Selective epigenetic control of retrotransposition in ... - SFC Bio

1 downloads 0 Views 540KB Size Report
Sep 6, 2009 - silenced by DNA methylation1,4, which in Arabidopsis is propa- gated at CG ... the presence of EVD extrachromosomal DNA in epi12 plants but.
Vol 461 | 17 September 2009 | doi:10.1038/nature08328

LETTERS Selective epigenetic control of retrotransposition in Arabidopsis Marie Mirouze1*, Jon Reinders1*, Etienne Bucher1, Taisuke Nishimura1, Korbinian Schneeberger2, Stephan Ossowski2, Jun Cao2, Detlef Weigel2, Jerzy Paszkowski1 & Olivier Mathieu1,3

Retrotransposons are mobile genetic elements that populate chromosomes, where the host largely controls their activities1–3. In plants and mammals, retrotransposons are transcriptionally silenced by DNA methylation1,4, which in Arabidopsis is propagated at CG dinucleotides by METHYLTRANSFERASE 1 (MET1)5. In met1 mutants, however, mobilization of retrotransposons is not observed, despite their transcriptional activation4–6. A posttranscriptional mechanism therefore seems to be preventing retrotransposition. Here we show that a copia-type retrotransposon (E´vade´, French for ‘fugitive’) evaded suppression of its movement during inbreeding of hybrid epigenomes consisting of met1- and wild-type-derived chromosomes. E´vade´ (EVD) reinsertions caused a series of developmental mutations that allowed its identification. Genetic testing of host control of the EVD life cycle showed that transcriptional suppression occurred by CG methylation supported by RNA-directed DNA methylation. On transcriptional reactivation, subsequent steps of the EVD cycle were inhibited by plant-specific RNA polymerase IV/V7,8 and the histone methyltransferase KRYPTONITE (KYP). Moreover, genome resequencing demonstrated retrotransposition of EVD but no other potentially active retroelements when this combination of epigenetic mechanisms was compromised. Our results demonstrate that epigenetic control of retrotransposons extends beyond transcriptional suppression and can be individualized for particular elements. We established a population of epigenetic recombinant inbred lines (epiRILs) derived from a cross between a wild-type (WT) and a met1 homozygous plant carrying a null mutation for MET1 (met1-3 mutant allele5). At the F8 generation of inbreeding by single-seed descent in the presence of MET1 (selected for in the F2), these lines displayed mosaic epigenomes9. We found rare individuals with aberrant development among the epiRILs. For example, in F8 line 12 (epi12), 4% of plants lacked male flower organs and petals (Fig. 1a), resembling mutations in the LEAFY (LFY) gene and subsequently confirmed by genetic mapping. A large insert at the LFY locus (Fig. 1b) was found to be a full-length Ty1/copia long terminal repeat (LTR) retrotransposon (5.3 kilobases (kb)) belonging to the ATCOPIA93 family (Supplementary Fig. 1) and encoded by the AT5G17125 locus (Fig. 1c), hereafter referred to as EVD. Two aberrant plants in epi49 (Fig. 1d, f) had an EVD insertion into the BRASSINOSTEROID INSENSITIVE 1 (BRI1) (Fig. 1e) and VARIEGATED 2 (VAR2) genes (Fig. 1g), respectively. Thus, EVD was independently activated in two different epiRILs and inserted into three unlinked loci (LFY on chromosome 5, BRI1 on chromosome 4, and VAR2 on chromosome 2). EVD was present as two nearly identical copies in WT plants and this was unaltered in the epiRIL parent met1 (Fig. 1h). In contrast, EVD was amplified in epi12

