| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
Plant Physiology 144:562-574 (2007) © 2007 American Society of Plant Biologists Using Genomics to Study Legume Seed Development1Department of Molecular, Cell, and Developmental Biology, University of California, Los Angeles, California 90095 (B.H.L., J.A.W., T.K., A.Q.B., R.B.G.); and Section of Plant Biology, Division of Biological Sciences, University of California, Davis, California 95616 (J.J.H.)
Seeds are essential for flowering plant reproduction because they protect, nourish, and contain the developing embryo that represents the next sporophytic generation. In addition, seeds contain energy resources that sustain the young sporophyte during germination before photosynthesis begins. In legumes, food reserves stored in embryonic cotyledons make seeds important as a food source for both human and animal consumption. For example, soybean (Glycine max) is now one of the most important seed crops in the world (Wilcox, 2004
Seed development is triggered by a novel double-fertilization process that leads to the differentiation of the embryo, endosperm, and seed coat, which are the major compartments of the seed (Fig. 1, A–C
; Goldberg et al., 1994
Many developmental and physiological events occur within each seed compartment during development (Fig. 1B) and are programmed, in part, by the activity of different genes (Goldberg et al., 1989  , 1994; Stangeland et al., 2003; Gehring et al., 2004; Haughn and Chaudhury, 2005). Seed development, therefore, is the result of a mosaic of distinct gene expression programs occurring in parallel in different seed compartments (e.g. embryo, endosperm, seed coat) as well as within specific regions and tissues (e.g. embryo proper, suspensor, epidermis). What these programs are and how they are integrated into unique regulatory networks within the plant genome remain major unanswered questions (Fig. 1D). Specifically, it is not yet known what genes in different seed compartments play important roles in cell fate specification, differentiation, and morphogenesis during early seed and embryo development. Molecular identification and characterization of these genes will help identify regulatory networks that program and coordinate the development of each seed compartment. In addition, it is not known what the functions are of many genes that are expressed in different seed compartments. Identifying the function of compartment-specific genes should provide new insight into their roles in seed development. At present, new genomic resources allow seed biologists to use global gene expression profiling and comparative genomics to answer many questions that only a short time ago seemed out of reach. These questions, and others (Fig. 1D), are challenging the field of seed biology, and their answers should provide new insights into the process of seed development and lead to improved seeds for human and animal consumption.
Legumes represent one of the largest and most diverse families of flowering plants, with approximately 20,000 species classified (Doyle and Luckow, 2003 ). There are three subfamilies in legumes and the largest, Papilionoideae, contains most of the model species in which different aspects of plant biology have been studied. The most common legume models are peanut (Arachis hypogaea), Lotus (Lotus japonicus), Medicago (Medicago truncatula), soybean (Glycine max), scarlet runner bean (SRB; Phaseolus coccineus), common bean (Phaseolus vulgaris), pea (Pisum sativum), and broad bean (Vicia faba). The latter five species have been used historically to study seed and embryo development (Goldberg et al., 1989; Johnson et al., 1994; Coste et al., 2001; Weterings et al., 2001; Weber et al., 2005).
Several features make legumes an excellent model system to study seed and embryo development. For example, many legumes, such as soybean and peanut, are food crops of major economic importance. The mature seeds of these legumes are rich in proteins, carbohydrates, and oils, and accumulate to high nutritional value. These stored seed food reserves make legumes, such as soybean, the second most important crop for human nutrition and animal feed (Rubel et al., 1972
One of the most fascinating characteristics of legumes is that collectively they produce a large range of seed sizes (Fig. 2A
). For example, some legume seeds are giants and are excellent models for developmental studies, particularly during early stages of seed development. The large size of SRB globular-stage seed and embryo allows manipulation and isolation of embryonic regions, such as the embryo proper and suspensor, using hand-dissection techniques (Walbot et al., 1972
A second novel feature of legumes is that their embryos show a wide range of morphological forms (Fig. 2B). For example, two closely related species, soybean and SRB, have morphologically distinct suspensors. The soybean suspensor is small, consisting of a few cells, whereas the SRB suspensor is much larger and contains several hundred cells (Fig. 2B). The variety in size and shape of legume seeds and embryos makes them excellent models for comparative morphological studies using a functional genomics approach. This strategy can lead to a better understanding of the function, evolution, and diversity of legume seeds and their corresponding compartments.
