ScRALF3, a secreted RALF-like peptide involved in cell–cell communication between the sporophyte and the female gametophyte in a solanaceous species

Authors

  • Eric Chevalier,

    1. Institut de Recherche en Biologie Végétale, Département de Sciences Biologiques Université de Montréal, Montréal, Québec, Canada
    Search for more papers by this author
    • These authors contributed equally to this paper.
  • Audrey Loubert-Hudon,

    1. Institut de Recherche en Biologie Végétale, Département de Sciences Biologiques Université de Montréal, Montréal, Québec, Canada
    Search for more papers by this author
    • These authors contributed equally to this paper.
  • Daniel P. Matton

    Corresponding author
    • Institut de Recherche en Biologie Végétale, Département de Sciences Biologiques Université de Montréal, Montréal, Québec, Canada
    Search for more papers by this author

For correspondence (e-mail dp.matton@umontreal.ca).

Summary

Small peptides have been shown to regulate numerous aspects of plant development through cell–cell communication. These signaling events are particularly important during reproduction, regulating gamete development and embryogenesis. Rapid alkalinization factor (RALF)-like genes, a large gene family that encodes secreted peptides, have specific or ubiquitous expression patterns. Previously, five RALF-like genes with potential involvement during reproduction were isolated from Solanum chacoense. Here, we show that ScRALF3 is an important peptide regulator of female gametophyte development. Its expression, which is auxin-inducible, is strictly regulated before and after fertilization. Down-regulation of ScRALF3 expression by RNA interference leads to the production of smaller fruits that produce fewer seeds, due to improper development of the embryo sacs. Defects include loss of embryo sac nuclei polarization, as well as an increase in asynchronous division, accounting for cellular dysfunctions and premature embryo sac development arrest during megagametogenesis. ScRALF3 is expressed in the sporophytic tissue surrounding the embryo sac, the integument and the nucellus, as revealed by in situ hybridization and GUS staining. As expected for a secreted peptide, fluorescence from an ScRALF3–GFP fusion construct is detected throughout the secretory pathway. Therefore, the ScRALF3 secreted peptide may be directly involved in the regulation of multiple aspects of cell–cell communication between the female gametophyte and its surrounding sporophytic tissue during ovule development.

Introduction

Coordination of development by cell–cell communication is essential to ensure the integrity of the plant and to make possible the production of a viable progeny. Stem cell maintenance, cell differentiation, stomatal development, recognition of pathogens and nodulation, pollen–pistil interactions, and numerous aspects of reproductive development rely heavily on intercellular communications (Matsubayashi and Sakagami, 2006; Higashiyama, 2010; Chevalier et al., 2011; Katsir et al., 2011; Marshall et al., 2011). Many interesting molecules, including small RNAs, reactive oxygen species, peptides and plant hormones, are involved in cellular signaling (reviewed by (Santner et al., 2009; Van Norman et al., 2011). Therefore, cell–cell communication may occur via long-distance signaling between organs or between cell layers, as within the same cell layers, to specify cellular identity and function.

During plant sexual reproduction, the pollen tube penetrates the ovule through the micropyle to reach the female gametophyte (FG) or embryo sac (ES). To allow double fertilization, the ES must have a precise organization, and intercellular communications are therefore essential during its development. The Polygonum-type developmental pattern, the most frequent within flowering plant species (Maheshwari, 1950; Friedman and Ryerson, 2009), may be separated in two major steps: megasporogenesis and megagametogenesis (Christensen et al., 1997). Megasporogenesis produces one functional megaspore (FM) within the nucellus cells after two meiotic divisions and programmed cell death of haploid spores. Concomitantly, integument(s) grow from the first and second layers of the primordium to cover/protect the nucellus and the ES (Smyth et al., 1990; Robinson-Beers et al., 1992). Megagametogenesis starts when the FM undergoes three successive rounds of mitosis within a syncytium to produce eight nuclei. The first mitotic division occurs in a chalaza/micropyle axis, leading to an equal number of nuclei at both ends of the ES. After repositioning of the nuclei, cellularization starts and the ES becomes a mature seven-cell structure with three antipodals (1n) at the chalazal end, a central cell (2n), and two synergids (1n) and the egg cell (1n) at the micropylar end. The antipodals degenerate in most cases before or after fertilization (Christensen et al., 1997). This architecture allows double fertilization with the two sperm cells to create a diploid zygote and a triploid endosperm to produce viable seeds.

In recent years, genetic studies in Arabidopsis thaliana have provided supporting evidence for a dynamic interplay between the ES cells. Degeneration of the central cell in agamous-like61/diana disrupts the identity of all gametophytic cells, showing that intercellular interactions are an important aspect in the regulation of FG development (Bemer et al., 2008). Programmed cell death of the antipodals appears to be regulated in part by this interaction with the central cell (Kagi et al., 2010). The egg cell also appears to be essential for the fate of all gametophytic cells. Originally, both gametic cells (egg and central cell) were thought to control cell fate maintenance within the ES through the LACHESIS (LIS)-dependent RNA processing pathway (Gross-Hardt et al., 2007; Moll et al., 2008). However, only specific down-regulation of LIS in the egg cell by RNAi affects cell fate maintenance of all the gametophytic cells, revealing a predominant role for the egg cell in this process (Volz et al., 2012).

According to the structure/function relationship between the ES and the ovule in double fertilization, the development of one is dependent on that of the other. Small RNAs are produced within the L1 cell layer of the primordium and transported symplastically into the internal layers to prevent differentiation of an additional FM from sporophytic tissue (Olmedo-Monfil et al., 2010). Moreover, in A. thaliana, an auxin gradient appears to control cell differentiation within the ES, with the highest concentration associated with synergid cell fate (Pagnussat et al., 2009). The eostre mutant supports this position-dependent signaling pathway, whereby a nucleus that is slightly mis-positioned with respect to the center of the ES after the eight-nucleate stage results in the development of two egg cells and one synergid (Pagnussat et al., 2007). Interestingly, auxin accumulates first at the tip of the nucellus (Pagnussat et al., 2009), suggesting possible cross-talk between the FG and the sporophytic tissue.