and epi49, with hybridization signals at the size expected for EVD extrachromosomal DNA (Fig. 1h and Supplementary Fig. 2). Movement of EVD was detected in 29 of the epiRILs examined for EVD mobilization and extrachromosomal DNA (Supplementary Fig. 3). Previously we reported active transposition of the DNA transposon CACTA in 18 lines of the same epiRIL population9. Because the two transposons were simultaneously active in only six lines (Supplementary Fig. 3), it is unlikely that a common regulatory switch controls the activity of both elements. To understand EVD mobilization in the epiRILs, its life cycle was examined in lines with active or immobile retrotransposons. Analysis of DNA methylation in the area of transcription initiation at the 59 LTR of EVD showed high CG methylation (mCG) levels in WT plants but mCG was erased in met1 and epi12 plants (Fig. 2a). However, m CG was retained in epi07, where there was no sign of EVD activity (Fig. 1h and Supplementary Fig. 2). Thus, met1 and WT parental epialleles of EVD were inherited in epi12 and epi07 plants, respectively. Remarkably, EVD 59 LTR methylation at CHH and CHG (H 5 A, T or C) sites was almost absent, even in WT plants (Fig. 2a). We examined whether erasure of mCG in met1 and epi12 allowed the production of RNA spanning the entire retrotransposon. Transcripts about 5 kb long corresponding to full-length EVD were present in met1 and epi12 plants but not in WT and epi07 plants (Fig. 2b). Two shorter non-polyadenylated transcripts were detected that may reflect aberrant or cleaved EVD RNAs (Fig. 2b), as observed previously for other retrotransposons10. Analyses of sequence polymorphism confirmed that all EVD transcripts originated from the AT5G17125 locus (Supplementary Fig. 4). Thus, EVD transcription was silenced at AT5G17125 by mCG in WT and epi07 plants. In the EVD life cycle, the long transcript should be translated and subjected to reverse transcription, leading to extrachromosomal DNA2. Southern blot analysis with non-restricted DNA revealed the presence of EVD extrachromosomal DNA in epi12 plants but not in met1 plants, although both the level and pattern of the EVD transcripts were very similar in the two lines (Fig. 2b, c). This suggested that translation and/or reverse transcription, which proceeded in epi12 plants, was blocked in met1 plants. Transposon silencing involves small RNAs (sRNAs) 24 and 21 nucleotides in length acting through RNA-directed DNA methylation (RdDM) or messenger RNA cleavage/translational inhibition, respectively11–13. The absence of EVD extrachromosomal DNA in met1 could result from translational inhibition, preventing accumulation of its reverse transcriptase. However, this is unlikely because 21-nucleotide sRNAs were present in both met1 and epi12 plants (Fig. 2d). Although EVD RNAs could be guided for cleavage by the 21-nucleotide sRNAs, they nevertheless accumulated. Moreover,

1 Department of Plant Biology, University of Geneva, Sciences III, 30 Quai Ernest-Ansermet, CH-1211 Geneva 4, Switzerland. 2Max Planck Institute for Developmental Biology, 72076 Tu¨bingen, Germany. 3Centre National de la Recherche Scientifique (CNRS), UMR 6247, GReD, INSERM U 931, 24 avenue des Landais, Clermont Universite´, 63177 Aubie`re, France. *These authors contributed equally to this work.

427 ©2009 Macmillan Publishers Limited. All rights reserved

LETTERS

LFY (AT5G61850)

kb

0.5 kb

2.4 1.8

1.6

c

epi12 ‘lfy’

H




EVD (AT5G17125)

‘lfy’

5.6

HH

0.5 kb

WT

EVD (AT5G17125)

8.9

3.6

0.5 kb

2.2

3.6

WT

‘WT’ ‘var2’

e

EVD (AT5G17125)

0.5 kb

3.4

W T m et 1 ep -3 i0 ep 7 i1 ep 2 i4 9

et 1 ep -3 i0 7 ep i1 2

m

24 21

epi07

epi12

kb 4.2 4.0 3.6

VAR2 (AT2G30950)

