Maintenance of DNA Methylation during the ArabidopsisLife Cycle Is Essential for Parental Imprinting W

Pauline E. Jullien,a Tetsu Kinoshita,b Nir Ohad,c and Frederic Bergera,1

a Chromatin and Reproduction Group, Temasek Lifesciences Laboratory, National University of Singapore, Singapore

117604, Republic of Singaporeb Integrated Genetics, National Institute of Genetics, Mishima 411-8540, Japanc Department of Plant Sciences, Faculty of Life Sciences, Tel Aviv University, Tel Aviv 69978, Israel

Imprinted genes are expressed predominantly from either their paternal or their maternal allele. To date, all imprinted genes

identified in plants are expressed in the endosperm. In Arabidopsis thaliana, maternal imprinting has been clearly

demonstrated for the Polycomb group gene MEDEA (MEA) and for FWA. Direct repeats upstream of FWA are subject to DNA

methylation. However, it is still not clear to what extent similar cis-acting elements may be part of a conserved molecular

mechanism controlling maternally imprinted genes. In this work, we show that the Polycomb group gene FERTILIZATION-

INDEPENDENT SEED2 (FIS2) is imprinted. Maintenance of FIS2 imprinting depends on DNA methylation, whereas loss of

DNA methylation does not affect MEA imprinting. DNA methylation targets a small region upstream of FIS2 distinct from the

target of DNA methylation associated with FWA. We show that FWA and FIS2 imprinting requires the maintenance of DNA

methylation throughout the plant life cycle, including male gametogenesis and endosperm development. Our data thus

demonstrate that parental genomic imprinting in plants depends on diverse cis-elements and mechanisms dependent or

independent of DNA methylation. We propose that imprinting has evolved under constraints linked to the evolution of plant

reproduction and not by the selection of a specific molecular mechanism.

INTRODUCTION

Reproduction of flowering plants is characterized by a double

fertilization event, which involves two identical sperm cells and

two female gametes, the egg cell and the central cell. Fertilization

of the egg cell leads to the development of the embryo, whereas

the second sperm cell independently fertilizes the central cell,

giving rise to the endosperm (Guignard, 1899; Maheshwari,

1950; Berger, 2003). The endosperm controls the supply of

maternal nutrients to the developing embryo and does not con-

tribute genetic material to the next generation. In Arabidopsis

thaliana and in maize (Zea mays), parental imprinted expression

has been detected only in the endosperm for a few genes

expressed only from their paternal or their maternal allele

(Berger, 2004; Gehring et al., 2004; Kohler and Grossniklaus,

2005). In Arabidopsis, imprinting was demonstrated using allele-

specific RT-PCR for the paternally expressed gene PHERES1

(Kohler et al., 2005) and the maternally expressed genes FWA

(Kinoshita et al., 2004) andMEDEA (MEA) (Kinoshita et al., 1999;

Vielle-Calzada et al., 1999). FWA andMEA are expressed during

neither vegetative development normale gametogenesis. During

female gametogenesis, FWA andMEA become transcriptionally

active, and both genes are expressed specifically in the central

cell (Choi et al., 2002; Kinoshita et al., 2004). After fertilization,

only the maternal alleles of FWA and MEA are expressed during

endosperm development, whereas their paternal alleles remain

silenced.

Imprinting of a gene thus results fromacombination of silencing

of the paternal allele and activation of expression of the maternal

allele. Silencing of the FWA paternal allele is likely correlated with

methylation of cytosine residues located in the direct repeats of

the 59 region mediated by the DNA methyltransferase MET1

(Kinoshita et al., 2004). The transcriptionally active FWAmaternal

allele in endosperm is characterized by demethylation of the

direct repeats. This demethylation may occur in the central cell

and might result from the activity of the DNA glycosylase DEME-

TER (DME). DME creates single-strand 59 nicks, and such DNA

breaks would be recognized by the DNA repair machinery, which

in turn replaces the methylated cytosine residue with unmethy-

lated ones (Choi et al., 2002; Gehring et al., 2006). Based on the

available knowledge regarding FWA imprinting, it was possible to

model a cycle for plant imprints (Berger, 2004; Kinoshita et al.,

2004). According to the model, FWA is silenced during the

vegetative phase by MET1. DME releases FWA silencing in the

central cell. After fertilization, the activated maternal FWA allele is

inherited by endosperm and is transcribed, whereas the paternal

allele remains silenced because it was inherited in an inactive

state from the male gamete. MET1 is active during male game-

togenesis (Saze et al., 2003), but its involvement in silencing of the

FWA paternal allele during male gametogenesis and endosperm

development has not beendemonstrated.Moreover, it is not clear

1 To whom correspondence should be addressed. E-mail fred@tll.org.sg; fax 65-68727007.The author responsible for distribution of materials integral to thefindings presented in this article in accordance with the policy describedin the Instructions for Authors (www.plantcell.org) is: Frederic Berger(fred@tll.org.sg).WOnline version contains Web-only data.Article, publication date, and citation information can be found atwww.plantcell.org/cgi/doi/10.1105/tpc.106.041178.

The Plant Cell, Vol. 18, 1360–1372, June 2006, www.plantcell.orgª 2006 American Society of Plant Biologists

whether the mode of regulation of FWA imprinting could apply to

other imprinted genes.

Regulation of MEA imprinting has been analyzed, but as

detailed below, the mechanism controlling MEA imprinting is

still unclear. DME is also essential for MEA transcriptional acti-

vation (Choi et al., 2002), and methylated CpG residues have

been identified in the MEA promoter (Xiao et al., 2003; Gehring

et al., 2006), suggesting thatMEA imprinting is controlled byDNA

methylation similar to FWA. DNA methylation could target puta-

tive CpG islands and transposons at the MEA locus (Xiao et al.,

2003; Spillane et al., 2004). However, such elements are not es-

sential for MEA imprinting, as suggested by conflicting genetic

experiments, which casts doubt on the role of MET1 in this

process (Vielle-Calzada et al., 1999; Luo et al., 2000; Vinkenoog

et al., 2000; Xiao et al., 2003). Although MET1 might play a

regulatory role in the control of imprinting of several genes

involved in endosperm development (Adams et al., 2000), its

direct involvement in the control ofMEA imprinting remains to be

tested.

Besides DNA methylation, histone methylation was shown

recently to be involved in the control ofMEA imprinting (Gehring

et al., 2006; Jullien et al., 2006) and likely also controls the

imprinting of PHERES1, encoding a MADS box transcription

factor (Kohler et al., 2005). MEA imprinting depends on tran-

scriptional repression of its paternal allele by methylation of the

Lys-27 of Histone3 by several Polycomb group complexes active

during the plant life cycle (Gehring et al., 2006; Jullien et al., 2006).

In endosperm, the PRC2 complex contains MEA (Grossniklaus

etal., 1998;Kiyosueetal., 1999; Luoetal., 1999) andacoreof three

additional proteins, the VEFS domain protein FERTILIZATION-

INDEPENDENT SEED2 (FIS2) (Luo et al., 1999), the WD40 pro-

tein FERTILIZATION-INDEPENDENT ENDOSPERM (FIE) (Luo

et al., 1999; Ohad et al., 1999), and the WD40 protein MULTI-

COPYSUPPRESSOROF IRA1 (MSI1) (Kohler et al., 2003; Guitton

et al., 2004). The mutants mea, fie, msi1, and fis2 exhibit similar

maternal defects in endosperm, and as a consequence, FIE,

MSI1, and FIS2 are anticipated to be subject to one or more of

the genetic mechanisms that regulate imprinting, including the

one presented for MEA (Chaudhury et al., 1997; Grossniklaus

et al., 1998; Kiyosue et al., 1999). Involvement of imprinting is

also supported for FIE and FIS2 by the monoallelic maternal

expression of transcriptional reporters for FIE–green fluorescent

protein (GFP) (Yadegari et al., 2000) and FIS2–b-glucuronidase

(GUS) (Luo et al., 2000), respectively. However, this transcrip-

tional control may affect only the corresponding transcriptional

reporter. Silencing in endosperm has been shown for the pater-

nal copy of reporter constructs inserted at several loci, leading to

the hypothesis of global silencing of the paternal genome in the

first days after fertilization (Vielle-Calzada et al., 2000). Alterna-

tively, the maternal effect associated with the loss-of-function

mutations could originate from abnormal development of the

female gametophyte or fromdosage effects.Hence, the imprinted

status of FIS2, FIE, and MSI1 remains unclear.

