Disruption of the pollen-expressed FERONIA homologs ANXUR1 and ANXUR2 triggers pollen tube discharge

In eukaryotes, complex and specialized mechanisms have evolved to achieve the union of male and female gametes at fertilization by regulating attraction, signal exchange and recognition between the gametes(Boavida et al., 2005; Dresselhaus, 2006; Hirohashi et al., 2008; Hiscock and Allen, 2008; Kothe, 2008; Lord and Russell, 2002). In plants, the gametes are not produced directly from the meiotic products as in animals, but rather from multicellular, haploid gametophytes. Sperm cells of flowering plants are non-motile and need to be delivered via the haploid male gametophyte (pollen) (McCormick,2004) to the female gametophytes (embryo sacs). The latter are enclosed by the diploid tissues of the ovules that are themselves deeply embedded in maternal tissues of the pistil. In compatible interactions, the pollen grain germinates on the stigma of the pistil producing a pollen tube(PT) that grows through the style into the transmitting tissue, from where it emerges and is guided to the micropylar opening of the ovule(Crawford and Yanofsky, 2008; Dumas and Rogowsky, 2008). In the plant model Arabidopsis, typically a single PT is guided to each ovule, enters the micropyle and penetrates the female gametophyte. In Arabidopsis, the female gametophyte is a seven-celled haploid structure that contains an egg cell flanked by two synergid cells, a central cell and three antipodals (Kägi and Gross-Hardt, 2007; Yadegari and Drews, 2004). The PT penetrates one of the two synergid cells,where it terminates its growth and ruptures to deliver the two sperm cells -processes collectively referred to as PT reception(Weterings and Russell, 2004). Subsequently, double fertilization initiates seed development: one of the sperm cells is transported to, and fuses with, the egg cell to form the zygote, whereas the second sperm fertilizes the central cell, thereby initiating endosperm development (Berger et al., 2008).

The precise guidance of the PT from the stigma to the female gametophyte and successful PT reception followed by gamete fusions require several complex interactions between the pollen and the female tissues(Johnson and Lord, 2006). The identification of a few sporophytic factors that provide guidance cues at early stages of PT growth has been reported (e.g. Dong et al., 2005; Palanivelu et al., 2003; Wolters-Arts et al., 1998). However, the signals produced by the female gametophyte itself, which controls the late stages of PT guidance, were unknown(Chen et al., 2007; Higashiyama et al., 2001; Marton et al., 2005) until the recent identification of the defensin-like LURE polypeptides, which serve as the synergid-secreted PT attractants in Torenia fournieri(Okuda et al., 2009).

As the PT reaches the micropyle, several interactions between the PT and the female gametophyte ensue. However, our knowledge about the molecular and genetic mechanisms that mediate the interactions between the gametophytes is very limited. In the female gametophytic Arabidopsis mutants feronia (fer) and sirène (srn), PTs reach the micropyle but do not arrest their growth and are unable to rupture(PT overgrowth phenotype), demonstrating that the female gametophyte controls PT reception (Huck et al.,2003; Rotman et al.,2003). This indicates that successful PT reception requires that female and male partners recognize each other(Huck et al., 2003; Rotman et al., 2003). Interestingly, FER, which is allelic to SRN, encodes a plasma membrane-localized, synergid-expressed receptor-like kinase(Escobar-Restrepo et al.,2007). Receptor-like serine-threonine kinases (RLKs) are transmembrane proteins that typically receive signals through their extracellular domain and subsequently activate signaling cascades via their intracellular kinase domain (Morillo and Tax, 2006). RLKs are one of the biggest gene families in plants,with more than 600 members in Arabidopsis(Shiu et al., 2004). FER belongs to the previously uncharacterized CrRLK1L subfamily, with 17 members in Arabidopsis (Hematy and Höfte, 2008). Another CrRLK1L subfamily member, THESEUS1(THE1), has been reported to work as a putative cell wall integrity sensor,inhibiting cell elongation in hypocotyls lacking cellulose synthase(Hematy et al., 2007). More recently, FER, THE1 and the new CrRLK1L subfamily member HERCULES1 (HERK1)have all been shown to promote cell elongation in leaves and leaf petioles(Guo et al., 2009). In light of these new findings, a model for the male-female gametophyte dialog at PT reception has emerged: in the female gametophyte, FER is able to sense either a ligand produced by the PT or a ligand resulting from the modification of the cell wall of either the synergid or PT upon contact, thus triggering a signaling cascade enabling the female gametophyte to prepare itself for fertilization; in return, the female gametophyte is thought to signal back to the PT to arrest its growth, rupture and deliver the sperm cells(Escobar-Restrepo et al.,2007; McCormick,2007).