Chromosome 1 AT1G34904

kb E


<


4.2 kb

>

0.6 kb EtBr

Figure 1 | Mobilization of EVD LTR retrotransposon in epiRILs. LFY, BRI1, VAR2 and EVD loci structures are represented with the following components: grey, white and yellow boxes and blue arrows for exons, introns, untranslated regions and long terminal repeats, respectively. HindIII (H) and EcoRV (E) restriction sites, fragment sizes, primers (black arrows) and probe position (black bar) are indicated. White scale bars represent 1 cm (1 mm in insets). a, ‘lfy’ phenotype in epi12. WT (top) and lfy (bottom) epi12 inflorescences and magnified flowers are shown. b, Southern blot analysis using HindIII and LFY probe on F2 progeny of the backcross between ‘lfy’ epi12 and WT plants with the WT or lfy phenotype (BC ‘WT’, and BC ‘lfy’, respectively). WT, met1-3 and epi12 ‘WT’ are controls. c, PCR analysis and sequencing of the LFY locus. d, ‘bri1’ phenotype in epi49. e, PCR analysis and sequencing of the BRI1 locus. f, ‘var2’ phenotype in epi49. g, PCR analysis and sequencing of the VAR2 locus. h, Southern blot analysis with EcoRV and EVD probe A (black bar). epi07, epi12 and epi49 (F8 generation) are compared with WT and met1-3 parents. The arrow and asterisk indicate extrachromosomal and internal EVD fragments, respectively. The ethidium bromide stained gel (EtBr) is shown as a loading control. Smears in epi12 and epi49 suggest somatic transposition. The restriction map shows ATCOPIA93 loci on chromosomes 1 and 5.

although EVD-specific 24-nucleotide sRNAs were ubiquitous (Fig. 2d), the non-CG methylation hallmark of RdDM was absent from the 59 LTR (Fig. 2a). Thus, these sRNAs are not channelled to RdDM. The lack of a correlation between the EVD-specific sRNAs and any steps in retrotransposon mobilization raises questions about their regulatory function. Because EVD seemed to be immobile in the strains parental to the epiRILs examined, the onset of its activity during inbreeding of epi12 was investigated. Beginning with the F4 generation, extrachromosomal DNA was associated with an increased EVD copy number (Fig. 2e). New insertions were detected only in F7 plants (Fig. 2e).

0.6

EtBr

Figure 2 | EVD life cycle in epiRILs and the met1-3 parent. a, Bisulphite DNA methylation analysis at the EVD 59 LTR. Methylation levels are given at CG, CHG and CHH sites. b, Northern blot analysis with total and poly(A)1 RNAs probed with an EVD full-length probe. ACTIN2 (ACT2) was used as loading control. c, Southern blot analysis with undigested DNA and EVD probe A (see Fig. 1h) revealed the presence of extrachromosomal EVD copies (EVD ecDNA) in epi12 plants but not in epi07 plants (both at F8) or in the parents (WT and met1-3). The ethidium bromide stained gel (EtBr) is shown as a loading control. d, EVD small RNAs detection with an LTR probe and internal probe (EVD probe B, see Supplementary Fig. 8 for probe information). miR159 and EtBr are shown as loading controls. e, EcoRVdigested DNA from successive epi07 and epi12 generations was hybridized with EVD probe A (legend as for Fig. 1h). Smears in epi12 suggest increasing somatic transposition.

Mobilization of EVD therefore seemed to be progressive during epiRIL inbreeding. We reported previously that self-propagation of met1 mutants leads to stochastically formed and irregularly inherited novel DNA methylation patterns14. We therefore examined EVD transposition in met1 mutants propagated by self-fertilization over four generations. Progressive EVD mobilization and accumulation of extrachromosomal DNA were detected from the second generation of homozygous met1 plants (Fig. 3a and Supplementary Fig. 5). EVD transposition varied between met1 lines (data not shown), probably as a consequence of stochastic changes in epigenetic marks characteristic of these strains14, which also occurred in the epiRILs9. Neither transposition nor extrachromosomal DNA accumulation was detected in the first generation of met1, suggesting that mechanisms other than m CG prevented EVD life-cycle progression. To define these mechanisms, the function of RdDM was examined by depletion of either RNA polymerases IV and V (Pol IV and Pol V) or DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2) by using mutations in NRPD2/NRPE2 (ref. 8) (nrpd2a-2) encoding the common subunit of Pol IV and Pol V, or in DRM2 (drm2-2)