Despite the fact that few signal molecules have been identified, intercellular interactions appear to occur between the ES and the surrounding maternal tissue. Comparative transcriptomic analysis of A. thaliana mutants lacking an ES (sporocyteless and coatlique) revealed up-regulation of 527 genes in the integuments, implying communication from the ES toward the sporophyte (Johnston et al., 2007). Signaling in the other direction (cross-talk) is also essential, as sporophytic mutations affecting the integuments result in ES developmental arrest. Mutants such as bel1, aintegumenta (ant) and inner no outer (ino), lacking both integuments (bel1 or ant) or only one (ino), show an arrest of ES development at the FM or eight-nucleate stage (Robinson-Beers et al., 1992; Baker et al., 1997; Christensen et al., 1997; Schneitz et al., 1997). So far, approximately 20 ovule sporophytic mutants with ES defects have been characterized, mostly with primary integument defects (Bencivenga et al., 2011).

Cysteine-rich peptides are an important class of small peptides that are abundant in reproductive tissues and mediate diverse aspects of cell–cell communication in plant reproduction and development, such as pollen-tube guidance (Silverstein et al., 2007; Okuda et al., 2009; Kanaoka et al., 2011; Marshall et al., 2011). Rapid alkalinization factor (RALF)-like peptides are also cysteine-rich peptides, and constitute a large multigenic family that is expressed in all tissues, with greater or lesser specificity depending on the gene (Olsen et al., 2002). Mature RALF peptides are processed from a precursor via subtilase activity near a dibasic site (Matos et al., 2008; Srivastava et al., 2009). Although some RALF-like genes are specifically expressed in the FG (Yu et al., 2005; Johnston et al., 2007), most currently described RALFs appear to be involved in plant growth processes. Incubation of tomato roots (Solanum lycopersicum) or A. thaliana seedlings with NtRALF from Nicotiana tabacum results in root growth arrest (Pearce et al., 2001), while silencing of NaRALF from Nicotiana attenuata leads to an increase in root growth (Wu et al., 2007). Similarly, over-expression of AtRALF1 in A. thaliana causes a semi-dwarf phenotype (Matos et al., 2008), and AtRALF23 over-expression inhibits hypocotyl elongation in seedlings, resulting in shorter and bushier plants (Srivastava et al., 2009). At the cellular level, some RALFs are involved in cell elongation, with the pollen-specific tomato SlPRALF regulating pollen tube elongation (Covey et al., 2010) and the sugarcane (Saccharum spp.) SacRALF1 inhibiting cell elongation of microcalli in cell suspension cultures (Mingossi et al., 2010). However, other RALFs, such as NtRALF, appear to block cell division, as labeling with 3H-thymidine is strongly inhibited in the tips of roots treated with exogenous RALF peptide (Moura et al., 2006). Most importantly, RALFs appear to act in a receptor-dependent pathway: (i) tobacco RALF accumulates in the cell wall (Escobar et al., 2003), (ii) tomato RALF binds to two transmembrane proteins of 25 and 120 kDa (Scheer et al., 2005), and (iii) AtRALF1 induces a Ca2+ increase and potentially act through an inositol triphosphate signaling pathway (Haruta et al., 2008).

In Solanum chacoense, five RALF-like genes have been isolated from fertilized ovules (Germain et al., 2005). In the present study, we present evidence for direct communication between the sporophyte and the female gametophyte through the action of a secreted ScRALF3 peptide that regulates various aspects of ES development. Although the gene is normally expressed within the sporophytic tissue of the ovule, ScRALF3 interference mutants displayed ES development delay or arrest. Therefore, a model is proposed in which ScRALF3 may act as a sporophytically expressed gene that regulates, at least in part, the architecture of the ES in a receptor-dependent signaling pathway.

Results

ScRALF3 transcripts are induced during fruit initiation

ScRALF3 was previously showed to be preferentially expressed in fertilized ovaries, with a sharp increase 2 days after pollination (Germain et al., 2005). To assess whether ScRALF3 expression correlates with fertilization or whether it may also be pollination-induced, an extended time-course analysis was performed (Figure 1a). Precise pollination timing was accomplished by manual pollination using pollen from a fully compatible S. chacoense genotype on the day of anthesis. ScRALF3 expression was not detected before 48 h, i.e. after fertilization, which occurs between 36 and 42 h post-pollination (Chantha et al., 2006; Tebbji et al., 2010). To determine whether ScRALF3 expression is regulated by fertilization or by fruit initiation, steady-state mRNA levels were compared in ovaries between 1 and 3 days after pollination or gibberellin treatment. Gibberellins are known to stimulate fruit development without fertilization in numerous plants, including solanaceous species (Vriezen et al., 2008; de Jong et al., 2009b). Application of a 1 mm GA3 solution directly onto the ovary led to development of small parthenocarpic fruit (data not shown). However, ScRALF3 mRNA levels were not increased by GA3 treatment (Figure 1b), leading to the conclusion that ScRALF3 expression correlates with post-fertilization events. Auxins are also known to initiate fruit development (de Jong et al., 2009b). Various concentrations of naphthaleneacetic acid (NAA) were tested for their ability to produce parthenocarpic fruit and to stimulate ScRALF3 expression. Only external treatments with 100 μm NAA (very weak expression) or a higher concentration (1 mm NAA, stronger expression) increased ScRALF3 expression (Figure 1b), and also led to initiation of fruit development in a concentration-dependent manner. An increase in ScRALF3 mRNA levels was observed 24 h after the first application of 1 mm NAA, and the level peaked after 3 days (Figure 1b). Therefore, we associate ScRALF3 expression with early post-fertilization events under the control of auxin.

Figure 1.

