Effect of GA3 Treatment on Seed Development and Seed-Related Gene Expression in Grape

Background The phytohormone gibberellic acid (GA3) is widely used in the table grape industry to induce seedlessness in seeded varieties. However, there is a paucity of information concerning the mechanisms by which GAs induce seedlessness in grapes. Methodology/Principal Findings In an effort to systematically analyze the cause of this GA3-induced seed abortion, we conducted an in depth characterization of two seeded grape cultivars (‘Kyoho’ and ‘Red Globe’), along with a seedless cultivar (‘Thompson Seedless’), following treatment with GA3. In a similar fashion to the seedless control, which exhibited GA3-induced abortion of the seeds 9 days after full bloom (DAF), both ‘Kyoho’ and ‘Red Globe’ seeded varieties exhibited complete abortion of the seeds 15 DAF when treated with GA3. Morphological analyses indicated that while fertilization appeared to occur normally following GA3 treatment, as well as in the untreated seedless control cultivar, seed growth eventually ceased. In addition, we found that GA3 application had an effect on redox homeostasis, which could potentially cause cell damage and subsequent seed abortion. Furthermore, we carried out an analysis of antioxidant enzyme activities, as well as transcript levels from various genes believed to be involved in seed development, and found several differences between GA3-treated and untreated controls. Conclusion Therefore, it seems that the mechanisms driving GA3-induced seedlessness are similar in both seeded and seedless cultivars, and that the observed abortion of seeds may result at least in part from a GA3-induced increase in cell damage caused by reactive oxygen species, a decrease in antioxidant enzymatic activities, and an alteration of the expression of genes related to seed development.


Introduction
Grapevine (Vitis vinifera L.) is one of the most widely cultivated and economically important fruit crops in the world. Seedlessness of the fruit, which occurs when grapes only produce traces of aborted seeds, no seeds, or a significantly reduced number of seeds, is an especially desirable trait for consumers in the table grape industry [1]. This aberration in seed development is caused by parthenocarpy or stenospermocarpy [2], with incomplete seed development occurring as a result of embryo abortion and endosperm breakdown caused by changes in physiological conditions brought on by high levels of growth-promoting phytohormones near or at bloom [3,4].
One such class of endogenous phytohormone is the gibberellic acids (GAs), which are tetracyclic diterpenoids that are essential to plant growth and development. GAs are involved in many aspects of plant growth, including stem and leaf elongation, flower induction, trichome, anther, fruit and seed development, and seed germination [5][6][7][8][9][10]. In addition to their roles in plant development, GAs have also been shown to have an effect on reactive oxygen and antioxidant activities within grape tissues [11,12].
Interestingly, many studies have demonstrated that seedlessness in grapes can be induced through the exogenous application of GA 3 . For example, Delaware grapevine (V. labrusca) produces seedless berries following the application of 100 ppm GA 3 solution 12-17 d before full bloom [13], with similar effects being noted in Muscat Bailey A (V. vinifera × V. labrusca) [14], as well as Emperatriz and Emperador (V. vinifera) cultivars [15]. This role of GAs is not restricted to grapes, but has been observed in many species. For example, application of GA has been found to induce seed abortion in sweet cherry and Clementine mandarin [16,17], while GAs (possibly GA 1 and/or GA 3 ) have also been shown to play an important role early in pea (Pisum sativum L.) and Arabidopsis thaliana seed development by regulating embryo and/or endosperm and pollen tube development [9,18,19].
Although GAs are used widely in the table grape industry in all major producing countries, there is a paucity of information concerning the mechanisms by which GAs induce seedlessness in grapes. While a handful of studies have been carried out on this subject, the majority of these have focused upon the effect of GAs on pollen tube growth. These studies have demonstrated that GA 3 applied to grape flowers before or during anthesis severely inhibits pollen germination and pollen tube growth [14], and this may possibly be due to the biosynthesis of pollen tube inhibitor(s), leading to the production of unfertilized ovules [20].
In an attempt to improve our understanding of the mechanisms by which GAs induce seedlessness in grape, we aimed to investigate the effect of pre-bloom GA 3 application on the size and shape of berries and seeds, the frequency of seedlessness in the berries, and embryonic development. In addition, we conducted an analysis of the transcript levels of genes related to seed development following treatment with GA 3 . Furthermore, since GAs have been shown to have an effect on reactive oxygen species (ROS), which at high levels signal programmed cell death [21], as well as antioxidant enzyme activities within grape tissues [11,12], we tested whether GA 3 -induced seed abortion was linked to changes in ROS, cell damage and/or antioxidant enzyme activities.