428 ©2009 Macmillan Publishers Limited. All rights reserved

LETTERS

NATURE | Vol 461 | 17 September 2009

c 60 50 40 30

EVD

+

met1/+ nrpd2a

met1 nrpd2a

met1 cmt3

met1 kyp

met1 drm2

nrpd2a-2

kyp-7

cmt3-11

met1-3 2nd

drm2-2

met1 nrpd2a

met1 nrpd2a

met1 nrpd2a

met1 kyp

met1 cmt3

nrpd2a-2

d

WT

met1 cmt3

met1 kyp

met1 drm2

nrpd2a-2

cmt3-11

kyp-7

drm2-2

met1-3 2nd

RT

WT

b

met1 drm2

met1 nrpd2a kyp-7

2.3 2.0

CHH

cmt3-11

20 10 0

met1-3 (1st)

CHG

drm2-2

5.3

nrpd2a-2

CG

met1 nrpd2a

kb

e 80 70

WT

Methylation of cytosines (%)

4th

3rd

2nd

1st

WT

met1-3

met1-3 2nd

a

f

1

2

3

4

5

3-1 EVD

kb 2-1



4-1

ACT2

+

3-2 3-3



4-3

5-3

4-2 5-1

5.3

3-4 5-2

2.3 2.0

Figure 3 | Genetic analysis of EVD transposition control. a, Southern blot analysis of EVD transposition in successive met1-3 generations. SspI-digested DNA was hybridized to an EVD fragment (probe C; see Supplementary Fig. 8). The arrow shows EVD extrachromosomal DNA. b, RT–PCR analysis of EVD transcripts. ACTIN2 (ACT2) was used to normalize RNA template amounts; negative controls lacked reverse transcriptase (RT2). c, Bisulphite DNA methylation at the EVD 59 LTR, as described in Fig. 2a. d, Southern blot

analysis of the indicated genotypes performed as in a. e, Three independent met1-3 nrpd2a-2 double-mutant plants show severe phenotypes compared with nrpd2a-2 and met1-3 single mutant siblings. Scale bar, 1 cm. f, Scheme of Arabidopsis chromosomes showing EVD retrotransposition events found in a met1/1 nrpd2a individual (filled circles). The original EVD locus is shown as a red square; red circles indicate EVD insertions verified by PCR assay (see Supplementary Fig. 7).

encoding the main de novo DNA methyltransferase in Arabidopsis. EVD transcription was not activated in nrpd2a or drm2 single mutants (Fig. 3b), which is consistent with unaltered DNA methylation at the EVD 59 LTR (Fig. 3c). However, met1 nrpd2a and met1 drm2 double mutants showed a synergistic increase in EVD transcript levels (Fig. 3b). Thus, in the absence of mCG, the RdDM pathway restrained the accumulation of EVD transcript, although this was not reflected by changes in DNA methylation (Fig. 3c). EVD extrachromosomal DNA accumulated and EVD transposed only in met1 nrpd2a plants but not in met1 drm2 plants (Fig. 3d and Supplementary Fig. 6), revealing a functional separation of DRM2 and Pol IV/V in the control of EVD transposition. Moreover, this control was achieved beyond transcription, with NRPD2/NRPE2 having a central function independent of DRM2 and, thus, probably of RdDM. Combinations of met1-3 and nrpd2a-2 alleles as double mutants were recovered only rarely. These plants displayed severe developmental abnormalities (Fig. 3e) reminiscent of met1 kyp mutants that we described previously14, which in addition to MET1 lacked the major histone H3 lysine 9 (H3K9) methyltransferase15. The met1 kyp double mutant also had high EVD activity (Fig. 3d) but, in contrast to nrpd2a, the kyp mutation combined with met1 did not influence EVD transcript level (Fig. 3b). Thus, KYP restrained EVD mobilization exclusively at the post-transcriptional level. This control must differ from the previously reported transcriptional restriction of the rice Tos17 copia-like retrotransposon by H3K9 methyltransferase SDG714, which altered DNA methylation levels16. It is possible that KYP methylates a protein important for translation and/or reverse transcription of EVD transcripts. Indeed, recent studies have identified non-histone substrates for lysine methyltransferases17–19. It has been shown that KYP-mediated H3K9 methylation directs non-CG DNA methylation by the plant-specific CHROMO-

