Tomato (Solanum lycopersicum L. cv. Micro-Tom) plants and transgenic lines in this background were grown in a greenhouse under standard conditions (16 h/8 h light/dark cycle, 80% relative humidity, 25˚C/18˚C day/night temperature). Blooming flowers were tagged and ripening stages were divided according to days post anthesis (DPA) and fruit color. For RNA extraction, pigment measurement and SAHH enzyme activity analyses, samples were immediately frozen in liquid nitrogen and then stored at -80˚C until further use. For ACC treatment and ethylene measurements, fresh material was used.

For expression pattern analysis, roots (R), stems (S), leaves (L), floral buds (B), flowers (FL), 16 DPA fruit, mature green fruit (MG), breaker fruit (Br), 2 days post breaker fruit (Br+2) and red ripe fruit (RR) were collected with liquid nitrogen and stored at -80˚C. Due to the high sequence identity and potential functional redundancy of SlSAHH genes (Li et al., 2015), primers were designed at 3′ terminal less conserved region of each cDNA. The sequences of primers were listed in Supplementary Table S1.

To investigate the expression of SlSAHH genes responded to different hormones, tomato fruits at mature green stage (MG, 35 DPA) were picked out. 100 μM ACC, 100 μM IAA, 100 μM GA3, 100 μM ABA or 0.4% ethephon in buffer solution was prepared in turn and hormone injection experiment was performed afterward (Orzaez et al., 2006; Yang et al., 2017). After treatment for 96 h, samples for each treatment were frozen in liquid nitrogen immediately and stored at -80˚C until RNA extraction.

ORF sequence of SlSAHH2 was downloaded from Sol Genomics Network¹ and amplified with primers SAHH2-F and SAHH2-R (listed in Supplementary Table S1). Afterward, plant binary vector pLP100 was chosen to construct over-expression vector pLP100-35S-SAHH2. Agrobacterium tumefaciens strain GV3101 was prepared for tomato transformation (Fillatti et al., 1987). The positive transgenic lines were screened by kanamycin (100 mg⋅L^-1) selection and GUS staining. After qPCR confirmation, two successful over-expression lines (OE-5# and OE-6#) in T2 and T3 generation were selected for further analysis.

A mature protein coding region of SAHH2 was amplified with primers listed in Supplementary Table S1 and then subcloned into pET-28a (+) vector (Novagen, Darmstadt, Germany) between the Sac I- and BamH I- sites with the poly-histidine at N-terminal. After that, a single colony of Escherichia coli BL21 (DE3) harboring the pET28a-SAHH2 vector was cultured in 5 mL LB medium with 100 μg⋅mL^-1 kanamycin and grown overnight at 37˚C. This culture was next extended to 100 mL shaken at 37˚C with 250 rpm until OD600 reached 0.6–0.8. Collect 5 mL of the bacteria liquid referred as 0 h and then add IPTG (CWBIO, China) into the left culture with a final concentration of 1 mM. Subsequently, the culture was shaken at 180 rpm at 30˚C for additional 2 h, 4 h, and 6 h. Bacteria liquid was collected by centrifugation at 4000 g for 10 min at 4˚C, washed twice and resuspended with 20 mM PB buffer (phosphate buffer, pH7.4). Then the solution of bacteria cells supplemented with 1 mg⋅mL^-1 lysozyme and 1 mM PMSF was sonicated on ice until the cell lysate clarified. At last, the supernatant after centrifugation and pellet resuspended with PB of different time points were all used to run SDS–PAGE. For purification of the recombinant protein SAHH2, a Ni-Agarose Kit for His tag soluble protein (CWBIO, China) was used with some modifications.