ScRALF3 is an auxin-responsive gene during fruit initiation. (a) Time-course analysis of ScRALF3 expression around fertilization (36–42 h). (b) Sensitivity of ScRALF3 expression to externally applied auxin and gibberellin. Hybridization with a ribosomal 18S probe shows equal loading. EtOH, ethanol control; GA3, gibberellin A3; NAA, naphthaleneacetic acid. h, hours after pollination; d, days after applied treatment.

The ScRALF3 promoter contains auxin regulatory elements

A 1.2 kb fragment of the ScRALF3 promoter was isolated by genome walking. We previously showed that ScRALF3 is a single-copy gene in S. chacoense (Germain et al., 2005). To assess allelic variation between ScRALF3 promoters, six amplicons of various lengths were cloned from the wild-type (WT) S. chacoense diploid genotype used (G4; S12S14 self-incompatibility alleles). Six primer pairs were used to amplify six overlapping lengths of the promoter (Table S2). Sequencing of five independent clones for each amplicon revealed two slightly different sequences in comparison with the original 1.2 kb promoter obtained by genome walking. The two allelic promoters shared 89.7% identity, with the differences being found mainly in repeated sequences.

Comparison with ScRALF3 orthologous sequences found in the genomes of Solanum tuberosum and Solanum lycopersicum showed a high degree of sequence identity and colinearity, except for an insertion in the S. chacoense RALF3 promoter. The ScRALF3 promoter was analyzed for known cis-regulatory elements (Table S1), especially auxin regulatory elements (Nemhauser et al., 2004; Berendzen et al., 2012). Although the canonical sequence for the AuxRE element (TGTCTC) was not found in the ScRALF3 promoter, four AuxRE variants (core TGTC elements and the AuxRE-related sequence AUX2) were found in the proximal promoter region, and two such variants were also found in the sequence shared by the two other Solanum species analyzed. Nemhauser et al. (2004) have shown that 42% of Arabidopsis auxin responsive promoters contain at least one pair of TGTC elements interspaced by 50 bp, which is the case in the ScRALF3 proximal promoter region. Other studies also revealed the requirement for a core TGTC element for ARF transcription factor binding (Ulmasov et al., 1995, 1997). One AuxRR core element (GGTCCAT), an auxin enhancer element (Sakai et al., 1996), was also found in the proximal promoter region. In addition to these specific auxin response elements, an extended DOF core motif (ACTTTA motif), which is the binding site of the NtBBF1 DOF factor that has been shown to be involved in auxin-induced gene expression in Nicotiana tabacum (Baumann et al., 1999) as well as Arabidopsis (Nemhauser et al., 2004), is found twice in the ScRALF3 promoter. Other motifs that have been recently shown to be enriched in auxin-regulated promoters (Berendzen et al., 2012) are also present in the ScRALF3 proximal promoter region (Table S1). As slightly divergent AuxRE elements have been found to confer auxin responsiveness in the presence of other elements, forming bipartite or tripartite modules (Berendzen et al., 2012; Walcher and Nemhauser, 2012), and exogenously applied auxin regulates ScRALF3 expression (Figure 1), the results suggest that ScRALF3 expression is under auxin control.

ScRALF3-silenced Solanum plants display a severe reduction in seed set

The region used for the RNAi construct (150 bp long) shared 24–51% nucleic acid identity between the five ScRALF members previously isolated (Germain et al., 2005), as well as eight RALFs newly identified by deep transcriptomic sequencing of S. chacoense tissues (Table S3), with no conserved stretches of more than 11 nt. When taking into account only the ScRALF genes expressed in ovule tissues, the size of these stretches is limited to 9 nt, further limiting a non-specific interference effect. Twelve plants had lower or non-detectable levels of ScRALF3 steady-state mRNA in T1 plants, as determined by semi-quantitative RT-PCR analyses (Figure 2a). To determine whether expression of other members of the RALF family may have been affected by the ScRALF3 RNAi construct, three lines (lines 7, 11 and 12) were analyzed for the expression of RALF-like genes in ovule tissues (Figure S1). None of the RALF-like genes tested showed signs of down-regulation that may have been caused by the ScRALF3 RNAi construct. A DNA gel-blot analysis was also performed using genomic DNA from four transgenic plants (lines 7, 11, 12 and 14), showing that these four RNAi lines derived from independent and unique tranformation events (Figure S2).

Figure 2.

Down-regulation of ScRALF3 by RNAi causes production of small fruits with fewer seeds. (a) ScRALF3 mRNA levels in selected RNAi lines determined by semi-quantitative RT-PCR and mean diameter (cm) of 24-day-old fruits. = 10–15 fruit per line. (b) Cross-sections of 30-day-old fruits from wild-type (G4) and selected transgenic plants taken together. (c) Time-course analysis (d, days) of fruit size development following pollination (= 25–30 fruit). (d) Seed number in 25-day-old fruit (= 15 fruits). Asterisks indicate statistically significant differences compared with wild-type (*< 0.05; **< 0.0001). Values in (c) and (d) are means ± standard deviation.

To assess whether ScRALF3 is required for reproductive development, fruit size was compared between WT non-transformed plants (G4) and the 12 RNAi lines, 24 days after pollination. Seven plants had significantly smaller fruits than WT (< 0.05), with a sampling of 10–15 fruits per line (Figure 2a,b). To further characterize this phenotype, we performed a time-course analysis on three independent transgenic plants (lines 11, 12 and 14) with a sample of 25–30 fruits per line. In these lines, ScRALF3 expression was not detected by RT-PCR analysis (Figure 2a). RNAi down-regulation of ScRALF3 slowed down development of the fruit, mainly between 10 and 15 days post-pollination (Figure 2c), and resulted in a marked seed set reduction of 65–80% compared to WT plants (Figure 2d). In comparison, all plants carrying an empty vector with a kanamycin resistance gene showed normal seed set (data not shown). Therefore, ScRALF3 is required for normal seed development.

