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).
1. Involvement of the synergids and central cell in PT attraction
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).
Figure 7. Possible models of cell–cell communication and signalling during double fertilization. (a) In the synergids, MYB98 regulates the expression of secreted peptides possibly involved in the attraction of the pollen tube. The central cell may act directly or indirectly in this process through the transcription factor CCG. This attractant changes the direction of pollen tube growth toward the micropyle by regulating the K+ transporters CHX21/23. The ANX1/2 receptor kinases are activated to maintain the integrity of the pollen tube. NORTIA (NTA) is associated with secretory vesicles until pollen tube (PT) reception. (b) FER/SRN is a receptor kinase involved in the arrest of the pollen tube. A GAP protein, LORELEI, may mediate its action. The FER/SRN pathway regulates vesicular trafficking by relocalization of NTA on the basal side of the synergids; it also interacts with the FIS pathway, possibly to prime the central cell to receive the sperm cell. The FER/SRN pathway may directly regulate synergid degeneration by targeting the mitochondria, or indirectly following the release of bursting factor(s) and the subsequent discharge of the pollen tube. (c) Synergid degeneration (dashed line) or regulation of vesicular trafficking causes the release of small signalling peptides, like ZmES4. This peptide is perceived by the PT and causes membrane depolarization, perhaps by inhibiting ANX1/2 signalling, leading to activation of the K+ transporter KZM1 and the Ca2+ transporter ACA9. This depolarization and a consequent water uptake would be responsible for PT bursting. (d) Alternatively, as observed by Hamamura et al. (2011), degeneration of the receptive synergid may occur upon PT discharge. An initial sperm cell fuses with the egg cell, while a second sperm cell fuses with the central cell, possibly via protein interaction involving GCS1/HAP2. An inhibitory signal is then produced by the egg cell to prevent polyspermy, although the fusion events appear to occur concomitantly.
Download figure to PowerPoint
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.
2. Signalling proteins involved in PT attraction
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.
3. Synergid–pollen tube interaction
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.
4. Cell–cell communication within the female gametophyte
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).
5. Interaction between the male and the female gametes
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.
6. Male cytoplasmic RNA transfer during fertilization
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).