In this work, we show that FIS2 is subject to parental genomic

imprinting. We study in detail the role of DNA methylation in the

transcriptional control of FIS2 and MEA and conclude that they

are distinct. We identify a cis-element upstream of FIS2 that is

targeted by MET1. This element is likely associated with the

control of FIS2 imprinting. Using a genetic approach, we inves-

tigated the role played byMET1 during the entire life cycle for the

control of FIS2 and FWA imprinting. To this end, we investigated

the effect of MET1 loss of function restricted to the vegetative

developmental phase (with the MET1as line) and the effect of

MET1 loss of function restricted to the gametogenesis phase

(with the met1-3/þ line). Based on our results, we propose a

revised model for the cycle of MET1-dependent imprinting in

plants.

RESULTS

FIS2 Expression Is Controlled by DNA

Methylation–Dependent Imprinting

To investigate the imprinted status of FIS2, we used polymor-

phisms in the FIS2 coding sequence between different wild-type

accessions to distinguish transcripts of each parental allele

(allele-specific RT-PCR) (Figure 1A). We performed allele-

specific RT-PCR on seeds originating from crosses between

wild-type accessions C24 and Columbia (Col), used alternately

asmale or female. Onewould expect that if FIS2 is not imprinted,

two transcripts should be detected. Contrary to this hypothesis,

we observed only expression ofmaternal FIS2 transcripts (Figure

1A). We observed monoallelic maternal expression of FIS2 up to

5 d after pollination (DAP) (see Supplemental Figure 1 online).

Maternal FIS2 transcripts stored in the ovules and inherited after

fertilization could be detected in developing seeds and mimic an

imprinted status. However, it is unlikely that such storedmaternal

transcripts could be still present at 5 DAP. It is likely that we

detected transcripts from active postzygotic transcription of the

FIS2 maternal allele in the endosperm. A putative general pater-

nal silencing has been observed for reporters of endosperm-

expressed genes for up to 4 DAP (Vielle-Calzada et al., 2000). It

could be hypothesized that such a general mechanism also

affects expression at genomic loci for genes expressed in

endosperm. During all stages of development examined, we

observed biparental expression of MINI3 (Figure 1A), a gene

expressed only in endosperm during the first 3 DAP (Luo et al.,

2005). These results and others (Berger, 2003; Kinoshita et al.,

2004) indicate that there is no general paternal silencing for

genes expressed in endosperm and that FIS2 monoallelic ex-

pression results from a specific silencing mechanism. In agree-

ment with these findings, Luo et al. (2000) previously showed that

the FIS2-GUS translational reporter is expressed in endosperm

only from thematernal allele.We concluded that the endogenous

FIS2 gene is imprinted.

We investigated the potential role played by MET1 in FIS2

imprinting. To this end, we used the loss-of-function allelemet1-3

in the Col background (Figure 1B). To avoid nonspecific cumu-

lative effects of inbred met1 loss-of-function lines (Saze et al.,

2003; Xiao et al., 2006), we isolatedmet1-3/met1-3 plants from a

segregating population of selfedmet1-3/þ plants. Loss of MET1

activity caused expression of the paternal allele of FIS2 in

crosses between wild-type female ovules and met1-3 pollen

(Figure 1B). To evaluate the impact of MET1 activity during

vegetative development, we used MET1as plants, which pro-

duce antisense MET1 mRNAs in vegetative tissues, leading to a

DNA Methylation and Imprinting 1361

loss of 70% of DNA methylation on symmetric CpG sites

(Finnegan et al., 1996). We observed that loss of MET1 activity

during vegetative development caused expression of the pater-

nal allele of FIS2 in crosses betweenwild-type female ovules and

MET1as pollen (Figure 1C). We concluded that imprinting of FIS2

depends directly or indirectly on MET1. DNA methylation is

propagated through replication in a semiconservative manner

(Finnegan et al., 2000). Because in the MET1as line MET1

antisense RNAs are produced only during vegetative develop-

ment, we propose that demethylation of FIS2 during vegetative

development is carried over to the endosperm, resulting in loss of

imprinting. DNA methylation likely interacts with mechanisms

that control other epigeneticmarks (Tariq et al., 2003; Chan et al.,

2005; Takeda and Paszkowski, 2006; Vire et al., 2006), and the

effect of met1-3 and MET1as could be indirect. Thus, we tested

whether other epigenetic mechanisms might be involved, in-

cluding DNAmethylation and histonemethylation. The DOMAIN-

REARRANGED METHYLTRANSFERASEs DRM1 and DRM2

control de novo methylation (Cao and Jacobsen, 2002a), and

CHROMOMETHYLASE3 (CMT3), which is unique to plants,

maintains methylation at asymmetric sites (Lindroth et al.,

2001; Cao and Jacobsen, 2002b). Loss-of-function mutants in

these genes do not influence FIS2 imprinting (Figure 1D). Tran-

scriptional gene silencing is also regulated by methylation of

histone H3. In Arabidopsis, Lys residues Lys-9 and Lys-27 of

histoneH3aremethylatedbyKRYPTONITE (KYP) (Jackson et al.,

2002) and by Polycomb group complexes (Bastow et al., 2004;

Jullien et al., 2006), respectively. FIS2 imprinting was not altered

in crosses between wild-type ovules and pollen from mutant

plants carrying loss-of-function kyp or from FIE cosuppressed

plants (Katz et al., 2004) (Figure 1D). Because we could not find

any effect of loss of epigenetic marks likely linked with the

maintenance of DNA methylation, we conclude that MET1 di-

rectly silences FIS2.

DNA Methylation Targets a Specific CpG-Rich Domain

Upstream of the FIS2 Gene

Loss of MET1 activity caused ectopic expression of FWA, as

reported previously (Saze et al., 2003; Kinoshita et al., 2004), but

surprisingly, it did not result in ectopic expression of FIS2 in

vegetative tissues (Figure 2A). The absence of endosperm-

specific transcriptional activators could explain the absence of

ectopic expression of FIS2 in vegetative met1 tissues. Never-

theless, we hypothesized that loss of function of MET1 causes a

loss of DNA methylation marks in FIS2 genomic sequences. To

test this hypothesis, we analyzed DNA methylation in the 59 re-

gion of FIS2 in leaf tissues. Using the methylation-sensitive

restriction enzyme McrBC, we identified a region affected by

DNA methylation (Figure 2B). Moreover, bisulfite sequencing

analyses identified a region at;2 kb upstream of the start codon

of FIS2 consisting of a 200-bp domain where most CpG sites

Figure 1. FIS2 Imprinting Is Controlled by DNA Methylation.

(A) Allele-specific RT-PCR on RNAs extracted from siliques obtained at

0.5, 1.5, and 3 DAP after crosses were made between wild-type Col and

wild-type C24. A 180-bp deletion in the FIS2 coding sequence of the

wild-type accession Col allows the distinction of the Col FIS2 transcript

from the FIS2 transcript in accession C24. Only the FIS2maternal allele is

expressed in developing seeds. By contrast, MINI3 is expressed from

both parental alleles.

(B) Allele-specific RT-PCR on RNAs extracted from siliques obtained at

4 DAP after crosses between wild-type ovules and pollen from met1-3/

met1-3 plants (Col) show reduced activity of the maintenance DNAMET1

during the entire life cycle. In contrast with wild-type crosses, the FIS2

paternal allele is expressed in developing seeds, which inherit a paternal

allele from met1-3/met1-3 plants.

(C) Allele-specific RT-PCR on RNAs extracted from siliques obtained at

5 DAP after reciprocal crosses between wild-type plants and MET1as

plants. The FIS2 paternal allele is expressed in developing seeds when

pollen is provided byMET1as plants, which show reduced activity of the

maintenance DNA MET1 during the vegetative developmental phase

(Finnegan et al., 1996).

(D) Allele-specific RT-PCR on RNAs extracted from siliques obtained at

5 DAP after crosses made between wild-type plants and different mutants

affected for histone methylation or for DNA methylation. A 180-bp

deletion in the FIS2 coding sequence of the wild-type accession Col

allows the distinction of the Col FIS2 transcript from the FIS2 transcript in

accessions C24, Landsberg erecta (Ler), and Wassilewskija. C24 is

presented as a control. Only the maternal allele of FIS2 is expressed in

seeds resulting from reciprocal crosses between the wild type and

homozygous mutants for other pathways controlling DNA methylation in

Arabidopsis cmt3 (Lindroth et al., 2001) and drm1 and drm2 (Cao and

Jacobsen, 2002a). KYP methylates Histone3 Lys-9 residues (Jackson

et al., 2002; Tariq et al., 2003), whereas the Polycomb group complex

containing FIEmethylates Histone3 Lys-27 residues (Bastow et al., 2004).