GA 3 -induced seedlessness in grape
To produce seedless grape berries in seeded cultivars ('Kyoho' and 'Red Globe'), floral clusters were treated with GA 3 18 d prior to full bloom in the case of 'Kyoho' and 16 d prior to full bloom in the case of 'Red Globe' ( Figure S1). As shown in Table 1, there was a significant increase in the incidence of seedless berries in both cultivars following GA 3 treatment compared to untreated controls. In the case of 'Kyoho', the frequency of seedless berries was 3.3% and 98.6% in control and GA 3 -treated plants, respectively, whereas 'Red Globe' generated 3.0% and 85.2% seedless berries in control and GA 3 -treated plants, respectively.

Effect of GA 3 treatment on the growth of berries and seeds at different developmental stages
To determine whether there were any morphological effects on either the berries or seeds following GA 3 treatment, measurements (berry weight, transverse diameter, longitudinal diameter, as well as seed weight) were carried out 3,9,15,21,27,33,39,45,51,57 and 63 d after full bloom (DAF) in the seeded cultivars 'Kyoho' and 'Red Globe', as well as the seedless 'Thompson Seedless' cultivar. As shown in Figure 1, patterns of change with respect to grape berry weight, transverse diameter, longitudinal diameter, and seed weight were similar in 'Kyoho' and 'Red Globe' cultivars following GA 3 treatment, but differed in 'Thompson Seedless' plants. In the case of the 'Kyoho' cultivar, berry weight, transverse diameter and longitudinal diameter from GA 3 -treated plants were significantly higher than those of untreated plants from 3 to 9 DAF, with the difference in longitudinal diameter lasting until 15 DAF (Figure 1A a, c-g). Interestingly, 'Kyoho' berry weight and transverse diameter became significantly smaller than untreated berries after 15 DAF, while this same trend was observed after 21 DAF for longitudinal diameter ( Figure 1A a, g). These results were very similar to those seen in 'Red Globe' berries ( Figure 1B). Seed weight in GA 3 -treated 'Kyoho' plants was similar to that of untreated seeds from 3 to 9 DAF (Fig. 1A b, e), but from 15 DAF onwards, seeds from treated plants were found to weigh significantly less than those from untreated plants (Fig. 1A b, f-g). Similarly, seeds from GA 3treated 'Red Globe' plants were found to be significantly smaller than those harvested from untreated plants from 9 DAF onwards ( Fig. 1B b, e-g).
In contrast to 'Kyoho' and 'Red Globe' cultivars, the weight, transverse diameter and longitudinal diameter of berries from 'Thompson Seedless' plants treated with GA 3 were always higher than those of untreated plants ( Figure 1C a, c-g). Seeds from GA 3 -treated 'Thompson Seedless' plants weighed significantly more than seeds from untreated plants at 3 DAF, were similar in weight to untreated controls at 9 DAF, and weighed significantly less than untreated controls from 15 to 45 DAF. From 51 DAF onwards, GA 3 -treated 'Thompson Seedless' plants produced seeds that were slightly, but not significantly, lower in weight than untreated control seeds from the same cultivar ( Fig. 1C b, e-g). Although the size of GA 3treated grape berries was different from that of untreated controls in all three cultivars, there was no effect on the content of soluble solids in grape berries after GA 3 application ( Figure  S2).

