Genetic engineering of parthenocarpic tomato plants using transient SlIAA9 knockdown by novel tissue-specific promoters

Parthenocarpy is the development of an ovary into a seedless fruit without pollination. The ubiquitous downregulation of SlIAA9 induces not only parthenocarpic fruit formation but also an abnormal vegetative phenotype. To make parthenocarpic transgenic tomato plants without unwanted phenotypes, we found two genes, namely, Solyc03g007780 and Solyc02g067760, expressed in ovary tissue but not in vegetative tissues. Solyc03g007780 was expressed in developing ovaries and anthers. Solyc02g067760 mRNA was detected in whole-flower tissues. The promoters of Solyc03g007780 (Psol80) and Solyc02g067760 (Psol60) predominantly induced the expression of genes in the ovule, placenta, endocarp and pollen and in whole-flower tissues, respectively. Psol80/60-SlIAA9i lines, created for SlIAA9-RNA interference controlled by two promoters, successfully formed parthenocarpic fruits without pleiotropic effects in vegetative tissues. Downregulation of SlIAA9, responsible for parthenocarpic fruit formation, was observed in ovules rather than ovaries in the Psol80/60-SlIAA9i lines. Although the weight of parthenocarpic fruits of the Psol80/60-SlIAA9i lines was lower than the weight of pollinated fruits of the wild type (WT), the parthenocarpic fruits presented redder and more saturated colors and higher levels of total soluble solids and titratable acidity than the WT fruits.

virus (P35S) are widely used 21 . However, when ubiquitous promoters are used for the expression of pleiotropic genes such as SlIAA9 or SlDELLA, they also cause unwanted phenotypes 18,22 . To overcome the disadvantages of these ubiquitous promoters, in recent years, many studies have reported the use of tissue-specific promoters in plant parts such as the seeds [23][24][25] , trichomes 26,27 , tubers 28 , roots 29,30 , pollen 31 , and petals 32 . However, research on promoters that drive the spatial and temporal expression of genes in reproductive tissues, especially in the ovary, are still rare.
Some researchers have previously reported the development of parthenocarpic tomatoes using a tissue-specific promoter and auxin biosynthesis gene combination in a heterologous system 13,15 . In research on a transgene-harboring construct consisting of an iaaM gene (which is involved in the biosynthesis of indole-3-acetic acid) from Pseudomonas syringae pv. savastanoi and the placental/ovule-specific DefH9 promoter from Antirrhinum majus, parthenocarpy was exhibited in transgenic tomatoes.
The objective of this study was to engineer parthenocarpic tomato plants without pleiotropic effects on vegetative tissues. To do so, we developed new, flower-specific promoters from tomatoes, with a special focus on the ovary tissues. Our transgenic plant lines created with novel, specific promoters and a SlIAA9-RNA interference (RNAi) construct showed a vegetative phenotype similar to that of the wild type, and they also successfully formed parthenocarpic fruits.

Results and Discussion
expression of Solyc03g007780 and Solyc02g067760 in various tissues. The first objective of our study was to develop an ovary-specific promoter. To identify the tissue-specific genes expressed in ovary tissues but not in vegetative tissues, we used RNA-seq data from 27 tissues of the tomato dwarf cultivar 'Micro-Tom' provided in a previous report 33 . Among the 505 genes specifically detected in the ovary samples, twenty-nine genes were selected and identified by comparison with an EST library representation of the Sol Genomics Network. Solyc03g007780 and Solyc02g067760 were selected by quantitative reverse transcription (qRT)-PCR analysis for further study.
Solyc03g007780 was not expressed in vegetative tissues such as the leaf, stem, and root of 7-and 30-day-old plants, but it was expressed in the ovaries at various time points and in the stamens when they reached a length of 5-6 mm and 1 day before anthesis (DBA) (Fig. 1a). In developing ovaries and pollinated ovaries (referred to as fruits in Fig. 1), the expression of Solyc03g007780 was initiated in 5-6 mm buds, and it gradually increased until 2 days after anthesis (DAA) and decreased from 3 DAA to 6 DAA. A previous report revealed that Solyc03g007780, which was designated an ovule-secreted protein (OSP), was expressed in ovules at a high level but was rarely expressed in other fruit or vegetative tissues, with the exception of fruits entering the ripening stage, in Solanum pimpinellifolium 34 . However, these expression data showed little difference in comparison with our data (e.g., the stamen in our data and the fruits entering the ripening stage in their data). Values are means ± SDs of three biological replicates. Different letters denote significant differences according to Tukey's HSD test at p < 0.01.