Female gametophyte development is impaired in mutant ovules

Reduced seed set may result from a defective ES or from early zygotic developmental defects. Clearing analyses of ovules at anthesis were performed to detect possible defects during ES development, which follows the Polygonum-type pattern in WT S. chacoense ovules (Figure 3a–k). In WT plants, a four-cell structure (two synergids, one egg cell and one central cell) characterizes the mature FG. Within the egg apparatus, the synergids may be distinguished from the egg cell as their nuclei are in close proximity to the micropyle, while the egg cell nucleus is close to the central cell nucleus, with a vacuole reaching to the micropyle entry (Figure 3j,k). However, in the RNAi lines, 61–77% of ovules showed premature arrest of their ES (Table 1), consistent with the number of missing seeds. In ES that continued their development, the embryos were viable (data not shown), indicating no disruption of zygote development in mutant fruits. Thus, defects detected during ES development appear to be solely responsible for the reduced seed set observed. Of the four RNAi lines analyzed, lines 7, 11 and 12 had approximately 50% of ovules arrested during the mitosis steps. In lines 11 and 12, 19–24% of ovules had a collapsed or empty ES, and this percentage increased to 50% for line 14. Together, these data suggest that ScRALF3 is required for progression through megagametogenesis, or to complete megasporogenesis.

Table 1. Immature embryo sac at anthesis in down-regulated ScRALF3 plants
GenotypeAbnormalFMMIMIIMIIIES UPNES
  1. Percentage of ES arrested at various developmental stages at anthesis. = 100 ovules per plant line. Abnormal sacs comprise collapsed or empty embryo sacs; FM, functional megaspore; MI/MII/MIII, mitosis I, II and III; ES UPN, embryo sac with unfused polar nuclei; ES, embryo sac.

G46292
RNAi3-72212366933
RNAi3-11241015175227
RNAi3-1219418296123
RNAi3-1450451139
Figure 3.

Embryo sac development within wild-type Solanum chacoense ovules. a-k, cleared ovules.(a) Archeosporial cell. (b) Megaspore mother cell. (c) Two haploid spores. (d) Tetrad of haploid spores. (e) Functional megaspore. (f) Two-nucleate stage. (g) Four-nucleate stage. (h) Eight-nucleate stage. (i) Cellularized ES with unfused polar nuclei. (j) Mature embryo sac. (k) Mature embryo sac: gray, maternal sporophytic tissue; yellow, central cell; green, synergids; blue, egg cell; n, nucellus; i, integument; c, chalaza; m, micropyle.

As anomalies in the sporophytic tissue of the ovule strongly affect FG development (Bencivenga et al., 2011), cleared ovules were carefully examined. No major developmental defect of the sporophytic tissue surrounding the ES was observed by differential interference contrast microscopy. Therefore, the phenotype is directly linked to a dysfunctional genetic and/or signaling pathway in the ES of the mutant lines.

In WT plants, ovule maturation is highly synchronous, and gametophyte development stages correlate well with bud size (Table 2). In 3–4 mm buds, the FG is progressing through megasporogenesis. A megaspore mother cell (MMC) has differentiated in the nucellus from an archeosporal cell (Figure 3a). At this stage, the integument is at the level of the MMC (solanaceous plants are unitegmic). As the integument continues to grow and reaches the upper part of the MMC, the MMC elongates, with a nucleus at the tip of the primordium (Figure 3b). The integument has almost covered all the nucellus when the FG starts its first meiosis, resulting in two tear-shaped cells that characterize this step (Figure 3c). When meiosis is complete, the integument fully covers the FG (Figure 3d) and degeneration of the three megaspores nearer to the micropylar pole occurs (data not shown). The remaining cell will eventually differentiate into an FM (Figure 3e). The FM nucleus relocates slowly toward the middle of the cell, and a first division on the chalaza/micropyle axis occurs when the flower buds reach 4–5 mm. The two nuclei then move, one to each pole of the ES (Figure 3f), followed by a second (Figure 3g) and a third (Figure 3h) mitosis in 5–6 mm flower buds. Cellularization starts in flower buds of 6–7 mm. The three nuclei located at the chalazal pole that give rise to the antipodals degenerate almost concomitantly with cellularization; thus, as for other Solanum spp., antipodals are ephemeral and rarely observed (Estrada-Luna et al., 2004). The polar nuclei then move closer to each other, and vacuoles develop inside the synergids and the egg cell (7–8 mm buds, Figure 3i). The two polar nuclei, often in close proximity to the egg cell nucleus, fuse to form a mature ES (Figure 3j,k).

Table 2. Megasporogenesis and megagametogenesis in wild-type (G4) and mutant plants (RNAi)Thumbnail image of

The same analysis was performed to assess at which stage the defect occurs in interference lines. We selected line 7 on the basis of its low percentage of degenerated ES, as well as line 11 on the basis of its lack of ScRALF3 expression and its similar phenotype to line 12. Early megasporogenesis appears to occur normally in mutant ovules; in 3–4 mm buds, the distribution of developing ES is within 95% of the one observed for WT ovules (Table 2). However, early on, irregularities are observed in the mutant FGs. Although 33% of the WT FGs began their mitotic cycle, only 21% (line 7) and 18% (line 11) of the mutant FGs proceeded beyond the FM stage in 4–5 mm buds. Slower progression at the end of megasporogenesis is observed in line 7, while more degenerated ES are observed in line 11. Later, although 62% of the WT FGs began cellularization, only 20% (line 7) and 5% (line 11) of the mutant FGs had reached this step in 6–7 mm buds (Table 2). In fact, only 33% (line 7) and 27% (line 11) of the mutant ES became fully mature at anthesis (Table 1). Therefore, the ES from these mutant lines are having difficulty progressing through the mitotic cycles, and are more likely to arrest earlier during megagametogenesis or to degenerate. The results are slightly different for line 14, in which degenerated ES clearly correlate with abnormal FM differentiation. Several shape abnormalities were observed, and their nucleus appears fractionated (Figure 4a,b). Also, in 5–6 mm buds, only 23% of the mutant FGs began the mitotic cycles, compared with 95% for the WT ovules. In these buds, the mutant ES are either stopped at the FM, or have degenerated (Table 2). In summary, ScRALF3 is important during all stages of megagametogenesis, but appears to play a key role early on, either to regulate proper differentiation of the FM and/or to prime the FM for the mitotic cycles.