METHYLASE 3 (CMT3)20. Mutations in KYP and CMT3 genes could therefore have a similar outcome in terms of EVD transposition. As shown for kyp, mutation of cmt3 did not influence the EVD 59 LTR methylation pattern or EVD transcript levels (Fig. 3b, c). However, in contrast with the met1 kyp double mutant, EVD extrachromosomal DNA accumulated to a low level in met1 cmt3 first-generation double mutants, and no EVD insertions were detected (Fig. 3d and Supplementary Fig. 6). Thus, the function of KYP in post-transcriptional control of EVD seems to be independent of CMT3. However, possible functional redundancy between CMT3 and DRM1 and/or DRM2 (ref. 21) could be contributing to EVD control. To address this, met1 drm1 drm2 and met1/1 drm1 drm2 cmt3 individuals were also tested (met1 drm1 drm2 cmt3 quadruple mutants are extremely rare and have severe developmental defects22). Only marginal redundancy between DRM1 and DRM2 in controlling EVD activity was revealed (Supplementary Fig. 6). Our results therefore support the hypothesis that Pol IV/V and KYP control retrotransposition posttranscriptionally through a mechanism that is independent of DNA methylation. We assumed that the epigenetic mechanisms involved in the control of EVD also protect the genome against mobilization of other transposons. We therefore sequenced the genome of a met1/MET1 nrpd2a/nrpd2a (met1/1 nrpd2a) mutant plant with high EVD activity (Fig. 3d and Supplementary Fig. 6). This could not be performed on a met1 nrpd2a double mutant because insufficient tissue was available (Fig. 3e). Sequencing identified new EVD insertion sites (Fig. 3f and Supplementary Fig. 7), but DNA transposons, including CACTA23, non-LTR retrotransposons and other families of potentially active LTR-containing retrotransposons24, remained immobile. These data suggest that EVD is the only transposon that moves in this genetic background. 429

©2009 Macmillan Publishers Limited. All rights reserved

LETTERS

NATURE | Vol 461 | 17 September 2009

Although DNA methylation has been associated with suppression of transposon-derived transcription, our results show that ‘classical’ epigenetic players also restrain subsequent steps of retrotransposition. Exemplified by EVD, we show that Pol IV/V and KYP are both required for post-transcriptional control after transcriptional reactivation. This regulation seems to apply only to EVD, suggesting that other retrotransposons remain immobilized by additional or alternative epigenetic mechanisms. In this respect, the SWI2/SNF2-like chromatinremodelling protein DECREASE IN DNA METHYLATION 1 (DDM1) seems to have a pivotal function in controlling retrotransposition; however, inbreeding of ddm1 mutants seems to be essential for the mobilization of retrotransposons, as shown in the accompanying paper24. This resembles met1 mutants14 and epiRILs9, in which inbreeding is associated with epigenetic instabilities and activation of transposition. In contrast, double-mutant combinations of met1 nrpd2a or met1 kyp, in which post-transcriptional control of EVD is compromised, activate transposition instantaneously and inbreeding is not required. It is now necessary to define how the host individualizes the control of different groups of transposons and how selectivity of this regulation has evolved.

9.

10. 11. 12. 13. 14.

15.

16. 17. 18.

19.

METHODS SUMMARY 20.

Genomic DNA analyses were performed by Southern blotting. DNA methylation levels at the EVD 59 LTR were assessed by bisulphite sequencing. Accumulation of EVD transcripts was determined by RT–PCR or northern blot analysis, and sRNA detection was performed by northern blotting. The genome of the met1/1 nrpd2a mutant line was analysed by Illumina sequencing-by-synthesis.

21.

Full Methods and any associated references are available in the online version of the paper at www.nature.com/nature.

23. 24.

Received 1 June; accepted 27 July 2009. Published online 6 September 2009. 1. 2.

3. 4. 5.

6. 7.

8.

22.