The paternal (p) and maternal (m) alleles are indicated. Glyceraldehyde-

3-phosphate dehydrogenase (GAPDH) was used as a loading control.

No-RT control assays were performed, but the data are not shown.

1362 The Plant Cell

weremethylated (30 of 38) (Figure 2C; see Supplemental Figure 2

online). Methylation of CpG residues in this region was severely

reduced inmet1-3 homozygous mutant leaves (see Supplemen-

tal Figure 2 online). We conclude that MET1 targets a specific

element in the 59 region thatmight control FIS2 silencing in leaves

as well as in pollen and endosperm.

Activation of the FIS2 Maternal Allele by DME in the

Central Cell

To investigate the role of DME on FIS2 expression, we tested the

effect of dme loss of function on FIS2 expression. To this end,

RT-PCRexperimentswere performedwithmRNAextracted from

buds and flowers, where DMEwas shown to be expressed (Choi

et al., 2002). We used FWA as a control for the action of the dme-4

allele, which was isolated independently from other dme alleles

(Guitton et al., 2004). FWA is expressed in the wild type before

and after fertilization (Kinoshita et al., 2004), whereas in dme-4, a

strong reduction of FWA transcription was observed (Figure 3A),

as shown previously with dme-1 (Kinoshita et al., 2004). Similarly,

dme-4 caused a reduction of FIS2 expression (Figure 3A), but the

level of FIS2 expression was only partially reduced. The limited

reduction of FIS2 expression by dme-4 was confirmed by anal-

ysis of the expression of the FIS2-GUS reporter, which includes

most of the FIS2 locus and is only maternally expressed in wild-

type endosperm (Figure 2C) (Luo et al., 2000). In self-pollinated

FIS2-GUS/FIS2-GUS, dme-4/DME plants, all ovules inherit FIS2-

GUS and 50% inherit the dme-4 mutation. Accordingly, we

observed a significant reduction of FIS2-GUS expression in 50%

of the ovules inwhichdme-4wasexpected to be present (Figures

3B to 3D) compared with wild-type ovules (Figures 3B and 3E).

However, half of dme-4 ovules still expressed the reporter

at lower levels than wild-type ovules (Figures 3B and 3D). After

fertilization, we observed a similar reduction of FIS2-GUS ex-

pression in endosperm, which inherits a maternal dme-4 (Figure

3B). Homozygous dme-4/dme-4 mutants are viable (Guitton

et al., 2004), whereas dme-2/dme-2 causes seed developmental

arrest (Choi et al., 2002). Hence, partial reduction of FIS2

expression in the dme-4 background could result from incom-

plete penetrance of the dme-4 mutation. However, a similar

partial reduction of FIS2-GUS expression was obtained with

the strong allele dme-2 (Choi et al., 2002) (Figure 3B). We

conclude that DME is required for the expression of FIS2 in

the central cell as well as in the endosperm. Persisting FIS2

expression in a fraction of dme ovules could be accounted for by

either redundant functions for DME or by additional activation

Figure 2. MET1 Targets a CpG-Enriched Region at the FIS2 Locus.

(A) RT-PCR analysis shows that imprinted genes FWA and FIS2 are not expressed in vegetative tissues of the wild-type accession Col. Reduced

methylation of DNA in nullmet1-3 homozygous mutant plants causes ectopic expression of FWA but not of FIS2 in leaves.met1-3/þ plants, which show

no ectopic expression of FWA, were used in the experiments presented in Figure 5A. GAPDH was used as a loading control.

(B) Analysis of DNA methylation using the methylation-sensitive restriction enzyme McrBC in wild-type Col leaves identifies a methylated region

spanning �3104 to �1867 in the 59 region of FIS2. The sensitivity to McrBC is decreased in the met1/met1 background, which indicates that MET1

controls DNA methylation in the �3104 to �1867 region. FWA repeats, which are methylated by MET1, were used as controls (Kinoshita et al., 2004).

(C) Bisulfite sequencing of the FIS2 59 region, which contains a 200-bp domain enriched in methylated cytosine residues predominantly on CpGs (see

Supplemental Figure 2 online for the full sequence). Analysis was done with genomic DNA from rosette leaves. The structure of the FIS2 genomic

sequence is depicted, with the start codon indicated by 0. Exons are indicated by black boxes, and the 59 and 39 untranslated regions are indicated by

white boxes. The extent of the partial translational fusion FIS2-GUS is indicated below the schematic genomic structure.

DNA Methylation and Imprinting 1363

mechanisms sufficient to overcome the repressive effect of DNA

methylation.

Imprinting of FWA and FIS2 in Endosperm Requires

the Maintenance of Silencing by MET1 during

Male Gametogenesis

In contrast with female gametogenesis, male gametogenesis

does not promote the expression of FWA, FIS2, and MEA (Luo

et al., 2000; Kinoshita et al., 2004). We have investigated how

silencing of FWA and FIS2 is maintained during male gameto-

genesis. Expression of MET1 has been detected in pollen, by

transcriptome analysis (Honys and Twell, 2004), and there is

evidence for MET1 function in the silencing of exogenous tran-

scriptional reporters during male gametogenesis (Saze et al.,

2003). Thus, we tested whether the loss of MET1 function

affected the expression of FWA and FIS2 paternal alleles in

pollen and later in endosperm. To this end, we used heterozy-

gousmet1-3/þ plants, which exhibit reduced MET1 activity only

during gametogenesis. In met1-3/þ plants, the mutant met1

allele is inherited by half of the microspores, which develop in

pollen grains after two cell divisions (McCormick, 2004). The

generative cell is expected to inherit a hemimethylated copy and

divides further to produce two male gametes, in which the

genomic DNA is demethylated (Figure 4A). Accordingly, we

observed ectopic expression of the transcriptional reporter

FWA-GFP in segregating pollen from met1-3/þ, FWA-GFP/

FWA-GFP plants (Figure 4B) (47.4% of pollen bearing FWA-GFP

fluorescence, SD ¼ 6.7, n ¼ 1372). Expression of FWA-GFP was

observed in the sperm cells, which indicates that met1 pollen

grains are able to provide a transcriptionally active FWA-GFP

copy (Figure 4B, inset).

Figure 3. Maternal Control of FIS2 Expression by DME.

(A) RT-PCR analyses performed on mRNA from buds (stages 11 and 12) and flowers (stage 13) (Smyth et al., 1990) show a lack of FWA expression and

reduced FIS2 expression in the dme-4 homozygous mutant compared with the wild-type C24 accession. GAPDH was used as a control.

(B) to (E) Effects of dme on FIS2-GUS expression.

(B) Percentage of ovules expressing the FIS2-GUS reporter construct in dme-2/þ and dme-4/þ before pollination (BP) and at 2 DAP. Three different

classes are observed: no expression (white), little expression (light blue), and wild-type expression (dark blue). Error bars correspond to the SD

associated with each set of measurements.

(C) to (E) Micrographs showing the three categories of FIS2-GUS expression in the dme-4/þ mutant background: no expression (white; [C]), little

expression (light blue; [D]), and wild-type expression (dark blue; [E]). Bars ¼ 20 mm.

1364 The Plant Cell

Our cytological observations were supported by RT-PCR

experiments showing ectopic expression of FWA in mRNA pop-

ulations extracted from stamens ofmet1-3/þ plants (Figure 4C).

