Role of cis-12-Oxo-Phytodienoic Acid in Tomato Embryo Development^,,

Abstract

Oxylipins including jasmonates are signaling compounds in plant growth, development, and responses to biotic and abiotic stresses. In Arabidopsis (Arabidopsis thaliana) most mutants affected in jasmonic acid (JA) biosynthesis and signaling are male sterile, whereas the JA-insensitive tomato (Solanum lycopersicum) mutant jai1 is female sterile. The diminished seed formation in jai1 together with the ovule-specific accumulation of the JA biosynthesis enzyme allene oxide cyclase (AOC), which correlates with elevated levels of JAs, suggest a role of oxylipins in tomato flower/seed development. Here, we show that 35S::SlAOC-RNAi lines with strongly reduced AOC in ovules exhibited reduced seed set similarly to the jai1 plants. Investigation of embryo development of wild-type tomato plants showed preferential occurrence of AOC promoter activity and AOC protein accumulation in the developing seed coat and the embryo, whereas 12-oxo-phytodienoic acid (OPDA) was the dominant oxylipin occurring nearly exclusively in the seed coat tissues. The OPDA- and JA-deficient mutant spr2 was delayed in embryo development and showed an increased programmed cell death in the developing seed coat and endosperm. In contrast, the mutant acx1a, which accumulates preferentially OPDA and residual amount of JA, developed embryos similar to the wild type, suggesting a role of OPDA in embryo development. Activity of the residual amount of JA in the acx1a mutant is highly improbable since the known reproductive phenotype of the JA-insensitive mutant jai1 could be rescued by wound-induced formation of OPDA. These data suggest a role of OPDA or an OPDA-related compound for proper embryo development possibly by regulating carbohydrate supply and detoxification.

Oxylipins are lipid-derived compounds such as jasmonic acid (JA), its precursor 12-oxo-phytodienoic acid (OPDA), and its metabolites like amino acid conjugates or hydroxylated derivatives. Oxylipins, preferentially JA and its Ile conjugate (JA-Ile), regulate various aspects of growth, development, and responses to biotic and abiotic stresses in plants (Wasternack, 2007; Browse and Howe, 2008; Howe and Jander, 2008; Browse, 2009a). All genes encoding enzymes of JA biosynthesis have been cloned from several plant species and the corresponding enzymes including isoforms have been characterized (Schaller and Stintzi, 2009). Enzymes active in OPDA formation such as a lipase, a 13-lipoxygenase, an allene oxide synthase, and an allene oxide cyclase (AOC) are located in the chloroplast, whereas OPDA reduction by the OPDA reductase3 (OPR3) and shortening of the carboxylic acid side chain by the fatty acid β-oxidation machinery takes place in peroxisomes. JA can be metabolized by several reactions. Among them conjugation with Ile catalyzed by JAR1, the JA amino acid conjugate synthase (Staswick and Tiryaki, 2004), leads to the most bioactive compound being crucial for the JA signaling cascade (Chung et al., 2009; Fonseca et al., 2009). All enzymes of jasmonate biosynthesis occur in resting, unstimulated cells. This allows JA formation within less than 5 min upon any stimulus that finally results in expression of JA biosynthesis genes and other JA-responsive genes (Glauser et al., 2009; Koo et al., 2009). The regulation of JA biosynthesis by substrate availability and by a positive feedback loop is well described by different assays including transgenic approaches (Laudert et al., 2000; Stenzel et al., 2003).

A role of jasmonates in wound- and herbivore-induced responses of leaves (for review, see Wasternack, 2007; Koo and Howe, 2009) and resistance to necrotrophic pathogens (Pieterse et al., 2009) is well documented. Much less is known about the function of jasmonates in development such as that of flowers and seeds. Due to the relatively high JA levels in plant reproductive tissues, an essential role in flower development was assumed (Creelman and Mullet, 1997; Miersch et al., 1998). Indeed, the characterization of Arabidopsis (Arabidopsis thaliana) mutants defective in JA biosynthesis or perception revealed a role of jasmonates in pollen development of Arabidopsis: The coronatine- and jasmonate-insensitive mutant coi1 (Xie et al., 1998) as well as the JA biosynthesis mutants fad3-2 fad7-2 fad8 (McConn and Browse, 1996), dad1 (Ishiguro et al., 2001), opr3 (Stintzi and Browse, 2000), and dde1 (Sanders et al., 2000) cannot produce viable pollen and are affected in stamen filament elongation. Therefore, all of them show an identical male-sterile phenotype (for review, see Wasternack and Hause, 2002; Browse, 2009b).

In tomato (Solanum lycopersicum), however, the function of JA in reproductive development differs from that in Arabidopsis. The JA-insensitive tomato jai1 mutant, homologous to coi1 in Arabidopsis, is female sterile due to a block in embryo development (Li et al., 2001, 2004). Pollen viability is also reduced in jai1, although there is enough fertile pollen for successful cross-fertilization to wild-type flowers (Li et al., 2004; B. McCaig, unpublished data). Although the degree of female sterility varies, at least one of the tomato mutants affected in JA biosynthesis also has reduced seed set. This concerns spr2, which is blocked at the plastidic ω-3-fatty acid desaturase FAD7 gene required for JA biosynthesis (Li et al., 2003), whereas acx1a, which is blocked in oxylipin β-oxidation during JA biosynthesis, is indistinguishable in seed set to wild type (Li et al., 2005). Under conditions that represent a generally low-stress environment insufficient to induce jasmonate production in wild-type plants, seed set of spr2 plants is strongly affected, whereas acx1a plants exhibit still sufficient seed set. In contrast, field-grown spr2 plants produce a higher number of seeds showing higher germination rates relative to seeds obtained from greenhouse-grown plants (Li et al., 2003, 2005). This suggests that seed set in spr2 is influenced by maternal growing conditions and that stress conditions might result in residual levels of JA or other oxylipins sufficient for female fertility (Li et al., 2003, 2005).

