The FERONIA Receptor-like Kinase Mediates Male-Female Interactions During Pollen Tube Reception

See allHide authors and affiliations

Science  03 Aug 2007:
Vol. 317, Issue 5838, pp. 656-660
DOI: 10.1126/science.1143562

This article has a correction. Please see:


In flowering plants, signaling between the male pollen tube and the synergid cells of the female gametophyte is required for fertilization. In the Arabidopsis thaliana mutant feronia (fer), fertilization is impaired; the pollen tube fails to arrest and thus continues to grow inside the female gametophyte. FER encodes a synergid-expressed, plasma membrane–localized receptor-like kinase. We found that the FER protein accumulates asymmetrically in the synergid membrane at the filiform apparatus. Interspecific crosses using pollen from Arabidopsis lyrata and Cardamine flexuosa on A. thaliana stigmas resulted in a fer-like phenotype that correlates with sequence divergence in the extracellular domain of FER. Our findings show that the female control of pollen tube reception is based on a FER-dependent signaling pathway, which may play a role in reproductive isolation barriers.

In contrast to animals, where the products of meiosis differentiate directly into gametes, the meiotic products of higher plants undergo further mitotic divisions to form multicellular haploid structures called gametophytes, which in turn produce the gametes. To accomplish fertilization, the gametophytes communicate with and recognize each other. In angiosperms, the male gametophyte (pollen) germinates on the stigma and the growing pollen tube delivers the two nonmotile sperm cells to the female gametophyte (embryo sac). Proper delivery depends on signals from the female gametophyte (1, 2). These chemotactic signals guide the pollen tube into the micropylar opening of the ovule, the reproductive structure that harbors the female gametophyte. In the majority of flowering plants, including Arabidopsis thaliana (Brassicaceae), the female gametophyte consists of seven cells: the egg cell, the two synergids (which lie just inside the micropylar opening of the ovule), the central cell, and the three antipodals (3) (Fig. 1A). In Torenia fournieri (Scrophulariaceae), the two synergids are necessary for pollen tube guidance (4). In most species, one of the synergids degenerates prior to or coincident with the pollen tube approaching the micropyle (5). The pollen tube grows into the degenerating synergid through the filiform apparatus, a structure formed by invaginations of the cell wall of both synergids (6). Pollen tube reception in the ovule involves the arrest of pollen tube growth (Fig. 1B), the rupture of the pollen tube, and the release of the sperm cells, which are subsequently targeted to the egg and central cells for fertilization (7). Once the pollen tube has ruptured, attraction of further pollen tubes ceases; thus, only a single pollen tube will normally enter each micropyle.

Fig. 1.

Pollen tube reception is mediated by the FERONIA RLK. (A) Diagram of a mature female gametophyte. Abbreviations: AC, antipodal cells; CC, central cell; EC, egg cell; SC, synergid cells; FA, filiform apparatus; MI, micropyle; PT, pollen tube. Scale bar, 30 μm. (B and C) Epifluorescence micrographs of Aniline Blue–stained ovules. Fluorescent signal indicates the presence of callose, a major component of the cell wall of the pollen tube. (B) A fertilized wild-type ovule. The white arrow indicates the site of pollen tube arrest. (C) An unfertilized fer gametophyte. The white arrow indicates the site of pollen tube arrest in wild-type ovules, the red arrow shows the pollen tube growing inside the female gametophyte, and arrowheads indicate two pollen tubes entering the same ovule. Scale bars, 30 μm. (D) Sequence difference between wild type and fer. Left: Denaturing high-performance liquid chromatography (DHPLC) chromatogram of a PCR fragment of At3g51550 amplified from DNA of Landsberg erecta (Ler), fer/FER (in Ler), and Columbia wild-type plants crossed with fer (Col × fer). The additional peaks (arrows) in the elution profiles from fer/FER plants reflect the sequence difference due to a 4-bp insertion in the fer allele. Right: Coordinates of the insertion and premature stop codon in fer. Amino acid abbreviations: F, Phe; L, Leu; N, Asn; P, Pro; R, Arg; T, Thr; V, Val; Y, Tyr. (E) Sequence difference between wild type and sir. Left: Sequence chromatogram of sir/SIR (in C24 accession); arrow shows a double peak indicating a sequence mismatch. Right: Sequence and coordinates of the same chromatogram region indicating a 1-bp deletion and subsequent frameshift that leads to a premature stop codon in sir. (F) Diagram of FER's predicted domains. SP, signal peptide; TM, transmembrane domain; KINASE, kinase domain. Black vertical bars, sites of fer and sir mutations; K, site-directed mutagenesis of the conserved Lys of the kinase domain; asterisks, experimentally located phosphorylation sites (19).

