- Email: [email protected]
Gene regulation: Better late than never?
bb10j06.qxd
10/5/00
2:00 pm
Page R365
Dispatch
R365
Gene regulation: Better late than never? Hong Ma
A new study has found that, in Arabidopsis, the paternal copies of many genes are delayed in expression during early seed development. The distribution of the genes and nature of their products suggest that this delayed expression of paternal alleles may be a global phenomenon. Address: Department of Biology and the Life Sciences Consortium, 504 Wartik Laboratory, Pennsylvania State University, Pennsylvania 16802, USA. E-mail: [email protected] Current Biology 2000, 10:R365–R368 0960-9822/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.
It is well established that, in animals, early embryo development requires maternal information stored in the unfertilized egg [1]. Zygotic gene expression is generally found to be activated at some specific time during early animal embryo development (for example, see [2,3]), often associated with a specific transition in global properties such as the length of the cell cycle. In contrast, it has long been thought that embryo development in plants requires little stored maternal information [4]. This view is supported by the well-known observation that plant somatic cells can readily form functional somatic embryos [5]. Despite this difference between animals and plants, recent work indicates that there are also important differences between the contributions of the maternal and paternal genomes during early plant development — in particular, VielleCalzada et al. [6] have found that the expression of paternal alleles is frequently delayed during seed development in the model plant Arabidopsis thaliana. Plants have alternating diploid and haploid generations. In the diploid sporophytic organism, cells undergo meiosis to produce spores, which develop into multicellular haploid gametophytes. The presence of a haploid generation provides a mechanism for the efficient removal of deleterious mutations. Cells in the gametophytes then differentiate into male and female gametes. In flowering plants, the male and female gametophytes are highly reduced in size and develop within the reproductive organs of the diploid sporophyte. The mature female gametophyte — the embryo sac — of flowering plants has seven cells, including a haploid egg cell and a binucleate (diploid) central cell; the mature male gametophyte — the pollen grain — contains two haploid sperm cells. Pollination delivers both sperm cells to the female gametophyte. The subsequent union of the egg cell and one of the sperm cells to form the zygote is accompanied by the fusion of the central cell with
the other sperm cell to produce the triploid precursor of the endosperm in a process called double fertilization [7]. The idea that plant embryo development requires little stored maternal information has led to the hypothesis that zygotic activation of gene expression occurs very early in plant seed development. In particular, the isolation of a large number of recessive embryo lethal mutations suggested that both maternal and paternal alleles of a large number of genes are required for embryo development [8]. For example, mutations of the EMB30/GNOM gene were found to affect the first zygotic division [9]. There has, however, been very little direct information on gene expression patterns during early seed development. In particular, it was not known whether maternal and paternal genomes are regulated distinctively. Recent studies in Arabidopsis have provided important first glimpses into this mystery [6,10–14], and suggest that maternal and paternal genomes are indeed under different regulation during seed development. Most strikingly, Vielle-Calzada et al. [6] have demonstrated that the expression of the paternal alleles of many genes are delayed during Arabidopsis seed development. The differential regulation of maternal and paternal alleles during Arabidopsis embryo development was demonstrated recently when mutations in the FIE, FIS2 and MEDEA genes were found to cause abnormal embryonic development only when the mutations were inherited from the mother, and not the father (Figure 1) [10–12]. When the maternal sporophyte is heterozygous for one of these mutations, half of the embryo sacs inherit the mutation, and the embryos that result from the egg cells of these embryo sacs will exhibit defective development regardless of whether the sperm carries a normal allele or not. Reciprocal crosses indicated that a functional male allele of these genes is not required. The genotype of the female gametophyte, irrespective of that of the male gametophyte, is therefore critical for normal embryo development, so that these are examples of gametophytic maternal-effect genes. It is possible that the FIE, FIS2 and MEDEA gene products are synthesized during early female gametophyte development and incorporated into the egg cell where their functions are required during early seed development after fertilization. At least in the case of MEDEA, however, it has been convincingly shown that, whereas the maternal allele is transcribed in the endosperm during early seed development, the paternal allele is not expressed in either endosperm or embryo [14]. The lack of expression of the paternal MEDEA allele is thus the
bb10j06.qxd
10/5/00
R366
2:00 pm
Page R366
Current Biology Vol 10 No 10
Figure 1 (a) Mutant Wild-type mother father a/a A/A
(b) Heterozygous mother a/A
A/A
Wild-type father
Meiosis + mitosis
Meiosis + fertilization
A/a Mutant phenotype
Mutant Wild-type mother father A/A a/a
Mutant a A Wild-type Wild-type A A Wild-type embryo sac pollen embryo sac pollen Fertilization Fertilization Mutant phenotype A/a Wild-type mother
A/A
A/A
a/A
Normal phenotype
Reciprocal crosses to detect maternal-effect mutations. (a) Crosses for maternal-effect mutations in animals, or those affecting the sporophytic generation in plants. (b) Crosses for plant gametophytic maternal-effect mutations. The part of a cross that results in a mutant phenotype is highlighted by a yellow box. The medea mutation exhibits this type of maternal effect throughout seed development and in the mature seed, whereas the emb30 mutation only shows maternal effect during early seed development.