A putative role of jasmonates in the development of female reproductive organs is suggested by the occurrence of AOC in ovules of developing wild-type flowers and the elevated levels of jasmonates in pistils (Hause et al., 2000; Stenzel et al., 2008). AOC catalyzes a crucial step in JA biosynthesis due to its enantiomer-specific catalysis and is encoded in tomato by a single-copy gene (Ziegler et al., 2000). Its preferential occurrence in ovules is reflected by the amount of OPDA, JA, and their amino acid conjugates and methyl esters, measured in pistils, all accumulating to high levels markedly exceeding that of leaves (Hause et al., 2000). The ovary-specific accumulation of jasmonates might result in expression of JA-induced genes, such as proteinase inhibitors (PIN2; Pena-Cortes et al., 1991), Leu amino peptidases (Fowler et al., 2009), endo-β-1,4-glucanase and γ-thionins (Milligan and Gasser, 1995), and defensins (Lay et al., 2003), all of them expressed in female flower organs. This suggests a specific role of JA and/or other oxylipins in ovules of tomato flowers.

The abundant accumulation of AOC protein in ovules, the preferential formation of oxylipins in pistils, and the female-sterile phenotype of jai1 prompted us to analyze the role of AOC and oxylipins in tomato embryo development. This idea was initially supported by the fact that sterility of jai1 may result from arrest in embryo/seed maturation, because jai1 ovules have still the ability to be fertilized followed by production of small aborted seeds (Li et al., 2004). It has been discussed that this mutant seems to be affected in SlCOI1/JA-dependent supply of carbon or nitrogen by the maternal tissue thereby altering embryo development and/or generation of vitamin(s) or cofactor(s) that requires transport from maternal tissues to the developing embryo (Li et al., 2004). This might be similar to the role of biotin synthesis in Arabidopsis embryogenesis (Patton et al., 1998). Furthermore, jai flowers are defective in the formation of JA-inducible secondary metabolites such as caffeoylputrescine (Chen et al., 2006).

Seed development in plants starts after double fertilization, leading to formation of the diploid embryo and the triploid endosperm, which nourishes and protects the developing embryo (Dumas and Rogowsky, 2008). The developing embryo and the endosperm are surrounded by the maternal tissue of the ovule, the nucellus, and the integuments, which are later transformed into the seed coat. Thus, the proper development of the embryo is influenced by tissues exhibiting potentially different genomes: the diploid embryo (haploid maternal and haploid paternal genomes), triploid endosperm (diploid maternal and haploid paternal genome), and diploid maternal tissue. Moreover, during seed growth, coordination is necessary between these three types of tissues. In Arabidopsis an interplay between embryo endosperm and developing seed coat during growth was shown to be essential for seed size determination (Berger et al., 2006).

To elucidate the putative function of jasmonates in embryo development, we generated 35S::SlAOC-RNAi plants and compared their phenotype to jai1. SlAOC promoter activity and AOC protein accumulation was analyzed in developing wild-type seeds and isolated embryos. Under the growth conditions used in our experiments, the spr2 mutant was found to be delayed in embryo development, whereas the acx1a mutant exhibited normal seed formation as previously published (Li et al., 2005). Therefore, we analyzed the oxylipin profile of developing wild-type seeds and programmed cell death (PCD) of developing wild-type and spr2 seeds. The delay of embryo development in spr2 plants was accompanied by an increased PCD in the developing seed coat and endosperm. Since OPDA was the major oxylipin occurring nearly exclusively in the seed coat of developing wild-type seeds, rescue of seed formation in the JA-insensitive mutant jai1 was performed by wound-induced OPDA formation. The data suggest a role of OPDA or a related compound originating from the maternal tissues for proper embryo development.

RESULTS

35S::SlAOC-RNAi Lines Exhibit a jai1-Similar Embryogenic Phenotype

We previously generated 35S::SlAOC-sense and -antisense lines to modulate JA levels in tomato. An up-regulation and down-regulation, respectively, of the SlAOC transcript was found in leaves (Stenzel et al., 2003). The SlAOC protein level was not diminished in ovules of antisense lines compared to wild-type ovules (Miersch et al., 2004). This prompted us to generate 35S::SlAOC-RNAi lines. Among 48 independent lines about 40 lines showed reduced SlAOC transcript (Supplemental Fig. S1A) and protein (Supplemental Fig. S1B) levels in leaves accompanied by decreased levels of OPDA and JA in unwounded leaves and more pronounced in wounded leaves (Supplemental Fig. S1C). Accordingly, ovules of developing flower buds showed nearly undetectable levels of SlAOC protein by immunocytochemical analysis of the RNAi lines (Fig. 1, C versus A), which was similar to that of the sections treated with preimmune serum (Fig. 1, A′–C′). This is different from jai1 plants, which showed SlAOC protein in ovules as well (Fig. 1B), pointing to a developmental regulation of SlAOC expression. Less than 5% of hand-pollinated flowers of RNAi lines developed into fruits (Fig. 1D), but these did not contain mature seeds (not shown). Even upon pollination of 35S::SlAOC-RNAi flowers with wild-type pollen, fruits were developed, but an arrest in seed development appeared. Most traits in the flower phenotype, such as size and number of petals or sepals were not altered. The arrest in seed development, however, is reminiscent to that of the female-sterile jai1 mutant (Li et al., 2001, 2004). The limited seed set of the 35S::SIAOC-RNAi lines did not allow a detailed analysis of developing seeds. Therefore, the phenotype of jai1 was analyzed cytologically by comparing developing seeds of wild-type plants (Fig. 2, A and B) and of jai1 mutant plants, either fertilized with wild-type pollen (Fig. 2, C and D) or self-fertilized (Fig. 2, E and F). Developing wild-type seeds are characterized by a well-developed embryo enclosed by endosperm and the developing seed coat. In contrast, embryos of jai1 plants were smaller and appeared poorly developed. Most obvious, the endosperm tissue was collapsed and seemed to be separated from the developing seed coat tissues (Fig. 2, C–E). The arrested embryo development was visible in jai1 plants fertilized with jai1 pollen, but also with wild-type pollen. This led to the conclusion that embryo development in tomato might be regulated by JA or related oxylipins synthesized and/or perceived in maternal tissues.

Figure 1.

