Transgenic plants overexpressing yeast spermidine synthase (ySpdSyn) were generated as described previously (Nambeesan et al., 2010). The ySpdSyn gene (ScSpe3, systematic name: YPR069C) was amplified from a yeast genomic library and cloned in the sense orientation between a CaMV35S promoter and the 3’ end of a pea rbcS-E9 gene in pKYLX71 (Nambeesan et al., 2010). This construct was introduced into disarmed Agrobacterium tumefaciens LBA4404, and Agrobacterium strains harboring the chimeric constructs were used for transformation of tomato cv. Ohio 8245 cotyledons (Tieman et al., 1992). Successful plantlets obtained were transplanted into potted soil and grown in the greenhouse.

Fifteen independent transgenic plants expressing CaMV35S-ySpdSyn were generated. Two independent transgenic lines C13, C14 that exhibited floral morphological defects, impaired fruit and seed set, and sterility were propagated as vegetative cuttings and characterized alongside the isogenic wild-type (WT) Ohio 8245 control.

For phenotypic measurements, 27–36 flowers were harvested from WT and transgenic plants, and the number of sepals, petals, and stamens were counted. Any abnormality in the stamen morphology was noted. The flower size was measured diagonally across the flower. The flowers were dissected to observe the morphology of the gynoecium. Pollen morphology from dissected stamens was observed under a light microscope (Olympus, Center Valley, PA).

For pollen germination assay, open flowers were hydrated for 1 h in a Petri dish with wet Kim wipes to maintain a relative humidity close to 100%. The hydrated pollen from the flowers was spread on a glass slide with 30 µl germination medium [20 mM MES, 2% sucrose, 15% PEG 4000, 1 mM KNO₃, 3 mM Ca(NO₃)₂, 0.8 mM MgSO₄, and 1.6 mM H₃BO_3, pH 6.0]. The pollen grains were incubated for 3 h in the dark at 25°C under high relative humidity by placing the glass slide on a moist filter paper in a Petri dish. Germination was terminated by the addition of 5 µl of phenol solution. Pollen germination was scored using a light microscope (Olympus, Center Valley, PA). Pollen tubes that were twice the diameter of the pollen grain were scored as positive for successful pollination. Pollen grains were examined from at least five flowers with 100–200 pollen counted per flower. Digital images were captured using the SPOT software (Draper, Utah).

Mature leaf and tissue from various stages of flower development, including early buds (0.1–0.7 mm in length), late buds (0.8–1.5 cm in length), and fully open flowers were harvested. Three independent biological replicates at each developmental stage were collected and were frozen immediately in liquid nitrogen and stored at −80°C until analyzed. All samples were processed and quantified in triplicates for polyamine measurement as described previously (Nambeesan et al., 2010).

Mature leaf and flowers were collected at two stages of development, ∼135 early buds (0.1–0.7 mm in length) and 75 late buds (0.8–1.5 cm in length) from WT, C13, and C14 lines. Total RNA was extracted, and cDNA synthesis was performed following the manufacture’s recommendation (Promega, Madison, WI). Briefly, total RNA (1 μg) was treated with DNase and ImPromII reverse transcriptase to perform reverse transcription. The cDNA was diluted five times, and 1 μl of diluted cDNA was used for semiquantitative reverse transcription PCR (RT-PCR) in a 10-μl total reaction volume. For each primer pair, cycle number was determined for exponential amplification, and all subsequent reactions were performed for the predetermined number of cycles. Each PCR reaction typically included the following cycles: 95°C (3 min); [95°C (30 s); 55/56°C (30 s); 70°C (1 min)] 25 cycles; 72°C, 10 min. Tomato ACTIN gene was used for normalization. After PCR amplification, the products were separated on a 1.2% agarose gel, stained with 0.5 μg/ml ethidium bromide, and images were captured and processed using Image J software (National Institutes of Health, Bethesda, Maryland). Primers used for PCR amplification are listed in Supplemental Table 1.

Quantitative RT-PCR analysis was performed using real-time PCR (Stratagene Mx3005P, San Diego, CA). All primers (Supplemental Table 1) along with control ACTIN (SlACTIN) were validated for relative gene expression analysis. One microliter of the cDNA was utilized in a 15-μl reaction using the SYBR green-PCR master mix (Applied Biosystems, Foster City, CA). Melting curve analysis was performed after the PCR reaction to determine specificity of the PCR products. At least three biological replicates were analyzed for gene expression analyses. Statistical analyses were performed using one-way analysis of variance to compare across transgenic C13 and C14 lines and WT separately for a given developmental timepoint or treatment, using JMP Pro 12 (SAS Institute, Cary, NC, USA). Means were separated using Tukey’s honestly significant difference (HSD) test (α = 0.05).