Heterozygous father
Meiosis + mitosis
Meiosis + fertilization A/a Normal phenotype Current Biology
Wild-type A a Mutant embryo sac pollen Fertilization
Wild-type A A Wild-type pollen embryo sac Fertilization
Normal A/a phenotype
molecular basis for the gametophytic maternal-effect of the MEDEA gene. This result indicates that there is differential gene expression during early seed development, depending on the parental origin of the alleles. Does this differential expression occur with other genes? Now the same group has presented an impressive set of results on such regulation of 20 genes from Arabidopsis [6]. First, they identified 19 genes that are expressed during early seed development, in embryo and/or endosperm, using a transposon-mediated insertional reporter gene — GUS, encoding β-glucuronidase — system for detection of gene expression [15]. To confirm that the GUS staining pattern truly represents the expression of the neighboring endogenous genes, in situ RNA hybridization was performed for one of the genes and found to reveal an identical pattern. To test whether both the maternal and paternal alleles are expressed, they performed reciprocal crosses between these transgenic GUS lines and wild-type plants; they found that for each and every one of these lines, GUS expression was detected when the transgene was inherited from the mother, and not when it was inherited from the father. Is the sex-specific differential expression of the GUS transgenes a specific property of the transgenes, or does it reflect a sex-specific gene regulation of endogenous genes? To distinguish these alternatives, Vielle-Calzada et al. [6] devised a clever test using single-nucleotide polymorphisms between different geographical strains (ecotypes) of Arabidopsis. One of the GUS lines has an insertion into the PROLIFERA
A/A
Normal phenotype
(PRL) gene [16], for which they also found a singlenucleotide polymorphism which creates a XhoI restrictionsite polymorphism between the Columbia and Landsberg erecta ecotypes. Using reverse transcription followed by the polymerase chain reaction (RT-PCR) and restriction analysis, they were able to show that, during the first three days of seed development, only expression of the maternal allele could be detected. They found that expression of the paternal allele began about four days after pollination, when the embryo has about 32–64 cells. So, at least for PRL, the sexspecific GUS expression during early seed development is consistent with the endogenous gene expression. How general is this pattern of regulation? Does it apply to just a specific subset of genes encoding particular types of protein? Or is this regulation found only with genes at specific chromosomal locations? Vielle-Calzada et al. [6] addressed these questions by determining the nature of the predicted gene products and their chromosomal map positions. They found that the genes showing delayedpaternal expression encode a variety of proteins, including likely transcriptional regulators, a predicted membrane protein, and a putative enzyme that catalyses conversions between protein secondary structures, as well as others with no similarity to known proteins. Furthermore, these genes map to different positions on four of five chromosomes. So the genes show neither functional similarity nor chromosomal clustering. Although it is too soon to be certain, the authors make a compelling argument for the global sex-specific regulation of gene expression during early seed development in Arabidopsis.
bb10j06.qxd
10/5/00
2:00 pm
Page R367
Dispatch
If there is indeed such a global regulation, then other known genes of early seed development should similarly show sex-specific expression. The EMB30/GNOM gene was previously shown to be required for early embryo development [9,17]. In some mutant embryos, the first division of the zygote is almost symmetric, unlike the normal first division, which establishes the apical-basal polarity in the early embryo. Self-pollination of the emb30/+ heterozygote results in one in four embryo lethals, indicating that both female and male alleles are needed for normal seed development. Using RT-PCR to detect allele-specific expression, however, Vielle-Calzada et al. [6] found that only the maternal allele of the EMB30 gene is expressed at one day after pollination. This suggests that EMB30 may in fact be a maternal-effect gene. To test this, Vielle-Calzada et al. [6] performed reciprocal crosses with emb30/+ and wild-type plants (Figure 1) [6]. When an emb30/+ plant was pollinated with wild-type pollen, the early morphological defects were not corrected by the paternal wild-type allele. In principle, this early defect could be due to a dosage effect where a single copy of normal copy is not sufficient. However, the reciprocal cross of a wild-type plant pollinated by pollen from an emb30/+ plant resulted in normal early seed development. The results of both crosses together indicate that EMB30 is indeed a maternal-effect gene for early seed development. If EMB30 is a strict gametophytic-maternal-effect gene, then 50% of the mature seeds from an emb30/+ plant should be defective in seed development, regardless of the source of the pollen. What, then, is the explanation for the observed 25% defective mature seeds from a self-pollination? It is possible that the delayed expression of the paternal allele is able to provide the necessary function to correct the early defects and ensure normal late development. If this is true, a cross of emb30/+ pollinated with wildtype pollen should produce 100% normal seeds instead of 50% normal seeds, even though there are early morphological defects as discussed above. Indeed 100% normal mature seeds were observed from such a cross [6]. So when a functional paternal allele is provided, the emb30 maternaleffect early morphological defects in the seeds are reversible, and the late seed development is normal. There are many other mutations which cause 25% lethal or defective embryos when the heterozygote is selfed, and these mutations are called embryo defective (emb) mutations [8]. It is possible that some or even many of the emb mutations that affect early seed development may be maternaleffect genes whose mutant phenotypes are rescued by the delayed expression of paternal alleles. For these genes, when the maternal allele is defective, the delayed expression of a normal paternal allele provides the essential function for normal seed development. Therefore, the delayed expression is still beneficial from a developmental perspective.
