INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Fruit development can be divided usually into three phases (Gillaspy et al., 1993). The earliest phase (up to around anthesis) involves the development of the ovary and the decision to abort or to continue with growth. It is currently accepted that the coordinated action of growth signals triggers fruit-set and growth, so in the absence of pollination and fertilization, the ovary will cease cell division and abscise (Gillaspy et al., 1993; García-Martínez and Hedden, 1997). Parthenocarpy is an alternative pathway to normal fruit-set and development, in which the ovary grows into fruit without fertilization and seed formation (Lukyanenko, 1991), and can be artificially induced by the application of hormones, mainly auxins and gibberellins (GAs) (Goodwin, 1978; Schwabe and Mills, 1981). In the case of natural parthenocarpy, it has been proposed to be the result of conditions that induce a threshold concentration of growth substances in the ovary sufficient to promote growth in the absence of pollination and fertilization (Nitsch, 1970).

There is some evidence supporting the hypothesis that GAs may control, at least partially, fruit development in seeded tomato (Lycopersicon esculentum Mill.). First, fruit growth can be induced by the application of GAs to unpollinated ovaries (Schwabe and Mills, 1981; Sawhney, 1984). Second, fruits containing GA-producing seeds are larger than those with GA-deficient seeds (Groot et al., 1987). Third, GAs of the early-13-hydroxylation (GA₁, GA₈, GA₁₇, GA₁₉, GA₂₀, GA₂₉, and GA₄₄) and of the non-13-hydroxylation (GA₉, GA₁₅, GA₂₄, and GA₂₅) biosynthetic pathways have been identified in developing pollinated fruits (Bohner et al., 1988; Koshioka et al., 1994). These GAs may be synthesized in the ovary itself, because genes coding for GA biosynthesis enzymes (copalyl diphosphate synthase, GA 20-oxidase, and GA 3-hydroxylase) are expressed in developing tomato flowers and fruit (Rebers et al., 1999). Parthenocarpy in tomato may also depend on GAs, because natural parthenocarpic fruits contain more GA-like substances than their non-parthenocarpic counterparts early after anthesis (Mapelli et al., 1978), and tomato fruits induced by the application of 4-chlorophenoxyacetic acid (an auxin) also contain high levels of GAs (Koshioka et al., 1994).

Natural parthenocarpy in tomato has been widely studied due to its potential use as a solution to poor fruit-set at unfavorable environmental conditions (George et al., 1984). Several genotypes carrying gene(s) inducing parthenocarpy have been described and selected in tomato (Philouze, 1983; George et al., 1984; Lukyanenko, 1991; Mazzucato et al., 1998). Among the different sources of natural parthenocarpy, the Russian cv Severianin obtained by Solovjeva (cited by Philouze, 1983) is of particular interest because of its strong expressivity, its facultative character, and its simple genetic control (Philouze et al., 1980). The capability of cv Severianin to set seedless fruits with complete locule fill under unfavorable environmental conditions is mainly due to the pat-2 gene (Philouze and Maisonneuve, 1978; Nuez et al., 1986; Vardy et al., 1989), which induces a different protein in the ovary after anthesis (Barg et al., 1990). The analysis by two-dimensional PAGE of in vitro translation products of RNAs from flowers and ovaries before anthesis also shows a differential expression associated to pat-2-induced parthenocarpy (Fos and Nuez, 1996; Fos and Nuez, 1997). This suggests that the molecular events observed in pat-2 ovaries before anthesis, which are associated with parthenocarpic fruit growth, may modify the hormone content of the ovary before pollination.