Thus, we usedmet1-3/þ plants to test whether loss of function of

MET1 during male gametogenesis was sufficient to allow the

expression of FWA paternal alleles in endosperm. In seeds re-

sulting from crosses between wild-type ovules and met1-3/þpollen, FWA transcripts were detected from both the paternal

andmaternal alleles (Figure 5A; note that polymorphism for Col is

represented by a double band). Accordingly, the paternal copy of

FWA-GFP is expressed in endosperm of seeds produced by

crosses between wild-type ovules and pollen from met1-3/þ,

FWA-GFP/FWA-GFP plants (Figure 5C) (46.7%, SD ¼ 5.2, n ¼358). By contrast, no expression was observed when FWA-GFP

was provided from a paternal wild-type background (Figure 5B;

n ¼ 126) (Kinoshita et al., 2004). We conclude that during

gametogenesis, MET1 maintains FWA silencing and that FWA

imprinting in endosperm requires MET1 activity during male

gametogenesis. The MEA paternal allele is imprinted in endo-

sperm (Kinoshita et al., 1999; Vielle-Calzada et al., 1999) and is

predominantly expressed from its maternal allele during the first

days after pollination (Jullien et al., 2006). We tested the effect of

MET1 on MEA paternal allele expression in endosperm. MEA

paternal allele expression was examined in seeds resulting from

crosses between wild-type ovules andmet1-3/þ pollen. In such

crosses, we detected a predominance of maternal MEA allele

expression, which is similar to the result obtained when crosses

were made with wild-type male accessions. We concluded that

MEA imprinting was not affected by loss of DNA methylation

during male gametogenesis (Figure 5D). A similar result was

recently obtained independently (Gehring et al., 2006). Thus,

MET1 does not appear to play a major role in MEA silencing

comparable to its role in FWA and FIS2 imprinting.

Similarly, we tested the impact ofmet1 loss of function during

male gametogenesis on FIS2 silencing in pollen and imprinting in

endosperm. FIS2 was ectopically expressed in stamens from

met1-3/þ plants (Figure 4C). Pollination of wild-type ovules with

met1-3/þ plants caused expression of the FIS2 paternal allele in

endosperm (Figure 5E), which indicates that FIS2 silencing by

MET1 duringmale gametogenesis is required for FIS2 imprinting.

Previously, it was shown that maternal inheritance of fis2 loss-of-

function alleles causes major defects in endosperm develop-

ment, including overexpression of the KS117 marker (Sørensen

et al., 2001). In addition, endosperm produced by fis2 mothers

fails to cellularize during the syncytial stage (Chaudhury et al.,

1997). In crosses between fis2 ovules andmet1 pollen, the wild-

type FIS2 paternal allele is activated bymet1 and is expected to

rescue the maternal loss-of-function fis2 allele. We next tested

whether in met1-3/þ plants the demethylated expressed pater-

nal allele of FIS2 is sufficient to rescue the inactive mutant fis2

allele provided maternally to the developing endosperm. Heter-

ozygousmet1-3/þ plants produce 50%met1 pollen. The KS117

expression pattern was rescued in 49% of seeds produced by

crosses with pollen from met1-3/þ plants and fis2/fis2, KS117/

KS117 plants used as females (SD ¼ 5.3, n ¼ 459) (Figure 5G). In

these crosses, 19% of seeds showed cellular endosperm similar

to wild-type seeds (SD ¼ 5.1, n ¼ 365) (Figure 5J). Hence, met1

loss of function during pollen development is sufficient to remove

Figure 4. Maintenance of FIS2 and FWA Silencing during Male Gametogenesis.

(A) During pollen development in met1-3/þ plants, half of haploid microspores produced by meiosis inherit the met1-3 mutation, and the other half

inherit the wild-type MET1 allele. All microspores inherit a methylated copy of the genome (gray rectangles). DNA replication in met1-3 microspores

produces hemimethylated DNA from the methylated template (gray/white rectangles). Hemimethylated DNA is inherited after mitosis by the vegetative

nucleus (V) and the nucleus of the generative cell (G), which divides a second time and produces the two sperm cells (S). After this division, one sperm

cell inherits a fully demethylated DNA copy (white rectangles) and the other sperm cell inherits hemimethylated DNA. The vegetative nucleus does not

participate in fertilization. During sperm cell maturation, DNA replicates (Durbarry et al., 2005), and at fertilization, all sperm frommet1-3microspores is

expected to contain at least one hemimethylated copy.

(B) Segregating population of pollen grains from FWA-GFP/FWA-GFP,met1-3/þ plants (confocal section). The inset represents a confocal section of a

pollen grain expressing FWA-GFP in the vegetative cell (V) and in the two sperm cells (s). Bar ¼ 5 mm.

(C) Ectopic expression of FWA and FIS2 observed by RT-PCR in RNA purified from stamens.GAPDHwas used as a control to assess equal amounts of

RNA loaded. DUO1 is expressed specifically in pollen (Rotman et al., 2005) and was used as a quality control for pollen mRNA extraction.

DNA Methylation and Imprinting 1365

the repressive epigenetic marks from the FIS2 paternal allele,

which remains active in endosperm.

We conclude that MET1 activity is essential to maintain the

silencing of FIS2 and FWA during male gametogenesis. If MET1

activity is absent during pollen development, the sperm cells

deliver a transcriptionally active paternal allele to endosperm and

imprinting of FIS2 and FWA in endosperm is compromised.

Regulation of MEA imprinting appears to depend on a distinct

regulatory mechanism independent of MET1. Instead, recent

evidence supports the notion thatMEA silencing relies on histone

methylation by Polycomb group complexes (Gehring et al., 2006;

Jullien et al., 2006).

MET1 Is Required for Silencing the FWA

Paternal Allele in Endosperm

We have shown that FIS2 and FWA imprinted status in endo-

sperm requires MET1 activity before fertilization. However, the

requirement of MET1 after fertilization and during endosperm

development has not yet been reported. Although the paternal

FWA-GFP copy provided bymet1 pollen becomes expressed in

endosperm, its expression gradually decreases in signal intensity

Figure 5. Imprinting of FWA and FIS2 in Endosperm Relies on the

Maintenance of Their Silencing during Male Gametogenesis.

The imprinting status of FWA,MEA, and FIS2was analyzed in developing

seeds resulting from crosses between wild-type ovules and pollen from

met1-3/þ plants.

(A) Allele-specific RT-PCR detecting expression of the paternal and

maternal alleles of FWA in seeds resulting from crosses between wild-

type ovules (Ler accession) and pollen frommet1-3/þ plants. (For the Col

accession, two bands were obtained because digestion of the restriction

polymorphism was partial; the lowest band was Col-specific [3 DAP].)

We used met1-3/þ plants, which did not show expression of FWA in

vegetative tissues (see Figure 2A). Although crosses between wild-type

Ler and Col showed only expression of the maternal FWA allele, the FWA

paternal allele was expressed in crosses with met1-3/þ pollen.

(B) and (C) Accordingly, at 2 DAP, the FWA-GFP reporter was expressed

paternally in endosperm when provided by pollen from met1-3/þ plants

(C) but not from wild-type plants (B).

(D) Allele-specific RT-PCR performed on crosses between wild-type

accessions RLD and Col detected expression of the maternalMEA allele,

typical of MEA imprinting (1 DAP). MEA imprinting was not altered if

met1-3/þ plants provided the MEA paternal allele.

(E) Crosses between wild-type accessions C24 and Col showed ex-

pression only from the maternal FIS2 allele typical of FIS2 imprinted

status in wild-type endosperm. By contrast, allele-specific RT-PCR

showed expression of the paternal and maternal alleles of FIS2 in seeds

resulting from crosses between ovules from wild-type C24 and pollen

from met1-3/þ plants (Col accession; 3 DAP).

(F) to (J) Rescue of endosperm development upon removal of the FIS2

paternal allele imprint in met1-3 pollen.

(F) and (H) GFP fluorescence of the enhancer trap marker KS117 was

increased uniformly in endosperm of the fis2 homozygous mutant (F)

compared with the wild type (H), in which it was restricted to the

endosperm posterior pole at 5 DAP.

(G) After pollination of fis2/fis2, KS117/KS117 plants by pollen from

met1-3/þ plants, half of the seeds showed a pattern of KS117 expres-

sion similar to the KS117 pattern in wild-type endosperm (asterisks).

Such seeds also showed a rescue of cytological alteration in fis2

endosperm ([I] and [J]).

(I) The phenotype conferred by fis2 is characterized by the absence of

cellularization of the syncytial endosperm and the enlargement of the

posterior pole opposite the embryo (Guitton et al., 2004).

(J) Rescue of such phenotypic traits was obtained in 19% of seeds

resulting from crosses between fis2/fis2 plants and met1-3/þ pollen.

Bars ¼ 50 mm ([B], [C], [I], and [J]) and 200 mm ([F] to [H]). GAPDH was

used as a control for the RT-PCR analyses ([A], [D], and [E]).