R367
What might be the mechanism that controls this delayed expression of paternal alleles? One possibility is genomic imprinting, which by definition affects expression of an allele according to its parent-of-origin. It was recently observed that only the maternal allele, and not the paternal allele, of MEDEA is active during seed development [14,18], suggesting that MEDEA might be imprinted. Critical support for the inference of imprinting was obtained by Vielle-Calzada et al. [14] when they found that the maternal MEDEA allele is transcribed after fertilization. They also showed that the silencing of the paternal MEDEA allele requires the zygotic function of the DDM1 gene, which encodes a protein similar to the SWI2/SNF2 family of DNA-dependent ATPases that are known to be important regulators of chromatin structure [14,19]. Mutations in DDM1 were found to reduce the DNA methylation levels of repetitive sequences. Silencing of MEDEA may therefore be regulated by chromatin structure and/or DNA methylation under the control of DDM1. As Vielle-Calzada et al. [6] point out, it is possible that the delayed expression of the paternal allele seen with many genes expressed during early seed development may also involve changes in chromatin structure or DNA methylation. For example, the tightly packed chromatin structure in the sperm nucleus may render the paternal alleles inaccessible to transcriptional activators. During the first few cell divisions during early seed development, the paternal chromatin may be loosened to allow gene expression. If the silencing of paternal alleles during early seed development observed by Vielle-Calzada et al. [6] is due to genomic imprinting, then the maternal alleles should be transcribed at this time. But unlike the case of MEDEA [14], transcription of the maternal alleles of these other differentially expressed genes has not been examined yet; for this reason, the authors were careful to make clear that their results do not distinguish between storage of maternal mRNAs that were transcribed (in the egg and/or central cells) before fertilization, on the one hand, and expression specifically of the maternal allele after fertilization, on the other. Direct measurement of transcription of the maternal alleles of these genes are needed to distinguish these possibilities. It is worth noting that MEDEA differs from EMB30 in that, unlike the latter, the former is not rescued by a normal paternal allele. In a zygotic ddm1 mutant, a derepressed paternal MEDEA allele was able to rescue the seed defect that would otherwise be caused by the mutant maternal medea allele [14]; this implies that silencing of the MEDEA paternal allele is normally more prolonged than is the case for the paternal alleles of the other genes that show parent-of-origin-specific expression. Two other genes, FIE and FIS2, show a failure of wild-type paternal alleles to rescue maternal defects, so these are likely silenced to an extent similar to MEDEA. There may, therefore, be mechanistic differences between the way
bb10j06.qxd
10/5/00
R368
2:00 pm
Page R368
Current Biology Vol 10 No 10
MEDEA, FIE and FIS2, on the one hand, and genes such as EMB30, on the other, are controlled. FIE, FIS2 and MEDEA all encode proteins related to the Polycomb group proteins, which are known to regulate long-term gene expression and chromatin structure in animals; it is possible, therefore, that these genes are required for either the delay in expression, or late activation, of paternal alleles during seed development. This can now be tested by crossing the GUS lines into fie, fis2 or medea mutants.
would be interesting to learn whether there are any paternal-effect genes, which are required regardless of the function of the maternal alleles. It is certain that the findings by Vielle-Calzada et al. [6] will stimulate further studies in this fascinating and rapidly growing field of plant development.
The delayed expression of the paternal genome during early seed development in principle extends the haploid stage of plant development, which as mentioned above can help eliminate deleterious mutations from the population and thereby increase the evolutionary fitness of the species. This strategy of removing mutations is effective only when a maternal-effect mutation is not paternally rescuable, such as in the case of the MEDEA gene. However, very few emb mutations are known to cause a true maternal-effect phenotype, as seen with the medea mutation. Some of the known emb mutations may be similar to the emb30 mutation and cause a maternal-effect defect only during early seed development; they were missed when only late development was examined. Other emb mutations may cause defects not during early seed development but in late seed development; they would not be maternaleffect mutations because paternal alleles are active when the gene is required. Therefore, true maternal-effect genes like MEDEA are probably rare. One explanation for the scarcity of true maternal-effect seed development mutations is that such mutations may also be required during the development of the haploid embryo sac, preventing them to be recognized as maternal effect mutations.