AOC protein occurs specifically in ovules of jai1 and wild-type plants, but diminished in 35S::SlAOC-RNAi lines, which are blocked in seed development. Immunocytochemical detection of SlAOC protein with an anti-SlAOC antibody in cross section of unpollinated flowers of 6-week-old plants of the wild-type MicroTom (A), jai1 (B), and a 35S::SlAOC-RNAi line (C). The corresponding controls with the preimmune serum are shown in A′ to C′. Beside a cross reactivity of the epidermal layer AOC protein is visible within ovules of wild type and jai1, but much less AOC protein was detected in ovules of the 35S::SlAOC-RNAi lines. In D the strong flower phenotype of one of the 35S::SlAOC-RNAi lines is shown. Here, most of the hand-pollinated flowers are blocked in development. The bars represent 50 μm (A–C′). [See online article for color version of this figure.]

Figure 2.

Seed development in the mutant jai1 is affected in endosperm and embryo morphology. Seeds (15 DAP) of wild type (cv MicroTom; A and B) and jai1 mutant plants (C–F), either fertilized with wild-type pollen (C and D) or self-fertilized (E and F), were processed for histological analyses as described in “Materials and Methods.” Longitudinal sections were stained with toluidine blue. Note the collapsed endosperm and the small embryo in seeds of the jai1 mutant. sc, Developing seed coat; es, endosperm; em, embryo. Bars represent 200 μm in A, C, and E, and 50 μm in B, D, and F. [See online article for color version of this figure.]

Early Stages of Embryo Development Exhibit SlAOC Promoter Activity and SlAOC Protein Accumulation in the Embryo and the Developing Seed Coat

The similarities between the reproductive phenotype of 35S::SlAOC-RNAi lines and jai1 plants led us to inspect AOC promoter activity and AOC protein accumulation, an essential enzyme in JA biosynthesis. Transgenic lines carrying a truncated SlAOC promoter upstream of the GUS reporter gene were generated and characterized elsewhere (Stenzel et al., 2008). With these lines SlAOC promoter activities tested during vegetative development and in response to environmental stimuli were identical to that detected in transgenic lines carrying the full-length promoter.

Here, developing seeds and embryos were analyzed in crosses SlAOC::GUS × SlAOC::GUS (Fig. 3A, top row for seeds, bottom row for isolated embryos) and wild type × SlAOC::GUS (Fig. 3A, middle row). GUS staining in intact developing seeds indicated strong SlAOC promoter activity from 10 d after pollination (DAP) onwards in the developing seed coat. The localized staining in embryos of seeds of wild-type plants fertilized with SlAOC::GUS pollen indicates that GUS staining originates also from filial tissue (Fig. 3B). This is supported by the fact that embryos at globular and torpedo stage, all isolated before staining, exhibited SlAOC promoter activity. Absence of SlAOC promoter activity from the endosperm could not be validated by GUS staining because of a possible diffusion of the dye from tissue exhibiting high GUS activity. In seed cross sections of different developing stages, however, immunocytochemical analysis indicated SlAOC protein in the embryo and the developing seed coat, but not in the endosperm (Fig. 3B), indicating that the SlAOC promoter activity visualized by GUS staining corresponds to AOC protein accumulation spatially and temporally. The strongest fluorescent signal indicative for occurrence of AOC protein appeared in embryos; the chalazal part of the seed showed strong staining as well (Fig. 3B). Therefore, both organs, the embryo and the developing seed coat, may have the capacity to synthesize JA.

Figure 3.

SlAOC promoter activity and SlAOC protein accumulation occur in early stages of tomato embryo development. A, High SlAOC promoter activity visualized by GUS staining appears in the globular (10 and 12 DAP) and the torpedo (18 and 21 DAP) stages of embryo development. In the top and the middle row complete seeds of self-fertilized SlAOC::GUS plants (MicroTom background) and wild type fertilized with pollen of SlAOC::GUS plants, respectively, are shown. In the bottom row isolated GUS-stained embryos of self-fertilized SlAOC::GUS plants are shown. B, Immunocytochemical detection of SlAOC protein in cross sections of seeds of the wild type (top row). The green fluorescence indicates occurrence of SlAOC protein in the embryo of the heart (15 DAP), the torpedo (18 and 21 DAP), the walking stick (24 DAP), and the curled cotyledon (27 DAP) stage, in the developing seed coat of each corresponding stage, but not in the endosperm. The bottom row shows isolated embryos of the corresponding stage. Bars represent 100 μm in A, and 250 μm in B.

A clearly detectable amount of AOC protein during any developmental stages is usually accompanied by elevated levels of JA, leading to accumulation of transcripts of JA-responsive genes, such as PIN2 (Wasternack, 2007). Transcripts of PIN2 were detectable by in situ hybridization in ovaries of flower buds, an organ unequivocally shown to contain high levels of jasmonates (Hause et al., 2000), but also in developing embryos of wild-type plants (Supplemental Fig. S2).

Embryos of the spr2 Mutant Are Delayed in Development, and spr2 Mutant Seeds Show Enhanced PCD in Developing Seed Coat and Endosperm

To validate the still correlative link between AOC occurrence, JA generation, and embryo development drawn from data obtained from wild-type plants, we analyzed the JA-deficient mutant spr2. This mutant was shown to be impaired in wound-induced JA accumulation and the production of the transmissible wound signal due to a defect in the plastid-located fatty acid desaturase that catalyzes the ω-3-desaturation of linoleic acid to the JA precursor, α-linolenic acid (α-LeA; Li et al., 2003). Consequently, wounded spr2 leaves contain only about 12% JA level from that of wounded wild-type leaves (Li et al., 2002), which is similar to that of spr2 flowers compared to the wild type and may attribute to the fertility of spr2 plants (Li et al., 2003). Here, we show, however, that the embryo development is also affected in spr2 plants. Under controlled greenhouse conditions wild-type seeds carried embryos in the curled cotyledon stage at 30 DAP (Fig. 4A), while embryo development of spr2 plants was delayed, showing embryos in the torpedo stage 30 DAP (Fig. 4D). Among 19 different spr2 plants and a total number of 436 inspected developing seeds about 70% exhibited this delay in embryo development, and most fruits did not contain mature seeds. In contrast, field-grown spr2 plants exhibited higher seed set (G. Howe, personal communication).

Figure 4.