Seed development following GA 3 application
To determine the effects of GA 3 treatment on seed development in grape, paraffin sections of GA 3 -treated and untreated 'Kyoho', 'Red Globe' and 'Thompson Seedless'  Figure S3). In the case of 'Kyoho' grape treated with GA 3 , normal seeds were evident up to 9 DAF (27 d after treatment) ( Figure 2H-L), but from 15 DAF (33 d after treatment) onwards, GA 3 -treated seeds stopped growing and embryo sacs degenerated ( Fig. 1Ab; Figure 2M-P). In the case of the 'Red Globe' cultivar, degenerated embryo sacs were observed in GA 3 -treated seeds 9 DAF (25 d after treatment) ( Figure S3H), while abnormal zygotes and endosperm nuclei were also observed in some cases at this time point ( Figure   S3I-J). As was the case for 'Kyoho', all embryo sacs of GA 3treated 'Red Globe' seeds had degenerated by 15 DAF (31 d after treatment) ( Figure S3K-L). In the 'Thompson Seedless' cultivar, normal zygotes, abnormal endosperm nuclei and the degeneration of embryo sacs were beginning to be apparent in GA 3 -treated seeds by 1 DAF (17 d after treatment), while the growth of GA 3 -treated seeds was completely halted and all embryo sacs had degenerated by 9 DAF (25 d after treatment) ( Fig. 1Cb; Figure 3I-L). In order to assay the effects of GA 3 treatment on anther development, paraffin sections of 'Kyoho', 'Red Globe' and 'Thompson Seedless' anthers were generated and it was found that there was no apparent difference between those treated with GA 3 and control untreated anthers in any of the three cultivars ( Figure S4). Vertical bars indicate standard errors. Asterisks indicate significant differences between GA 3 treated and untreated control samples (*P < 0.05, **P < 0.01, Student's t-test). In each column, the first and second photos represent untreated control and GA 3 -treated berries, respectively, the third and fourth photos represent longitudinal sections of untreated control and GA 3 -treated berries, respectively, and the fifth and sixth photo represent untreated control and GA 3 -treated seeds, respectively. doi: 10.1371/journal.pone.0080044.g001

Changes in H 2 O 2 and MDA content following GA 3 treatment
It has been proposed that ROS, such as H 2 O 2 , play two very opposing roles in living organisms: at low levels they act as a signaling pathway to activate defense responses, while at high levels they exacerbate stress damage and signal programmed cell death [21]. One manner in which ROS can cause cell damage is through the lipid peroxidation pathway [22]. Since MDA content is a direct measure of lipid peroxidation, it was utilized as a marker to assess the degree of cell damage [23]. Therefore, we analyzed H 2 O 2 accumulation and MDA content in all three grape cultivars following GA 3 treatment from 1 to 75 d following application for the 'Kyoho' cultivar and 1 to 73 d following application for the 'Red Globe' and 'Thompson Seedless' cultivars.
As shown in Figure 4, peaks in the accumulation of H 2 O 2 were noted 12 d after GA 3 Figure 4F) were similar in pattern to those observed with respect to H 2 O 2 accumulation. In 'Kyoho' berries, peaks in MDA content were observed 6 d, 18 d and 39 d after GA 3 treatment. Conversely, MDA content in both 'Red Globe' and 'Thompson Seedless' berries peaked at 6 d following treatment, with levels declining thereafter. Interestingly, in all three cultivars, MDA content 6 d after treatment was always significantly higher than in the untreated control.

Changes in the activities of various antioxidant enzymes in response to GA 3 treatment
In order to prevent damage caused by high levels of ROS, cells must maintain a state of redox homeostasis. It has been suggested that this may be attained through interactions that occur between ROS and scavenger antioxidant enzymes. We analyzed the activities of three such enzymes (SOD, CAT and POD) in flowers/berries harvested between 1 and 75 d following GA 3 treatment for the 'Kyoho' cultivar or 1 and 73 d following GA 3 treatment for 'Red Globe' and 'Thompson Seedless' cultivars. In all cases, their activities were found to differ between treated and untreated controls ( Figure 4G Similarly, the activity of CAT in the flowers of all three cultivars tested 1 d after GA 3 treatment were diminished compared to untreated controls, however, this difference was significant only in the 'Red Globe' cultivar ( Figure 4J-L). These values remained at least somewhat lowered in GA 3 -treated samples until 30 d following treatment in 'Kyoho', 12 d following treatment in 'Red Globe', and 22 d following treatment (6 DAF) in 'Thompson Seedless'.
In the case of POD activity, untreated control samples from all three cultivars exhibited a large peak 12 d after treatment in 'Kyoho' and 'Red Globe' berries, or 22 d after treatment (6 DAF) in 'Thompson Seedless' berries ( Figure 4M-O). This peak was significantly reduced in GA 3 -treated samples from all three cultivars ( Figure 4M-O).