References
If delayed expression of paternal alleles is indeed a general phenomenon, true of most genes, then genetic studies of embryo and seed development will need to be re-examined in light of the new results. Because maternal-effect and zygotic mutations — which produce 50% and 25% defective seeds after self-crossing, respectively — are not always easily distinguishable, given that mutagenized plants tend to have reduced seed set, reciprocal crosses need to be performed to identify maternal-effect mutations. Furthermore, some mutations may exhibit a maternal-effect defect only during early seed development, as in the case of the emb30 mutation, and must be recognized by examining early seed development rather than mature seeds. In this second group, late expression of the paternal alleles are expected to rescue the mutant defect. Furthermore, it is not known whether the delay in paternal genome expression is simply a passive consequence of tightly packed sperm chromosomes, or a result of active regulation important for normal seed development. In other words, is gene dosage important in early seed development? Artificially accelerated expression of paternal alleles could be used to address this question. Finally, it
Acknowledgements I thank Ueli Grossniklaus and Jean-Philippe Vielle-Calzada for helpful comments on the manuscript.
1. Dworkin MB, Dworkin-Rastl E: Functions of maternal mRNA in early development. Mol Reprod Dev 1990, 26:261-297. 2. Zalokar M: Autoradiographic study of protein and RNA formation during early development in Drosophila eggs. Dev Biol 1976, 49:425-437. 3. Nothias JY, Majumder S, Kaneko KJ, DePamphilis ML: Regulation of gene expression at the beginning of mammalian development. J Biol Chem 1995, 270:22077-22080. 4. Walbot V: Sources and consequences of phenotypic and genotypic plasticity in flowering plants. Trends Plant Sci 1996, 1:27-32. 5. de Vries SC, de Jong AJ, van Engelen FA: Formation of embryogenic cells in plant tissue cultures. Biochem Soc Symp 1994, 60:43-50. 6. Vielle-Calzada JP, Basker R, Grossniklaus U: Delayed activation of the paternal genome during seed development. Nature 2000, 404:91-94. 7. van Went JL, Willemse MTM: In Embryology of Angiosperms. Edited by Johri B. Berlin: Springer; 1984:273-318. 8. Meinke DW: Seed development. In Arabidopsis thaliana. Edited by Meyerowitz EM, Somerville CR. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1994:253-295. 9. Mayer U, Büttner G, Jürgens G: Apical-basal pattern formation in the Arabidopsis embryo: studies on the role of the gnom gene. Development 1993, 117:149-162. 10. Ohad N, Margossian L, Hsu Y-C, Williams C, Repetti P, Fisher R: A mutation that allows endosperm development without fertilization. Proc Natl Acad Sci USA 1996, 93:5319-5324. 11. Chaudhury AM, Ming L, Miller C, Craig S, Dennis E, Peacock WJ: Fertilization-independent seed development in Arabidopsis thaliana. Proc Natl Acad Sci USA 1997, 94:4223-4228. 12. Grossniklaus U, Vielle-Calzada JP, Hoeppner MA, Gagliano WB: Maternal control of embryogenesis by MEDEA, a polycomb group gene in Arabidopsis. Science 1998, 280:446-450. 13. Scott RJ, Spielman M, Bailey J, Dickinson HG: Parent-of-origin effects on seed development in Arabidopsis thaliana. Development 1998, 125:3329-3341. 14. Vielle-Calzada JP, Thomas J, Spillane C, Coluccio A, Hoeppner MA, Grossniklaus U: Maintenance of genomic imprinting at the Arabidopsis medea locus requires zygotic DDM1 activity. Genes Dev 1999, 13:2971-2982. 15. Sundaresan V, Springer P, Volpe T, Haward S, Jones J, Dean C, Ma H, Martienssen R: Patterns of gene action in plant development revealed by enhancer trap and gene trap transposable elements. Genes Dev 1995, 9:1797-1810. 16. Springer PS, McCombie WR, Sundaresan V, Martienssen RA: Gene trap tagging of PROLIFERA, an essential MCM2-3-5-like gene in Arabidopsis. Science 1995, 268:877-880. 17. Shevell DE, Leu W-M, Gillmor CS, Xia G, Feldmann KA, Chua N-H: EMB30 is essential for normal cell division, cell expansion, and cell adhesion in Arabidopsis and encodes a protein that has similarity to Sec7. Cell 1994, 77:1051-1062. 18. Kinoshita T, Yadegari R, Harada JJ, Goldberg RB, Fischer RL: Imprinting of the MEDEA polycomb gene in the Arabidopsis endosperm. Plant Cell 1999, 11:1945-1952. 19. Jeddeloh, JA, Stokes TL, Richards EJ: Maintenance of genomic methylation requires a SWI2/SNF2-like protein. Nat Genet 1999, 22:94-97.
