Synergetic effect of the Onion CHI gene on the PAP1 regulatory gene for enhancing the flavonoid profile of tomato skin

Tomatoes are known to have ameliorative effects on cardiovascular disease and cancer. The nutritional value of tomatoes can be enhanced by increasing flavonoids content through genetic modification. The regulatory gene PAP1 (production of anthocyanin pigment 1) from Arabidopsis is reported to increase initial flavonoid flux and anthocyanin content. The structural gene CHI from Alium cepa increases flavonol content. However, the number of structural genes that can be transferred to plants is limited. To solve this problem, for the first time, we produced gene stacking transgenic tomato, in which Arabidopsis PAP1 (production of anthocyanin pigment 1) was stacked with an onion CHI by crossing. This procedure resulted in increased rutin and total anthocyanin content of as much as 130 and 30 times more, respectively, than the content in wild tomato skin, compared with 2.3 and 3 times more flavonol content, and 1 and 1.5 times more anthocyanin content in unstacked FLS and PAP1 tomatoes, respectively.


Generation of CHI-FLS-and PAP1 -expressing tomato plants. Initially, 15 transgenic lines for each
FLS and PAP1 were generated. Among them, 9 morphologically normal and healthy lines were selected and subjected to further analysis. The 15 lines were selected on 100 mg/L kanamycin selection analysis before DNA and RNA amplification. All transgenic tomatoes were confirmed by DNA amplification.
The shapes of all the selected transgenic tomatoes were indistinguishable from those of the wild type ( Fig. 2A,B and C). There were no statistical differences for fruit weight and number of fruits per plant between transgenic plants having different transgenes and wild type plants ( Table 1). The CHI lines were used from previous work 30,31 . By crossing, we obtained 4 lines of FLS x CHI and 5 lines of PAP1 x CHI. All stacked lines were confirmed by PCR using genomic DNA ( Fig. 3A and B) as the template. The expression of genes was confirmed by RT-PCR (Fig. 3C~F). DNA and RT-PCR were performed by each gene primer for every transgenic plant before further experiments to prevent segregation because each gene was in different vector.
The petioles of approximately 1 month old plants harboring both PAP1 and CHI had dark purple spots and lasted for c.a. 3 weeks (Fig. 2D). Molecular work. Gene expression was confirmed by RT-PCR (Fig. 3C~F). All the 32 T0 lines of FLS, PAP1, and CHI were confirmed by DNA amplification. All the T1 lines of each gene were checked by DNA and RT-PCR analysis. Consistent with single insertion, each T1 line showed a 3:1 segregation ratio. Over 90% of the F1   generation lines of FLS x CHI and PAP1 x CHI had both genes integrated. Also, this segregation ratio is the same in the F2 generation. All stacked and unstacked genes were stably transmitted to the next generation. Figure 3C~F shows a typical expression pattern of the CHI, FLS, and PAP1 genes in CHI x FLS and CHI x PAP1. They exhibited no distinguishable phenotypes between lines regardless of flavonoid content and inserted genes. The expression of each FLS, CHI and PAP1 was independently strong enough on lines, flavonol and anthocyanin content. The CHI and PAP1 lines were selected for Southern blot analysis because the stacked lines between them show the highest total flavonol content. Among the five CHI lines, CHI-06 and CHI-08 were selected for Southern blot analysis (Fig. 3G) because they showed the highest content of the major flavonols and total flavonols from previous work 30,31 when stacked with the PAP1 and FLS lines. The genomic DNAs from randomly selected CHI and PAP1 transgenic plants were digested with HINDII to include inserted T-DNA and genomic DNA and hybridized with CHI and PAP1 probe. All of the detected fragments with CHI and PAP1 probe had the expected fragment size of larger than 3.0k. These lines demonstrated single copy insertion.
Major flavonoids. First, we detected the rutin content in CHI, FLS, and PAP1 tomatoes. All the FLS-, CHI-, and PAP1-expressing tomatoes displayed significantly higher rutin content as compared with wild-type tomatoes (Fig. 4). Among these transgenic plants, the CHI-expressing tomatoes exhibited the highest rutin content as compared with other transgenic and wild-type tomatoes (Fig. 4). The CHI lines used for crossing with the FLS and PAP1 lines were selected to compare genetic effects. Rutin concentration in the unstacked T1 generation of CHI-expressing tomatoes exhibited greater variation than that in the FLS-and PAP1-expressing tomatoes. The FLS and CHI lines exhibited significant differences between lines, but PAP1 showed no differences between lines regarding rutin concentration.