Embryos of the spr2 mutant are blocked in development and exhibit enhanced PCD in the developing seed coat and endosperm. At 30 DAP, embryos of the wild-type Castlemart were in the curled cotyledon stage (A), whereas embryos of the spr2 mutant were arrested in development (D). PCD was analyzed during development of seeds of the wild type (B and C) and the spr2 mutant (E and F). Cross sections of the torpedo (18 and 21 DAP) and the curled cotyledon (28 DAP) stage were analyzed by the TUNEL assay (B and E) and concomitant DAPI staining (C and F). The wild-type tissues show few TUNEL-positive signals only (B). In contrast, a high number of TUNEL-positive signals appeared nearly exclusively in the seed coat tissues and in the endosperm of the spr2 mutant (E), but not in the embryo. DAPI staining of the corresponding cross sections of the wild type (C) and spr2 mutant (F) indicates the number of nuclei. sc, Developing seed coat; es, endosperm; em, embryo. Bars represent 1 mm in A and D, and 100 μm in B, D, E, and F.

To characterize seed development of spr2 plants in comparison to wild-type plants, terminal deoxynucleotidyl-transferase-mediated dUTP nick end labeling (TUNEL) analyses were performed to visualize cells undergoing PCD. Using this method, nonrandom, internucleosomal fragmentation of nuclear DNA becomes visible, which is one of the basic features of PCD and occurs as a result of specific endonuclease activation prior to condensation of nuclear chromatin (Greenberg, 1996). In positive controls generated by treatment of sections with DNase I to achieve appropriate fragmentation, all nuclei of a section showed TUNEL-positive signals (Supplemental Fig. S3). In cross sections of wild-type seeds being in preceding stages of embryo development, few positive TUNEL signals occurred nearly exclusively in the developing seed coat and the endosperm of the wild-type seeds, whereas no signal was detectable in embryo tissue (Fig. 4B). Embryo tissue of spr2 mutant also did not show label, but the developing seed coat and endosperm of spr2 mutant seeds exhibited a significantly higher number of positive TUNEL signals than the wild type (Fig. 4E). The concomitant 4,6-diamidino-phenylindole (DAPI) staining (Fig. 4, C and, F) clearly indicates that a similar number of nuclei were visible in cross sections of endosperm and seed coat tissues of both plant genotypes, thereby showing that the difference in the number of TUNEL- positive signals was not due to different numbers of nuclei analyzed. These data suggest that JA or any other oxylipin formed downstream of α-LeA is necessary for seed coat and endosperm function and might have a role to avoid PCD in endosperm tissue, thereby leading to proper embryo development.

The JA-Deficient and OPDA-Forming acx1a Mutant Exhibits Embryo Development Similar to the Wild Type

The spr2 mutant is affected in ω-3-fatty acid desaturase activity (Li et al., 2003). Consequently, other oxylipins besides JA might be also diminished, thus being responsible for the embryogenic phenotype of spr2 plants. To confine the number of putative oxylipins downstream of the ω-3-fatty acid desaturase reaction, we inspected acx1 plants. This mutant is affected in ACX1A, one of the two genes known so far that encode acyl-CoA oxidases in tomato (Li et al., 2005). ACX1A catalyzes the first step in peroxisomal fatty acid β-oxidation and is involved in JA biosynthesis. Wild-type embryos in the curled cotyledon stage (Fig. 5A) at 30 DAP were phenotypically similar to acx1a mutant embryos (Fig. 5B), thereby correlating with the previous observation that acx1a plants did not show a diminished seed production (Li et al., 2005). Inspection of cross sections of seeds at 30 DAP did not show differences between tissues of the seed coat, endosperm, and embryo of the wild type (Fig. 5C) and the acx1a mutant (Fig. 5D). These data suggest that an oxylipin formed within the lipoxygenase pathway downstream of α-LeA and upstream of JA is necessary and sufficient for proper embryo development. Seeds from 35S::AOC-RNAi lines aborted during development as described above. This raises the question whether a product generated downstream of AOC but prior to ACX1A, such as OPDA, is likely to be essential for seed development.

Figure 5.

The JA-deficient and OPDA-forming acx1a mutant exhibits an embryo development similar to the wild type. Embryos from wild-type (A) and mutant acx1a (B) were prepared from seeds at 30 DAP and showed all the curled cotyledon stage. The cross sections of wild-type (C) and mutant (D) seed stained with toluidine blue did not show differences in morphology. Bars represent 1 mm in A and B, and 100 μm in C and D.

OPDA Is the Most Abundant Oxylipin and Accumulates Nearly Exclusively in the Seed Coat Tissues

To analyze whether OPDA might play a role for proper seed development, we quantified OPDA and several jasmonates in intact developing seeds, in isolated seed coats and embryos at 27 DAP, and in developing fruits of wild-type plants (Fig. 6). Indeed, OPDA was the most prominent oxylipin detectable in seeds from 15 to 27 DAP (Fig. 6A). Dinor-OPDA, JA, 11-OH-JA, and 12-OH-JA were detected only at negligible levels. After mechanical dissection of wild-type seeds at 27 DAP into seed coat and embryo, OPDA was found preferentially in the seed coat tissues and to low amount only in the embryo (Fig. 6B). An equivalent analysis of oxylipin levels in seed coat tissues and embryos of spr2 seeds could not be done due to their arrest in embryo development. However, in the few spr2 seeds generated we measured an OPDA level in the range of about 10% of that detected in wild-type seeds (Table I). We tried to rescue the reproductive phenotype of spr2 plants by daily injection of 200 μL of a 100 μm OPDA solution into fruits between 15 and 21 DAP. A rescue appeared only occasionally to less than 10% obviously due to metabolic conversion of OPDA to JA by the enzymatic steps downstream of the ω-3-fatty acid desaturase being active in the spr2 mutant. More than 90% of the injected OPDA were found to be converted to JA and 12-hydroxy-JA. This conversion was similarly recorded upon injection of D₅-OPDA, resulting in deuterated JA and deuterated 12-hydroxy-JA. This corresponds to the strong shift into JA and 12-hydroxy-JA upon wounding of tomato leaves (Miersch et al., 2008). Nevertheless, the results shown in Figure 6 and Table I demonstrate that OPDA is the major oxylipin in developing seeds and accumulates preferentially in the seed coat tissues.

Figure 6.

Oxylipin accumulation in seeds and fruits of wild-type tomato. Seeds (A) and fruits (C) of wild-type plants were collected at the time points indicated, and seeds at 27 DAP were mechanically dissected into seed coat and embryo (B). Content of OPDA (gray), dn-OPDA (black), JA (white), 12-OH-JA (light gray), and 11-OH-JA (dark gray) were determined as described in “Material and Methods.” OPDA is the most prominent compound in developing seeds, thereby accumulating nearly exclusively in the seed coat as shown for 27 DAP (B).