Measurement of SAHH activity was carried out following published methods (Wolfson et al., 1986), with slight modifications. In vitro experiment, the purified 5, 10, 15, 20 and 25 μg recombinant protein removing salts and other small molecules through Sephadex G25 columns were used to assay the hydrolysis activities. Reactions were conducted at 25˚C for 15 min in 1 mL reaction mixture contained: 50 mM HEPES-KOH (pH 7.8), 1 mM EDTA (pH 8.0), 0.1 mM DTNB (Sigma), 0.1 mM SAH (Sigma), and required dose of recombinant protein SlSAHH2. In the reaction assay, Hcy was used as a reducing reagent for 5, 5′dithiobis-(2-nitrobenzoic acid) (DTNB) to DTNB-thiolate, which resulted in an increase in the absorbance of the reaction mixture at 412 nm. In vivo experiment, 100 mg tissues from WT and transgenic lines at different stages were ground in ice-cold HEPES buffer (pH7.8, 50 mM HEPES, 5 mM DTT, 1 mM Na₂EDTA, 5 mM ascorbic acid, 10 mM boric acid, 20 mM Na-metabisulfate and 4% Polyvinylpyrrolidone) and extracts were collected by centrifugation at 4˚C for 5 min. The supernatant (0.5 mg) was used for enzyme activity assays. The experiments were performed in three repetitions with six replicates each.

Tomato flowers were tagged and fruit ripening time was observed. On-vine ripening period was expressed as the number of days needed from anthesis to breaker (DPA). The experiment was carried out with nine individual plants for each line and repeated for three generations (T1–T3).

In order to determine ethylene production level, pre-weighed fruits at breaker stage were harvested and placed in open 50 mL air-tight containers for 3 h to avoid the effect of ethylene emission caused by picking. Jars were then sealed with paraffin wax and incubated at room temperate for 16 h. Afterward, 1 mL of headspace gas was injected into a gas chromatograph (Varian CP-3800 GC gas chromatograph, United States) fitted with an activated alumina column and a flame ionization detector. Reagent-grade ethylene standards were used for evaluating ethylene content and ethylene production in fruit was calculated with normalization of fruit weight (Chung et al., 2010). Three biological replicates were adopted and each replicate contained at least 10 fruits.

Five gram fruit pericarp at 43- and 46- DPA were prepared in a beaker. Little amount of methanol was added into the beaker and then stirred with a glass rod adequately. After that, filtered the solution with filter paper and repeated the above steps until the extract became colorless. Next, discard the extraction and extract the left residue with little amount of petroleum ether several times until the extraction became colorless again. Collect the extraction which was the tomato red pigment extract for following experiment. For standard curve drawing, different concentration of the sultan I solution was used. For lycopene content measurement, 1–2 ml extractions together with absolute alcohol supplied was used. The UV spectra were monitored at 485 nm. Two independent experiments were performed as biological replicate for each sample with three technological replicates.

For ACC treatment experiment, tomato fruit were harvested at breaker stage and injected with a buffer solution (pH 5.6) contained MES (10 mM), sorbitol (3% w/v) and ACC (100 μM) as described above. Briefly, fruits were injected with a 1 ml syringe containing a 0.5 mm needle, inserted 3–4 mm into the fruit tissue from the stylar apex. The solution was gently injected into the fruit until the buffer ran off the hydathodes and stylar apex at the tip of the sepals. Only completely infiltrated fruits were used for next experiments. Then fruits were incubated in a culture room at 26˚C, under 16 h light/8 h dark cycles with light intensity of 100 μmol m^-2 s^-1. After 96 h, the difference of changes in color was observed and fruits pericarps were frozen at -80˚C until further analysis.

For ethylene triple response assay, seeds of WT and transgenic lines were sterilized and sown on MS medium with 1 μM ACC or not. Then the seeds were all cultured at 25˚C in the dark. Root and Hypocotyl elongation were observed and measured 7 days post sowing. For each line, at least 30 seedlings were measured. To clarify the molecular mechanism, the expression of E4, E8, ACO1, ACO3 and ACS2 was detected by real-time PCR.

For each line, three independent biological replicates were used and six fruits collected from different plants were referred to as one sample. Total RNAs of samples were extracted using Trizol (Invitrogen) according to the manufacturer’s instructions. Reverse transcription of the first-strand cDNA was performed with RevertAid^TM First Strand cDNA Synthesis Kit (Fermentas, United Kingdom). Gene-specific primers were designed with the software of Primer Express 5.0 and SlActin (Solyc03g078400) was used as internal control. QPCR was performed using the SyBR Green PCR Master Mix (CWBIO, China) in a 25 μL total sample volume (1.0 μL of cDNA, 1.0 μL of primers, 12.5 μL of 2×SYBR Mix Taq and 10.5 μL of distilled water). Reaction was performed with an initial incubation at 95˚C for 20 s, followed by 40 cycles of 95˚C for 3 s and 60˚C for 30 s with Bio-Rad CFX connect (Bio-Rad, United States). The cycle threshold (Ct) 2^-Δ(ΔCt) method was adopted for relative quantification the specific mRNA levels (Livak and Schmittgen, 2001). Primers used for real-time PCR were all shown in Supplementary Table S1.