The FLS x CHI and PAP1 x CHI transgenic tomato lines crossed with the CHI-08 line exhibited the highest rutin, quercetin glucoside, and kaempferol rutinoside content (Fig. 5) in each stacked tomato compared with CHI lines. While the wild, FLS, and PAP1 transgenics exhibited unmeasureable traces of quercetin-glucoside, quercetin-glucoside content in the F01C08 and P06C08 lines were 4.11 and 6.44 mg/g, respectively. The CHI lines exhibited the largest difference in production of kaempferol rutinoside when compared with the FLS and PAP1 lines crossed with the CHI lines. Both the CHI 06 and 08 lines crossed with FLS and PAP1 exhibited considerable differences, but the CHI 04 lines exhibited no difference when crossed with FLS. The high flavonol phenotype was maintained in mature fruit of hemizygous T1 and homozygous T2 individuals of the CHI x FLS and CHI x PAP1 lines, indicating that the high-flavonoid phenotype was inherited stably to the next generations (Table 1).
Naringenin chalcone is a precursor of naringenin converted by CHI (Fig. 1). The content of naringenin chalcone in CHI 06 and 08 was significantly lower than in CHI 01, 02, or 04. It showed an inverse relationship with rutin. When the lines were crossed with the FLS and PAP1 lines, the inverse relationship became less clear. The  lines crossed with the FLS and PAP1 lines exhibited consistently less variation than the wild, FLS, and PAP1 lines due to overexpression of CHI.
Minor flavonoids. Chlorogenic acid, caffeic acid, cumaric acid, and sinapic acid are upstream of CHI, while luteolin-7-O-glucoside and myricetin are in the downstream (Fig. 1). These minor flavonoids are subject to more complex variation (Fig. 6) than the major flavonoids (Fig. 5). The chlorogenic acid, caffeic acid, and cumaric acid content of all unstacked PAP1 lines increased significantly compared with the wild lines. Sinapic acid increased to an average of 0.008 mg/100 g in the PAP1 line and 0.037 mg/100 g and 0.031 mg/100 g in the P06C08 and P16C06 lines, respectively, from zero in the wild type. The content of luteolin-7-O is highest in the CHI 06 and 08 lines. Neither the unstacked FLS and PAP1 lines nor these lines stacked with CHI had a significant effect on enhancing the luteolin-7-O flavonol content. Regarding myricetin content, all CHI stacked lines with FLS and PAP1 exhibited higher content than the wild, FLS, and PAP1 lines.
Total flavonol content, antioxidant activity, and total anthocyanin content. The lines with highest rutin content in each genotype were selected for analysis. Differences in total flavonol content, anti-oxidant activity, and total anthocyanin content between stacked lines and their parents were measured ( Table 2). The methanol-and ethanol (1:1)-extracted antioxidant activity of CHI x PAP1 increased as much as 53 times compared with the wild type. Hence, the most abundant flavonol in tomato peel was rutin, and the total flavonol content was in accordance with the rutin content. Only the CHI x PAP1 lines contributed to the increase in total anthocyanin content. Stability of flavonol expression between F1 and F2. Two CHI lines used for crossing and two crossed lines from each genotype were selected to test stability between generations. There was no significant difference between F1 and F2 by t-tests for rutin, quercetin-3-B-D-glucoside, or kaempferol rutinoside ( Table 3). The order of flavonol content across lines and genotypes was not changed in any generations.