Development of parthenocarpic transgenic tomato plants.
The downregulation or loss of function of the Aux/IAA transcription factor SlIAA9, a negative regulator of the auxin response, resulted in the development of parthenocarpic fruits 17,18 . However, due to the ubiquitous role of SlIAA9, the constitutive knockdown or knockout of this gene also results in abnormal vegetative phenotypes, such as fused leaves with simple lobes. To obtain parthenocarpic tomato plants without the pleiotropic effect on vegetative tissues, Psol80 or Psol60 was used to drive the ovary-specific transcription of the RNAi constructs for SlIAA9 knockdown ( Supplementary  Fig. S1). To prevent off-target effects on other Aux/IAA genes, identical SlIAA9 mRNA region that verified targeting accuracy in a previous report 18 was selected through further confirmation by the BLAST algorithm of the Sol Genomics Network. Through Agrobacterium-mediated transformation into 'Micro-Tom' cultivars, three independent RNAi lines (lines 4-17, 4-23, and 4-25) for SlIAA9 downregulation by Psol80 (Psol80-SlIAA9i) and two lines (lines 6-10 and 6-18) for SlIAA9 downregulation by Psol60 (Psol60-SlIAA9i) were obtained.
First, we evaluated parthenocarpic fruit formation after emasculation. In our experiment, growing ovary followed by color change from green to reddish was considered as a fruit. Both the Psol80-SlIAA9i and Psol60-SlIAA9i lines successfully formed parthenocarpic fruits, similar to the iaa9-3 line 17 , SlIAA9 mutant with malfunction of SlIAA9 protein, used as a control line with parthenocarpy (Fig. 3). On the other hand, most emasculated flowers of the WT withered and fell within 2 weeks without developing into fruits. Only very few unpollinated WT ovaries (approximately 2% of the unpollinated ovaries) developed into abnormal fruits with an orange color, spongy pericarp, and weight under 0.3 g. The percentage of parthenocarpic fruit formation was 63% in the iaa9-3 mutant and ranged from 62 to 68% in the Psol80-SlIAA9i lines (4-17, 23, and 25) ( Table 1). The Psol60-SlIAA9i lines showed relatively low parthenocarpic efficiency, measuring 40% (6-10) and 44% (6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18), compared with the efficiency of other tested lines. When two flowers per flower truss were emasculated, the weight of parthenocarpic fruits from the iaa9-3 mutant was similar to that of fruits from pollinated ovaries of the WT, whereas the weight of the parthenocarpic fruits of the Psol80-SlIAA9i lines was up to 45% of that of the pollinated fruits of the WT. Furthermore, the parthenocarpic fruits of the Psol60-SlIAA9i lines exhibited a much lower weight than the pollinated fruits of the WT (up to approximately 20%).