Since the first report of the fer/srn mutants, two other female gametophytic Arabidopsis mutants that specifically disrupt PT reception have been characterized, namely lorelei, which affects a putative glucosylphosphatidylinositol-anchored protein (GAP)(Capron et al., 2008), and scylla, for which the corresponding gene has not yet been identified(Rotman et al., 2008). Besides these three strictly female gametophytic mutants, the peroxin loss-of-function mutant abstinence by mutual consent (amc) displays a similar PT overgrowth phenotype, but only when a mutant PT interacts with a mutant female gametophyte, suggesting that the male-female communication required for PT reception relies on some components of identical nature in both gametophytes (Boisson-Dernier et al.,2008). Although a male signaling pathway controlling PT reception is expected to exist, to date no component acting strictly in the male gametophyte has been identified.

Here, we report that ANXUR1 and ANXUR2, the pollen-expressed homologs most closely related to FER, display a polar localization at the plasma membrane of the PT tip, and function redundantly to maintain PT integrity during growth,most likely by regulating the timing of PT discharge. Thus, our study provides the first genetic evidence of a male counterpart to FER, as a signaling process that functions in the male gametophyte to potentially control sperm cell discharge and delivery.

Primers used in this study are listed in Table 1. The binary vectors constructed and described below were introduced into Agrobacterium tumefaciens strain GV3101 by electroporation, which was then used to transform the wild type by floral dipping(Clough and Bent, 1998).

Arabidopsis thaliana plants (accession Columbia-0) were grown in a Conviron growth chamber (Controlled Environments, Winnipeg, Canada) or in a growth room in plastic pots filled with ready-to-use soil (Professional Blend,Sunshine, Canada, or ED73 Universal Erde, Germany). After sowing, pots were kept at 4°C for 2-3 days. Growing conditions were 22°C, 60% humidity with a 16-hour light/8-hour dark photoperiod regime at ∼75 μmol m^-2 s^-1.

anx1-1 and anx1-2 mutants of the ANX1 gene(At3g04690), as well as anx2-1 and anx2-2 mutants of the ANX2 gene (At5g28680), were obtained from the Arabidopsis Biological Resource Center (ABRC) and correspond to the SALK_016179 (1398 bp downstream of ATG), SALK_045687 (2145 bp downstream of ATG), SALK_127359 (570 bp upstream of ATG) and SALK_133057 (1833 downstream of ATG) lines, respectively. Genotyping PCR reactions for anx1-1 and anx1-2 were performed with primer pairs 1F1/1R1, 1F1/LBa1, 1F2/1R2 and 1F2/LBa1. For anx2-1 and anx2-2 genotyping, primer pairs 2F1/2R1, 2F1/LBa1, 2F2/2R2 and 2F2/LBa1 were used.

All in vitro pollen growth assays were performed as described(Boavida and McCormick, 2007). Flowers were incubated at 22°C for 30 minutes in moisture incubation boxes, then brushed onto slides containing germination medium [0.01% boric acid (w/v), 5 mM CaCl₂, 5 mM KCl, 1 mM MgSO₄, 10%sucrose, pH 7.5, 1.5% low-melting point agarose]. Slides in moisture incubation boxes were pre-incubated for 30-45 minutes at 30°C before returning them to 22°C for 30 minutes to several hours. Plasmolysis was induced by incubating growing pollen tubes with 40% sucrose-containing liquid germination medium for 20 minutes prior to observation. Pollen grains and tubes were imaged with a Leica DM6000 digital microscope using Leica Application Suite Advanced Fluorescence software (Leica Microsystems,Mannheim, Germany). Pollen germination counts and tube length measurements were performed manually using ImageJ 1.40g software(http://rsb.info.nih.gov/ij).

Aniline Blue staining was performed as described(Huck et al., 2003). Minimal pollination experiments were performed using an eyelash to deposit 4-16 pollen grains on the stigma of receiving pistils.

RT-PCR analyses were performed as described(Boisson-Dernier et al., 2008). Primer pairs used for ACTIN7 (At5g09810), ANX1(At3g04690), ANX2 (At5g28680) and FER(At3g51550) were ACT7-F/ACT7-R, 1F1/1R1, 2F2/2R2 and FER-RTF/FER-RTR.