Discussion
The ectopic expression of our onion CHI gene resulted in a significant increase in rutin, as expected based on previous work 31 . The rise in total flavonol accumulation was comprised mainly of increases in the accumulation of rutin, quercetin glucoside, and kaempferol-rutinoside in the peel tissues. This result was consistent with that reported in Verhoeyen et al. 22 . All these compounds have the same precursor, dihydrokampferol converted by FLS. The ectopic expression of the petunia FLS gene by itself does not increase the flavonol level 22 . The petunia FLS is effective only when stacked with CHI and CHS. However, our onion FLS increased the rutin content by a factor of as much as 3.5. Our cultivar may have a naringenin chalcone pool, which is the product of CHS, that can supply enough substrate to FLS and CHI. The onion CHI was used on the assumption that onion has a robust flavonol pathway because onion exhibited the highest reported quercetin content in a survey of 28 vegetables and 9 fruits 35 . The quercetin constitutes more than 80% of the total flavonoids in onion 33 . Rutin is the glycoside form of quercetin. In both CHI/PAP1 and CHI/ FLS lines, the rutin content increased significantly in comparison with the CHI-only plants of its parent, which indicated that the CHI gene transmitted to the next generation successfully. The line difference was greater than the genotype difference. The order of rutin content was CHI08 > CHI06 > CHI 04. In both CHI/FLS and CHI/ PAP1 lines, the order of CHI lines for rutin did not change. It is possible that the effect of CHI is greater than that of FLS or PAP1. The average of rutin content in CHII/PAP1 was greater than in CHI/FLS even though the rutin content in FLS only was almost the same as in PAP1 only. It has been reported that PAP1 upregulates PAL, CHS, The content of quercetin-glucoside and kaempferol rutinoside showed more complex variations than that of rutin. However, lines crossed with CHI08 and CHI06 had the highest and second highest content, respectively, of these flavonols in both CHI/PAP1 and CHI/FLS genotypes, and the same content as in the CHI-only tomato. Not all CHI/PAP1 lines showed increased quercetin-glucoside content compared with CHI-only lines. Flavonol content in a stacked genotype depends on how the regulatory gene works to move the flavonoid flux. When CHI is crossed with another regulatory gene, Del/Ros, the content of quercetin-glucoside in CHI/Del/Ros is lower than in CHI-only lines 31 . The Del/Ros converted the flavonol substrate from flavonol to anthocyanin 31 . In that case, the CHI08 line showed also the highest anthocyanin content and the CHI06 line shows the second highest 31 .
The level of naringenin chalcone accumulation was depleted in the high-flavonol fruit when compared with the wild type. The high-flavonol lines, CHI-06 and 08, exhibited the lowest content of naringenin chalcone, which was converted to naringenin by CHI. In high-flavonol transgenic tomato favoring the petunia CHI gene, naringenin and naringenin chalcone have a negative correlation 20 . The proportional increase in rutin, quercetin-glucoside, and kaempferol rutinoside was much greater than the decrease in naringenin chalcone in the high-flavonol lines. Verhoeyen (2002) suggested that the ectopic expression of CHI utilized the naringenin chalcone pool and that depletion of naringenin chalcone removed a point of negative feedback, in the form of increased flux, from the pathway 22 .
The PAP1-transferred tomato showed a pale pinkish color on the shoot and some pink spots in the tomato fruit in the green state 17 . However, the color of the fruit was the same across the genotypes even though there was a 6-fold increase in a CHI/PAP1 line. the typical red color in tomato comes from lycopene 31 . The color of anthocyanin might be covered by lycopene 17 in this experiment. The transcription factors in flavonoid biosynthesis, including PAP1, often work in a complex combinatorial way and can also change the expression of other regulatory factors to enable a cell-specific accumulation of pigments 37 . Not only regulatory genes, but also structural genes such as CHI and FLS, express in tissue-specific ways 38 . Most flavonol and anthocyanin are located in tomato peel rather than flesh 22 . Flavonol alone 1 or anthocyanin alone 27 increase in both peel and flesh with one and two regulatory genes, respectively. Both flavonol and anthocyanin increased in both flesh with both two regulatory genes and one structural gene 31 . However, in this report, a single regulatory gene and a single structural gene were used to increase both flavonol and anthocyanin in tomato peel.
Overall, the PAP1 x CHI lines were more effective than the FLS x CHI lines in terms of flavonol production. Even though the PAP1 gene upregulates the DFR gene 17 , the upregulation of DFR is not active enough to exploit  most of the flavonol flux converted by ectopic expression of CHI. For the most part, some minor phenolics, upstream of CHI, exhibited very little increase in the CHI x PAP1 lines. This might be due to the movement of the flavonoid flux by the activation of CHS by PAP1 and the ectopic expression of CHI. There are two groups involved in flavonoid pathways. One is an early biosynthetic gene such as CHS-, CHI-or F3H-regulated R2R3-MYB regulatory genes such as MYB12. The other is a late biosynthetic gene such as DFR, regulated by a ternary transcription factor including PAP1 39 . MYB12-inserted tomato increased in flavonol content only 1 . In this experiment, the onion CHI removed the bottleneck blocking early biosynthetic genes and the PAP1 upregulated late biosynthetic genes. This resulted in both higher flavonol content and higher anthocyanin content. However, the upregulation of late biosynthetic genes was not strong enough to convert all flavonol to anthocyanin. The onion CHI stacked with two regulatory genes, Del/Ros showed the same effect of increasing both flavonol and anthocyanin content 31 .