Previous researchers revealed that the pollen and ovules of SlIAA9-downregulated lines were fertile, although a small percentage of pollinated flowers developed into seeded fruits (35%) when an SlIAA9-downregulated line was used as the female recipient 18 . During self-pollination of the iaa9-3 mutant, we also observed that some fruits from pollinated ovaries did not have any seeds despite several rounds of artificial pollination. To determine whether parthenocarpic fruit set in Psol80/60-SlIAA9i lines also has an effect on seeded fruit formation, we measured the efficiency of fruit www.nature.com/scientificreports www.nature.com/scientificreports/ formation and fruit weight using six self-pollinated flowers per plant. While most of the pollinated ovaries of the WT (95%) developed into seeded fruits, i.e., fruits resulting from pollination, the iaa9-3 mutant produced a small number of seeded fruits (67%) with fewer seeds than the pollinated fruits of the WT (Table 1). However, the Psol80/60-SlIAA9i lines showed efficiencies of seeded fruit formation and seed number per fruit that were similar to those of the WT. These results indicated that the Psol80-SlIAA9i and Psol60-SlIAA9i lines showed facultative parthenocarpy, in contrast to the iaa9-3 mutant. Previous researchers reported that the early pollinated fruits of SlIAA9-downregulated lines were larger than those of the WT 39 . However, the weight of pollinated mature fruits from SlIAA9-inhibited lines (Psol80/60-SlIAA9i lines and iaa9-3 mutant) was similar to that of pollinated fruits of the WT (Table 1). www.nature.com/scientificreports www.nature.com/scientificreports/ engineered parthenocarpic tomato plants exhibit a Wt-like vegetative phenotype. In a previous report, the inhibition of SlIAA9 also caused altered vegetative tissue morphology, such as fused leaves, long internodes and increased lateral shoot numbers, corresponding to ubiquitous expression of SlIAA9 17,18 . In this study, while the iaa9-3 mutant showed an abnormal plant form with a sparse leaf arrangement, the Psol80/60-SlIAA9i lines (4-17, 4-23, 4-25, 6-10, and 6-18) presented plant forms similar to those of the WT (Fig. 4a). To present the difference in plant form numerically, we counted the number of compound leaves and flower trusses on 50-day-old plants. All Psol80-SlIAA9i and Psol60-SlIAA9i lines formed similar numbers of compound leaves and flower trusses compared to those of WT (Fig. 4b). However, the iaa9-3 mutant formed significantly fewer flower trusses and compound leaves. In addition, the WT and all Psol80/60-SlIAA9i lines exhibited pinnately compound leaves with lobed leaflets, whereas the iaa9-3 mutant formed fused leaves with simple lobes (Fig. 4c). Because the downregulated level of SlIAA9 in leaves showed a positive correlation with the severity of altered leaf morphology 18 , we also examined the mRNA level of SlIAA9 in leaves to explain the normal leaf morphology of the Psol80/60-SlIAA9i lines. Consistent with the WT-like leaf phenotype, the Psol80/60-SlIAA9i lines showed SlIAA9 mRNA levels similar to those of the WT, in contrast to the iaa9-3 mutant, which presented no intact SlIAA9 mRNA in leaves (Fig. 4d). Furthermore, Solyc03g007780 and Solyc02g067760 expression was also not detected in the expanding leaves of the corresponding transgenic lines ( Supplementary Fig. S3). These results suggested that the mRNA level of SlIAA9 was not affected by SlIAA9i in leaves of the Psol80/60-SlIAA9i lines.
Tissue specific downregulation of SlIAA9 is responsible for parthenocarpic fruit formation in Psol80/60-SlIAA9i lines. Knockdown of SlIAA9 mRNA mimics increased auxin followed by subsequent parthenocarpic fruit formation 16,18 . In unpollinated ovaries, SlIAA9 mRNA was more strongly distributed in the ovule and placenta than in other ovary tissues 39 , implying that the transcript level of SlIAA9 in those tissues is important for parthenocarpic fruit set by SlIAA9 inhibition. Consistent with this interpretation, increased IAA in the placental/ovule tissue was sufficient for induction of parthenocarpic fruit set 14,15 . In the present study, Psol80 and Psol60 also induce the expression of the genes in ovules and placenta (Fig. 2b). Therefore, we assessed the mRNA level of SlIAA9 in ovules of Psol80/60-SlIAA9i lines. The level of SlIAA9 mRNA was significantly or slightly lower in the ovules of the Psol80-SlIAA9i lines and Psol60-SlIAA9i lines, respectively, than in the ovules www.nature.com/scientificreports www.nature.com/scientificreports/ of the unpollinated or pollinated WT (Fig. 5a). This result suggested that the parthenocarpic fruit formation was induced by downregulation of SlIAA9 mRNA -in the Psol80/60-SlIAA9i lines.