All experiments were repeated three times independently and all results were reproducible. Statistical results were presented as means ± standard error. To compare group differences, two-tailed Student’s t-tests were used. P-values less than 0.05 were recognized as significant.

To explore the expression profile of SlSAHH gene family, real-time PCR was performed in roots (R), stems (S), leaves (L), buds (B), flowers (FL), and fruit at different ripening stages of wild-type tomato. Although functional redundancy existed among the three genes (Li et al., 2015), their expression patterns were quite different. SlSAHH1 was highly expressed in stem, buds and flowers, with a rapid decline in all fruit tissues (Figure 1A). Interestingly, SlSAHH2 showed quite different expression pattern. The mRNA level was lower in roots, stems, leaves, buds, flowers and 16DPA fruit, while highly accumulated during fruit ripening especially at breaker stage (Figure 1B). Additionally, the expression profile of SlSAHH3 was distinct from the first two members. High transcriptional level can be seen in the tissues of roots, leaves and flowers and the maximum expression level was displayed in stem, but its transcriptional level was lower in ripe fruit (Figure 1C). These results indicated that the expression patterns of the three SlSAHH genes were all different and SlSAHH2 may function in the process of fruit ripening.

To determine whether the expression of SlSAHH gene family members could be regulated by phytohormone, qPCR was conducted with wild-type MG fruit treated by exogenous ethephon, ACC, GA3, IAA and ABA. As shown in Figure 2A, the expression of SlSAHH1 was induced by all these hormones treatment. The mRNA level of SlSAHH3 changed little upon ethephon, ACC, GA3 and IAA treatment. Similarly, both SlSAHH1 and SlSAHH3 were induced significantly after ABA treatment (Figure 2C). Interestingly, the mRNA level of SlSAHH2 increased significantly (16-folds) with the stimulation of exogenous ethylene. Also, the expression of SlSAHH2 can be up-regulated by IAA and ABA and down-regulated by GA3 (Figure 2B). These findings indicated that SlSAHH family genes could be regulated by phytohormones and the expression of SlSAHH2 was strongly induced by ethylene.

To further investigate the function of SlSAHH2, transgenic tomato lines were created by over expressing SlSAHH2 using its full-length cDNA with pLP100 vector under CaMV 35S promoter. WT and five different transgenic lines were grown under the same condition. For qPCR analysis in leaves, SlSAHH2 mRNA level was up-regulated significantly in two independent lines (OE-5# and OE-6#) without affecting the expression of other two homologous genes (Figure 3A). To establish the correlation between SlSAHH2 and ripening time, the expression of SlSAHH2 and its corresponding SAHH enzyme activity were detected in flower and early development fruit. According to the higher mRNA level of SlSAHH2 in transgenic lines (Figure 3B), SAHH enzyme activity enhanced about 20–30% in transgenic IMG fruit than in WT (Figure 3C). However, SAHH enzyme activity showed no significant change in transgenic flower than in WT (Figure 3C). Also, SAHH enzyme activity remained higher in transgenic breaker fruit although the SlSAHH2 mRNA level showed no significant change compared to WT (Figure 3D). So the inconsistency between mRNA level and enzyme activity existed. To give strong evidence that SlSAHH2 can functioned as an enzyme, recombinant protein SlSAHH2 (theoretical molecular weight: 56.48 kDa) was obtained in E. coli. It was induced strongly by IPTG at 6 h in supernatant and the optimized concentration of imidazole for purification was 200 mM (Figure 3E and Supplementary Figure S1). The enzyme activity enhanced nearly linearly in a dose-dependent way of SAHH2, which indicated that SlSAHH2 can hydrolyze substrate in vitro (Figure 3F).

