Spring production was evaluated in trials performed in two different locations: Monsampolo and Pontecagnano. Spring production was divided into early production, consisting of the first four harvests, and total production, consisting of sixteen harvests. Early spring production corresponds to the cultivation period from March to the first half of May, with temperatures somewhat low for fruit-set and growth. During this period, the average minimum and maximum temperatures ranged from 7° to 17°C in southern Italy (Pontecagnano) and from 14° to 17°C in central Italy (Monsampolo). The transgenic parthenocarpic hybrids gave a significantly higher yield, on the average a six-fold increase during early production, in comparison to their controls (Table 1). The increment in the number of fruits per plant and the higher average weight of the fruits were the main causes of the increased early spring production of the parthenocarpic hybrids. The suboptimal/adverse environmental conditions did not affect the growth of GM parthenocarpic fruits and the average fruit weight was significantly higher in the GM eggplants in comparison to untransformed controls. The traditional parthenocarpic cultivar Talina produced fruits with an average weight similar to those of GM eggplant fruits (Table 1). However, due to the higher number of fruits per plant, the productivity of the transgenic parthenocarpic hybrids was also increased, by an average of five-fold, when compared to cv Talina (Table 1). During the whole harvesting period, the GM fruits were always seedless (Fig. 1) and were normal in both size and shape.

The increased productivity of GM hybrids characterised both the early spring production (i.e. the first four harvests) and the whole spring production cycle (i.e. sixteen harvests). During the whole spring production cycle, the hybrids P1 and P2 gave an average yield that was 46% higher with respect to the corresponding control C1 (Table 1). The hybrid P5 gave a 37% higher yield with respect to its control C2. The average total number of fruits produced per plant in both locations was similar in all the hybrids (8–9 fruits/plant). However, the higher average weight of the GM fruits led to a higher total yield of transgenic hybrids with respect to their controls. When considering the whole spring cultivation cycle, the parthenocarpic cultivar Talina gave a total production that was not significantly different from either of the three GM hybrids (Table 1).

Summer production was evaluated in an open field trial carried out during the optimal period of eggplant cultivation. Plants were transplanted on May 20^th and the last harvest took place on September 11^th. The early production of the transgenic hybrids, consisting of the first three harvests, was significantly higher than that of the untransformed hybrids (Table 2): P1 and P10 hybrids yielded, respectively, 36% and 76% more than their corresponding controls, C1 and C10. The difference in productivity between P1 and C1 hybrids, which have long-shaped fruits, was caused by the higher average weight of GM fruits. When comparing P10 and C10 hybrids, which have sub-oval fruits, the higher yield obtained with GM plants during the early harvesting period was due to the increased number of fruits per plant.

Consisting of ten harvests, the total production of P1 hybrids was 37% higher than control C1 eggplants (Table 2). The difference in total yield between P1 and control C1 hybrids was statistically significant and due both to the higher number of fruits/plant and to the increased weight of GM fruits. It is noteworthy to point out that when considered individually, neither trait (number of fruits/plant or fruit weight) showed statistically significant differences between the GM and untransformed plants (Table 2). Although higher in P10 hybrids in comparison to its control C10, the total yield (the number and average weight of the fruits) was not significantly different between the two. During the whole harvesting period, fruits from both P1 and P10 parthenocarpic hybrids were always seedless (Fig. 2), whilst control fruits always had seeds. Therefore, under open field cultivation, the DefH9-iaaM gene had a positive influence on fruit quality, as GM DefH9-iaaM fruits were always seedless. Fruit quality affects the economic value of eggplant production.

Although the environmental conditions were optimal and consequently did not affect negatively fruit-set, the DefH9-iaaM gene caused both faster development and growth of the fruits as indicated by the increased early-summer production (the first three harvests). In this regard, it is worthwhile to stress that expression of the DefH9-iaaM gene takes place in the placenta and ovules before pollen development. As a consequence, in GM parthenocarpic plants fruits are seedless and fruit development initiates well before non-GM controls 11.

In all trials we have never used homozygous lines because growers mostly employ F1 hybrids. The use of hemizygous primary transformants as pollinator plants allowed us to obtain in rather short time, by in vivo selection for kanamycin resistance, F1 plants transgenic for the DefH9-iaaM gene. Young, healthy and vigorously growing plants did not produce seeds. However, it is possible to obtain seeds from aged DefH9-iaaM transgenic plants both by selfing and crossing. By exploiting the delayed female fertility we have produced the homozygous plants needed as male parents for rapid seed multiplication of F1 eggplant hybrids. Therefore, the female sterility of young and mature plants is not an insuperable hindrance for mass propagation and commercial fruit production.

The DefH9 gene is expressed specifically in the placenta and ovules during early phases of flower development 6. To determine whether expression of the DefH9-iaaM gene also occurs during later stages of fruit growth and whether it is influenced by pollen fertilization, mRNA from transgenic flowers and fruits obtained either from emasculated or hand pollinated flowers was analyzed by RT-PCR at different stages of development, until the fruit attained a length of 28 cm. An amplicon of 161 bp corresponding to the spliced DefH9-iaaM mRNA was detected in all stages analyzed (Fig. 3, lanes 2–6). Thus, the presence of the mRNA of the DefH9-iaaM gene and consequently its action is most likely not limited to early stages of flower and fruit development. Pollination did not affect the steady state level of DefH9-iaaM mRNA (Fig. 3, compare lane 5 and lane 6).

Treatment with auxin often causes parthenocarpic development in several plant species [for review, see: 12]. However, in some species and/or varieties, to efficiently sustain fruit growth, the hormonal treatment of the flowers must be repeated 13. The finding that DefH9-iaaM mRNA is also present during later stages of fruit development is consistent with the interpretation that in DefH9-iaaM parthenocarpic plants, the placenta, the ovules, and the tissues derived therefrom are a source of auxin during the whole growth of the fruit. As a consequence, they efficiently sustain fruit growth.

