Tomato plants (Solanum lycopersicum cv. Micro Tom) and BEL11 RNAi transgenic plants from the previous study (30) were grown in a greenhouse at 25°C with 16 h light and 8 h dark alternating. Flowers were tagged at anthesis. Clean and healthy fruit samples were immediately used for experiments or stored at –80°C after quick freezing in liquid nitrogen until they were used for measurement.

pTRV1 and pTRV2 vectors were subjected to this experiment described by Liu et al. (31). A 420-bp specific fragment of the SlBEL11 gene corresponding to bases 778–1197 of the SlBEL11 gene, a 540-bp fragment of the SlPDS gene located in bases 764–1303 of the PDS gene, a 340-bp fragment of the SlLCY-b1 gene located in bases 537–876 of the SlLCY-b1 gene, and a 339-bp fragment of the SlLCY-b2 gene located in bases 571–909 of the SlLCY-b2 gene were PCR amplified from tomato cDNA using Kod-plus polymerase (TOYOBO, Japan) and primers shown in Supplementary Table 1. The PCR amplified products then were ligated into the pTRV2 plasmids which were digested by SacI and XhoI. Then, pTRV1 vector plasmid and the recombinant plasmids identified by sequencing were transferred into Agrobacterium tumefaciens strain GV3101. The preparation of agrobacterium infection solution and the method of infecting tomato sprouts were described by Meng et al. (30). Tomato sprouts infiltrated with empty pTRV2 and pTRV2-PDS were set as negative and positive controls (31).

The extraction of total RNA was completed according to the manufacturers’ method described by the RNAprep Pure Plant Kit (TIANGEN Biotech, Hainan, China). In this experiment, containers and consumables were treated with RNase inactivation to ensure the integrity of the extracted RNA. The next steps were required to be completed on ice to confirm that the samples were not degraded. The RNA was subjected to cDNA synthesis with Fastking One Step RT-PCR Kit (TIANGEN Biotech, Hainan, China) following the instructions of the reagent manufacturer. qRT-PCR was conducted to examine the mRNA levels of the genes. In total, 1 μL of cDNA, together with 5 μL of SYBR green mix (TIANGEN Biotech, Hainan, China), 0.3 μL each forward and reverse primer, were added into the PCR mixtures (with the total volume of 10 μL). The PCR reactions were performed in A CFX Connect Detection System (Bio-Rad, CA, USA). Actin was selected as the reference gene. The method of 2^–ΔΔCt was adopted for calculation of results. The primer sequences were provided in Supplementary Table 1.

Fresh samples were pulverized with liquid nitrogen. In total, 50 (±1) mg of sample powder was extracted for 24 h with 1 mL of extraction solution (methanol:water:formic acid = 75:23:2) and vortexed for 15 min. After centrifugation, add 0.5 ml extraction solution to the residue, repeat shaking and centrifugation, and combine the supernatant. Concentrate and dry in the concentrator at 4°C, reconstitute with 200 μL 80% methanol. Centrifuge again, store the supernatant in the bottle for LC-MS/MS analysis. The mobile phase consisted of 0.04% acetic acid in water (solvent A) and 0.04% acetic acid in acetonitrile (solvent B). The flow rate was 0.4 ml/min and the column temperature was 40°C. The parameters of the gradient separation were as follows: increase 5% B–95% B in 10 min, hold for 1 min and change back to a 5% B concentration in 6 s, hold for 2.9 min.

The coding sequence of the SlBEL11 HD-box was PCR amplified. Then, the coding sequence was cloned into pGEX-4T-1 which was fused with a glutathione S-transferase (GST) tag. The recombinant construct identified by sequencing was transformed into Escherichia coli strain Rosetta (DE3). Protein expression and purification were completed referring to the method reported by Xiao et al. (32).

The probes containing TTGACTTGACatagtGTCA motif sequence from the promoter of SlLCY-b2 were biotinylated using a DNA 3′ End Biotinylation Kit (Thermo Fisher Scientific, Waltham, USA). The same sequence without labeled DNA fragment was a competitor. The TTGACTTGACatagtGTCA DNA fragments were modified into TTTTTTTTTTatagtTTTT and used as a mutant probe in the assay. EMSA was performed according to the manufacturer’s protocol provided by A LightShift Chemiluminescent EMSA kit (Thermo Scientific).