Eleven independent events in transgenic tomato lines positive for the expression of CaMV35S‐ySpdSyn were evaluated in greenhouse. Five of them—C5, C11, C12, C13, and C14—were found severely impaired in fruit set. They varied from having no seeds to very few seeds and exhibited partial parthenocarpy. Other four independent transgenic lines—C1, C3, C4, and C15—exhibited normal fruit set, but their seed number/fruit varied being 18, 15, and 30% lower in C1, C3, and C4, respectively, except for C15 in which seed numbers were similar to the WT control ( Table 1). Variable phenotypes, including impaired fruit and seed set and seed germination among the independent transgenic plants, may have resulted from the influence of transgene integration site(s) (De Buck et al., 2013).

Expression of ySpdSyn and endogenous SlSpdSyn in tomato leaves of C13 and C14 T₀ lines is shown in Figure 1A. Both transgenic lines expressed ySpdSyn and SlSpdSyn, while, as expected, WT leaves expressed only the endogenous SlSpdSyn gene. The SlSpdSyn transcript levels remained unaltered in C13 and C14 lines compared to WT leaves, which indicated lack of endogenous gene silencing (Figure 1A). Quantification of different polyamines in these leaves indicated that Spd levels increased by 1.3- and 1.7-fold in C13 and C14 lines, respectively, as compared to the WT (Figure 1B ). However, the transgenic event caused a 2.1- and 1.5-fold decrease in Spm in C13 and C14, respectively, compared to WT plants. Leaves from C14 lines exhibited significant 1.4-fold increase in Put compared to WT leaves, whereas C13 leaves did not show any change in Put levels compared to WT leaves (Figure 1B).

The floral and fruit morphology in transgenic and WT plants was evaluated to determine the basis of impaired fruit and seed set in C13 and C14 lines. Flower size or floral organ number did not differ significantly among these lines compared to WT plants. All genotypes had five sepals per flower. However, the average petal number was significantly different in line C13 (6.6 petal/flower) as compared to WT and C14 which had 5.8 and 5.3 petals/flower, respectively (Figure 2D). Similarly, the average number of stamens in C13 (7 per flower) was significantly higher than the WT and C14 lines that had 5.8 and 5.4 stamens per flower, respectively (Figure 2D). However, C13 line had severe morphological defects, with ∼17% flowers having 8–10 stamens in comparison to 5.4 in the WT. Most flowers from C14 and C13, however, did not form a normal staminal cone ( Figure 2A–C ). Dissection of flowers from transgenic lines revealed that carpels from C13 (Figure 2F) and C14 flowers (Figure 2G) were significantly enlarged as compared to the WT flowers (Figure 2E). Occasionally, fusion of multiple gynoecia at the base was observed in C13 flowers. The length of the style was reduced in C13 and C14 flowers (Figures 2E–G) with the tip of the style close to the stigmatic surface curved in transgenic flowers (Figures 2E–G). Other phenotypes in transgenic flowers included delayed abscission of petals during fruit growth (Figure 2K), multiple fruit set in the same flower by fusion of multiple gynoecia at the base (Figures 2L, M), and altered fruit shape with multiple locules ( Figure 2N).

We next investigated pollen morphology and germination to determine the basis of differences in fruit set in transgenic lines. WT pollen appeared spherical and intact, and 96% of the pollen displayed a normal morphology ( Figure 2H ). However, ∼68% of pollen from C13 and 37% from C14 flowers exhibited a collapsed morphology ( Figures 2I, J ). In vitro pollen germination assays indicated reduced pollen germination in C13 and C14 lines compared with WT. Among the intact-appearing pollen, 79% of WT and only 17% of C13 and 40% of C14 pollen grains germinated (data not shown). None of the abnormal looking pollen grains germinated (data not shown). The reciprocal crosses using C13 and C14 pollen on WT stigma did not result in fruit set, but pollination of C13 and C14 using WT pollen resulted in 100 and 80% parthenocarpic fruit set, respectively (data not shown), Among the fruit that set seed in C14, only one seed was present. These results indicated that fruit development occurred without fertilization. Reduced seed set in C14 indicated partial female sterility.