To generate the p35S-ANX1-YFP and p35S-ANX2-YFP constructs, full-length ANX1 and ANX2 DNA sequences were amplified from BACs F7O18 and F4I4 (ABRC) with the primer pairs 1-BPs/1-BPas and 2-BPs/2-BPas,respectively. These fragments were cloned into pDNR221 (Invitrogen), sequenced and then recombined into the binary pXCSG-YFP(Feys et al., 2005). The pXCSG-YFP vector containing the plasma membrane-targeted FLS2-YFP fusion(Robatzek et al., 2006), and the pH35YG vector (Kubo et al.,2005) containing the 35S-YFP, were used as positive controls for membrane localization and cytosol-nuclear localizations, respectively. Tobacco leaf infiltration was performed as described(Walter et al., 2004). Fluorescence images were acquired by spinning-disc confocal microscopy as described (Boisson-Dernier et al.,2008)

To generate the pACA9-ANX1-YFP and pACA9-ANX2-YFP constructs, full-length ANX1 and ANX2 DNA sequences were amplified with the primer pairs 1-SalI/1-SpeI and 2-SalI/2-SpeI, respectively. Fragments were ligated into pGEM-T Easy (Promega, Madison, WI, USA), sequenced and then cloned into the SalI/SpeI-cut binary ps779 [ACA9-promoter-TAP2(YFP)](Myers et al., 2009) vector that confers hygromycin resistance. Fourteen (pACA9-ANX1-YFP) and six(pACA9-ANX2-YFP) independent T1 transgenic lines were selected for their strong expression, and all exhibited a similar expression pattern. After 5 hours of growth in vitro, growing pollen tubes were imaged as described above.

FER is highly expressed in the synergids of the female gametophyte and in a variety of vegetative tissues, but not in the male gametophyte(Escobar-Restrepo et al.,2007; Guo et al.,2009). According to publicly available ATH1 microarray data from various organs (Zimmermann et al.,2004) and pollen transcriptome studies(Becker et al., 2003; Honys and Twell, 2004; Pina et al., 2005), the two closest homologs of FER in Arabidopsis, At3g04690 and At5g28680 (Hematy and Höfte,2008) (see Fig. S1A in the supplementary material), appear to be predominantly expressed in the male gametophyte(Fig. 1A), but not in the sperm cells (Borges et al., 2008). The strong and preferential expression of these two FER homologs in pollen was confirmed by RT-PCR analysis(Fig. 1B). Based on their complementary expression pattern, At3g04690 and At5g28680were renamed ANXUR1 (ANX1) and ANXUR2(ANX2), respectively, after the male consort of the goddess Feronia in ancient Italy (Funke,1801). Consistent with this, ANX1 and ANX2 proteins were both identified in the pollen proteome (Grobei et al., 2009).

All the CrRLK1L members are predicted to have transmembrane domains(Hematy and Höfte, 2008). THE1 and HERK1 are localized uniformly to the plasma membrane of hypocotyl epidermal cells (Hematy et al.,2007; Guo et al.,2009), whereas FER displays a uniform plasma membrane localization pattern in leaf epidermis, but a polar localization within the synergids towards the filiform apparatus(Escobar-Restrepo et al.,2007). We first transiently expressed the ANX1-YFP (yellow fluorescent protein) and ANX2-YFP fusion proteins in tobacco epidermal cells under the control of the constitutive 35S promoter. Consistent with plasma membrane localization, ANX1-YFP and ANX2-YFP displayed the same fluorescence pattern at the cell periphery, similar to that of a membrane-bound flagellin receptor FLS2-YFP fusion protein (Robatzek et al., 2006) (see Fig. S1B in the supplementary material).