1366 The Plant Cell

between 2 and 4 DAP (Figures 6A to 6C) and becomes gradually

undetectable after 6 DAP (data not shown). We hypothesized

that this resulted from progressive silencing of the paternal FWA-

GFP copy by maternally provided MET1 activity in met1/þendosperm. According to this hypothesis, a loss of function of

MET1 in met1/met1 endosperm is expected to sustain the

expression of a paternally provided copy of FWA-GFP. Trans-

mission of met1-3 through male and female gametes follows

Mendelian proportions (see Supplemental Table 1 online), and

we expected production of 25% met1/met1 seeds in crosses

betweenmet1/þ plants. We performed crosses between ovules

of met1-3/þ plants and pollen from met1/þ, FWA-GFP/FWA-

GFP plants, leading to 25% met1/met1 seeds and 25% met1/þseeds with a paternal FWA-GFP copy from met1 pollen. We

observed two distinct classes of seeds expressing FWA-GFP

either in met1-3/þ or in met1/met1 endosperm. One-quarter of

seeds showed low expression levels comparable to those of

seeds withmet1/þ endosperm (Figures 6D and 6E), and another

one-quarter of seeds showed high levels of FWA-GFP expres-

sion at 4 DAP (Figures 6D and 6F). Given the Mendelian propor-

tion of each seed class, the latter class presumably corresponds

to met1/met1 endosperm with no MET1 activity, in which pro-

gressive silencing of a demethylated FWA-GFP paternal copy

does not occur. These results suggest that MET1 is continuously

required for silencing of the paternal FWA allele during endosperm

development to maintain FWA imprinted status.

DISCUSSION

Imprinting of FIS2 and FWA Relies on MET1 Activity during

Male Gametogenesis

We have shown that FIS2 undergoes monoallelic maternal

expression in endosperm. A gene is considered to be imprinted

Figure 6. Maintenance of FWA Imprinting in Endosperm Depends on MET1.

(A) At 2 DAP, after fertilization of wild-type ovules with pollen frommet1-3/þ, FWA-GFP/FWA-GFP plants, expression of the paternally provided allele of

FWA-GFP is observed in endosperm in half of the seeds.

(B) and (C) The intensity of FWA-GFP expression decreases during endosperm development from 2 DAP (B) to 4 DAP (C).

(D) In contrast with the gradual silencing of FWA-GFP paternal expression in met1-3/þ endosperm, one-quarter of the seeds show expression of the

FWA-GFP paternal allele in crosses between ovules from met1-3/þ plants and pollen carrying met1-3/þ and FWA-GFP/FWA-GFP.

(E) and (F) One-quarter of the seeds express low levels of FWA-GFP (E) and are presumed to be heterozygous for met1-3. Another one-quarter of the

seeds show higher sustained expression of FWA-GFP (F) and are presumably homozygous for met1-3 in endosperm.

Bars ¼ 80 mm ([B], [C], [E], and [F]). The number of seeds observed for each cross is indicated above each bar. SD is indicated for each measurement.

HO, homozygous.

DNA Methylation and Imprinting 1367

when it is expressed after fertilization and primarily fromone of its

two parental alleles (Vielle-Calzada et al., 1999). We detected

FIS2 maternal transcripts up to 5 d after pollination and showed

that the FIS2 paternal allele can be expressed in endosperm

duringmet1 loss of function in gametogenesis. Thus, FIS2 is ac-

tively transcribed after fertilization and is expressed only from its

maternal allele. We conclude that FIS2 is a maternally expressed

imprinted gene. MET1 activity is necessary during endosperm

development for sustained silencing of the paternal copy of FWA.

We anticipate a similar dependence of the FIS2 paternal allele

silencing on MET1 activity in endosperm. In conclusion, when

MET1 activity is removed, as in met1-3 mutants, the expression

of MET1-dependent imprinted genes is maintained beyond the

wild-type schedule for both parental alleles.

Continuous DNA methylation by MET1 at specific sites during

vegetative development maintains the silencing of both parental

FWA and FIS2 alleles. The imprinted genes are expressed during

female gamete maturation, whereas they remain silent during

male gametogenesis. This imbalance of gene expression in

gametes prefigures imprinting in endosperm after fertilization.

In the central cell, silencing of imprinted genes expressed in

endosperm would be removed via DNA demethylation mediated

by the glycosylase DME. This was shown for FWA (Kinoshita

et al., 2004), and our data indicate that DME also acts on

FIS2 together with other unknown activators, because FIS2

expression is only partially suppressed in dme ovules. We have

shown that MET1 is required during male gametogenesis

to maintain FWA and FIS2 silencing. This provides an example

for the role played by MET1 during male gametogenesis, which

was originally demonstrated with an exogenous reporter con-

struct (Saze et al., 2003). Furthermore, we have shown that

sperm cells deficient for met1 provide activated FIS2 paternal

alleles, which are able to rescue endosperm defects caused

by the maternally provided fis2 null alleles. In conclusion, im-

printing results from the maintenance of DNA methylation

in sperm cells, whereas DNA methylation is removed in the

female gamete (Figure 7). Our model thus extends and comple-

ments the model proposed previously (Berger, 2004; Kinoshita

et al., 2004).

Similarly, in mammals, loss of function of de novo DNA

methylation during gametogenesis caused a lack of imprints

(epigenetic marks responsible for silencing) establishment in

mice (Monk, 1990; Kaneda et al., 2004). Thus, during gameto-

genesis, crucial steps are made necessary for imprinting genes

in both mammals and plants, but the establishment of imprints

occurs through distinct DNA methylation mechanisms. In con-

trast with plants, extensive demethylation occurs in the mam-

malian germ line and erases parental imprints (Delaval and Feil,

2004). Imprints are reestablished in the male and female germ

lines by a specific de novoDNAmethyltransferase (Kaneda et al.,

2004) and not by the maintenance DNA methyltransferase, as in

flowering plants.

Figure 7. Model for the Dual Control of Parental Genomic Imprinting by DNA Methylation in Plants.

The genes FIS2 and FWA are imprinted in endosperm and silenced by the continuous action of MET1, which targets methylation (pink circles) to an

element in the promoter specific for each gene. MET1-dependent silencing of FIS2 and FWA is required during the vegetative phase, male

gametogenesis, and endosperm development, when it maintains silencing of the paternal (p) allele. FIS2 and FWA expression is activated during female

gametogenesis by the DNA glycosylase DME, leading to the expression of their maternal (m) allele in endosperm after fertilization. In endosperm, the

paternal copy remains silenced through the continuous action of MET1, whereas the maternal copy is active, leading to an imprinted expression.

1368 The Plant Cell

MET1 Controls Imprinting in Arabidopsis via

Diverse cis-Elements

In spite of the similarities between the expression patterns of

maternally imprinted genes identified in Arabidopsis, we show

that imprinting mechanisms of different genes are distinct. Our

results demonstrate that MET1 plays a major role in FIS2 and

FWA imprinting, but some aspects of MEA imprinting do not

appear to depend critically on MET1. Silencing of the MEA

paternal allele is not mediated by MET1 but rather depends on

the continuous action of histone methylation mediated by Poly-

comb group complexes (Gehring et al., 2006; Jullien et al., 2006).

Nevertheless, activation of the maternal allele of MEA depends

on the removal of MET1-dependent DNA methylation (Gehring

et al., 2006). It is not clear, however, whether DME also leads to

the removal of Polycomb group–dependent histone methylation

marks. The regulation of MEA imprinting also could involve

DDM1, a SWI2/SNF2 factor, which interferes with MEA function

(Vielle-Calzada et al., 1999) and interacts with DNA methyl

binding proteins (Zemach et al., 2005).

MET1 was shown to target tandem repeats in the promoter of

FWA and to regulate its imprinting (Kinoshita et al., 2004), al-

though a direct connection between these two events remains to

be demonstrated. Additional silencing mechanisms are involved

in the silencing of FWA (Chan et al., 2004), but it is not clear

whether they control FWA imprinting. Upstream of the FIS2

coding sequence, we could detect neither repetitive sequence

nor transposonswith bioinformatics analysis, which successfully

detected such elements in the MEA promoter (Spillane et al.,

2004). Rather, we show that methylation is directed to a 200-bp

region upstream of the FIS2 coding sequence. A putative MET1-

targeted control element has been detected upstream of one

of the maize homologues of FIE, which is also imprinted and

expressed in endosperm (Danilevskaya et al., 2003). This ele-

ment consists of CpG-enriched sequences different from the

cis-elements identified in Arabidopsis for FIS2. Based on the

limited data available, we are able to conclude that imprinting in

flowering plants relies on several types of cis-elements and

several silencing mechanisms. It is also becoming clear that

imprinting in mammals does not rely on a single type of epige-

netic transcriptional regulation. Both maintenance and de novo

DNA methyltransferase are involved, as well as Polycomb group

activities.