Trials were carried out in central Italy (Monsampolo del Tronto-AP) and in southern Italy (Pontecagnano-SA) (approval of the Italian Ministry of Health N° B/IT/97-29). The greenhouses were rather similar and made of galvanized steel and covered with plastic polyethylene (0.12 mm thick). An apparatus for drip-irrigation was used and the soil was completely mulched. A complete randomized block design with three replicate hybrid genotypes was adopted. Each experimental plot measured 3.12 m² and contained eight plants in a double row. Transplanting was performed on March 3^rd in southern Italy and on March 27^th in central Italy. The P1, P2, P5, C1, C2 and the commercial Talina hybrids were employed. Transgenic parthenocarpic hybrids P1, P2 and P5 were obtained by crossing (as male parent) the primary transgenic plant DR2 iaaM #34-1 with the line Tal 1/1 (P1), the primary transgenic plant DR2 iaaM #28-1 with Tal 1/1 (P2) and the transgenic plant Tal 1/1 iaaM #1-1 with the line Tina (P5). The hybrids P1 and P2 are homologous to C1 (DR2 × Tal1/1), except for the presence of the DefH9-iaaM gene integrated in their genome. The transgenic hybrid P5 is homologous to its untransformed control C2 (Tal1/1 × Tina). DR2 and Tina are parental lines obtained through classical breeding, Tal1/1 is a double haploid line derived from anther culture of the F1 commercial cultivar Talina. The segregation of the marker gene nptII was checked by spraying the plants with kanamycin 14 and allowed for the conclusion that the transgenes segregate as a single locus in the backcrossed progenies of the three independent events analyzed (Tal iaaM 1-1: χ² = 0.01065, P = 0.917; DR2 iaaM 34-1: χ² = 0.0496 P = 0.824; DR2 iaaM 28-1 χ² = 0.06467 P = 0.799). Southern blot analysis showed that DR2 iaaM 28-1 and 34-1 had a single copy of the transgene, while Tal iaaM 1-1 had three copies of the transgene (Fig 4). Since the interaction genotype/location was not significant for the yield, the data were computed as average of the two locations and subjected to analysis of variance according to a randomized complete block design. Duncan's Multiple Range Test (P = 0.05) was used for means separations.

The open field trial was carried out under open field conditions at Monsampolo del Tronto (approval of the Italian Ministry of Health B/IT/99/21). Two transgenic parthenocarpic hybrids were tested: the hybrid P1 (Tal1/1 × DR2 iaaM #34-1) with elongated fruits and the hybrid P10 (UGA × Tal1/1 iaaM #1-1) with sub-oval fruits were compared to their homologous non-transgenic controls C1 (DR2 × Tal1/1) and C10 (UGA × Tal1/1), respectively. The UGA line has oval dark purplish fruits and it has been provided by Dr. S.C. Phatak. A complete randomized block design with the hybrids replicated four times was adopted. Each experimental plot measured 11.7 m² and contained 30 plants in a double row. Transplanting was performed on May 10^th.

Early spring production consisted of the first four harvests (i.e. 4 out of 16 harvests during the whole production cycle), while early summer production, whose cultivation cycle consists of ten harvests, corresponds to the first three harvests. For all trials the number and weight of fruits were recorded. In addition, fruit sample for each harvest and replication was cut to check for the presence of seeds. Data were computed for the early harvesting time and for the whole harvesting season. Analysis of variance was performed according to a randomized complete block design. Duncan's Multiple Range Test (P = 0.05) was used for means separations.

High-molecular-weight DNA was isolated from the young leaves of transgenic and untransformed eggplants by using Plant DNAzol (Invitrogen), according to the manufacturer's instructions. Ten μg of DNA was digested overnight with 80 units of KpnI (Promega) in a volume of 500 μl, separated on a 0.7% agarose gel and transferred to Hybond N (Amersham Pharmacia Biotech). A 1350 bp fragment of the DefH9 gene was used as a radiolabeled probe. The membrane was hybridized overnight in 5X SSC/50% formamide (Sigma) at 42°C and washed two times for 15 min. in 2 × SSC/0.1% SDS, and two times for 15 min. in 0.1 × SSC/0.1% SDS at 42°C. Signals were detected using Kodak X-OMAT AR5 film (Sigma).

Semiquantitative (competitive) PCR analysis was carried out for 38 cycles (annealing temperature 63°C) using as template cDNA (8 ng) obtained by priming poly(A)⁺ mRNA with an iaaM specific primer (5'-AATAGCTGCCTATGCCTCCCGTCAT-3'). The mRNA was extracted from either young flower buds (5,8,11 mm) or eggplant fruits (placental tissue from fruits either 40 or 280 mm long). As an internal standard, 0.5 fg of a 600 bp long DefH9 cDNA fragment was used in the PCR assay. To amplify the 161 bp long amplicon an iaaM specific primer (5'-GGGTGAATTAAAATGGTCATACAT-3') and a DefH9 specific primer (5'-CTTTGGAACTCGTGTTGAGCTCTCA-3') were used. For the internal standard, a 3' primer (5'-TGAGCATTGATCTCCTGAGTGGTGT-3') together with the DefH9 specific primer were used to produce the 351 bp long amplicon. The PCR assays were performed with a thermostable DNA polymerase mixture (Expand High Fidelity PCR system, Roche) in presence of 3 μCi of ³²P dCTP. The intensity of the bands was quantified by using an Instant Imager (Packard, Meriden, CT).