Figure 4.

Loss of embryo sac polarization in mutant ovules is associated with a wide range of nuclei distribution. a-m, cleared ovules.(a) Degenerated embryo sac. (b) Degeneration of the functional megaspore (FM). (c) Division of the FM perpendicular to the micropylar/chalazal axis (dashed line). c, chalaza; m, micropyle. (d–h) Abnormal nuclei positioning at the four-nucleate stage: (d) stacked image; (e–h) individual focal planes. (i–m) Clustered nuclei at the eight-nucleate stage: (i) stacked image; (j–m) individual focal planes. (n) Phenotypic analysis of mutant embryo sacs (four-nucleate stage) showed no preference with regard to nuclei positioning (= 100). False colors are used to emphasize the nuclei in different focal planes. Scale bars = 10 μm.

The embryo sacs of ScRALF3-silenced plants display abnormal nuclear distribution and asynchronous nuclear divisions

Organization of the ES was not necessarily normal in arrested or developing mutant ovules. Instead of having an equal number of nuclei at each pole, the nuclei were either aligned (Figure 4d–h) or clustered (Figure 4i–m) within the syncytium following mitosis I, II or III. These observations suggest that the mitotic divisions are disorganized, and that the ES is no longer polarized in a chalazal/micropylar axis. To further characterize the lack of polarization inside the ES in mutant ovules, localization of nuclei was assessed within the syncytium of 100 ovules at mitosis II for the WT and two mutants (lines 7 and 12). This stage was selected for the analysis for two reasons: (i) there was still a sufficient number of FGs reaching this stage even in mutant ovules, and (ii) fewer misinterpretations may occur as the result of incomplete nuclei migration from a first mitosis in progress. In WT plants, 95% of the FGs showed a normal segregation pattern (two nuclei at each pole). In mutant lines 7 and 12, only approximately 35% of the FGs showed a normal nuclear segregation. Instead, the four nuclei were either localized on the chalazal or the micropylar side, or clustered in the middle of the syncytium, without any preference (Figure 4n). This random nuclei distribution appeared to be closely related to a lack of ES polarization, as the nucleus of the FM in some mutant ovules divided perpendicularly to the chalazal/micropylar axis (Figure 4c); this was not observed in WT ovules. Furthermore, abnormal numbers of nuclei (5–7) within the syncytium were more frequent than in WT plants, also suggesting an effect on mitotic division synchronicity. For example, in Figure 5(a–e), only two nuclei progressed through the third mitosis, leading to a syncytium containing six nuclei. When comparing one hundred ovules in WT versus mutant plants with FGs containing 5–8 nuclei (Figure 5f), 26% of the WT ovules showed asynchrony (5–7 nuclei); this percentage reached 53% (line 2) and 70% (line 7) in mutant ovules. Loss of molecular regulation of divisions or an indirect consequence of the abnormal nuclei distribution may explain the increased asynchrony in mutant FGs. Nevertheless, these data suggest that ScRALF3 is involved in polarization of the syncytium, influencing the migration, distribution and division of the nuclei.

Figure 5.

Asynchronous divisions within the syncytium of mutant ovules. (a–e) Six-nucleate syncytium clustered on one side of the ES: (a) stacked image; (b–e) individual focal planes. False colors are used to emphasize the nuclei in different focal planes. Scale bar = 10 μm. (f) Phenotypic analysis of mutant embryo sacs proceeding through the last mitosis showed an abnormal number of nuclei (5–7) (= 50). Color coding refers to the nuclei number and their positioning. Mycropylar (M), central (Ct), chalazal (Ch), normal (N).

ScRALF3 is mostly expressed in the sporophytic tissue of the ovule

To assess how down-regulation of the ScRALF3 transcripts affects FG development, an RNA expression analysis on ovaries from various developmental stages was performed (Figure 6a). Steady-state ScRALF3 mRNA levels are high in young buds and slowly decrease until anthesis, concomitant with the progression of megagametogenesis. As previously described (Germain et al., 2005), ScRALF3 expression is then up-regulated after fertilization. RNA in situ analyses revealed that the ScRALF3 mRNA is mostly localized in the sporophytic tissue of the ovule, mainly in the integument, but also in the placenta vasculature (Figure 6b–e,g–i). In smaller buds (≤4 mm), ScRALF3 expression was observed in the young primordium (Figure 6b), then throughout the growing integument around immature ES (Figure 6c). As ovules grew (buds > 4 mm), ScRALF3 expression became restricted to few structures, notably the tip of the integument (Figure 6d,e arrow). However, lower ScRALF3 expression remained throughout the integument, as revealed by the use of a higher probe concentration (Figure 6i). Therefore, a gradient in ScRALF3 expression is established within the developing ovules, with a higher concentration at the tip of the integument. Analysis of transgenic plants harboring a ScRALF3 promoter–GUS fusion confirmed the results obtained by in situ hybridization. Down-regulation of ScRALF3 expression gradually occurs as the ovules get closer to anthesis (Figure 7a–f). ScRALF3 is expressed in the chalaza of the primordium (Figure 7a), then within the integument (Figure 7b–d). ScRALF3 is also expressed within the nucellus cells near the micropyle, during degeneration of the non-functional megaspores (Figure 7d). Subtle expression remained during megagametogenesis, mostly at the tip of the integument and in the funiculus (Figure 7e), disappearing at anthesis (Figure 7f). Together, this suggests that ScRALF3 is a sporophytically expressed gene, with an apparent polarized expression around the developing embryo sac. Peak expression of ScRALF3 occurs early during ovule development, mostly during megasporogenesis and preceding the appearance of any visible phenotype.

