Introduction

A subset of animal and plant genes is expressed in a parent-of-origin-specific manner either from the paternal or maternal allele. This phenomenon, called imprinting, was first discovered in maize as a maternal control of the kernel aleurone color (Kermicle 1970). Later, imprinting was described in mice through pronuclear transplantation experiments (Surani and Barton 1983; Reik and Walter 2001). To date, more than 80 imprinted genes have been described in mammals (Morison et al. 2005).

In plants imprinting has been found so far only in the endosperm, a terminal nutritive tissue that develops after double fertilization from the fertilized central cell. The endosperm provides nutrients for developing embryos in dicots and for germinating seedlings in monocots (Kranz et al. 1998; Walbot and Evans 2003). A dozen or so imprinted genes have been discovered in plants, of which the majority are maternally expressed and paternally silenced in the endosperm (Alleman and Doctor 2000; Baroux et al. 2002; Gehring et al. 2004a, b). So far, only the Arabidopsis MADS-box gene PHERES1 was shown to be expressed paternally due to the repressing activity of the Polycomb Group (PcG) protein, MEDEA (Köhler et al. 2003a, b, 2005).

Plant PcG proteins are important components of parent-of-origin control of gene expression and have been reported recently to be involved also in the self-regulation of imprinting (Baroux et al. 2006; Gehring et al. 2006; Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). PcG proteins form complexes that are able to silence genes and maintain their silencing over many cell divisions by a mechanism that relies mainly on histone modifications at the repressed locus (Delaval and Feil 2004). The Drosophila Enhancer of Zeste-Extra Sex Combs [E(Z)-ESC] PcG complex, for example, functions as a repressor of a number of homeotic genes (Sathe and Harte 1995). The E(Z)-ESC PcG complex appears to be conserved, but more complex in plants (Grossniklaus and Schneitz 1998; Grossniklaus et al. 1998; Luo et al. 1999; Ohad et al. 1999; Springer et al. 2002; Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). In Arabidopsis the Fertilization-independent seed (FIS) complex that controls seed development is composed of Medea (MEA), Fertilization-independent endosperm (FIE), Fertilization-independent seed2 (FIS2) and Multicopysupressor of IRA 1 (MSI1), which are homologs of the Drosophila E(Z), ESC, Suppressor of Zeste12 and CAF1/P55 PcG proteins, respectively (Grossniklaus and Schneitz 1998; Grossniklaus et al. 1998; Luo et al. 1999; Ohad et al. 1999; Köhler et al. 2003a, b). MEA and FIS2 are imprinted genes and are expressed maternally in the endosperm (Grossniklaus and Schneitz 1998; Kinoshita et al. 1999; Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). These two genes as well as the imprinted gene FWA, which displays an endosperm-specific expression, have been studied intensively to elucidate the mechanism of imprinting in Arabidopsis (Kinoshita et al. 2004; Baroux et al. 2006; Gehring et al. 2006; Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). It was shown that the maintenance of FWA and FIS2 imprinting depends on DNA methylation (Kinoshita et al. 2004; Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). In contrast, the MEA paternal allele is silenced by the MEA-FIE PcG complex due to histone methylation and thus is independent of DNA methylation (Baroux et al. 2006; Gehring et al. 2006; Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). However, MEA, FWA and FIS2 imprinting share a common feature such as the activation of maternal alleles in the central cell due to demethylation by Demeter (DME) DNA glycosylase before fertilization (Choi et al. 2002; Kinoshita et al. 2004; Gehring et al. 2006; Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). After fertilization, the maternal allele remains transcriptionally active in the endosperm, whereas the paternal allele remains silent as it was in the sperm. Because the endosperm is a terminal tissue and is not transmitted to the next generation, the epigenetic state of the imprinted genes does not require re-setting of the imprinting marks and a model for the one way control of imprinting in Arabidopsis was proposed (Kinoshita et al. 2004).