Next, we investigated the subcellular localization of ANX-RLKs in pollen,in which they are expected to function, by stably transforming Arabidopsis with ANX1-YFP and ANX2-YFP driven by the strong pollen-specific ACA9 promoter(Schiott et al., 2004). The ANX1-YFP and ANX2-YFP expression patterns were analyzed in the growing PTs (as evidenced by time-lapse imaging) of fourteen and six T1 transgenic lines with good expression, respectively. These lines did not exhibit any obvious fertilization-related phenotypes that might otherwise confound our analyses. As controls, we used transgenic Arabidopsis lines expressing ACA9-YFP and GFP-CNGC18 in PTs. ACA9-YFP and GFP-CNGC18 have been reported to display a uniform and a polar plasma membrane localization at the tip of growing PTs,respectively (Frietsch et al.,2007; Schiott et al.,2004). In contrast to the uniform plasma membrane expression pattern of the ACA9-YFP fusion protein, but similar to GFP-CNGC18 expression,PTs expressing ANX1-YFP and ANX2-YFP exhibited the most intense fluorescence signal at the cell periphery of the tips(Fig. 2). The enrichment in the membrane at the tip, but not in the shank, of PTs for ANX1-YFP, ANX2-YFP and GFP-CNGC18 was also evidenced by analysis of fluorescence intensity across the growing PT shanks and tips (Fig. 2). Moreover, PT plasmolysis confirmed localization of ACA9-YFP,ANX1-YFP and ANX2-YFP at the plasma membrane (see Fig. S2A in the supplementary material). After plasmolysis, a weak residual fluorescence signal was consistently observed close to the cell wall for PTs expressing ANX1-YFP and ANX2-YFP but not ACA9-YFP (see Fig. S2A in the supplementary material). In germinating pollen grains, polar localization of the ANX1-YFP and ANX2-YFP fusion proteins was also observed in the freshly emerged tip, but did not appear to precede the appearance of the first bulge (see Fig. S2B in the supplementary material). Moreover, fluorescence of ANX-YFP fusion proteins was associated with very dynamic small vesicles (see Movies 1 and 2 in the supplementary material), consistent with possible exocytosis at the tip subapical region and endocytosis at the apex(Zonia and Munnik, 2009). In dead PTs, in which growth and cytoplasmic streaming had stopped, the polar distribution of ANX-YFP fusion proteins towards the tip was completely abolished (see Fig. S2C in the supplementary material). In conclusion,ANX1-YFP and ANX2-YFP fusion proteins display an active polar localization at the plasma membrane of the growing PT tip.

To further elucidate the role of the ANX-RLKs, we isolated two single T-DNA insertional mutants for each of the ANX-RLKs, namely anx1-1 and anx1-2 for ANX1 and anx2-1 and anx2-2 for ANX2 (Fig. 3A), from the Salk Institute Genomic Analysis Laboratory Database (SIGnAL; La Jolla, CA,USA) (Alonso et al., 2003). RT-PCR analysis from open-flower cDNAs indicated that for each of the anx1 and anx2 alleles, the T-DNA insertions disrupted ANX-RLK gene expression (Fig. 3B). We anticipated that ANX1 and ANX2 would function redundantly in pollen because (1) ANX1 and ANX2 share 85.6% identity at the amino acid level and define a subgroup of the CrRLK1L subfamily(Hematy and Höfte, 2008);(2) the ANX1-YFP and ANX2-YFP fusion proteins show overlapping localization patterns; and (3) the ANX-RLK genes are highly and preferentially expressed in pollen. Indeed, despite the lack of ANX-RLK gene expression, none of the four single T-DNA insertion mutants displayed any obvious phenotype in vegetative or reproductive tissues, seed set, or in PT growth assays in vitro and in vivo(see below).

By crossing the single anx1 and anx2 mutants, we attempted to generate two independent anx1-1 anx2-1 and anx1-2 anx2-2 double mutants, as well as the semi-independent anx1-1 anx2-2 double mutant. However, after genotyping 175 progeny of the F1 crosses, no double-homozygous mutants could be identified. Therefore, we decided to first isolate plants homozygous for one anx mutation and heterozygous for the second. Interestingly, all these plants, namely anx1-1/anx1-1 anx2-1/ANX2, anx1-1/ANX1 anx2-2/anx2-2 and anx1-2/anx1-2 anx2-2/ANX2, exhibited a moderately reduced seed set compared with the wild type (see Table S1 in the supplementary material). Furthermore, upon self-fertilization, these plants displayed a segregation ratio of the anx1 anx2 mutants to wild type of ∼1:1(Table 2)(χ²=1.073, 0.3>P>0.2), indicative of a gametophytic defect (Howden et al.,1998; Moore et al.,1997). To determine the transmission efficiency (TE) of the anx1 anx2 mutations, we performed reciprocal crosses of the independent anx1-1/anx1-1 anx2-1/ANX2 and anx1-2/anx1-2 anx2-2/ANX2 mutants to the wild type. Although transmission through the female gametophyte was not significantly affected(Table 3) (TE_F=78%for anx1-1 anx2-1, χ²=1.11, 0.3>P>0.2;TE_F=86% for anx1-2 anx2-2, χ²=0.45,0.7>P>0.5), we did not observe transmission of the anx1-1 anx2-1 (n=99) and anx1-2 anx2-2 (n=90)mutations through the male gametophyte(Table 3), demonstrating that anx1 anx2 double mutants are specific to the male gametophyte.