Figure 6.

Regulation of ScRALF3 expression within the sporophytic tissue during ovule development. (a) RNA gel blot of ScRALF3 transcripts in de-pericarped ovaries during ovule and fruit development. mm, bud size; DAP, days after pollination. (b–e) In situ localization of ScRALF3 transcripts using an ScRALF3 antisense probe (50 ng) for ovaries from (b) 3 mm, (c) 4 mm, (d) 5 mm and (e) 6 mm buds. (f) In situ localization of ScRALF3 transcripts using an ScRALF3 sense probe (50 ng) for ovaries from 5 mm buds. (g, h) Close-up of ovules of 5–6 mm buds. (i, j) In situ localization using (i) an ScRALF3 antisense probe (200 ng) and (j) its corresponding sense probe for ovaries from 4 to 5 mm buds. Scale bars = 100 μm.

Figure 7.

Expression of ScRALF3 in the sporophytic tissue of the ovule as revealed by GUS staining. (a) Staining in the chalaza of the developing primordium. (b, c) Re-localization of ScRALF3 expression within the integument during megasporogenesis. (d) Expression in the integument, and partially in the nucellus, as the haploid spores degenerate. (e) Subtle expression at the tip of the integument and in the chalaza during megagametogenesis (FM shown here). (f) ScRALF3 expression decreases over the mature stage. Scale bars = 20 μm. n, nucellus; i, integument; c, chalaza; m, micropyle; MMC, megaspore mother cell; FM, functional megaspore.

ScRALF3 is a secreted peptide

To assess whether the ScRALF3 peptide acts as a signal molecule between the sporophyte and the gametophyte, we determined its subcellular localization by fluorescence microscopy following transient expression in onion cells (Figure 8). As expected from the presence of a signal peptide at the ScRALF3 N-terminus (Figure 8a,b), cells transformed with the full ScRALF3–GFP construct showed tangled lines or a meshed network of GFP fluorescence, as well as fluorescent dots moving along the mesh (Figure 8d,e). This pattern is reminiscent of Golgi bodies moving along ER strands (Hawes et al., 1998; Nelson et al., 2007), as revealed using ER and Golgi markers (Figure 8g,h). Transient expression of an ScRALF3–GFPHDEL construct also confirmed its translocation to the ER (Figure 8c,f). Stably transformed S. chacoense plants were also produced using these constructs, but it was impossible to clearly observe the GFP signal due to high autofluorescence background in leaves.

Figure 8.

ScRALF3 passes through the secretory pathway in an onion cell transient expression assay. (a) The ScRALF3 signal peptide (SP) restricts GFP to the mesh network of the ER. (b) Without the SP, a blurred signal in the cytoplasm is observed, with strong fluorescence within the nucleus. (c, f) The presence of the ER retention signal HDEL increases fluorescence in the ER. (d) ScRALF3 in Golgi bodies (arrow) moving along the ER strand. The numbers in yellow indicate time in seconds (s). (e) ScRALF3 in the ER. (g, h) ER marker (g) and Golgi marker (h) (Nelson et al., 2007). (i) Chimeric proteins used in the transient expression assay. Images were taken either with an epifluorescent microscope (a–d) or a confocal microscope (e–h).

Nonetheless, these transgenic lines constitutively expressing ScRALF3 (R3) or ScRALF3-HDEL (R3-HDEL) were used for fusion protein analyses using an anti-GFP antibody (Figure 9). A very weak signal at approximately 44 kDa was associated with the full chimeric protein (42 kDa), and revealed possible constitutive secretion of the protein in the R3 leaf protein extract (Figure 9a). In line with this, there is a strong accumulation of an approximately 30 kDa protein, often associated with the core GFP, a fragment of the protein that is released after partial digestion within the vacuole or the apoplast (Batoko et al., 2000; Jung et al., 2011). In contrast, in the R3–HDEL leaf extracts, the full protein accumulated to higher levels than the core GFP, with an additional signal at approximately 35 kDa (Figure 9b). This approximately 35 kDa band is consistent with processing of the ScRALF3 pro-region through the action of subtilisin-related proteases (Germain et al., 2005), and corresponds well with the expected approximately 36 kDa mature peptide. Concomitant with the decrease in core GFP accumulation, the chimeric protein is stabilized when kept within the secretory pathway. These results support those obtained by transient expression in onion cells, in which a better signal to background ratio was observed in R3–HDEL cells. Also, we suspect that ScRALF3 is processed within the median- or trans-Golgi apparatus, as the HDEL retention signal is used to return proteins to the ER from the cis-Golgi and thus, fewer proteins would be processed. In summary, ScRALF3 is associated with the secretory pathway.

Figure 9.

ScRALF3 is processed within the secretory pathway in S. chacoense. (a) ScRALF3 is constitutively secreted in the apoplast as revealed by accumulation of core GFP (approximately 29 kDa). (b) The ER retention signal stabilizes the unprocessed (44 kDa) and processed (36 kDa) ScRALF3 chimeric proteins. Three independent transgenic plants were analyzed for each construct. Coomassie blue staining was used as a loading control. See also Movies S1–S4.

Discussion

Cell–cell communication is essential during ovule and early fruit development to ensure a viable progeny (Nowack et al., 2010; Chevalier et al., 2011). ScRALF3, a small secreted peptide for which transcripts are down-regulated in the ovule until anthesis and up-regulated after fertilization, is thus a good candidate for contribution to these signaling events.

