In our experiment, plants of Solanum lycopersicon Mill. cv. Ailsa Craig (AC), a near-isogenic tomato line, as the wild type plants. Seeds of tomato were kindly provided by Dr. James J. Giovannoni (Boyce Thompson Institute for Plant Research, Cornell University, Ithaca, NY, USA). The plants were grown in a greenhouse using standard cultural practices, with regular additions of fertilizer and supplementary lighting when required. For plants grown in vitro, seeds were germinated on MS medium (Murashige and Skoog, 1962). Conditions in the culture chamber room were set as follows: 16-h-day/8-h-night cycle, 25/18°C day/night temperature, 80% relative humidity. The transgenic plants were grown in peat-based compost supplemented with fertilizer in greenhouses equipped with heating and cooling systems and supplemental lighting. Plants of the first generation (T0) came from tissue culture, and plants of the second generation (T1) were from seedlings. Flowers were tagged at anthesis to determine maturity and ripening-stage. The ripening stages of tomato fruits were divided according to days after anthesis (dpa) and fruit color. Immediately upon harvesting, seeds and locular gel were removed and the pericarp frozen with liquid nitrogen, and stored at 80°C until further use.

RNA isolation from pericarp of the fruits was conducted using the method described by Moore et al. (2005). Then, 1 mg of total RNA was treated with DNase I and used to synthesize first-strand cDNA through reverse transcription-PCR using TransScript First-Strand cDNA Synthesis SuperMix (TransGen Biotech) with tailed oligo(dT)18 primer. A full-length cDNA sequence of SlTDT was amplified from fruit cDNA using primers of SlTDT-F (5′-ATGAATGGAGATCATCATGA-3′) and SlTDT-R (5′-CTAGCCACACATATGCAAA-3′) through TransStart FastPfu PCR SuperMix (TransGen Biotech). The amplified products were ligated into a pEASY-Blunt vector and subsequently transformed into Escherichia coli Trans1-T1 (TransGen Biotech). Positive colonies were selected and confirmed by sequencing (BGI, Beijing, China).

Sequence alignment of SlTDT and tonoplast dicarboxylate transporters from other organisms was performed using ClustalX (version 2.1) software (Livak and Schmittgen, 2001) with default multiple parameters and PAM series protein weight matrix. The alignment was then imported into MEGA (version 5.2) software and the phylogenetic tree was constructed by the neighbor-joining statistical method with 1000 bootstrap replicates.

To construct the 35S::GFP-SlTDT, the full-length SlTDT cDNA was cloned into the plant expression vector pCAMBIA2300 downstream of GFP. Transient expression of the GFP-SlTDT fusion construct in tobacco was performed as described with slight modifications (Batoko et al., 2000). Individual Agrobacterium colonies were cultured at 28°C to the stationary phase, washed, and resuspended in infiltration medium containing 10 mM MES (pH 5.6), 10 mM MgCl₂, and 200 mM acetosyringone to an OD 600 of 0.5. Then the suspension was injected into the abaxial epidermis of leaves of 5–6-week-old Nicotiana tabacum plants using a needleless syringe. Plants were incubated for 2–3 days at 25°C under constant light and transformed leaves were used to extract protoplasts and vacuoles for confocal microscopy.

To generate the SlTDT overexpression (OE) construct, the full-length cDNA (nucleotides 1-1659) was inserted into the plant binary vector pBI121 containing the cauliflower mosaic virus 35S promoter. To construct the RNA interference (RNAi) plasmid, a 363 bp SlTDT fragment was amplified from tomato cDNA using the primer pair: 5′-GTGCTGGTGATGGAACTG-3′ and 5′-GTAGGAGGGGATGGATGT-3′. The resulting product was cloned into the PCR8/GW/TOPO Gateway entry vector (Invitrogen). The cloned fragment was subsequently transferred into the destination vector pK7GWIWG2 (Karimi et al., 2002) using the LR Clonase II enzyme (Invitrogen). All constructs generated were confirmed by sequencing using universal primers (BGI, Guangzhou, China).

The above constructs were transformed into Agrobacterium tumefaciens GV3101 by electroporation, respectively, and Agrobacterium-mediated transformation was performed following the method described previously (Meissner et al., 1997, 2000). The hardened putative transgenics were shifted to the greenhouse and confirmed by PCR analysis using primers NPTII-F (5′-TCGGCTATGACTGGGCACAACAG-3′) and NPTII-R (5′-GGCGATACCGTAAAGCACGAGGAA-3′). Three independent transformants (T0) were identified by their ability to grow on kanamycin and the transcript levels of SlTDT were analyzed by RT-qPCR. Transformed seeds of T0 plants were selected on half MS medium containing 100 mg L^-1 kanamycin. Kanamycin-resistant plants (T1) were transferred to soil and grown to maturity in a phytotron (22°C, 16 h light/8 h dark photoperiod) and used for phenotype analysis.

For expression analysis in different tissues, total RNA was also extracted from roots, stems, young leaves, mature leaves and flowers, respectively, using the RNAiso reagent (TaKaRa, Japan), then used as the template for reverse-transcribed via PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa Corp., Dalian, P.R. China). Real-time quantitative PCR (RT-qPCR) assays was performed using SYBR Premix Ex Taq (TaKaRa Corp., Dalian, P.R. China) and the StepOne Plus Real-time PCR system (Applied Biosystems, Foster City, CA, USA). All PCR primers (Supplementary Table S1) were designed using Primer Express 3.0 (Applied Biosystems, Foster City, CA, USA), and primer specificity was determined by melting curve analysis. The relative expression levels were estimated using the 2^-ΔΔCt method (Livak and Schmittgen, 2001). β-Actin (Accession number: AB35991) was used as the endogenous control for quantifying mRNA levels. Three independent biological replicates were analyzed for each sample.

Metabolite analysis by Gas chromatography-mass spectrometer (GC-MS) was carried out according to a reported method (Batoko et al., 2000). with minor modifications to assay the changes in metabolites among the samples. Peel of 15 T1 tomato fruit and flesh (300 mg) of wild type was homogenized individually using a ball mill pre-cooled with liquid nitrogen and extracted in 2700 μL of methanol; 300 μL of internal standard (0.2 mg/mL ribitol) was subsequently added as a quantification standard. The mixture was extracted for 15 min at 70°C and centrifuged at 2200 rpm. The upper (methanol) phase was removed, concentrated and dried in a vacuum. The residue was dissolved in 50 μL of 20 mg/mL pyridine hydrochloride, and incubated for 30 min at 50°C. The sample was treated with 50 μL of BSTFA (1% TMCS) for 40 min at 60°C.

Exactly 1 μL of each sample was accurately absorbed and injected into the gas chromatograph through a fused-silica capillary column (30 m × 0.25 mm × 0.25 μm; DB-5 MS stationary phase). The injector temperature was 250°C and the carrier gas had a flow rate of 1.0 mL/min. The column temperature was held at 100°C for 1 min, increased to 184°C at a temperature gradient of 3°C /min, increased to 190°C at 0.5°C /min, held for 1 min, increased to 280°C at 15°C/min and then held for 5 min. Helium (purity 99.999%) was used as the carrier gas with a flow rate of 1 mL/min. The significant MS operating parameters were as follows: ionization voltage was 70 Ev (electron impact ionization), ion source temperature was 200°C and the interface temperature was 250°C. TIC (total ion current) spectra were recorded in the mass range of 45–600 atomic mass units (amu) in scanning mode.

To identify these unknown vacuolar-transporter-like sequences, we performed a Blast search against the database of tomato with the sequences of AtTDT (AT5G4756). We found that the tomato genome contains a single copy of an ORF (Solyc11g012360) exhibiting highest homologies with its Arabidopsis counterpart. The corresponding cDNAs comprising the entire ORF were amplified from the cDNA, which has been deposited in GenBank (accession No. KC733165). The putative protein comprising 532 amino acid in length, is predicted to contain 13 transmembrane domains, and possesses a calculated molecular mass of 60.6 kDa and an isoelectric point (pI) of 7.18 (Figure 1A). Sequence analysis revealed that the deduced amino acid sequence of SlTDT shared roughly 96, 83, and 63% identity with homologous sequences from potato (StTDT, XM_006352290.2), tobacco (NtTDT, XM_009630263.1), and A. thaliana (AtTDT, NM_124129.3), respectively (Figure 1A). To compare the amino acid similarity between the predicted SlTDT peptides and the TDTs isolated from other plant species, an unrooted phylogenetic tree was constructed using the program MEGA 5.0 (Figure 1B). This result showed that SlTDT was most closely related to the functional transporter StTDT from potato fruit. Analysis of the deduced amino acid sequences by TMHMM revealed the presence of 13 transmembrane domains (Figure 2).

To investigate the subcellular localization of SlTDT, we transiently expressed a GFP-SlTDT fusion construct under the control of the CaMV35S promoter in the epidermal cells of tobacco leaves. Confocal microscopy of transformed tobacco leaves revealed the vacuolar localization of SlTDT in plant. Figure 3 shows GFP-SlTDT fluorescence in the stomatal apparatus of tobacco leaves. Clear fluorescence of the fusion protein was found in the tonoplast, labeling the vacuole membrane and not the plasma membrane. In epidermal cells transformed with the control vector pCAMBIA2300-eGFP, green fluorescence was distributed in both the cytoplasm and nucleus of tobacco epidermal cells (Figure 3). In contrast, green fluorescence was exclusively detected in the vacuole membrane of cells transformed with the fusion plasmid of pCAMBIA2300eGFP-SlTDT (Figure 3). These results showed clearly that SlTDT localized to the vacuole membrane.

To investigate changes in SlTDT expression level, qRT-PCR was performed using RNA isolated from roots, stems, young leaves, mature leaves, flowers and fruits at different ripening stages (Figure 4). Transcript level of SlTDT differed among the various organs, with its transcript being most abundant in young leaves, and increased during fruit ripening, and then dropped (Figure 4).

To further explore the function of SlTDT, a 35S:SlTDT overexpression construct and a SlTDT RNAi construct were separately introduced into tomato by agrobacterium-mediated method. Three independent transgenic lines corresponding to each construct were identified by their ability to grow on kanamycin. Transcript expression of SlTDT in kanamycin-resistant plants (T0) was analyzed by RT-qPCR and two overexpression lines (OE2, OE3) and two RNAi lines (R1, R2) were selected for further study according to their transcript levels of SlTDT (Supplementary Figure S1). We observed that the transgenic SlTDT T1 plants grown in phytotron showed normal growth phenotype and the size and ripening process of their fruits were similar to those of WT plants (Supplementary Figures S2 and S3).

GC-MS analysis revealed that fruit of SlTDT overexpression lines exhibited higher malate level and lower citrate level at orange ripe stage compared with that of the control. By contrast, silencing of SlTDT led to lower malate level and higher citrate level than that in control (Figure 5). Moreover, overexpression of SlTDT significantly increased the content of glutamate, while repression of SlTDT reduced the content of glutamate.