In the assay of SlIAA9 transcript levels in ovaries, the transcript level of SlIAA9 in unpollinated ovaries of the WT continuously decreased between 0 and 6 DAA without ovary growth (Fig. 5b). However, the level dramatically increased 2 DAA and remained at that level with ovary development into fruit by pollination or 2,4-D treatment. The level of SlIAA9 mRNA in developing ovaries of the Psol80/60-SlIAA9i lines was not increased compared with that of unpollinated ovaries of the WT between 0 and 4 DAA, but it rapidly increased 6 DAA in the Psol80-SlIAA9i lines or 12 DAA in the Psol60-SlIAA9i lines.
Surprisingly, an obvious decrease in SlIAA9 mRNA level by SlIAA9 RNAi in ovaries of the Psol80/60-SlIAA9i lines was not observed. To describe the insufficient downregulation of SlIAA9 in ovaries, the activity of Psol80/60 had to be considered. As shown in Fig. 2b and Supplementary Fig. S2b, Psol80 induced the transcription of the gene in the ovule and adjacent tissues, which occupied a small portion of the entire ovary (containing the columella, ovary wall, placenta, and ovule), resulting in low expression in the entire ovary. Psol60 also conferred much lower expression to the controlled gene compared with P35S. These results implied that the distribution or the amount of SlIAA9 dsRNA expressed by Psol80/60 was not sufficient to cause a remarkable decrease in SlIAA9 mRNA level in the entire developing ovary. Meanwhile, despite the development of ovaries into fruits in the Psol80/60-SlIAA9i lines (Fig. 3A), the increase in SlIAA9 mRNA level was delayed until 6 DAA in the Psol80-SlIAA9i lines and 12 DAA in the Psol60-SlIAA9i lines. These expression patterns showed an inverse correlation with the expression pattern of Solyc03g007780 in the pollinated and unpollinated ovaries of WT and Solyc02g067760 in the unpollinated ovaries of WT, reflecting the activity of each promoter (Fig. 1, Supplementary  Fig. S4). Solyc03g007780 was expressed considerably from 0 DAA to 4 DAA, and at the same time, the SlIAA9 mRNA level was not increased. However, at the time point when the expression of Solyc03g007780 was extremely reduced (6 DAA), the transcripts of SlIAA9 mRNA increased. The same expression pattern was observed in ovaries 12 DAA in the Psol60-SlIAA9i lines. These results suggest that the increase of SlIAA9 mRNA by ovary development was partially suppressed during RNAi constructs for SlIAA9 knockdown expressed by two promoters.
According to a previous report, the percentage of parthenocarpy in SlIAA9-downregulated tomato lines was negatively correlated with the level of SlIAA9 mRNA 18 . In our research, we also observed that parthenocarpic lines with lower levels of SlIAA9 mRNA in ovules showed a higher efficiency of parthenocarpy (Table 1, Fig. 5a). In addition, we found a negative correlation between the transcript level of SlIAA9 in ovules and the weight of the mature fruits; i.e., strongly downregulated lines showed larger fruit formation (Table 1, Fig. 5a,).
fruit quality and yield of parthenocarpic fruits in Psol80/60-SlIAA9i lines. We also examined the fruit quality and yield of parthenocarpic fruits of the SlIAA9-inhibited lines. There was no significant difference in pH between the parthenocarpic fruits of SlIAA9-inhibited lines and the pollinated fruits of the WT. However, in the titratable acidity (TA) assay, parthenocarpic fruits showed increased TA compared with pollinated fruits of the WT (Table 2). While the parthenocarpic fruits of the Psol80-SlIAA9i lines and iaa9-3 mutant presented slightly increased TA, those of the Psol60-SlIAA9i lines had much higher TA. Although the value of TA is mainly www.nature.com/scientificreports www.nature.com/scientificreports/ determined by concentrations of citric acid, malic acid, and phosphate in tomato fruits, phosphate concentrations showed no difference with between cultivars with various TA 40 . Therefore, the higher TA suggests that the parthenocarpic fruits had high acid concentrations compared with that in the pollinated fruits of the WT.