10/5/00
2:00 pm
Page R365
Dispatch
R365
Gene regulation: Better late than never? Hong Ma
A new study has found that, in Arabidopsis, the paternal copies of many genes are delayed in expression during early seed development. The distribution of the genes and nature of their products suggest that this delayed expression of paternal alleles may be a global phenomenon. Address: Department of Biology and the Life Sciences Consortium, 504 Wartik Laboratory, Pennsylvania State University, Pennsylvania 16802, USA. E-mail: [email protected] Current Biology 2000, 10:R365–R368 0960-9822/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved.
It is well established that, in animals, early embryo development requires maternal information stored in the unfertilized egg [1]. Zygotic gene expression is generally found to be activated at some specific time during early animal embryo development (for example, see [2,3]), often associated with a specific transition in global properties such as the length of the cell cycle. In contrast, it has long been thought that embryo development in plants requires little stored maternal information [4]. This view is supported by the well-known observation that plant somatic cells can readily form functional somatic embryos [5]. Despite this difference between animals and plants, recent work indicates that there are also important differences between the contributions of the maternal and paternal genomes during early plant development — in particular, VielleCalzada et al. [6] have found that the expression of paternal alleles is frequently delayed during seed development in the model plant Arabidopsis thaliana. Plants have alternating diploid and haploid generations. In the diploid sporophytic organism, cells undergo meiosis to produce spores, which develop into multicellular haploid gametophytes. The presence of a haploid generation provides a mechanism for the efficient removal of deleterious mutations. Cells in the gametophytes then differentiate into male and female gametes. In flowering plants, the male and female gametophytes are highly reduced in size and develop within the reproductive organs of the diploid sporophyte. The mature female gametophyte — the embryo sac — of flowering plants has seven cells, including a haploid egg cell and a binucleate (diploid) central cell; the mature male gametophyte — the pollen grain — contains two haploid sperm cells. Pollination delivers both sperm cells to the female gametophyte. The subsequent union of the egg cell and one of the sperm cells to form the zygote is accompanied by the fusion of the central cell with
the other sperm cell to produce the triploid precursor of the endosperm in a process called double fertilization [7]. The idea that plant embryo development requires little stored maternal information has led to the hypothesis that zygotic activation of gene expression occurs very early in plant seed development. In particular, the isolation of a large number of recessive embryo lethal mutations suggested that both maternal and paternal alleles of a large number of genes are required for embryo development [8]. For example, mutations of the EMB30/GNOM gene were found to affect the first zygotic division [9]. There has, however, been very little direct information on gene expression patterns during early seed development. In particular, it was not known whether maternal and paternal genomes are regulated distinctively. Recent studies in Arabidopsis have provided important first glimpses into this mystery [6,10–14], and suggest that maternal and paternal genomes are indeed under different regulation during seed development. Most strikingly, Vielle-Calzada et al. [6] have demonstrated that the expression of the paternal alleles of many genes are delayed during Arabidopsis seed development. The differential regulation of maternal and paternal alleles during Arabidopsis embryo development was demonstrated recently when mutations in the FIE, FIS2 and MEDEA genes were found to cause abnormal embryonic development only when the mutations were inherited from the mother, and not the father (Figure 1) [10–12]. When the maternal sporophyte is heterozygous for one of these mutations, half of the embryo sacs inherit the mutation, and the embryos that result from the egg cells of these embryo sacs will exhibit defective development regardless of whether the sperm carries a normal allele or not. Reciprocal crosses indicated that a functional male allele of these genes is not required. The genotype of the female gametophyte, irrespective of that of the male gametophyte, is therefore critical for normal embryo development, so that these are examples of gametophytic maternal-effect genes. It is possible that the FIE, FIS2 and MEDEA gene products are synthesized during early female gametophyte development and incorporated into the egg cell where their functions are required during early seed development after fertilization. At least in the case of MEDEA, however, it has been convincingly shown that, whereas the maternal allele is transcribed in the endosperm during early seed development, the paternal allele is not expressed in either endosperm or embryo [14]. The lack of expression of the paternal MEDEA allele is thus the
bb10j06.qxd
10/5/00
R366
2:00 pm
Page R366
Current Biology Vol 10 No 10
Figure 1 (a) Mutant Wild-type mother father a/a A/A
(b) Heterozygous mother a/A
A/A
Wild-type father
Meiosis + mitosis
Meiosis + fertilization
A/a Mutant phenotype
Mutant Wild-type mother father A/A a/a
Mutant a A Wild-type Wild-type A A Wild-type embryo sac pollen embryo sac pollen Fertilization Fertilization Mutant phenotype A/a Wild-type mother
A/A
A/A
a/A
Normal phenotype
Reciprocal crosses to detect maternal-effect mutations. (a) Crosses for maternal-effect mutations in animals, or those affecting the sporophytic generation in plants. (b) Crosses for plant gametophytic maternal-effect mutations. The part of a cross that results in a mutant phenotype is highlighted by a yellow box. The medea mutation exhibits this type of maternal effect throughout seed development and in the mature seed, whereas the emb30 mutation only shows maternal effect during early seed development.
