Os05g0445900

Please input one-sentence summary here.

Annotated Information

Function

Cytosine DNA methylation, which occurs in the CG, CHG(H = A, C or T) and CHH sequence contexts, is an epigenetic modification in plants. As well, there are some proteins coded for DNA demethylation. The genes and the corresponding encoded enzymes that mediate DNA methylation and demethylation have been characterized mainly in Arabidopsis. The Arabidopsis enzymes that mediate 5-methylcytosine (5-meC) DNA demethylation are DEMETER (DME), REPRESSOR OFSILENCING 1 (ROS1),DEMETER-LIKE 2 (DML2) and DEMETER-LIKE 3 (DML3). Phylogenetic analysis revealed that the rice (Oryza sativa)genome encodes six putative bi-functional DNA glycosylases that mediate cytosine DNA demethylation: four ROS1orthologs and twoDML3orthologs, but no DME orthologs.

1.It has been demonstrated that rice ROS1a protein is a bi-functional DNA glycosylase/lyase for 5-meC DNA demethylation, although biochemical characterization of the ROS1a enzyme remains to be performed to confirm this.Rice ROS1a is toxic to E. coli containing 5-meC in its genome. When ROS1a cDNA was expressed under the control of an IPTG (isopropyl-β-d-thiogalactopyranoside)-inducible promoter, ROS1a was toxic to an E. coli dcm+ strain containing 5-meC in an IPTG-dependent manner, and was less toxic to a dcm− mutant without 5-meC.

2. ROS1a is the most abundantly expressed gene in tissues. It was recently reported that null mutants of ROS1c, which encodes a 5-meC DNA glycosylase/lyase, show no effects on transmission of the null alleles and produce a small portion of wrinkled seeds.And the null mutation, ros1a-GUS1, was hardly ever transmitted to progeny.Even in the presence of the wild-type paternal ROS1a allele, the maternal nullros1a-GUS1allele caused failure of early stage endosperm development, indicating non-equivalent contribution of maternal and paternal ROS1a to endosperm development.

Figure.  Histochemical GUS staining patterns in ROS1a/ros1a-GUS1 plants. (a) Longitudinal section through basal shoot meristem region of mature plants. Insets are magnified views of shoot apical and lateral (upper) and inflorescence (lower) meristems. (b) Flowers. (c) Anthers. Approximately half of the pollen present showed GUS staining (see also Figure S7b). The inset is a magnified view. (d) Pistils. Insets are ovules before pollination exposed by removing the carpel. The upper and lower insets show a GUS-stained ovule and its sibling, which displayed no GUS staining, respectively. Before pollination, ovules containing gametes with the wild-type ROS1a allele were morphologically indistinguishable from those with ros1a-GUS1 allele. (e) Schematic representation of a rice female gametophyte enclosed by the maternal tissues of the ovule: ovary tissue (light brown), integument (dark brown), and nucellus (gray). The haploid female gametophyte consists of the egg apparatus (green), including the egg cell and two synergids, the central cell (white), which contains large vacuoles and thin lines of cytoplasm (pink), and antipodals (blue). (f,g) Differential interference contrast micrographs of GUS-stained female gametophytes of ROS1a/ros1a-GUS1 plants. To make the borders of different tissues clear, white lines are drawn on the image in (f), which corresponds to the area enclosed by the red line in (e). (h) Differential interference contrast micrograph of a sibling female gametophyte displaying no GUS staining. Arrows indicate GUS-stained meristems (a), pollen (b) and the probable egg apparatus (d). In (b), the arrowhead indicates the GUS-stained lodicule. AP, antipodals; PN, polar nuclei. The images in (g,h) correspond to the area enclosed by the blue line in (e). Scale bars = 1 mm [a,b, inset in (a)], 500 μm [c,d], 100 μm [inset in (d)] and 50 μm [inset in (c) and f-h].