Despite the original discovery of imprinting in maize, the paucity of information about the molecular mechanisms of imprinting is available in this species. Maternal demethylation of certain alleles of zein and α-tubulin genes have been reported in the maize endosperm (Lund and Ciceri et al. 1995; Lund and Messing et al. 1995; Alleman and Doctor 2000). Extensive maternal hypomethylation was detected in the maize endosperm by a PCR-based genomic scan (Lauria et al. 2004). However, DNA methylation of locus-specific imprinted genes has not been reported in maize until recently (Gutierrez-Marcos et al. 2006). Previously, we identified two FIE homologs in maize, Fie1 and Fie2, that show distinct expression patterns and imprinting during kernel development (Springer et al. 2002; Danilevskaya et al. 2003). Fie1 expression is restricted to the endosperm and shows mono-allelic expression from the maternal allele, whereas, Fie2 is broadly expressed in many tissues and shows bi-allelic expression in the embryo and the endosperm at later stages (Danilevskaya et al. 2003; Gutierrez-Marcos et al. 2003). Based on the pattern of expression we proposed that Fie1 and Fie2 may have evolved divergent functions in maize. In this study, we examined Fie1 and Fie2 expression in the isolated gametes before fertilization, and in zygotes and developing kernels at the early stages after fertilization. We also investigated DNA methylation of Fie1 and Fie2 and report the demethylation of the maternal Fie1 allele, but not of the paternal allele after fertilization and the silencing by DNA metylation of Fie1 in other tissues including the embryo and endosperm.

Materials and methods

Plant material

Reciprocal crosses were performed between maize inbred lines B73 and Mo17. Embryo and endosperm tissues were dissected from kernels at 14 days after pollination (DAP). Pericarp was removed from the kernels and was not included in DNA extraction. Tissues were frozen in liquid nitrogen and stored at −80°C. Pollen and leaf tissue were obtained from mature B73 or Mo17 plants. Inbred line A188 was used for isolation of individual gametes.

DNA isolation and MSRE-PCR

Genomic DNA was isolated from 10 mg to 20 mg of frozen embryo and endosperm tissues using Puregene™ DNA isolation components (Gentra Systems, Minneapolis, MN). About 0.5 μg of genomic DNA was digested at 37°C overnight in a 50 μl reaction with 25 U of either HpaII or MspI (New England Biolabs, Beverly, MA) for MSRE-PCR (Liang et al. 2004). After an overnight incubation, 25 additional units of the appropriate restriction enzyme were added and the reaction was incubated for 2 h at 37°C. PCR amplification was performed using Expand Long Template DNA polymerase (Roche, Germany). About 2 μl of the restriction enzyme reaction was used for PCR amplification in a 50 μl volume. The PCR conditions were 95°C for 5 min followed by 35 cycles at 95°C for 45 s, 60°C for 45 s, 72°C for 1 min and a final extension of 72°C for 10 min.

Bisulfite sequencing

Bisulfite treatment (Engemann et al. 2001) was performed with the EZ DNA Methylation Kit™ (Zymo Research, Orange, CA). In a 50 μl reaction, 1 μg of genomic DNA was treated according to the manufacturer’s recommendations. After bisulfite treatment, PCR was performed using gene specific primers for the bisulfite converted DNA sequence (Details in Suppl. Materials). The PCR conditions used were the same as outlined for MSRE-PCR. PCR products were subcloned into pCR®4-TOPO vector (Invitrogen, Carlsbad, CA) and sequenced with M13F and M13R primers.

Single Cell RT-PCR (SC RT-PCR)

Cells of the female gametophyte before and after fertilization as well as sperm cells were isolated from the maize inbred line A188 as described by Cordts et al. (2001). SC RT-PCR was performed according to Richert et al. (1996) with one modification. A multiplex reverse transcription reaction was conducted on each cell either with reverse R-fie1 and R-fie2 primers for Fie1 and Fie2, respectively, in addition to the reverse GAPDH (glyceraldehyde-3-phosphate dehydrogenase) primer Gap2. Generated cDNA was split into two equal amounts to amplify Fie and GAPDH transcripts separately during 40 PCR cycles using primers described in Table S1. Gels were blotted and hybridized as described (Dresselhaus et al. 1999) with radio-labeled Fie1 and Fie2-specific probes.