In an effort to provide further evidence that changes were occurring with respect to SOD, CAT and POD in response to GA 3 treatment in 'Kyoho', 'Red Globe' and 'Thompson Seedless' flowers, we analyzed the expression of their corresponding genes using semi-quantitative RT-PCR. As shown in Figure 5 (see also Figure S5), SOD1 expression decreased in 'Kyoho' and 'Red Globe' flowers following GA 3 application, while levels increased initially in the 'Thompson Seedless' cultivar. In the case of SOD2, expression appeared to decrease in 'Kyoho' flowers following treatment, while expression increased initially and then decreased following GA 3 treatment, in 'Red Globe' and 'Thompson Seedless' flowers. SOD3 transcript levels were increased 0.5 h and 1 h following treatment and declined thereafter in all three cultivars.
Similar patterns were observed with the expression of catalase (CAT1 and CAT2) and peroxidase genes (POD); CAT1 and CAT2 were both initially up-regulated and subsequently down-regulated in the flowers of all three cultivars, with a similar pattern being noted for POD expression in 'Red Globe' and 'Thompson Seedless' flowers. Conversely, POD expression in 'Kyoho' flowers simply appeared to be down-regulated beginning 3 h post-treatment ( Figure 5, Figure  S5). These results correlate well with the enzymatic activity results, whereby in all three cultivars SOD, CAT and POD activities in GA 3 -treated samples (with the exception of POD activity in 'Red Globe') tended to be reduced compared to untreated samples ( Figure 4G-O).

Expression of genes related to seed development following GA 3 treatment
To analyze the effect of GA 3 application on grape genes that are putative orthologs of Arabidopsis genes known to be involved in seed development, we assayed the expression of  Figure S6). In flowers, the expression patterns of AG, LEC2-like, and TTG2-like were similar in 'Kyoho' and 'Thompson Seedless' cultivars, whereby they appeared to be initially up-regulated and then subsequently down-regulated following treatment. This same pattern was observed for VAL3-like expression in the flowers of all three cultivars. Conversely, in 'Red Globe' flowers, the expression of AG, LEC2-like, and TTG2-like exhibited downregulation post-GA 3 treatment. In the seeds, these four genes all exhibited up-regulation in 'Kyoho' following treatment while in 'Thompson Seedless' they were all initially up-regulated and then subsequently down-regulated after GA 3 application. In 'Red Globe' seeds, AG and VAL3-like were initially upregulated and subsequently down-regulated while TTG2-like and LEC2-like were simply up-regulated following treatment. In the case of AP2, 'Kyoho' flowers, as well as seeds from all three cultivars, exhibited up-regulation following treatment, while both the flowers and seeds of all three cultivars displayed down-regulation of LEC1-like after GA 3 application. MEA-like appeared to be down-regulated following treatment in the flowers of all three cultivars, while in the seeds, all three cultivars exhibited an initial up-regulation and subsequent down-regulation. In the case of GASA4-like, expression appeared to be up-regulated in 'Kyoho' flowers following

Discussion
Large fruit size, as well as reduced seed number and size, are major goals in the improvement of fruit crop species [1]. At present, exogenous GA 3 application is often used as a strategy to induce seedlessness of seeded grapes; however, very little is known concerning the mechanism by which this occurs. Therefore, we endeavoured to investigate the effects of GA 3 application before bloom on berry and seed development in two seeded ('Kyoho' and 'Red Globe') and one seedless ('Thompson Seedless') cultivars, as well as its effect on antioxidant enzyme activity and the expression of genes involved in seed development.