Slotkin, R. K. & Martienssen, R. A. Transposable elements and the epigenetic regulation of the genome. Nature Rev. Genet. 8, 272–285 (2007). Beauregard, A., Curcio, M. J. & Belfort, M. The take and give between retrotransposable elements and their hosts. Annu. Rev. Genet. 42, 587–617 (2008). Goodier, J. L. & Kazazian, H. H. Retrotransposons revisited: the restraint and rehabilitation of parasites. Cell 135, 23–35 (2008). Zhang, X. et al. Genome-wide high-resolution mapping and functional analysis of DNA methylation in Arabidopsis. Cell 126, 1189–1201 (2006). Saze, H., Mittelsten Scheid, O. & Paszkowski, J. Maintenance of CpG methylation is essential for epigenetic inheritance during plant gametogenesis. Nature Genet. 34, 65–69 (2003). Lister, R. et al. Highly integrated single-base resolution maps of the epigenome in Arabidopsis. Cell 133, 523–536 (2008). Pontier, D. et al. Reinforcement of silencing at transposons and highly repeated sequences requires the concerted action of two distinct RNA polymerases IV in Arabidopsis. Genes Dev. 19, 2030–2040 (2005). Wierzbicki, A. T., Haag, J. R. & Pikaard, C. S. Noncoding transcription by RNA polymerase Pol IVb/Pol V mediates transcriptional silencing of overlapping and adjacent genes. Cell 135, 635–648 (2008).

Reinders, J. et al. Compromised stability of DNA methylation and transposon immobilization in mosaic Arabidopsis epigenomes. Genes Dev. 23, 939–950 (2009). Hirochika, H., Okamoto, H. & Kakutani, T. Silencing of retrotransposons in Arabidopsis and reactivation by the ddm1 mutation. Plant Cell 12, 357–369 (2000). Matzke, M. A., Kanno, T., Huettel, B., Daxinger, L. & Matzke, A. J. Targets of RNAdirected DNA methylation. Curr. Opin. Plant Biol. 10, 512–519 (2007). Brodersen, P. et al. Widespread translational inhibition by plant miRNAs and siRNAs. Science 320, 1185–1190 (2008). Slotkin, R. K. et al. Epigenetic reprogramming and small RNA silencing of transposable elements in pollen. Cell 136, 461–472 (2009). Mathieu, O., Reinders, J., Caikovski, M., Smathajitt, C. & Paszkowski, J. Transgenerational stability of the Arabidopsis epigenome is coordinated by CG methylation. Cell 130, 851–862 (2007). Jackson, J. P. et al. Dimethylation of histone H3 lysine 9 is a critical mark for DNA methylation and gene silencing in Arabidopsis thaliana. Chromosoma 112, 308–315 (2004). Ding, Y. et al. SDG714, a histone H3K9 methyltransferase, is involved in Tos17 DNA methylation and transposition in rice. Plant Cell 19, 9–22 (2007). Chuikov, S. et al. Regulation of p53 activity through lysine methylation. Nature 432, 353–360 (2004). Kouskouti, A., Scheer, E., Staub, A., Tora, L. & Talianidis, I. Gene-specific modulation of TAF10 function by SET9-mediated methylation. Mol. Cell 14, 175–182 (2004). Sampath, S. C. et al. Methylation of a histone mimic within the histone methyltransferase G9a regulates protein complex assembly. Mol. Cell 27, 596–608 (2007). Jackson, J. P., Lindroth, A., Cao, X. & Jacobsen, S. E. Control of CpNpG DNA methylation by the KRYPTONITE histone H3 methyltransferase. Nature 416, 556–560 (2002). Cao, X. et al. Role of the DRM and CMT3 methyltransferases in RNA-directed DNA methylation. Curr. Biol. 13, 2212–2217 (2003). Zhang, X. & Jacobsen, S. E. Genetic analyses of DNA methyltransferases in Arabidopsis thaliana. Cold Spring Harb. Symp. Quant. Biol. 71, 439–447 (2006). Miura, A. et al. Mobilization of transposons by a mutation abolishing full DNA methylation in Arabidopsis. Nature 411, 212–214 (2001). Tsukahara, S. et al. Burst of retrotransposition reproduced in Arabidopsis. Nature doi:10.1038/nature08351 (this issue).