In this assay, the SlLCY-b2 promoter was cloned into the pGreenII 0800-LUC dual reporter gene vector (33). The SlBEL11 gene was cloned into the pEAQ vector (34) as effector. The constructed SlLCY-b2-pro-LUC effector and pEAQ-HT-SlBEL11 reporter plasmids were co-transformation of tobacco with Agrobacterium tumefaciens strain GV3101. The method was subject to Yin et al. (35). The Agrobacterium cells that has been transformed into the corresponding expression vector were inserted into the LB liquid medium containing kanamycin and gentamicin for cultivation (28°C, 240 rpm, 12–24 h). The Agrobacterium cells were collected and suspended in osmotic solution (10 mM MES, 10 mM MgCl₂, and 150 μM acetosyringone, pH 5.6), and adjusted to an appropriate concentration. The bacteria containing effector and reporter were mixed in a ratio of 9:1, and then the mixture was injected into tobacco leaves and cultured for 2–3 days. The ratio LUC/REN of two luciferase enzymes (LUC and REN) in leaves was determined by the dual luciferase reporting and analysis system (Promega). Each assay should contain at least 6 measurements.

The significance between the data was analyzed by t-test. The bar and line charts were plotted using GraphPad Prism 9 software.

The SlBEL11 belongs to the BEL1-like Homeodomain (BEL1-like) TFs family with high expression abundance in the red stage and involved in the regulation of chlorophyll synthesis and fruit pigmentation in tomato fruit (30). Compared with the control fruit, in addition to the observation that the SlBEL11 RNAi tomato fruits accumulate more chlorophyll at mature green stage, we also found that it still remained yellow at the red stage (Figure 1D). To further explore its possible role in tomato fruit, virus-induced gene silencing (VIGS) analysis was executed. In this experiment, a clearly demarcated red-yellow region was detected in fruits infected with the virus containing the TRV-SlPDS vector compared to tobacco rattle virus (TRV)-infected control and even red fruits. Among them, the tomato fruit injected with TRV-SlPDS was used as a positive control (Figure 1A). Similar to TRV-SlPDS tomato, tomato fruits injected with TRV-SlBEL11-containing vector virus also displayed mottled yellow and red areas, which were also separated by distinct borders. Gene expression analysis using quantitative real-time PCR (qRT-PCR) showed a reduction of approximately 80 and 90%, respectively, in SlPDS and SlBEL11 transcripts in the yellow areas of the TRV-SlPDS and TRV-SlBEL11 fruits, compared with the red TRV-Control fruits (Figures 1B, C). This mean that SlBEL11 may be involved in the regulation of carotenoid synthesis in tomato fruit.

In order to verify the hypothesis, we first carried out experiments to determine whether there was a difference in carotenoid content in the SlBEL11 RNAi and WT tomatoes. In this study, we found that the total carotenoid content in SlBEL11 RNAi tomatoes were much higher than that of WT at Br + 5 stage (Table 1). Compared to WT fruits, RNAi#012 and RNAi#0512 fruits showed a 90 and 54% reduction in phytoene, which was derived from upstream of the carotenoid synthesis pathway, respectively (Table 1). Whereas, the accumulation of substances downstream of carotenoid metabolism pathway was more than that of WT (Figure 1E). It is worth mentioning that large amounts of lutein accumulated in SlBEL11 silencing fruits and the content of lutein estimated in RNAi#012 (218.65 ng⋅g^–1) and RNAi#0512 fruits (166.67 ng⋅g^–1), was four times and three times that of WT fruits (56.67 ng⋅g^–1), respectively. Though the SlBEL11 RNAi tomatoes showed comparable lycopene content to that of WT plants, the silencing of SlBEL11 gene resulted in a large increase of carotenoids including zeaxanthin, violaxanthin, and neoxanthin in SlBEL11 RNAi tomatoes. Remarkably, the most striking is that the accumulation of violaxanthin and neoxanthin in the RNAi#012 tomato fruits was detected to be approximately 42 and 35 times that of WT, respectively. Overall, the loss of SlBEL11 function tremendously affected carotenoid metabolism, which led to abnormalities in the color of the tomato fruit.

To explain how SlBEL11 regulated carotenoid synthesis at the genetic level, we examined the transcript levels of main enzyme genes from carotenoid biosynthesis pathway (PSY1, PSY2, PDS, ZDS, CRTISO, LCY-b1/b2, CYC-b, and LCY-e) by qRT-PCR (Figure 2). In the SlBEL11 RNAi tomato lines, the expression levels of PSY1, PSY2, PDS, and ZDS were much lower than those in the WT line (Figure 2B). These enzyme genes mainly came from the upstream of carotenoid biosynthesis pathway. On the contrary, the expression levels of LCY-b2 and LCY-e genes, which were derived from the downstream of the carotenoid metabolic pathway, were detected to be strikingly increased in the SlBEL11 RNAi tomato lines. Remarkably, the expression level of LCY-b2 transcript was elevated about nine and fivefold in RNAi#012 and RNAi#0512 tomato fruits respectively, compared to WT fruits. The obvious up-regulation of the LCY-b2 gene was consistent with the marked increase in the content of its catalytic products especially zeaxanthin, violaxanthin and neoxanthin. Similarly, the transcriptional level of LCY-e increased by threefold in both RNAi#012 and RNAi#0512 transgenic lines, which may be the reason for the massive accumulation of lutein downstream of the carotenoid metabolism pathway. Hence, silencing of SlBEL11 in tomato enhanced carotenoid content downstream, which may be caused by its transcriptional regulation of carotenoid pathway genes.