Parthenocarpic tomato fruits induced by increased IAA contents in the ovules have similar fruit weights, slightly increased levels of total soluble solids (TSS) compared with those of pollinated fruits of the WT 13 . The parthenocarpic fruits of iaa9-3 mutant also showed higher levels of total soluble solids (TSS) than the fruits of the WT (Table 2). In particular, low-weight parthenocarpic fruits from the Psol80/60-SlIAA9i lines exhibited much higher TSS levels than pollinated fruits of the WT. Generally, the Psol80/60-SlIAA9i lines formed lower-weight parthenocarpic fruits than the iaa-3 mutant line, and the fruits of these lines had higher TA and TSS values. This result also suggested that low-weight parthenocarpic fruits had higher TA and TSS values.
In the evaluation of fruit color, the parthenocarpic fruits of the iaa9-3 mutant and Psol80/60-SlIAA9i lines presented higher chroma and hue values, indicating more color saturation and a redder color than observed for the pollinated WT fruits 41 (Table 3). The more saturated and redder color of parthenocarpic fruits in the Psol80/60-SlIAA9i lines also suggested that the fruits of these lines were induced by successful SlIAA9 inhibition, similar to the case of parthenocarpic fruits in the iaa9-3 mutant. In addition, since lycopene is the main carotenoid responsible for the red color of tomato fruits 42 , we also measured the lycopene and β-carotene contents in www.nature.com/scientificreports www.nature.com/scientificreports/ the parthenocarpic fruits of SlIAA9-inhibited lines (Table 2). However, there was no difference in β-carotene or lycopene content compared with the content in pollinated fruits from the WT. Additionally, the L* values, which express the lightness of fruits and have positive correlations with total carotenoid and lycopene contents 43 , did not show significant differences between the WT and SlIAA9-inhibited lines (Table 3). These results suggested that the saturated red color of parthenocarpic fruits in the SlIAA9-inhibited line was caused by other pigments, not a high lycopene content.
To compare the yield of parthenocarpic fruits of Psol80/60-SlIAA9i lines to that of pollinated fruits of the WT, all the flowers from the 4th flower truss were pollinated or emasculated for 4 weeks. The yield of the iaa9-3 mutant was up to 73% of that of the WT, whereas that of the Psol80-SlIAA9i and Psol60-SlIAA9i lines was up to 76% and 24%, respectively, despite these lines expressing a WT-like leaf number and morphology (Table 1, Fig. 4b,c). This change in yield is thought to be the result of relatively low parthenocarpic fruit set efficiency and fruit weight due to insufficient downregulation of SlIAA9 in the ovule.
The objective of this study was to make parthenocarpic tomato plant without abnormal vegetative phenotype. In this study, the promoters of Solyc03g007780 and Solyc02g067760 induced tissue specific expression of genes.   DNA and RNA extraction and cDNA synthesis. Genomic DNA and total RNA were extracted using a DNeasy or RNeasy plant mini kit (Qiagen) according to the manufacturer's instructions. Total RNA was treated with a DNA-free RNA kit (Zymo Research) to eliminate DNA contamination. First-strand cDNA was synthesized from 1 μg of total RNA using the PrimeScript II 1st strand cDNA Synthesis Kit (Takara-bio) and oligo dT primers. The synthesized cDNA was subjected to PCR to confirm genomic DNA-free cDNA with a set of SGN-U314153 (CAC) gene primers: forward, CCTCCGTTGTGATGTAACTGG, and reverse, ATTGGTGGAAAGTAACATCATCG 44 .

Vector construction and tomato transformation.