In terms of seed growth, GA 3 application had a significant influence on seed abortion in 'Kyoho' and 'Red Globe' cultivars (Table 1, Figure 1A-B), and also affected the early stages of Results of semi-quantitative RT-PCR assays were quantified using GeneTools software, and log-transformed values of the relative expression levels of genes encoding antioxidant enzymes following GA 3 treatment compared to untreated controls were used for hierarchical cluster analysis. The color scale represents relative expression levels with red denoting up-regulation and green denoting down-regulation. Sampling times are indicated at the top of the figure; F represents floral tissue; K represents the 'Kyoho' cultivar; R represents the 'Red Globe' cultivar; T represents the 'Thompson seedless' cultivar. doi: 10.1371/journal.pone.0080044.g005 Figure 6. Hierarchical clustering of genes related to seed development in three grape cultivars following GA 3 treatment. Results of semi-quantitative RT-PCR assays were quantified using GeneTools software, and log-transformed values of the relative expression levels of genes involved in seed development following GA 3 treatments compared to untreated controls were used for hierarchical cluster analysis. The color scale represents relative expression levels with red denoting up-regulation and green denoting down-regulation. Sampling times are indicated at the top of the figure; F represents floral tissue; K represents the 'Kyoho' cultivar; S represents seed tissue; R represents the 'Red Globe' cultivar; T represents the 'Thompson seedless' cultivar. doi: 10.1371/journal.pone.0080044.g006 seed development in 'Thompson Seedless' grapes ( Figure 1C). Exogenous application of GA 3 is known to impair fertilization by either inducing seed abortion or reducing pollen tube growth in grapes and mandarin [14,17,20]. However, since our results demonstrated that seed growth in GA 3 -treated samples ceased after full bloom (Figure 1), it seems likely that in this case ovules were in fact being fertilized prior to seed abortion. This is also known to be the case in stenospermocarpic seedless grape cultivars, such as 'Thompson Seedless', whereby pollination and fertilization occur but the embryos are aborted [24]. Recently, it has been reported that embryo abortion in 'Thompson Seedless' began between 40-55 DAF [25]. However, seed abortion in GA 3 -treated samples of both seeded and seedless cultivars in the present study occurred between 9 and 15 DAF (Figures 1-3, Figure S3). These results demonstrate that exogenous GA 3 application does not precisely imitate the natural process of embryo abortion that occurs in stenospermocarpic grapes.
In order to determine whether fertilization did indeed occur prior to seed abortion following GA 3 application, we analyzed the structural development of seeds in the three cultivars following treatment. In a similar fashion to untreated controls, ovule structure prior to pollination was not affected by GA 3 application, and both zygotes and endosperm nuclei were observed in all three cultivars after full bloom following treatment ( Figure 2, Figure 3, Figure S3), which suggests that the ovules had undergone fertilization normally. Subsequently, however, abnormal zygotes and endosperm nuclei became apparent, with complete abortion occurring 15 DAF in 'Kyoho' and 'Red Globe' cultivars (33 and 31 d after treatment, respectively), and 9 DAF (25 d after treatment) in the 'Thompson Seedless' cultivar following GA 3 treatment ( Figure  2, Figure 3, Figure S3). These results correspond well with the reduction in seed size noted in 'Kyoho' and 'Red Globe' cultivars at 15 DAF (33 and 31 d after treatment, respectively) ( Figure 1) and suggest that instead of impairing fertilization, exogenous GA 3 application is somehow interfering with seed development.
Interestingly, GA 3 treatment was found to have contrasting effects on mean berry weights of seeded and seedless cultivars, inhibiting growth in seeded 'Kyoho' and 'Red Globe' cultivars while stimulating growth in 'Thompson Seedless' (Figure 1). These results were consistent with previous findings in which growth was inhibited in the seeded grape cultivar 'Emperador' and promoted in the seedless cultivar 'Emperatriz' upon GA 3 application [15].
Recently, it has been reported that reactive oxygen species (ROS), including superoxide anions and H 2 O 2 , act as regulators of growth and development in plants [26]. These molecules carry out their function through their interaction with, and damage caused to, substrates such as various enzymes and nucleic acids, serving as signaling molecules that induce cell death. Since the production of H 2 O 2 has been found to be induced by GA [27], it seemed reasonable to suppose that GA 3 -induced seed abortion may result from the accumulation of ROS. Furthermore, since lipid peroxidation products can be utilized as biomarkers for oxidative injury, we also analyzed MDA content in flowers/berries to assess the amount of lipid peroxidation, and thus the degree of cell damage, following GA 3 treatment [23]. In this study, H 2 O 2 content in GA 3 -treated samples was significantly higher than that in untreated samples 12 d after GA 3 treatment (in 'Kyoho', a second peak also occurred 30 d after GA 3 treatment), while MDA content peaked 6 d after treatment, which in both cases was earlier than in untreated controls (Figure 4). The fact that H 2 O 2 levels peaked after MDA upon GA 3 treatment are consistent with previous findings that suggest that the accumulation of aldehydes (such as MDA) above a certain threshold level leads to the production of ROS (such as H 2 O 2 ) [28,29].
