- American Society of Plant Biologists
The genetic and developmental processes that control grain and seed development have become a fertile area of research in recent years. Insights have been obtained on how two haploid genomes come together to form the diploid zygote and triploid endosperm inside the maternal organs that sustain them (Chaudhury et al., 2001). The female gametophyte controls embryo and endosperm development, but only a small number of genes have been identified in which seed viability depends on the genotype of the maternal allele. A recently cloned gene involved in seed development celebrates Demeter, the Greek goddess of fertility and giver of grains (who was known in Rome as Ceres) (Figure 1) . Choi et al. (2002) describe the isolation and characterization of DEMETER (DME), a gene whose maternal allele is essential for seed viability. DME joins the three FIS (fertilization-independent seed) Polycomb group genes (Ohad et al., 1996; Chaudhury et al., 1997; Grossniklaus et al., 2001) as genes whose wild-type maternal alleles are necessary for the proper development of the female gametophyte and seed development. Genomic imprinting is an epigenetic mechanism that determines the expression or repression of genes according to parental origin. DME provides new understanding by showing how a previously isolated gene, MEDEA (MEA), might be imprinted (for a review on imprinting, see Baroux et al., 2002).
Demeter of Cnidos, Goddess of Fertility and Mother of Corn. © Photo © Maicar Förlag-GML (http://homepage.mac.com/cparada/GML/).
Three FIS genes (MEA/FIS1, FIS2, and FIE/FIS3) were identified in a genetic screen designed to isolate mutants that showed autonomous seed development and thus decouple seed development from fertilization (Peacock et al., 1995; Ohad et al., 1996; Chaudhury et al., 1997). Apart from their autonomous phenotype, these gametophytic mutants showed a parent-of-origin effect. When the mutation was carried by the male parent, the developing seeds were normal, but when the mutation was carried by the female parent, seed development was abnormal even with a wild-type pollen parent (Ohad et al., 1996; Chaudhury et al., 1997). In an independent screen, a tagged mutant with a phenotype similar to that of the fis mutants was found and named after the goddess Medea (Grossniklaus et al., 1998). MEA is identical to FIS1 and was shown to be an imprinted gene. At the molecular level, the paternally derived gene was not transcribed in the developing endosperm, whereas the maternally derived gene was expressed (Grossniklaus et al., 1998; Kinoshita et al., 1999). The FIS2 gene also was shown to be imprinted using a reporter gene to analyze expression (Luo et al., 2000).
Interest in the relationships between seed development and parent-of-origin effects has been stimulated with the characterization of DME, which has been shown to be a direct regulator of MEA (Choi et al., 2002). Three mutant alleles of DME were isolated by screening diverse T-DNA Arabidopsis libraries for 50% seed abortion. dme-1 is a weak allele and probably produces a slightly truncated polypeptide. In seeds that inherit a maternal copy of dme-1, the embryo aborts but the endosperm is enlarged. When inherited paternally, dme-1 has no effect on seed viability. A paternally inherited, wild-type copy of DME is unable to rescue the development of seeds carrying a maternal dme-1 allele, showing that seed viability is dependent entirely on a wild-type maternal DME allele. Such parent-of-origin effects on seed viability are similar to those reported previously for MEA, FIS2, and FIE (reviewed by Chaudhury et al., 2001). Homozygous dme-1 plants develop normally except that most seeds abort. In addition, abnormalities in flower development and leaf and stem morphology were observed occasionally, but these were not transmitted to the next generation. This finding suggests a role for DME in floral development in Arabidopsis.
Using promoter fusions to reporter genes, Choi et al. (2002) found that before fertilization, DME is expressed in the two polar nuclei and the two synergid cells of the female gametophyte. No expression was reported in other reproductive tissues, even those that appear to be somewhat affected in the homozygous dme-1 plant. The expression of DME ceases rapidly after fertilization, and expression of a DME reporter gene was not detected before the first division of the primary endosperm nucleus.
When RNA levels of MEA, FIS2, or FIE were examined in the presence of dme-1, only MEA expression was disrupted, suggesting that the expression of MEA is dependent on DME function. Before fertilization, MEA is expressed in the two polar nuclei before their fusion to form the double-haploid nucleus of the central cell in the embryo sac. MEA continues to be expressed after fertilization in the endosperm nuclei formed by the division of the now triploid central cell nucleus. It also was shown that the expression of MEA is dependent on the presence of a functional copy of DME in the female gametophyte. In ovules that inherited both the dme-1 mutation and a MEA reporter construct, there was no expression of the MEA reporter. From these and other results, Choi et al. (2002) concluded that the expression of DME in the central cell before fertilization is necessary and sufficient for the transcription of MEA in the endosperm after fertilization. The parent-of-origin effects of dme mutants on seed viability can be explained simply by a model in which the expression of the maternal MEA allele is controlled positively by DME expression in the central cell before fertilization.