DNA methylation targets several types of elements in imprint-

ing control regions, which can be related to altered transposable

elements and CpG-enriched regions (Constancia et al., 2004;

Delaval and Feil, 2004). The diversity of cis control elements and

trans-acting mechanisms involved in imprinting found in flower-

ing plants and mammals contrasts with the relative similarity of

functions for some targets of imprinting. Most imprinted genes

identified in flowering plants play an essential role in endosperm

development and thus indirectly control embryo nutrition in the

developing seed. In animals, several imprinted genes were found

to have an essential function in reproduction and to affect

interactions between the mother and the developing embryo or

infant (Constancia et al., 2004). It is thus likely that the evolution of

imprinting did not depend on selection of a molecular epigenetic

mechanism but more likely stemmed from biological constraints

linked to reproductive innovations. Selective mechanisms might

have used available silencingmachineries and targeted potential

cis-elements already present at the loci of some genes important

for reproduction. This evolutionary process eventually led to the

actual state of diverse cis and trans control elements associated

with imprinting.

METHODS

Plant Materials and Growth Conditions

The MET1 antisense line of Arabidopsis thaliana was provided by J.

Finnegan (Finnegan et al., 1996). This line is characterized by a loss of

70% CpGmethylation and by a dominant effect of the MET1 antisense in

vegetative diploid tissues. As the line is maintained by selfing, it likely

accumulates epimutations of unknown nature, which may interfere with

our experiments. Hence, for most experiments, we used themet1-3 (Col)

line provided by J. Paszkowski (Saze et al., 2003) and took care to

propagate the met1-3 line as a heterozygote. Homozygous mutants

cmt3, drm1, drm2, and kyp (Ler) came from S.E. Jacobsen’s laboratory

(Lindroth et al., 2001; Cao and Jacobsen, 2002a; Jackson et al., 2002).

Themutantdme-2 (Col) was provided byR. Fischer (Choi et al., 2002). The

mutants dme-4 and fis2-6 (C24) were characterized previously in the

laboratory of F.B. (Guitton et al., 2004). FIEcs (Col) plants were charac-

terized previously by N.O.’s group (Katz et al., 2004). We used met1-3/þplants in which neither FIS2 nor FWA transcripts could be detected in

vegetative tissues (Figure 2A), which indicated that the parent plants were

not affected by the mutation during the vegetative phase but only during

the production of male gametes.

The KS117 line was identified after a screen in Jim Haseloff’s enhancer

trap line collection (Haseloff, 1999). The transgenic reporter line FIS2-

GUS was kindly provided by A. Chaudhury (Chaudhury et al., 1997; Luo

et al., 2000). The FWA-GFP fusion was described previously (Kinoshita

et al., 2004).

After 3 d at 48C in the dark, seeds were germinated and grown on soil.

Plants were cultured in a growth chamber under short days (8 h of light at

208C/16 h of dark at 168C; 60 to 70% RH) until rosettes were formed.

Plants were transferred to long days at 208C (14 h of light/10 h of dark) to

induce flowering and grown until seeds were harvested. Crosses were

performed by emasculating flowers and pollinating them manually. Si-

liques were harvested at various times after pollination, as indicated in

figure legends and in the text.

RT-PCR Analyses

Sample tissues were collected from Arabidopsis plants and immediately

frozen in liquid nitrogen. Tissues were ground, and total RNA was

prepared using the RNeasy mini kit (Qiagen). DNase treatment was

done on 2 mg of total RNA using the DNase free kit (Ambion).

For reverse transcription, 500 ng of total RNA was incubated for 1 h at

428C with 200 units of RevertAid Moloney murine leukemia virus reverse

transcriptase (Fermentas) in a 20-mL reaction mixture containing 4 mM

oligo(dT) 12 primer, appropriate reaction buffer, 1 mM deoxynucleotide

triphosphate, and 40 units of recombinant RNasin ribonuclease inhibitor

(Promega). The reaction was stopped by incubation at 708C for 10 min.

Equal amounts of RT products were used to perform subsequent

PCRs. Primers used to amplify FIS2were Fis2-R5018 (59-CCTGCATTGTT-

TGGGAGTGATAGAA-39) and Fis2-F3412 (59-GGATGATGTAGGAAATC-

CCCAATTGAGCCCTTTG-39). Allele-specific RT-PCR of FWA was

performed as described (Kinoshita et al., 2004). Allele-specific RT-PCR

ofMEAwas performed as described (Kinoshita et al., 1999). Primers used

to amplify FWAwere FWA-RTf2 (59-GTTACATGGATTGAACAAGCGG-39)

DNA Methylation and Imprinting 1369

and FWA-RTr2 (59-ACCTTGAATGAGTGCAGCAGTTG-39). Primers

used to amplify the control, GAPDH RNA, were GAPDH39 (59-GTAGC-

CCCACTCGTTGTCGTA-39) and GAPDH59 (59-AGGGTGGTGCCAA-

GAAGGTTG-39). A negative control without RT was used for each

RT-PCR experiment. We designed primers spanning introns, such that

genome-specific and cDNA-specific PCR products could easily be

distinguished by size differences.

Microscopy

FIS2-GUS expression was analyzed as described by Jullien et al. (2006).

KS117 GFP fluorescence was analyzed in 5-DAP seeds with a stereomi-

croscope (MZ16FA; Leica). GUS-stained developing seeds or pistils

cleared with a derivative of Hoyer’s medium were observed with differ-

ential interference contrast optics. In developing seeds mounted in 50%

glycerol solution, FWA-GFP fluorescence was analyzed with a GFP-

specific filter set with a320 planapo objective (DM6000 B; Leica). Images

were acquired with a DXM1200F digital camera (Nikon) and processed

using Metamorph (version 6.2; Universal Imaging). The settings for ob-

servation were saved and applied to all seeds observed to allow com-

parison of fluorescence intensities at 2 and 4 DAP. Pollen was prepared

as described (Rotman et al., 2005). In pollen, FWA-GFP fluorescence was

analyzed by confocal microscopy (LSM 510) with a340 water-immersion

objective, and imageswere further processedwith Adobe Photoshop and

Adobe Illustrator.

For each measurement of FWA-GFP and FIS2-GUS in developing

seeds, counting was established on individual siliques. Measurements of

FWA-GFP expression in pollen were performed by placing several an-

thers on a slide. Fifteen individual fields were observed, each containing

20 to 50 pollen grains, which were scored, thus establishing the data set

for determining the percentage of positive expressing pollen. Standard

deviation was calculated for each genetic background used based on the

average obtained per silique or per field of pollen grains observed.

McrBC Analysis

DNA methylation status in the FIS2 promoter was analyzed by PCR

amplification of DNA that had beenpretreatedwithMcrBCendonuclease,

which digests methylated DNA (New England Biolabs). One hundred

nanograms of genomic DNA from rosette leaves of met1-3 and Col-0

plants was digested overnight with 10 units of McrBC, according to the

manufacturer’s instructions. Ten nanograms of template DNA was then

amplified by PCR with ExTaq polymerase (Takara) using the following

primers: F3697Tf (59-AAAGAGTTATGGGYYGAAG-39) and F3697Tr

(59-GGGCAGAAACATGGTCCA-39), FIS2.-3104BisBf (59-ACAARTCAC-

AACCAAAACCTTAA-39) and FIS2.-2312r (59-TGCCAAATAGCACAATGA-

GGA-39), FIS2.-3104BisBf and FIS2.-1867r (59-ATGTTGCGCCTTCACCA-

CTT-39), and FIS2.-1218BisBf (59-TCCARTCCACTATTCTTTACTCTT-39)

and FIS2.-16r (59-GTAGTTGAATCTTATTTTCCCACCTGA-39).

Bisulfite Sequencing

Bisulfite sequencing was performed as described (Paulin et al., 1998).

