, it remains unclear whether and how the H3K27me3 modification is removed in plants.

Methylated histone

d

c

35S

REF6H246A -YFP-HA

Methylated histone

JmjC

ZnF

H3K27me2 YFP-HA

H3K27me1

0.8 0.6

H3K27me2

0.4 0.2 0

H3K27me3 H3K27me2

Figure 1  REF6 is an H3K27me3/2 demethylase. (a) Diagrams of tagged REF6 and REF6H246A. (b) Overexpression of REF6-YFP-HA reduced H3K27me3 and H3K27me2 but not H3K27me1. (c) Statistical analysis of b. (d) Overexpression of REF6H246A-YFP-HA has no effect on H3K27 methylation. (e) Statistical analysis of d. In b and d, histone methylation is shown in red (right). We visualized nuclei transfected with REF6-YFP-HA or REF6H246A-YFP-HA by YFP signal (green; middle). Nuclei were stained with DAPI (blue; left). Arrows indicate transfected nuclei. Scale bars, 2 µm. We observed more than fifty pairs of transfected nuclei versus non-transfected nuclei in the same field of view and the results were consistent with those shown in b and d. For each of the quantifications shown in c and e, we analyzed 25 regions by comparing the integrated histone modification staining density of transfected nuclei to that of non-transfected nuclei. Error bars, s.d. (f) REF6-YFP-HA demethylates H3K27me3 and H3K27me2 in vitro. The in vitro demethylation mixture was separated by SDS-PAGE and immunoblotted using the antibodies specified on the right. We re-probed the membrane blotted with H3K27me3 with anti-H3 to confirm equal loading.

e

H3K27me1 WT REF6H246A-YFP-HA

H3K27me3

1.0

H3K27me2

0.8 0.6

H3K27me1

0.4

H3K4me3

0.2 0

f

e R YF EF P- 6H A

REF6H246A-YFP-HA

on

35S

H3K27me3

1.0

N

H3K27me3

Relative intensity

REF6-YFP-HA H246A

JmjN

DAPI

WT REF6-YFP-HA

Relative intensity

© 2011 Nature America, Inc. All rights reserved.

Falong Lu1,2,4, Xia Cui1,4, Shuaibin Zhang1,2, Thomas Jenuwein3 & Xiaofeng Cao1

H3K27me3 H3K27me2

H3K9me3 H3K36me3 H3

1State

Key Laboratory of Plant Genomics and National Center for Plant Gene Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, China. 2Graduate School of the Chinese Academy of Sciences, Beijing, China. 3Max-Planck Institute of Immunobiology and Epigenetics, Freiburg, Germany. 4These authors contributed equally to this work. Correspondence should be addressed to X. Cao ([email protected]). Received 7 December 2010; accepted 12 May 2011; published online 5 June 2011; doi:10.1038/ng.854

Nature Genetics  VOLUME 43 | NUMBER 7 | JULY 2011

715

letters

Col

lhp1

b

REF6ox

f o

p

Col

6 4 2

i

REF6ox-6

REF6ox-17

j

k

m

l

Col lhp1 REF6ox-6 REF6ox-17

AP1

AP3

PI

AG

SEP3

FT

SOC1

t

C

0

30 25 20 15 10 5 0

h

6o xR A6 EF 6o xre f6

8

s

e

q

lhp1

Relative transcript level

Cell area 2 (1,000 µm )

© 2011 Nature America, Inc. All rights reserved.

r

g

d

R A1 EF

n

c

ol

a

H3K27me3 H3K27me2

Figure 2  REF6ox plants show similar phenotypes to H3K27me3 silencing–deficient mutants. (a) Phenotype of Col, lhp1 and two REF6ox plants grown under long-day conditions. (b–e) A weak REF6ox plant shows reduced apical dominance (b), and a terminal flower (c–e), with carpelloid sepal (d) and no petals (e). (f–h) Stronger REF6ox plants that are dwarfed with short stems and unelongated pedicels (flower stems) are shown from weak to strong. (i–m) The strongest mutants show embryonic flowering (i) or embryonic flowering-like structures (j,l). The inside structures of j and l are shown enlarged in k and m. (n–q) Scanning electronic microscopy shows leaf epidermal cells of Col (n), lhp1 (o) and two REF6ox plants (p,q). White scale bars, 5 mm; yellow scale bars, 0.2 mm. (r) Mean size of the epidermal cells shown above. Error bars, s.d. (s) Expression of H3K27me3 target genes in REF6ox plants determined by RT-qPCR. Expression levels were normalized to Tubulin2. Error bars, s.d. (t) H3 lysine methylation status in two strong REF6ox plants and a ref6 loss-of-function mutant as determined by immunoblot with the antibodies specified on the right. Immunoblotting with H3 antibody showed equal H3 loading.