Historically, legumes have been used to address important questions of seed and embryo development. In fact, early work with legumes contributed to the development of major ideas in biology. For example, during the early 1800s, Matthias Schleiden used several legumes, including Medicago and Vicia, to investigate the endosperm and describe the process of seed development (Schleiden and Vogel, 1838 , 1842). These studies contributed to his role in establishing the cell theory. In the mid-1800s, Gregor Mendel used peas to study the inheritance of phenotypic variation, including seed color and shape, leading to his Laws of Inheritance and the establishment of modern-day genetics (Mendel, 1865).
From the late 1800s to the middle of the 1900s, legumes were used to describe the processes of seed and embryo development, including the cellular events that occur before and after fertilization, early embryo cell cleavages, and endosperm differentiation. For example, Guignard's compendium of more than 40 legume species described the rich diversity of legume embryo and suspensor morphologies (Fig. 2B; Guignard, 1882
Studies on legume seed formation transitioned from descriptive anatomy to experiments at the molecular, biochemical, and physiological levels during the 1970s (Dure, 1975
During this same period, our laboratory used RNA-excess DNA-RNA hybridization experiments to show that approximately 14,000 to 18,000 diverse mRNAs are present in soybean embryos at different developmental stages (Goldberg et al., 1981b
Research on genes active in legume seed development exploded during the late 1970s and 1980s when it became possible to clone and study individual mRNAs and genes and reinsert them into plants using newly developed transformation techniques (Bevan et al., 1983
At present, the remarkable development of new genomic resources makes it possible to study legume gene expression during seed and embryo development at a global level. Currently, Medicago, Lotus, and soybean cDNA and oligonucleotide microarrays are available (Endo et al., 2002
Our laboratory has developed a genomics strategy to begin to identify the regulatory networks that program legume seed and embryo development (Fig. 3 ). One aspect of this strategy is to use the giant SRB embryo, pioneered by Sussex and colleagues (Walbot et al., 1972 ; Sussex et al., 1973), as an entry point to dissect the molecular events that occur during early embryogenesis (Fig. 4
). More recently, we incorporated soybean into our strategy (Fig. 5
) because it contrasts with SRB in terms of early embryo morphology (Fig. 2B) and because our laboratory has worked on soybean embryo development for more than 25 years (Goldberg et al., 1989). Recent development of soybean and SRB genomic resources allows both of these legumes to be used as excellent systems to identify genes that play important roles in the differentiation of unique seed and embryo compartments and to build compartment-specific regulatory networks that are crucial for seed formation (Figs. 4–6
).
Our laboratory has utilized SRB as a model system (Weterings et al., 2001 ) to identify genes and regulatory networks programming early embryo developmental events using a genomics approach (Figs. 3 and 4). We have focused on the question of how the embryo-proper and suspensor regions are specified from the apical and basal cells of a two-cell embryo, respectively (Fig. 4A; Weterings et al., 2001). SRB is unique in this regard because of its giant seed and large embryo (Fig. 2A), permitting hand dissection of embryo-proper and suspensor regions at early developmental stages (Fig. 4, B and C; Walbot et al., 1972; Sussex et al., 1973; Weterings et al., 2001). We prepared cDNA libraries from hand-dissected embryo-proper and suspensor regions of SRB globular-stage embryos (Fig. 3; Weterings et al., 2001) and ESTs were sequenced to determine what genes are active in the embryo proper and suspensor. Our SRB early embryo EST database is available at both http://estdb.biology.ucla.edu/PcEST and GenBank (accession nos. CA896559–CA916678). Figure 4D shows the distribution of >16,000 SRB suspensor ESTs grouped by functional categories, illustrating the large diversity of genes that are active in a terminally differentiated suspensor. Surprisingly, more than 300 suspensor transcription factor ESTs were identified that are distributed into a variety of transcription factor gene families (A.Q. Bui, B.H. Le, and R.B. Goldberg, unpublished data), providing a glimpse into the spectrum of regulatory genes active in one region of a legume embryo shortly after fertilization. What roles these transcription factors play in suspensor differentiation and function remain to be determined.
As part of our strategy, we carried out real-time quantitative reverse transcription (qRT)-PCR and in situ hybridization experiments (Fig. 3) to quantify and localize the accumulation of SRB mRNAs during early embryo development. One advantage of using giant SRB embryos is that qRT-PCR can be used to quantitate mRNA levels in different regions of a single embryo. For example, PcL1L mRNA uncovered in our EST studies is localized throughout the globular-stage SRB embryo (Fig. 4E, inset; Kwong et al., 2003
Even though the giant SRB embryo is a novel system for studying the specification of embryo regions early in development, this legume does have some limitations for global studies of seed formation. For example, although we have produced an EST dataset for SRB globular-stage embryo regions (http://estdb.biology.ucla.edu/PcEST), few other genomic resources are available. In addition, transformation procedures have not yet been developed for SRB. Finally, because SRB is not a major food crop, it is unlikely that a genome project will be carried out to sequence the SRB genome. To complement our use of SRB as a gene discovery engine and overcome these deficiencies, we decided to go back to the future (Goldberg et al., 1989 ) and use soybean to dissect genes important for making a seed. At the present time, a large number of genomic resources have been developed for soybean, including microarrays, EST databases, and genome sequences (Shoemaker et al., 2002; Vodkin et al., 2004; Jackson et al., 2006). In addition, transformation procedures are well established for soybean, making it feasible to address questions of gene function (Ko et al., 2006; Olhoft et al., 2006). Finally, the unique morphological differences between soybean and SRB embryos (e.g. suspensor size and shape) should permit structure-function questions of legume embryo diversity to be studied (Fig. 2B).
Although we have been able to take advantage of the giant SRB embryo to dissect by hand embryo-proper and suspensor regions (Figs. 3 and 4C), this approach is time consuming and not practical with legumes that have smaller embryos, such as soybean (Fig. 2B). One way to overcome the limitations of hand dissection and to be able to isolate different regions from any legume seed and embryo regardless of size (Fig. 2) is to make use of spectacular progress in laser capture microdissection (LCM) technology (Day et al., 2005 We have been using LCM with soybean seeds to identify all the genes required to make a seed (Figs. 3 and 5). In combination with GeneChip technology, we can investigate the global gene activity profiles in different compartments of the entire seed. For example, we used LCM to isolate the endosperm, suspensor, embryo proper, endothelium, inner integument, outer integument, epidermis, and hilum from a globular-stage soybean seed (Fig. 5, B and C). We hybridized RNAs isolated from each of these seed regions, as well as from intact globular-stage seeds (Fig. 5, A–C), with soybean Affymetrix GeneChips (J.A. Wagmaister, X. Wang, A.Q. Bui, B.H. Le, and R.B. Goldberg, unpublished data). We then compared the spectrum of diverse transcripts present in the entire globular-stage soybean seed (Fig. 5A) to those obtained with the eight laser-captured seed regions (Fig. 5, B and C). These data are available at http://estdb.biology.ucla.edu/seed as part of our National Science Foundation (NSF) Plant Genome Research Project.
Approximately 20,000 diverse transcripts were found to be present in the whole-mount globular-stage soybean seed (Fig. 5A), a value close to that which we obtained more than a quarter of a century ago using Rot curve hybridization technology (Goldberg et al., 1981b We estimated independently that there are approximately 22,000 diverse transcripts present in a soybean globular-stage seed (Fig. 5, B and C) by taking the union of each individual seed mRNA set captured by LCM (Fig. 5C) in close agreement with the result obtained with intact globular-stage seeds (Fig. 5A). These data indicate that (1) at least 20,000 to 22,000 diverse mRNAs are required to make a globular-stage soybean seed; (2) the majority of the diverse mRNAs present in each globular-stage seed region are shared with other regions; and (3) there are small sets of seed region-specific mRNAs. We used hierarchical clustering to determine whether sets of mRNAs that are shared between different soybean globular-stage seed regions are coregulated at a quantitative level (Fig. 5, E and F). Our analyses identified groups of shared coregulated mRNAs that accumulate at a higher level in a particular seed region (Fig. 5E). For example, approximately 65 suspensor transcripts accumulate at a 2-fold or higher level in the suspensor compared with other seed regions (Fig. 5F). Finally, comparison of diverse mRNAs present in each region of a soybean globular-stage seed identified small sets of region-specific transcripts. For example, 74 mRNAs were detected in the suspensor that are undetectable in other seed regions at the level of the GeneChip (Fig. 5C; J.A. Wagmaister, X. Wang, A.Q. Bui, B.H. Le, and R.B. Goldberg, unpublished data). Real-time qRT-PCR experiments validated the GeneChip specificity of two of these suspensor transcripts—an osmotin-like 34 mRNA and a mRNA encoding a member of the NAM transcription factor family (Fig. 5G; C. Cheng, J.A. Wagmaister, A.Q. Bui, and R.B. Goldberg, unpublished data). Taken together, this example illustrates that it is possible to combine LCM and GeneChip technologies to profile the spectrum of mRNAs that are present in any legume seed compartment and region throughout development (Fig. 3). The challenge will be to identify which mRNAs play a critical role in the differentiation of each seed region and how their corresponding genes are organized into regulatory networks in the soybean genome (Fig. 3).
Legumes exhibit a wide range of diversity in seed size and embryo morphology (Fig. 2), providing an outstanding opportunity to use LCM and functional genomics to compare the mRNA sets present in the different legume seeds. For example, it should be possible to isolate and compare the RNA sets present in legume suspensors that vary greatly in size and form (e.g. SRB, Medicago, Lotus, and soybean; Fig. 2B) and address the question of what role, if any, variation in suspensor size plays in legume embryo development. Although it is unlikely that GeneChips will be constructed containing diverse mRNAs for each legume species, it is possible to take advantage of the close relationship between legumes at the DNA and RNA levels to use cross-species hybridization approaches to compare gene activity within the seed regions of any legume. Cross-species hybridization approaches have been applied successfully in plants, animals, and fungi where species have diverged from a common ancestor more than 75 million years ago, such as pig and human (Chismar et al., 2002 ; Moody et al., 2002; Adjaye et al., 2004; Becher et al., 2004; Ji et al., 2004; Wang et al., 2004; Chalmers et al., 2005; Nowrousian et al., 2005). This evolutionary distance is greater than that separating SRB, Medicago, Lotus, and soybean, which have been shown to diverge from a common ancestor approximately 54 million years ago (Fig. 6; Lavin et al., 2005). To test this approach, we carried out cross-species hybridization using laser-captured SRB suspensors and embryo-proper RNAs hybridized with soybean GeneChips (Fig. 4, G and H). Our results indicated that most diverse SRB embryo mRNAs that are conserved enough to be detected by the soybean GeneChip are shared by SRB embryo-proper and suspensor regions (Fig. 4H). By contrast, small sets of mRNAs, including those encoding transcription factors, are specific to each region of the globular-stage SRB embryo at the level of the GeneChip (Fig. 4H). The results from these LCM cross-species hybridization studies complement those generated by sequencing hand-dissected SRB embryo-proper and suspensor ESTs (Figs. 3 and 4D). In addition, they identified 1,000 new embryo proper- and suspensor-specific mRNAs, including those encoding approximately 60 transcription factors that might play key roles in embryo region specification during early embryogenesis (Fig. 4A). Our data indicate that cross-species hybridization using mRNAs from diverse legumes can be successful in identifying genes active during seed development at a global level. Coupling LCM with cross-species hybridization using available legume GeneChips and microarrays (e.g. soybean) should provide an entry point for identifying genes that play important roles in the development of model legume seeds, such as soybean, Medicago, and Lotus, as well as in those of nonmodel legumes where few genomic resources are available (Fig. 2B).