Recent work in A. thaliana has shown that pollen tube reception is controlled by female gametophytic factors (8, 9). In heterozygous feronia (fer) and sirène (sir) mutants, all female gametophytes develop normally but half of the ovules remain unfertilized. In these ovules, the pollen tube continues to grow inside the female gametophyte, fails to arrest its growth, and does not rupture to release the sperm cells (Fig. 1C). Pollen from these mutants, however, can fertilize wild-type ovules, indicating that pollen tube growth and sperm delivery are unaffected (8, 9). Detailed cytological and ultrastructural analyses have shown that the synergids are normally specified and differentiated (8), which suggests that these mutants identify components of an active signaling process in which female gametophytic factors control pollen tube behavior and hence fertilization. Only a small number of factors involved in pollen germination, pollen tube growth, and guidance are known at the molecular level (1015). The molecular events underlying the female control of pollen tube reception are unknown. Here, we show that FER encodes a kinase-active receptor-like kinase, which is localized to the plasma membrane and, in particular, is asymmetrically localized to the filiform apparatus of the synergid cells. Inter-specific crosses indicate that the interaction of the FER receptor-like kinase with a putative ligand on the pollen tube may be involved in reproductive isolation barriers.

To gain insight into the molecular basis of the signaling event underlying pollen tube reception, we molecularly cloned FER with the use of positional methods (16). We mapped the mutation to At3g51550, which in the fer mutant contains a 4–base pair (bp) insertion corresponding to the footprint left after excision of a Ds transposon (Fig. 1D). Because sir shows the same phenotype, we sequenced the coding region of At3g51550 in the mutant and found a single base pair deletion (Fig. 1E). Both mutations lead to frameshifts that result in premature stop codons; therefore, FER and SIR are allelic.

A genomic fragment carrying At3g51550 plus 2.6 kb and 1.2 kb of upstream and downstream sequences, respectively, complements the fer phenotype (table S1), demonstrating that At3g51550 corresponds to the FER gene (fig. S1, A and B). The FER open reading frame contains a single 175-bp intron in the 5′ untranslated region and produces a transcript of 2682 bp, which encodes a putative receptor-like serine-threonine kinase (RLK) (Fig. 1F).

FER, which is a unique gene in A. thaliana, belongs to the CrRLK1L-1 subfamily of kinases, no members of which have a known function (Fig. 2A) (17). Plant RLKs belong to a monophyletic gene family with more than 600 members (18). RLKs are transmembrane proteins that receive signals through an extracellular domain and subsequently activate signaling cascades via their intracellular kinase domain, a molecular function consistent with the role of FER in a signaling process.

Fig. 2.

FERONIA, a plant-specific, widely expressed RLK, phosphorylates itself. (A) Unrooted tree showing FER and homologs from different species. The FER homologs, whose clade is supported by a 100% bootstrap value, are in the shaded area. The other plant species together with A. thaliana in this clade are A. lyrata, B. oleracea, B. rapa, and C. flexuosa homologs 1, 2, 3, and 4 (Brassicaceae), Oryza sativa (Os., Poaceae), and Populus trichocarpa (Salicaceae). The rest of the tree contains the A. thaliana members of the CrRLK1L-1 subfamily of kinases. The branch length scale bar represents 0.1 substitutions per site. (B) FER autophosphorylates during in vitro kinase assays. Top: Coomassie-stained gel of proteins used in the kinase assays. Bottom: Phosphoimager scan of gel in top panel. GST and GST-FERKR (FER Lys563 → Arg in Ler, FERLys565 → Arg in Col-0) have no kinase activity, whereas GST-FERwt is autophosphorylated. (C) Quantification of FER transcripts in leaves, closed flower buds, open flowers, and siliques (collected 1 to 4 days after hand pollination) and mature pollen grains. Transcript levels were normalized to 18S ribosomal RNA; means and SEs of three independent experiments are shown.