Table I. Levels of OPDA, JA, and JA-Ile and transcript accumulation of SlACX1A and SlACX1B in seeds of wild type (Castlemart), spr2, and acx1a at 21 DAP.

Genotype	OPDAa	JAa	JA-Ilea	ACX1Ab	ACX1Bb
Wild type	3,298 ± 1,029	393 ± 183	183 ± 540	0.443 ± 0.097	8·× 10−5 ± 5 ×10−5
spr2	188 ± 178	29 ± 16	n.d.c	n.d.c	n.d.c
acx1a	5,224 ± 815	67 ± 41	16 ± 9	0.426 ± 0.030	20·× 10−5 ± 9 × 10−5

^aMean ± sd, given in pmol/g fresh wt, n = 6.

^bMean ± sd of relative expression levels in relation to the reference gene SlTIP4l, n = 3.

^cNot determined.

Properties of acx1a and jai1 Plants: A Role of OPDA in Embryo Development?

The delayed embryo development in spr2 seeds (Fig. 4D), and the normal embryo development in the JA-deficient and OPDA-forming acx1a mutant as well as the nearly exclusive accumulation of OPDA in the developing seed coat prompted us to analyze levels of OPDA, JA, and JA-Ile in developing seeds of wild type, acx1a, and spr2. In complete wild-type seeds (cv Castlemart) high levels of OPDA, JA, and JA-Ile were found (Table I). The corresponding amounts in acx1a seeds were similar for OPDA, and levels of JA and JA-Ile were lower than in wild type, but still measurable (Table I). These data indicate the expected accumulation of OPDA and deficiency in JA/JA-Ile in acx1a seeds. There is, however, like in acx1a leaves (Li et al., 2005), a residual amount of JA/JA-Ile that may attribute to the normal embryo development of acx1a mutant. The few spr2 seeds generated contained less than 10% OPDA and JA compared to the wild type, thus corresponding to the above-mentioned percentage for spr2 flowers (Li et al., 2003).

To test putative contribution of the two ACX1 family members, expression of ACX1A and ACX1B was recorded by quantitative reverse transcription (RT)-PCR (Table I). Both genes were expressed in developing seeds of wild type and acx1a. The relative transcript accumulations of ACX1B were, however, about 3 orders of magnitude lower than those of ACX1A in both genotypes. Consequently, a contribution of ACX1B on the observed residual JA formation in acx1a mutant is highly unlikely.

To find out whether the normal embryo development in acx1a mutant was caused by the residual amount of JA/JA-Ile or by elevated levels of OPDA, we analyzed jai1 plants. The lack of JA/JA-Ile perception in this mutant (Li et al., 2004; Yan et al., 2009; Sheard et al., 2010) should allow to test OPDA activity independently of JA/JA-Ile by injection of OPDA into fruits. Since the success of such a treatment, however, might be dependent of the right time window and position for application, we took advantage from the known wound-induced generation of OPDA and JA/JA-Ile (Fig. 7). Although genes encoding JA biosynthetic enzymes are known to be regulated by COI1/JA (Reymond et al., 2000), their developmentally regulated occurrence in flower organs of jai1 (Fig. 1; Miersch et al., 2004) allows to induce elevated OPDA and JA levels by wounding. Beginning at the onset of flowering, all leaves of jai1 plants were daily wounded by squeezing with a forceps, and subsequently mature fruits were checked for the occurrence of seeds. In nonwounded jai1 plants all fruits tested were without any seeds (n > 60, example shown in Fig. 7A). Daily repeated wounding of leaves led to generation of seeds in 20 out of 23 fruits checked. The seed number varied from one seed per fruit up to 25 seeds per fruit, resulting in a mean of 5.87 ± 5.79 seeds per fruit (n = 23, example shown in Fig. 7B). To check whether leaf wounding resulted in an increase of OPDA in fruits, OPDA levels were determined from fruits of nonwounded and repeatedly leaf-wounded jai1 plants (Fig. 7C) and compared with fruits of nonwounded wild-type plants (Fig. 6C). Young, green fruits of a size of 3- to 6-mm diameter were collected from five jai1 plants each and subjected to OPDA determination. Although not statistically significant (P = 0.06, Student’s t test), jai1 fruits of wounded plants contained more OPDA than fruits of nonwounded plants, whereas nonwounded wild-type fruits contained about 5-fold-higher OPDA level than unwounded jai1 fruits as expected (Fig. 6C versus 7C). This suggests that the reproductive jai1 phenotype was rescued by endogenously formed OPDA or an OPDA-like compound. Putatively formed JA and JA-Ile are excluded to be active compounds due to the block in JA/JA-Ile perception (Li et al., 2004).

Figure 7.

Seed set and OPDA content in fruits of jai1 plants after leaf wounding. Mature fruit of nontreated (A) and wounded (B) plants as well as OPDA levels in green fruits harvested from nontreated and wounded plants (C). Daily leaf wounding performed for 5 d after onset of flowering resulted in formation of seeds and enhanced OPDA levels in jai1 fruits. For comparison of OPDA content of nonwounded wild-type fruits see Figure 6C. In C the mean and sd of five independent determinations are shown. [See online article for color version of this figure.]

DISCUSSION

A role of jasmonates in flower development was initially shown by identification of Arabidopsis mutants affected in JA biosynthesis such as dad1, fad 3-2 fad 7-2 fad 8, allene oxide synthase, and opr3 or in JA signaling such as coi1-1 and jin1 (Browse, 2009b). Most of them are male sterile due to diminished stamen filament elongation and loss of viable pollen. In case of loss of function of the F-box protein COI1 there is a functional diversity between Arabidopsis and tomato. In contrast to the exclusive male sterility of coi1 plants of Arabidopsis (Xie et al., 1998), the jai1 plants affected in the tomato homolog of COI1 are less affected in pollen viability, but are additionally female sterile (Li et al., 2004). The female-sterile phenotype of jai1 plants might be caused by a diminished expression of JA-responsive genes, and including JA biosynthesis genes (Li et al., 2004), since latter are expressed in a COI1-dependent manner (Reymond et al., 2000; Devoto et al., 2005). This might result in two simultaneous effects, diminished biosynthesis of JAs and related oxylipins as well as the loss of JA-responsive gene products.