There are 15 potentially active JmjC domain histone demethylases in Arabidopsis9. We systematically overexpressed these proteins to identify any that reduced the level of H3K27me3. REF6 specifically demethylated H3K27me3/2 (Fig. 1a–c). REF6 promotes flowering and, when overexpressed, leads to activation of the floral integrators FT (FLOWERING LOCUS T) and SOC1 (SUPPRESSOR OF CO OVEREXPRESSION 1)16. REF6 is homologous in the JmjN and JmjC domains to metazoan KDM4 proteins, which have H3K9me2/3 and H3K36me2/3 demethylase activity9. The C-terminal part of REF6 contains four C2H2 zinc finger domains (4 × zf-C2H2), which are absent from JMJ13 and human KDM4 proteins in this subgroup9. In cells where REF6-YFP-HA was overexpressed, H3K27me3 and H3K27me2 signals, but not H3K27me1 signals, were markedly reduced (Fig. 1b,c), whereas there were no differences in mono-, dior tri-methylation levels at H3K9, H3K4 or H3K36 (Supplementary Fig. 1). The effects of REF6-YFP-HA on H3K27me3/2 levels were abolished when His246, a conserved iron-binding amino acid, was replaced by alanine (Fig. 1d,e). In addition, immunoaffinitypurified REF6-YFP-HA demethylated oligonucleosomes at H3K27me3/2, but not at H3K27me1, H3K4me3, H3K9me3 or H3K36me3 in vitro (Fig. 1f). However, it has been reported that recombinant REF6 can demethylate H3K4me3/2 and H3K36me3/2 716

H3K27me1 H3K4me3 H3K4me2 H3K4me1 H3K9me2 H3K9me1 H3K36me3 H3K36me2 H3

(ref. 17). The different activities seen may be caused by the presence of different co-factors, as has been shown for LSD1 (ref. 18). Together, our in vivo and in vitro results show that REF6 is an intrinsic H3K27me3/2-specific demethylase. H3K27me3 provides a cellular memory that allows the cell to maintain the repressed transcriptional states of target genes. In Arabidopsis, most of the H3K27me3 mark is bound by LIKE HETEROCHROMATIN PROTEIN 1 (LHP1), also known as TERMINAL FLOWER 2 (TFL2), which efficiently represses gene expression19,20. Impairment of either H3K27me3 or LHP1 causes reactivation of silenced genes and diverse developmental defects2. REF6ox (REF6-YFP-HA-­overexpressing) transgenic plants showed early flowering with upward curling leaves and various degrees of pleiotropic phenotypes (Fig. 2a–m and Supplementary Fig. 2). These phenotypes were reminiscent of the lhp1 mutant21 or emf2 (embryonic flowering 2), which is defective in a core component of the PRC2 complex22,23. To confirm the similarity between REF6ox plants and mutants that are defective in H3K27me3-mediated gene silencing, we extended our analysis to the cellular and molecular level. The leaf surface cells of REF6ox plants showed reduced size, similar to those of lhp1 mutants (Fig. 2n–r). At the molecular level, several H3K27me3 target genes, such as the MADS-box floral organ identity genes AP1 (APETALA1), AP3 (APETALA3), PI (PISTILLATA), VOLUME 43 | NUMBER 7 | JULY 2011  Nature Genetics

letters a

d

40

60

30

40

20

20

10

C o re l f6 re cl f6 f cl f

H3K27me3 hypermethylated in ref6

Transcription repressed in ref6

611

45

78

b

0

FT

0

AP1

All genes

15

10

5

−5

–45

P = 6.2 × 10

0

d

REF6ox-6 REF6ox-17

1

– 23,307,000 23,307,500 23,308,000 23,308,500

Relative transcript level

ref6

Fold enrichment

25

2

10

Expression

Col ref6

Col

+

5

log2 (RPKM)

H3K27me3

20 15 10 5 0

4 3 2 1 0

2

TCH4

Nature Genetics  VOLUME 43 | NUMBER 7 | JULY 2011

1

AP3

PI

AG

SEP3

CLF to regulate plant development23,28. We evaluated the genetic relationship between REF6 and H3K27 methyltransferases by generating a ref6 clf swn triple mutant. The ref6 clf swn triple mutant was essentially identical to clf swn mutants in phenotype, which suggests that clf swn might be epistatic to ref6 (Fig. 3a,b). We also generated a ref6 clf double mutant, whose phenotype was similar to that of clf in flowering time (Fig. 3c,d) but weaker than that of clf in leaf curling and reduced fertility (Fig. 3c and Supplementary Fig. 5), indicating that ref6 partially suppresses clf. This is consistent with the fact that CLF mutation only partially decreases H3K27me3 levels23,26. In addition, at the molecular level, the changes were similar to our phenotypic observations (Fig. 3e). Together, these genetic inter­ actions show that REF6 functions antagonistically with H3K27me3

0

c

ref6 clf Col ref6 clf ref6 clf clf swn ref6 clf swn

REF6 targets % of genes with given RPKM

© 2011 Nature America, Inc. All rights reserved.