To determine the activity of the predicted intracellular kinase domain, we tested whether FER autophosphorylates in an in vitro kinase assay. The predicted intracellular domain (FERwt) and a kinase-inactive version (FERKR) were fused to glutathione S-transferase (GST). GST and GST-FERKR exhibited no kinase activity, whereas GST-FERwt was autophosphorylated (Fig. 2B). In addition, experimentally verified phosphorylation sites were present in the FER kinase domain (19) (Fig. 1F). Taken together, these findings show that FER encodes a kinase-active RLK involved in a novel signaling pathway that plays a critical role in the last stages of the communication between female and male gametophytes required for fertilization.

To evaluate whether the expression of FER is consistent with its proposed role in pollen tube reception, we examined the temporal and spatial expression pattern of FER. Quantitative real-time reverse transcription polymerase chain reaction (RT-PCR) revealed FER mRNA throughout the mature plant—specifically in leaves, buds, flowers, and siliques—but it was not detected in mature pollen (Fig. 2C). By in situ hybridization, FER transcripts were detected in floral apices, young ovule primordia, and young anthers with immature pollen (Fig. 3, A to C). In older anthers harboring mature pollen, FER transcript was not detected (Fig. 3D), consistent with the quantitative real-time RT-PCR experiments and the female-specificroleof FER in fertilization. In emasculated flowers, a very weak FER signal was detected throughout mature unfertilized ovules, and a stronger signal could be detected in the synergid cells (Fig. 3E). After fertilization, FER transcripts were detected in globular embryos (Fig. 3G), consistent with the finding that in rare cases where fertilization of fer gametophytes is achieved, the resulting homozygous embryos abort (8). In our complementation experiments, both the defect in pollen tube reception and embryo lethality were rescued by a wild-type FER transgene, demonstrating that FER is also required after fertilization (table S1). A genomic fragment containing the putative FER promoter (1.3 kb upstream of the start codon) was fused to the bacterial uidA gene encoding β-glucuronidase (GUS) to confirm the weak expression of FER observed in mature female gametophytes. Using a chromogenic substrate for GUS, we found that the FER promoter is highly active in the synergid cells (Fig. 3, I and J). Thus, the spatiotemporal expression pattern of FER is consistent with a function in pollen tube reception.

Fig. 3.

FERONIA is expressed in developing primordia and synergid cells. (A to H) In situ hybridization with an antisense FER probe [(A) to (E), (G)] or with a sense probe [(F) and (H)]. FER mRNA is present in the floral apex (A) and in ovule primordia (B). FER mRNA is expressed in anthers and microspores during early pollen development (C), but the transcript levels in mature pollen (D) are below the limit of detection (inset shows the sense control). (E) FER transcript can be detected in the synergid cells of the female gametophyte. (F) Sense control in the female gametophyte. (G) After fertilization, FER mRNA was detected in globular embryos. (H) Sense control in a fertilized ovule with a globular embryo. (I and J) Analysis of the FER upstream transcriptional regulatory region. (I) pFER::GUS is active in both synergid cells. (J) Higher magnification of the synergids from another ovule. Scale bars, 100 μm (A), 30 μm [(B), (E), (F), (I), (J)], 15 μm (C), 70 μm (D), and 50 μm [(G) and (H)]. Abbreviations: FP, floral apex; OP, ovule primordia; CW, carpel wall; MP, microspore cells; TP, tapetum cells; PG, pollen grain; CC, central cell; EC, egg cell; SC, synergid cell; GE, globular-stage embryo.