As shown in Table 2, one double-homozygous anx1-2 anx2-2 mutant plant was recovered from the progeny of a self-fertilized anx1-2/anx1-2 anx2-2/ANX2 plant. Although no difference was observed in the vegetative tissues of anx1-2 anx2-2 compared with the wild type, anx1-2 anx2-2 was almost completely sterile, producing only 11 seeds from 277 siliques(Fig. 3C). However, unlike male-sterile dde2-2 mutant flowers, which display a defect in filament elongation and anther dehiscence(von Malek et al., 2002), anx1-2 anx2-2 flowers were normal, with obvious signs of dehiscence and self-pollination (Fig. 3D). When wild-type pollen was deposited on anx1-2 anx2-2 pistils,siliques elongated (Fig. 3C,arrows) (n=14 crosses) and were filled normally (see Table S1 in the supplementary material), consistent with anx1 anx2 mutations not affecting the female gametophyte. Conversely, deposition of anx1-2 anx2-2 pollen on dde2-2 pistils did not produce any seeds,whereas pollen from anx1-1/anx1-1 anx2-1/ANX2 and anx1-2/anx1-2 anx2-2/ANX2 plants led to silique elongation and a normal seed set (see Fig. S3 and Table S1 in the supplementary material) (n=12 crosses for each genotype). When anx1-1/anx1-1 anx2-1/ANX2 and anx1-2/anx1-2 anx2-2/ANX2 plants were manually self-pollinated or used either as male donor or female receiver in crosses with the wild type, the resulting siliques were always filled normally (see Table S1 in the supplementary material). This suggests that during natural self-pollination of these plants (∼80% seed set), single anx mutant pollen can only partially compensate for the inability of anx double-mutant pollen to reach female gametophytes(see below), whereas the large excess of pollen present during manual pollination leads to full compensation (∼92% seed set).

In conclusion, our data provide compelling genetic evidence that the anx1 anx2 double mutant is male sterile, with very rare transmission of the anx1 anx2 mutation through the male gametophyte. Thus, ANX1 and ANX2 function redundantly in the male gametophyte.

To investigate the reason for the transmission failure of anx1 anx2 mutations through the male gametophyte, we first carried out in vitro PT growth assays for the different single and double anxmutants and visualized pollen growth 5 hours after incubation. Surprisingly,unlike the wild type and single anx mutants, approximately half of the germinated pollen from anx1-1/anx1-1 anx2-1/ANX2 and anx1-2/anx1-2 anx2-2/ANX2 plants had burst in vitro after forming a bulge, whereas the other half formed normally growing PTs(Fig. 4A; Table 4). Germinated pollen from the two independent anx1-1/anx1-1 anx2-1/ANX2 and anx1-2/anx1-2 anx2-2/ANX2 mutant alleles was either able to produce normal PTs or to burst in the 1:1 ratio expected for segregation of the anx1 anx2 mutations (Table 4) (for anx1-1/anx1-1 anx2-1/ANX2,χ ²=0.03, 0.9>P>0.8; for anx1-2/anx1-2 anx2-2/ANX2, χ²=0.018, 0.95>P>0.9). Consistently, all the germinated pollen from the double-homozygous anx1-2 anx2-2 plants burst to leave traces of cytoplasmic and membrane contents outside the pollen (Fig. 4A; Table 4). Time-lapse imaging of pollen grown in vitro indicated that discharge of the anx1 anx2mutant pollen started after 1 hour of incubation on germination medium and occurred within 1 second (Fig. 4B; see Movie 3 and Fig. S4A in the supplementary material). The frequency of pollen discharge gradually increased from 0% initially(n=198) to 39.1% at 2 hours (n=289), 59.9% at 3 hours(n=142), 73% at 4 hours (n=100), and eventually reached 93.2% at 8 hours of in vitro incubation (n=192). Closer examination of the anx1 anx2 mutant pollen 3 hours after incubation indicated that 59.8% of the grains that had burst had formed a bulge (between 2 and 5μm in length), 13.8% an emerging tip (between 5 and 10 μm) and 3.4% a clear PT longer than 10 μm (see Fig. S4B in the supplementary material)before discharging (n=174). We could not discriminate between the absence of a bulge and bulge formation behind the grain (and therefore not visible) in the remaining 23% of the pollen grains that had burst. When an emerging tip or PT was formed, first the tube ceased its growth, then the discharge occurred in the subapical region of the tip, rather than at the apex itself (Fig. 4B; see Movie 3 in the supplementary material). Astonishingly, pollen discharge did not always result in immediate death of the pollen, as evidenced by residual cytoplasmic streaming that could last for several hours and, occasionally, pollen could discharge several times (see Fig. S4A and Movie 4 in the supplementary material). These results provide strong evidence that disruption of ANX1 and ANX2 triggers spontaneous discharge of the germinating pollen in vitro.

