SUN Regulates Vegetative and Reproductive Organ Shape by Changing Cell Division Patterns 1[C][W][OA]

One of the major genes controlling the elongated fruit shape of tomato ( Solanum lycopersicum ) is SUN . In this study, we explored the roles of SUN in vegetative and reproductive development using near isogenic lines (NILs) that differ at the sun locus, and SUN overexpressors in both the wild species LA1589 ( Solanum pimpinellifolium ) and the cultivar Sun1642 background. Our results demonstrate that SUN controls tomato shape through redistribution of mass that is mediated by increased cell division in the longitudinal and decreased cell division in the transverse direction of the fruit. The expression of SUN is positively correlated with slender phenotypes in cotyledon, leaﬂet, and ﬂoral organs, an elongated ovary, and negatively correlated with seed weight. Overexpression of SUN leads to more extreme phenotypes than those shown in the NILs and include thinner leaf rachises and stems, twisted leaf rachises, increased serrations of the leaﬂets, and dramatically increased elongation at the proximal end of the ovary and fruit. In situ hybridizations of the NILs showed that SUN is expressed throughout the ovary and young fruit, particularly in the vascular tissues and placenta surface, and in the ovules and developing seed. The phenotypic effects resulting from high expression of SUN suggest that the gene is involved in several plant developmental processes.

Tomato (Solanum lycopersicum) accessions feature a variety of fruit shapes and sizes (Paran and van der Knaap, 2007). Genes controlling fruit morphology offer important insights into the patterning of the organ and mechanisms by which organ shape and size are realized. One of the major tomato fruit shape genes is SUN, which, when expressed at high levels in the fruit, leads to an elongated shape (Xiao et al., 2008). The mutation that led to the identification of SUN was a gene duplication event mediated by the retrotransposon, Rider. The duplicated gene was placed in a novel genome environment, leading to high expression in the fruit (Xiao et al., 2008;Jiang et al., 2009). When overexpressing SUN under the control of the cauliflower mosaic virus 35S promoter in tomato, the transgenic plants produce extremely elongated and often seedless fruits (Xiao et al., 2008). SUN encodes a protein belonging to the IQD family and is characterized by the conserved IQ67 motif that is involved in calmodulin binding Levy et al., 2005;Xiao et al., 2008). The function of this family of proteins is poorly understood. Overexpression of the Arabidopsis (Arabidopsis thaliana) gene AtIQD1 increases the production of the secondary metabolite glucosinolate , whereas the high expression of SUN leads to elongated fruit shape. However, the biochemical mechanisms by which these phenotypes are realized are unknown. Moreover, gene expression studies in tomato did not show dramatic differences in the tomato with or without SUN ). Yet, the parthenocarpic fruit development associated with SUN overexpression led us to hypothesize that SUN may be involved in the production of a hormone or secondary metabolite that affects the auxin pathway either directly or indirectly (Xiao et al., 2008).
It has been hypothesized more than 200 years ago that carpels are modified leaves (Goethe, 1970;Coen, 2001). The notion has gained strong support from studies on flower and fruit development, indicating that floral organs and leaves are partly interchangeable through modifying only a small set of regulatory genes (Honma and Goto, 2001;Pelaz et al., 2001;Alonso-Cantabrana et al., 2007;Østergaard, 2009). Several tomato genes known for their roles in fruit set and growth also exert effects on leaf morphology (Jones et al., 2002;Wang et al., 2005;de Jong et al., 2009b;Molesini et al., 2009). However, it is not clear whether, in addition to fruit shape, SUN also affects the phenotype of vegetative parts of tomato.
In this study, we evaluated the vegetative and reproductive phenotypes of the near isogenic lines (NILs) that differ at the sun locus as well as lines overexpressing SUN under the control of the 35S promoter in two genetic backgrounds: LA1589 (Solanum pimpinellifolium) and Sun1642 (tomato). Our results show that SUN has no significant effects on fruit weight, and that it regulates tomato fruit shape by rearrangement of fruit mass via an altered cell division pattern. In addition to fruit shape, SUN also controls floral organ and leaf morphology. High expression of SUN led to slender lateral organ shape, as well as thinner leaf rachis and stem. The involvement of SUN in multiple developmental processes suggests it may play a role in the basic programs of plant growth.