Leaf number

25 AG (AGAMOUS) and SEP3 (SEPALLATA3), 20 which are normally expressed in flowers, 15 were ectopically activated in the seedlings of REF6ox plants. Other genes such as FT and 10 SOC1, which are expressed at low levels in 5 leaves, were derepressed. These molecular 0 patterns resemble those of the lhp1 mutant (Fig. 2s). We also observed derepression of the meristem function genes KNAT1 (knotted-like from Arabidopsis thaliana 1), KNAT6, CUC1 (CUP-SHAPED COTYLEDON 1), CUC2 and CUC3, the expressions of which are tightly restricted by H3K27me3 (Supplementary Fig. 3)24,25. H3K27me3 ChIP analysis revealed clear reduction of H3K27me3 at these loci (Supplementary Fig. 4). Finally, using two strong REF6ox transgenic plants (the phenotypes of which were similar to those shown in Fig. 2g,h) we showed that H3K27me3 and H3K27me2 underwent strong global reduction, whereas the methylation of H3K4, H3K9 and H3K36 did not (Fig. 2t). These results support our findings that REF6 functions by removing H3K27me3/2. In Arabidopsis, CLF is the main H3K27me3 methyltransferase. It directly mediates the repression of FT, FLC (FLOWERING LOCUS C) and AG by H3K27me3 and thereby controls flowering time, leaf morphology and floral organogenesis26,27. SWN acts redundantly with

e

80

Relative transcript level

ref6 clf swn Cauline leaf Rosette leaf

35 30

a

clf

clf swn

b

ref6

c

Col

Figure 3  Genetic interaction between H3K27me3 methyltransferases and REF6. (a,b) The shoot parts of clf swn double (a) and ref6 clf swn triple mutants (b) show a callus-like structure. Scale bars, 1 mm. (c) Phenotypes of Col, ref6, clf and ref6 clf plants. Scale bar, 2 cm. (d) Assessment of flowering time by counting leaf numbers in bolting plants. All the above plants were grown under long-day conditions. (e) Expression of H3K27me3 target genes in the plants shown in a–c determined by RT-qPCR. Expression levels were normalized to Tubulin2. Error bars, s.d.

TCH4

Figure 4  REF6 mutation causes H3K27me3 hypermethylation of several hundred endogenous genes. (a) Diagram of the significant overlap between genes downregulated more than 20.6-fold and H3K27me3 hypermethylated genes in ref6. Chromatin and RNA were from 10-day-old seedlings planted on Murashige-Skoog (MS) plates. (b) Plot of all expressed genes in Col seedlings and H3K27me3 hypermethylated genes in ref6. Total transcript expression levels are quantified as reads per kb of exon model per million mapped reads (RPKM) derived from 10-day-old Col seedlings from a previous study33. REF6 activates TCH4 expression by removing the H3K27me3 mark. (c) H3K27me3 ChIP-Seq data for the TCH4 locus (left). Two subregions (1 and 2) were validated by qPCR using ChIP samples of the other biological replicate (right). We used two intergenic regions without H3K27me3 modification as background controls. Plus and minus signs indicate the two DNA strands; TCH4 is on the minus strand. (d) TCH4 expression was validated by RT-qPCR using RNA samples of the other biological replicate. Expression was normalized to Tubulin2. Error bars, s.d.

717

letters

KDM4

SUV39

MEA SWN

CLF

HP1 H3K9me3 Metazoan

REF6

LHP1 H3K27me3 Arabidopsis

© 2011 Nature America, Inc. All rights reserved.

Figure 5  Diagram comparing the biochemical roles of HP1 and KDM4 in metazoans (left) with LHP1 and REF6 in Arabidopsis (right).