Wild-type tomato plants show a preferential accumulation of AOC protein in ovules and elevated levels of JA and other oxylipins in the ovule-containing pistils of flower buds (Fig. 1; Hause et al., 2000). Similarly, the AOC gene is up-regulated in pollinated flowers (Wang et al., 2009). These data and the female-sterile phenotype of jai1 plants prompted us to investigate the role of JA in embryo development. Therefore, we generated 35S::SIAOC-RNAi lines that showed an early flower abortion and an impairment in seed development (Fig. 1). Cytological analysis of seeds of 35S::SIAOC-RNAi plants could not be performed due to this strong phenotype. By analogy, however, a cytological comparison of wild-type and jai1 plants (Fig. 2) was done because the phenotype of 35S::SIAOC-RNAi plants is reminiscent to jai1 plants. Seeds of jai1 plants showed small and poorly developed embryos.

In tomato AOC is encoded by a single-copy gene (Ziegler et al., 2000) with high promoter activity in young seeds, vascular bundles, and green fruits (Stenzel et al., 2008). There is a close correlation between AOC promoter activity, AOC protein accumulation, and JA levels in different developmental stages of tomato and in response to various environmental stimuli (Hause et al., 2000; Stenzel et al., 2003, 2008). Similarly, AOC promoter activity and AOC protein accumulation closely correlated from heart stage of embryo development onwards (Fig. 3). The AOC protein accumulated in the embryo and developing seed coat, but not in the endosperm. The levels of oxylipins, however, accumulated with a peak at 21 DAP nearly exclusively in the seed coat tissues. The JA precursor OPDA was the dominant compound detectable in wild-type seeds (Fig. 6). This suggests that decreased formation of OPDA or a OPDA-like compound in the 35S::SIAOC-RNAi plants attributes to the observed embryogenic phenotype in these transgenic lines. At least for wounded leaves strongly decreased levels of JA and OPDA in the transgenic lines could be shown (Supplemental Fig. S1). Obtaining similar data for the seeds of the 35S::SIAOC-RNAi plants was impossible due to the low seed set. Therefore, we inspected spr2 plants that are affected in the formation of α-LeA. Under controlled conditions preventing environmental stress, spr2 plants showed a delay in embryo development thus supporting the assumption that any compound generated downstream of α-LeA is necessary for proper embryo development. The residual amount of JA detected in spr2 flowers (18% of the wild type) was assumed to be sufficient to promote fertility (Li et al., 2003). OPDA levels were not checked. Since spr2 plants are affected in the plastid ω-3-fatty acid desaturase, they should have JA and OPDA deficiency. Indeed, we could detect much less amounts of both compounds in the few spr2 seeds generated. Consequently, the observed delay in development of spr2 embryos might be caused by deficiency in JA and/or OPDA. The delay led to impaired seed development, resulting in about 30% developed seeds in comparison to wild type. This reduced seed development of spr2 plants appeared under greenhouse conditions used here. In field-grown spr2 plants, however, where stress-induced generation of oxylipins may occur, a significantly higher seed set was observed (G. Howe, personal communication). This is consistent with the assumption that for seed development a threshold of JA and/or OPDA is necessary. Regulation of plant development by a threshold of signaling compounds is a common phenomenon (Boss et al., 2010).

Taking into account the 35S::SIAOC-RNAi plants, compounds essential for embryo development could be narrowed down to any compound downstream of the AOC-catalyzed step. The normal seed development, however, in acx1a plants, which showed elevated levels of OPDA but diminished JA levels, suggests that OPDA and/or its metabolites are active in normal embryo development. The acx1a embryos were similar under our growth conditions to the wild type in all developmental stages as exemplified and shown for the curled cotyledon stage (Fig. 5). Cross sections of seeds in the curled cotyledon stage showed no differences in morphology of seed coat, endosperm, and embryo between wild type and acx1a. An interpretation, however, of the acx1a embryo phenotype in terms of preferential role of OPDA, is hampered by the fact that acx1a seeds showed a residual level of JA/JA-Ile that corresponds to data for leaves (Li et al., 2005). It is unlikely that the second ACX1 gene, ACX1B, attributes to this residual amount due to (1) its very low expression compared to that of ACX1A in the acx1a mutant and (2) the preference of ACX1A for C18 cyclopentanoid-CoAs that are the required JA precursors (Li et al., 2005). It cannot be excluded so far that additional ACX genes homologous to AtACX1, AtACX3, and AtACX4 and detected in the tomato EST database may code for enzymes active in JA biosynthesis (Li et al., 2005).

Like the spr2 plants, which show residual amount of JA/JA-Ile and partial fertility (Li et al., 2003), residual amount of JA/JA-Ile in acx1a plants may account for the observed proper embryo development. The following observations, however, make an exclusive activity of JA/JA-Ile in embryo development highly improbable: (1) the defect in embryo development of jai1 could be rescued by wound-induced endogenous formation of OPDA, and (2) although both spr2 and acx1a plants have residual levels of JA/JA-Ile, only acx1a plants exhibiting elevated OPDA levels showed normal embryo development. Obviously, acx1a plants have the required threshold of OPDA. In contrast, jai1 plants do not seem to have a sufficient amount of OPDA without additional stress treatment. In nonstressed jai1 plants, missing JA perception may lead to diminished OPDA levels, since the well-known positive feedback of JA/JA-Ile on oxylipin biosynthesis cannot be active (Wasternack, 2007). Only a stress treatment, such as repeatedly performed wounding, causes an OPDA accumulation sufficient for seed formation in jai1 fruits. To date, there is no hint how OPDA is perceived, but its perception via the JA receptor/coreceptor complex can be clearly excluded (Thines et al., 2007; Katsir et al., 2008; Fonseca et al., 2009; Yan et al., 2009; Sheard et al., 2010).

It is worthwhile to mention, although not directly comparable, that in the moss Physcomitrella patens, which is able to form OPDA but not JA, the aoc knockout plants of P. patens exhibit reduced fertility (Stumpe et al., 2010). Except the membrane-bound OPDA occurring exclusively in Arabidopsis species and forming the so-called Arabidopsides, there is no report so far on OPDA metabolites, such as amino acid conjugates. A recent inspection of P. patens (Stumpe et al., 2010) and some higher plant species (R. Kramell, personal communication) failed to detect such OPDA derivatives.