Supplementary Information is linked to the online version of the paper at www.nature.com/nature. Acknowledgements We thank all members of the Paszkowski laboratory for discussion; L. Broger, M. Freyre, J. Nicolet, C. Me´gies and J. Lafleuriel for technical assistance; M. Dapp and X. Zhang for met1/1 drm1/1 drm2/1 cmt3/1 material; C. Lanz for help in preparing the Illumina libraries and running the instruments and P. King and E. Lacombe for critically reading the manuscript. This work was supported by grants from the Swiss National Science Foundation (3100A0-102107), the European Commission through The Epigenome (LSHG-CT2004-503433) and Targeted Gene Integration in Plants (TAGIP; 018785) and the Max Planck Society. Author Contributions M.M., J.R., J.P. and O.M. conceived the study. M.M., J.R., E.B., T.N. and O.M performed the experiments. K.S., S.O., J.C. and D.W. contributed Illumina sequencing-by-synthesis data and analysis. M.M., J.R., J.P. and O.M. wrote the paper with contributions from all co-authors. Author Information Reprints and permissions information is available at www.nature.com/reprints. Correspondence and requests for materials should be addressed to J.P. ([email protected]) or O.M. ([email protected]).

430 ©2009 Macmillan Publishers Limited. All rights reserved

doi:10.1038/nature08328

METHODS Plant material. EpiRILs9, met1-3 (ref. 5), kyp-7 (ref. 25), drm1-2 (ref. 21), drm2-2 (ref. 26), cmt3-11 (ref. 26) and nrpd2a-2 (ref. 27) are all in the Col-0 background. Plants were grown in soil in long-day conditions (growth chamber, 16 h light, 8 h dark, 22 uC). Genomic DNA analysis. DNA was extracted from leaf tissue with the MaxiPrep kit (Qiagen). Southern blotting was performed as described previously14 (see Supplementary Fig. 8 and Supplementary Table 1 for probe and primer details, respectively). DNA methylation analysis. Bisulphite conversion was perfomed as described previously28, using the Epitect kit (Qiagen). PCR products were cloned using the pGEMT vector (Promega). For each sample, 8–12 clones were sequenced. Sequencing data were analysed with CyMATE (http://www.gmi.oeaw.ac.at/en/ cymate-index/cymate-v2/). Primers are listed in Supplementary Table 1. Expression analysis. RNA was isolated from leaves by using TRI Reagent (Sigma), and RT–PCR was performed as described previously14. For northern blot and sRNA analysis, total RNA was isolated from flower tissue and enriched for sRNAs with the use of the mirVana miRNA isolation kit (Ambion). Northern blotting for large RNAs was performed with 10 mg of RNAs. Poly(A)1 RNAs were purified with the Oligotex kit (Qiagen), and 160-ng aliquots were used for northern blot analysis. sRNAs were detected as described previously29,30. EVD probes. The probe localizations are described in Supplementary Fig. 8. All EVD probes, except EVD probe C, were prepared as gel-purified PCR products (primers listed in Supplementary Table 1). For EVD probe C, the full-length EVD element was amplified from Col-0 genomic DNA using primers EVD_XhoI and EVD_XmaI (see Supplementary Table 1 for sequence information) and cloned into the pBSK(2) plasmid vector, and the probe was subsequently obtained by digesting pBSK:EVD with XhoI and SspI and gel-purifying the resulting ,2-kb fragment. Illumina sequencing-by-synthesis (SBS). Genomic DNA libraries from the original Col-0 and the derived met1/1 nrpd2a mutant line were prepared as follows. Leaf tissue (0.5 g portions) was ground to a fine powder in liquid N2 and transferred to a tube containing 10 ml of ice-cold nuclei extraction buffer (10 mM Tris-HC1 pH 9.5, 10 mM EDTA pH 8.0, 100 mM KC1, 500 mM sucrose, 4 mM spermidine, 1 mM spermine, 0.1% 2-mercaptoethanol). The homogenized tissues were filtered through two layers of Miracloth (Calbiochem), and 2 ml of lysis buffer (10% Triton X-100 in nuclei extraction buffer) was added to the filtered homogenate and kept on ice for 2 min before centrifugation at 2,000g for 10 min at 4 uC. The nuclei pellet was resuspended in 500 ml of CTAB extraction buffer (100 mM Tris-HC1 pH 7.5, 0.7 M NaCl, 10 mM EDTA pH 8.0, 1% CTAB, 1% 2-mercaptoethanol) and incubated for 30 min at 60 uC. After extraction with chloroform/isopentanol (24:1), DNA was precipitated with propan-2-ol and washed with 75% ethanol. Paired-end SBS library preparation, cluster generation and SBS sequencing were performed as described previously31, with the modification that a paired-end DNA sample kit (catalogue no. PE-1021001; Illumina Inc.), a paired-end cluster generation kit (catalogue no. PE-1031002l; Illumina) and a sequencing kit (catalogue no. FC-204-2036; Illumina) were used on a Genome Analyser II (catalogue no. SY-301-1201; Illumina). DNA