Since excess levels of ROS cause damage, cells must attempt to maintain a state of redox homeostasis. One model of how this is attained suggests that interactions occur between ROS and antioxidant enzymes, which then act as a metabolic interface for signals derived from both metabolism and the environment. Indeed, it has been reported that the extent to which ROS accumulate is determined by the antioxidative system, which enables organisms to maintain proteins and other cellular components in an active state [30]. Interestingly, the antioxidant enzyme genes tested here (SOD1, SOD2, SOD3, CAT1, CAT2 and POD) tended to be down-regulated in flowers after treatment with GA 3 ( Figure 5, Figure S5). In line with this, their corresponding enzyme activities (SOD, CAT, and POD) were reduced after GA 3 application (Figure 4). This correlates well with previous findings whereby GA 3 treatment was found to down-regulate the expression of SODs [12] and decrease the activities of CAT and SOD [31]. All of the GA 3induced changes in H 2 O 2 and MDA content, as well as antioxidant enzyme activities, occurred prior to GA 3 -induced seed abortion.
It has been reported previously that hydroxyl radicals facilitate cell enlargement [32,33], and that the GA 3 -induced modulation of cell redox homeostasis may play a role in the determination of the shape and size of grape berries [12]. This could also be the case in this study, as we noted increased size of GA 3 -treated berries in all three cultivars prior to seed abortion ( Figure 1) and during the same time frame as changes in H 2 O 2 and MDA content, as well as antioxidant enzyme activities, were observed ( Figure 4). Taken together, these results suggest that GA 3 application interferes with redox homeostasis by decreasing the scavenging activity of ROS detoxification enzymes, thus leading to an accumulation of ROS and cell damage within seeds.
To determine whether the observed changes in seed development upon GA 3 application could also be due, at least in part, to variation in the expression of genes involved in seed development, we carried out semi-quantitative RT-PCR using flower and seed tissue and tested a selection of genes ( Figure  6, Figure S6) that have been characterized previously in species such as Arabidopsis, Petunia and rice [34]. In Arabidopsis, the MEDEA (MEA) gene regulates cell proliferation in both the embryo and the endosperm by impacting gametophytic maternal control during seed development [35]. Mutations in this gene have been found to induce two distinct phenotypes: silique elongation without fertilization [36,37] or seed abortion [38]. In Arabidopsis, expression of the MEA gene was lower in the early stages of seed development and increased subsequently [39]. In this study, expression of the grape MEA-like gene was found to increase initially, then decline in GA 3 -treated flowers and seeds ( Figure 6, Figure S6), which could be related to seed abortion in GA 3 -treated plants.
GASA4 regulates floral meristem identity and also positively influences both seed size and total seed yield [40]. It has been reported previously that the GASA4 gene is up-regulated by GAs in meristematic regions and flower buds in Arabidopsis, but down-regulated in cotyledons and leaves [41]. In 'Kyoho' and 'Red Globe' seeds, we found the grape GASA4-like gene to be down-regulated following GA 3 treatment (Figure 6, Figure  S6). Similarly, LEC1 is an important regulator of embryo development that activates the transcription of genes required for both embryo morphogenesis and cellular differentiation [42]. Moreover, LEC1 RNA accumulates only during seed development in embryonic and endosperm tissue [42]. In grape, LEC1-like exhibited down-regulation in both flowers and seeds following GA 3 treatment ( Figure 6, Figure S6). Therefore, it is possible that the GA 3 -induced down-regulation of both GASA4-like and LEC1-like may be detrimental to seed development and contribute to GA 3 -induced seed abortion. Conversely, in Arabidopsis, mutation of the AP2 gene causes the production of large seeds and an increase in both embryo cell number and cell size [43,44]. In grape, AP2 expression was up-regulated by GA 3 (Figure 6, Figure S6), which would likely interfere with embryo development and thus may have played an important role in GA 3 -induced seedlessness.