In the present study, the possible role of GAs in parthenocarpic tomato fruit development controlled by pat-2 has been investigated at stages before fruit-set using near-isogenic lines. The inhibition of fruit-set by paclobutrazol and its reversal by GA₃ suggested that fruit growth of both seeded wild-type (wt) and parthenocarpic pat-2 tomato depends on GAs. The quantification of endogenous GA levels in developing tomato ovaries showed the accumulation of very large amounts of GA₂₀ in pat-2 ovaries before anthesis. However, GA₂₀ may not be active per se, because the application of an inhibitor of 2-oxoglutarate-dependent dioxygenases decreased parthenocarpic fruit development. These results suggest that the parthenocarpic capability of pat-2 ovaries may be the result of the accumulation of GA₂₀, leading to an early higher synthesis of active GA in the absence of pollination and fertilization.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Plant Material

Two non-parthenocarpic tomato (Lycopersicon esculentum Mill.) lines, Madrigal (MA/wt) and Cuarenteno (CU/wt), and their corresponding near-isogenic parthenocarpic lines carrying the pat-2 gene, Par54-11 (MA/pat-2) and Par14-11 (CU/pat-2), were used in the experiments. The parthenocarpic lines were generated in our laboratory from crosses of MA/wt and CU/wt with the Russian cv Severianin (source of pat-2) (Nuez et al., 1985). Line MA/pat-2 has been backcrossed to the non-parthenocarpic line MA/wt four times, and line CU/pat-2 is an F₄ selection that was then backcrossed three times with the non-parthenocarpic CU/wt line.

The plants were grown during autumn-winter in an air-conditioned greenhouse set at an average temperature of 10°C in large containers (6 m × 40 cm × 50 cm depth; 18 plants per container) with sand (experiment of Fig. 1), or in 25-L pots containing a peat:soil (1:1) mixture irrigated with nutrient solution. The temperatures fluctuated according to the environment, but extremes were never higher than 21°C (day) or lower than 6°C (night).

View larger version (22K): [in this window] [in a new window]   Figure 1.   Development of MA/wt, MA/pat-2, CU/wt, and CU/pat-2 fruits. The points are means ± SE of eight to 12 developing fruits. , Self pollinated flowers; , flowers emasculated and unpollinated; , ovaries from emasculated flowers treated with GA3 (2 µg ovary1).

Ovaries from flowers at the flower bud (FB), preanthesis (PR), and anthesis (AN) stage, as previously described (Fos and Nuez, 1996), were collected from the second to sixth clusters for the determination of GAs (Table I). The ovaries were frozen immediately in liquid N₂ and stored at 80°C until extraction.

Plant Hormone Application

Flowers from the second to fourth clusters were emasculated 1 d before anthesis to prevent self-pollination, and 2 to 3 d later the ovaries were treated with different amounts of GA₁ (Dr. Michel Beale, Long Ashton Research Station, Long Ashton, UK), GA₃ (Fluka, Sigma-Aldrich Química SA, Alcobendas, Spain), or GA₂₀ (purchased from Dr. L. Mander, Australian National University, Canberra) at the concentrations specified in each experiment (in 20 µL of 5% [v/v] methanol containing 0.1% [v/v] Tween 80, applied directly to the ovary). Paclobutrazol (Imperial Chemical Industries, Bracknell, UK; 5 µg per ovary), an inhibitor of ent-kaurene oxidase, and 3,5-dioxo-4-butyryl-cyclohexane carboxylic acid ethyl ester (LAB 198 999 [LAB], BASF, Limburgerhof, Germany; at 1, 5, and 10 mM), an acylcyclohexadione derivative that inhibits 2-oxoglutarate-dependent dioxygenases (Rademacher et al., 1992), were applied to pollinated MA/wt and unpollinated MA/pat-2 ovaries alone or together with GAs. Control ovaries were treated with the same volume of solvent solution. Each treatment was carried out on nine to 12 ovaries (three to four plants, one cluster per plant, and three ovaries per cluster).