­ ethyltransferases and further support the hypothesis that REF6 m functions as an H3K27me3 demethylase in vivo. To test the global effects of the removal of H3K27me3 by REF6 in an unbiased manner, we performed ChIP-Seq analysis using H3K27me3 antibodies in wild-type Col and ref6. In Col, we identified 6,634 genes that underwent H3K27me3 methylation (Supplementary Table 1), which largely overlapped with those identified in a previous ChIPchip study1 (4,255 out of 4,979, 85.5%; Supplementary Fig. 6a). In addition, the patterns of wild-type H3K27me3 within the known H3K27me3 target genes are highly similar to those of the previous study1 (Supplementary Fig. 6b–d). Compared to Col, 530 regions covering 656 genes in the ref6 mutant showed more than a threefold increase in H3K27me3. Only one region, covering two genes, showed a more than threefold decrease in H3K27me3 in the ref6 mutant (Supplementary Table 2). This further illuminates the role of REF6 in H3K27me3 demethylation and indicates that most of the 656 genes should be direct targets of REF6. We confirmed binding of REF6 at 13 genes by ChIP-quantative PCR (qPCR) using antibodies to HA in a REF6øREF6-HA transgenic line that complemented the ref6-1 phenotype (Supplementary Figs. 7,8). Gene Ontology analysis revealed that in terms of molecular function, these genes were mostly enriched in transcription factor activity (P = 8.5 × 10−16), and in terms of biological processes, they are enriched in functions related to developmental processes (P = 1.1 × 10−6) and responses to organic substances (P = 1.1 × 10−6; Supplementary Fig. 9 and Supplementary Table 3). Thus, REF6 is implicated in a wide range of roles in development and responses to stimuli by removing the H3K27me3 mark. To investigate whether increased H3K27me3 influenced transcription, we performed an expression analysis using Affymetrix ATH1 arrays and identified 122 upregulated and 123 downregulated genes in the ref6 mutant (Supplementary Table 4). Forty-five of the 123 downregulated genes overlapped with the H3K27me3 hypermethylated genes (P = 6.2 × 10−45; Fig. 4a and Supplementary Table 4). By contrast, none of the 122 upregulated genes was among the hypermethylated genes. This suggests that increased H3K27me3 leads to decreased gene expression. Overall, the expression of the 656 REF6 target genes was fourfold lower than that of all the expressed genes in Col (Fig. 4b). The 45 downregulated target genes showed intermediate expression (Supplementary Fig. 10), which is consistent with the detection range of expression microarray technology29. Our data are consistent with a positive correlation between increased H3K27me3 and decreased gene expression. Six of the 45 downregulated genes, TCH4 (TOUCH 4), At2g41640, At3g04210, At4g00950, At4g23290 and At5g57800, can be induced by treatment with the plant hormone brassinosteroid30. REF6 can be recruited by bri1-EMS-SUPPRESSOR1 (BES1) to activate brassinosteroid responsive genes, and loss of REF6 results in a brassinosteroid 718

deficiency phenotype31. Our ChIP-Seq analysis revealed increased H3K27me3 at these six brassinosteroid-responsive loci (Fig. 4c and Supplementary Fig. 11). RT-qPCR showed that the expression of TCH4 was downregulated in ref6 plants but was upregulated in REF6ox plants in which the H3K27me3 level at this locus was decreased (Fig. 4d and Supplementary Fig. 4b). These data indicate that brassinosteroid-dependent activation of these genes requires H3K27me3 demethylation mediated by REF6 and suggest that removal of a repressive histone mark may be an important mechanism for activating genes in response to environmental and developmental cues in plants. In conclusion, we have identified a previously unidentified H3K27me3 demethylase and filled a major gap in our understanding of the regulation of H3K27me3 in plants (Fig. 5). The genome-wide H3K27me3 profiling and transcriptomic analysis indicated that REF6 contributes to gene activation by removing repressive H3K27me3 marks. Before this study, the molecular basis of PcG derepression was understood only in metazoans. Whether there is a uniform theme in counteracting PcG activity remains a major question in the field of epigenetics4. Our work shows that plants and metazoans use conserved mechanisms to regulate H3K27me3 dynamics but use distinct subfamilies of enzymes. In Arabidopsis, REF6 has two close homologs: EARLY FLOWERING 6 (ELF6) and JMJ13, which are potential H3K27me3 demethylases that act redundantly with REF6. REF6 is similar in sequence to metazoan H3K9me2/3 and H3K36me2/3 demethylases. In metazoans, H3K9me3 recruits HP1 to regulate gene expression (Fig. 5). The Arabidopsis genome harbors little H3K9me3 methylation32, and LHP1, the Arabidopsis homolog of HP1, binds to H3K27me3 rather than to H3K9me3 in vivo to repress gene expression. This suggests that REF6, along with LHP1, may have acquired the ability to act in H3K27me3-mediated gene silencing during evolution. Methods Methods and any associated references are available in the online ­version of the paper at http://www.nature.com/naturegenetics/. Accession codes. The ChIP-Seq dataset and ATH1 expression array dataset are deposited in Gene Expression Omnibus (GEO) under the accession code GSE25447. Note: Supplementary information is available on the Nature Genetics website. Acknowledgments We thank our colleagues for comments and advice. We thank L. Gu for technical help in handling genomic datasets, Q. Zhu for technical help, the Arabidopsis Biological Resource Center for T-DNA insertion lines and I. Hanson for editing. This work was supported by the National Basic Research Program of China (grants 2009CB941500 and 2011CB915400 to X. Cao), and by the National Natural Science Foundation of China (grants 30930048 and 30921061 to X. Cao and 30971619 to X. Cui). AUTHOR CONTRIBUTIONS F.L., X. Cui and X. Cao conceived and designed the study. F.L., X. Cui and S.Z. performed the experiments. T.J. contributed essential reagents and edited the manuscript. F.L., X. Cui and X. Cao analyzed data and wrote the paper. COMPETING FINANCIAL INTERESTS The authors declare no competing financial interests. Published online at http://www.nature.com/naturegenetics/. Reprints and permissions information is available online at http://www.nature.com/ reprints/index.html. 1. Zhang, X. et al. Whole-genome analysis of histone H3 lysine 27 trimethylation in Arabidopsis. PLoS Biol. 5, e129 (2007).

VOLUME 43 | NUMBER 7 | JULY 2011  Nature Genetics

© 2011 Nature America, Inc. All rights reserved.

letters 2. Hennig, L. & Derkacheva, M. Diversity of Polycomb group complexes in plants: same rules, different players? Trends Genet. 25, 414–423 (2009). 3. Pien, S. & Grossniklaus, U. Polycomb group and trithorax group proteins in Arabidopsis. Biochim. Biophys. Acta 1769, 375–382 (2007). 4. Sawarkar, R. & Paro, R. Interpretation of developmental signaling at chromatin: the Polycomb perspective. Dev. Cell 19, 651–661 (2010). 5. Schatlowski, N., Creasey, K., Goodrich, J. & Schubert, D. Keeping plants in shape: polycomb-group genes and histone methylation. Semin. Cell Dev. Biol. 19, 547–553 (2008). 6. Finnegan, E.J. & Dennis, E.S. Vernalization-induced trimethylation of histone H3 lysine 27 at FLC is not maintained in mitotically quiescent cells. Curr. Biol. 17, 1978–1983 (2007). 7. Kwon, C.S., Lee, D., Choi, G. & Chung, W.I. Histone occupancy-dependent and -independent removal of H3K27 trimethylation at cold-responsive genes in Arabidopsis. Plant J. 60, 112–121 (2009). 8. Charron, J.B., He, H., Elling, A.A. & Deng, X.W. Dynamic landscapes of four histone modifications during deetiolation in Arabidopsis. Plant Cell 21, 3732–3748 (2009). 9. Lu, F. et al. Comparative analysis of JmjC domain-containing proteins reveals the potential histone demethylases in Arabidopsis and rice. J. Integr. Plant Biol. 50, 886–896 (2008). 10. Agger, K., Christensen, J., Cloos, P.A. & Helin, K. The emerging functions of histone demethylases. Curr. Opin. Genet. Dev. 18, 159–168 (2008). 11. Liu, C., Lu, F., Cui, X. & Cao, X. Histone methylation in higher plants. Annu. Rev. Plant Biol. 61, 395–420 (2010). 12. Inagaki, S. et al. Autocatalytic differentiation of epigenetic modifications within the Arabidopsis genome. EMBO J. 29, 3496–3506 (2010). 13. Saze, H., Shiraishi, A., Miura, A. & Kakutani, T. Control of genic DNA methylation by a jmjC domain-containing protein in Arabidopsis thaliana. Science 319, 462– 465 (2008). 14. Lu, F., Cui, X., Zhang, S., Liu, C. & Cao, X. JMJ14 is an H3K4 demethylase regulating flowering time in Arabidopsis. Cell Res. 20, 387–390 (2010). 15. Searle, I.R., Pontes, O., Melnyk, C.W., Smith, L.M. & Baulcombe, D.C. JMJ14, a JmjC domain protein, is required for RNA silencing and cell-to-cell movement of an RNA silencing signal in Arabidopsis. Genes Dev. 24, 986–991 (2010). 16. Noh, B. et al. Divergent roles of a pair of homologous jumonji/zinc-finger-class transcription factor proteins in the regulation of Arabidopsis flowering time. Plant Cell 16, 2601–2613 (2004). 17. Ko, J.H. et al. Growth habit determination by the balance of histone methylation activities in Arabidopsis. EMBO J. 29, 3208–3215 (2010).