DNA was purified from rosette leaves. After chemical bisulfite reaction,

the top strand of the FIS2 promoter from �2185 to �1768 (relative to the

translation start) was amplified with primers FIS2.-2185BisTf (59-AGG-

TYYAATYGYATATTTATTTAGGGTTTYGGGT-39) and FIS2.-1768BisTr

(59-TCCTACATTTTAATAAAATATTACTRAATCTAARCA-39), and thebottom

strand from�2110 to�1670 was amplified with primers FIS2.-2110BisBf

(59-TACCAAACCCRAARAARAAAAATTTACAA-39) and FIS2.-1670BisBr

(59-TGATGGYAGTAGAGATTATAAGAAAAGA-39). The amplified PCR

fragments were gel-purified and cloned into pT7Blue plasmid (Novagen),

and then six to nine independent cloneswere sequenced. TheASA1 gene

was used as a positive control for the bisulfite chemical reaction

(Jeddeloh et al., 1998). We used met1-3 homozygous mutants segre-

gated from met1-3/þ heterozygotes because these plants do not accu-

mulate epimutations, in contrast with MET1as selfed plants.

Accession Numbers

Arabidopsis Genome Initiative numbers for the genes used in this

study are At1g02580 (MEA), At2g35670 (FIS2), At4g25530 (FWA), and

At5g04560 (DME).

Supplemental Data

The following materials are available in the online version of this article.

Supplemental Figure 1. Analysis of FIS2 Imprinting at 3 and 5 DAP.

Supplemental Figure 2. Bisulfite Sequencing of the FIS2 59 Region

Contains a 200-bp Domain Enriched in Methylated Cytosine Residues

Predominantly on CpGs.

Supplemental Table 1. Resistance to BASTA Is Associated with

met1-3.

ACKNOWLEDGMENTS

We thank all contributors of materials cited in Methods and the

Nottingham Arabidopsis Resource Centre and ABRC. P.E.J. and F.B.

were supported by the Temasek Lifesciences Laboratory and by the

National University of Singapore. N.O. was supported by the Israel

Science Foundation (Grant 574-04) and by the U.S.–Israel Binational

Agricultural Research and Development Fund (Grant IS-3604-04c).

Received January 16, 2006; revised March 24, 2006; accepted April 11,

2006; published April 28, 2006.

REFERENCES

Adams, S., Vinkenoog, R., Spielman, M., Dickinson, H.G., and Scott,

R.J. (2000). Parent-of-origin effects on seed development in Arabidop-

sis thaliana require DNA methylation. Development 127, 2493–2502.

Bastow, R., Mylne, J.S., Lister, C., Lippman, Z., Martienssen, R.A.,

and Dean, C. (2004). Vernalization requires epigenetic silencing of

FLC by histone methylation. Nature 427, 164–167.

Berger, F. (2003). Endosperm: The crossroad of seed development.

Curr. Opin. Plant Biol. 6, 42–50.

Berger, F. (2004). Plant sciences. Imprinting—A green variation. Science

303, 483–485.

Cao, X., and Jacobsen, S.E. (2002a). Role of the Arabidopsis DRM

methyltransferases in de novo DNA methylation and gene silencing.

Curr. Biol. 12, 1138–1144.

Cao, X., and Jacobsen, S.E. (2002b). Locus-specific control of asym-

metric and CpNpG methylation by the DRM and CMT3 methyltrans-

ferase genes. Proc. Natl. Acad. Sci. USA 99 (suppl. 4), 16491–16498.

Chan, S.W., Henderson, I.R., and Jacobsen, S.E. (2005). Gardening

the genome: DNA methylation in Arabidopsis thaliana. Nat. Rev.

Genet. 6, 351–360.

Chan, S.W., Zilberman, D., Xie, Z., Johansen, L.K., Carrington, J.C.,

and Jacobsen, S.E. (2004). RNA silencing genes control de novo

DNA methylation. Science 303, 1336.

Chaudhury, A.M., Ming, L., Miller, C., Craig, S., Dennis, E.S., and

Peacock, W.J. (1997). Fertilization-independent seed development in

Arabidopsis thaliana. Proc. Natl. Acad. Sci. USA 94, 4223–4228.

1370 The Plant Cell

Choi, Y., Gehring, M., Johnson, L., Hannon, M., Harada, J.J., Gold-

berg, R.B., Jacobsen, S.E., and Fischer, R.L. (2002). DEMETER, a

DNA glycosylase domain protein, is required for endosperm gene

imprinting and seed viability in Arabidopsis. Cell 110, 33–42.

Constancia, M., Kelsey, G., and Reik, W. (2004). Resourceful imprint-

ing. Nature 432, 53–57.

Danilevskaya, O.N., Hermon, P., Hantke, S., Muszynski, M.G.,

Kollipara, K., and Ananiev, E.V. (2003). Duplicated fie genes in

maize: Expression pattern and imprinting suggest distinct functions.

Plant Cell 15, 425–438.

Delaval, K., and Feil, R. (2004). Epigenetic regulation of mammalian

genomic imprinting. Curr. Opin. Genet. Dev. 14, 188–195.

Durbarry, A., Vizir, I., and Twell, D. (2005). Male germ line development

in Arabidopsis. duo pollen mutants reveal gametophytic regulators of

generative cell cycle progression. Plant Physiol. 137, 297–307.

Finnegan, E.J., Peacock, W.J., and Dennis, E.S. (1996). Reduced DNA

methylation in Arabidopsis thaliana results in abnormal plant devel-

opment. Proc. Natl. Acad. Sci. USA 93, 8449–8454.

Finnegan, E.J., Peacock, W.J., and Dennis, E.S. (2000). DNA meth-

ylation, a key regulator of plant development and other processes.

Curr. Opin. Genet. Dev. 10, 217–223.

Gehring, M., Choi, Y., and Fischer, R.L. (2004). Imprinting and seed

development. Plant Cell 16 (suppl.), S203–S213.

Gehring, M., Huh, J.H., Hsieh, T.F., Penterman, J., Choi, Y., Harada,

J.J., Goldberg, R.B., and Fischer, R.L. (2006). DEMETER DNA

glycosylase establishes MEDEA polycomb gene self-imprinting by

allele-specific demethylation. Cell 124, 495–506.

Grossniklaus, U., Vielle-Calzada, J.P., Hoeppner, M.A., and Gagliano,

W.B. (1998). Maternal control of embryogenesis by MEDEA, a poly-

comb group gene in Arabidopsis. Science 280, 446–450.

Guignard, M.L. (1899). Sur les antherozoıdes et la double copulation

sexuelle chez les vegetaux angiospermes. Rev. Gen. Bot. 11, 129–135.

Guitton, A.E., Page, D.R., Chambrier, P., Lionnet, C., Faure, J.E.,

Grossniklaus, U., and Berger, F. (2004). Identification of new mem-

bers of Fertilisation Independent Seed polycomb group pathway

involved in the control of seed development in Arabidopsis thaliana.

Development 131, 2971–2981.

Haseloff, J. (1999). GFP variants for multispectral imaging of living cells.

Methods Cell Biol. 58, 139–151.

Honys, D., and Twell, D. (2004). Transcriptome analysis of haploid male

gametophyte development in Arabidopsis. Genome Biol. 5, R85.

Jackson, J.P., Lindroth, A.M., Cao, X., and Jacobsen, S.E. (2002).

Control of CpNpG DNA methylation by the KRYPTONITE histone H3

methyltransferase. Nature 416, 556–560.

Jeddeloh, J.A., Bender, J., and Richards, E.J. (1998). The DNA

methylation locus DDM1 is required for maintenance of gene silencing

in Arabidopsis. Genes Dev. 12, 1714–1725.

Jullien, P.E., Katz, A., Oliva, M., Ohad, N., and Berger, F. (2006).

Polycomb group complexes self-regulate imprinting of the polycomb

group gene MEDEA in Arabidopsis. Curr. Biol. 16, 486–492.

Kaneda, M., Okano, M., Hata, K., Sado, T., Tsujimoto, N., Li, E., and

Sasaki, H. (2004). Essential role for de novo DNA methyltransferase

Dnmt3a in paternal and maternal imprinting. Nature 429, 900–903.

Katz, A., Oliva, M., Mosquna, A., Hakim, O., and Ohad, N. (2004). FIE and

CURLY LEAF polycomb proteins interact in the regulation of homeobox

gene expression during sporophyte development. Plant J. 37, 707–719.

Kinoshita, T., Miura, A., Choi, Y., Kinoshita, Y., Cao, X., Jacobsen,

S.E., Fischer, R.L., and Kakutani, T. (2004). One-way control of FWA

imprinting in Arabidopsis endosperm by DNA methylation. Science

303, 521–523.

Kinoshita, T., Yadegari, R., Harada, J.J., Goldberg, R.B., and

Fischer, R.L. (1999). Imprinting of the MEDEA polycomb gene in

the Arabidopsis endosperm. Plant Cell 11, 1945–1952.