The correlative data given here for OPDA raise the question, which processes might be affected by OPDA in the well-documented communication between the maternal tissues, such as integument and nucellus, and the filial tissues, such as endosperm and embryo. This repeatedly described communication is necessary for proper seed development and involves auxin and ethylene as preferential signals (Balbi and Lomax, 2003; Wang et al., 2009). It might be, however, negatively affected by insufficient levels of OPDA. This assumption is supported by the fact that developing spr2 seeds exhibited increased PCD in the developing seed coat and endosperm (Fig. 4). Since seed coat tissues and endosperm of wild type revealed only few cells undergoing PCD, enhanced PCD in spr2 seeds may lead to or go along with delayed seed development. Although not checked by the TUNEL assay, the clearly aborted/shrunken endosperm in seeds of nontreated jai1 plants (Fig. 2) points to an enhanced PCD in the nonstressed mutant. This fits into the hypothesis that seed-coat-generated OPDA as well as OPDA generated by wounding in jai1 may attribute to the known up-regulation of stress-responsive genes such as those encoding heat shock proteins or glutathione-S-transferases, all involved in detoxification and specifically up-regulated by OPDA (Taki et al., 2005; Mueller et al., 2008). In light of these data, seed-coat-generated OPDA may attribute to detoxification of a tissue necessary for nutritional supply of the developing embryo.

Another scenario might result from the loss of defense gene expression of OPDA- and JA-induced genes, all affected in jai1, 35S::SIAOC-RNAi, and spr2. In developing flowers and seeds of tomato, gene expression of JA-inducible defense proteins such as PIN2 and DEFENSIN2 is up-regulated (Li et al., 2004; Stotz et al., 2009). Wild-type flowers accumulate more than 200 μg/mL soluble PIN2 protein, which is the most abundant JA-inducible protein of tomato. In jai1 flowers, however, PIN2 is undetectable (Li et al., 2001, 2004). A scenario of JA activity was suggested for flower development before anthesis: An increased JA level upon developmental AOC up-regulation in ovules (Hause et al., 2000) might be followed by PIN2 expression as detected by in situ hybridization (Supplemental Fig. S2), which in turn may lead to the decreased herbivore infestation in flowers (Damle et al., 2005). A similar scenario, which may occur in seed development, could be an OPDA-induced DEFENSIN formation known to affect seed production and embryo development (Stotz et al., 2009). Such a dual function of defense proteins in defense and development was repeatedly observed. It is tempting to speculate that seed-coat-generated OPDA might be active by inducing DEFENSIN formation in the endosperm.

Another picture could be an OPDA-dependent alteration of carbon availability during seed formation. In tomato a JA-inducible cell wall invertase (LIN5) is preferentially expressed in gynoecia and was suggested to function in carbohydrate supply of this flower organ (Godt and Roitsch, 1997). Indeed, RNAi lines significantly silenced in LIN5 expression showed decreased seed size, seed production per fruit and fruit yield, and decreased content of hexoses (Zanor et al., 2009). Interestingly, in LIN5-silenced lines the content of oxylipins measured as JA content was decreased down to about 20% of the wild type. Thus, it is tempting to speculate that JA biosynthesis is required spatially and temporally for proper generation of sugars that in turn are required for development of tomato embryos/fruits. Another essential factor might be proper amino acid levels that are remarkably decreased in transgenic JA-deficient tomato fruits (Kausch et al., 2012).

In tomato flowers the LIN5-dependent generation of hexoses (Zanor et al., 2009) is regulated posttranslationally by a LIN5-specific inhibitor protein INVINH1, which gene is coexpressed with LIN5 within the gynoecia (Jin et al., 2009). Silencing of INVINH1 allowed extra invertase activity that led to more hexoses being available for developing seeds and was followed by an increase in seed weight and size (Jin et al., 2009). Interestingly, such invertase inhibitors are down-regulated by OPDA (Taki et al., 2005). Consequently, a high level of OPDA as shown here for the developing seed coat may down-regulate inhibitor expression, thereby allowing LIN5 to generate hexoses for proper embryo development, whereas spr2 seeds are unable to repress inhibitor expression due to OPDA deficiency.

The proposed role of OPDA in embryo development described here is an additional example similar to OPDA-specific, but JA-independent gene expression (Taki et al., 2005; Mueller et al., 2008; Ribot et al., 2008; Böttcher and Pollmann, 2009), protein pattern (Dueckershoff et al., 2008), and other processes mediated preferentially by OPDA such as fertility of P. patens (Stumpe et al., 2010), phytochrome A signaling and shade avoidance (Robson et al., 2010), or hypocotyl growth inhibition at reduced V-ATPase activity (Brüx et al., 2008). More recently, genetic and biochemical evidence showed that inhibition of seed germination well known for a long time as a characteristic response upon JA/JA-Ile treatment is not caused by the applied compounds but is mediated by endogenous accumulation of OPDA during seed development in a COI1-independent manner (Dave et al., 2011). Such JA-independent changes are further supported by a recent proteome analysis of wounded Arabidopsis leaves. Here, about 95% of the wound-stimulated changes occurred in the absence of JA and included preferentially redox-controlled enzymes in glutathione synthesis and transporters (Gfeller et al., 2011). The increasing number of examples on JA-independent OPDA-specific responses addresses the question for future work on OPDA perception and OPDA signaling.

MATERIALS AND METHODS

Plant Material, Growth Conditions, and Treatments

Tomato (Solanum lycopersicon) cv MicroTom, cv Lukullus, cv Castlemart, the mutants jai1 (in cv MicroTom), spr2, and acx1a (in cv Castlemart), and transgenic lines of cv MicroTom carrying a 35S::SlAOC-RNAi construct (see below) or an SlAOC::GUS construct (Stenzel et al., 2008) were grown in greenhouse as described (Wasternack et al., 1998). The flowers were hand pollinated and seeds were prepared from the emerging fruits at different DAP. Wounding of jai1 plants was performed by squeezing of fully developed leaves with tweezers every day in a time period of 5 d after onset of flowering.

Quantitative RT-PCR

Total RNA was prepared from 100 mg of isolated seeds (21 DAP) using the Plant RNeasy extraction kit (Qiagen, www.qiagen.com) with additional DNase digestion step (RNase-free DNase set; Qiagen). For real-time RT-PCR analyses, 1 μg of total RNA was converted into cDNA with M-MLV reverse transcriptase, RNase H minus, point mutant (Promega, www.promega.com) according to the manufacturer’s protocol using oligo(dT)19 primer. cDNA was diluted to fixed quantities (15 ng per reaction of reverse-transcribed total RNA).

For semiquantitative detection of SlACX1A and SlACX1B transcripts a SYBR green-based PCR assay (Applied Biosystems, www.appliedbiosystems.com) was performed. The following primers and annealing temperatures were used: ACX1A forward 5′-CAA ATG CTG TAT CAC TGG TTG ATG-3′, ACX1A reverse 5′-CCA TCA TAA CGT CCA AGA ATT GAA-3′, 60°C; ACX1B forward 5′-GGG CCT TGT ATA CTC AGG TTT G-3′, ACX1B reverse 5′-TTT CGG GTA CAC AAA TGA ACA GCA-3′, 60°C. To normalize transcript levels for differences in the efficiency of cDNA synthesis, transcript levels of the constitutively expressed SlTIP4l were measured using the following primers and temperature: TIP4l forward 5′-ATG GAG TTT TTG AGT CTT CTG C-3′, TIP4l reverse 5′-GCT GCG TTT CTG GCT TAG G-3′, 60°C (Expósito-Rodríguez et al., 2008). The reactions were performed in a Mx 3000P QPCR system (Stratagene, www.stratagene.com) with the following protocol: 95°C for 10 min, followed by 40 cycles of 95°C for 30 s and 60°C for 1 min, and 95°C for 1 min, and a subsequent standard dissociation protocol. As control for genomic DNA contamination, 15 ng of total nontranscribed RNA was used under the same conditions as described above. All assays were performed from three biological replicates and in three technical replicates each. Relative expression levels were calculated by the comparative Ct method including normalization to the constitutively expressed gene.

Extraction and Quantitative Analysis of Oxylipins

Seeds of at least five different fruits of at least three different plants were collected at different DAP. Seeds from 27 DAP were used for mechanical separation of seed coat tissues (integument and nucellus) and embryos using a stereo microscope. All materials (about 50–500 mg fresh weight) were immediately frozen in liquid nitrogen. Extraction, purification, and quantification of OPDA, dinor-OPDA, JA, 11-OH-JA, and 12-OH-JA were performed as described (Stenzel et al., 2003; Miersch et al., 2008).

Generation and Characterization of 35S::SlAOC-RNAi Lines

For the SlAOC-repression constructs, an inverted-repeat construct was created using a BamHI/XhoI fragment (423 bp) from the 3′ end of the SlAOC-cDNA. The inverted fragments were separated by a spacer containing 1,079 bp of the GUS gene. The RNAi construct was inserted into pBI101.1 (Clonetech, www.clonetech.com) between the cauliflower mosaic virus 35S promoter and the NOS terminator, checked by sequencing, and transformed into Agrobacterium tumefaciens strain GV3101. Transgenic tomato cv MicroTom were generated by Agrobacterium-mediated gene transfer essentially as described for generation of SlAOC::GUS lines (Stenzel et al., 2008). Analyses of integration numbers by Southern blot, level of SlAOC mRNA and protein were performed using primary transformants and T1 lines as described previously (Stenzel et al., 2003, 2008).

Histochemical Analyses

GUS activity analyses in intact seeds and isolated embryos were performed according to Stenzel et al. (2008). For analysis of seed morphology, developing seeds of wild-type (cv MicroTom) and jai1 plants were fixed in 3% (v/v) sodium-cacodylate-buffered glutaraldehyde (pH 7.2), postfixed with 1% (w/v) OsO4 solution, dehydrated in an ethanol series, and embedded in epoxy resin (Spurr, 1969). Semithin sections (1-μm thickness) were stained with toluidine blue. For immunocytochemical analysis of SlAOC, in situ hybridization, and TUNEL assays, tissues were fixed with 4% (w/v) paraformaldehyde/0.1% (v/v) Triton X-100 in phosphate-buffered saline (PBS) immediately after harvest and embedded in polyethylene glycol 1500 as described (Isayenkov et al., 2005).

Immunodecoration and visualization of AOC protein was performed on 2-μm-thick sections with an antibody raised against recombinant SlAOC followed by an anti-rabbit IgG antibody either conjugated with AlexaFluor 488 (Invitrogen, www.invitrogen.com) or conjugated with alkaline phosphatase (Chemicon, www.chemicon.com) as described (Hause et al., 2000).

In situ hybridization at 5-μm-thick sections was carried out as outlined previously (Tretner et al., 2008) using digoxigenin-labeled antisense and sense RNA probes obtained from PIN2 cDNA by in vitro transcription.

For TUNEL assay, 2-μm-thick sections were pretreated with 1 μg mL⁻¹ proteinase K in 10 mm Tris-HCl (pH 8.0) at 37°C for 10 min. After washing with PBS, TUNEL reaction was carried out at 37°C for 60 min using the Fluorescein FragEL DNA fragmentation detection kit (Calbiochem, www.merck-chemicals.com). Positive controls were generated by pretreatment of sections with 1 μg mL⁻¹ DNase I in PBS at 37°C for 15 min. After washing, sections were stained with 1 μg mL⁻¹ DAPI (Sigma-Aldrich, www.sigmaaldrich.com) in PBS.

All sections were observed with an AxioImager (Zeiss, www.zeiss.de), pictures were taken with an Axiocam (Zeiss), and processed through Photoshop 8.0.1 (Adobe, www.adobe.com).

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers AY817109.1 (ACX1A), AY817110.1 (ACX1B), X94946.1 (PIN2), and SGN-U321250 (TIP41).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. SlAOC mRNA and protein as well as OPDA and JA levels are decreased in wounded leaves of 35S::SlAOC-RNAi lines compared to the wild-type MicroTom.

• Supplemental Figure S2. The JA-responsive PIN2 gene is preferentially expressed in ovules of unpollinated flowers and in developing embryos.

• Supplemental Figure S3. Positive controls for TUNEL analysis in seeds of the wild-type Castlemart and the spr2 mutant.