Real-time quantitative RT-PCR (qRT-PCR)

Developing kernels were collected every 24 h from the ear base harvested from the B73 inbred pollinated by the Mo17 inbred in the field. Total RNA was isolated from whole kernels using TRIzol® Reagent with Phase Lock Gel-Heavy (Eppendorf North America, Inc., Molecular Research Center, Inc.). Total RNA was treated with DNase I (Invitrogen, Carlsbad, CA) and reverse-transcribed by using the Taqman Reverse Transcription kit (Applied Biosystems, Foster City, CA) according to the manufacturer’s instructions. The final RNA concentration in the cDNA synthesis reaction was 30 ng/μl. The absence of genomic DNA contamination was confirmed by the PCR without the reverse transcriptase. To ensure gene-specific amplification, primers and 5’ FAM labeled LNA probes (Suppl. Table S2) were designed against 3’ UTR of Fie1 and Fie2 using Universal ProbeLibrary (Roche Diagnostics, Indianapolis, IN). Primers and the 5’ VIC labeled MGB probe (Table S2) for β-Actin were designed using Primer Express 2.0 software (Applied Biosystems, Foster City, CA). The expression of each gene was assayed in triplicate in a total volume of 20 μl containing 1× Taqman master mix, 300 nM forward and reverses primers, 100 nM probes and 3 μl of the 1:4 diluted cDNA. The PCR thermal cycling parameters were 50°C for 2 min, 95°C for 10 min followed by 40 cycles of 95°C for 15 s and 60°C for 1 min. All assays were run on an ABI 7500 Real Time PCR System (Applied Biosystems, Foster City, CA). Transcript levels of Fie1 and Fie2 were measured relatively to the endogenous reference β-Actin with the Ct method as described by the manufacturer. The PCR amplification efficiency was determined by measuring a series of input cDNA concentrations. The PCR efficiency for Fie1, Fie2 and β-Actin was 1.01, 1.03 and 1.02, respectively.

Results

The maternal Fie1 allele is activated in the endosperm after fertilization whereas Fie2 is expressed in both gametes and zygotes

As shown previously, Fie1 transcript was detected exclusively in the endosperm where it is expressed from the maternal allele. Fie2 is expressed in a broad set of tissues including ovules before fertilization (Danilevskaya et al. 2003; Gutierrez-Marcos et al. 2003). However, Fie1 and Fie2 expression was never analyzed in gametes. To address this issue, transcription of Fie1 and Fie2 was examined in manually isolated gametes and zygotes by single cell RT-PCR (SC RT-PCR) (Richert et al. 1996; Cordts et al. 2001). Fie1 mRNA was neither detected in the individual gametes nor in zygotes 12 h after in vitro pollination, which is around 6 h after fertilization (Fig. 1A). In contrast, Fie2 transcript is present in all gametic cells with the highest transcript level in the central cell (Fig. 1B). To detect the onset of Fie1 activation in the endosperm, developing kernels were collected every 24 h after pollination and analyzed by quantitative real-time RT-PCR (qRT-PCR) for Fie1 and Fie2 expression. Fie1 transcript was at background level in the ovules and in developing kernels at 1 DAP. Fie1 mRNA was first detected around 2 DAP (Fig. 1C and Suppl. Tables S1 and S2). Because fertilization occurs around 16–24 h after in vivo pollination (Kiesselbach 1999), we conclude that Fie1 activation takes place approximately 24–32 h after fertilization. The highest Fie1 mRNA level was detected in kernels at 10–11 DAP and gradually decreased at later stages (Fig. 1C). In contrast, Fie2 transcript was present in mature ovules and its level slightly increased at 1–4 DAP, but decreased at later stages. Overall, Fie2 mRNA level was significantly lower than Fie1. At 10 DAP, expression of Fie1 is about 80 times higher than that of Fie2.

