Identiﬁcation of genes expressed in the angiosperm female gametophyte

Until recently, identiﬁcation of gene regulatory networks controlling the development of the angiosperm female gametophyte has presented a signiﬁcant challenge to the plant biology community. The angiosperm female gametophyte is fairly inaccessible because it is a highly reduced structure relative to the sporophyte and is embedded within multiple layers of the sporophytic tissue of the ovule. Moreover, although mutations affecting the female gametophyte can be readily isolated, their analysis can be difﬁcult because most affect genes involved in basic cellular processes that are also required in the diploid sporophyte. In recent years, expression-based approaches in multiple species have begun to uncover gene sets expressed in speciﬁc female gametophyte cells as a means of identifying regulatory networks controlling cell differentiation in the female gametophyte. Here, recent efforts to identify and analyse gene expression programmes in the Arabidopsis female gametophyte are reviewed.

The female gametophyte is essential for angiosperm reproduction The plant life cycle alternates between a haploid gametophyte generation and a diploid sporophyte generation. In plants, specialized diploid cells (mother cells) within sporophytes undergo meiosis and give rise to haploid spores. Spores undergo cell proliferation and differentiation to develop into multicellular haploid gametophytes. A major function of the gametophyte generation is to produce haploid gametes. Fusion of egg and sperm gives rise to the sporophyte, thereby completing the life cycle (Raven et al., 2005).
In lower plants, gametophytes typically are the dominant and free-living generation. In contrast, angiosperms have dramatically reduced gametophytes that are composed of very few cells and are embedded within the sexual organs of the flower. Angiosperms have two gametophytes: the female gametophyte (also referred to as the embryo sac or megagametophyte) and the male gametophyte (also referred to as the pollen grain or microgametophyte). The male gametophyte develops within the stamen's anther and is composed of two sperm cells encased within a vegetative cell (McCormick, 2004;Singh et al., 2008;Borg and Twell, 2010). The female gametophyte develops within the ovule, which is found within the carpels of the flower. The angiosperm female gametophyte most commonly consists of one egg cell, one central cell, two synergid cells, and three antipodal cells (Drews and Yadegari, 2002;Colombo et al., 2008a;Sundaresan and Alandete-Saez, 2010;Yang et al., 2010).
The female and male gametophytes are essential for the angiosperm reproductive process. Sexual reproduction is initiated when pollen is transferred from anther to stigma. Soon thereafter, the male gametophyte forms a pollen tube that grows through the carpel's internal tissues and eventually releases its two sperm cells into one of the two synergid cells. Soon after arrival, the pollen tube ceases growth and the synergid cell penetrated by the pollen tube undergoes cell death. Next, one sperm cell migrates to the egg cell and the other to the central cell. When the plasma membranes of the male gametes fuse with those of the egg and central cells, the sperm nuclei are transmitted into these cells for karyogamy. Following fertilization, the egg and central cell give rise to the seed's embryo and endosperm, respectively (Lord and Russell, 2002;Weterings and Russell, 2004;Berger et al., 2008;Sprunck, 2010).
Female gametophyte cells control many steps of the fertilization process. During pollen tube growth, the synergid cells produce a guidance cue that directs pollen tube growth to the ovule and female gametophyte Okuda et al., 2009). Female gametophyte cells (mainly the synergid cells) contain factors that control arrest of pollen tube growth and release of pollen tube contents (Huck et al., 2003;Rotman et al., 2003Rotman et al., , 2008Escobar-Restrepo et al., 2007;Capron et al., 2008;Boisson-Dernier et al., 2009;Miyazaki et al., 2009;Tsukamoto et al., 2010). Upon fertilization, the ovule is induced to develop into a seed. Central cell-expressed gene products control the activation of endosperm development via epigenetic mechanisms (Kohler and Grossniklaus, 2005;Huh et al., 2008;Berger and Chaudhury, 2009;Jullien and Berger, 2009). Also, factors present in the egg cell may regulate the activation of embryo development, although these factors have yet to be identified (Curtis and Grossniklaus, 2008). Finally, genetic studies indicate that the female gametophyte plays a role in maternal control of embryo and endosperm development following fertilization (Evans and Kermicle, 2001;Grini et al., 2002).
The mechanisms that control the differentiation of the female gametophyte and its individual cell types remain to be determined. However, recent work using molecular and genetic approaches has identified auxin as a critical signal for the proper development of the asymmetric structure of the female gametophyte in Arabidopsis (Pagnussat et al., 2009). Independent processes have also been shown to control cell identity particularly in the micropylar cells of the female gametophyte (Gross-Hardt et al., 2007;Pagnussat et al., 2007;Moll et al., 2008). Recent efforts in deciphering the regulatory networks that control female gametophyte cell differentiation and development are summarized below.
Identification of genes expressed in the Arabidopsis female gametophyte As discussed above, each of the female gametophyte's four cell types contains unique structural, developmental, and physiological features. Our goal is to understand the gene regulatory circuitry operating in these cells that causes them to acquire their unique features and functions during cell differentiation. As a first step toward dissecting these generegulatory networks, genes expressed in the Arabidopsis female gametophyte were identified using differential expression screens.
The general strategy used to identify female gametophyteexpressed genes was to identify mRNAs present in normal ovules but not in mutant ovules lacking female gametophytes. male sterility1 (ms1) (Thorlby et al., 1997;Wilson et al., 2001;Ito et al., 2007) was used as the source of normal ovules, and determinant infertile1 (dif1) (Bai et al., 1999;Bhatt et al., 1999;Cai et al., 2003) as the source of mutant ovules lacking female gametophytes. ms1 is male sterile but undergoes normal ovule and female gametophyte development (Steffen et al., 2008). dif1 is a recessive sporophytic mutation and all ovules lack female gametophytes in homozygous dif1 mutants. The absence of a female gametophyte in dif1 ovules results from a meiosis defect: DIF1 encodes a homologue of yeast REC8/RAD21 cohesin genes and the dif1 mutation affects chromosomal segregation during meiosis (Bai et al., 1999;Bhatt et al., 1999;Cai et al., 2003).
Female gametophyte-expressed genes were screened for with Affymetrix Arabidopsis ATH1 Genome Arrays using probes synthesized from the ms1 and dif1 ovule RNAs. These assays identified 86 genes whose expression was potentially enhanced in wild-type ovules. To validate this expression pattern, real-time reverse transcription-PCR (RT-PCR) assays were carried out with RNA from independently harvested ovules. Using the criteria of >8-fold change, 71 of the 86 genes were validated as having reduced expression in dif1 ovules (Steffen et al., 2007).
To identify the female gametophyte cells that express these genes, 45 of the 71 genes were selected and transgenic Arabidopsis lines containing promoter:GFP (Green Florescent Protein) gene fusions were analysed (Steffen et al., 2007). Of the 45 genes analysed, 43 showed expression in the female gametophyte. The female gametophyte expression patterns of these genes fell into five general categories: expression in the egg cell only (one gene), expression predominantly or exclusively in the synergid cells (17 genes), expression predominantly or exclusively in the central cell (11 genes), expression in the antipodal cells only (11 genes), and expression in multiple female gametophyte cell types (three genes).
Similar approaches have been used by other groups to identify female gametophyte-expressed genes (Yu et al., 2005;Johnston et al., 2007;Jones-Rhoades et al., 2007). More recently, these analyses have been complemented by the identification of individual-cell transcriptomes of the Arabidopsis female gametophyte using laser-capture microdissection coupled to microarray analysis (Wuest et al., 2010). A summary of these expression patterns is shown in Table 1.

Identification of transcription factor genes expressed in the female gametophyte
The microarray screen used detected only one gene encoding a transcription factor (Steffen et al., 2008). Although there has been no systematic study of transcription factor gene mRNA levels in plants, most mammalian and yeast transcription factor genes produce mRNAs that occur at low levels (Holland, 2002;Vaquerizas et al., 2009). Thus, the low number of transcription factor genes identified in the above-mentioned microarray-based approach was probably due to insufficient sensitivity. This is especially relevant to the general approach used as the female gametophyte cells comprise only seven of ;1,000 cells in a mature ovule-in effect, diluting the female gametophyte RNAs ;150-fold. As an alternative, in a second screen, real-time RT-PCR was used to detect individual transcription factor gene mRNAs in ovules or ovaries followed by reporter gene assays for analysis of cell-specific expression (Kasahara et al., 2005;Steffen et al., 2008;Wang et al., 2010).
Real-time RT-PCR assays were performed with nearly 1, 400 Arabidopsis transcription factor genes using RNA from ms1 and dif1 ovules or ovaries (Wang et al., 2010). These assays identified 69 genes with elevated mRNA levels in ms1 as compared with dif1 mutant ovaries. Using additional biological replicates and more stringent statistical analyses, 26 genes were confirmed to be down-regulated in dif1 ovaries (Wang et al., 2010).
To determine the female gametophyte cells within which these genes are expressed, promoter fusion analyses with single or double nuclear-localized GFP constructs were used (Wang et al., 2010). This approach was taken because many transcription factor genes produced weak or ambiguous levels of promoter-reporter activity using standard, untargeted, cytosolic GFP (cGFP) as a reporter (Wang et al., 2010). Moreover, for genes expressed in multiple cell types within the female gametophyte, determination of the exact patterns using cGFP reporters proved difficult due to low resolution of the cGFP signal. This was particularly the case for genes with reporter activity in the egg cell, the central cell, and/or the synergid cells.
Using promoter:GFP gene fusions, the expression of 18 transcription factor genes was assayed (Wang et al., 2010). Of these, 17 showed expression in the female gametophyte. Ten genes were exclusively or predominantly expressed in the egg cell (two genes), the central cell (two genes), or the antipodal cells (six genes), whereas the remaining seven genes were expressed in two (four genes) or more (three genes) cell types, usually with one cell type showing a higher level of expression. The expression of a subset of these genes was also analysed during megagametogenesis and early seed development. For example, 12 genes were transcriptionally active in the developing embryo and/or endosperm.
These real-time RT-PCR assays identified a number of transcription factor genes that had not been identified in other published studies, presumably because their mRNA levels are too low for effective detection using standard microarray-based approaches (Wang et al., 2010). Therefore, a combined expression-based analysis of regulatory and non-regulatory gene sets that are expressed in specific cell types (Table 1) should enable identification of the

Analysis of the synergid cell and central cell gene networks
Functional studies have been carried out with two of the transcription factor genes identified in the screens discussed above. These genes are MYB98 (Kasahara et al., 2005;Punwani et al., 2007Punwani et al., , 2008 and AGL61/DIANA (Bemer et al., 2008;Steffen et al., 2008). As discussed below, analysis of these genes provided initial insight into the gene networks of the synergid and central cells.
MYB98 encodes an R2R3 MYB protein. In the context of the ovule, the MYB98 gene is expressed predominantly in the synergid cells and myb98 mutations affect the synergid cells specifically. myb98 synergid cells fail to attract pollen tubes and have an abnormal filiform apparatus, which is a thickened and elaborated cell wall located at the micropylar poles of these cells. However, with the exception of the filiform apparatus defect, myb98 synergid cells are otherwise normal. These observations suggest that MYB98 regulates a subcircuit of the synergid gene regulatory network that functions to activate the expression of genes required for pollen tube guidance and filiform apparatus formation (Kasahara et al., 2005;Punwani et al., 2007).
To identify genes regulated by MYB98, the synergidexpressed genes identified in the microarray screen were tested for reduced expression in myb98 female gametophytes. Specifically, the 17 genes expressed predominantly or exclusively in the synergid cells plus one gene expressed highly in both the synergid cells and the central cell were tested. Of the 18 genes tested, 16 exhibited reduced expression in myb98 ovules. Most of these genes encode cysteine-rich proteins within four gene families (CRP810, CRP3700, CRP3730, and CRP3740). Further analysis by us and others (Jones-Rhoades et al., 2007) showed that the MYB98 subcircuit comprises at least 83 downstream genes including 48 genes within the CRP810, CRP3700, CRP3730, and CRP3740 families. All genes tested exhibited no mutant phenotype, most probably due to functional redundancy. However, five of six proteins tested localized to the filiform apparatus, suggesting that they play a role in formation or function of this structure (Punwani et al., 2007(Punwani et al., , 2008.
The DNA-binding properties of MYB98 were also examined in vitro and in vivo. Using electrophoretic mobility shift assays, it was shown that MYB98 binds to a specific DNA sequence (TAAC) in vitro. Through promoter dissection of several of the downstream genes, it was demonstrated that the in vivo MYB98 binding site is GTAACNT (N¼any nucleotide). Many of the genes downstream of MYB98 contain the GTAACNT element, and site-directed mutagenesis of these elements abolishes synergid expression, very strongly suggesting that the corresponding genes are direct targets of MYB98 (Punwani et al., 2007(Punwani et al., , 2008. The second transcription factor gene analysed was AGL61, which encodes a type I MADS transcription factor. During ovule and seed development, AGL61 is expressed exclusively in the central cell and early endosperm. The agl61 mutation affects the central cell specifically. agl61 central cells exhibit an overall reduction in size and a reduced or absent vacuole. agl61 central cells fail to give rise to endosperm when fertilized with wild-type pollen. These observations suggest that AGL61 controls the expression of central cell-expressed genes that are required for development of the central cell into endosperm following fertilization (Bemer et al., 2008;Steffen et al., 2008).
MADS-domain proteins often function as heterodimers with other MADS-domain proteins (de Folter and Angenent, 2006). The expression pattern and mutant phenotype of AGL61 are similar to those of AGL80, which were previously analysed (Portereiko et al., 2006). These observations suggest that AGL61 may function as a heterodimer with AGL80 within the central cell. AGL61-AGL80 interactions were determined in yeast two-hybrid assays and in bimolecular fluorescence complementation assays using Arabidopsis protoplasts. In both assays, a specific interaction was observed, suggesting very strongly that an AGL61-AGL80 heterodimer functions in the central cell (Bemer et al., 2008;Steffen et al., 2008).
To identify genes regulated by AGL61 and AGL80, several of the central cell-expressed genes identified in the microarray screen were tested for reduced expression in agl61 and agl80 female gametophytes. Of four genes tested, three showed reduced expression in both agl61 and agl80 central cells. During the course of these studies, it was also found that several synergid-and antipodal-expressed genes are ectopically expressed in agl61 and agl80 central cells. Together, these data suggest very strongly that an AGL61-AGL80 heterodimer functions to both activate and repress genes during central cell development (Bemer et al., 2008;Steffen et al., 2008).

Summary
Each of the female gametophyte's four cell types contains unique features that are essential for angiosperm reproduction. As a first step toward understanding the gene networks that cause these cells to acquire their unique features and functions during cell differentiation, large numbers of both regulatory and non-regulatory gene sets expressed in the Arabidopsis female gametophyte were identified (summarized in Table 1). Functional analysis of two of these genes, MYB98 and AGL61, has provided an initial view of the gene networks operating in the synergid and central cells. Analysis of the gene networks operating in the other female gametophyte cells is currently in progress.