A small-fruit phenotype was observed following down-regulation of the ScRALF3 transcript in transgenic plants. Recently, Bonghi et al. (2011) used a microarray approach to analyse the peach transcriptome, revealing that auxin, cytokinin and gibberellin are good signaling candidates, acting directly or indirectly on the cross-talk between the seed and the pericarp regulating fruit growth (Bonghi et al., 2011). In line with this, cis-regulatory elements found in the ScRALF3 promoter suggested that ScRALF3 expression is directly or indirectly driven by auxin, as shown in Figure 1. ScRALF3 expression following pollination is similar to the pattern of auxin accumulation observed in young tomato fruit (Mapelli et al., 1978), which mainly stimulates cell division (Bünger-Kibler and Bangerth, 1983). The phenotype, which appeared 10–15 days after pollination, may be caused by inhibition of cell division in the first 10–14 days, preceding the 6–7 weeks of cell expansion that are typical of many solanaceous fruits (Gillaspy et al., 1993; de Jong et al., 2009a). However, fewer seeds are also produced within the growing fruit, and a positive correlation between final fruit size and seed set is well established (Varga and Bruinsma, 1976). Production of hormones by the embryo (seed) is required to stimulate fruit growth, explaining the small-fruit phenotype in parthenocarpic fruits (Mapelli et al., 1978; Sjut and Bangerth, 1983), and probably also in the ScRALF3 mutant lines.

Cellular analysis of ScRALF3 mutant ovules revealed a high proportion of arrested ES during megagametogenesis, accounting for the lower seed set observed. The mutant syncytia showed abnormal segregation of their nuclei and asynchronous division, cellular defects that are partially reminiscent of cytoskeleton disruption. In maize, the abnormal microtubular organization within indeterminate gametophyte1 mutant ES is not associated with cell division arrest (Huang and Sheridan, 1994, 1996). In Arabidopsis, kinesin mutants cause mis-positioning of nuclei that occurs later during megagametogenesis (FG5 stage) (Tanaka et al., 2004). Thus, microtubules are important for FG organization, and ScRALF3 may be required for cell organization within the developing ES. The arrest in gametophyte development may also be explained by other general cell mechanism defects affecting cell signaling and/or DNA remodeling. In Arabidopsis, mutations in a transcription factor (agl23), a histone acetyltransferase (ham1 ham2) or an E3 ligase (rhf1a/rhf2a) also lead to partial arrest at the FG1 stage (Colombo et al., 2008; Latrasse et al., 2008; Liu et al., 2008).

The gradual down-regulation of ScRALF3 expression during FG development and the various phenotypes observed suggest that ScRALF3 may effectively regulate a general cell mechanism affecting progression through cell division, rather than a stage transition. For example, the CKI1 signaling pathway regulates vacuole development within the syncytium, indirectly regulating the organization of the ES at several steps of megagametogenesis; mutations affecting the pathway lead to a later arrest at stage FG5 (Pischke et al., 2002; Hejatko et al., 2003). Also, a mutation in a lysophosphatidylacyltransferase causes mis-development of the ER within the FG; differences in the mutant ovule are apparent at the four-cell stage (FG7), and dominant expression of the gene occurs during the four- and eight-nucleate stages (FG3–FG5) (Kim et al., 2005). Over-expression of ScRALF3 had no effect on seed set and FG maturation (data not shown): ScRALF3 expression during megasporogenesisis is therefore more important than its down-regulation during megagametogenesis, and may be required to prime the FM for future divisions.

ScRALF3 may act in a non-cell autonomous pathway to regulate FG development. Recently, a sporophytic cytokinin receptor has been shown to regulate FG development, which suggests an interesting dialog between the sporophyte and the gametophyte (Kinoshita-Tsujimura and Kakimoto, 2011). Moreover, it was shown that auxin may be produced first in the sporophyte, and then synthesized within the FG in a concentration-dependent manner, regulating the cell fate of the ES (Pagnussat et al., 2009). Tissue localization of the ScRALF3 transcript showed a strong accumulation in the sporophytic tissue surrounding the FG. However, this strong sporophytic signal may mask a fainter gametophytic signal. ScRALF3 is also a secreted peptide that is processed within the secretory pathway. ScRALF3 may therefore act like a signalling peptide between the sporophyte and the gametophyte, especially as the observed FG phenotype is similar to that observed in mutants lacking integuments (Bencivenga et al., 2011).

In a proposed model, ScRALF3 is a sporophytically expressed signaling factor regulating gametophytic cell organization in a receptor-dependent signaling pathway. Outward signaling has already been shown to have a predominant effect on the cell architecture. ROPs (Rho of plants) represent one such principal regulator that control cytoskeletal organization in plants (Mucha et al., 2011), and are thought to act downstream of receptor-like kinases, especially during pollen tube growth (Kaothien et al., 2005; Zhang and McCormick, 2007). Moreover, brassinosteroids control plant growth by modulating microtubular organization, possibly through transcriptional regulation (Catterou et al., 2001; Vert et al., 2005). In addition, characterization of a dynamin mutant supports a model in which external signals may be important to regulate FG development, in line with our model. This mutant shows arrest prior to the first division, without abnormalities of the membrane, supporting a signaling defect rather than a structural defect, two processes that are regulated by dynamins (Backues et al., 2010).

In summary, the ScRALF3 secreted peptide, for which the gene is transcribed within the sporophytic tissue surrounding the young ES, affects FG development. Cellular analysis of the mutant revealed an abnormal organization within the syncytium, with aberrant cell division, mis-positioning of the nuclei and loss of polarity. As suggested in a previous comparative transcriptomic analysis (Johnston et al., 2007), ScRALF3 may thus be a candidate peptide regulating cell–cell communication between the sporophyte and the gametophyte.

Experimental procedures

Materials

All plant material has been described previously (Germain et al., 2005). The primers used are listed in Table S2. Differential interference contrast and fluorescence microscopy were performed using a Zeiss Axio Imager M1 microscope; pictures were taken using a Zeiss AxiocamHRc camera (Zeiss, http://www.zeiss.com). Confocal microscopy was performed using a Zeiss LSM 510 META microscope with an argon ion laser emitting at 488 nm and an LP 505 emission filter.

