|II.||Signal transduction during ovule and embryo sac development||14|
|III.||Signalling during double fertilization||20|
|II.||Signal transduction during ovule and embryo sac development||14|
|III.||Signalling during double fertilization||20|
Cell–cell communication pervades every aspect of the life of a plant. It is particularly crucial for the development of the gametes and their subtle interaction leading to double fertilization. The ovule is composed of a funiculus, one or two integuments, and a gametophyte surrounded by nucellus tissue. Proper ovule and embryo sac development are critical to reproductive success. To allow fertilization, the correct relative positioning and differentiation of the embryo sac cells are essential. Integument development is also intimately linked with the normal development of the female gametophyte; the sporophyte and gametophyte are not fully independent tissues. Inside the gametophyte, numerous signs of cell–cell communication take place throughout development, including cell fate patterning, fertilization and the early stages of embryogenesis. This review highlights the current evidence of cell–cell communication and signalling elements based on structural and physiological observations as well as the description and characterization of mutants in structurally specific genes. By combining data from different species, models of cell–cell interactions have been built, particularly for the establishment of the germline, for the progression through megagametogenesis and for double fertilization.
Sexual reproduction is a key element in the life cycle of higher plants. Within the flower, the stamens produce the pollen, in which the haploid sperm cells (1n) are contained (Fig. 1). The carpel is the female organ in which double fertilization, the interaction of one sperm cell with the egg cell and a second sperm cell with the central cell of the embryo sac, takes place. The carpel includes the stigma, which promotes the deposition and germination of pollen; the style, which plays an essential role in the elongation and guidance of the pollen tube (PT); and the ovary, which protects the ovules. Within the ovules, a highly organized structure called an embryo sac includes the four different cell types required to complete the double fertilization. These are the egg cell (1n) and central cell (2n), which will develop into an embryo (2n) and the endosperm (3n), respectively, after fertilization, as well as synergids (1n) and antipodals (1n). Each ovule becomes a seed, and each ovary a fruit. The seed contains the embryo that will, under the correct conditions, germinate to generate a new organism.
Proper ovule and embryo sac development are critical to reproductive success. The multiple cell layers or cell types of the ovule and the embryo sac form the various components of the future seed. Several cell–cell interactions must take place within the ovule: first, to ensure the appropriate division and differentiation of cells and tissues to generate a complete and fertile embryo sac; secondly, to synchronize the double-fertilization process; and thirdly, to assure the development of a viable seed. This review will cover the various signalling elements identified thus far as affecting plant reproduction with respect to the ovule, as well as evidence of cell–cell communication discovered via the phenotypic characterization of several mutants.
In Arabidopsis thaliana, the mature gynoecium consists of two fused carpels whose locules are separated by a central septum. The central region of the septum forms the transmitting tissue, while its peripheral cells are fused with the inner wall of the ovary. Placental tissue differentiates along the septum adjacent to the lateral wall of the ovary. The ovule primordia begin to emerge from the placenta during stage 9 of floral development (Smyth et al., 1990; Robinson-Beers et al., 1992). The inner and outer integuments then appear on the surface of each primordium, originating from the chalaza, which separates the apical nucellus from the funiculus. The integuments grow and cover the nucellus, leaving a small opening at the apical end, the micropyle. The funiculus contains vasculature which supplies nutrients to the ovule and the embryo, and which, in part, determines the orientation of the micropyle. The integuments are required to protect the embryo sac, contribute to the positioning of the ovule and eventually form the seed wall. The nucellus provides the initial cell, which will differentiate into a megasporocyte or megaspore mother cell (MMC), starting the germline (Fig. 2a) (Bowman, 1994; Skinner et al., 2004). Several transcription factors have been characterized as playing a role in the development and cell fate patterning of the ovule. Although several genetic interactions have been identified, their regulation remains largely unknown. At this level, the signalling pathways involved in the cell fate patterning of the ovule are poorly understood. Nevertheless, cell–cell communication between cell layers of the nucellus has been documented.
The integuments differentiate from the L1 cell layer at the chalazal end of the primordium (Fig. 2a). In Arabidopsis, however, the discovery of the receptor-like kinase STRUBBELIG (SUB) shows that the L2/L3 cell layers are involved in the differentiation of the outer integuments. In fact, sub-1 ovules, characterized by an incomplete coverage of the outer integument resulting in a finger-like clamps surface, exhibit disturbances during the initiation of these tissues (Chevalier et al., 2005). The SUB protein is only present in L2/L3 cell layers. Expression of SUB under the control of specific promoters (AINTEGUMENTA and WUSCHEL) in the nucellus and L2/L3 cell layers is able to recover the sub-1 mutant phenotype. In some cases, WUS::SUB:GFP is restricted to the nucellar epidermis, and the sub-1 mutant phenotype is not rescued (Yadav et al., 2008). These results demonstrate that SUB acts via a noncell autonomous pathway and is involved in a signalling network between the inner cell layers of the primordium required for the initiation of the outer integuments (Fig. 3).
The search for Arabidopsis mutants with phenotypes similar to sub-1 led to the identification of three new genes: DETORQUEO, QUIRKY (QKY) and ZERZAUST (Fulton et al., 2009). The analysis of all combinations of double mutants revealed that the three had considerable functional redundancy, but also specific functions. The genes do not seem to be epistatic and acting in a simple linear signalling pathway, but their complex phenotypes make it difficult to assess the additive or synergistic effects among them. QKY encodes a protein with four cytoplasmic C2 domains that is anchored to the membrane by its C-terminal domain. Despite the limited sequence homology, QKY appears to belong to the MCTP (multiple C2 domain proteins and transmembrane region) family. MCTPs play a role in both exocytosis and membrane repair, functions that are conserved throughout several of the living kingdoms (Fulton et al., 2009). Thus, a model was proposed (Fig. 3) in which the receptor kinase SUB may affect QKY activity in the inner layers of the primordium and regulate the exocytosis of some factors affecting the cell fate of the L1 cell layer (Fulton et al., 2010).