SUN Controls Fruit Shape by Rearranging Fruit Mass
NILs that differ at sun in both the LA1589 and the Sun1642 backgrounds showed a noticeable difference in fruit shape (Fig. 1). Sun1642ee and LA1589ee carry the gene duplication and produce elongated fruits, while Sun1642pp and LA1589pp are wild type at the locus and carry round fruits. Overexpression of SUN under the control of the 35S promoter in the roundfruited background (Sun1642ox and LA1589ox) led to extremely elongated fruit shape. The parthenocarpic fruits from the overexpressors exhibited unusual proximal end elongation and often a pronounced tip (Fig.  1). To determine whether differences in fruit shape were accompanied by differences in fruit weight, seed weight, and seed number, we evaluated these traits in the NILs in both field-and greenhouse-grown plants. These traits were not evaluated in the lines that express SUN under the control of the 35S promoter because of the often seedless nature of these fruit and the irregular fruit set. Fruit weight and seed number per fruit did not differ significantly between the NILs in both the LA1589 and Sun1642 backgrounds, despite environmental effects (note that the differences between experiments were significant). This result indicated that the difference in fruit shape controlled by SUN was neither due to changes in seed number nor changes in fruit weight. Seed weight on the other hand differed between the NILs but only consistently in the LA1589 background (Table I). This result suggested that SUN might control seed development in addition to fruit shape. However, the effect on seed weight could also be due to other genes at the locus that are differentially expressed, for example DEFL1 (Xiao et al., 2008). To evaluate whether SUN itself affected seed weight, transgenic lines that express SUN under its own promoter were evaluated. From six independent transgenic lines, three appeared to show a reduction in seed weight compared to control albeit that this was only significant in one transgenic line (Supplemental Table S1).
Since fruit weight was not affected by SUN, we hypothesized that shape was controlled by the rearrangement of fruit mass. To explore the tissue and cellular basis of the mass rearrangement, we evaluated the length, width, and thickness of the pericarp and septum, as well as cell number and cell size in these tissues. Fruits were collected from sun NILs in LA1589 background at 7 DPA. At this time point, the fruit shape index (length/width ratio) difference between the NILs is at the maximum level . A difference was found in septum morphology such that the septum of LA1589ee was much more elongated and narrower than LA1589pp ( Fig. 2A; Table II). In LA1589ee, more cells were found in the longitudinal direction of the septum and less cells in the transverse direction compared to LA1589pp. On the other hand, septum cell size was not significantly altered between the NILs ( Fig. 2A; Table II). With respect to the pericarp, the longitudinal perimeter and the number of cells in this dimension were increased in LA1589ee, whereas cell size was not. In the transverse direction, cell number along the perimeter was not significantly different, and the increased pericarp perimeter in LA1589pp was due to the enlarged mesocarp cell size (Table II). This finding was supported by the notion that while the pericarp was slightly thicker in LA1589pp than in LA1589ee, the number of cell layers was not altered and therefore cell size was altered to account for difference in thickness. It should be noted that at this stage in fruit, mesocarp cell size increases dramatically . We cannot rule out the possibility that the differences in cell size of the mesocarp cells in the transverse direction is due to slightly slower fruit development in LA1589ee compared to LA1589pp. Regardless, these results indicated that the largest change in fruit ontogeny arose from changes in cell proliferation, resulting in a significantly higher number of cells in the longitudinal direction and less cells in the transverse plane.
Since ovary shape is already slightly different at the time of anthesis (Xiao et al., 2008), we evaluated whether changes in cell morphology were detected at that time. The analyses were based on three ovaries sectioned longitudinally, through the center, and five ovaries sectioned transversely through the center for each genotype. The ovary length and width, and the septum length were significantly different between LA1589ee and LA1589pp (Supplemental Table S2). Although the differences in cell size and the number of cells along the longitudinal direction were not significant, we noticed a trend of more cells in the longitudinal direction of LA1589ee septum (Supplemental Table S2). Similar to 7-DPA fruit, distal and proximal end areas were also longer in LA1589ee compared to  LA1589pp ovaries at anthesis (Table II; Supplemental  Table S2). The septum cell number in the transverse direction was significantly higher in LA1589pp, but this was offset by reduction in cell size, leading to equal septum width in both NILs (Supplemental Table  S2). Nevertheless, this result indicated that reduced cell number in the transverse direction of LA1589ee septum was already observed in anthesis-stage ovaries.
To evaluate further whether changes in cell division occurred prior to or after pollination, we compared septum and pericarp morphology at anthesis and 7 DPA. In the longitudinal direction, septum cell number increased 63% from anthesis to 7 DPA in LA1589ee and only 28% in LA1589pp (Fig. 2B). Similarly, in the pericarp, cell number increased 49% and 12% in LA1589ee and LA1589pp, respectively. In the transverse direction, differences in cell number were observed as early as at anthesis, and transverse cell number increases in septum and pericarp after anthesis until 7 DPA were comparable in both NILs (Fig. 2B). This result suggested that while changes in cell division along the medio-lateral axis were already executed at the time of anthesis, changes in cell division along the apical-basal axis were executed shortly after pollination. In sum, the changes in the direction of cell division in septum and pericarp led to a greatly elongated fruit shape (Fig. 1).

