Members of the tomato FRUITFULL MADS-box family regulate style abscission and fruit ripening

Summary The FRUITFULL genes regulate tomato fruit ripening in a redundant manner. They are functional diversely in fruit shape regulation through affecting cellular differentiation and expansion.


Introduction
Members of MADS-box transcription factor families, which are widely distributed among animal and plant taxa, contain a conserved domain of approximately 60 amino acids in the N-terminal region, named the MADS-box (M) domain. Plant MADS-box gene families have been studied in detail and are now associated with a wealth of genetic and evolutionary data (Ferrario et al., 2004). In addition to the M domain, most of the plant MADS-box proteins contain three other domains: the Intervening (I) domain, the Keratin (K) domain, and the C-terminal (C) domain (Kaufmann et al., 2005). The MADS-box domain is involved in DNA binding and dimerization, while the I domain is responsible for DNAbinding specificity during dimer formation and the K domain mediates protein-protein interactions. Compared with the other domains, the C domain shows less sequence conservation and it has been found to be essential for ternary complex formation and transcriptional activation (Davies et al., 1996;Liu et al., 2008). MADS-box genes also function in seed pigmentation and endothelium development (TRANSPARENT TESTA16) (Nesi et al., 2002), root development (AGL12 and AGL17) (Rounsley et al., 1995;Tapia-Lopez et al., 2008), and fruit formation (SHP1, SHP2, and FUL) (Ferrandiz et al., 2000;Gu et al., 1998;Liljegren et al., 2000).
Other than these Arabidopsis genes, another well studied MADS-box gene is MADS-RIN, which was identified after screening a series of tomato mutants (Vrebalov et al., 2002). Fruits from the loss-of-function rin (ripening inhibitor) mutant exhibit perturbed ripening, as shown by the absence of a respiratory climacteric and associated ethylene evolution, as well as reduced softening, carotenoid accumulation, and the production of flavour compounds. While rin fruits are sensitive to ethylene, their ripening is not induced by exogenous ethylene (Vrebalov et al., 2002). In strawberry, a non-climacteric fruit, silencing of a tomato MADS-RIN homologue, FaMADS9, also leads to the inhibition of normal development and ripening (Seymour et al., 2011). These results suggest that MADS-RIN genes play a conserved function in ripening regulation in both climacteric and non-climacteric fruits (Seymour et al., 2011).
The mechanism by which MADS-RIN proteins regulate fruit ripening has been studied and it was reported that the tomato MADS-RIN protein influences gene expression by binding to CArG box sequences in the promoter regions of many ripening-related genes (Martel et al., 2011;Fujisawa et al., 2012;Qin et al., 2012). MADS-domain proteins can form hetero/homo-dimers, or higher order complexes, through protein-protein interactions (Favaro et al., 2002;Shchennikova et al., 2004;de Folter and Angenent, 2006) and these associations are essential for their function (Leseberg et al., 2008). A yeast two-hybrid screen of a prey library constructed from red ripe tomato fruit, with MADS-RIN as the bait, identified several other MADS-box proteins, including TDR4/FUL1, MBP7/FUL2, TAGL1, TAG1, MBP21, and TDR5 as putative MADS-RIN protein interactors (Martel et al., 2011). Similar to MADS-RIN, one of these proteins, TAGL1, also functions in tomato fruit ripening regulation and TAGL1-RNAi transgenic lines showed defects in fruit ripening (Vrebalov et al., 2009). This result suggests that MADS-RIN forms complexes with other proteins to control fruit ripening.
The interaction partners of MADS-RIN, TDR4/FUL1 and MBP7/FUL2, are close homologues of the Arabidopsis protein FRUITFULL, which has been suggested to regulate the transcription of genes required for cellular differentiation during fruit development (Gu et al., 1998). In addition, VmTDR4, a FRUITFULL homologue from bilberry (Vaccinium myrtillus) was found to regulate anthocyanin biosynthesis during fruit ripening (Jaakola et al., 2010). These results suggest that FUL1 and FUL2 might similarly play important roles during tomato fruit development, a hypothesis that was tested in this current study through the functional analysis of transgenic plants. As reported by Bemer and by Burko, it was confirmed that FUL1 and FUL2 played redundant roles in tomato fruit pigmentation accumulation and that FUL2 was involved in tomato leaf development (Bemer et al., 2012;Burko et al., 2013). Furthermore, our data showed that FUL1 and FUL2 play distinct roles in regulating cellular differentiation and expansion. The over-expression of FUL2, but not FUL1, affected fruit shape, resulted in a thinner fruit pericarp, and changed the number of layers of cells in the stem. These results demonstrate that FUL MADS-box transcription factors have diverse functions in growth and developmental regulation in tomato.