Nature Genetics  VOLUME 43 | NUMBER 7 | JULY 2011

18. Lan, F., Nottke, A.C. & Shi, Y. Mechanisms involved in the regulation of histone lysine demethylases. Curr. Opin. Cell Biol. 20, 316–325 (2008). 19. Zhang, X. et al. The Arabidopsis LHP1 protein colocalizes with histone H3 Lys27 trimethylation. Nat. Struct. Mol. Biol. 14, 869–871 (2007). 20. Turck, F. et al. Arabidopsis TFL2/LHP1 specifically associates with genes marked by trimethylation of histone H3 lysine 27. PLoS Genet. 3, e86 (2007). 21. Larsson, A.S., Landberg, K. & Meeks-Wagner, D.R. The TERMINAL FLOWER2 (TFL2) gene controls the reproductive transition and meristem identity in Arabidopsis thaliana. Genetics 149, 597–605 (1998). 22. Gaudin, V. et al. Mutations in LIKE HETEROCHROMATIN PROTEIN 1 affect flowering time and plant architecture in Arabidopsis. Development 128, 4847–4858 (2001). 23. Chanvivattana, Y. et al. Interaction of Polycomb-group proteins controlling flowering in Arabidopsis. Development 131, 5263–5276 (2004). 24. Shen, W.-H. & Xu, L. Chromatin Remodeling in Stem Cell Maintenance in Arabidopsis thaliana. Mol. Plant 2, 600–609 (2009). 25. Chen, D., Molitor, A., Liu, C. & Shen, W.H. The Arabidopsis PRC1-like ring-finger proteins are necessary for repression of embryonic traits during vegetative growth. Cell Res. 20, 1332–1344 (2010). 26. Jiang, D., Wang, Y. & He, Y. Repression of FLOWERING LOCUS C and FLOWERING LOCUS T by the Arabidopsis Polycomb repressive complex 2 components. PLoS ONE 3, e3404 (2008). 27. Goodrich, J. et al. A Polycomb-group gene regulates homeotic gene expression in Arabidopsis. Nature 386, 44–51 (1997). 28. Schubert, D. et al. Silencing by plant Polycomb-group genes requires dispersed trimethylation of histone H3 at lysine 27. EMBO J. 25, 4638–4649 (2006). 29. Wang, Z., Gerstein, M. & Snyder, M. RNA-Seq: a revolutionary tool for transcriptomics. Nat. Rev. Genet. 10, 57–63 (2009). 30. Nemhauser, J.L., Mockler, T.C. & Chory, J. Interdependency of brassinosteroid and auxin signaling in Arabidopsis. PLoS Biol. 2, E258 (2004). 31. Yu, X., Li, L., Guo, M., Chory, J. & Yin, Y. Modulation of brassinosteroid-regulated gene expression by Jumonji domain-containing proteins ELF6 and REF6 in Arabidopsis. Proc. Natl. Acad. Sci. USA 105, 7618–7623 (2008). 32. Johnson, L. et al. Mass spectrometry analysis of Arabidopsis histone H3 reveals distinct combinations of post-translational modifications. Nucleic Acids Res. 32, 6511–6518 (2004). 33. Deng, X. et al. Arginine methylation mediated by the Arabidopsis homolog of PRMT5 is essential for proper pre-mRNA splicing. Proc. Natl. Acad. Sci. USA 107, 19114–19119 (2010).

719

ONLINE METHODS

© 2011 Nature America, Inc. All rights reserved.

Plant materials. Nicotiana benthamiana was grown at 25 °C under long-day (16 h light, 8 h dark) conditions. All Arabidopsis materials were grown at 23 °C under long-day conditions if not specified. The ref6 mutant used here was ref6-3 (SAIL_747_A07) if not specified16. clf was clf-28 (SALK_139371)34 and swn was swn-4 (SALK_109121)35. The lhp1 mutant used in this study was tfl2-1 (ref. 21). Transgenes. Primer sequences can be found in Supplementary Table 5. The full length REF6 genomic coding sequence without its stop codon was first cloned into pENTR/D-TOPO (Invitrogen) using primers cx3592 and cx3593 and then introduced by LR reaction (recombination reaction between attL and attR sites using the Gateway technology from Invitrogen) into pEarleyGate101 (pEG101) vector with a C-terminal YFP-HA tag (REF6-pEG101)36. REF6H246A was made by site-directed mutagenesis using a QuickChange kit (Stratagene) with primers cx5264 and cx5265. The full length REF6 genomic coding sequence including 2 kb of promoter but without the stop codon was first cloned into pENTR/DTOPO (Invitrogen) using primers cx4046 and cx3593 and then introduced by LR reaction into pEarleyGate301 (pEG301) vector with a C-terminal HA tag (REF6:REF6-HA)36. The constructs were transformed into Agrobacterium tumefaciens cells (strain EHA105). Agrobacteria were then infiltrated into N. benthamiana leaves as described37,38, or the constructs were stably transformed into Arabidopsis wild-type Col using the floral dip method39. The REF6: REF6-HA transgene was introduced into ref6-1 (SALK_001018)16. Forty-eight hours after infiltration, tobacco leaves were harvested for nuclei isolation and immunostaining or immunoprecipitation. T1 Arabidopsis seeds were planted in soil and selected by spraying with 100 mg/l BASTA 3–5 d after germination. In vivo histone demethylation assay. The demethylation assay was carried out as described11,12,14. Half of each tobacco leaf was infiltrated with A. tumefaciens EHA105 strains containing REF6-pEG101 or REF6H246A-pEG101 to express REF6-YFP-HA or REF6H246A-YFP-HA. The other leaf halves, without infiltration, were used as controls. Forty-eight hours after infiltration, the whole tobacco leaves were cut into pieces of about 0.5 cm × 1 cm and fixed in cold 4% paraformaldehyde in Tris-HCl buffer (10 mM Tris pH 7.5 at 25 °C, 100 mM NaCl, 10 mM EDTA) for 20 min. The leaves were then washed twice using ice-cold Tris-HCl buffer for 10 min each. Isolation and immunostaining of nuclei were carried out as described14. In brief, nuclei were released by chopping leaves in LB01 buffer (15 mM Tris-HCl pH 7.5, 2 mM EDTA, 0.5 mM spermine, 80 mM KCl, 20 mM NaCl, 0.1% Triton X-100) until very fine and were filtered through a cell strainer cup (BD falcon). Nuclei in the flow through were then diluted 1:4 in sorting buffer (100 mM Tris pH 7.5, 50 mM KCl, 2 mM MgCl2, 0.05% Tween20, 5% sucrose), spotted onto microscopy slides and air dried. After fixation with 4% paraformaldehyde in PBS buffer (10 mM sodium phosphate, pH 7.0, 143 mM NaCl), the slides were used for immunostaining according to the manufacturer’s instructions for the Alexa Fluor SFX kit (Invitrogen). Immunolabeling was performed using histone methylationspecific antibodies (H3K4me3: Abcam ab8580, 1:100; H3K4me2: Millipore 07-030, 1:1,250; H3K4me1: Millipore 07-436, 1:100; H3K9me3: Millipore 07-442, 1:100; H3K9me2: Millipore 07-441, 1:200; H3K9me1: Millipore 07450, 1:100; H3K27me3: from T. Jenuwein, 1:200; H3K27me2: Millipore 07-452, 1:100; H3K27me1: Millipore 07-448, 1:100; H3K36me3: Abcam ab9050, 1:100; H3K36me2: Millipore 07-274, 1:100; H3K36me1: Millipore 07-548, 1:100). The modified histones were revealed by Alexa Fluor 555-conjugated goat anti-rabbit (1:200, Invitrogen). Transfected cells were revealed by monitoring the YFP signal. After staining, the slides were mounted in VECTASHIELD mounting medium with DAPI (Vector Laboratory) and then photographed with an OLYMPUS BX51 fluorescence microscope. More than fifty pairs of transfected nuclei versus non-transfected nuclei in the same field of view were observed. Quantification was performed using ImageJ software by comparing histone methylation integrated staining density of nuclei overexpressing REF6-YFP-HA or REF6H246A-YFP-HA to that of the local neighboring wildtype nuclei12. In vitro demethylation assay. REF6-YFP-HA was immunoaffinity purified from transiently expressing tobacco leaves. The demethylation assay was carried out as previously described14. In brief, enzyme and Hela oligonucleosomes

Nature Genetics

were incubated in reaction buffer (20 mM HEPES-NaOH, 150 mM NaCl, 50 µM Fe(NH4)2(SO4)2, 1 mM α-ketoglutarate, 2 mM ascorbate, pH 8.0) for 4 h at 30 °C. The reaction product was analyzed by protein blotting using modification-specific antibodies. Cell size quantification. Leaf epidermal cell size was measured using the fifth rosette leaf of 4-week-old wild-type, REF6ox and lhp1 seedlings. Epidermal cells from the middle part of the leaves were examined by scanning electron microscopy and then measured using ImageJ 1.43. At least ten epidermal cells were measured per strain. Transcript level analysis. Total RNA was extracted 10 d after germination (DAG) from whole seedlings grown under long-day conditions on MS plates using TRIzol reagent (Invitrogen). For RT-qPCR, reverse transcriptase reactions were performed using oligo dT primers. Quantitative PCR was performed using a CFX96 Real-time PCR Instrument (Bio-Rad) with the SYBR Green (Invitrogen, S-7567) method (final concentration of SYBR Green used in reaction mix was 0.15×). Primers for qPCR can be found in Supplementary Table 5. The AP1, AP3, PI, AG and SEP3 qPCR primers were as described40. For genome-wide expression analysis of ref6, two replicates of Col and ref6 samples were analyzed on Affymetrix ATH1 arrays by an Affymetrix service facility (CapitalBio Corporation) according to the manufacturer’s protocols. Genes showing a 20.6-fold change with a q-value < 0.05 were considered to be differentially expressed. Total histone protein blot. The shoot parts of 3-week-old plants in soil were ground to a fine powder in liquid nitrogen and resuspended in 3× SDS loading buffer. The mixture was sonicated for six bursts of 10 s and then boiled at 99 °C for 5–10 min. Insoluble material was removed by centrifugation for 10 min at 15,000 g. The supernatants were then used for protein blot using the antibodies listed below. Anti-H3 immunoblot was used as a loading control. Antibodies: anti-H3: ab1791 (Abcam), anti-H3K27me3: 07-449 (Millipore), anti-H3K27me2: 07-452 (Millipore), anti-H3K27me1: 07-448 (Millipore), anti-H3K4me3: 07-473 (Millipore), anti-H3K4me2: 07-030 (Millipore), anti-H3K4me1: MABI0302 (Wako), anti-H3K9me2: 07-441 (Millipore), anti-H3K9me1: MABI0306 (Wako), anti-H3K36me3: ab9050 (Abcam), antiH3K36me2: 07-274 (Millipore). Chromatin immunoprecipitation (ChIP). ChIP was performed essentially as described with the minor modification that the crosslinking was performed on ten DAG Arabidopsis seedlings powdered in liquid nitrogen using a pestle and mortar for H3K27me3 ChIP14. Antibodies used for ChIP were anti-H3K27me3 from T. Jenuwein, anti-H3K27me3: 07-449 (Millipore), antiH3: ab1791 (Abcam) and anti-HA: H6908 (Sigma). The ChIPed DNA was then used for Illumina single-end sequencing or qPCR. The primers used for qPCR can be found in Supplementary Table 5. Two intergenic regions without H3K27me3 modification were used as H3K27me3-negative controls (NCs: NC2 and NC3). ChIP-Seq analysis. ChIPed DNA was ligated with Illumina single-end genome sequencing adaptors, size fractionated to obtain 300-bp fragments, PCR amplified and sequenced according to standard protocols (single end 36 cycles). The reads were aligned to the TAIR9 assembly using the ELAND extended program (Illumina), allowing up to two mismatches; reads that mapped to multiple loci were discarded. The MACS program41 was used to shift the reads and convert the data to WIG format. WIG files were visualized with the Integrated Genome Browser (IGB)42. H3K27me3-enriched domains were called with the SICER program43. The ChIPDiff program44 was used for quantitative comparisons of H3K27me3 levels in Col and ref6. All three of the data processing programs remove redundant reads to avoid PCR bias. Finally the gene list in Supplementary Table 2 was manually checked to avoid false positives caused by the 1-kb window size used in the ChIPDiff program. H3K27me3 target genes (Col_Zhang in Supplementary Fig. 4a) were obtained from reference 1. RNA level plot. Recent advances in transcript detection technology (RNASeq) make it possible to accurately compare relative transcript levels within one transcriptome using reads per kb of exon model per million mapped reads

doi:10.1038/ng.854

(RPKM) as an indicator45. The RPKM data were obtained from RNA-Seq data of ten DAG Col samples similar to those used in our transcriptome analysis33. All the expressed genes (genes with at least one valid sequence in the RNA-Seq dataset) were used as the Col dataset. x axis, expression level (log2(RPKM)); y axis, percentage of genes with given expression level. Flowering time assessment. Mutants and control plants were grown in soil side by side in a light-tight growth room at 23 °C under long-day conditions. Flowering time was assessed by counting the number of rosette and cauline leaves when the plants flowered. At least fifteen plants were counted for each line.

© 2011 Nature America, Inc. All rights reserved.

34. Doyle, M.R. & Amasino, R.M. A single amino acid change in the enhancer of zeste ortholog CURLY LEAF results in vernalization-independent, rapid flowering in Arabidopsis. Plant Physiol. 151, 1688–1697 (2009). 35. Wang, D., Tyson, M.D., Jackson, S.S. & Yadegari, R. Partially redundant functions of two SET-domain polycomb-group proteins in controlling initiation of seed development in Arabidopsis. Proc. Natl. Acad. Sci. USA 103, 13244–13249 (2006). 36. Earley, K.W. et al. Gateway-compatible vectors for plant functional genomics and proteomics. Plant J. 45, 616–629 (2006).

37. English, J.J., Davenport, G.F., Elmayan, T., Vaucheret, H. & Baulcombe, D.C. Requirement of sense transcription for homology-dependent virus resistance and trans-inactivation. Plant J. 12, 597–603 (1997). 38. Zhang, Y. et al. SDIR1 is a RING finger E3 ligase that positively regulates stressresponsive abscisic acid signaling in Arabidopsis. Plant Cell 19, 1912–1929 (2007). 39. Clough, S.J. & Bent, A.F. Floral dip: a simplified method for Agrobacterium-mediated transformation of Arabidopsis thaliana. Plant J. 16, 735–743 (1998). 40. Kotake, T., Takada, S., Nakahigashi, K., Ohto, M. & Goto, K. Arabidopsis TERMINAL FLOWER 2 gene encodes a heterochromatin protein 1 homolog and represses both FLOWERING LOCUS T to regulate flowering time and several floral homeotic genes. Plant Cell Physiol. 44, 555–564 (2003). 41. Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008). 42. Nicol, J.W., Helt, G.A., Blanchard, S.G. Jr., Raja, A. & Loraine, A.E. The Integrated Genome Browser: free software for distribution and exploration of genome-scale datasets. Bioinformatics 25, 2730–2731 (2009). 43. Zang, C. et al. A clustering approach for identification of enriched domains from histone modification ChIP-Seq data. Bioinformatics 25, 1952–1958 (2009). 44. Xu, H., Wei, C.L., Lin, F. & Sung, W.K. An HMM approach to genome-wide identification of differential histone modification sites from ChIP-seq data. Bioinformatics 24, 2344–2349 (2008). 45. Mortazavi, A., Williams, B.A., McCue, K., Schaeffer, L. & Wold, B. Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat. Methods 5, 621–628 (2008).

doi:10.1038/ng.854

Nature Genetics