Heterozygous father
Meiosis + mitosis
Meiosis + fertilization A/a Normal phenotype Current Biology
Wild-type A a Mutant embryo sac pollen Fertilization
Wild-type A A Wild-type pollen embryo sac Fertilization
Normal A/a phenotype
molecular basis for the gametophytic maternal-effect of the MEDEA gene. This result indicates that there is differential gene expression during early seed development, depending on the parental origin of the alleles. Does this differential expression occur with other genes? Now the same group has presented an impressive set of results on such regulation of 20 genes from Arabidopsis [6]. First, they identified 19 genes that are expressed during early seed development, in embryo and/or endosperm, using a transposon-mediated insertional reporter gene — GUS, encoding β-glucuronidase — system for detection of gene expression [15]. To confirm that the GUS staining pattern truly represents the expression of the neighboring endogenous genes, in situ RNA hybridization was performed for one of the genes and found to reveal an identical pattern. To test whether both the maternal and paternal alleles are expressed, they performed reciprocal crosses between these transgenic GUS lines and wild-type plants; they found that for each and every one of these lines, GUS expression was detected when the transgene was inherited from the mother, and not when it was inherited from the father. Is the sex-specific differential expression of the GUS transgenes a specific property of the transgenes, or does it reflect a sex-specific gene regulation of endogenous genes? To distinguish these alternatives, Vielle-Calzada et al. [6] devised a clever test using single-nucleotide polymorphisms between different geographical strains (ecotypes) of Arabidopsis. One of the GUS lines has an insertion into the PROLIFERA
A/A
Normal phenotype
(PRL) gene [16], for which they also found a singlenucleotide polymorphism which creates a XhoI restrictionsite polymorphism between the Columbia and Landsberg erecta ecotypes. Using reverse transcription followed by the polymerase chain reaction (RT-PCR) and restriction analysis, they were able to show that, during the first three days of seed development, only expression of the maternal allele could be detected. They found that expression of the paternal allele began about four days after pollination, when the embryo has about 32–64 cells. So, at least for PRL, the sexspecific GUS expression during early seed development is consistent with the endogenous gene expression. How general is this pattern of regulation? Does it apply to just a specific subset of genes encoding particular types of protein? Or is this regulation found only with genes at specific chromosomal locations? Vielle-Calzada et al. [6] addressed these questions by determining the nature of the predicted gene products and their chromosomal map positions. They found that the genes showing delayedpaternal expression encode a variety of proteins, including likely transcriptional regulators, a predicted membrane protein, and a putative enzyme that catalyses conversions between protein secondary structures, as well as others with no similarity to known proteins. Furthermore, these genes map to different positions on four of five chromosomes. So the genes show neither functional similarity nor chromosomal clustering. Although it is too soon to be certain, the authors make a compelling argument for the global sex-specific regulation of gene expression during early seed development in Arabidopsis.
bb10j06.qxd
10/5/00
2:00 pm
Page R367
Dispatch
If there is indeed such a global regulation, then other known genes of early seed development should similarly show sex-specific expression. The EMB30/GNOM gene was previously shown to be required for early embryo development [9,17]. In some mutant embryos, the first division of the zygote is almost symmetric, unlike the normal first division, which establishes the apical-basal polarity in the early embryo. Self-pollination of the emb30/+ heterozygote results in one in four embryo lethals, indicating that both female and male alleles are needed for normal seed development. Using RT-PCR to detect allele-specific expression, however, Vielle-Calzada et al. [6] found that only the maternal allele of the EMB30 gene is expressed at one day after pollination. This suggests that EMB30 may in fact be a maternal-effect gene. To test this, Vielle-Calzada et al. [6] performed reciprocal crosses with emb30/+ and wild-type plants (Figure 1) [6]. When an emb30/+ plant was pollinated with wild-type pollen, the early morphological defects were not corrected by the paternal wild-type allele. In principle, this early defect could be due to a dosage effect where a single copy of normal copy is not sufficient. However, the reciprocal cross of a wild-type plant pollinated by pollen from an emb30/+ plant resulted in normal early seed development. The results of both crosses together indicate that EMB30 is indeed a maternal-effect gene for early seed development. If EMB30 is a strict gametophytic-maternal-effect gene, then 50% of the mature seeds from an emb30/+ plant should be defective in seed development, regardless of the source of the pollen. What, then, is the explanation for the observed 25% defective mature seeds from a self-pollination? It is possible that the delayed expression of the paternal allele is able to provide the necessary function to correct the early defects and ensure normal late development. If this is true, a cross of emb30/+ pollinated with wildtype pollen should produce 100% normal seeds instead of 50% normal seeds, even though there are early morphological defects as discussed above. Indeed 100% normal mature seeds were observed from such a cross [6]. So when a functional paternal allele is provided, the emb30 maternaleffect early morphological defects in the seeds are reversible, and the late seed development is normal. There are many other mutations which cause 25% lethal or defective embryos when the heterozygote is selfed, and these mutations are called embryo defective (emb) mutations [8]. It is possible that some or even many of the emb mutations that affect early seed development may be maternaleffect genes whose mutant phenotypes are rescued by the delayed expression of paternal alleles. For these genes, when the maternal allele is defective, the delayed expression of a normal paternal allele provides the essential function for normal seed development. Therefore, the delayed expression is still beneficial from a developmental perspective.