Mutually Exclusive Expression Patterns of SUN and DEFL1
To investigate the tissue-specific expression of SUN, we performed RNA in situ hybridization in ovaries and young fruits of sun NILs in the LA1589 background. We also analyzed the expression pattern of DEFL1, a gene that is expressed in round fruit but not in elongated fruit due to the gene duplication event at the locus (Xiao et al., 2008). In addition to using DEFL1 as control probe for hybridization (we expected no hybridization of DEFL1 when SUN is expressed and vice versa) we wanted to test the hypothesis that the promoter of DEFL1 is driving expression of SUN, which would be consistent with previous findings (Xiao et al., 2008). To visualize the internal structure along different axes, hybridizations of SUN and DEFL1 were conducted using longitudinally cut sections as well as transverse sections. SUN was expressed in LA1589ee, but barely detected in LA1589pp (Fig. 3, A, C, and E). Conversely, DEFL1 transcript was only accumulating in LA1589pp (Fig. 3, B, D, and F). At 2 d before anthesis, SUN was highly expressed in vascular tissues and embryo sac of the ovule (Fig. 3A). At anthesis and 2 DPA at the onset of embryo development , SUN transcript was localized in the vascular tissues of the sepals and most parts of the ovary, including ovary wall, placenta surface, and seed (Fig. 3, C and E). The central region of the columella had relatively low signals except for where the vascular bundles are located (Fig. 3, A, C, and E). At 6 d after fertilization when the embryo is at the globular stage , high levels of SUN transcript were found in the pericarp and developing embryo of LA1589ee (Supplemental Fig. S1). The dark staining in septum and the inner layers of pericarp was due to the accumulation of starch grains (Supplemental Fig. S1, inserts). At all of these stages, the expression patterns of DEFL1 in LA1589pp ovary and fruit mirrored those of SUN in LA1589ee ( Fig. 3; Supplemental Fig. S1). The expression of SUN and DEFL1 in vegetative parts and floral organs was also analyzed by northern blot. At the time of anthesis, the highest levels of SUN transcript in flower were found in sepals, petals, and ovaries of LA1589ee and Sun1642ee (Fig. 4A). DEFL1, on the other hand, showed highest expression in the same organs of LA1589pp and Sun1642pp, and was barely detected in LA1589ee and Sun1642ee. Different vegetative tissues were collected from young seedlings at 7 d after germination. We found SUN was expressed the highest in hypocotyl and shoot apex of LA1589ee and Sun1642ee, while DEFL1was present in the same tissues from LA1589pp and Sun1642pp (Fig.  4B). The expression of SUN in other vegetative parts of LA1589ee and Sun1642ee was at lower levels. We also detected a low level of SUN expression in the floral organs and roots of LA1589pp and Sun1642pp (Fig. 4). The expression of SUN in these tissues was presumably derived from the ancestral gene on chromosome 10.
The RNA in situ hybridization indicated that the expression pattern of SUN was ubiquitous throughout the ovary around the anthesis stage and young developing fruit, and not, as might have been expected, expressed at higher levels in septum tissues, where the morphology was changed most dramatically in the fruit of the NILs. Consistent with the previous studies (Xiao et al., 2008, our analyses showed SUN and DEFL1 were expressed in similar tissue types and in a mutually exclusive manner. The latter result clearly demonstrated that the promoter of DEFL1 was driving SUN expression on chromosome 7.

