Polycomb group-dependent imprinting of the actin regulator AtFH5regulates morphogenesis in Arabidopsis thaliana

Parental genomic imprinting - the preferential expression of one of the two parental alleles of a gene - has been described in flowering plants and mammals (Feil and Berger,2007). In plants, distinct fertilizations involving two pairs of male and female gametes produce the plant embryo and the endosperm. The endosperm controls the supply of maternal nutrients to the embryo(Berger et al., 2008). The endosperm is the only tissue for which imprinted genes have been identified in plants (Kinoshita et al.,2008). In maize, only maternally expressed imprinted genes have been identified (Gutierrez-Marcos et al.,2004; Gutierrez-Marcos et al.,2006; Hermon et al.,2007; Kermicle,1970). In Arabidopsis, among the five imprinted genes identified, PHERES1 (PHE1) is paternally expressed(Makarevich et al., 2006),whereas MEDEA (MEA)(Kinoshita et al., 1999), FERTILIZATION INDEPENDENT SEED 2 (FIS2)(Jullien et al., 2006b), FLOWERING WAGENINGEN (FWA)(Kinoshita et al., 2004) and MATERNALLY EXPRESSED PAB-C TERMINAL (MPC)(Tiwari et al., 2008) are expressed maternally. DNA methylation of cis-elements in the 5′control region is responsible for the silencing of MPC, FWA and FIS2. By contrast, methylation of histone H3 lysine 27 (H3K27)residues is essential for silencing the non-expressed allele of MEAand PHE1 (Gehring et al.,2006; Jullien et al.,2006a; Makarevich et al.,2006), although the imprint of PHE1 further requires DNA methylation at 3′ sites (Makarevich et al., 2008). MEA and FIS2 are core members of the endosperm-specific FERTILIZATION INDEPENDENT SEED (FIS) Polycomb group(PcG) complex that also includes FERTILIZATION INDEPENDENT ENDOSPERM(FIE) and MULTICOPY SUPPRESSOR OF IRA1 (MSI1)(Guitton and Berger, 2005).

The wild-type endosperm posterior pole (also called the chalazal pole) is distinguished from the peripheral and anterior (micropylar) domains of the endosperm by a multinucleate structure termed the cyst(Boisnard-Lorig et al., 2001; Brown et al., 1999; Scott et al., 1998). The cyst develops from the migration of nuclei from the peripheral endosperm(Guitton et al., 2004). Early endosperm syncytial development ends when cellularization partitions the syncytium into mononucleate cells, but cellularization does not occur in the posterior pole (Brown et al.,1999; Sorensen et al.,2002). The endosperm of fis mutants is characterized by multiple defects, including enhanced proliferation, much enlarged posterior structures and the absence of cellularization(Guitton et al., 2004; Kiyosue et al., 1999; Kohler et al., 2003; Luo et al., 1999). This pleiotropic phenotype might be the consequence of the maintenance of a juvenile developmental program (Ingouff et al., 2005b). Characterized targets of the FIS PcG complex are transcriptional regulators (Makarevich et al., 2006). The pathways downstream of this transcriptional regulation are unknown and the targets, the functions of which explain the fis mutant phenotype, have not been identified.

We have previously characterized ARABIDOPSIS FORMIN HOMOLOGUE 5(AtFH5), which is expressed in the endosperm(Ingouff et al., 2005a). Formins are actin-nucleating agents that are involved in cell polarity and cytokinesis throughout eukaryotes (Kovar,2006; Wallar and Alberts,2003). Insertional mutants, atfh5-1 and atfh5-2,are defective in endosperm posterior pole structures(Ingouff et al., 2005a). We report that AtFH5 is a new imprinted gene. The FIS PcG activity restricts expression of AtFH5 to the maternal allele and in the posterior endosperm. We identify a cis-element that is targeted by PcG activity and responsible for AtFH5 imprinting. We further study the genetic interaction between FIS genes and AtFH5 and assess the relationships between PcG-mediated regulation of AtFH5 and endosperm morphogenesis.

Seeds of all ecotypes were obtained from the ArabidopsisBiological Resource Center(www.Arabidopsis.org). atfh5 mutants (Ingouff et al.,2005a) and the fis PcG mutants mea-6 and fis2-6 (Boisnard-Lorig et al.,2001; Guitton et al.,2004) were described previously.

Seeds were imaged using DIC optics and Feulgen staining as reported previously (Boisnard-Lorig et al.,2001; Guitton et al.,2004), using a Leica DM600 microscope for DIC optics and a Zeiss LSM 510 META system (Zeiss, Jena, Germany; 40× Plan objective).

RNA extraction and treatments were as described previously(Ingouff et al., 2005a). For quantitative PCR, tissues were collected from dissected seedlings 10 days after planting on MS-Agar (Sigma). SYBR Green Master Mix was used according to the manufacturer's instructions in a 7900HT Fast Real-Time PCR apparatus(Applied Biosystems, Carlsbad, CA, USA)

Extraction of chromatin from seeds 4-5 days after pollination was conducted using a ChIP Assay Kit (Upstate, Lake Placid, NY, USA) according to the manufacturer's instructions, using antibodies against di- and trimethylated histone H3 (Lys27) (Upstate, Cornell, Ithaca, NY, USA). To detect pull-down of the AtFH5 promoter, semi-quantitative PCR reactions were employed,with 25 cycles at an annealing temperature of 58°C (55°C for ACT11) and using Illustra Taq polymerase (GE Healthcare, Chalfont St Giles, UK).

Isolated H2B-mRFP PCR product including a 35S terminator was fused to the indicated AtFH5 promoter regions and 5′UTR by marker-fusion PCR (Kitazono et al.,2002) then recombined with Alligator2(Bensmihen et al., 2004) by Gateway cloning (Invitrogen, Carlsbad, CA, USA). Alligator2 was first digested with HindIII and EcoRV to remove the 2×35S promoter,ATG and 3×HA tag.

We fused the endogenous Arabidopsis thaliana AtFH5 promoter to a construct encoding the nuclear reporter histone H2B-red fluorescent protein(pAtFH5::H2B-mRFP) to study the expression of AtFH5 in the gametes prior to fertilization. We could not detect expression of pAtFH5::H2B-mRFP in the female gametes(Fig. 1A). pAtFH5::H2B-mRFP expression in the pollen was limited to the vegetative nucleus, and was not observed in the two male gametes(Fig. 1B). From these observations, we concluded that AtFH5 is not expressed prior to fertilization in either the male or female gametes.

AtFH5 is expressed only in the endosperm during seed development and not in the embryo nor in the surrounding maternal seed integuments(Ingouff et al., 2005a). Hence, we were able to study the parental origin of the expression of AtFH5 in endosperm by RT-PCR using isolated seeds. We identified a sequence polymorphism between the Landsberg erecta (Ler) and C24 Arabidopsis accessions (Fig. 1C). If AtFH5 were expressed from both the maternal and paternal alleles, we would have detected AtFH5 mRNA from either accession. However, we detected only transcripts from the maternal allele of AtFH5 in crosses between Ler and C24 parents(Fig. 1C). Since AtFH5was not expressed in the central cell prior to fertilization, we conclude that the AtFH5 transcripts we detected do not originate from the central cell and that AtFH5 is a maternally expressed imprinted gene.

Expression of the paternal allele of AtFH5 was regained by maternal loss of PcG function in the endosperm(Fig. 1C). This result suggested that the maternal FIS PcG complex maintains silencing of the paternal allele of AtFH5. Other PcG imprinted targets, MEAand PHE1, are suppressed by PcG complexes active in vegetative tissues (Jullien et al.,2006a; Katz et al.,2004; Makarevich et al.,2006; Schubert et al.,2006). In accordance, compromised PcG activity caused by co-suppression of FIE or by loss-of-function of the PcG gene SWINGER (SWN) caused a significant increase of AtFH5 expression. In the sporophyte, AtFH5 expression was detectable in the roots, but not in above-ground tissues(Fig. 2B; see Fig. S1 in the supplementary material). Loss of FIE or SWN resulted in an up-to-fourfold increase in AtFH5 expression in the shoot, as compared with that in the root, where the PcG genes were not repressed(Fig. 2C).

To assess whether the PcG-dependent histone methylation on lysine residue 27 was deposited on the AtFH5 promoter, we performed a chromatin immunoprecipitation (ChIP) analysis to detect di- and trimethylation at H3K27(H3K27me2,3). H3K27 methylation was detected within a 342 bp region of the AtFH5 promoter, but not in the coding region(Fig. 2C). From the above results, we concluded that PcG complexes are likely to deposit marks on the promoter of AtFH5 that are involved in the repression of its transcription in vegetative tissues and gametes.

To assess the function of the region marked by H3K27me2,3 in the AtFH5 promoter we constructed transgenic plant lines expressing a series of H2B-mRFP reporters driven by sequences derived from the AtFH5 promoter (Fig. 2A). We identified the region CUT4, which includes a 417 bp region around the site marked by H3K27me2,3 enrichment(Fig. 2D), that was sufficient to recapitulate the AtFH5 expression pattern(Fig. 2E-H). In order to test whether this element was sufficient to confer imprinted expression, we used the CUT4 transgenic reporter lines as male or female in crosses with wild-type plants (Fig. 3). H2B-mRFP was expressed maternally when CUT4::H2B-mRFP transgenic lines were pollinated with wild-type plants(Fig. 3A). When the same reporter lines were used as pollen donors, however, no RFP was detected in the endosperm (Fig. 3B). The expression of CUT4::H2B-mRFP thus recapitulates the imprinted expression of the endogenous AtFH5 locus.

Loss of PcG activity in mea and fis2 mutants caused paternal expression of CUT4::H2B-mRFP in the endosperm(Fig. 3C,D). Our results thus indicate that the AtFH5 promoter alone is capable of directing FIS PcG-dependent imprinted expression of a reporter gene. We conclude that PcG-mediated silencing relies on sequence elements in the AtFH5promoter. In Drosophila, PcG activity is targeted by cis-acting Polycomb response elements (PREs) (Chan et al., 1994; Ringrose et al.,2003). The limited dissection of the AtFH5 promoter suggests that it is likely to contain a PRE.

In addition, whereas CUT4::H2B-mRFP expression is confined to the posterior pole in the wild type (Fig. 2F and Fig. 3A),the reduction of PcG activity caused ectopic expression of maternally and paternally contributed CUT4::H2B-mRFP in the anterior and peripheral endosperm (Fig. 3C,D). The MEA and FIS2 components of the PcG complex that is active in endosperm are expressed throughout endosperm until 2 days after fertilization,when their expression becomes confined to the posterior pole(Luo et al., 2000). Our results suggest that AtFH5 expression is repressed by PcG activity in the anterior and peripheral endosperm. In the posterior, MEA, FIS2and AtFH5 are co-expressed, suggesting that a repressor of PcG activity might be expressed at this location, resulting in AtFH5expression.

We studied the genetic interaction between FIS PcG regulation and AtFH5 to uncover the role played by AtFH5 in the complex endosperm phenotype caused by FIS gene maternal inheritance. We observed the endosperm phenotype in seeds produced by self-fertilization in mea-6/mea-6; AtFH5/atfh5-1 plants. In comparison to the wild type (Fig. 4A), mea/mea seeds were larger, showed uncellularized endosperm, larger chalazal cysts and ectopic cysts in the peripheral endosperm (Fig. 4B)(n>500). In atfh5/atfh5 seeds, the posterior pole is missing, as described previously(Ingouff et al., 2005a)(Fig. 4C) (n>500). Since loss of FIS activity in this background results in the ectopic expression of AtFH5 from either parental allele, we expected to observe a distinct double-mutant phenotype in those 25% of seeds that are homozygous for the atfh5 mutation. We found that 26±3% of seeds produced by mea-6/mea-6; AtFH5/atfh5-1 plants (n=303 seeds) and 23±4%of seeds produced by mea-6/mea-6; AtFH5/atfh5-2 plants (n=319 seeds) showed an absence of cellularization and the overgrowth of endosperm typical of the fis mutant phenotype, but no large and ectopic cysts, which are also characteristic of the fis mutant phenotype(Fig. 4D) (n=303). The absence of posterior cyst in a fis mutant background correlated with a 25% increase in seed lethality in mea-6/mea-6; AtFH5/atfh5-1 and mea-6/mea-6; AtFH5/atfh5-2 plants(Fig. 4E,F). As a result, we could not obtain double-homozygous mea/mea; atfh5/atfh5 plants. These observations support the hypothesis that the enlargement of the chalazal cyst observed in the mea endosperm specifically depends on the presence of AtFH5. AtFH5 is required for the construction of the posterior endosperm and mea causes ectopic expression of AtFH5 outside of the posterior endosperm. Presumably, ectopic actin organization by AtFH5delocalizes the recruitment of nuclei from the peripheral endosperm to the posterior pole and causes ectopic cyst formation in fis mutant endosperm. We thus conclude that AtFH5 control of endosperm posterior pole morphogenesis is directly regulated by PcG.

We have identified AtFH5 as a new imprinted gene in Arabidopsis. However, the imprinted status of AtFH5 is not prefigured by a sex-specific pattern of expression in the gametes. This implies that specific activation of the maternal allele during female gametogenesis is not required for AtFH5 imprinting, in contrast to other maternally expressed imprinted genes studied in Arabidopsis. The imprinted status is thus defined by silencing of the paternal allele followed by zygotic activation of the maternal copy. Similarly, most imprinted genes in mammals are not expressed during gametogenesis when the sex-dependent silencing occurs, and their imprinted status is revealed only later in specific tissues (Reik et al.,2001). AtFH5 is silenced by PcG activity in vegetative tissues and in endosperm. It is likely that the repressive H3K27me2,3 marks are removed during female gametogenesis and that some transcriptional mechanism enables AtFH5 expression only after fertilization in endosperm.

In animals and plants the genes marked by PcG complexes and associated with developmental regulation encode transcription factors that are important to establish patterning and cell fate(Ringrose and Paro, 2007; Schubert et al., 2005). Here,we show that PcG complexes may also directly control cytoskeletal genes required for proper morphogenesis. In support of this hypothesis, the Drosophila formin Diaphanous(Schwartz et al., 2006) and the mammalian diaphanous homologs (Bracken et al., 2006) were also identified as potential direct PcG targets in genome-wide profiling experiments. It is thus conceivable that a developmental mechanism has been conserved or selected independently during the evolution of multi-cellular eukaryotes, whereby PcG complexes regulate the transcription of genes that encode both transcription factors and structural molecules in order to create a patterned cellular context for physical processes involving cell migration and morphogenesis.