3.Rare transmission of the ros1a-GUS1 allele to progeny.To obtain homozygous knock-in plants, we self-pollinated the isolated T0 plants. Of the 250 fully grown seeds selected, 232 germinated, and the resulting T1 seedlings were genotyped by PCR analysis. Surprisingly, only two were ROS1a/ros1a-GUS1 plants, and the remaining 230 seedlings were homozygous for the wild-type allele (ROS1a/ROS1a) (Table 1). Moreover, the ros1a-GUS1 allele in the two heterozygous T1 plants was not transmitted to progeny, confirming that the ros1a-GUS1 allele is rarely transmittable to progeny. Close inspection revealed that the T0 panicles comprised three types of grain: empty grains with only infertile flower remnants , grains with normal-shaped seeds , and grains with deformed seeds containing severely under-developed and non-starch-producing endosperm . Approximately equal numbers of normal-shaped and deformed seeds were observed (54:56, 1:1, χ2 = 0.036, P > 0.8; Table 2). Embryos in the deformed seeds always displayed GUS-positive staining, but none of the normal-shaped seeds showed GUS-positive patterns, indicating that the ros1a-GUS1 allele co-segregated with the deformed seed phenotype and that the normal-shaped seeds were ROS1a/ROS1a (Table 2 and Figure 5b,d). In the analysis of the segregation ratio of ros1a-GUS1 in seedlings (Table 1), deformed seeds bearing the ros1a-GUS1 allele were probably excluded from the analysis because only fully grown seeds were used.

Expression

Rice contains four ROS1 orthologs and two DML3 orthologs tentatively named ROS1a–d and DML3a and DML3b, that contain characteristic DNA glycosylase domains flanked by conserved domains of unknown functions. Of these, ROS1ais the longest gene, comprising 17 exons that encode a protein with 1952 amino acids, and 5¢ and 3¢ UTRs of 73 and 607 bp, respectively.RT-PCR analysis revealed thatROS1awas expressed in all vegetative and reproductive tissues tested . Quantitative RT-PCR analysis revealed that ROS1ais the most extensively expressed gene among the four genes(ROS1a, ROS1c, ROS1d and DML3a) expressed in five selected tissues examined, including anthers and pistils,whereas ROS1b and DML3bare scarcely expressed in these tissues. Interestingly, moderate levels of transcripts for ROS1c, ROS1dandDML3awere detected in pistils and immature seeds 2 days after pollination.

Evolution

Phylogenetic analysis revealed that the rice (Oryza sativa) genome encodes six putative bi-functional DNA glycosylases that mediate cytosine DNA demethylation: four ROS1 orthologs and two DML3 orthologs, but no DME orthologs (Zemach et al., 2010). Rice endosperm DNA is hypomethylated in all sequence contexts, implying that hypomethylation in rice endosperm relies on some of these DNA glycosylases or alternative biochemical mechanisms (Zemach et al., 2010). In this study, to characterize the function of one of the four rice ROS1 orthologs, tentatively named ROS1a (LOC_Os01g11900.1), that resides on chromosome 1, we used homologous recombination-promoted knock-in targeting with positive/negative selection (Yamauchi et al., 2009) to obtain a mutant that disrupts ROS1a by fusion of its endogenous promoter with the GUS reporter gene encoding β-glucuronidase. We reproducibly obtained T0 plants with the null knock-in allele, ros1a-GUS1, in the heterozygous condition, and detected GUS expression in the T0 plants in the shoot apical, lateral and inflorescence meristems, as well as in both female and male gametophytes before fertilization. The ros1a-GUS1 allele was hardly transmitted to the next generation; neither the maternal nor the paternal ros1a-GUS1 allele was virtually found in the progeny. The results indicate that ROS1a, presumably through DNA demethylation, is indispensable in both gametophytes, and that the null allele of ROS1a is difficult to isolate by conventional mutagenesis techniques, in which mutants are usually obtained as segregants in the progeny population.

summary

DME is preferentially expressed in the homodiploid central cell of the female gametophyte before fertilization, and promotes maternal allele-specific global hypomethylation, leading to maternal allele-specific expression of imprinted genes in the endosperm, including the Polycomb Repressive Complex 2 (PRC2) genes MEDEA (MEA) and FERTILIZATION INDEPENDENT SEED 2 , and to maternal allele-specific expression of transposons in the endosperm. Expression of these imprinted genes is prerequisite for normal endosperm development. mea and fis2 mutants exhibited characteristic maternal defects in the endosperm similar to the dme mutant, because PRC2 represses autonomous growth of unfertilized endosperm and overgrowth of fertilized endosperm . In rice, GUS staining was detected in the central cells of ROS1a/ros1a-GUS1 plants , indicating that ROS1a is expressed in the central cell of the female gametophyte. Although many maternally expressed imprinted genes were identified in the rice endosperm, very few were in common with those in Arabidopsis . Expression of FERTILIZATION INDEPENDENT ENDOSPERM 1 (FIE1) (LOC_Os08g04290), one such imprinted gene encoding a component of PRC2 in rice, appears to correlate with DNA demethylation at its 5′ region, but an fie1 disruptant showed no defects in endosperm development . The results imply that certain roles of PRC2 in rice endosperm development differ from those of Arabidopsis. Although we could not demonstrate ROS1a-promoted DNA demethylation, ROS1a is postulated to activate the imprinted genes through DNA demethylation in order to facilitate normal endosperm development. GUS staining was also detected in the antipodals and the egg apparatus comprising two synergids and an egg cell whose genetic information is transmittable to progeny (Figure 4f,g). The molecular mechanisms of ROS1a function in these cells remain to be elucidated.

Although DME in Arabidopsis is predominantly expressed in the central cell of the female gametophyte, and is only marginally expressed to demethylate some imprinted genes and transposons in the vegetative cell of the male gametophyte , comparable amounts of ROS1a were expressed in both the female and male gametophytes , and the paternal ros1a-GUS1 allele could not be transmitted to progeny . Although the molecular mechanisms for the inability to transmit the paternal ros1a-GUS1 allele are unknown, ros1a-GUS1 pollen germinated less efficiently than did ROS1a pollen. Reduced germination and transmission of dme pollen were also observed in certain Arabidopsis ecotypes (Col-gl and Col-0) . The vegetative nucleus in pollen supports the sperm cell before fertilization and shows reduced DNA methylation in Arabidopsis, resulting in transient reactivation of diverse transposons and generation of small interfering RNAs that may ensure the genomic integrity of sperm cells by silencing transposons in the sperm cells . Whether similar DNA demethylation occurs in the vegetative nucleus in rice pollen, and whether the ROS1a protein promotes such demethylation, remain to be determined. If such DNA demethylation occurs in the vegetative nucleus, it will be interesting to determine whether such demethylation directly correlates with the intransmittable nature of the paternal ros1a-GUS1 allele .

Quantitative RT-PCR analysis revealed that ROS1a, ROS1c, ROS1d and DML3a are expressed in five selected tissues tested, including anthers and pistils, and ROS1a is the most abundantly expressed gene in these tissues . The same analysis showed that ROS1c, ROS1d and DML3a are moderately expressed in both pistils and immature seeds 2 days after pollination. It was recently reported that null mutants of ROS1c, which encodes a 5-meC DNA glycosylase/lyase, show no effects on transmission of the null alleles and produce a small portion of wrinkled seeds . Their results are in sharp contrast to those for the null allele of ROS1a described here, and indicate that moderate expression of ROS1c, ROS1d and DML3a in pistils could not compensate for the loss of ROS1a. It would be interesting to determine whether ROS1a and any of the other three genes are expressed simultaneously in the same cells in the pistil. If ROS1a and ROS1c are expressed in the same cells, it would then be interesting to determine whether the different transmittability of their null alleles may be explained simply by their expression levels or whether it is due to intrinsic differences in their molecular functions.

The null ros1a-GUS1 allele was not transmitted to progeny . In conventional mutagenesis procedures, mutants are usually identified and isolated as segregants in the progeny population. Such procedures would not have enabled isolation of the ros1a-GUS1 allele. In contrast, our homologous recombination-promoted gene targeting procedure allowed T0 transgenic plants to be reproducibly obtained with the exactly anticipated structure of the targeted gene in the heterozygous condition. Moreover, the knock-in targeting strategy allowed detection of spatio-temporal ROS1a expression and tracking of the transmitted ros1a allele. However, because GUS staining was necessary to detect ROS1a expression, it was difficult to analyze in vivo and/or to isolate viable tissues that express GUS for detailed examination. Utilization of other markers such as green fluorescent protein, in combination with other techniques including laser micro-dissection and/or fluorescence-activated cell sorting, would facilitate further characterization of ROS1a function in endosperm development and pollen gametophytic transmission.

Labs working on this gene

1.National Institute for Basic Biology, Okazaki 444-8585, Japan

2.Graduate School of Science and Technology, Niigata University, Niigata 950-2181, Japan

3.Graduate School of Nutritional and Environmental Sciences, Graduate School of Pharmaceutical Sciences, and Global Center of Excellence Program, University of Shizuoka, Shizuoka 422-8526, Japan

References

1.A. Zemach et al., Local DNA hypomethylation activates genes in rice endosperm. Proc Natl Acad Sci U S A 107, 18729 (Oct 26, 2010).

2.A. Ono et al., A null mutation of ROS1a for DNA demethylation in rice is not transmittable to progeny. Plant J 71, 564 (Aug, 2012).