R367
What might be the mechanism that controls this delayed expression of paternal alleles? One possibility is genomic imprinting, which by definition affects expression of an allele according to its parent-of-origin. It was recently observed that only the maternal allele, and not the paternal allele, of MEDEA is active during seed development [14,18], suggesting that MEDEA might be imprinted. Critical support for the inference of imprinting was obtained by Vielle-Calzada et al. [14] when they found that the maternal MEDEA allele is transcribed after fertilization. They also showed that the silencing of the paternal MEDEA allele requires the zygotic function of the DDM1 gene, which encodes a protein similar to the SWI2/SNF2 family of DNA-dependent ATPases that are known to be important regulators of chromatin structure [14,19]. Mutations in DDM1 were found to reduce the DNA methylation levels of repetitive sequences. Silencing of MEDEA may therefore be regulated by chromatin structure and/or DNA methylation under the control of DDM1. As Vielle-Calzada et al. [6] point out, it is possible that the delayed expression of the paternal allele seen with many genes expressed during early seed development may also involve changes in chromatin structure or DNA methylation. For example, the tightly packed chromatin structure in the sperm nucleus may render the paternal alleles inaccessible to transcriptional activators. During the first few cell divisions during early seed development, the paternal chromatin may be loosened to allow gene expression. If the silencing of paternal alleles during early seed development observed by Vielle-Calzada et al. [6] is due to genomic imprinting, then the maternal alleles should be transcribed at this time. But unlike the case of MEDEA [14], transcription of the maternal alleles of these other differentially expressed genes has not been examined yet; for this reason, the authors were careful to make clear that their results do not distinguish between storage of maternal mRNAs that were transcribed (in the egg and/or central cells) before fertilization, on the one hand, and expression specifically of the maternal allele after fertilization, on the other. Direct measurement of transcription of the maternal alleles of these genes are needed to distinguish these possibilities. It is worth noting that MEDEA differs from EMB30 in that, unlike the latter, the former is not rescued by a normal paternal allele. In a zygotic ddm1 mutant, a derepressed paternal MEDEA allele was able to rescue the seed defect that would otherwise be caused by the mutant maternal medea allele [14]; this implies that silencing of the MEDEA paternal allele is normally more prolonged than is the case for the paternal alleles of the other genes that show parent-of-origin-specific expression. Two other genes, FIE and FIS2, show a failure of wild-type paternal alleles to rescue maternal defects, so these are likely silenced to an extent similar to MEDEA. There may, therefore, be mechanistic differences between the way
bb10j06.qxd
10/5/00
R368
2:00 pm
Page R368
Current Biology Vol 10 No 10
MEDEA, FIE and FIS2, on the one hand, and genes such as EMB30, on the other, are controlled. FIE, FIS2 and MEDEA all encode proteins related to the Polycomb group proteins, which are known to regulate long-term gene expression and chromatin structure in animals; it is possible, therefore, that these genes are required for either the delay in expression, or late activation, of paternal alleles during seed development. This can now be tested by crossing the GUS lines into fie, fis2 or medea mutants.
would be interesting to learn whether there are any paternal-effect genes, which are required regardless of the function of the maternal alleles. It is certain that the findings by Vielle-Calzada et al. [6] will stimulate further studies in this fascinating and rapidly growing field of plant development.
The delayed expression of the paternal genome during early seed development in principle extends the haploid stage of plant development, which as mentioned above can help eliminate deleterious mutations from the population and thereby increase the evolutionary fitness of the species. This strategy of removing mutations is effective only when a maternal-effect mutation is not paternally rescuable, such as in the case of the MEDEA gene. However, very few emb mutations are known to cause a true maternal-effect phenotype, as seen with the medea mutation. Some of the known emb mutations may be similar to the emb30 mutation and cause a maternal-effect defect only during early seed development; they were missed when only late development was examined. Other emb mutations may cause defects not during early seed development but in late seed development; they would not be maternaleffect mutations because paternal alleles are active when the gene is required. Therefore, true maternal-effect genes like MEDEA are probably rare. One explanation for the scarcity of true maternal-effect seed development mutations is that such mutations may also be required during the development of the haploid embryo sac, preventing them to be recognized as maternal effect mutations.