SUN Affects Lateral Organ Shapes
To determine whether SUN affects the shape of other floral organs, we evaluated stamen, petal, and sepal shape in the NILs that differ at sun in both Sun1642 and LA1589 backgrounds. As shown in Table  III, sepals from LA1589ee and Sun1642ee were more elongated than those of LA1589pp and Sun1642pp. Stamen and petal shape was also altered but only significantly in the Sun1642 background (Table III). These results indicated that SUN controls floral organ shape in addition to fruit shape. Overexpression of SUN led to very elongated ovaries compared to the controls in both backgrounds ( Fig. 5; Supplemental  Fig. S2). Also, overexpression of SUN led to increased shape index of the other floral organs in both backgrounds (Table III). Moreover, in the LA1589 background the petals and stamens of SUN overexpressors were often fused with the ovary at the proximal end, suggesting a reduction in organ separation.
The effect of SUN on vegetative development was already noticeable shortly after seed germination (Fig.  6A). Lines that overexpress SUN showed dramatic effects on cotyledon and leaf shape (Fig. 6A). LA1589ee and Sun1642ee had slightly more elongated cotyledons than LA1589pp and Sun1642pp, and those of the SUN overexpressors were more slender and tapered at the distal end (Fig. 6, B and C; Table III). SUN did not affect leaflet number (Fig. 6, D and E), but higher levels of SUN expression led to altered leaflet margin characteristics (Fig. 6, B and C). In Sun1642 background, high levels of SUN expression were also associated with fusion between the rachis and the first pair of lateral leaflets, which were almost always found on the overexpressors and occasionally on Sun1642ee plants (Fig. 6D, inserts). We analyzed terminal leaflet shape based on the fifth true leaves on plants at 30 d after germination. The leaflet margin of the overexpressors was extremely serrated with an increased number of teeth and no proximal end indentation (Fig. 6, B and C; Table III; Supplemental Fig.  S3A). However, the tooth numbers of terminal leaflets from the NILs were similar. The difference in regard to leaf margin feature between the NILs was that LA1589ee and Sun1642ee had more pointed tips, as well as a reduced proximal indentation area (Fig. 6, B and C; Table  III). We also noticed that the shape indices of terminal leaflets from LA1589ee and Sun1642ee were higher and thus were more elongated than those of LA1589pp and Sun1642pp (Table III). Close-up views of vein structures showed SUN overexpressors had altered vein patterning with reduced angles between the secondary and tertiary vein and between the tertiary and quaternary veins (Supplemental Fig. S3B). It is conceivable that such changed vein structure was due to a less-expanded intervein area. In the overexpressors, more higher-order marginal veins ended freely in the lamina (Supplemental Fig. S3C), which could explain increased tooth number.
The leaves of the overexpressors were nearly always twisted in the rachis and petiole, often causing the leaflets to be positioned with the abaxial side facing upwards (Fig. 6, D and E). Cross sections of rachis taken between the first and second pairs of lateral leaflets showed that the rachis of overexpressors was much thinner (Fig. 7A). Although to a less extent, the leaf rachises in the LA1589ee and Sun1642ee were also twisted (Fig. 7, B and D). Closer inspections of the NILs revealed that the twisting rachis was associated with altered epidermal cell morphology. The abaxial epidermal cells tended to be less elongated and more disorganized in LA1589ee, Sun1642ee, and the SUN overexpressors compared to LA1589pp and Sun1642pp (Fig. 7, C and E). Disorderly arrangement of epidermal cells was also found in the adaxial side of the leaf rachises from LA1589ee and overexpressors in LA1589 background (Supplemental Fig. S4, A-C). The severe reduction in abaxial cell length could explain the downward curving of overexpressor leaf at an early developmental stage (Supplemental Fig. S4, D and E). The distal angle of leaflet was measured at the position of 10% above the tip by the Tomato Analyzer software. b The proximal indentation was evaluated by the indentation area relative to total fruit area.

