Plant epigenetics: MEDEA's children take centre stage

It is increasingly recognised that epigenetic modifications — mitotically or meiotically heritable changes in gene function that do not result from alterations in DNA sequence — are important in development. In plants and animals, the Polycomb-group (Pc-G) genes mediate a wide variety of epigenetic phenomena, including X chromosome inactivation, genomic imprinting and transgene silencing [1]. They generally repress transcription of their target genes, and are thought to act at the level of chromatin structure to confer mitotically heritable repression. In Arabidopsis, three Pc-G genes, collectively termed the ‘FERTILISATION INDEPENDENT SEED’ development (FIS) genes, have attracted intense interest for their role in controlling seed development [2,3]. Progress in understanding how the FIS genes act has been hindered because their targets were unknown. Kohler and colleagues [4] now report an important advance, the identification of a target gene that is directly regulated by the FIS proteins. Plants reproduce sexually by producing seeds. The seed contains the diploid embryo and also a second zygotic tissue, the endosperm (Figure 1). The endosperm transmits nutrients and signals from the surrounding maternal tissue to the embryo, and it is ephemeral in Arabidopsis, being largely consumed by the embryo during seed maturation. The endosperm is economically important as in some plants, particularly the cereals, it is more persistent and forms the bulk of the mature seed. The three FIS genes — FIS2, MEDEA (MEA) and FERTILIZATION INDEPENDENT ENDOSPERM (FIE) — are all required maternally for seed viability. Thus, all seeds that inherit a mutant fis allele maternally abort, irrespective of the genotype of the paternal allele [2,3]. By contrast, seeds that inherit a wild-type maternal allele and a mutant paternal allele develop normally. One possible explanation for these parent-of-origin effects is that the FIS genes are imprinted, so that only the maternal alleles are active during endosperm and/or embryo development. The paternal alleles are silent and therefore can not rescue maternal mutant alleles. An alternative explanation is that the FIS gene products are required in the female gametophyte (Figure 1) before fertilisation for normal embryo and/or endosperm development. The expression of the FIS genes is consistent with both possibilities. All are expressed before fertilisation in the female gametophyte. After fertilisation, expression is exclusively from the maternal allele during early seed development, but later it becomes bi-allelic [5–8]. Imprinting may be most relevant for MEA, as the paternal allele is silenced more persistently than the paternal alleles of the other FIS genes, and genetic studies suggest that reactivation of the paternal MEA allele in seeds can rescue the mea phenotype [5,7]. Characterisation of the DEMETER (DME) gene has identified an important extra level in the regulatory hierarchy controlling seed development. dme mutations have similar parent-of-origin effects to fis alleles, so that only the maternal allele of DME is essential for seed development [9]. Unlike the FIS genes, DME is expressed before fertilisation in the female gametophyte, but not subsequently during seed development. DME appears to act at least partially through MEA, as it is required to activate the maternal MEA allele in the gametophyte and subsequently in the seed (Figure 2). DME encodes a DNA glycosylase, and a related protein has been shown to regulate gene expression by reducing DNA methylation at target gene promoters [10]. It is unlikely, however, that DME acts directly on methylation of MEA, because this was not altered in dme mutants [9]. To gain further insight into the regulatory pathway controlling seed development, Kohler et al. [4] conducted transcriptional profiling of mea and fie mutants. RNA samples were extracted from siliques containing very early stage mea and fie mutant seeds. This strategy minimized secondary effects, as mutant seeds at this developmental stage appear morphologically similar to wild type. One of the genes that was commonly derepressed in the mutants was found to encode a MADS box class transcription factor and named PHERES1 (PHE1) after the Greek myth in which Medea murdered her sons Pheres and Meidos. MADS box genes have also been implicated as the targets of several other plant Pc-G members [11,12]. PHE1 represents a distinct and evolutionary more ancient class of MADS box protein, however, and is the first of this class to be functionally characterised in plants. To confirm that PHE1 is regulated by the FIS genes, Kohler et al. [4] compared the spatial and temporal pattern of PHE1 transcription in wild type and fis class mutants using in situ hybridisation and PHE1 PROMOTOR::GUS reporter genes. PHE1 is not expressed in the female gametophyte, but is detectable 1–2 days after pollination in the embryo and endosperm in both fis mutants and wild type. Later in seed development, expression of PHE1 becomes restricted to the chalazal region of the endosperm in wild type, whereas in Current Biology, Vol. 13, R638–R640, August 19, 2003, ©2003 Elsevier Science Ltd. All rights reserved. DOI 10.1016/S0960-9822(03)00569-4