In 'Thompson Seedless' and 'Kyoho' flowers and seeds, as well as 'Red Globe' seeds, the regulation of LEC2-like, AG and TTG2-like in response to GA 3 was similar in that they tended to be initially up-regulated and then down-regulated following application. Since LEC1, which is an activator of seed development, is known to positively regulate LEC2 in many plants [45], we expected that LEC2-like expression would follow a similar pattern to LEC1-like in grape following GA 3 treatment. As expected, the grape LEC2-like gene was downregulated following GA 3 treatment in flowers, which resembled our results for the LEC1-like gene ( Figure 6, Figure S6). However, in 'Kyoho' and 'Red Globe' seeds, the response of LEC2-like expression to GA 3 application did not correspond with LEC1-like as it was up-regulated ( Figure 6, Figure S6). This up-regulation was also converse to the putative activating function of the grape LEC2-like gene, which suggests that this gene may play a slightly different role in grape [46].
Genes of the AGAMOUS (AG) subfamily have been shown to play crucial roles in reproductive organ identity, as well as fruit and seed development [47,48]. In agreement with our results for 'Kyoho' and 'Thompson Seedless' cultivars, it has been found previously that AG is up-regulated in inflorescence apices after GA treatment [49]. Furthermore, TTG2-like, which has been implicated in integument cell elongation and endosperm growth in Arabidopsis [50], was mainly downregulated in floral tissues and up-regulated in seeds following GA 3 application. Taken together, these expression results lend credence to the suggestion that GA 3 -induced seed abortion is a complex process, with changes in the expression of genes related to seed development playing a role.
In the present study, we investigated the effects of GA 3 application before full bloom on berry and seed development in two seeded ('Kyoho' and 'Red Globe') and one seedless ('Thompson Seedless') cultivar, as well as its effect on ROS content and antioxidant enzyme activity, and the expression of genes involved in seed development. Our findings suggest that GA 3 -induced seed abortion is a result of abnormal seed development and may be caused, at least in part, by an impairment of redox homeostasis in flowers/berries resulting in oxidative damage to the seeds, as well as by changes in the expression of genes related to seed development. While these findings have enhanced our knowledge of how seedlessness is induced by exogenous GA 3 application in grape, further analyses will be required to provide a more in depth understanding of the exact mechanism of GA 3 -induced seedlessness due to the apparent complexity of the mechanism involved.

Plant Material and GA 3 treatment
Seeded grape cultivars 'Kyoho' (Vitis vinifera × V. labrusca) and 'Red Globe' (V. vinifera), as well as the seedless 'Thompson Seedless' (V. vinifera) cultivar, were grown in an 8year-old vineyard situated in an experimental field of Northwest A&F University, Yangling, Shaanxi, China (34° 20' N, 108°24' E). Both treated and untreated control experiments consisted of three biological replicates, with five vines per replicate, respectively. Fifteen clusters were kept on each vine, and treatment was carried out at 18 d before full bloom in 'Kyoho' and 16 d before full bloom in 'Red Globe' and ' Thompson Seedless' cultivars. Initially, clusters were soaked in 0.05% Tween-20 (Roche, Basel, Switzerland) for 3 s to enhance subsequent uptake of the phytohormone. GA 3 treatment was then carried out by soaking clusters in 100 mg L -1 GA 3 (Sigma-Aldrich, St. Louis, MO, USA) dissolved in a small amount of 100% ethanol for 5 s ( Figure S1) [13]. Untreated control clusters were subject to the same process without GA 3 .

Analysis of seed development
For morphological analyses of developing seeds, flowers, young grape berries and seeds were collected one day after GA 3 treatment, every two subsequent days for 18 d, and then every three days until 39 d after application. Three biological replicates were harvested in each case. Following fixation in Carnoy's fluid (3:1 anhydrous alcohol/glacial acetic acid) for 12-24 h at room temperature, the samples were dehydrated through an ethanol series (30%, 50% and 70%) and stored in 70% ethanol at 4°C until further use [51]. Samples stored at 4°C were further dehydrated with a graded ethanol series (85%, 95% and 100%), infiltrated with xylene, and embedded in paraffin (Taiva, Hubei, China). Sections 8-10 μm in thickness were transferred onto poly-L-Lys-coated glass slides (WHB, Shanghai, China), deparaffinized with xylene, and re-hydrated through an ethanol series (100%, 95%, 85%, 70%, 50% and 30%). The resulting sections were stained with Ehrlich's haematoxylin (Saichi, Shanghai, China) for 30 minutes at room temperature, dehydrated with an ethanol series, infiltrated with xylene, sealed with resinene (XT, Shanxi, China), and finally mounted beneath a coverslip. Slides were observed using an optical microscope (OLYMPUS BH-2, Japan) and photos were obtained using an attached digital camera (OLYMPUS DP72, Japan).
Total SOD activity was assayed by monitoring the inhibition of the photochemical reduction of nitro blue tetrazolium (NBT; Amresco, Solon, OH, USA), which produces a blue colouration in the presence of reactive oxygen, using a UV-1700 UV-Vis Spectrophotometer (Shimadzu, Kyoto, Japan) as described previously [54]. In brief, 3 ml reaction mixtures comprising 50 mM sodium phosphate buffer (pH 7.0), 13 mM methionine, 75 μM NBT, 2 μM riboflavin (Amresco), 0.1 mM EDTA, and 100 μl enzyme extract were illuminated for 20 min at an intensity of 4000 lx using a GXZ-280B illumination incubator (Ningbo Jiangnan Instrument Factory, Ningbo, China) and then assayed at 560 nm. One unit of SOD activity was defined as the amount of enzyme required to cause 50% inhibition of NBT reduction at 560 nm.
CAT activity was determined by monitoring the consumption of H 2 O 2 at 240 nm for 3 min as described previously [55]. In each case, reaction mixtures comprised 50 mM sodium phosphate buffer (pH 7.0), 10 mM H 2 O 2 and 200 μl of enzyme extract in a 3 ml volume. CAT activity was reported as 0.01 times the amount of change in absorption per min per g FW.
The activity of POD was assayed by adding 100 μl aliquots of tissue extracts to 3 ml of assay solution, consisting of 13 mM guaiacol (Amresco), 5 mM H 2 O 2 and 50 mM sodium phosphate buffer (pH 7.0) [56]. Increases in optical density were then measured at 470 nm for 3 min at 25°C using a UV-1700 UV-Vis Spectrophotometer (Shimadzu). POD activity was expressed as the change in absorbance per min per g FW. . First-strand cDNA was synthesized using 500 ng total RNA as template along with PrimeScript TM RTase and an oligo dT primer (TaKaRa Biotechnology, Dalian, China). Subsequent PCR assays were carried out in a final volume of 20 µl including 1µl of cDNA template, 0.4 μmol L -1 gene-specific primers, and 2x HotMaster Taq PCR MasterMix (Tiangen Biotech Co. Ltd., Beijing, China). The sequences of the gene-specific primers utilized in this study can be found in Table S1, with the exception of those used for amplification of the grape CAT1, SOD1, SOD2 and SOD3 genes, which have been described previously [57]. Since several of the genes tested had not previously been identified in grape, the amino acid sequences of Arabidopsis GASA4 (CAA66909.1), LEC1 (NP_173616.2), LEC2

Gene expression analysis by semi-quantitative RT-PCR
(NP_564304.1), MEA (NP_563658.1), TTG2 (NP_001078015.1), and VAL3 (NP_193886.2) were used as query sequences to search for putative orthologs in the grape genome database, GENOSCOPE (http:// www.genoscope.cns.fr/spip/), as well as the National Center for Biotechnology Information (NCBI, http://www.ncbi.nlm.nih.gov). The grape Actin1 gene fragment (GenBank Accession number AY680701) was amplified with primers F (5'-GAT TCT GGT GAT GGT GTG AGT-3') and R (5'-GAC AAT TTC CCG TTC AGC AGT-3') as an internal standard. Thermal parameters for amplification were as follows: 94°C for 3 min, 25 cycles of 94°C for 30 s, 58°C for 30 s and 72°C for 30 s, followed by a final elongation at 72°C for 5 min. Three independent biological replicates and technical replicates, respectively, were performed in each case. Subsequently, 10 μl of the resulting PCR products were separated on a 1.2% (w/v) agarose gel, stained with ethidium bromide and photographed under UV illumination. The results of all semi-quantitative RT-PCR assays were quantified using GeneTools software (Syngene, Frederick, MD, USA) and log-transformed values of relative expression levels following GA 3 treatment compared to untreated controls were used for hierarchical cluster analysis with the MultiExperiment Viewer (MeV version 4.8.1, http:// www.tm4.org/).

Statistical analysis
Data were analyzed using student's t-tests carried out using SPSS software (SPSS 17.0 ® , Chicago, IL, USA).