We observed strong expression of PAP1 in fruit without environmental stress. In addition to the effect on flavonoid, the PAP1 tends to respond to its environment, with the expression level increasing by exposure to light 17 . In Arabidopsis, the anthocyanin level increases by osmotic pressure 29 . Herbivory suppressed the PAP1-induced increase of transcripts of flavonoid biosynthetic genes in tobacco 40 . However, the PAP1 expressed well in tomato fruit, increasing anthocyanin content without stress 17 .
In this antioxidant activity test, the main factor affecting the increase of antioxidant activity was the presence of rutin which is the most abundant flavonol in our transgenic unstacked and stacked lines and is a strong antioxidant [41][42][43] . The antioxidant activity of rutin increases with any increase in rutin concentration 44 . Also, the antioxidant activity of plant extracts from ARTEMISIA VULGARIS, of which the main flavonoid is rutin 45 , increases with the concentration of plant extracts, in the same manner as that of rutin 46 . In this experiment, the r-square of the regression between rutin content and antioxidant activity is 0.74 (data not shown). Even though total flavonol, rutin and anthocyanin content increased, the lycopene content did not change.
For the first time, we confirmed that combinations of one structural gene CHI and regulatory gene PAP1 enhanced flavonoid production tremendously. There were approximately 130 and 30 times more of the major flavonols, rutin and total anthocyanin, respectively, in tomato peel of CHI/PAP1 compared with the content in wild tomato peel. The research work provides very important information to improve flavonol content in tomato peel, which will add more nutritional value for tomato.

Methods
Vector construction. The FLS and CHI genes were cloned from red onion. RNA was extracted with the RNeasy plant mini kit from QIAGEN (Valencia, CA, U.S.A). cDNA was made with the Advantage RT-for-PCR Kit from Clontech (Mountain View, CA, U.S.A). The primer sequences for CHI and FLS cloning were CHI forward 5′-ATGGAAGCAGTGACAAAGTT-3′, CHI reverse 5′ T CATGAAAGCACCGGTAACT 3′ FLS forward 5′ ATGGAAGTAGAGAGAGTGCAGGCGA 3′, and FLS reverse 5′ TTACTGAGGAAGTTTATTAATTTTG 3′. The joined two vector with pE1775 47 were transferred to E. coli (DH5α) and Agrobacterium (LBA1775). The PAP1 and FLS vector was constructed following the methods of published reports 34,36 with a 35 s promoter. The plasmids containing FLS and CHI were introduced into A. tumefaciens using the freeze-thaw method 48 . The PAP1 vector harboring PAP1 gene was provided from Vikram et al. 36 49 . Agrobacterium tumefaciens LBA 4404 was used to generate stable transgenic plants. Tomato transformation was performed via the Agrobacterium-mediated transformation method using cotyledon and hypocotyl explants, as described in Park et al. 50 . After inoculation with A. tumefaciens, the plant cultures were maintained at 25 °C under a 16-h photoperiod. After 6 to 8 weeks, regenerated shoots were transferred to a rooting medium for 6 more weeks. The temperature of the greenhouse was maintained within a range of 25 °C to 30 °C. All genes mentioned above were transferred to the Rubion tomato cultivar. DNA isolation and Southern blot analysis. The Southern blot procedure was modified from Wu et al. 51 . Tomato genomic DNA was extracted from leaf tissue using the Qiagen Plant DNA extraction kit. DNA (10 µg) from the CHI lines and the PAP1 lines were digested with XbaI and BamHI, respectively. The digested DNA was separated by electrophoresis and blotted onto a nylon membrane (Zeta-probe GT membrane, Bio-Rad Laboratories, Hercules, CA), following the manufacturer's instructions. The probe for the CHI and PAP1 genes was isolated from vectors harboring each gene 51  HPLC analysis. One gram of peel was frozen in liquid nitrogen and macerated in a round-bottom 15 ml tube with a plastic pestle. The samples were extracted with 4.8 ml of 62.5% methanol and 1.2 ml 6 M HCl for 60 min at 45 °C. The extracts were cooled on ice and sonicated at temperature for 45 min. The samples were centrifuged at 13,000 RPM for 20 min. The supernatant was filtered with a 0.45 μm filter. The extraction procedure was based on Muir et al. 20 .