Extraction and Purification of GAs

Frozen material, previously ground in a mortar with liquid N₂, was homogenized with a polytron homogenizer (PT-10, Kinematica, Kriens-Luzern, Switzerland) in cold 80% (v/v) methanol (1:25, w/v), stirred for 12 h at 4°C, and re-extracted twice for 30 min with one-half volume of methanol. To determine recoveries during extraction and purification and to allow GAs to be located during HPLC, the following radioactive GAs were added to the homogenized samples: [³H₂]GA₁ (1.39 TBq mmol¹), 16,17-dihydro[³H₂]GA₁₉ (2.45 TBq mmol¹), and [³H₂]GA₂₀ (1.41 TBq mmol¹). Methanol was removed under vacuum at 40°C, and the aqueous residue purified by solvent partitioning QAE-Sephadex A-25 (Pharmacia, Milton Keynes, UK) chromatography and C₁₈-BondElut (500 mg; Varian, Harbor City, CA), as described by García-Martínez et al. (1991).

Reverse-Phase HPLC

Dried samples were dissolved in 10% (v/v) methanol (0.4 mL), injected onto a 4-µm C₁₈ column (15 cm long, 3.9 mm i.d.; NovaPak, Millipore, Milford, MA), and eluted with a linear gradient of 10% to 100% (v/v) methanol containing 50 µL L¹ (v/v) acetic acid over 40 min at 1 mL min¹. Fractions were grouped according to retention times of radioactive GA markers and taken to dryness in vacuo. Retention times of standards were 17 to 18 min for GA₁, 24 to 25 min for GA₂₀, and 26 to 28 min for dihydro-GA₁₉.

GA Identification

GAs were identified in extracts of non-parthenocarpic MA ovaries, collected from developing flowers before pollination (from FB to AN stage). A total mixed population of 365 ovaries (3,535 mg fresh weight [FW]) were extracted and purified as described above. Dried samples from pooled HPLC fractions were dissolved in 0.5 mL of methanol and methylated with ethereal diazomethane for 15 min at room temperature. They were then dried in vacuo and trimethylsilylated with N-methyl-n-trimethylsilyltrifluoroacetamide (Pierce, Rockford, IL) for 30 min at 90°C. GA₂₀ was also identified in extracts from a pool of MA/pat-2 and CU/pat-2 ovaries (267 and 268, respectively) collected from flowers from FB to AN.

The analyses were carried out with a gas chromatograph (model 5890, Hewlett-Packard, Palo Alto, CA) coupled to a mass-selective detector (model 5971A, Hewlett-Packard) using an HP-1 capillary column (25 m long, 0.2-mm i.d., 0.33-µm film thickness), under the conditions described by Hedden et al. (1989). The samples (1 µL) were co-injected with 0.1 µL of Parafilm extract in n-hexane to determine the Kovats Retention Index (KRI) values (Gaskin et al., 1971). The initial oven temperature was maintained at 60°C for 1 min and then increased at 20°C min¹ to 240°C and at 4°C min¹ to 300°C. The MS ion source was operated at 70 eV and 175°C to 185°C, the injector temperature at 220°C, and the interface temperature at 280°C. The samples were run using the selected ion monitoring (SIM) mode.

GA Quantification

The samples were extracted and purified as described above. The number and FW of ovaries collected for GA quantification are shown in Table I, and aliquots of approximately 300 mg FW were used for quantification. For these analyses, the appropriate amounts of the following internal standards were added to the samples at the time of extraction: [17-²H₂]GA₁, [17-²H₂]GA₃, [17-²H₂]GA₈, [17-²H₂]GA₁₉, [17-²H₂]GA₂₀, and [17-²H₂]GA₂₉ (purchased from Prof. L. Mander, Australian National University, Canberra).

The chromatographic and spectrometric conditions described above were also used in quantitative analysis. The concentrations of GAs in the original extracts were determined from the peak area ratios of the following pairs of ions: 506/508 for GA₁, 504/506 for GA₃, 594/596 for GA₈, 434/436 for GA₁₉, 418/420 for GA₂₀, and 506/508 for GA₂₉ by references to calibration curves. Additional ions were also monitored to confirm the identify of the GAs quantified.