Materials and methods
Plant material and growth conditions All tomato (Solanum lycopersicum cv. Ailsa Craig) plants, including transgenic lines and a mutant line homozygous for the rin mutation in the Ailsa Craig background, were grown in a glasshouse under natural daylight and with 60-75% relative humidity and ambient temperature (>20 °C). The tomato transgenic lines were advanced to the T 2 generation. Flowers were tagged at the full-bloom stage to synchronize developmental comparisons. For analysis, 1, 7, and 14 days post-anthesis (DPA) corresponded to ovaries of 1, 7, and 14 DPA, respectively. The fruit stages used were immature green (IG), mature green (MG), breaker (BR), yellow ripe (YR), and red ripe (RR), which were picked at approximate 28, 35, 38, 41, and 44 DPA, respectively.

Construct recombination and plant transformations
For the FUL1/FUL2 double-silencing construct, a 362 bp fragment of the FUL1-IK domain was amplified by FUL1 sequence-specific primers (amplified with the primers in Supplementary Table S1 at JXB online). Fragments were incorporated into the pDONR221 vectors using the Clonase BP reaction (Invitrogen). The LR reaction (Invitrogen) was performed subsequently to incorporate the fragments into the pHELLSGATE8 vector. For the FUL1 and FUL2 over-expression constructs, the coding regions of FUL1 and FUL2 were amplified and cloned into the XbaI and KpnI sites of the pMV2 vector (modified from pHELLSGATE2). The expressions of transgene were driven using the CaMV 35S promoter in both vector systems.
For the Pro FUL2 ::GUS construct, a genomic DNA (gDNA) sequence (from -2518 to -24 bp) upstream of the FUL2 coding sequence was amplified using sequence-specific primers. The primers were fused with the attB1 and attB2 sites for recombination. The resulting fragment was recombined into the pDONR221 vector using BP recombinase and the LR reaction was performed to incorporate the fragment into the pV3P vector (modified from pHELLS-GATE2) which contains the glucuronidase synthase (GUS) coding sequences. All of the recombinant constructs were transformed into the Agrobacterium strain C58 by electroporation, and subsequently transformed into tomato cotyledon explants.

Ethylene assay
To measure ethylene production, BR stage fruits of approximately the same size were harvested and kept in sealed containers for 4 h and left open for 20 h each day (16 d) at room temperature (each biological replicate contained three fruits). Ethylene was measured in the headspace of the sealed containers by sampling with a syringe. Measurements were performed as described by Vrebalov et al. (2009). All samples involved three technical and three biological replicates.
Gene expression analysis RNA from various tissues of wild-type and transgenic plants were isolated using TRIzol ® 117 reagent (Invitrogen, USA). For cDNA synthesis, 3 µg RNA was used with M-MLV reverse transcriptase (Toyobo, Japan) according to the manufacturer's instructions. The cDNA concentrations were normalized to actin expression levels for RT-PCR analysis. Primers of the ripeningrelated genes are listed in Supplementary Table S1 at JXB online. The actin gene was used as an internal control for quantitative real-time PCR (qPCR), which was performed using the Power SYBR Premix Ex Taq kit and the TaKaRa two-step method (TaKaRa, Japan). PCR products were quantified using the Roche LightCycler 480 Real-Time PCR Detection System and the SYBR Green I Master Kit (Roche, Switzerland). The wild-type and transgenic plants were represented by three biological replicates for each sample.

Paraffin section
Light microscopic observation of paraffin sections was used to measure the number of cell layers and cell sizes in stems and pericarp of both wild-type and transgenic plants. For the fruit material, 14 DPA and BR stage fruits were harvested and at least nine pericarp sections were isolated. For the stem samples, the 3rd, 13th, and 17th internodes were selected and at least nine sections were measured and harvested for analysis. Paraffin sections were prepared as described by Yang et al. (2011) and the number of cell layers was counted manually. Photomicrographs were taken using an Olympus microscope.

GUS staining
Slices of fruits and stems from the transgenic lines transformed with the native promoter of FUL2-driving expression of the GUS gene (Pro FUL2 ::GUS) were stained with a GUS staining solution (100 mM sodium phosphate buffer) to evaluate GUS activity. Staining was allowed to proceed for 5 h at 37 °C in the dark and then washed with a graded ethanol series at room temperature for decolorization and observation by light microscopy (OLYMPUS SZX12).

FUL1 and FUL2 show distinct expression patterns in tomato tissues
Previous studies indicate that expression of MADS-RIN is induced coincident with ripening and peaks at the red ripe stage (Vrebalov et al., 2002). For comparative purposes, realtime quantitative PCR (qPCR) was performed to investigate the expression levels of FUL1, FUL2, and RIN in different tissues and fruit developmental stages of the tomato cultivar Ailsa Craig (Fig. 1). The results showed that the expression of FUL1 was similar to RIN, and was very low in roots, stems, leaves, and in the early stages of fruit development, but increased rapidly at the breaker (BR) stage, which marked the onset of ripening. This suggested that FUL1 might have a similar function to RIN and regulate fruit ripening. On the other hand, the expression of FUL2 was high in flowers and throughout fruit development, as well as in stems and leaves. Indeed, FUL2 was expressed in all tissues tested except roots, indicating that it functions in both reproductive and vegetative organs. It was noted that Bemer and colleagues reported similar expression patterns for FUL1 and FUL2 (Bemer et al., 2012). Taken together, the contrasting spatial and temporal expression patterns of FUL1 and FUL2 indicated that FUL1 and FUL2 may have distinct functions in the regulation of tomato plant growth.

Over-expression of tomato FUL2 inhibits the abscission of styles from fruit leading to a 'pointed tip' phenotype
To gain an insight into the functions of FUL1 and FUL2, vectors were constructed to mediate their suppression (RNAi) or over-expression and transformed into tomato (cv. Ailsa Craig). At least 15 independent transgenic lines were obtained corresponding to each vector. Nine of the 15 FUL2 over-expressing transgenic lines showed an obvious alteration in fruit shape with a prominent pointed tip at the blossom end of the fruit ( Fig. 2A), of which three lines (FUL2-OE-3, 10, and 13) were selected for transcriptional analysis. Expression of FUL2 in fruit of these lines was 7-28-fold higher than that in the wild type (Fig. 2B). The developmental onset of the pointed tip phenotype was assessed using stereomicroscopy and at 1 DPA, an early stage of fruit development, styles and ovaries of the FUL2-OE and the wild type showed no visible differences. Subsequently, at 7 DPA, the style of the wild type typically wilted from the tip to the base and abscised while the FUL2-OE lines showed a similar abscission phenotype, but with a visibly enlarged base. However, by 14 DPA, rather than showing abscission, as is typical of the wild type, the style of FUL2-OE was still connected to the ovary and the visibly enlarged basal region developed into a pointed tip (Fig. 2C), suggesting that this phenotype of FUL2-OE fruit is related to an abnormally persistent style.
To evaluate the abnormal style development further, a series of paraffin sections of ovaries at different developmental stages was prepared for microscopic observation. While the intact style typically abscised from the wild-type fruit, the style of FUL2-OE remained attached to the swollen ovary (young fruit) at 14 DPA with no obvious abscission zone (AZ) formation, and normal cell development was observed in the pointed tip (Fig. 2D). These results suggested that FUL2 may regulate style abscission by influencing ovary cell development, rather than AZ formation. In order to confirm that the abnormal development was the principal cause of the pointed tip phenotype, the growth was compared of FUL2-OE fruits from which the style had been removed following pollination and fertilization or had been left in place. Two weeks after pollination, no pointed tip was observed on the style-removed FUL2-OE fruits while, as expected, the non-treated FUL2-OE had the pointed tip phenotype (Fig. 2E). These results indicated that FUL2 affected style abscission by regulating ovary development, resulting in a pointed tip phenotype.
To obtain further support for this conclusion, the expression of FUL2 during fruit set was investigated using the GUS reporter gene driven by the presumed FUL2 promoter in transgenic tomato plants (Pro FUL2 ::GUS). GUS staining of 1 DPA Pro FUL2 ::GUS flowers showed that the FUL2 promoter had high expression activity in sepals and the stigma. Within the style, GUS activity was detected in the upper portion but not the base (Fig. 2F). This suggested that FUL2 negatively regulates the abscission of the pistil from tomato fruit at an early developmental stage and that reducing its expression in the pistil after pollination and fertilization is necessary to promote pistil abscission.

Over-expression of FUL2 suppresses tomato shoot growth
In addition to the fruit pointed tip phenotype, it was observed that the shoots of the FUL2-OE lines had smaller leaves and thinner stems than those of wild-type plants (Fig. 3A). In order to understand better how FUL2 influences main stem growth, the diameters of FUL2-OE main stems at different internodes were measured. Young stem diameters (the 17th internode) showed no obvious difference between FUL2-OE and wild-type plants, but stems of FUL2-OE plants were substantially narrower than those of the wild type in older regions. This difference increased with age and was approximately 5 mm at the 3rd internode (Fig. 3B). Moreover, the average cell size in the stem at the 3rd internode of FUL2-OE plants was smaller (Fig. 3C), indicating that FUL2 may block cell expansion in the stem during shoot growth.
In order to elucidate further the regulation of cell expansion in the stem by FUL2, GUS expression was assessed in transverse sections of the 17th stem internodes of Pro FUL2 ::GUS transgenic lines. GUS staining revealed that the FUL2 gene showed no obvious expression in the pith, but high expression in the epidermis, cambium, and vascular tissues (Fig. 3D). The cambium is a layer of actively dividing cells and this cell division determines stem diameter. These results therefore support the idea that FUL2 regulates stem growth by repressing cell division in the cambium and, consequently, the number of cell layers. Taken together, our data indicate that FUL2 controls shoot growth by regulating both cell expansion and cell division.

Over-expression of FUL2 alters tomato pericarp structure
Shelf life is an important agronomic trait for many species of fleshy fruits, and one of the notable characteristics of the tomato rin mutant is substantially extending fruit shelf life (Giovannoni, 2007). To determine whether FUL2 potentially interacts with RIN to regulate tomato fruit shelf life, a storage experiment was performed with FUL2-OE fruits. Both wild-type and FUL2-OE fruits with peduncles were harvested at the BR stage and stored at room temperature for 45 d. As expected, FUL2-OE fruits dehydrated much more slowly than those of the wild type (Fig. 4A). A previous study suggested that silencing FUL1/FUL2 expression in MicroTom fruit influences the expression of genes related to cuticle biosynthesis and post-harvest water loss, but does not affect pericarp development (Bemer et al., 2012). Our results further suggest that FUL2, but not FUL1, functions in modulating post-harvest water loss.
In addition to the slower post-harvest dehydration, FUL2-OE fruits showed the thinner pericarp. The smaller size and the fewer number of pericarp cells were observed by using paraffin wax in the FUL2-OE pericarp (Fig. 4B, C, D). Moreover, this indicated that over-expressing FUL2 inhibited pericarp cell expansion, possibly through a similar mechanism to that in the stem. In accordance with the high expression of FUL2 in the stem, it was also highly expressed in parenchyma cells of the fruit pericarp (Fig. 4F). These results suggested that FUL2 functions in cell division and expansion in both vegetative and reproductive organs. Plantcell-wall-modifying proteins known as expansins have been reported to play roles in fruit softening and organ abscission, as well as affecting organ size and morphology by regulating cell growth (Brummell et al., 1999;Cho and Cosgrove, 2000). Given the reduced cell size of the FUL2-OE fruit, it was investigated whether the expression of the SlEXP1, an expansin gene that is abundantly and specifically expressed during fruit ripening (Brummell et al., 1999), was influenced by FUL2. It was determined that, indeed, the transcript levels of SlEXP1 were significantly reduced in FUL2-OE fruit at the mature green (MG) and BR stages (Fig. 4E).

Double silencing of FUL1 and FUL2 suppressed tomato fruit ripening
Since FUL1 and FUL2 share a high degree of sequence similarity, and a high expression level in the pericarp during ripening, it was hypothesized that they might have similar functions. To test this, a sequence domain that is conserved between the FUL1 and FUL2 proteins was used as the basis to develop a strategy to doubly silence FUL1/FUL2 by RNAi. Screening of the resulting transgenic lines indicated that this approach effectively suppressed the expression of both FUL1 and FUL2 (Fig. 5A).
As previously described by Bemer et al. (2012) the FUL1/FUL2 doubly silenced transgenic lines showed bright yellow fruit during ripening (Fig. 5C). However, unlike the previous report, it was observed that fruits from the FUL1/FUL2 doubly silenced lines showed blocked ethylene production compared with the wild type, which had a typical peak (6 ml kg -1 h -1 ) at BR+3 d. The pattern of ethylene biosynthesis by the transgenic fruits was similar to that of the rin mutant, which produced <1 ml kg -1 h -1 (Fig. 5B). The discrepancy between our results and those of Bemer et al. (2012) may be explained by differences in genotype, since Ailsa Craig was used here, while Bemer et al. (2012) used the dwarf mutant MicroTom as experimental material.
The climacteric burst of ethylene production in tomato fruit is largely driven by the ethylene biosynthetic genes ACO1, ACS2, and ACS4 (Barry and Giovannoni, 2007), and their transcription is inhibited by the rin mutation (Martel et al., 2011). The transcript levels of these genes were evaluated in the FUL1/FUL2 doubly silenced fruit to establish whether this might explain the reduced ethylene production. Compared with wild-type fruit, the transcript abundance of ACS2 in the transgenic fruit was reduced by 10-fold and 5-fold at the BR and YR stages, respectively, and ACO1 showed a similar reduction (Fig. 5D). By contrast, transcript levels of ACS4 showed no significant suppression at the BR stage, but a sharp decrease at the YR stage. Recently, Shima and colleagues also confirmed that both FUL1 and FUL2 can bind to the CArG-box sites of the ACS2 promoter in a RIN-dependent manner (Shima et al., 2013). These results indicate that FUL proteins regulate ethylene biosynthesis through enhancing the expression of ACS2 and ACO1 during tomato fruit ripening.

Discussion
FUL1 and FUL2 regulate tomato fruit ripening in a functionally redundant manner FUL1 was first identified in tomato as a floral identity gene (Pnueli et al., 1991), but the expression pattern of FUL1 suggests a possible role in fruit development and ripening (Busi et al., 2003;Eriksson et al., 2004). Moreover, it has previously been reported that MADS-RIN directly regulates the expression of the FUL1 gene (Fujisawa et al., 2012;Martel et al., 2011). These results suggested that FUL1 might be a candidate gene for regulating tomato fruit ripening, an idea that is supported by the observation that a FUL1 homologue plays a central role in bilberry fruit ripening (Jaakola et al., 2010). However, it was previously reported that suppressing  a single tomato FUL gene resulted in no obvious phenotype (see Supplementary Fig. S1 at JXB online), but the double knock-out of FUL1 and FUL2 led to obvious phenotypes, such as yellow ripe fruits (Bemer et al., 2012), and severely perturbed ethylene production. These results suggest that FUL1 and FUL2 contribute to tomato fruit ripening in a redundant manner. Our data also suggest that the functions of FUL1 and FUL2 have diversified since FUL2, but not FUL1, was observed to affect style abscission and cell expansion. Transgenic plants over-expressing either FUL1 or FUL2 were generated and, interestingly, FUL2 over-expressing plants showed pleiotropic effects, including altered fruit shape (Fig. 2), stem structure (Fig. 3), and pericarp structure (Fig. 4), but FUL1 over-expressing plants had no obvious phenotypic change.
In the Arabidopsis ful mutant, both cell expansion and differentiation were affected in the carpel valves which have broader and rounder cells and fewer cell layers than those of the wild type (Gu et al., 1998). These results imply that the FRUITFULL protein has conserved functions in different plant species with regard to regulating cell expansion, differentiation, and fruit ripening. Our data suggest that FUL1 has lost some of the functions that have been retained by FUL2. This functional loss may be related to differences in protein structure. Leseberg et al. (2008) analysed the protein-protein interactions of 16 tomato MADS-box proteins with the MADS-domain removed and showed that FUL2 could interact with RIN, TM3, JOINTLESS, MBP18, MBP24, MBP21, MADS1, TAG1, and TM5, but FUL1 could only interact with TM3, JOINTLESS, and RIN. The K domain of MADS-box transcription factors is responsible for the specificity of the DNA-binding dimer formation and it is proposed that structural differences in the K domains of FUL1 and FUL2 may be responsible for their functional divergence.

FUL2 regulates the abscission of styles from tomato fruit
Following pollination and fertilization, the style normally withers and abscises from the fruit. However, it was observed that this process was inhibited in fruit from the FUL2-OE transgenic lines. The whole style of FUL2-OE failed to abscise during fruit set, resulting in fruit with a pointed tip at the blossom end (Fig. 2). This suggests that FUL2 regulates style abscission from tomato fruit at an early developmental stage. Inhibition of style abscission in FUL2-OE plants led to a fruit pointed tip phenotype. Seven phenotypically similar mutants (nipple-tip (n), n-2, n-3, n-4, persistent style (pst), beaked (bk), and bk-2) have previously been reported in tomato (Barten et al., 1994). Of these, n has been assigned to tomato chromosome 5 (Mutschler et al., 1987), and pst and bk mapped to chromosomes 7 and 2, respectively (Rick and Butler, 1956), but none of the corresponding genes had been cloned at this time. Accordingly, FUL2 is not a candidate gene for the n, or bk genes as it is located on chromosome 3. These results also show that development of the tomato pistil is regulated by multiple genes but more experiments are needed to elucidate any potential relationship between FUL2 and other 'nipple tip' genes.
How does FUL2 regulate tomato cell development? Previous studies showed that expansin proteins may function not only in fruit softening and organ abscission, but may also regulate organ size and morphology. The expansin gene AtEXP10 has been shown to encode a cell-wall-loosening protein that regulates organ size, morphology, and abscission and suppression of AtEXP10 led to shorter petioles and leaf blades and a reduction in petiole cell size (Cho and Cosgrove, 2000). Indeed, a thinner pericarp with smaller cell sizes of FUL2-OE tomato fruits was observed compared with the wild type, as well as suppressed expression of the expansin gene SlEXP1 (Fig. 4E). Thus, FUL2 might affect tomato style abscission and cell growth through regulating the expression of expansin genes, and it is proposed that the repressed expression of different expansin genes might also contribute to the thinner pericarp, stem, and small leaf phenotypes of the FUL2-OE transgenic lines.

FUL proteins, TAGL1 and MADS-RIN may form higher order complexes to regulate tomato fruit ripening
MADS domains can bind to the CArG-box sites in the promoter region of many genes (Messenguy and Dubois, 2003), and in dimers of MADS-box proteins, both of the MADS domains take part in binding to a single DNA site (Pellegrini et al., 1995;Santelli and Richmond, 2000). The heterodimer of the MADS-box proteins MC and JOINTLESS show stronger CArG-box DNA motif binding ability than MC and JOINTLESS alone, and the DNA-binding specificities of the heterodimer may be different from the homodimers of MC and JOINTLESS. This suggests that the heterodimerization of MADS-box proteins may be important for their physiological functions. TAGL1 and FULs can interact with the MADS-RIN protein, and all of them play important roles in tomato fruit ripening. It may be that both TAGL1 and FULs can form heterodimers with MADS-RIN, or form higher order complexes to regulate fruit ripening through an enhanced CArG-box DNA motif binding ability or altered DNA-binding specificities, and defects in any one of them perturbs tomato fruit ripening.
Similar to the rin mutant, TAGL1-RNAi and FUL1/FUL2-RNAi transgenic fruit showed a complete block in ethylene production and repressed expression of ACS2 (Barry et al., 2000;Vrebalov et al., 2002). This suggests that TAGL1, the two FUL genes and RIN may regulate ethylene biosynthesis through a similar, or even an identical pathway. It has been confirmed that RIN, TAGL1 and FULs can bind to the ACS2 promoter directly (Ito et al., 2008;Itkin et al., 2009;Shima et al., 2013), and the heterodimerization of RIN with TAGL1 may regulate tomato fruit ripening by enhancing the ability of RIN and TAGL1 to bind to the ACS2 promoter. If this is true, FULs may also form heterodimers with RIN, and this heterodimerization may enhance the ability of the RIN and FULs proteins to interact with the ACS2 promoter directly and compensate for the functional defect caused by the failed heterodimerization of RIN and TAGL1. However, it is known that knocking down the expression of either TAGL1 or the two FUL genes in tomato represses the ethylene burst during fruit ripening, which suggests that the interactions of RIN and TAGL1, or RIN and FULs, may not result in functional complementation. This raises the question, how do these three genes regulate autocatalytic ethylene production during tomato fruit ripening? One hypothesis is that RIN, TAGL1, and FULs can form heterodimers to regulate the expression of their own specific ripening related genes, and they can also form D-FUL-E higher order complexes to regulate the expression of ACS2 during tomato fruit ripening. Indeed, yeast three-hybrid assays have shown that TAGL1, RIN, and FUL1 can form D-FUL-E higher order complexes (Leseberg et al., 2008). Thus, the regulation of gene expression during fruit development and ripening by MADS box transcription factors is highly complex and much still remains to be learnt about the mode of action of RIN in regulating fruit ripening.

Supplementary data
Supplementary data can be found at JXB online.
Supplementary Table S1. Primers used in this study. Supplementary Fig. S1. FUL1 and FUL2 regulated fruit development and ripening.