Choi et al. (2002) also present molecular data to support this model. Although MEA and DME are not normally expressed in the leaf, MEA expression can be detected in leaves in which DME is expressed ectopically under the control of the Cauliflower mosaic virus (CaMV) promoter. When such plants were pollinated with wild-type pollen, RNA from both maternal and paternal copies of MEA was detected in the endosperm. The paternal copy of MEA was activated after fertilization, because the CaMV promoter drives DME expression in the endosperm. These observations are consistent with DME being a positive regulator of MEA and with the hypothesis that the paternal copy of MEA normally is silent in the endosperm as a result of the restricted expression pattern of DME. Thus, the mechanism for the imprinting of MEA is not the silencing of the paternal copy but the activation of the maternal copy.
The deduced amino acid sequence suggests that DME is a large, monofunctional DNA glycosylase domain protein of 1729 amino acids. DNA glycosylases are DNA repair proteins that excise mismatched, modified, or damaged bases by cleaving the N-glycosidic bond between the base and the sugar-phosphate backbone of the DNA. Base excision results in an abasic site that is mutagenic and must be repaired. Repair is initiated with strand cleavage 5′ to the abasic site to produce a 3′ hydroxyl that is recognized by a specialized DNA repair polymerase. Unlike bifunctional DNA glycosylases, monofunc-tional DNA glycosylases such as DME do not have DNA-nicking activity. Strand cleavage is performed by an apurinic or apyrimidinic endonuclease, which creates a single-strand nick at abasic sites. Repair is completed by sealing the nick with a DNA ligase. Choi et al. (2002) suggest that nicks at specific sites in DNA might constitute an essential feature in the control of chromatin structure and gene expression.
The authors also present evidence to support the hypothesis that DME activity leads to a single-strand nick in the vicinity of the MEA promoter. Using a sensitive PCR-based assay for nicks produced in vivo, it was shown that nicking occurs in the sense strand within 2 kb of the start of MEA transcription, but only if the CaMV: DME transgene also is present in the genome. There was no detectable nicking within 14 kb of the sense strand of MEA in the absence of the CaMV:DME transgene. The bands produced by the nicking assay were of different sizes, suggesting that DME produces nicks in the MEA promoter at many different locations (Choi et al., 2002).
In the absence of pollination, the endosperm develops autonomously in all alleles of the three FIS genes that have been examined, although the penetrance is variable (Ohad et al., 1996; Chaudhury et al., 1997). Because the maternal expression of DME is essential for MEA expression, dme mutants also should show autonomous endosperm development. However, this question and several others about the mechanism of DME function remain unanswered. For example, how does a single-stranded nick induced by DME in the MEA promoter result in activation? How does DME recognize the MEA promoter? Why is the transcription of the maternal DME gene turned off soon after fertilization, resulting in the paternal copy of MEA remaining silent in triploid endosperm nuclei?
It remains to be determined what role, if any, DME plays in seed development other than controlling MEA expression. The floral and vegetative pleiotropic phenotypes observed in dme-1 homozygous mutants, and the presence of DME RNA in stems, suggest that DME functions more broadly in plant development. But as shown by Choi et al. (2002), DME does not affect the expression of the other two FIS genes that also are imprinted. Three other Arabidopsis proteins are closely related to DME. An intriguing possibility is that the DME-like genes control the imprinting of other genes.
In both the fis1 and fis2 mutants, the maternal defect can be rescued by pollen that carries a nonfunctional fis1 or fis2 allele, provided that the DNA of the pollen donor also is hypomethylated (Luo et al., 2000). This fact indicates that, at least in certain circumstances, rescue of the fis maternal defect can occur without the paternal activation of MEA or FIS2. The work of Choi et al. (2002) suggests that another mechanism for the rescue of the mea maternal defect is through the ectopic expression of DME in the pollen donor, resulting in the premature activation of the paternal MEA gene.
The characterization of DME has initiated a new area of investigation by linking a presumptive DNA metabolism enzyme to a mechanism of imprinting. Future work may elucidate the mechanism by which FIS2, FIE, and other genes are imprinted. Under the gaze of many goddesses, this fertile area of investigation surely will bring many new surprises.
Acknowledgments
We thank Jean Broadhvest for helpful comments.