Fig. 1
figure 1

Fie1 and Fie2 expression in gametes and developing kernels. (A) RT-PCR analysis of Fie1 and (B) Fie2 expression in 15 sperm cells (SC), 3 central cells (CC), 3 egg cells (EC), and 3 zygotes (Z) collected 12 h after in vitro pollination (around 6 h after fertilization). Fie PCR products were blotted and hybridized with a gene-specific probe. DNA fragment sizes (kb) are indicated by arrows. This experiment was repeated twice to confirm the obtained result. (C) Quantification of the Fie1 and Fie2 mRNAs in developing kernels by qRT-PCR. Relative amounts were calculated and normalized with respect to actin transcript levels (=100%). Data shown represent mean values obtained from three independent amplification reactions, and the error bars indicate the standard error of the mean. The original numbers are shown in Suppl. Table S1

Full size image

Fie1 is methylated in most tissues whereas Fie2 is not methylated

To examine the role of DNA methylation in determining the expression pattern of the Fie1 (GenBank Accession #AY150645) and Fie2 (GenBank Accession #AY150646) genes, we analyzed their methylation status in different tissues. Methylation sensitive restriction enzyme digestion PCR (MSRE-PCR) analysis with the isoschizomers HpaII and MspI was performed to assess the methylation pattern across the two genes (Liang et al. 2004). Both enzymes recognize CCGG sites but HpaII does not cut DNA if either cytosine is methylated. MspI does not cut DNA if the external cytosine is methylated. Fie1 contains seven CCGG sites distributed throughout the gene and Fie2 contains ten CCGG sites clustered in upstream and at exon 1 (Fig. 2A and Suppl. Table S3). Using both enzymes for MSRE-PCR analysis, we found that CCGG sites at exon 1 and exon 7 were methylated in the embryo, endosperm and leaf DNA (Fig. 2B). CCGG sites at exons 11–13 showed a low level of methylation. DNA methylation was not detected in Fie2 (Fig. 2B).

Fig. 2
figure 2

Methyl-sensitive restriction-enzyme-dependent (MSRE) PCR of Fie1 and Fie2. (A) Genomic maps of Fie1 (GenBank accession #AY150645) and Fie2 (GenBank accession #AY150646) are shown with MspI/HpaII restriction sites. Exons are depicted by black arrows. Introns are shown with lines. Retrotransposon RIRE is depicted by the shaded arrow. Gene specific primers, shown with small arrows, were used to amplify four regions of Fie1 and one region of Fie2 containing MspI/HpaII sites. Primers are marked with location relative to the ATG start codon. (B) Agarose gel with PCR products of the five analyzed regions of the Fie genes. Genomic DNA was isolated from embryo and endosperm tissue from reciprocal crosses of inbred lines B73 and Mo17 (female shown first). DNA was treated with MspI or HpaII restriction enzyme prior to amplification. The presence of PCR products indicates methylation of MspI/HpaII sites within the region. Undigested DNA was used as a positive control

Full size image

Bisulfite sequencing (Engemann et al. 2001) was applied to quantify Fie1 and Fie2 DNA methylation at CpG and CpNpG sites in the embryo, endosperm, pollen and leaf tissues. The promoter region was of particular interest because it lacks HpaII/MspI sites. The promoter, exon 1 and the region including exons 6–8 of Fie1 are shown in Fig. 3A. Endosperm DNA revealed a low level of methylation ranging from 20% to 40% at CG sites and from 14% to 16% at CNG sites (Fig. 3A and Suppl. Table S4) in all segments analyzed including the promoter. In contrast, Fie1 DNA from embryo, pollen and leaves was methylated up to 58–97% at CG-sites and 50–75% at CNG sites (Fig. 3A and Suppl. Table S4). Combining MSRE-PCR and bisulfite sequencing allowed the conclusion that the Fie1 gene is methylated across the promoter region and approximately half of the coding region up to exon 8. Significant cytosine methylation was not detected at the Fie2 gene (Fig. 3B).

Fig. 3
figure 3

Bisulfite DNA methylation pattern of Fie1 and Fie2 in different maize tissues. (A) Genomic structure of Fie1 is shown above methylation profiles. Exons are depicted with shaded boxes. Introns are shown by solid black lines. The transcription start point is marked with an arrowhead and two SNPs in exon 1 between parental inbred lines B73 and Mo17 are marked with asterisks. Regions used for bisulfite sequencing are underlined. Percent methylation at CG, CNG and asymmetric sites in endosperm, embryo, pollen and leaf is indicated. (B) Genomic structure and methylation profile of Fie2. Description as in (A). Detailed information about numbers and clones analyzed and primers used are provided in Suppl. Tables S3 and S4

Full size image

Maternal Fie1 allele is hypomethylated in the endosperm

To discriminate maternal and paternal alleles in bisulfite treated DNA molecules, we took advantage of two SNPs in exon 1 in inbred lines B73 and Mo17 (Fig. 4A). Due to the preferential amplification of the maternal molecules from the endosperm DNA (Fig. 3A) we have sequenced 116 clones amplified from exon 1 using endosperm DNA from reciprocal B73 × Mo17 crosses. 102 clones were of maternal and 14 of paternal origin. We found that maternal endosperm molecules were hypomethylated (2.8% methylated cytosines at CG sites and 3.3% at CNG) compared to corresponding paternal endosperm molecules (65.7% at CG and 49.2% at CNG). Maternal DNA isolated from embryos displayed a high methylation content (64.0% at CG and 56.7% at CNG) similar to corresponding paternal DNA regions (70.0% at CG, 63.9% at CNG) (Fig. 4B and Suppl. Table S4). Because the maternal endosperm alleles have been preferentially amplified, we designed an additional experiment to confirm methylation at the paternal allele. Embryo and endosperm DNA from reciprocal crosses of B73 and Mo17 were digested to completion with HpaII and amplified across HpaII sites and SNPs in exon 1 with gene-specific primers extended with T3 and T7 adapters (Fig. 4A). PCR fragments obtained were directly sequenced with T3 and T7. Sequencing chromatograms of PCR products from digested and undigested embryo DNA showed the presence of SNPs from both parents, B73 and Mo17 (Fig. 4C), indicating that both parental alleles are methylated in the embryo. Conversely, the chromatograms of PCR products generated from digested endosperm DNA showed the presence of paternal SNPs but a complete absence of maternal SNPs (Fig. 4D). Undigested endosperm DNA showed a mixture of traces from both parents. These results confirm that the paternal Fie1 allele is methylated in the endosperm and therefore protected from digestion by HpaII. The maternal alleles are unmethylated and can be digested by HpaII. In summary, the maternal Fie1 allele is hypomethylated in the endosperm but methylated in the embryo. The expression of Fie1 is thus strongly correlated with a loss of DNA methylation of the maternal allele in the endosperm. The paternal allele is methylated in the endosperm and all other tissues investigated.