Statistical Methods

Statistical treatments of the data were made by analysis of variance using the Fisher's LSD procedure to discriminate among the means (Statgraphics Plus program, version 3.1 for Windows, Statistical Graphics, Rockville, MD).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Growth of Non-Parthenocarpic and Parthenocarpic Ovaries

Unpollinated MA/wt and CU/wt ovaries grew very little compared with pollinated ovaries, and abscised about 10 d after anthesis (Fig. 1). In contrast, unpollinated MA/pat-2 and CU/pat-2 ovaries developed similarly to pollinated wt ovaries, leading to mature fruits essentially identical to wt seeded fruits except for the absence of seeds (Fig. 1). It is already known that the final weight of seedless and seeded pat-2 fruits in both the MA and CU backgrounds are similar (Nuez et al., 1991).

The application of GA₃ (2 µg) induced set of unpollinated MA/wt and CU/wt ovaries, and the final size and weight of parthenocarpic wt fruits induced by GA₃ were similar to those developed from pollinated ovaries in CU plants, but significantly smaller in MA plants (Fig. 1). The application of GA₃ did not alter the growth of unpollinated MA/pat-2 and CU/pat-2 ovaries (Fig. 1).

Inhibition of Tomato Fruit Growth with Paclobutrazol

The means of the final diameters and weights of pollinated MA/wt and parthenocarpic MA/pat-2 fruits were not significantly different (Table II). However, the application of paclobutrazol (5 µg) prevented the growth of both pollinated MA/wt (all the ovaries abscised) and unpollinated MA/pat-2 (only one ovary developed, although significantly less than the control) ovaries and the application of GA₃ (2 µg) fully reverted this inhibition (Table II).

Identification of GAs in Tomato Ovaries before Pollination

The identification of GAs in an extract of developing tomato ovaries (a mixed population from FB to AN) of the non-parthenocarpic line MA/wt was carried out by comparing the relative abundance of at least five characteristic ions and KRI values with those of authentic standards. The following GAs of the early-13-hydroxylation biosynthetic pathway could be identified: GA₁, GA₈, GA₁₉, GA₂₀, GA₂₉, GA₄₄, and GA₅₃. GA₃, as well as two members of the non-13-hydroxylation pathway, GA₉ and GA₃₄, were also identified in young ovaries. In addition, a compound with the same KRI as GA₈₁, a 2-hydroxy derivative of GA₂₀, was also tentatively identified for the first time in tomato (Table III).

Quantification of GAs from Tomato Ovaries

The content (nanograms per gram FW) of GA₁, GA₈, GA₁₉, GA₂₀, and GA₂₉ in ovaries from flowers at three developmental stages (FB, PR, and AN) of the non-parthenocarpic tomato lines MA/wt and CU/wt and their respective near-isogenic parthenocarpic (pat-2) lines were determined by GC-SIM (Fig. 2). GA₃ was also quantitated but it was always at very low level or undetected (results not presented).

View larger version (39K): [in this window] [in a new window]   Figure 2.   Concentration of GA1, GA8, GA19, GA20, and GA29 in unpollinated ovaries of non-parthenocarpic (MA/wt and CU/wt) and parthenocarpic (MA/pat-2 and CU/pat-2) near-isogenic lines of tomato at three developmental stages. FB, Flower bud; PR, preanthesis; AN, anthesis. White bars, wt; shaded bars, pat-2.

In non-parthenocarpic ovaries, the highest GA concentrations were found at the FB stage: about 40 to 50 ng g¹ FW for GA₁₉, 10 to 15 ng g  FW for GA₂₀, 4 to 6 ng g¹ FW for GA₂₉, 2 to 4 ng g¹ FW for GA₁, and 7 to 17 ng g¹ FW for GA₈. The concentrations of all GAs decreased progressively during ovary development in the two non-parthenocarpic lines. In the case of GA₁₉, its content was higher in non-parthenocarpic than in pat-2 ovaries (three to five times higher with the MA background, and two to four times higher with the CU background), except at the FB stage with the MA background, which were very similar (Fig. 2). In contrast, the level of GA₂₀, the immediate product of GA₁₉ metabolism was clearly higher in pat-2 than in wt ovaries at the three stages of flower development analyzed: between 10 and 160 times with the MA background, and between five and 50 times higher with the CU background (Fig. 2). The concentration of GA₂₀ was particularly high and constant (about 100 ng g¹ FW) in pat-2 ovaries with MA background. The content of GA₁ was higher in pat-2 ovaries with MA (about 2- to 4-fold), but not with CU background (where similar or higher GA₁ levels were found in non-parthenocarpic ovaries). The content of GA₈ was lower in pat-2 than in non-parthenocarpic ovaries in MA background at the FB stage (about 3-fold), and similar at later developmental stages (Fig. 2), whereas with the CU background it was higher (about 2-fold at PR and AN stages) in pat-2 ovaries. The differences in the contents of GA₂₉ between non-parthenocarpic and pat-2 ovaries were either not consistent (MA background) or not significant (CU background, except at PR stage, where the pat-2 ovaries contained about three times less GA₂₉).