Overexpression of SUN Leads to Altered Plant Architecture
We did not observe significant alterations in overall plant growth between the sun NILs in both backgrounds, albeit that LA1589ee and Sun1642ee were slightly shorter than LA1589pp and Sun1642pp (Fig. 8, A and B). The shorter stature of LA1589ee and Sun1642ee was not due to decreased internode length (Table III), but changed internode angles, leading to zigzag stems. Overexpression of SUN caused dramatic changes in the architecture of the whole plant. These plants displayed prostrate growth and twisted stems (Fig. 8, A and B).
As mentioned above, Sun1642ee and SUN overexpressing plants had thinner and twisted leaf rachises (Fig. 7A). We further compared the stem morphologies among the NILs and SUN overexpressors in the Sun1642  background. Cross sections of hypocotyls and stems at different positions were taken from 10-to 12-leaf-stage plants. We found altered stem anatomy in Sun1642ee and the overexpressors, especially the latter (Figs. 8C; Supplemental Fig. S5). The SUN overexpressing plants had reduced pith and more laterally expanded cortex cells compared to Sun1642pp (Fig. 8C). We also noticed that in the longitudinal direction, the cortex cells of Sun1642pp were arranged in several continuous strands, whereas those of the overexpressor were less elongated and disorganized (data not shown). A proposed helical growth model suggests this type of cortex cell morphology may underlie the twisting of the stem (Furutani et al., 2000;Ishida et al., 2007). Since the inner layer cells (cortex cells) have stronger isotropic expansion than the epidermal cells do, when there is a need to balance differences in longitudinal length, the outermost cell layer would tend to twist (Furutani et al., 2000;Ishida et al., 2007). In all, the stem phenotypes due to high SUN expression were similar to what we found in the leaf rachis and this could be explained by the helical growth model.

DISCUSSION
The locus sun was identified for its role in controlling tomato fruit shape (Van der Knaap and Tanksley, 2001;Xiao et al., 2008Xiao et al., , 2009. In this study, using the NILs that differ at sun and SUN overexpressing lines, we further revealed that SUN is involved in controlling tomato fruit shape through redistribution of mass, and also plays important roles in the growth of other floral organs and in the vegetative development of the plant.

SUN Exerts Its Effect on Fruit Shape through a Change in Cell Division Patterns
The wild species accession LA1589 produces round and spherical fruits. The fruit of the sun NILs harbor- ing the elongated-fruited allele has increased length and reduced width compared to the fruit from the NILs having the round-fruited allele, suggesting that SUN promotes the longitudinal growth and decreases transverse growth. Since fruit weight is not affected by SUN, it is likely that it controls fruit shape through redistribution of mass. This proposed action of SUN on redirecting the dimensional growth of the fruit is supported by the observation that more cells were found in the longitudinal direction along the septum and pericarp of the 7-DPA fruits of LA1589ee, and fewer cells in the transverse direction (Table II; Fig. 2). Thus, SUN seems to induce the longitudinal cell division and to repress cell division in the transverse direction. We hypothesize that this is the result of a change of cell division planes.

SUN Affects Multiple Development Processes Reminiscent of Alterations in Auxin Transport
Ectopic expression of SUN under 35S promoter leads to fruit set without seed development. Production of pathenocarpic fruits can be triggered by exogenous application of plant hormone auxin or GA (de Jong et al., 2009a). Auxin and GA act synergistically in fruit initiation (Serrani et al., 2008;Vriezen et al., 2008), but they have different effects on parthenocarpic fruit growth in terms of ovule and locular tissue development (Serrani et al., 2007;de Jong et al., 2009a). The fruits of SUN overexpressors resemble the parthenocarpic fruits induced by auxin that have gel tissue filled in the locular cavities and produce pseudoembryos (de Jong et al., 2009a). It has been shown that inhibition of polar auxin transport (but not high auxin concentration) changes the orientation of cell division in tobacco (Nicotiana tabacum) cells (Petrásek et al., 2002;Campanoni et al., 2003). Thus, it is possible that SUN alters the axiality of cell division in tomato fruit by affecting, directly or indirectly, auxin transport.
In addition to fruit and leaf shape indices, SUN affects morphology by increasing serration at the leaflet margins (Fig. 5, B and C). In Arabidopsis, defining and positioning of teeth requires two key processes: polar auxin transport by PIN1 that creates local auxin maxima at the convergence points to allow serration outgrowth (Hay et al., 2006;Scarpella et al., 2006) and the activity of CUC2 in the sinus, leading to regional growth repression and/or extension of tooth outgrowth (Nikovics et al., 2006;Kawamura et al., 2010). SUN overexpressing plants have more teeth on the leaflets (Fig. 5, B and C; Supplemental Fig. S3), which could be the result of increased number of auxin maxima sites at the margin. Scarpella et al. (2006) found that more convergence points in the epidermis can be promoted by exogenous auxin application. Therefore, it is possible that overexpressing SUN leads to high auxin levels in the marginal area of the leaflet. LA1589ee and Sun1642ee do not have more teeth on leaflets than LA1589pp and Sun1642pp, but reduce leaflet proximal end outgrowth and increase sinus indentation, which resemble the phenotypes caused by elevated activity of CUC2 and GOBLET in Arabidopsis and tomato, respectively (Nikovics et al., 2006;Berger et al., 2009). Auxin induces posttranscriptional repression of CUC2 by activating MIR164A (Nikovics et al., 2006;Bilsborough et al., 2011). Our observation suggests SUN may play a role in disturbing normal auxin distribution in leaf, for example between tips (auxin maxima) and sinuses, and further affecting the activity of genes involved in leaf margin growth. Fusions between rachis and lateral leaflets (Fig. 6D), and at the proximal end of flower organs are also suggestive of perturbed auxin transport. Koenig et al. (2009) propose that during secondary morphogenesis, reduced auxin level and auxin response in tomato leaf rachis are necessary for preventing blade outgrowth along the rachis. Arrested polar auxin transport by N-1-naphthylphthalamic acid in maize (Zea mays) shoot also causes marginless leaf bases (Scanlon, 2003). Twisted leaf rachis and stem are indications that SUN could also affect auxin transport or response.
It has also been proposed that auxin distribution may underlie twisted growth in plants (Ishida et al., 2007). For example, twisting and asymmetric organ development of Arabidopsis mutants, tornado1 (trn1), trn2, and twisted dwarf1, is accompanied by changed auxin transport patterns (Cnops et al., 2000;Bouchard et al., 2006;Cnops et al., 2006). Since SUN expression does not lead to differences in expression of auxin signaling and biosynthesis-related genes , and considering the fruit and vegetative phenotypes discussed above, it is conceivable that if SUN interacts with the auxin pathway, this is more likely via altering auxin transport than auxin signaling.