References
If delayed expression of paternal alleles is indeed a general phenomenon, true of most genes, then genetic studies of embryo and seed development will need to be re-examined in light of the new results. Because maternal-effect and zygotic mutations — which produce 50% and 25% defective seeds after self-crossing, respectively — are not always easily distinguishable, given that mutagenized plants tend to have reduced seed set, reciprocal crosses need to be performed to identify maternal-effect mutations. Furthermore, some mutations may exhibit a maternal-effect defect only during early seed development, as in the case of the emb30 mutation, and must be recognized by examining early seed development rather than mature seeds. In this second group, late expression of the paternal alleles are expected to rescue the mutant defect. Furthermore, it is not known whether the delay in paternal genome expression is simply a passive consequence of tightly packed sperm chromosomes, or a result of active regulation important for normal seed development. In other words, is gene dosage important in early seed development? Artificially accelerated expression of paternal alleles could be used to address this question. Finally, it
Acknowledgements I thank Ueli Grossniklaus and Jean-Philippe Vielle-Calzada for helpful comments on the manuscript.
1. Dworkin MB, Dworkin-Rastl E: Functions of maternal mRNA in early development. Mol Reprod Dev 1990, 26:261-297. 2. Zalokar M: Autoradiographic study of protein and RNA formation during early development in Drosophila eggs. Dev Biol 1976, 49:425-437. 3. Nothias JY, Majumder S, Kaneko KJ, DePamphilis ML: Regulation of gene expression at the beginning of mammalian development. J Biol Chem 1995, 270:22077-22080. 4. Walbot V: Sources and consequences of phenotypic and genotypic plasticity in flowering plants. Trends Plant Sci 1996, 1:27-32. 5. de Vries SC, de Jong AJ, van Engelen FA: Formation of embryogenic cells in plant tissue cultures. Biochem Soc Symp 1994, 60:43-50. 6. Vielle-Calzada JP, Basker R, Grossniklaus U: Delayed activation of the paternal genome during seed development. Nature 2000, 404:91-94. 7. van Went JL, Willemse MTM: In Embryology of Angiosperms. Edited by Johri B. Berlin: Springer; 1984:273-318. 8. Meinke DW: Seed development. In Arabidopsis thaliana. Edited by Meyerowitz EM, Somerville CR. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 1994:253-295. 9. Mayer U, Büttner G, Jürgens G: Apical-basal pattern formation in the Arabidopsis embryo: studies on the role of the gnom gene. Development 1993, 117:149-162. 10. Ohad N, Margossian L, Hsu Y-C, Williams C, Repetti P, Fisher R: A mutation that allows endosperm development without fertilization. Proc Natl Acad Sci USA 1996, 93:5319-5324. 11. Chaudhury AM, Ming L, Miller C, Craig S, Dennis E, Peacock WJ: Fertilization-independent seed development in Arabidopsis thaliana. Proc Natl Acad Sci USA 1997, 94:4223-4228. 12. Grossniklaus U, Vielle-Calzada JP, Hoeppner MA, Gagliano WB: Maternal control of embryogenesis by MEDEA, a polycomb group gene in Arabidopsis. Science 1998, 280:446-450. 13. Scott RJ, Spielman M, Bailey J, Dickinson HG: Parent-of-origin effects on seed development in Arabidopsis thaliana. Development 1998, 125:3329-3341. 14. Vielle-Calzada JP, Thomas J, Spillane C, Coluccio A, Hoeppner MA, Grossniklaus U: Maintenance of genomic imprinting at the Arabidopsis medea locus requires zygotic DDM1 activity. Genes Dev 1999, 13:2971-2982. 15. Sundaresan V, Springer P, Volpe T, Haward S, Jones J, Dean C, Ma H, Martienssen R: Patterns of gene action in plant development revealed by enhancer trap and gene trap transposable elements. Genes Dev 1995, 9:1797-1810. 16. Springer PS, McCombie WR, Sundaresan V, Martienssen RA: Gene trap tagging of PROLIFERA, an essential MCM2-3-5-like gene in Arabidopsis. Science 1995, 268:877-880. 17. Shevell DE, Leu W-M, Gillmor CS, Xia G, Feldmann KA, Chua N-H: EMB30 is essential for normal cell division, cell expansion, and cell adhesion in Arabidopsis and encodes a protein that has similarity to Sec7. Cell 1994, 77:1051-1062. 18. Kinoshita T, Yadegari R, Harada JJ, Goldberg RB, Fischer RL: Imprinting of the MEDEA polycomb gene in the Arabidopsis endosperm. Plant Cell 1999, 11:1945-1952. 19. Jeddeloh, JA, Stokes TL, Richards EJ: Maintenance of genomic methylation requires a SWI2/SNF2-like protein. Nat Genet 1999, 22:94-97.