To check that the m/z ion 418, used to quantitate GA₂₀, was not contaminated in pat-2 ovaries with a similar ion from another compound with the same retention time as GA₂₀, extracts from pools of MA/pat-2 and CU/pat-2 ovaries (from FB to AN flowers) were analyzed by GC-SIM measuring the abundance of the same ions used to identify GA₂₀ in MA/wt ovaries (Table II). The results showed clearly the presence of uncontaminated GA₂₀ in both kinds of pat-2 ovaries (Table IV).

Effect of GA₁, GA₂₀, and LAB on the Growth of Unpollinated Tomato Ovaries

The application of GA₁ (0.02, 0.2, and 2 µg) induced parthenocarpic growth of unpollinated MA/wt ovaries in a dose-responsive manner, and 2 µg of GA₂₀ had an effect similar to the same amount of GA₁ (Table V). Lower amounts of GA₂₀ (0.2 and 0.02 µg) were not active. The effect of GA₂₀ on parthenocarpic growth of MA/wt ovaries was not affected by the simultaneous application of 1 mM LAB, but it was significantly reduced by 5 mM LAB (Table V), suggesting that GA₂₀ needs to be metabolized to be active. A higher concentration of the inhibitor (10 mM) did not elicit further reduction of fruit-set and growth.

In MA/pat-2 ovaries, the application of GA₁ or GA₂₀ (2 µg) did not affect the growth of parthenocarpic fruits developed in the absence of pollination (Table VI, experiment I), as was observed previously with GA₃ (Fig. 1). The smaller final size of the fruits in this experiment may be due to the use of 25-L pots with a peat/soil mixture rather than large containers with sand. LAB (1 mM) did not have any effect when applied together with 2 µg GA₂₀, as occurred in MA/wt, whereas 5 mM LAB also significantly reduced fruit-set and growth (Table VI, experiment I). In a separate experiment, the application of 5 mM LAB alone also inhibited significantly fruit-set and growth, and this inhibition was fully reverted by the simultaneous application of 2 µg of GA₁ (Table VI, experiment II).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The application of GA₃ induced fruit-set and development of unpollinated ovaries in the non-parthenocarpic tomato MA/wt and CU/wt lines (Fig. 1), which is in agreement with previous results in different tomato cultivars (Bünger-Kibler and Bangerth, 1982; Sjut and Bangerth, 1982; Alabadí et al., 1996). GA₁ and GA₂₀ were also active in inducing parthenocarpic fruit development, at least in MA/wt (Table V). The application of paclobutrazol, an inhibitor of GA biosynthesis, negated the development of seeded MA/wt fruits, and this effect was fully counteracted by GA₃ (Table II). These results suggest that GAs may be involved in tomato fruit development. The final size of GA₃-induced fruits of the MA/wt tomato line were smaller than seeded fruits, but those of the CU/wt tomato line were not. This indicates that, depending on the genetic background, more than a single application of GA₃ may be necessary to mimic the effect of pollination and fertilization. Alternatively, other growth factors (e.g. auxins; Gillaspy et al., 1993) in addition to GAs may also be requested for the normal fruit growth of tomato. Unpollinated ovaries of the MA/pat-2 and CU/pat-2 lines developed similarly to pollinated ovaries of their near-isogenic wt lines, and their growth was not altered by GA₃ treatment (Fig. 1). The application of paclobutrazol prevented the growth of seedless parthenocarpic MA/pat-2 ovaries, and this effect was negated by the application of GA₃ (Table II). This indicates that parthenocarpic fruit-set and development in pat-2 also depend on GAs, and suggests that the capacity of pat-2 ovaries to set and develop may be due to their higher content of active GAs in the absence of pollination and/or fertilization, as found for the level of GA-like substances in pat (another parthenocarpic gene) tomato fruits at early stages of development (Mapelli et al., 1978).

The following members of the early-13-hydroxylation pathway of GA biosynthesis: GA₅₃, GA₄₄, GA₁₉, GA₂₀, GA₂₉, GA₁, and GA₈, were identified in developing non-parthenocarpic (MA/wt) tomato ovaries before anthesis (Table III). These GAs, with the exception of GA₅₃, plus GA₁₇ were previously identified in the pericarp of 20-d-old fruits (Bohner et al., 1988) and in entire 10-d-old pollinated fruits (Koshioka et al., 1994). GA₃, previously identified in vegetative tissues of tomato (Grünzweig et al., 1997), has now been found in unpollinated ovaries and, as in maize, may come from GA₂₀ via GA₅ (Fujioka et al., 1990). The identification of GA₉ and GA₃₄ (Table II), two members of the non-13-hydroxylation pathway, supports that both biosynthetic pathways are present simultaneously in the ovaries before pollination (see Fig. 3). This is in contrast to previous results showing that GA₁₅, GA₂₄, and GA₂₅ (from the non-13-hydroxylation pathway) were only present in seeds of 20-d-old pollinated fruits (Bohner et al., 1988).

View larger version (12K): [in this window] [in a new window]   Figure 3.   Possible pathways of GA biosynthesis in developing unpollinated tomato ovaries. The GAs in parentheses have not been identified in unpollinated ovaries.

It has also been reported that GAs could not be detected in pollinated tomato ovaries until several days after anthesis (Koshioka et al., 1994). The presence of GAs in unpollinated ovaries, however, as shown in this work, suggests that the developing ovaries are capable of GA biosynthesis. This idea is supported by the recent observation that genes coding for copalyl diphosphate synthase, GA 20-oxidase, and GA 3-hydroxylase are expressed during flower and early fruit development in tomato, and that a GA 20-oxidase gene (Le20ox-2) is localized in the placenta tissue of young, unpollinated tomato ovaries (Rebers et al., 1999). Our results also suggest that the GA biosynthetic capability of the ovary may be reduced or disappear once it has developed completely, around the time of anthesis, and be triggered again as a consequence of pollination, because the concentration of the quantitated GAs decreased progressively during ovary development in wt tomato plants (Fig. 2). This agrees with previous results in pea, in which GA₈, GA₁₉, and GA₂₀, present in emasculated, unpollinated ovaries, rapidly decreased in the absence of pollination (García-Martínez et al., 1991).

The quantitation of GAs showed that the concentration of GA₂₀ in pat-2 ovaries before pollination was much higher than in non-parthenocarpic ovaries (Fig. 2). This difference was found at the three stages of ovary development studied (FB, PR, and AN) and with two different genetic backgrounds (MA and CU). The accumulation of GA₂₀ in pat-2 ovaries was confirmed by identifying this GA by GC-SIM in both MA/pat-2 and CU/pat-2 ovaries (Table IV). Although the levels of GA₂₉ and GA₈ catabolite are not known, the higher content of GA₂₀ may not be due to a blockage of GA₂₀ metabolism. First, there was not, as would be expected if the metabolism of GA₂₀ to GA₂₉ were decreased, a large reduction in the amount of GA₂₉ in pat-2 ovaries (Fig. 2). Second, the higher amount of GA₁ (the product of 3-hydroxylation) in MA/pat-2 ovaries and a similar or higher amount of GA₁ plus GA₈ (the metabolic product of GA₁) in CU/pat-2 ovaries do not support a blockage of GA₂₀ to GA₁ metabolism. The accumulation of GA₂₀ in pat-2 ovaries may rather be due to a higher GA 20-oxidase activity, since the level of GA₁₉ (the immediate biosynthetic precursor of GA₂₀) in pat-2 ovaries was generally much lower than in non-parthenocarpic ovaries (Fig. 2). The recent isolation of three GA 20-oxidase cDNA clones of tomato, whose transcripts were detected in flower buds and during early fruit development (Rebers et al., 1999), should allow testing whether the expression of these genes is enhanced in pat-2 ovaries.

The application of GA₁ and GA₂₀ to unpollinated MA/wt tomato ovaries induced parthenocarpic fruit growth (Table V), but did not affect the parthenocarpic growth of unpollinated pat-2 ovaries (Table VI). This indicates that the endogenous content of active GA may be saturating in pat-2 ovaries. However, GA₂₀ does not seem to be active by itself because the simultaneous application of LAB, an inhibitor of GA₂₀ metabolism, blocked the induction by GA₂₀ of parthenocarpy in wt ovaries (Table V). Although the application of 5 mM LAB also inhibited the parthenocarpy of unpollinated pat-2 ovaries, simultaneous treatment with GA₁ reversed this effect but a GA₂₀ treatment was ineffective (Table VI). This strongly suggests that the effect of 5 mM LAB was due to its blockage of GA₂₀ metabolism. Furthermore, the inhibition of MA/pat-2 ovary growth by LAB (Table VI) also suggests that the accumulated GA₂₀ needs to be converted to an active GA for the expression of parthenocarpy of pat-2 ovaries.

It has been shown that GA₁ is the active GA in the control of stem elongation in many species (Graebe, 1987), and there is considerable evidence suggesting that it also regulates fruit-set and development in pea (García-Martínez and Hedden, 1997). GA₂₀ is the immediate precursor of GA₁ in many species (Graebe, 1987). The metabolic biosynthetic pathway of GA₁ in reproductive tomato tissues, however, is still unknown, and GA₄ as an alternative precursor, as suggested in pea fruit (Rodrigo et al., 1997), cannot be fully discarded. However, the low GA levels of members of the GA₄ biosynthetic pathway in young tomato fruit (Bohner et al., 1988; Koshioka et al., 1994) suggest that the early-13-hydroxylation pathway is the main operative pathway in the ovary and fruit (Fig. 3). The MA/pat-2 unpollinated ovaries had a higher GA₁ content than MA/wt ovaries (Fig. 2). In CU/pat-2 ovaries, however, the content of GA₁ was not higher than in the corresponding non-parthenocarpic isoline, but they contained more GA₈, the metabolic product of GA₁, and the amount of GA₁ plus GA₈ was higher at the PR and AN stages (Fig. 2).

A correlation between the level of [³H]GA₈ produced from applied [³H]GA₂₀ and pea internode elongation has also been found (Ingram et al., 1986), indicating that the content of GA₈ may be a measure of the previous GA₁ content in the tissues. This suggests that pat-2 unpollinated ovaries with the CU background may also synthesize more GA₁ than wt ovaries and that the flux of GA₁ may be important to induce parthenocarpic growth. Therefore, pat-2, whose expression can be detected before pollination (Fos and Nuez, 1996), may induce parthenocarpy by increasing the content or production of active GA in unpollinated ovaries, probably as a result of the high accumulation of GA₂₀. A similar situation occurs in the pea sln mutant, in which the blockage of conversion of GA₂₀ to GA₂₉ and GA₂₉ catabolite produces the accumulation of GA₂₀ in the mature seeds, leading to a high GA₁ content in young seedlings and to a slender phenotype (Ross et al., 1995).

In conclusion, the results presented show that the pat-2 mutation produces the accumulation of GA₂₀ in unpollinated tomato ovaries. Natural parthenocarpy induced by pat-2 is probably due to a higher synthesis of active GA (e.g. GA₁) in the developing unpollinated ovaries as a result of the accumulation of GA₂₀.