Kiyosue, T., Ohad, N., Yadegari, R., Hannon, M., Dinneny, J., Wells,

D., Katz, A., Margossian, L., Harada, J.J., Goldberg, R.B., and

Fischer, R.L. (1999). Control of fertilization-independent endosperm

development by the MEDEA polycomb gene in Arabidopsis. Proc.

Natl. Acad. Sci. USA 96, 4186–4191.

Kohler, C., and Grossniklaus, U. (2005). Seed development and

genomic imprinting in plants. Prog. Mol. Subcell. Biol. 38, 237–262.

Kohler, C., Hennig, L., Bouveret, R., Gheyselinck, J., Grossniklaus,

U., and Gruissem, W. (2003). ArabidopsisMSI1 is a component of the

MEA/FIE polycomb group complex and required for seed develop-

ment. EMBO J. 22, 4804–4814.

Kohler, C., Page, D.R., Gagliardini, V., and Grossniklaus, U. (2005).

The Arabidopsis thaliana MEDEA polycomb group protein controls

expression of PHERES1 by parental imprinting. Nat. Genet. 37, 28–30.

Lindroth, A.M., Cao, X., Jackson, J.P., Zilberman, D., McCallum,

C.M., Henikoff, S., and Jacobsen, S.E. (2001). Requirement of

CHROMOMETHYLASE3 for maintenance of CpXpG methylation.

Science 292, 2077–2080.

Luo, M., Bilodeau, P., Dennis, E.S., Peacock, W.J., and Chaudhury,

A. (2000). Expression and parent-of-origin effects for FIS2, MEA, and

FIE in the endosperm and embryo of developing Arabidopsis seeds.

Proc. Natl. Acad. Sci. USA 97, 10637–10642.

Luo, M., Bilodeau, P., Koltunow, A., Dennis, E.S., Peacock, W.J., and

Chaudhury, A.M. (1999). Genes controlling fertilization-independent

seed development in Arabidopsis thaliana. Proc. Natl. Acad. Sci. USA

96, 296–301.

Luo, M., Dennis, E.S., Berger, F., Peacock, W.J., and Chaudhury, A.

(2005). MINISEED3 (MINI3), a WRKY family gene, and HAIKU2 (IKU2),

a leucine-rich repeat (LRR) kinase gene, are regulators of seed size in

Arabidopsis. Proc. Natl. Acad. Sci. USA 102, 17531–17536.

Maheshwari, P. (1950). An Introduction to the Embryology of Angio-

sperms. (New York: McGraw-Hill).

McCormick, S. (2004). Control of male gametophyte development.

Plant Cell 16 (suppl.), S142–S153.

Monk, M. (1990). Changes in DNA methylation during mouse embryonic

development in relation to X-chromosome activity and imprinting.

Philos. Trans. R. Soc. Lond. B Biol. Sci. 326, 299–312.

Ohad, N., Yadegari, R., Margossian, L., Hannon, M., Michaeli, D.,

Harada, J.J., Goldberg, R.B., and Fischer, R.L. (1999). Mutations in

FIE, a WD polycomb group gene, allow endosperm development

without fertilization. Plant Cell 11, 407–416.

Paulin, R., Grigg, G.W., Davey, M.W., and Piper, A.A. (1998). Urea

improves efficiency of bisulphite-mediated sequencing of 59-methyl-

cytosine in genomic DNA. Nucleic Acids Res. 26, 5009–5010.

Rotman, N., Durbarry, A., Wardle, A., Yang, W.C., Chaboud, A.,

Faure, J.E., Berger, F., and Twell, D. (2005). A novel class of MYB

factors controls sperm-cell formation in plants. Curr. Biol. 15,

244–248.

Saze, H., Scheid, O.M., and Paszkowski, J. (2003). Maintenance of

CpG methylation is essential for epigenetic inheritance during plant

gametogenesis. Nat. Genet. 34, 65–69.

Smyth, D.R., Bowman, J.L., and Meyerowitz, E.M. (1990). Early flower

development in Arabidopsis. Plant Cell 2, 755–767.

Sørensen, M.B., Chaudhury, A.M., Robert, H., Bancharel, E., and

Berger, F. (2001). Polycomb group genes control pattern formation in

plant seed. Curr. Biol. 11, 277–281.

Spillane, C., Baroux, C., Escobar-Restrepo, J.M., Page, D.R.,

Laoueille, S., and Grossniklaus, U. (2004). Transposons and tandem

repeats are not involved in the control of genomic imprinting at the

MEDEA locus in Arabidopsis. Cold Spring Harb. Symp. Quant. Biol.

69, 465–475.

Takeda, S., and Paszkowski, J. (2006). DNA methylation and epigenetic

inheritance during plant gametogenesis. Chromosoma 115, 27–35.

DNA Methylation and Imprinting 1371

Tariq, M., Saze, H., Probst, A.V., Lichota, J., Habu, Y., and

Paszkowski, J. (2003). Erasure of CpG methylation in Arabidopsis

alters patterns of histone H3 methylation in heterochromatin. Proc.

Natl. Acad. Sci. USA 100, 8823–8827.

Vielle-Calzada, J.P., Baskar, R., and Grossniklaus, U. (2000). Delayed

activation of the paternal genome during seed development. Nature

404, 91–94.

Vielle-Calzada, J.P., Thomas, J., Spillane, C., Coluccio, A., Hoeppner,

M.A., and Grossniklaus, U. (1999). Maintenance of genomic imprint-

ing at the Arabidopsis medea locus requires zygotic DDM1 activity.

Genes Dev. 13, 2971–2982.

Vinkenoog, R., Spielman, M., Adams, S., Fischer, R.L., Dickinson,

H.G., and Scott, R.J. (2000). Hypomethylation promotes autonomous

endosperm development and rescues postfertilization lethality in fie

mutants. Plant Cell 12, 2271–2282.

Vire, E., et al. (2006). The polycomb group protein EZH2 directly

controls DNA methylation. Nature 439, 871–874.

Xiao, W., Custard, K.D., Brown, R.C., Lemmon, B.E., Harada, J.J.,

Goldberg, R.B., and Fischer, R.L. (2006). DNA methylation is critical

for Arabidopsis embryogenesis and seed viability. Plant Cell 18,

805–814.

Xiao, W., Gehring, M., Choi, Y., Margossian, L., Pu, H., Harada, J.J.,

Goldberg, R.B., Pennell, R.I., and Fischer, R.L. (2003). Imprinting of

the MEA polycomb gene is controlled by antagonism between MET1

methyltransferase and DME glycosylase. Dev. Cell 5, 891–901.

Yadegari, R., Kinoshita, T., Lotan, O., Cohen, G., Katz, A., Choi, Y.,

Nakashima, K., Harada, J.J., Goldberg, R.B., Fischer, R.L., and

Ohad, N. (2000). Mutations in the FIE and MEA genes that encode

interacting polycomb proteins cause parent-of-origin effects on seed

development by distinct mechanisms. Plant Cell 12, 2367–2381.

Zemach, A., Li, Y., Wayburn, B., Ben-Meir, H., Kiss, V., Avivi, Y.,

Kalchenko, V., Jacobsen, S.E., and Grafi, G. (2005). DDM1 binds

Arabidopsis methyl-CpG binding domain proteins and affects their

subnuclear localization. Plant Cell 17, 1549–1558.

1372 The Plant Cell

DOI 10.1105/tpc.106.041178; originally published online April 28, 2006; 2006;18;1360-1372Plant Cell

Pauline E. Jullien, Tetsu Kinoshita, Nir Ohad and Frédéric BergerImprinting

Life Cycle Is Essential for ParentalArabidopsisMaintenance of DNA Methylation during the

 This information is current as of May 1, 2020

 

Supplemental Data /content/suppl/2006/04/28/tpc.106.041178.DC1.html

References /content/18/6/1360.full.html#ref-list-1

This article cites 57 articles, 29 of which can be accessed free at:

Permissions https://www.copyright.com/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X

eTOCs http://www.plantcell.org/cgi/alerts/ctmain

Sign up for eTOCs at:

CiteTrack Alerts http://www.plantcell.org/cgi/alerts/ctmain

Sign up for CiteTrack Alerts at:

Subscription Information http://www.aspb.org/publications/subscriptions.cfm

is available at:Plant Physiology and The Plant CellSubscription Information for

ADVANCING THE SCIENCE OF PLANT BIOLOGY © American Society of Plant Biologists