Fig. 4
figure 4

Demethylation of the maternal Fie1 allele in the endosperm. (A) Parental Fie1 alleles in reciprocal crosses were distinguished by two SNPs in exon 1 between inbred lines B73 and Mo17, respectively. Gene specific primers with T3 or T7 extensions were designed around SNPs and two linked HpaII sites for direct sequencing. (B) Percent methylation at CG, CNG and asymmetric sites of the Fie1 maternal and paternal alleles from the endosperm and embryo (Suppl. Table S4). Alleles were discriminated by SNPs in the exon 1. (C, D) Chromatograms of PCR products amplified from HpaII-digested and undigested embryo (C) and endosperm (D) DNA as indicated. Arrows point to the SNPs on chromatogram traces

Full size image

Discussion

The maize genome contains two homologous Fie genes which have distinct expression pattern (Danilevskaya et al. 2003). Fie1 is a highly regulated gene, expressed exclusively in the endosperm, where its expression is controlled by imprinting, resulting in mono-allelic expression of the maternal but not the paternal allele. In contrast, Fie2 shows bi-allelic expression in many tissues including embryo and endosperm at later stages of development. The Arabidopsis FIE gene functions as a repressor of endosperm development in the central cell prior to fertilization (Ohad et al. 1999) and seems to be involved in the regulation of the ontogenic sequence of endosperm development after fertilization (Ingouff et al. 2005). To elucidate functions of maize Fie genes prior to fertilization, we examined their expression pattern in manually isolated gametes from maize. Neither gene is expressed in the sperm. Out of both Fie genes, only Fie2 transcript was detected in the central cell supporting its putative function as a repressor of endosperm development before fertilization. At a lower level, Fie2 transcription was also detected in the egg cell and after fertilization in the zygote and developing endosperm. This expression pattern might be attributed to additional functions of Fie2 during kernel development similar to that of Arabidopsis FIE. Knock-out mutants should now be analyzed to study Fie2 functions. However, such mutants are not available yet.

The absence of Fie1 transcript in the central cell and the egg cell indicates that this gene does not play a role in the regulation of pre-fertilization events in the embryo sac. Rather it might have very specific function(s) during endosperm development. Endosperm development in maize and other cereals is characterized by distinct changes of the cell cycle pattern. The primary endosperm nucleus, resulting from fusion of a sperm with the two polar nuclei of the central cell, divides mitotically within 3–5 h after fertilization without cytokinesis forming a multinucleate syncytium (Kiesselbach 1999). Cell wall deposition is activated around 3 DAP and is completed at 4–5 DAP. Mitotic proliferation continues up to 10–15 DAP until the cell cycle switches to endoreduplication (Kowles and Phillips 1988). The transition from mitotic divisions to endoreduplication is thought to be controlled by parental imprinting (Dilkes et al. 2002; Leblanc et al. 2002). Expression of Fie1 begins approximately 24–32 h after fertilization reaching the highest level at 10–11 DAP coinciding with the described switch of the cell cycle from mitotic divisions to endoreduplication. As a PcG protein, Fie1 might regulate imprinting of other genes and it is therefore tempting to speculate that Fie1 might have a function in the maternal control of the transition to endoreduplication in the maize endosperm. Rice and sorghum genomes also contain two FIE homologues (Lai et al. 2004), suggesting that Fie genes in cereals might have evolved distinct functions compared to the single Arabidopsis FIE gene. It is interesting now to study whether rice and sorghum Fie1-like genes are also regulated by imprinting as is the case for the maize Fie1 gene. This would indicate conserved functions in the grasses.

The distinct feature of the Fie1 gene is its mono-allelic expression from the maternal allele in the endosperm at all stages of development (Danilevskaya et al. 2003). Fie2 shows bi-allelic expression in the embryo and in the endosperm at later stages. However, in the early endosperm at 6 DAP, Fie2 has been shown to be expressed maternally (Danilevskaya et al. 2003; Gutierrez-Marcos et al. 2003). To understand the role of DNA methylation in tissue-specific expression and imprinting, we have examined the methylation status of both Fie genes. Fie1 was found to be methylated in all tissues tested, which is consistent with its restricted expression pattern and indicates that the methylated silent state is the default for Fie1. DNA methylation was not detected in the Fie2 gene, which is consistent with its broad expression in many tissues. However, recently Fie2 methylation of the paternal allele was observed in 6 DAP endosperm indicating that the transient methylation of this gene takes place only during early stages of endosperm development (Gutierrez-Marcos et al. 2006), as the gene is no longer methylated at 14 DAP (this report).