We determined the levels of free polyamines (Put, Spd, and Spm) in fully opened flowers from WT and four CaMV-ySpdSyn lines to evaluate if altered PAs homeostasis was associated with the observed phenotypes ( Figure 3 ). The steady-state Spd level in the flowers of other transgenic lines, viz., C4, C15, and WT plants was similar; these genotypes had normal fruit and seed set ( Figure 3). The flowers of C13 and C14 lines exhibited parthenocarpic phenotype and had significantly lower Spd levels than the WT flowers (Figures 2 and 3 ). Notably, the Spd/Spm ratio in the flowers of C4 and C15 plants was 8.8 and 7.0, respectively, and only 2.9 and 3.1 in C13 and C14 flowers, respectively (Figure 3). Similarly, the Spd/Put ratio was significantly higher in C4 and C15 flowers compared to the WT flowers but significantly lower in C13 to C14 flowers. This pattern was also reflected in the Spd/TPA (Figure 3). Notably, the ratio of Put/TPA was significantly lower in C4 and C15 flowers and significantly higher in C13 and C14 flowers compared to WT flowers, respectively. Consistent patterns for Spm/TPA ratio were not observed for transgenic flowers from the four independent lines examined (Figure 3). Taken together, these results suggest that reduced Spd and Spd/Put ratios are associated with the flower phenotype and may therefore affect fruit and seed set in tomato.

Expression patterns of ySpdSyn and endogenous SlSpdSyn were compared at various stages of flower development including early and late bud stages and fully open flowers. The ySpdSyn transcripts were present at all the stages of flower development in the C13 and C14 lines and, as expected, not in the WT (Figure 4A). As anticipated, SlSpdSyn transcript was present at all the stages of flower development, indicating the lack of silencing of the endogenous or ySpdSyn transgene in the transgenic lines (Figure 4A).

We quantified the levels of Put, Spd, and Spm in early and late flower buds to determine if the observed changes in various PA levels described in fully opened WT flower were also present during the flower development in C13 and C14 genotypes that had abnormal fruit and seed set (Figure 4B). The Spd level was not significantly different at early and late bud stages in the C13 line but was significantly lower in the C14 line at both stages of bud development. Furthermore, Spm levels in the C13 and C14 genotypes were significantly lower than the WT at early and late bud stages of flower development, whereas TPA and Put levels were lower at the latter stage.

The Spd/Put ratio was significantly higher at early bud stage in the C13 and C14 lines but not at late bud stage as compared with the WT. Spd/Spm ratio was significantly higher in the early and late bud stage in C13, whereas it decreased at the late bud stage in C14 line ( Figure 4C ). A significant decrease in Spm/Put ratio in both the C13 and C14 genotypes compared to WT was apparent only at the early bud development stage ( Figure 4). Although the ratio of Spd/TPA was higher at the early bud stage, however, in general, Put/TPA and Spm/TPA ratios in both genotypes at the early and late bud stages of development were not different from the WT.

Flower initiation and development are known to be regulated by the hormone gibberellins (GAs) (Davis, 2009). Therefore, we surmised that PAs and GAs may interact to regulate floral programs. To understand such an association between GAs and PAs, we analyzed the expression of tomato DELLA (SlDELLA) and GA-biosynthesis genes ( Figure 5). DELLA expression decreased significantly at early stages of bud development in C13 and C14 lines being 1.8- and 1.4-fold lower in C13 and C14, respectively, as compared to WT ( Figure 5A).

The expression of GA20-oxidase1 (GA20ox1) decreased by 2.2-fold in the C13 line at the early bud stage, decreasing further by 5.6-fold in the late bud stage ( Figure 5B ). A similar pattern was obtained for flower bud development in C14 line. Expression of GA20-oxidase2 (GA20ox2) was not significantly different at the early bud stage of both the transgenic lines; however, at the late bud stage, GA20ox2 expression was elevated ∼2- and ∼2.5-fold in the C13 and C14 lines, respectively ( Figure 5C ). In contrast, GA3ox2 expression was significantly downregulated ∼4-fold in both C13 and C14 lines at the early bud stage with no significant differences at the late bud stage between the WT and two transgenic lines ( Figure 5D ).