While the sequence of events during embryo sac development can vary among species, Polygonum-type development is the most commonly observed route in angiosperms (Friedman & Ryerson, 2009). The female gametophyte, or embryo sac, is formed by the differentiation of a nucellus cell into an archesporial cell (2n), which then acquires the MMC identity (2n). The MMC undergoes meiosis to form a tetrad of megaspores (1n). Three of these cells, usually those at the distal end of the primordium, will degenerate, and the fourth will become the functional megaspore. These steps constitute megasporogenesis (Fig. 2b) (Christensen et al., 1997). The functional megaspore (FG1) then goes through three successive rounds of mitosis without cytokinesis, generating the eight nuclei necessary to complete the differentiation of a mature embryo sac. This syncytium (FG5) has a polarity linked to the positioning of nuclei during mitosis. Some nuclei migrate, reposition themselves and merge: six haploid cells and one diploid cell are formed after cellularization, each with a specific cellular identity. The embryo sac has two synergid cells (1n) and one egg cell (1n) at its micropylar pole, plus a central cell (2n) and three antipodal cells (1n) at its chalazal pole. The antipodal cells may degenerate to form a mature embryo sac of four cells (FG7). These steps constitute megagametogenesis (Christensen et al., 1997).
A sporophytic cell in the nucellus must differentiate into an archesporial cell to start a germline. Several cells appear to acquire this cellular identity within the primordium. Indeed, in rice, the MEIOSIS ARRESTED AT LEPTOTENE1 (MEL1) gene, which encodes an ARGONAUTE-like protein, is expressed in all cells of the inner layers of the nucellus; its expression is then limited to the MMC and down-regulated once the first meiosis starts. The mel1 mutant embryo sacs do not complete megasporogenesis, but show no abnormality in the initiation and establishment of the germline (Nonomura et al., 2007). Thus, the inner cell layers of the nucellus differentiate into archesporial cells, which potentially become immature MMCs, although only one becomes the mature MMC. This model is valid if cell–cell communication occurs; for example, signal emission by the mature MMC that abolishes the acquisition of MMC identity by neighbouring cells (Fig. 4). But which signalling molecules could potentially be involved in this lateral inhibition mechanism?
In rice (Oryza sativa), MULTIPLE SPOROCYTE 1 (MSP1) encodes a receptor-like kinase with expression limited to cells surrounding the MMC. In the msp1 loss-of-function mutant, several MMCs appear in the nucellus. This suggests that MSP1 activates a signalling pathway in companion archesporial cells that inhibits MMC differentiation (Nonomura et al., 2003). TAPETUM DETERMINANT-LIKE 1A (OsTDL1A) is a peptide binding the MSP1 receptor. OsTDL1A is coexpressed with MSP1 in the ovule, and interference with OsTDL1A expression causes the appearance of several MMCs within the nucellus (Zhao et al., 2008). The OsTDL1A/MSP1 interaction may therefore be involved in a cell autonomous pathway that restricts the number of MMCs. It is possible, however, that other, unidentified, partners may bind to this complex and participate in lateral inhibition.
In maize (Zea mays), MULTIPLE ARCHESPORIAL CELL 1 (MAC1) is required for the establishment of the female germline. Several MMCs develop in the mac1 mutant and undergo meiosis to form tetrads. Within a tetrad, several megaspores can survive and generate embryo sacs, potentially causing infertility. Although the mutation has been localized on chromosome 10, the nature of the gene is unknown (Sheridan et al., 1996). While the phenotype associated with MAC1 is similar to that of MSP1, the MAC1 gene is not the MSP1 orthologue in maize (Ma et al., 2007). Nevertheless, MAC1 remains a good candidate to support the model of lateral inhibition exerted by the MMC.
Signalling by small RNAs also affects the establishment of the female germline. In Arabidopsis, AGO9 is another ARGONAUTE-like protein which is expressed in the L1 cell layer of the nucellus. AGO9 is associated with small RNAs (ta-siRNAs) that specifically target long primary RNAs (TAS genes) not predicted to encode proteins (Vaucheret, 2006). The absence of AGO9 leads to additional functional megaspores in the nucellus, which directly differentiate out of a sporophytic cell. Likewise, mutations in RNA-DEPENDENT RNA POLYMERASE 6 (RDR6) and SUPPRESSOR OF GENE SILENCING 3 (SGS3), proteins involved in both the biogenesis of ta-siRNAs and their activity in adjacent cells, cause the same phenotype (Olmedo-Monfil et al., 2010). Together, these results support cell–cell communication between the L1 cell layer and somatic companion cells by transport of small RNAs, limiting the number of gametic cells within the ovule. Isolation of the small RNAs showed that their primary targets are transposable elements (Olmedo-Monfil et al., 2010).
Aside from the nuclear protein SPOROCYTELESS (SPL), which initiates sporogenesis in Arabidopsis, we still do not know which signal initiates the establishment of the MMC (Yang et al., 1999). Cell–cell communications are numerous and diverse, and several mechanisms are in place to limit the proliferation of the germline. Are those pathways sufficient in limiting the establishment of multiple MMCs? It is unlikely, as the neighbouring cells must receive a signal indicating that an adjacent cell has initiated megasporogenesis.
After meiosis, in most flowering plants, the chalazal megaspore differentiates into the functional megaspore, while others enter into programmed cell death (Fig. 2b). Thus, the germ cells react according to their position within the ovule. The MMC already displays signs of polarity through the positioning of organelles, the deposition of callose, and the arrangement of the microtubule cytoskeleton. This polarity persists during megasporogenesis. The Arabidopsis switch1/dyad mutant produces two diploid megaspores, but only the chalazal one expresses specific markers of the functional megaspore (Motamayor et al., 2000; Ravi et al., 2008). These observations suggest that cell–cell communication involved in polarization takes place between the germline and the neighbouring sporophytic cells at an early stage of megasporogenesis. Again, we do not yet know the components of the possible signalling pathway involved.