RNA extraction and Northern hybridization

Total RNA was extracted using TRIzol reagent (Invitrogen, http://www.invitrogen.com). RNA preparation, transfer and hybridization were performed as described previously by Germain et al. (2005). Radioactive probes were produced using a High Prime DNA labeling kit (Roche, http://www.roche-applied-science.com) and [α-32P]dATP (Perkin Elmer, http://www.perkinelmer.com).

Hormone treatment

Treatments consisted of pipetting 2 μl of the appropriate solution on top of the ovary at anthesis (method adapted from de Jong et al., 2009b; Vriezen et al., 2008). Approximately 20 flowers were treated. Ovaries were dissected between 6 h and 3 days after the initial treatment for RNA extraction. Some treated flowers were left on the plant to monitor the production of parthenocarpic fruit.

Genome walking and allele isolation

Solanum chacoense genomic DNA was isolated using a plant DNA extraction kit (Qiagen, http://www.qiagen.com). The promoter region was isolated using a Clontech genome walker kit (http://www.clontech.com). DraI and EcoRV were used to create the genomic library. The secondary PCR products were cloned into pCR4-TOPO using a TOPO TA cloning kit (Invitrogen). Sequencing reactions were performed at the Université de Montréal Genomic Platform (Montréal, Canada). Primers were designed based on the longest DNA sequence obtained to isolate the various alleles (Table S1). The sequences were cloned into pDONRtm/Zeo (Invitrogen). The Genbank accession number for the ScRALF3 promoter is KC222019.

Plant transformation and selection

Transgenic plants were obtained by leaf agroinfiltration and callus regeneration as described previously (Matton et al., 1997) using Agrobacterium tumefaciens strain LBA4404 carrying the RNAi vector pK7GWIWG2(I)-ScRALF3. GATEWAY® technology using pDONRtm/Zeo (Invitrogen) as the entry vector and pK7GWIWG2(I) (Karimi et al., 2002) as the destination vector was used for cloning. Fifteen independent shoots were chosen and grown until maturity in a greenhouse. Twelve plants showing no signs of tetraploidy were chosen for phenotype analysis. Semi-quantitative RT-PCR was performed using 600 ng RNA from 4 days post-pollination ovaries using the M-MLV reverse transcriptase kit (Invitrogen) according to the manufacturer's instructions. Forty-two cycles of PCR were required to amplify ScRALF3 and UBIQUITIN using HOTSTART Taq polymerase (Bioshop Canada Inc., http://www.bioshopcanada.com). Statistical analyses were performed using the GraphPad software t-test calculator (GraphPad Software Inc., http://www.graphpad.com).

Ovule clearing

Bud size measurements were taken at the baseline of the buds for developmental classification. Ovaries without the pericarp were fixed in FAA overnight (1% formaldehyde/0.5% glacial acetic acid/50% ethanol) (Fisher Scientific, http://www.fishersci.com). The ovaries were incubated in 100% ethanol for 1 h, and transferred into methyl salicylate/EtOH solutions with increasing ratios of methyl salicylate/EtOH for 30 min each (1:3, 1:1 and 3:1). The tissues were kept in 100% methyl salicylate (Sigma-Aldrich, http://www.sigmaaldrich.com) at room temperature until observation.

In situ hybridization

The method used for in situ hybridization was adapted from Lantin et al. (1999). VistaVision HistoBond microscope slides (VWR, https://ca.vwr.com) were used with 10 μm tissue sections. The riboprobes were synthesized from the ScRALF3 cDNA (Germain et al., 2005) using digoxigenin-11-UTP and a MAXIscript T7/T3 kit (Invitrogen).

GUS staining

Transgenic plants were obtained as described above. GATEWAY® technology, using pDONRtm/Zeo (Invitrogen) as the entry vector and pMDC162 (Curtis and Grossniklaus, 2003) as the destination vector, was used for cloning. The transgenic plants were selected by PCR screening for the hygromycin resistance gene. Ten independent lines were carefully examined using the GUS staining protocol described by Weigel and Glazebrook (2002). Ovule clearing was performed to observe GUS expression within the ovules. The tissues were kept in methyl salicylate (Sigma-Aldrich) at 4°C for less than 2 days to avoid fading of the GUS stain.

Onion cell bombardment

GATEWAY® technology, using pDONRtm/Zeo (Invitrogen) as the entry vector and pMDC83 or pMDC201 (Curtis and Grossniklaus, 2003), containing a GFP and a GFP6HDEL marker, respectively, as the destination vector, was used for cloning. Onion cells were transiently transformed by microparticle bombardment (Germain et al., 2008).

Western blotting

Leaf proteins were extracted using 100 mM Tris/HCl, pH 8, 0.1% SDS, 2% β-mercaptoethanol and 1× Roche Complete mini protease inhibitor, then quantified by the Bradford assay. Protein samples (25 μg) were denatured in Laemmli sample buffer at 95°C/5 min, subjected to 12.5% SDS–PAGE as described by Laemmli (1970), then transferred to a polyvinyldene fluoride (PVDF) membrane (GE Healthcare, http://www.gelifesciences.com). Signal detection using anti-GFP (1:1000) (Roche) and anti-mouse IgG antibodies conjugated with horseradish peroxidase (1:5000) (Sigma) was performed using Amersham™ ECL™ Plus (GE Healthcare) detection reagents according to the manufacturer's instructions, and exposed on a BioflexEcono film (Mandel, http://www.interscience.com).

Acknowledgments

This work was supported by the Natural Sciences and Engineering Research Council of Canada and the Canada Research Chair program. E.C. is the recipient of PhD fellowships from the Natural Sciences and Engineering Research Council of Canada and from Le Fonds Québécois de la Recherche sur la Nature et les Technologies. A.L-H. is the recipient of a MSc fellowship from Le Fonds Québécois de la Recherche sur la Nature et les Technologies.

Ancillary