Plant Materials
NILs were developed from crosses between tomato (Solanum lycopersicum) var. Sun1642 and the wild species Solanum pimpinellifolium LA1589, to ensure a small introgression of approximately 60 kb (Xiao et al., 2008). Recombinants with small introgressions were identified during the positional cloning of the gene and facilitated by the high recombination rate around the locus. The resulting two sets of NILs were used for phenotypic analysis. The SUN overexpressors in LA1589 genetic background were described previously, and SUN overexpressors in Sun1642 background were obtained by transforming pEK69 into round-fruited NIL Sun1642pp as described previously (Xiao et al., 2008). Transgenic plants and LA1589 sun NILs were grown in greenhouse, whereas Sun1642 NILs were grown in greenhouse and fields in Wooster, OH.

Flower and Fruit Measurements
Fruit measurements were taken of eight to 10 fruits per plant. Ten or more measurements were taken for other morphological traits. To ensure sufficient seed set, open flowers were tagged and hand pollinated.
Anthesis-staged flowers were used for measurement of floral organ shapes. Different floral organs were dissected and images were taken under a dissection microscope (NILs and overexpressors in LA1589 background) or by scanner (NILs and overexpressors in Sun1642 background). The organ shapes were measured by ImageJ (http://rsb.info.nih. gov/ij/) on the images taken. Mature fruit shape was measured by TomatoAnalyzer (Brewer et al., 2006; http://www.oardc.ohio-state.edu/ vanderknaap/tomato_analyzer.htm). Seed numbers were counted per fruit, and 100 seeds were weighed for seed weight.

Leaf Measurements
For internode length, leaflet shape, and leaflet number, three measurements were conducted on each plant of a total of five plants for each genotype. Terminal leaflet shape index and distal angle were measured by TomatoAnalyzer (Brewer et al., 2006). The leaf shape index refers to the ratio of height to width, and the distal angle of leaflet was measured at the position of 10% above the tip. For cotyledon shape analysis of NILs and overexpressors, cotyledons were collected from two-to three-leaf-stage seedlings. For leaflet vein morphology, terminal leaflets from wild-type plants and overexpressors were fixed in Carnoy's fluid containing acetic acid, absolute ethanol, and chloroform (1:6:3 by volume). Images were taken by scanner.

Hypocotyl and Stem Sectioning
Hypocotyl and stem cross sections were taken from the middle of hypocotyls and between subsequent nodes of 12-leaf-stage NILs and overexpressors in Sun1642 background. Fresh-cut 300-mm-thick sections were made using the 4000 automatic oscillating tissue slicer (Electron Microscopy Sciences). Sections were stained with 0.5% toluidine blue for a few minutes and washed with water. Images were then taken using a dissecting microscope (Leica MZFLIII) equipped with the Spot RTKE 7.2 color mosaic camera.