To investigate the subcellular localization of FER, we bombarded onion epidermal cells with a FER-GFP construct driven by the FER promoter (pFER::FER-GFP). The FER-GFP (green fluorescent protein) fusion protein was localized to the plasma membrane (Fig. 4, A and B), in contrast to GFP driven by the constitutive 35S promoter (35S::GFP), which is detected throughout the cell (Fig. 4, C and E). The pFER::FER-GFP construct was also stably transformed into A. thaliana and fully complemented the fer mutation (table S1), and the fusion protein was localized at the plasma membrane in leaf epidermal cells (Fig. 4D). Unfertilized ovules accumulated high levels of GFP signal in the lower part of the synergids (Fig. 4F) where the filiform apparatus is located. Weaker GFP fluorescence was detected in the membranes of the synergid cells and in the surrounding maternal sporophytic cells, as well as faintly in the egg cell (Fig. 4G). Because the filiform apparatus is a structure rich in plasma membrane, we tested whether the asymmetric distribution of FER might simply be due to an enrichment of plasma membrane in the area. Therefore, we compared the distribution of FER to that of another GFP fusion construct that has a plasma membrane localization motif and is expressed in the female gametophyte (pAtD123::EGFP-AtROP6C). Under the same conditions, FER-GFP levels were much higher in the filiform apparatus than in the rest of the synergids' cell membranes when compared to EGFP-AtROP6C (Fig. 4H). This finding suggests that FER is polarly transported to the filiform apparatus. Taken together, the data are consistent with a model in which FER, localized in the filiform apparatus, binds a putative ligand on the approaching male gametophyte, which then triggers the molecular events involved in pollen tube reception.

Fig. 4.

FERONIA is a cell membrane– localized RLK targeted to the filiform apparatus, and sequence divergence in its extracellular domain correlates with feronia-like phenotypes in interspecific crosses. (A) Transmission image of a plasmolyzed onion epidermal cell transiently expressing FER-GFP under FER promoter (pFER::FER-GFP). Black arrow points to cell wall; white arrow points to cell membrane. (B) Confocal laser scanning microscopy (CLSM) single optical section of (A) with FER-GFP localized at the periphery of the cell membrane. (C) Epifluorescence micro-graph of an onion cell transiently expressing (35S::GFP). (D) CLSM single optical section of leaf epidermis of an A. thaliana plant stably transformed with pFER::FER-GFP. (E) A. thaliana leaf epidermal cell transiently expressing (35S::GFP). (F) Ovule from the same plant as in (D) under CLSM; GFP signal in green, chlorophyll autofluorescence in red. (G) Maximum projection of several sections of the micropyle area of ovule in (F) showing FER-GFP accumulation in the filiform apparatus of the synergids. (H) Maximum projection of several sections of pAtD123::EGFP-AtROP6C female gametophyte. Upper arrow indicates upper part of synergids; lower arrow indicates region of filiform apparatus. (I) Gametophyte of a wildtype A. thaliana plant pollinated with pollen from A. lyrata, showing pollen tube overgrowth (white arrow). (J) Gametophyte of a wild-type A. thaliana plant pollinated with pollen from C. flexuosa, showing pollen tube overgrowth (white arrow). (K) Ka/Ks ratios calculated using a sliding window analysis with windows of 86 codons and a step size of 43 codons of the coding region of FER (excluding the signal peptide) in A. thaliana, A. lyrata, and C. flexuosa homolog 1. The least divergent homolog of C. flexuosa is shown. Scale bars, 30 μm [(A), (B), (F), (I), (J)], 60 μm(C), 10 μm [(D) and (E)], 20 μm [(G) and (H)]. SC, synergid cell; FA, filiform apparatus; EC, egg cell.