fragments of ,300 bp were used for library production and a 42-imaging-cycle recipe was used in SBS sequencing. Image analysis of the output of ten lanes from two Genome Analyser II runs was performed with GAPipeline version 1.0, with default parameters, resulting in 6.81 3 107 and 7.72 3 107 read pairs for the mutant and background samples, respectively. Unfiltered reads were used as input for the SHORE pipeline, version 0.2.4beta (http://1001genomes.org/ downloads)31. Reads were subjected to quality filtering, trimming and subsequent alignment against the unmasked A. thaliana reference sequence32 using GenomeMapper in accordance with the SHORE manual. Average genome read coverage was about 163 and 173 for mutant and background samples, respectively. To detect the locations of unknown retroelement insertions, we focused on read pairs derived from the edge of the new insertion sites, where one of the read pairs is sampled from the DNA near the insertion and the other from DNA within the insertion. The expectancy is that such read pairs feature one read with a unique alignment in the vicinity of the insertion site, and the other read aligns repetitively to the origin of the retroelement as well as to all its homologues in the genome sequence. We therefore analysed all read pairs in which one read mapped uniquely in the genome, whereas the read pair featured multiple alignments. We clustered all read pairs for which the unique alignments of the two reads were not farther apart than 500 bp. Additionally, at least 25% of the repetitive mappings of their read pairs did not exceed a pairwise distance of 300 bp. Any 500-bp window featuring a cluster with more than 14 members was considered to be near a potential insertion site, resulting in 547 such sites. From these we excluded clusters that were also discovered in the reads derived from the genomic background sequencing, that featured one of the repetitive alignments nearby, that had their insertion site in centromeric regions or that did not share similar repetitive alignments with other clusters and were thus probably false positives. 25. Mathieu, O., Probst, A. & Paszkowski, J. Distinct regulation of histone H3 methylation at lysines 27 and 9 by CpG methylation in Arabidopsis. EMBO J. 24, 2783–2791 (2005). 26. Chan, S. W. et al. RNAi, DRD1, and histone methylation actively target developmentally important non-CG DNA methylation in Arabidopsis. PLoS Genet. 2, e83 (2006). 27. Onodera, Y. et al. Plant nuclear RNA polymerase IV mediates siRNA and DNA methylation-dependent heterochromatin formation. Cell 120, 613–622 (2005). 28. Reinders, J. et al. Genome-wide, high-resolution DNA methylation profiling using bisulfite-mediated cytosine conversion. Genome Res. 18, 469–476 (2008). 29. Pall, G. S., Codony-Servat, C., Byrne, J., Ritchie, L. & Hamilton, A. Carbodiimidemediated cross-linking of RNA to nylon membranes improves the detection of siRNA, miRNA and piRNA by northern blot. Nucleic Acids Res. 35, e60 (2007). 30. Kanno, T. et al. A structural-maintenance-of-chromosomes hinge domaincontaining protein is required for RNA-directed DNA methylation. Nature Genet. 40, 670–675 (2008). 31. Ossowski, S. et al. Sequencing of natural strains of Arabidopsis thaliana with short reads. Genome Res. 18, 2024–2033 (2008). 32. The Arabidopsis Genome Initiative. Analysis of the genome sequence of the flowering plant Arabidopsis thaliana. Nature 408, 796–815 (2000).

©2009 Macmillan Publishers Limited. All rights reserved