During fruit development, the time from blossom to breaker stage was calculated (DPA) by tagging the flowers. In present study, the whole ripening time was shortened in OE fruit than in WT. For instance, at 41 DPA, the WT fruit was still be mature green while the OE-5# fruit was in Br+3 and the OE-6# fruit was in breaker stage (Figure 4A). The ripening process was significantly accelerated in OE-5# (about 5 days) and OE-6# (about 2 days) than in WT (Figure 4B). As the main content of carotenoids, lycopene content was extracted and determined. The accumulation of lycopene in OE fruit was much higher at both 43- and 46-DPA (Figure 4C). Phytoene synthase 1 plays a role in rate-limiting step of carotenoid synthesis during tomato fruit ripening. Consistent with lycopene content, the expression of Phytoene synthetase1 coding gene SlPSY1 was up-regulated in OE fruit at 43 DPA (Figure 4D). The decreased expression of SlPSY1 existed in OE fruit at 46 DPA may be because of negative feedback. So SlSAHH2 was accounting for enhanced lycopene in ripening fruit.

Considering tomato fruit ripening was predominantly controlled by ethylene, the expression of several ripening regulators and ethylene related genes was detected. At breaker stage, although ethylene production just increased significantly in OE-6# transgenic line (Figure 5A), the expression of two ethylene inducible genes (E4 and E8) and three ethylene biosynthesis genes (SlACO1, SlACO3 and SlACS2) increased significantly in both two transgenic lines (Figures 5D–F). Additionally, at early fruit developmental stages of transgenic lines, the expression of these genes was initially suppressed in flower and then began to increase in IMG fruit (Supplementary Figure S2), displaying the gradually enhanced expression pattern with fruit development and ripening. The transcriptional levels of several important ripening regulators were also influenced as expected. The expression of RIN, TAGL1 and CNR were up-regulated. As a negative regulator, AP2a was inhibited in SlSAHH2-OE fruit. At the same time, NOR was also down-regulated because it was a positive regulator of AP2a (Figure 5G). The results suggested that over-expression of SlSAHH2 in tomato influenced the expression of ripening regulators and ethylene related genes significantly, which even enhanced ethylene production.

Fruit ripening was a process of being sensitive to ethylene. To investigate ethylene sensitivity in WT and transgenic fruit, ethylene precursor 1-aminocyclopropane-1-carboxylate (ACC) was used for treatment. After treatment for 96 h, color of OE fruit pericarp turned much more quickly (Figure 6A). When fruit became a little orange in OE-5# and OE-6#, the color in WT fruit was still pale yellow. These results indicated that pigmentation of OE fruit was partly dependent on ethylene. It was concluded that OE fruit was much more sensitive to ethylene which can accelerate the process of ripening. QPCR results indicated that E4, E8, SlACO1, SlACO3 and SlACS2 were all up-regulated in OE-5# and OE-6# fruit after ACC treatment, which was in accordance with the above observation (Figure 6B).

Furthermore, experiments were conducted in non-fruit tissue to confirm ethylene sensitivity. The transcriptional levels of SlSAHH2 increased substantially at 3 h, while evidently decreased at 1 and 6 h post ethephon treatment in WT seedlings, revealing that the regulation of ethylene on SlSAHH2 in non-fruit tissue was effective (Figure 7A). In ethylene triple response assays, WT, OE-5# and OE-6# seeds were germinated on Murashige and Skoog medium with or without the supplement of ACC. The elongation of roots and hypocotyls was measured 7 days post sowing. It was demonstrated that the average length of transgenic hypocotyls and roots was shorter than WT no matter in the absence (0 μM) or presence (1.0 μM) of ACC (Figures 7B,C). We detected the mRNA level of E4, E8, SlACO1, SlACO3 and SlACS2 in seedlings with or without ACC treatment. SlACO1 showed prominent up-regulation in OE-5# and OE-6# before treatment. Except for SlACS2, other genes were all up-regulated significantly after ACC treatment (Figure 7D). The qPCR results were basically in line with the morphologic changes of seedlings, revealing that the transgenic lines were also sensitive to ethylene in non-fruit tissues.