Because a signal transduction cascade initiated by the interaction of the FER RLK with a putative pollen ligand seems necessary for fertilization, it is possible that changes in the components of this interaction could act as reproductive isolation barriers. To some extent, the interaction between the pollen tube and the synergid is similar to sperm-egg interactions in animals (20), but it occurs at the level of gametophytes rather than gametes. For example, divergence in the protein sequence of the putative ligand-binding extracellular domain of FER or changes in the pollen ligand could prevent interspecific fertilization events. Interestingly, in interspecific crosses of Rhododendron species, pollen tube growth does not arrest. This results in pollen tube overgrowth in the female gametophyte, similar to the phenotype observed in fer (21, 22). If our model is correct, an interspecific cross where wild-type A. thaliana is pollinated by a related Brassicaceae species with a sufficiently divergent ligand might also produce a fer-like phenotype. Therefore, we performed inter-specific crosses between A. thaliana and different Brassicaceae species. In most crosses, the pollen germinated but failed to grow toward the ovules, as reported previously (2326). In crosses with the close relative Arabidopsis lyrata, pollen tubes were correctly targeted to the ovule but only 43.9% of the embryo sacs received pollen tubes and were fertilized; 5.2% of the ovules were fertilized with more than one pollen tube in the same micropyle, showing pollen tube overgrowth inside the female gametophyte (fig. S1C); and 50.9% of the ovules had pollen tubes that continued to grow inside the female gametophyte, as was observed in fer mutants (Fig. 4I) (n = 813 ovules). In crosses with a more distant relative, Cardamine flexuosa, only 7.7% of A. thaliana ovules attracted pollen tubes; of these, 1.6% had pollen tubes that entered the micropyle and stopped as previously reported (27), 0.8% had pollen tubes coiling outside the micropyle, and in 5.3% of the ovules the pollen tubes entered the micropyles but reception failed, resulting in a fer-like phenotype (n = 363 ovules) (Fig. 4J). In summary, around 50% and 70% of the A. lyrata and C. flexuosa pollen tubes that entered a wild-type A. thaliana embryo sac, respectively, displayed a fer-like phenotype.

Given the involvement of the FER signal transduction cascade in the interaction between the pollen tube and the receptive synergid, it is likely that FER plays a role in the failed pollen tube reception in these interspecific crosses. In the A. thaliana × A. lyrata crosses, about half of the pollen tube reception events were normal, demonstrating the capacity of the A. lyrata pollen tube for a normal reception response. However, in the same cross, fer-like phenotypes were observed. This suggests two possible reasons for failed receptions: (i) They are caused by a divergent ligand that is inefficiently recognized by the receptor domain of the A. thaliana FER RLK, or (ii) they result from the presence of two polymorphic forms of the putative ligand, one that is and one that is not recognized by the A. thaliana FER RLK. In the case of C. flexuosa, the ligand or ligands are predicted to have diverged to a degree that they cannot be recognized by the A. thaliana FER RLK, thereby causing the fer-like phenotype in A. thaliana ovules. We expect that a divergence in the sequence of the putative ligand would also be reflected in sequence divergence in FER's extracellular domain, which is proposed to interact with this ligand.

To test this hypothesis, we isolated FER homologs in these species and calculated the ratio between the number of nonsynonymous substitutions per nonsynonymous site (Ka) and the number of synonymous substitutions per synonymous site (Ks) (28) as an indication of their divergence in pairwise comparisons between A. thaliana and A. lyrata (one FER homolog) or C. flexuosa (four FER homologs) (Fig. 4K and fig. S1D). In all comparisons, higher Ka/Ks values occurred in the putative ligand-binding extracellular region, which indicates that this domain shows the greatest degree of amino acid diversification and evolves faster than the highly conserved intracellular kinase domain (29). The high sequence divergence in the extracellular domain of FER may contribute to reproductive isolation between two species, as has been proposed for other genes involved in recognition at fertilization (30).

Our data suggest that FER acts in the filiform apparatus to control the behavior of the pollen tube to achieve fertilization. We propose that the interaction between the putative male ligand and the extracellular domain of the FER RLK triggers a signal transduction cascade inside the synergid cell. A subsequent signal then feeds back from the synergid to the pollen tube, causing growth arrest and the release of the sperm cells. Conceptually, this process is similar to the signaling events occurring during the self-incompatibility reaction in Brassica spp. (31). In an incompatible pollination, a pollen ligand interacts with a stigma-expressed RLK, inducing a signaling cascade in female papillar cells, which then signal back to the pollen and inhibit its germination. Further studies of the FER signaling pathway will help to uncover the molecular mechanism of fertilization and reproductive isolation in plants. Furthermore, the manipulation of the FER pathway might allow the generation of hybrids between otherwise incompatible species.

Supporting Online Material

Materials and Methods

Fig. S1

Table S1


References and Notes

View Abstract

Navigate This Article