Using bisulfite sequencing, we found a high level of Fie1 methylation at the promoter region, exon 1 and exons 6–8 in DNA samples isolated from embryos, pollen and leaves. A lower level of DNA methylation was detected in the endosperm, where only the maternal Fie1 alleles are expressed. Our results further showed that the maternal Fie1 allele is hypomethylated in the endosperm, but the paternal allele is hypermethylated. In DNA extracted from embryos, both maternal and paternal alleles are methylated at the same level.

This methylation pattern strongly correlates with Fie1 expression. Fie1 is methylated and silent in most tissues except the endosperm where the maternal allele is demethylated and transcribed.

Because Fie1 transcript was not detected in the central cell, it was unclear when demethylation of the Fie1 gene actually occurs. According to a recent study by Gutierrez-Marcos et al. (2006), Fie1 is methylated in the sperm and the egg cell, but hypomethylated in the central cell. Thus the demethylation of the Fie1 gene occurs before fertilization in the central cell. Despite its demethylation in the central cell, Fie1 becomes transcriptional active only in the primary endosperm after fertilization suggesting that demethylation is necessary but not sufficient for its activity. Endosperm-specific factors are apparently required to activate transcription of the maternal Fie1 gene in the endosperm. Demethylation and transcriptional activation of the maternal alleles of MEA (Xiao et al. 2003; Kinoshita et al. 2004), FWA (Kinoshita et al. 2004) and FIS2 (Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006) occur during female gametogenesis in the central cell by the antagonistic activity of MET1 methyltransferase and DME, a DNA glycosylase with a 5-methylcytosine excising activity (Choi et al. 2002; Gehring et al. 2006). DME is specifically expressed in the central cell preceding fertilization, erasing methylation marks set up by MET1 on MEA, FWA and FIS2 (Xiao et al. 2003; Kinoshita et al. 2004; Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). The DME-like genes might play similar roles by erasing methylation marks on imprinted genes in maize as well.

Until recently, the molecular mechanisms of imprinting in plants have most extensively been studied in Arabidopsis. These studies revealed two types of imprinted genes. MEA, for example, shows bi-allelic expression in many tissues except the endosperm, where the gene is maternally expressed. Other genes, like FWA and FIS, are silent and methylated in all tissues except the endosperm, suggesting that methylation is the default state. The Fie1 default state is also methylated and the maternal allele is activated due to its demethylation in the central cell prior to fertilization. We found that Fie1 is methylated across the extended ∼4 kb segment including the promoter and coding region up to exon 7. This is significantly deviating from imprinted genes in Arabidopsis, where methylation is directed to specific segments, for example, the tandem repeats in the promoter of FWA (Kinoshita et al. 2004, 2007) and a 200-bp upstream segment in FIS2, respectively (Jullien and Katz et al. 2006; Jullien and Kinoshita et al. 2006). There are no direct or inverted repeats, which could act as cis-acting methylation elements in the Fie1 gene. Recently a DMR (Differentially Methylated Region) was identified in Fie1 at the 5’ promoter region and in exon 1. However, in the central cell and the 6 DAP endosperm, DNA methylation was detected only in the promoter segment of the DMR but not in exon 1 and other downstream sequences (Gutierrez-Marcos et al. 2006). In contrast, we have detected that the paternal Fie1 allele is methylated over the same extended region including the promoter, exon 1 and exons 6–8 in the 14 DAP endosperm as it was in the pollen. The different methylation pattern of Fie1 at 6DAP (Gutierrez-Marcos et al. 2006) and 14 DAP endosperm (this report) may reflect the complex dynamics of demethylation and de novo methylation of the paternal Fie1 allele that occurs during endosperm development in maize. Our study of the imprinted Fie1 gene in maize provides new evidence for diversity and adds further complexity of imprinting mechanisms in plants.