To further explore the potential regulatory mechanisms of increased content of specific carotenoids and upregulation of specific genes from carotenoid biosynthetic pathway in SlBEL11 RNAi lines, we performed electrophoretic mobility-shift assay (EMSA). We found one possible motif TTGACTTGACatagtGTCA in the LCY-b2 promoter (Figure 3B). In EMSA, the capacity of the SlBEL11 protein of regulating the LCY-b2 promoter was analyzed. As shown in Figure 3, recombinant SlBEL11 protein physically bound to the LCY-b2 promoter fragment and caused changes in mobility. After adding more untagged competitor protein with the same sequence, the shifted band disappeared. In contrast, GST-SlBEL11 still showed bands with labeled probes when excess mutant probe was added (Figure 3B). Overall, in this experiment, SlBEL11 can directly bind to the TTGACTTGACatagtGTCA motif in the LCY-b2 promoter.

Through EMSA experiment, it has been determined that SlBEL11 can bind LCY-b2 promoter. But whether it has the activity of inhibiting or activating the LCY-b2 gene promoter needs to be further determined. To this end, the dual luciferase reporter assays were performed to examine whether it functioned to repress the LCY-b2 gene promoter. The dual luciferase reporter plasmids contained the SlLCY-b2 promoter were fused to LUC, which contained the TTGACTTGACatagtGTCA element. The CaMV35S promoter-driven REN served as an internal control, while the pEAQ-HT-SlBEL11 plasmid was used as an effector (Figure 4A). Compared with the control, expression of SlBEL11 significantly repressed the promoter activity of SlLCY-b2 (Figure 4B). Intriguingly, when the TTGACTTGACatagtGTCA element in promoter of SlLCY-b2 was mutated into TTTTTTTTTTatagtTTTT sequence (Figure 4C), there was no significant change in LUC/REN ratio (Figure 4D). From this we can conclude that the TTGACTTGACatagtGTCA element underlay the transcriptional repressive activity of SlBEL11 on the SlLCY-b2 gene promoter.

Comparison of the amino acid sequence of SlLCY-b2 with known lycopene beta cyclase homologous protein sequences revealed a high degree of homology to other LCY-bs (Supplementary Figure 1). Particularly, SlLCY-b2 showed 87% identity with the amino acid sequence of SlLCY-b1, which predicted functional similarity between the two genes. Studies have reported that up-regulation of the SlLCY-b1 gene caused fruit orange pigmentation, which was accompanied by increased β-carotene content. Further VIGS experiment was conducted to investigate the possible function of SlLCY-b2. In this experiment, silencing of the positive control PDS gene in Micro-Tom tomato plants results in aberrant carotenoid biosynthesis, conferring a leaf photo-bleaching phenotype (Figure 5A), which ensured that TRV clones caused gene silencing. Tomato infected with TRV-SlLCY-b1 developed a yellow spotted phenotype in leaves about 6-week post-Agro-infiltration (Figure 5A). When SlLCY-b1 and SlLCY-b2 genes were silenced at the same time, tomato leaves developed a more pronounced yellow phenotype, suggesting that carotenoid synthesis was affected seriously (Figure 5A). However, no apparent phenotype was detected in tomato leaves infiltrated with TRV-SlLCY-b2. Then, RT-qPCR was performed to analyze the effect of VIGS on SlPDS, SlLCY-b1, SlLCY-b2. In tomato plants infiltrated with TRV-PDS, the transcript level of the PDS gene were detected to be reduced by 96% compared with the TRV infected controls, which was consistent with the photo-bleaching phenotype of PDS-silenced tomato plants (Figures 5A, B). Contrasted with TRV-Control plants, the gene expression of LCY-b1 in the leaves of TRV-LCYb1 tomato plants was down-regulated by about 99%, which was comparable to the expression level of LCY-b1 in TRV-LCYb1 + b2 plants (Figures 5C, E). However, in TRV-LCYb2 tomato plants, the expression level of the LCYb2 gene changed slightly, with a silencing level of 34%, which may result in invisible phenotype in its leaves (Figure 5D). In TRV-LCYb1 + b2 tomato plants, LCY-b1 and LCY-b2 genes were silenced to varying degrees, and the silencing efficiencies were 99% and 54%, respectively, relative to the control plants (Figures 5E, F). These results may suggest that down-regulation of the LCY-b2 gene, resulting in inhibition of carotenoid biosynthesis, was partly responsible for the yellowing phenotype of TRV-LCYb1 + b2 tomato plants.