Scanning Electron Microscopy
Leaf rachises of the sixth leaves and lateral shoots were collected from 4-week-old plants. Hand-pollinated fruits of the sun NILs in LA1589 background were harvested at 7 d after pollination. The fruits were sliced through the middle in the longitudinal or transverse direction. All samples were immediately infiltrated and fixed with 3% gluteraldehyde, 2% paraformaldehyde in 0.1 M potassium phosphate buffer pH 7.4 for 2 h at room temperature, and then overnight at 4°C. Samples were then washed four times with phosphate buffer and dehydrated through an ethanol series, critical point dried, mounted, and coated with platinum as described previously . Samples were viewed and images recorded with a Hitachi 3500N scanning electron microscope (Hitachi High Technologies America, Inc.) under high vacuum. Cell counting and measurements were done on the scanning electron microscopy micrographs using ImageJ software.

Northern Blot
Total RNA for various tomato tissues were extracted using Trizol reagent (Invitrogen Inc.) as previously described (Xiao et al., 2008). In all RNA preparations, tissues were quickly frozen in liquid nitrogen after harvest and then stored in freezer (280°C) until RNA extraction. Ten micrograms of total RNA per sample was separated in 1.2% agarose gel in 13 MOPS buffer containing formaldehyde, then transferred to Hybond N membrane (Amersham Biosciences). Hybridizations were performed at 42°C in formamidecontaining hybridization buffer sequentially with radiolabeled gene-specific probes as previously described (Xiao et al., 2008). eIF4a6 or actin were used as loading controls.

In Situ Hybridization
RNA in situ hybridization was performed using the protocol reported by Balasubramanian and Schneitz (2002) with minor modification. To generate RNA probes for in situ hybridization, we amplified linear templates for SUN and DEFL1 from cDNA using following primers: SUN, 5#-ATGGGAAA-GCGAAGAAAC-3# and 5#-AATAGCAGCTTGTTTGCGTT-3# and DEFL1, 5#-GGCACAATCCATTCGTTTCT-3# and 5#-GTCTAGTGCACCATAATTGC-3#. An antisense or sense RNA probe was created by adding a sequence containing the T7 promoter to the 5# of the reverse or forward primer, respectively. Probes were labeled by in vitro transcription with T7 polymerase using a digoxigenin (DIG) RNA labeling kit (Roche). Flowers at 9 and 4 d before anthesis were vacuum infiltrated and fixed with 4% paraformaldehyde in 13 phosphate-buffered saline pH 7.0. Samples were washed with 13 phosphate-buffered saline, dehydrated with ethanol series and histoclear, and finally infiltrated and embedded in paraffin (Polyscience). Ten-microgram sections were taken with a microtome (American Optical Spencer 820). Sections were deparaffinized and rehydrated before acetylation, and then dehydrated. The slides were hybridized overnight with DIG-labled RNA probes at 55°C, and washed with 0.23 saline-sodium citrate buffer and blocked with blocking solution (Boehringer). DIG-labeled RNA was detected by alkaline phosphatase-conjugated antibody (Anti-Digoxigenin-AP Fab fragments, Roche). The antibody (1:1,000) was applied to the slides, incubated at room temperature, and washed off, followed by color reaction with 4-nitroblue-tetrazolium-chloride/5-bromo-4-chloro-3-indolyl-phosphate solution (Roche). The slides were mounted and sealed. The images were taken using a microscope (Leica DM IRB) equipped with a digital camera (Q Imaging Retiga 2000).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Expressions of SUN and DEFL1 mRNA in 6-DPA fruits of sun NILs in the LA1589 background.
Supplemental Figure S2. Ovary morphology of SUN overexpressors in Sun1642 background.
Supplemental Figure S3. Vein patterns of terminal leaflets from wild type and SUN overexpressing plants in LA1589 background.
Supplemental Figure S4. Epidermal cell morphology of leaf rachises from sun NILs and overexpressors.
Supplemental Figure S5. Hypocotyl and stem morphology of 12-leaf-stage sun NILs and overexpressors in Sun1642 background.
Supplemental Table S1. Comparison of fruit components between pHX4 transgenic lines in LA1589 background.
Supplemental Table S2. Light microscopy analysis of anthesis ovaries from the LA1589 sun NILs.