To corroborate the in vitro defect of anx1 anx2 pollen, an in vivo analysis of PT growth was carried out using Aniline Blue staining. First,pollination experiments with a limited amount of pollen were performed, with∼4-16 anx1-1/anx1-1, anx2-1/anx2-1, anx1-1/anx1-1 anx2-1/ANX2 and anx1-2/anx1-2 anx2-2/ANX2 mutant pollen grains deposited on wild-type pistils. Eighteen hours after pollination, pistils were stained with Aniline Blue to reveal callose deposition in the PTs. First, no difference was observed for the pollen germination rate on the stigma between single and double segregating anx mutants(Table 5; Fig. 5A), indicating that anx1 anx2 pollen grains are able to germinate and produce PTs in vivo. Secondly, we attempted to visually locate the tip of each PT in order to assign it to either of two classes: (1) wandering in the transmitting tract or(2) targeting an ovule. When pollen grains from single anx mutants were used, tips could be clearly identified for ∼86% of the growing PTs(Table 5; Fig. 5A, left panel). By contrast, tips could be located for only ∼47% of anx1-1/anx1-1 anx2-1/ANX2 and anx1-2/anx1-2 anx2-2/ANX2 mutant PTs(Table 5; Fig. 5A, right panel). These results suggest that anx1 anx2 mutant PTs are unable to reach the locules of the ovary.

To confirm this, normal pollination of wild-type pistils by pollen of either wild-type (n=12 pistils) or double-homozygous anx1-2 anx2-2 (n=12 pistils) plants was performed. Eighteen hours after pollination, wild-type PTs had grown through the entire style and transmitting tract of all pistils and were targeting the ovules normally(Fig. 5B, left panel). By contrast, the majority of anx1-2 anx2-2 PTs were arrested between the stigma and the style or in the style itself, with a few exceptional PTs(n=12 out of more than 500) that reached the top of the ovary locule(Fig. 5B, right panel). However, within the twelve pollinated pistils, not a single female gametophyte was targeted by anx1-2 anx2-2 PTs, consistent with the fact that the anx1-2 anx2-2 plant produced only 11 seeds. In conclusion, our results strongly suggest that anx1 anx2 mutant pollen grains are able to germinate and produce PTs that will eventually burst in the style, long before reaching and fertilizing the female gametophytes.

Our analyses have shown that ANX1 and ANX2 play a role in PT discharge, but they are not required for pollen germination, as in vitro no fewer than 77% of the anx1 anx2 double-mutant pollen at least formed a bulge prior to bursting. Consistent with this, anx1 anx2double-mutant pollen grains germinate normally on the stigmatic cells in vivo. However, in vitro, as soon as they have established polar tip growth, the pollen bulges, tips or tubes rapidly and systematically explode, such that PTs longer than 10 μm are an exception. These results indicate that ANX-RLKs are required for the maintenance of PT growth in vitro. In vivo, anx1 anx2 PTs can grow over a longer distance (>60 μm), as they generally grow through the entire stigma, arresting in the style. This observation illustrates the beneficial effect of the female sporophytic tissues on PT growth during compatible interactions(Crawford and Yanofsky, 2008; Higashiyama et al., 1998; Johnson and Lord, 2006; Palanivelu and Preuss, 2006). However, in the in vivo analyses, anx1 anx2 PTs very rarely reached the locules of the ovary (n=12 out of more than 500) and none of these PTs targeted a female gametophyte. Consistent with this, the double-homozygous anx1-2 anx2-2 mutant plant produced only 11 seeds out of 277 siliques analyzed. In segregating anx double mutant plants, single anx mutant PTs can compensate for the inability of double anx mutant PTs to reach the female gametophytes (see Table S1 in the supplementary material).

To date, only the Arabidopsis vanguard1 (vgd1) mutant has been reported to exhibit a phenotype similar to that triggered by the disruption of the ANX-RLKs, although the defects appear milder(Jiang et al., 2005). In vivo, vgd1 mutant PTs germinate normally and arrest only in the transmitting tract. However, vgd1 PTs more easily and frequently reach the ovary locules and female gametophytes than anx1 anx2 PTs. Consequently, each silique of homozygous vgd1 mutant plants produces a few seeds (Jiang et al.,2005). Although less well documented, vgd1 mutant pollen has also been reported to germinate and burst in vitro(Jiang et al., 2005). Interestingly, VGD1 is strongly expressed in pollen and encodes a wall-localized pectin methylesterase (PME; E.C. 3.1.1.11)(Jiang et al., 2005). PMEs catalyze the specific demethylesterification of the linear homopolymer(1,4)-linked-α-d-galacturonic acid homogalaturonan (HGA), a major pectic constituent of the cell wall(Micheli, 2001; Pelloux et al., 2007; Willats et al., 2001). HGA is deposited in a highly methyl-esterified form in the cell wall and is subsequently demethylesterified in muro by PMEs, the activities of which are regulated by PME inhibitor (PMEI) proteins(Juge, 2006). The degree of esterification of pectins is essential for cell wall mechanics, as unesterified pectins are able to bind Ca²⁺ and induce pectin gelation, which rigidifies the cell wall(Willats et al., 2001). Consistent with this, the PME multigene family functions in a wide range of organs, tissues and processes related to plant defense and cell elongation(Pelloux et al., 2007). Besides VGD1, the involvement of PMEs and PMEIs in PT growth is well documented (for reviews, see Chebli and Geitmann, 2007; Cheung and Wu,2008; Krichevsky et al.,2007; Zonia and Munnik,2009). Indeed, pectins are the major, polarly distributed constituent of the PT cell wall (Li et al., 1994). At the tip of the PT, where expansibility is required for polarized tip growth and where pectins are secreted, the methyl-esterified pectic `loose' forms are predominant, whereas in the lateral regions of the tube wall that need to be more resistant, esterified 'rigid' pectins are prevalent (Bosch and Hepler,2005; Bosh et al., 2005; Li et al., 1994; Parre and Geitmann,2005). Interestingly, this polar distribution could well be the consequence of the interactions between PMEs and PMEIs at the PT tip, whereas in the lateral region of the PT wall only PMEs appear to be present(Röckel et al.,2008).

Upon in vitro germination, anx1 anx2 PT growth prematurely terminates, followed by tip rupture. Despite bursting, cytoplasmic streaming remained active and new bulges were often produced, which subsequently burst. This suggests that ANX-RLKs are not driving PT growth per se, but are rather required to sustain PT growth and prevent discharge. With the identification of more than 50 mutations that affect pollen germination, PT growth and shape(Cheung and Wu, 2008), it is surprising that only vgd1 and anx1 anx2 mutations systematically lead to PT bursting. Although unlikely, the in vitro and in vivo phenotypic similarities of anx and vgd1 PTs could be circumstantial. However, given the importance of the balance between esterified and de-esterified pectins for PT growth, how the ANX-RLKs and PME/PMEIs are interrelated is of interest. One among many possible models is that once polar tip growth has been established, ANX-RLKs are secreted to the plasma membrane at the PT tip to monitor the cell wall status or sense external stimuli. This information is relayed to the interior by the ANX-RLKs to control the synthesis of specific cell wall material via regulation of cell wall modifying enzymes such as PMEs and PMEIs. Therefore, it would be interesting to analyze the cell wall composition and mechanical properties of anx and vgd1 mutant PTs.

Our results show that anx1 anx2 mutant PTs burst in vitro and are most likely to do so in vivo as well. Time-lapse imaging indicates that this process is explosive and rapid, as the expulsed intracellular content can regularly be found as far as 40 μm from the rupture site and occurs within 1 second. Moreover, when PTs were formed in vitro, they first ceased to grow then burst at the subapical region of the PT tip. Interestingly, these characteristics are reminiscent of the explosive discharge of PTs in the synergid cells of the female gametophyte during PT reception(Higashiyama et al., 2000; Palanivelu and Preuss, 2006; Rotman et al., 2003; Sandaklie-Nikolova et al.,2007).

ANX1 and ANX2 are the closest homologs of FER in Arabidopsis,displaying an overall 55.1% identity at the amino acid level to FER, which increases up to 77.5% for the kinase domain (see Fig. S1A in the supplementary material). FER is strongly expressed in the synergids of the female gametophyte and encodes a RLK that is polarly localized towards the filiform apparatus, which is the entry point of the PT into the synergid(Escobar-Restrepo et al.,2007). Moreover, fer mutant female gametophytes are unable to be fertilized, as PTs approaching them do not arrest their growth and fail to rupture, showing that FER inhibits PT growth and positively regulates PT discharge (Huck et al., 2003; Rotman et al.,2003). Besides its strong expression in synergids, FER is also widely expressed in vegetative tissues but not in pollen(Escobar-Restrepo et al.,2007; Guo et al.,2009). Conversely, ANX1 and ANX2 are strongly expressed in the male gametophyte and ANX-YFP fusion proteins are polarly localized in the plasma membrane of the growing PT tip, the contact point with the filiform apparatus during PT reception. Therefore, one might expect that ANX1 and ANX2 play a similar role in the PT as FER does in the female gametophyte, i.e. to negatively regulate PT growth. Surprisingly, we provide compelling evidence that ANX1 and ANX2 function in exactly the opposite way,as disruption of ANX1 and ANX2 leads to PT growth arrest and discharge. Therefore, we propose a new model for PT reception based on the phenotypes of the1, herk1, fer (see below) and anx mutants, as well as on live imaging of PT reception in Arabidopsis(Escobar-Restrepo et al.,2007; Huck et al.,2003; Palanivelu and Preuss,2006; Rotman et al.,2003; Sandaklie-Nikolova et al., 2007). Once the polar tip growth is established, we propose that ANX-RLKs are secreted to the PT tip to monitor the status of the cell wall and sustain growth within the female tissues until PTs reach the female gametophytes. Upon contact at the filiform apparatus, first, the female FER-dependent signaling cascade is activated allowing the female gametophyte to prepare itself for fertilization. Subsequently, the male ANX-dependent signaling process is deactivated, enabling the PT to arrest growth, to rupture and deliver the sperm cells to effect fertilization. Alternatively, ANX-RLKs could also function to maintain PT integrity during growth within the female tissues without playing any specific role during PT reception.

ANX1 and ANX2 define a subgroup of the receptor-like serine-threonine kinase CrRLK1 subfamily in Arabidopsis(Hematy and Höfte, 2008)and have close homologs in various dicotyledonous and monocotyledonous species(see Fig. S5 in the supplementary material). Out of the 17 Arabidopsis CrRLK1 subfamily members, five (ANX1, ANX2, FER,HERK1 and THE1) have been characterized and all are proposed to function in the control of cell elongation during various processes. HERK1, THE1 and FER have been reported to be required for optimum cell elongation in hypocotyls and petioles, as herk1 the1double-mutant and FER knockdown transgenic lines all exhibit dwarf phenotypes (Guo et al., 2009). However, disruption of THE1 in several cellulose synthase-deficient backgrounds promotes cell elongation(Hematy et al., 2007),demonstrating that THE1 can regulate cell elongation both positively and negatively. Unlike HERK1, THE1 and FER, ANX1 and ANX2 appear to function specifically in the male gametophyte but not in vegetative tissues. An anx1 anx2 double-mutant plant was neither stunted nor did it exhibit any vegetative phenotypes, consistent with the preferential expression of ANX1 and ANX2 in pollen. However,we cannot exclude the possibility that anx1 anx2 mutants exhibit discrete vegetative phenotypes or that the role of ANX1 and ANX2 during vegetative growth can only be revealed in some cell wall-damaged mutant backgrounds, as has been shown for THE1(Hematy et al., 2007). Interestingly, in addition to the similar phenotypes of anx1 anx2 and PME-deficient vgd1 mutant PTs, THE1 and HERK1 have been reported to influence the expression of some cell wall-modifying enzymes,including expansins and PMEs (Guo et al.,2009; Hematy et al.,2007).

The extracellular domains of the CrRLK1L subfamily members do not exhibit any homology with domains of known function and no ligands have yet been identified for any of its members. Therefore, the nature of the ligand(s) for the extracellular domains of CrRLK1L subfamily RLKs cannot be predicted. The identification of the first ligand, as well as the intracellular interacting partners, for any CrRLK1L subfamily member will undoubtedly shed light on the very important mechanisms that tightly control the relationship between cell wall composition, cell elongation and cell-cell communication during plant growth.

Saori Miyazaki and colleagues have recently reported similar findings on ANXUR1 and ANXUR2(Miyazaki et al., 2009).