The functional megaspore (FG1) undergoes three successive rounds of mitosis to form the eight nuclei essential to the formation of the mature embryo sac (FG7, Fig. 2b). Signalling proteins are also involved in this process. In Solanum chacoense, FRK1, a MAPKKK, is critical to development beyond stage FG1 (Gray-Mitsumune et al., 2006; Lafleur, 2010). In A. thaliana, the protein AGP18, a classical arabinogalactan protein, was found to regulate this same process (Acosta-Garcia & Vielle-Calzada, 2004). Similarly, the CYTOKININ INDEPENDENT 1 (CKI1) signalling pathway and a kinesin-MAPKKK pathway may be important in the complete cellularization and positioning of nuclei during stage FG5.
CKI1 was isolated by activation tagging in an Arabidopsis mutant showing a phenotype associated with a constitutive cytokinin signalling pathway. This histidine kinase receptor (AHK) was initially seen as part of the cytokinin receptor family (Kakimoto, 1996; Hwang & Sheen, 2001; Zheng et al., 2006), but was later shown to be unable to bind cytokinin (Yamada et al., 2001). Infertility occurs in the cki1-5/CKI1 and cki1-6/CKI1 mutants, and the T-DNA alleles are never transmitted through the female gametophyte (Pischke et al., 2002). The abnormalities appear after the FG4 stage; at stage FG5, embryo sacs have an unusual morphology, with additional vacuoles and poor positioning of the nuclei. At stage FG7, some embryo sacs are completely degenerated, while others contain several small vacuoles separated by many cytoplasmic strands (Pischke et al., 2002). Further study of a mutant obtained by transposon insertion, cki1-i/CKI1, suggests that the integrity of the vacuole is affected first; if this structure does not form properly, degeneration of the embryo sac results (Hejatko et al., 2003). Although cytokinin is not directly involved in the progression through stage FG5, signalling proteins downstream of the AHKs, such as the ARABIDOPSIS HISTIDINE PHOSPHOTRANSFER PROTEINS (AHPs) and the ARABIDOPSIS RESPONSE REGULATORS (ARRs), do appear to play a role (Fig. 5a). Thus, CKI1 uses elements downstream from the cytokinin receptors to control female gametophyte development (Deng et al., 2010).
A mitogen-activated protein kinase (MAPK) pathway may also be involved in the cellularization/differentiation which occurs during stage FG5 (Fig. 5b). In tobacco and Arabidopsis, the NACK-PQR pathway, a NACK1 (NPK1-activating kinesin-like protein) and mitogen-activated protein (MAP) cascade, plays an essential role in sporophytic cytokinesis (Krysan et al., 2002; Nishihama et al., 2002; Soyano et al., 2003). In Arabidopsis, the triple mutant for the ANP MAPKKK, anp1anp2anp3, is not transmitted through the male and female gametes (Krysan et al., 2002). The viability of the female gametes is also severely affected in the mkk6-2/anq-2 mutant (Takahashi et al., 2010). AtNACK2/STUD/TETRASPORE and AtNACK1/HINKEL genes, both encoding kinesins, have redundant functions during female gametophyte development; in the double heterozygote mutant atnack1-1/+; atnack2-1/+, 25% of the eggs show a nucleus cluster advanced to the middle of the embryo sac at stage FG5 (Tanaka et al., 2004). Investigation of components of the NACK-PQR MAPK pathway in the completion of female gametophyte development would be interesting, as AtNACK2/STUD/TETRASPORE, AtANP3,AtMKK6 and AtMPK4 define a putative signalling module that regulates male-specific meiotic cytokinesis in Arabidopsis (Zeng et al., 2011).
Cells within the embryo sac must be adequately differentiated to allow for its proper functioning. For example, in the Arabidopsis eostre mutant, there are two egg cells and only one synergid, rather than two synergids and one egg cell. Although this differentiation defect only mildly affects the attraction and guidance of the PT through the micropyle, it can lead to the formation of two zygotes without an endosperm (Pagnussat et al., 2007). Similarly, improper differentiation of the central cell can cause problems during fertilization (Portereiko et al., 2006; Bemer et al., 2008; Steffen et al., 2008). Two models have been proposed to explain the cell fate of nuclei within the female gametophyte: the auxin gradient within the embryo sac and the LACHESIS pathway. These models are not mutually exclusive; both posit that all gametophytic cells have the potential to become a gametic cell. The former deals with cell fate determination, while the latter explains the maintenance of proper differentiation within the embryo sac.
In Arabidopsis, analysis of the wild-type and the eostre mutant supports the idea that the proper positioning of nuclei within the female gametophyte is a prerequisite in the cell fate specification. By the FG5 stage (Fig. 2b), we can predict cell fate specification of the three nuclei located at the extremity of the wild-type syncytium: two nuclei are side by side at the micropylar end, becoming the synergids, while another nucleus is positioned slightly toward the middle of the embryo sac, becoming an egg cell. In the eostre mutant, the positioning of these nuclei is reversed, as is their cell fate (Pagnussat et al., 2007). In the maize indeterminate gametophyte 1 (ig1) mutant, additional divisions occur which are associated with supernumerary cells whose fate also seems to depend on their positioning (Evans, 2007). Thus, the nuclei located at the micropylar end become synergids, followed by the egg cell, central cell and antipodal cells as we move along the longitudinal axis (Fig. 6). This cell fate specification and organization within the syncytium is crucial to facilitate the PT reception, as well as the reception of the sperm cells following the discharge of the PT (Hamamura et al., 2011).
An auxin gradient appears to play a major role in cell fate determination within the syncytium (Pagnussat et al., 2009). Using a reporter gene sensitive to auxin (DR5::GFP), a high concentration of the hormone was visualized at the micropylar end of the Arabidopsis embryo sac. The experimental down-regulation of the AUXIN REGULATING FACTORS (ARFs) within the embryo sac simulates a low concentration of auxin. In this case, the synergid cell fate is compromised and two egg-like cells are formed. Similarly, the quadruple mutant of the auxin F-box receptors tir1afb1afb2afb3 displays defective embryo sacs in which synergid cells express an egg cell marker. The YUCCA auxin biosynthesis genes (YUC) appear to be preferentially expressed on the micropylar side of the syncytium. Their expression throughout the megagametophyte in transgenic overexpression studies stimulates the expression of synergid markers in the egg cell, central cell and antipodal cells, as well as the expression of an egg cell marker within the antipodal cells; no abnormality occurs in the expression of the central cell marker (Pagnussat et al., 2009). Put simply, the signalling pathway of auxin, including its biosynthesis, is required for the differentiation of cells within the embryo sac in a concentration-dependent manner, although it does not affect the positioning of nuclei (Pagnussat et al., 2009).
Angiosperms have developed a second molecular pathway involved in determining cell fate within the embryo sac: a lateral inhibition mechanism involving cell–cell signalling. By screening Arabidopsis mutants that exhibited delocalization of an egg cell marker, lachesis (lis), clotho and atropos were identified (Fig. 6). In these mutants, the synergids and the central cell express an egg cell marker, while the antipodal cells express a central cell marker (Gross-Hardt et al., 2007; Moll et al., 2008). The associated genes encode proteins that are components of the spliceosome, and their interactions allow the localization of the complex into the nucleus. LIS is expressed exclusively in gametic cells (egg and central cell). Thus, the spliceosome may be involved in processing specific genes within the gametic cells, directly or indirectly producing a signal molecule. These molecular signals then inhibit gamete differentiation in the accessory cells and are involved in communication between the egg cell and the central cell (Gross-Hardt et al., 2007; Moll et al., 2008). The nature of these signals is unknown, as are the specific mRNA targets of the spliceosomal proteins.
The role of gametic cells in the specification of the other cells within the embryo sac is also supported by the characterization of various Arabidopsis mutants. The agamous-like61/diana mutant exhibits early degeneration of the central cell after fusion of the polar nuclei (Fig. 6). After degeneration, there is a widespread expression of synergid and egg cell markers throughout the chalazal pole, coupled with a broader expression of antipodal markers at the micropylar pole (Bemer et al., 2008). This supports a major role for the central cell in the determination and maintenance of cellular identity in adjacent cells. More recently, the characterization of the fiona/syco-1 mutant has produced further proof of cell–cell communication inside the embryo sac (Fig. 6). The fiona/syco-1 mutation affects the integrity of the mitochondria of the central cell, as well as the fusion of polar nuclei, without broadly affecting central cell identity. SYCO is specifically expressed in the central cell, but the fiona/syco-1 mutation increases the life span of the antipodals without changing their identity (Kagi et al., 2010), as observed in lis and clotho mutants (Gross-Hardt et al., 2007; Moll et al., 2008). Thus, the mitochondria of the central cell would appear to be required for the fusion of the polar nuclei in these cells, as well as for the control of the programmed cell death of the antipodals. Interestingly, the wild-type paternal allele is sufficient to restore fusion of the polar nuclei, and leads to normal seed development (Kagi et al., 2010).
The fact that auxin first accumulates in the nucellus before forming a gradient in the embryo sac (Pagnussat et al., 2009) supports the idea that there is communication between the sporophytic tissue and the female gametophyte. This cell–cell communication occurs not only during initiation of the germline, but throughout its development. In fact, several studies of Arabidopsis mutants have suggested that the development of the sporophytic and gametophytic tissues during ovule development is synchronized. Recessive mutants with phenotypes related to ovule development are divided into three categories: those with defects in sporophytic and gametophytic tissues, those with defects during megasporogenesis, and those with defects during megagametogenesis (Schneitz et al., 1997). Many gametophytic mutants exist without abnormalities in sporophytic tissue structures, but the reverse condition, sporophytic ovule mutants without abnormalities in gametophyte development, is not observed. This suggests an interaction between the two tissues in the sporophyte to gametophyte direction (Schneitz et al., 1997). For example, in the mutants bel1 and aintegumenta, in which the morphogenesis and identity of the integuments are disrupted (Fig. 6), embryo sac development is arrested (Robinson-Beers et al., 1992). The expression of BELL1 and AINTEGUMENTA is sporophyte-specific (Reiser et al., 1995; Elliott et al., 1996; Klucher et al., 1996); the gametophytic phenotype may be explained by sporophytic signals acting on embryo sac development. At the transcriptional level, however, the absence of an embryo sac causes sporophytic genes to be up-regulated, suggesting that communication is reciprocal (Johnston et al., 2007). Since the presence or absence of the gametophyte does not appear to affect the development of the sporophyte, this suggests the production of signals by the gametophyte to negatively control sporophytic signalling acting on it.
Recently, three Arabidopsis sporophytic cytokinin receptors, CYTOKININ RESPONSE 1 (CRE1), AHK2 and AHK3, have been found to be putative signalling candidates in this sporophyte–gametophyte interaction. The triple mutants of these (cre1-12 ahk2-2tk ahk3-3) principally produce ovules lacking an embryo sac. In a few cases, premature arrest of the female gametophyte is observed or the ovules are slightly smaller compared with wild-type (Kinoshita-Tsujimura & Kakimoto, 2011). The inguments grow normally, suggesting that these receptors facilitate gametophyte development through cell–cell communication with the sporophyte.
Once the embryo sac is fully developed, fertilization can take place. The female reproductive structures are actively involved in the process of double fertilization. Communication between the style and the PT via multiple types of signalling molecules, such as chemocyanin and arabinogalactan proteins in Arabidopsis, guide the PT to the ovule (Cheung et al., 1995; Kim et al., 2003). Sporophytic signal molecules like GABA, can also be produced by the integuments of the ovule to attract PTs (Palanivelu et al., 2003). Signals stemming from the gametophyte, however, are ultimately required to guide the PT through the micropyle, as demonstrated from embryo sac-less mutants (Hulskamp et al., 1995). These signals are also temporally linked to the double fertilization, since fertilized ovules lose their attractiveness, as shown in Torenia fournieri and Arabidopsis (Higashiyama et al., 1998; Palanivelu & Preuss, 2006).
The histological configuration and location of the synergids, as well as the entry point of PTs in the embryo sac, are clues to the role played by the synergids in PT guidance (Huang & Russell, 1992; Russell, 1992, 1996; Higashiyama et al., 1998). In T. fournieri, laser ablation of synergids completely abolishes attraction of PTs, a phenomenon not observed by the removal of gametic cells (egg and central cell) or a single synergid (Higashiyama et al., 2001). In Arabidopsis, the synergid-expressed transcription factor MYB98 was first identified as an underexpressed gene in the mutant determinant infertile 1 (dif1), which does not form an embryo sac (Kasahara et al., 2005). The myb98 mutant shows abnormalities in the integrity of the filiform apparatus and in the attraction of PTs (Kasahara et al., 2005). The discovery of this mutant supports the active role of synergids in PT attraction via the expression of genes encoding attractive peptides under MYB98 regulation (Fig. 7a) (Punwani et al., 2007).
In the majority of cases, ablation of the central cell affects the integrity of the embryo sac. We cannot, therefore, assert its nonparticipation in PT attraction (Higashiyama et al., 2001). The Arabidopsis central cell guidance (ccg) mutant was isolated as a female gametophyte mutant with Mendelian segregation distortion that also affects guidance of the PT (Chen et al., 2007). The CCG gene encodes a transcription factor of the TFIIB family specific to the central cell (Fig. 7a). While differentiation and the integrity of the embryo sac are not affected by the mutation, CCG seems to be involved in the production of signals by the central cell to attract the PT or to coordinate the expression of the chemoattractants from the synergid cells (Chen et al., 2007). However, the fusion of the polar nuclei is not a decisive event regulating PT guidance, since Arabidopsis mutants defective in this step may either fail (Shimizu & Okada, 2000; Shimizu et al., 2008) or succeed (Christensen et al., 2002; Maruyama et al., 2010) in attracting PTs. Nevertheless, when cell fate determination of the central cell does not occur properly, as in the agl61/diana and agl80 insertional mutants, mutant ovules are still able to attract the PT (Portereiko et al., 2006; Bemer et al., 2008; Steffen et al., 2008). This suggests that a basal level of cell fate acquisition is sufficient for PT attraction. A correlation between the level of differentiation of the central cell and the expression of CCG throughout female gametophyte development could bring new insights into this matter. The involvement of the central cell in PT attraction may be indirect, as the central cell and the egg cell can exchange molecules of < 10 kD via the plasmodesmata before anthesis (Han et al., 2000).
Does the female gametophyte play a role in PT repulsion? This phenomenon has not been observed in Arabidopsis mutants myb98 and matagama1/matagama3/ccg, in which it is common to see more than one PT at the entrance of the micropyle in the absence of micropylar guidance (Shimizu & Okada, 2000; Kasahara et al., 2005; Chen et al., 2007; Shimizu et al., 2008). In wild-type Arabidopsis ovules, the behaviour of PTs near the micropyle reflects the initiation of repulsion soon after a PT enters via the micropyle and long before it reaches the female gametophyte (Palanivelu & Preuss, 2006). This indicates that either the sporophyte or the female gametophyte emits a repulsion signal when it detects the PT, or the PT issues this repellent after detecting the attractant molecule (Palanivelu & Preuss, 2006). The signals involved are still unknown.
The cysteine-rich proteins (CRPs) are an important class of small peptides that are abundant in reproductive tissues (Silverstein et al., 2007). In T. fournieri, 16 CRP genes are overrepresented in the synergids. Among them, two peptides, LURE1 and LURE2, are able to attract the PT in a semi-in vivo assay and are immunolocalized at the surface of the filiform apparatus (Fig. 7a). Microinjection of antisense morpholino oligonucleotides against LURE1 and 2 in the central cell of the T. fournieri-protruding embryo sac leads to their diffusion in the neighbouring gametophytic cells and to a decrease in PT attraction (Okuda et al., 2009). In maize, another gene, ZmEA1, expressed by the egg cell and synergids, encodes a nonCRP peptide localized at the filiform apparatus (Fig. 7a). The embryo sacs of plants underexpressing ZmEA1 show no structural abnormalities, but display reduced seed set and decreased PT targeting to the synergids (Marton et al., 2005).
Semi-in vivo approaches have also shown that PT attraction is species-specific and thus may act as an interspecific reproductive barrier (Higashiyama et al., 2006; Palanivelu & Preuss, 2006). Proteins belonging to the CRP class are known to evolve very rapidly and therefore make prime candidates as species-specific chemoattractants (Silverstein et al., 2007), although a number of different signals may have been recruited in plants to fulfil this purpose.
Signalling pathways regulating proper PT targeting following the perception of attracting signals are only beginning to be understood. Two Arabidopsis potassium transporter homologues, CHX21 and CHX23, localized in the PT endoplasmic reticulum, are required for PT targeting to the ovule. Double-mutant (chx21/chx23) PTs germinate and extend normally within the transmitting tract, but fail to turn towards the ovules, having no apparent funicular or micropylar guidance (Lu et al., 2011). A simple model could be that a localized change in pH and/or cation level is directed by transporters acting downstream of a transduction cascade in response to guidance cues (Fig. 7a). The characterization of an Arabidopsis sperm cell-specific transmembrane protein, GCS1/HAP2, suggests the involvement of the sperm cell in PT guidance (von Besser et al., 2006; Mori et al., 2006). When a wild-type plant is pollinated with heterozygous LAT52::GUS pollen grains, c. 50% of ovules show staining related to the presence of the β-glucuronidase (GUS). When the hap2-1/+ mutant is pollinated with those pollen grains, despite the normal growth of all PTs, c. 25% of ovules show the specific GUS staining, suggesting that only the HAP2 pollen is able to reach the synergid and burst (von Besser et al., 2006). However, based on current data, it is still difficult to propose a model in which the sperm cell plays a key role in this process.
In Arabidopsis, the following sequence of events is observed: (1) the PT reaches the synergid; (2) the PT continues to grow around the synergid; (3) the synergid degenerates; (4) the PT discharges; and (5) the released sperm cells fuse with the central cell and the egg cell (Sandaklie-Nikolova et al., 2007). Recent live-cell imaging observations suggest, however, that the breakdown of the receptive synergid cell occurs concomitantly with PT discharge (Hamamura et al., 2011). This implies that the interaction of the PT and the synergid induces a signalling cascade resulting in synergid degeneration and the bursting of the PT; or, alternatively, that the force generated by PT discharge can cause the breakdown of the synergid.
Communication between the PT and the synergid may initially occur via the activation of a receptor kinase at the surface of the synergid. In a genetic screen of Arabidopsis insertional mutants with distorted segregation, which was run in parallel with cytological screening of sterile mutants without structural abnormalities in the embryo sac and pollen, feronia (fer) & sirène (srn) mutants were isolated (Fig. 7b) (Huck et al., 2003; Rotman et al., 2003). In both cases, the embryo sac attracts PTs, but the synergids are not recognized by the PT, which continues to grow inside the embryo sac. In some cases, supernumerary PTs are found inside the embryo sac (Huck et al., 2003; Rotman et al., 2003). The in-depth characterization of the fer mutant showed that FER and SRN are allelic (Escobar-Restrepo et al., 2007). FER/SRN encodes a receptor kinase that belongs to the CrRLK1L-1 family. FER/SRN is weakly expressed in the mature embryo sac with a stronger expression in the synergids, although it is expressed in other tissues as well. FER/SRN is localized to the plasma membrane of the synergids at the filiform apparatus. Using FER homologues from numerous species, it was found that the putative ligand-binding extracellular region showed the greatest degree of amino acid diversification, indicating that this domain was evolving faster than the rest of the protein. Since pollen from closely related species can be attracted, but not recognized by the synergids, FER may act as a reproductive barrier at the PT reception step (Escobar-Restrepo et al., 2007).
Lorelei (lre) mutants, in which a potential GAP protein anchored to the membrane is disrupted, exhibit a similar phenotype, although less penetrant (fertilization abnormalities in 50% of homozygous mutants) (Capron et al., 2008; Tsukamoto et al., 2010). The low penetrance of the phenotype may be the result of functional redundancy among three Arabidopsis paralogues (LLGs), although this would be unlikely, since the paralogues show stronger expression in the PT, while LRE is expressed in the synergids of the mature embryo sac (Capron et al., 2008). Furthermore, the lre-5;llg1 double mutant does not show a more severe phenotype than lre (Tsukamoto et al., 2010). Taken together, these results demonstrate the involvement of a serine/threonine kinase receptor in the reception of a PT and in the activation of a signalling cascade, possibly involving a GAP protein, in the synergid (Fig. 7b). This activation may in turn activate the degeneration of the synergid, as well as the release of factors affecting PT growth and integrity. Notably, Arabidopsis synergid degeneration requires GFA2, a protein similar to the yeast mitochondrial Mdj1p chaperone (Christensen et al., 2002). This suggests a role for the mitochondria in triggering synergid degeneration and their potential as a target by pathways such as the FER pathway (Fig. 7b). However, the fact that there is no fertilization in the gfa2 mutant could support a model in which the mitochondria are involved in a signalling pathway leading to the release of pollen bursting factors.
The involvement of the FER pathway in the release of pollen bursting factors is supported by another Arabidopsis female gametophytic mutant named nortia (nta). This mutant displays reduced fertility as well as PT overgrowth in the synergids (Kessler et al., 2010). The NTA gene, or AtMLO7, codes for a member of the Mildew resistance locus o (MLO) family. These MLO proteins, originally discovered as being required for powdery mildew susceptibility in barley, typically have seven transmembrane domains, preceded by a signal peptide and followed by a C-terminal calcium-binding domain. Arabidopsis plants stably transformed with a pNTA::NTA-GFP fusion construct revealed a punctate pattern throughout the cytoplasm of the synergids in mature but unfertilized female gametophytes. Interestingly, upon PT arrival at the micropyle, NTA-GFP was localized to the basal half of the synergids. However, polar localization of NTA-GFP upon PT arrival was not observed in fer mutant embryo sacs (Kessler et al., 2010). Since MLO proteins appear to modulate SNARE-dependent and vesicular transport-associated processes at the plasma membrane (Bhat et al., 2005), NTA could be involved in delivering regulatory or signalling proteins to the plasma membrane. The FER receptor-kinase could thus initiate a signalling cascade upon PT arrival, leading to the relocalization of NTA-containing vesicles to the filiform apparatus (Fig. 7b).
Signalling proteins are also involved in the integrity of the PT. ANXUR1 (ANX1) and ANX2 are Arabidopsis pollen-specific receptor kinases paralogous to FER (Fig. 7c). The anx1anx2 mutant is almost 100% male sterile. In in vitro assays, mutant PTs burst shortly after germination, while in in vivo assays, the tube burst within the stigma (Boisson-Dernier et al., 2009; Miyazaki et al., 2009). The anx1anx2 phenotype suggests that the ANXUR receptors are involved in the activation of a pathway maintaining the integrity of the PT. A signal released by the synergid would cause ANXUR inactivation, thus leading to tube burst and sperm cell discharge. However, the early bursting phenotype made a direct observation of this conclusion impossible.
The recent discovery of the EMBRYO SAC4 (ZmES4) peptide, in maize, supports this signalling model. ZmES4, a member of the large CRP family, is normally stored in secretory vesicles of the egg apparatus cells until fertilization. ZmES4-RNAi plants show no defects in PT attraction, but the tubes do not burst and will continue to grow in or around the egg apparatus. In vitro, ZmES4 causes PTs to rupture. Significant depolarization of the PT membrane occurs involving KZM1, a pollen-expressed potassium import channel (Amien et al., 2010). This supports a model in which the synergid perceives the PT and liberates a small peptide, thereby stopping growth and bursting the PT through depolarization of its membrane (Fig. 7c).
An Arabidopsis calcium pump on the plasma membrane of the PT, AUTOINHIBITED Ca2+ATPase9 (ACA9), also affects the growth and reception of the PT (Fig. 7c). It has been proposed that this pump controls a calcium signalling pathway by modulating Ca2+ oscillations. Although the aca9 phenotype is similar to fer/srn, PTs stop at the synergid, suggesting that this pump is involved in the bursting rather than the halting of the tube. Thus, the arrest and rupture of the PT appear to be regulated via different signalling pathways (Schiott et al., 2004).
Another link for the importance of organelles in PT reception is revealed in the Arabidopsis abstinence by mutual consent (amc) mutant (Boisson-Dernier et al., 2008). The AMC gene (also known as Aberrant Peroxisome Morphology 2 or APM2; Mano et al., 2006) codes for an atypical peroxin, a peroxisome biogenesis factor that is targeted to the peroxisome. When an amc PT encounters an amc female gametophyte, a PT overgrowth phenotype without sperm cell discharge is observed, and often, multiple amc PTs can target an amc ovule. Genetic interaction between the FER and APM2/AMC pathways reveals a partial but significant additive effect for the amc and fer mutations in amc/+ and fer/+ double mutant plants, suggesting that the two pathways are most likely independent. In an amc background, peroxisome formation is severely affected. Thus, functional peroxisomes must be present in either the male or female gametophyte to enable PT reception. The authors postulate that, in amc gametophytes, mislocalization of a protein normally targeted to the peroxisome could affect PT reception. Since peroxisomes are involved in the production of numerous signalling molecules, including jasmonic acid, salicylic acid, auxin (IAA), reactive oxygen species (ROS) and nitric oxide (NO), this opens up the scene for possible roles of these molecules in the male–female gametophyte dialogue.
The FERTILIZATION INDEPENDENT SEED (FIS) pathway, which inhibits the proliferation of the endosperm, also appears to be linked to the signalling pathway involved in PT reception through FER/SRN (Fig. 7b) (Rotman et al., 2008). With the aim of identifying the components of the FER signalling pathway, a screen was carried out to identify other Arabidopsis mutants with abnormal PT reception. The scylla mutant was identified, in which the autonomous development of endosperm was also observed. This led to a reassessment of the fer/srn phenotype, which demonstrated that a very small percentage of fer/srn ovules present the autonomous endosperm development typical of fis mutants. Conversely, multicopy suppressor of ira1 (msi1) mutants (MSI1 is part of the FIS polycomb complex) have a very low percentage of ovules with the fer/srn phenotype. The msi1;fer/srn double mutant shows a synergistic effect. This indicates that the FER/SRN signalling pathway of the synergid interacts with the FIS signalling pathway of the central cell, and that this interaction controls the release of sperm cells and the development of the endosperm (Rotman et al., 2008). In other words, during fertilization, the synergids and central cell are in constant communication.
Communication between the egg cell and central cell may persist during fertilization, possibly to synchronize the development of the embryo and endosperm. During fertilization, one sperm cell fertilizes the egg cell to initiate the development of the diploid zygote, while another fuses with the central cell to initiate the development of a triploid endosperm. The discovery of the Arabidopsis cdc2a and F-box-like17 (fbl17) mutants allowed researchers to dissect the phenomenon of double fertilization. During pollen development, microspores undergo a first mitosis to produce a generative cell enveloped within a vegetative cell. The former undergoes a second mitosis to produce the two sperm cells needed for double fertilization. In cdc2a and fbl17 mutants, this second mitosis does not occur, and a single sperm-like cell is present in mature pollen (Iwakawa et al., 2006; Nowack et al., 2006; Kim et al., 2008; Gusti et al., 2009). Surprisingly, both the development of the zygote and the endosperm is initiated following fertilization of the egg cell (Nowack et al., 2006; Kim et al., 2008; Gusti et al., 2009). Thus, a positive signal from the fertilized egg cell to the central cell is produced, thereby triggering endosperm development (Nowack et al., 2006). However, recent analyses have questioned these findings. For the cdc2a mutant, in some cases, the generative cell undergoes division within the growing PT, and the second sperm cell is able to merge with the central cell. The karyogamy between the sperm cell and the central cell is incomplete, leading to seed abortion. Is there a positive signal emitted from the egg cell? Apparently not, as evidenced by the presence of only the zygote in some cdc2a ovules (Aw et al., 2010), as well as only two zygotes in the mutant eostre (Pagnussat et al., 2007), without any proliferation of the endosperm. The fusion of the sperm cell with the central cell is sufficient to initiate division of the central cell, which could be done via a signalling pathway involving membrane proteins, such as GENERATIVE CELL SPECIFIC1 (GCS1/HAP2) (von Besser et al., 2006; Mori et al., 2006), or via cytoplasmic signals (Bayer et al., 2009) (see Section 5 below).
When sperm cells are released into the degenerated synergid, they merge with the central cell and the egg cell. Recognition proteins are essential at this point to activate cell fusion. GCS1/HAP2, a sperm cell membrane protein, is involved in the interaction of the sperm cell with the central cell and egg cell (Fig. 7d) (Wong et al., 2010). GCS1/HAP2 was originally isolated by differential display from Lilium longiflorum pollen as being highly specific to the generative cell, and indeed, the protein gradually accumulates during pollen development, particularly during the advanced bicellular stage (Mori et al., 2006). In Arabidopsis, the sperm cells of the gcs1/hap2 mutant are able to penetrate the embryo sac, but do not fuse with the central cell and egg cell (von Besser et al., 2006; Mori et al., 2006).
In Arabidopsis, the cdc2a and fbl17 mutant phenotypes suggested a preferential interaction between the sperm cell and the egg cell (Nowack et al., 2006; Gusti et al., 2009), although this finding was not supported by a recent analysis of the cdc2a mutant, which detected no preference (Aw et al., 2010). The single sperm cell of the chromatin assembly factor1 (caf1) mutant can also merge with either the central cell or the egg cell (Chen et al., 2008). However, using the GCS1/HAP2 promoter to express the DTA (Diphteria Toxin Fragment A) toxin in the generative cell blocked the division that would have generated the two sperm cells. The resulting single cell expresses markers characteristic of the sperm cell and is able to fertilize either the egg cell or the central cell, with a strong preference (9 : 1) for the central cell (Frank & Johnson, 2009). Discrepancies between results regarding preferential fertilization of the egg or central cells from the two sperm cells could be related to the use of mutants. Very recent data using wild-type plants under physiological conditions may have resolved this issue (Hamamura et al., 2011). Using a photobleaching procedure to follow the fate of each labelled sperm cell, the study of Hamamura et al. (2011) did not reveal any preferential fertilization, and concluded that the two sperm cells are functionally equivalent.
On the opposite end of the spectrum, subjecting the embryo sac to several male gametes is useful in determining the egg and/or central cell involvement in the polyspermy block phenomenon. The tetraspore (tes) mutant of Arabidopsis can produce pollen with up to four sperm cells with different ploidy levels (Spielman et al., 1997). However, fertilization with tes mutant pollen causes the abortion of the seed by genetic imbalance as a result of imprinting. The phenotype is rescued by using a methyltransferase1 (met1) mutant background, in which the imprinting mechanism is inhibited (Scott et al., 2008). Thus, the karyotype of the endosperm and the embryo could be characterized, and only the karyotype of the endosperm showed signs of multiple fertilizations. In other words, if the egg and central cell are subjected to several male gametes, only the central cell becomes polyploid. The fertilized egg cell must therefore produce a signal that blocks polyspermy (Scott et al., 2008). Hamamura et al. (2011) addressed this issue by observing the time required for gamete fusion and found no significant difference between egg cell and central cell plasmogamy, suggesting that polyspermy block may be concurrent.
When the zygote is formed from the fertilized egg cell, the single cell extends to nearly three times its initial length, at which point an asymmetric division takes place, forming a round apical cell and an elongated basal cell. While the apical cell becomes the embryo proper, the basal cell forms the suspensor. Bayer et al. (2009) found that the YODA MAPKKK may control the asymmetric division of the zygote, for which its activity depends on a male cytoplasmic RNA transfer occurring during fertilization. Previously, Lukowitz et al. (2004) had demonstrated that activation and regulation of YODA (YDA) MAPKKK activity are required for proper differentiation of the zygote. Although the yda mutant has defects other than those observed during embryogenesis, the short suspensor (ssp) mutant phenotype mimics yda only in that respect. While SSP is transcribed in mature pollen, it is transiently translated in the central cell and the egg cell after fertilization. SSP encodes a membrane-bound protein which is part of the PELLE/IRAK superfamily. The yda;ssp double mutant is anatomically identical to the single yda mutant, while the expression of the hyperactive form of YODA reverses the ssp phenotype. Together, these double mutants suggest that SSP acts upstream from YODA to control asymmetric division of the zygote (Bayer et al., 2009). YODA is a MAPKKK located upstream from the MKK4/MKK5 and MPK3/MPK6, all of which are involved in stomatal development (Wang et al., 2007). It will be interesting to further investigate, via temporal and spatial knockout, the involvement of this signalling pathway in embryo development. The MPK3 and MPK6 kinases have already been shown to be involved in ovule development, as the mpk3(+/−)/mpk6 shows normal megasporogenesis and megagametogenesis, but impaired development of the integuments, leading to female sterility (Wang et al., 2008).
In the last decade, considerable progress was made in the area of cell–cell communication related to female gametophyte development and double fertilization. Forward genetics was shown to be a powerful tool for the study of plant reproduction, and several regulatory genes were isolated. These genes variously encode proteins involved in signalling, genetic regulation and the structural integrity of the cell. Most importantly, several newly characterized mutants are specific to a cell type and have been useful in dissecting the interplay among the cells of the ovule, the female gametophyte and those tissues playing a role in double fertilization. Thus, although little is known about the molecules involved in cell–cell communication, genetic evidence supports their existence. The female gametophyte has revealed surprising cell–cell communication mechanisms. Cells are able to communicate amongst themselves using the typical ligand/receptor kinase signalling pathways, MAPK and histidine phosphorelay transduction cascades, but are also able to carry out atypical signalling through the transfer of RNA molecules. The greater use of proteomic tools to characterize protein–protein interactions with the aim of completing the numerous hypothetical signalling pathways discovered via mutant analysis will surely be part of the next frontier in our understanding of the intricacies of cell–cell communication in sexual plant reproduction.
This work was supported by the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canada Research Chair program. E.C. is a recipient of PhD fellowships from NSERC and from Le Fonds Québécois de la Recherche sur la Nature et les Technologies (FQRNT); A.L-H. is a recipient of an MSc fellowship from FQRNT; and E.L.Z. is a recipient of a PhD fellowship from NSERC.