Using the genomics strategy that we developed to study the early stages of legume seed and embryo development (Fig. 3), we identified genes, including those encoding transcription factors, that are active specifically in the embryo proper and suspensor of SRB and soybean globular-stage embryos (Figs. 4, D and H, and 5, D and G). We also identified genes that are active specifically in other compartments of the seed (e.g. endosperm, integuments, hilum; Fig. 5C). What DNA sequences and transcription factors regulate compartment-specific genes within a seed and how compartment-specific genes are organized into regulatory networks within a plant genome are important questions of seed biology (Fig. 1D).
As a first step to uncover regulatory networks that operate within legume seeds, we used in situ hybridization to identify mRNAs from our EST database that accumulate specifically in the SRB suspensor (Figs. 3 and 4). For example, G564, C541, PCS1511, and PCEP3567 mRNAs accumulate at a high level in the suspensor of SRB globular-stage embryos (Fig. 6, A–D; Weterings et al., 2001
We showed that approximately 4.2 kb of the G564 upstream region activates transcription in the suspensor of transgenic tobacco globular-stage embryos, demonstrating that suspensor-specific expression is controlled primarily at the transcriptional level (Fig. 6E; Weterings et al., 2001
The spectacular increase in legume genome sequences enables comparative approaches to be used to identify conserved cis-regulatory sequences among related legume species. For example, we uncovered G564 orthologs in soybean, Lotus, and Medicago (Fig. 6, H and I; T. Kawashima and R.B. Goldberg, unpublished data). Soybean separated from SRB approximately 19 million years ago and from Lotus and Medicago approximately 54 million years ago (Fig. 6H; Lavin et al., 2005 One or more of the suspensor-specific transcription factors that we identified using EST and LCM-GeneChip analyses (Figs. 4 and 5) might interact with the conserved 10-bp motif and other cis-regulatory sequences to control transcription in the suspensor of SRB and other legumes. Similarly, transcription factors specific to other seed compartments (Fig. 5) might play an important role in controlling transcription in different parts of the seed. To date, the molecular mechanisms by which region-specific transcription factors are interconnected to form seed regulatory networks remain unknown. Studying the function of region-specific transcription factors is essential for understanding the importance of these proteins in seed development and for uncovering downstream target genes to construct seed gene regulatory networks (Fig. 3).
Advances in soybean transformation procedures (Ko et al., 2006
The study of legume seed development has become exciting due to the availability of new genomic resources and sophisticated techniques, such as LCM and RNA profiling using GeneChip arrays. We have identified genes that are unique to a particular seed compartment and that are coregulated within the context of the soybean globular-stage seed (Fig. 5, C, E, and F). The completion of the soybean genome sequence (Jackson et al., 2006 ) should allow us to identify conserved motifs among the upstream regions of these unique and coregulated genes, thus facilitating the identification of compartment-specific cis-regulatory sequences that connect seed genes into regulatory networks. In addition to the soybean genome sequence, other legume genome sequences will be available soon (Broughton et al., 2003; Gepts et al., 2005; Young et al., 2005). Genome sequences from diverse legume species will provide an invaluable resource for comparative analysis to identify conserved cis-regulatory sequences that connect seed genes into regulatory networks, an approach that has been successful in other eukaryotes, such as the sea urchin (Yuh et al., 2002; Bolouri and Davidson, 2003). Comparative analysis between legume genome sequences, combined with the GeneChip and EST data obtained from seed and embryo compartments of SRB and soybean (Figs. 4 and 5) and other plants (e.g. Arabidopsis; Casson et al., 2005), should facilitate the discovery of genes essential for seed and embryo development, including those important for specific legume traits, such as seed size and embryo morphology (Fig. 2). In addition, comparative analysis of legume genomes with other nonlegume genomes, such as Arabidopsis, rice, and poplar (Populus spp.), will advance the discovery of genes important for seed development in flowering plants (Graham et al., 2004; Zhu et al., 2005). Finally, once a soybean whole-genome GeneChip becomes available, the entire mRNA profiles of all seed and embryo compartments can be determined, completing the identification of seed- and embryo-specific genes. Taken together, remarkable advances in genomic resources will allow us to answer questions regarding seed and embryo development (Fig. 1D) that were not possible only a few years ago. It is now becoming realistic in this genomic era to understand what genes and regulatory networks are required to make a legume seed. Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers CA896559 to CA916678 (ESTs) and AF325187 (G564).
We are grateful to all the members of our laboratory, past and present, who have helped to establish soybean and SRB as powerful systems to investigate seed development. We particularly acknowledge Dr. Koen Weterings, Dr. Yuping Bi, and Dr. Xingjun Wang for contributing to many of the experiments summarized in this Update. In addition, we thank Ms. Chen Cheng for carrying out the real-time qRT-PCR experiment presented in Figure 5G. We would also like to acknowledge Ian Sussex, Roger Beachy, Tim Hall, Maarten Chrispeels, Niels Nielsen, Lila Vodkin, Don Boulter, T.J. Higgins, Klaus Muntz, and Uli Wobus whose laboratories helped to provide a foundation for understanding gene activity during legume seed development. Received March 28, 2007; accepted April 18, 2007; published June 6, 2007.
1 This work was supported by the National Science Foundation Plant Genome Program (grant no. DBI–0501720), the Department of Energy (grant no. DE–FG03–97ER20263), and Ceres Inc. T.K. is a recipient of a Nakajima Foundation predoctoral fellowship. 
2 These authors contributed equally to the article. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Robert B. Goldberg (bobg{at}ucla.edu). www.plantphysiol.org/cgi/doi/10.1104/pp.107.100362 * Corresponding author; e-mail bobg{at}ucla.edu; fax 310–825–8201.
Adams CA, Norby SW, Rinne RW (1982) Protein modification and utilization of starch in soybean (Glycine max (L.) Merr.) seed saturation. J Exp Bot 33: 279–287 Adjaye J, Herwig R, Herrmann D, Wruck W, Benkahla A, Brink TC, Nowak M, Carnwath JW, Hultschig C, Niemann H, et al (2004) Cross-species hybridisation of human and bovine orthologous genes on high density cDNA microarrays. BMC Genomics 5: 83[CrossRef][Medline] Alonso JM, Stepanova AN, Leisse TJ, Kim CJ, Chen HM, Shinn P, Stevenson DK, Zimmerman J, Barajas P, Cheuk R, et al (2003) Genome-wide insertional mutagenesis of Arabidopsis thaliana. Science 301: 653–657 Alpi A, Lorenzi R, Cionini PG, Bennici A, Damato F (1979) Identification of gibberellin A1 in the embryo suspensor of Phaseolus coccineus. Planta 147: 225–228[CrossRef][ISI] Amore G, Davidson EH (2006) cis-Regulatory control of cyclophilin, a member of the ETS-DRI skeletogenic gene battery in the sea urchin embryo. Dev Biol 293: 555–564[CrossRef][ISI][Medline] |
TOP
MAJOR QUESTIONS REMAIN...
LEGUMES ARE AN EXCELLENT...
-conglycinin, glycinin, Kunitz trypsin inhibitor, and lectin, showed that their mRNA accumulation patterns are regulated temporally and spatially (Goldberg et al., 1981a