The HPLC analysis was modified from the published paper 52 . The HPLC system has an autosampler (SpectraSYSTEM AS1000, Thermo Separation Products, San Jose, CA, USA), a pump (HP 1050, Hewlett Packard, Palo Alto, CA, USA), an integrator (HP 3396, Hewlett Packard, Palo Alto, CA, USA), and an UV/VIS detector (Acutect 500, Thermo Separation Products, San Jose, CA, USA). A 5 μL sample was injected into the HPLC column (Discovery BIO Wide Bore C18, 15 cm × 4.6 mm, 5 μm, Supelco, Inc., Bellefonte, PA, USA) with a guard column (Discovery BIO Wide Bore C18, 2 cm × 4 mm, 5 μm, Supelco, Inc., Bellefonte, PA, USA). Stacking genes by crossing. The T2 plants harboring CHI and FLS gene were used as parent. The anthers were removed from unopened flowers one day before anthesis. The next day before noon, the pollens were collected with forceps. The emasculated flowers were pollinated with forceps. After pollination, the forceps were rinsed in a solution of 70% alcohol and wiped with a tissue.
Total flavonoid and anthocyanin content. To record total flavonoid content, the samples before HPLC injection were measured by a 361 nm photospectrometer known as the Nanodrop (Thermo Scientific, Wilmington, DE, USA). Rutin was used as the standard. Anthocyanin content was measured with minor modifications 53 . Tomato peel was ground in volume HCl 0.5% (v/v) in methanol. One volume of chloroform was added to the extract to remove chlorophylls. The mixture was centrifuged at 14,000 g for 1 min. Anthocyanins containing phase were recovered and absorption was determined spectrophotometrically at 544 nm with the Nanodrop. Antioxidant activity. The antioxidant capacity of tomato was measured by the modified 2,20 -azino-bis( 3-ethylbenzthiazoline-6-sulphonic acid) or ABTS method 25,29,30 . Antioxidants were extracted with a 5 ml extraction solution [methonal/ethanol (70/29.5/0.5, v/v/v)] from 1 g of tomato peel samples. The extract containing antioxidants was incubated in darkness at −20 °C overnight. Subsequently, the solution was centrifuged at 1,000 rpm for 2 min. ABTS [(2.5 mM) (Roche Diagnostics, Indianapolis, IN, USA)] stock solution was prepared and about 0.4 g of MnO 2 (Acros Organics, Belgium) was added to the stock solution to generate ABTS radical cation (ABTS*). Excess MnO 2 was removed using a 0.2 mM disk filter (Millipore Corp., Bedford, MA, USA). The ABTS* solution was incubated at 30 °C in a water bath and was diluted to an absorbance of 0.7 (±0.02) at 730 nm using 5 mM phosphate buffer saline [pH 7.4 and ionic strength (150 mM NaCl)]. 100 mL of the extract was added to 1 mL of the ABTS* solution and vortexed for 10 s. The absorbance of the mixture was measured at 730 nm in a spectrophotometer (U-1100, Hitachi Ltd. Japan) after a 1-min reaction period. A Trolox [(6-hydroxy-2,5,7,8-t etramethylchroman-2-carboxyl acid) (Acros Organics, Belgium)] standard curve was prepared using a 0.5-mM stock solution.
Statistical analysis. All data were analyzed using SAS (Version 9.1, Cary, N.C., U.S.A.) 54 . For mean separation, Tukey's test was used. Analysis of variance was performed using the GLM procedure. Significant differences were determined at the 95% confidence level (P < 0.05). Each line had 4-6 plants. Two-to-three pooled tomatoes were collected from each plant for every line 31 .
Data availability. All data generated or analyzed during this study are included in this published article and available from the corresponding author on reasonable request.