To obtain the promoter region and SlIAA9 fragments, the genomic DNA or cDNA from the ovaries was amplified in 50 μl of PCR mixture with 5 μl of 10x buffer, 5 μl of 2 mM dNTPs, 1 U of KOD Plus Neo (Toyobo), and 1 μl of 10 μM primers (Supplementary Table S1). The thermal cycling procedure consisted of 5 min at 98 °C, 28 cycles of amplification consisting of denaturation for 30 s at 98 °C, annealing for 30 s at 60 °C, and extension for 1 min at 68 °C, and a final extension at 68 °C for 3 min. For the promoter:GUS assay, promoter amplicons with HindIII (5′) and SalI (3′) restriction sites for Solyc03g007780 or with SalI (5′) and BamHI (3′) restriction sites for Solyc02g067760 were cloned into the pBI101-GUS vector. For the RNAi vector, an amplified promoter with Avr2 (5′) and Xho1 (3′) was inserted into the pBI-sense, antisense gateway vector (Inplanta Innovations) instead of P35S. Subsequently, 715 bp SlIAA9 amplicons were cloned into the pCR8/GW/TOPO vector and inserted into the pBI-sense, antisense gateway vector using Gateway LR Clonase (Thermo Fisher). The vector construct used in this study is shown in Supplementary Fig. S1.
Transgenic tomato production was conducted by Agrobacterium-mediated transformation according to a method previously described 45 . The transgenic lines were subjected to PCR using promoter and gene-specific primer sets and a Southern blot analysis to find the homozygous-independent lines.
Quantitative reverse transcription pcR. The expression levels of the genes were assessed by quantitative reverse transcription PCR (qRT-PCR) with a Dice Real-Time thermal cycler and SYBR Premix Ex Taq II following the manufacturer's protocol (Takara-bio). Diluted cDNA was used as a template in 25 μl of PCR-amplification reaction mixture containing 12.5 μl of SYBR Premix Ex Taq II (Takara-bio) and 1 μl of the 10 μM primer set. A dissociation curve analysis was also performed to confirm primer compatibility. The relative expression of the genes of interest was measured by the standard curve method with three biological replicates. SAND was used as a reference gene 44 . The primer set used for qRT-PCR is described in Supplementary Table S2. GUS and histological assay. Histochemical ß-glucuronidase (GUS) analysis was performed with 5-bromo-4-chloro-3-indolyl-b-D-glucuronide (X-Gluc) using a previously described method, with a slight modification: phosphate buffer (pH of 8.0) and 20% methanol 35,36,38 . The tissues were incubated in the GUS staining solution at room temperature for 16 h, followed by washing with 70% ethanol to eliminate chlorophyll. Paraffin sectioning was performed according to previously described methods 46 .  Table 3. Color index (L*, a*, chroma, and hue) of mature fruits. Each data point represents the mean ± standard deviation (n = 20 except n = 8 of 6 lines). Different letters denote significant differences between factors according to Tukey's HSD test at p < 0.01.
Scientific RepoRtS | (2019) 9:18871 | https://doi.org/10.1038/s41598-019-55400-7 www.nature.com/scientificreports www.nature.com/scientificreports/ fruit characteristic analysis. For fruit analysis, twenty flowers per plant were emasculated or pollinated, and six fruits per plant were maintained for the experiment. Forty-eight-day-old fruits were subjected to color and weight measurements and stored at −80 °C for further experiments. The color was rated on the middle surface of the fruits with a Minolta Color Reader CR-10 (Konicaminolta). The color parameters describing lightness indicate the range from black (0) to white (100) (L*) and the color direction on the red to green scale (a*) and on the yellow to blue scale (b*). The chroma (color saturation) was calculated as (a* 2 + b* 2 ) 1/2 . The hue (h, term used for the classification of red, yellow, blue, and green) was measured by taking the arctan(b/a).
Frozen fruits were ground into a fine power and used to measure the total soluble solids (TSS), pH, titratable acidity (TA), and β-carotene and lycopene contents. The TSS was measured using a PAL-J refractometer (Atago). The pH and TA were analyzed using 1 g of fruit powder homogenized into 10 ml of distilled water according to a previously described method 47 . Oxalic acid was used as a standard for the titrant calculation. The β-carotene and lycopene contents were measured and calculated according to a previously described method 48 . Supernatant from the extract of 300 mg of fruit powder with 3 mL of acetone/hexane (4:6, v/v) was tested for its absorption using a Beckman DU 640 UV/Vis spectrophotometer (Beckman Coulter). The β-carotene (C CAR ) and lycopene (C LYC ) concentrations were calculated using the following equations: