• Open Access

Pathway engineering for healthy phytochemicals leading to the production of novel flavonoids in tomato fruit

Authors


* Correspondence (fax +31 317 41 80 94; e-mail elio.schijlen@wur.nl)

Summary

Flavonoids are a large family of plant polyphenolic secondary metabolites. Although they are widespread throughout the plant kingdom, some flavonoid classes are specific for only a few plant species. Due to their presumed health benefits there is growing interest in the development of food crops with tailor-made levels and composition of flavonoids, designed to exert an optimal biological effect. In order to explore the possibilities of flavonoid engineering in tomato fruits, we have targeted this pathway towards classes of potentially healthy flavonoids which are novel for tomato. Using structural flavonoid genes (encoding stilbene synthase, chalcone synthase, chalcone reductase, chalcone isomerase and flavone synthase) from different plant sources, we were able to produce transgenic tomatoes accumulating new phytochemicals. Biochemical analysis showed that the fruit peel contained high levels of stilbenes (resveratrol and piceid), deoxychalcones (butein and isoliquiritigenin), flavones (luteolin-7-glucoside and luteolin aglycon) and flavonols (quercetin glycosides and kaempferol glycosides). Using an online high-performance liquid chromatography (HPLC) antioxidant detection system, we demonstrated that, due to the presence of the novel flavonoids, the transgenic tomato fruits displayed altered antioxidant profiles. In addition, total antioxidant capacity of tomato fruit peel with high levels of flavones and flavonols increased more than threefold. These results on genetic engineering of flavonoids in tomato fruit demonstrate the possibilities to change the levels and composition of health-related polyphenols in a crop plant and provide more insight in the genetic and biochemical regulation of the flavonoid pathway within this worldwide important vegetable.

Introduction

Flavonoids form a large family of polyphenolic plant secondary metabolites. Based on their chemical structure, flavonoids can be divided in several classes, including chalcones, flavanones, dihydroflavonols, flavonols, flavones isoflavonoids and anthocyanins (Koes et al., 1994; Figure 1). Since flavonoids are highly abundant in fruits and vegetables they form an integral part of the human diet (Hertog et al., 1993).

Figure 1.

Schematic overview of the flavonoid pathway. In wild-type tomato fruit peel the flavonoid pathway is active leading to the accumulation of 6-hydroxychalcones and flavonols (solid black arrows). Over-expression of flavonoid pathway genes (black) led to a stimulated flux through the pathway (bold black arrow) and the introduction of new branches of this pathway (bold grey arrows). The dotted arrow represents the pathway leading to anthocyanins, which normally is not active in tomato fruit. Abbreviations: CHS, chalcone synthase; STS, stilbene synthase; CHR, chalcone reductase; CHI, chalcone isomerase; F3H, flavanone 3 hydroxylase; FNS, flavone synthase; FLS, flavonol synthase; DFR, dihydroflavonol reductase; ANS, anthocyanidin synthase.

It is widely accepted that a healthy diet is an important factor in preventing chronic diseases, such as cancer and coronary heart diseases. Increasing evidence from epidemiological and animal feeding studies, as well as from in vitro studies, suggests that dietary polyphenols are likely candidates for the beneficial effects of nutrition on the prevention of chronic diseases (Middleton et al., 2000). Both the basic skeleton (e.g. position of free OH-groups) and modification patterns (e.g. position of glycosylation) appear to determine the bioactivity and bioavailability of flavonoids in humans (Ross and Kasum, 2002). Depending on their structure, flavonoids are ascribed to exert a large diversity of health-related effects, including the scavenging of damaging free radicals and modifying the activity of enzymes. Consequently, the ingestion of different flavonoid species may be more beneficial to human health than ingestion of only a few species. However, in most fruit and vegetable crop plants, only one or a few flavonoid classes are represented by a limited set of flavonoids species. For instance, tomato fruit contains only chalcones (naringenin-chalcone) and flavonols (quercetin and kaempferol glycosides) (Muir et al., 2001).

One particular group of polyphenols for which a positive effect on health has gained increasing support during the last decade (Renaud and de Lorgeril, 1992; Renaud et al., 1998) is that of the flavonoid-related stilbenes, specifically trans-resveratrol and its glucoside piceid (Constant, 1997). Red wine is the main dietary source of stilbenes and the inverse correlation observed between cardiac mortality and regular intake of red wine has largely been attributed to these polyphenolic compounds (Hung et al., 2000). Indeed, stilbenes display numerous biological activities in vitro, such as antioxidant activity, inhibition of inflammation and antitumour activity, mediated through cell cycle regulation, induction of differentiation and apoptosis (reviewed by Pervaiz, 2003). At present, the most promising experimental data point towards a cancer chemopreventive activity of resveratrol in vivo, thereby providing a high clinical potential of this natural compound (Pervaiz, 2003).

In addition to resveratrol, several classes of flavonoids have been shown to protect against cancer in vitro and in various animal models. These flavonoids include deoxychalcones (e.g. isoliquiritigenin and butein), which are only present in legumes (Dixon et al., 1995), and flavones (e.g. apigenin and luteolin), which are more commonly found throughout the plant kingdom, but are present in only a few food crops (Forkmann and Martens, 2001). Inhibition of proliferation by cell cycle arrest, modulation of signal transduction pathways and induction of apoptosis are several possible mechanisms by which these flavonoids exert their chemopreventive activity against several cancer cell lines (Yin et al., 2001; Ko et al., 2002; O’Prey et al., 2003; Van Dross et al., 2003; Takahashi et al., 2004; Wang et al., 2004; Martens and Mithöfer, 2005).

Due to their presumed health benefits there is growing interest in the development of food crops with tailor-made levels and composition of flavonoids, designed to exert an optimal bioavailability or biological effect. Genetic engineering is a promising tool to modify the composition of flavonoids in food crops. Previously, we and others have used several transgenic approaches to increase the levels of the endogenous flavonols quercetin and kaempferol in the peel and flesh of tomato fruits. Ectopic expression of a single CHI gene from Petunia resulted in a tissue-specific increase of total flavonols in the fruit peel. This was mainly due to an accumulation of the flavonols rutin (quercetin 3-rutinoside), quercitrin (quercetin-3-glucoside), and to smaller but still substantial increases in kaempferol glycosides. In these high-flavonol transformants, naringenin chalcone levels were strongly reduced, suggesting that the natural naringenin chalcone pool was utilized by CHI to stimulate the flux through the endogenous flavonoid pathway (Muir et al., 2001; Verhoeyen et al., 2002).

Up-regulation of the flavonoid pathway in tomato fruit flesh, a tissue that normally does not produce flavonoids, was achieved by the introduction and coordinate expression of the maize regulatory genes Lc and C1. Total flavonol content of ripe transgenic tomatoes overexpressing Lc/C1 was about 20-fold higher than that of the control fruits, mainly due to accumulation of kaempferol glycosides. (Bovy et al., 2002; Le Gall et al., 2003). Alternatively, RNAi-mediated suppression of the tomato regulatory gene DET1 resulted in a ‘high pigment’ fruit phenotype, consisting of an up to 3.5-fold increase in flavonoid content in addition to enhanced carotenoid levels (Davuluri et al., 2005). Recently, Giovinazzo et al. (2005) reported on the production of stilbenes in transgenic tomato, suggesting that it is indeed possible to introduce new branches of the flavonoid pathway, at least at its first step, by introducing foreign structural genes. However, branching off further downstream in the pathway by structural genes has not been reported so far.

In order to further explore the possibilities of flavonoid engineering in tomato fruits, we have targeted this pathway towards classes of flavonoids, which are normally not present in tomato. Using structural genes from several plant sources and combinations thereof, we were able to produce transgenic tomatoes accumulating high levels of stilbenes, deoxychalcones or flavones. These fruits displayed altered antioxidant profiles and an up to threefold increase in total antioxidant activity of the fruit peel.

Results and discussion

Strategy to produce novel flavonoids in tomato

In cultivated tomato fruits flavonoids are only found in the peel. The main flavonoids accumulating are naringenin-chalcone and the flavonol rutin (quercetin-3-rutinoside). In fruit flesh, the pathway is hardly active and structural genes required for the production of flavonoids are expressed at very low levels, relative to fruit peel (Muir et al., 2001; Bovy et al., 2002) (Figure 1).

In order to introduce new flavonoid biosynthetic branches, three binary gene constructs were made (Figure 2). To target the pathway towards the production of stilbenes, we introduced a gene construct expressing the grape stilbene synthase (STS) cDNA under control of the constitutive cauliflower mosaic virus double 35S promoter (CaMV d35S). This enzyme acts independent of CHS (Schröder and Schröder, 1990) and therefore it was expected that expression of the corresponding gene alone would result in the accumulation of stilbenes in both peel and flesh.

Figure 2.

Schematic overview of gene constructs. In all gene constructs, transgene expression was under control of the CaMV double 35S promoter (Pd35S) and terminated by the Agrobacterium tumefaciens nos terminator (Tnos). Abbreviations of transgenes: STS, stilbene synthase; CHS, chalcone synthase; CHR, chalcone reductase; CHI, Chalcone isomerase; FNS, flavone synthase; RB, right border; LB, left border.

However, to produce deoxychalcones independently of the endogenous tomato CHS gene-expression levels during the tomato fruit ripening, we reasoned that both chalcone synthase (CHS) and chalcone reductase (CHR) genes were required (Figure 1). Therefore, a double gene construct was made, consisting of the Petunia CHS1 cDNA and the alfalfa CHR cDNA (Ballance and Dixon, 1995), both expressed under control of the d35S promoter.

The third construct aimed at producing flavones, which are produced from flavanones through the action of flavone synthase (FNS). We reasoned that at least two genes are required to target the pathway in tomato peel towards flavones: a gene encoding chalcone isomerase (CHI) to relief the block in the pathway in the fruit peel at CHI (Muir et al., 2001) and an FNS gene to convert the CHI product 2S-naringenin into flavones (Figure 1). So, the third construct used consisted of the Petunia CHI gene (Muir et al., 2001) in combination with the Gerbera FNS-II gene (Martens and Forkmann, 1999), both expressed under control of the d35S promoter.

These three gene constructs were expressed in tomato cultivar Moneymaker through Agrobacterium tumefaciens-mediated transformation. In total 10 STS, 16 CHS/CHR and 17 CHI/FNS primary transformants were obtained carrying one to three copies of these transgenes. Plants were grown to maturity and fruits were harvested when visually ripe.

From each plant at least three fruits were harvested, and peel tissue was screened for flavonoid composition by HPLC (high-performance liquid chromatography). This initial screening revealed that several transgenic lines accumulated the expected compounds in the peel of their fruits: stilbenes in STS, deoxychalcones in CHS/CHR and flavones in CHI/FNS transgenic plants. Transgenic plants producing highest levels of stilbenes (three plants), deoxychalcones (six plants), and flavones (five plants) were selected for detailed molecular and biochemical analyses. Of each selected transgenic plant and of a wild-type plant three cuttings were made and all plants were simultaneously grown to maturity. Ripe fruits from individual plants were pooled and peel tissue was analysed for the composition and levels of flavonoids by HPLC (Figure 3).

Figure 3.

Flavonoid composition in tomato fruit peel (mg/kg fresh weight). Upper panel: levels of the endogenous flavonoids naringenin chalcone (nc) and rutin, determined by HPLC analysis of STS (left), CHS/CHR (middle) and CHI/FNS over-expressing lines compared to wild-type (*n = 2). Middle panel: levels of stilbenes (resveratrol and piceid), deoxychalcones (butein and isoliquiritigenin), flavones (luteolin and luteolin-7-glucoside) and quercetin aglycone found in, respectively, STS (left), CHS/CHR (middle) and CHI/FNS (right) overexpressing tomato fruit peel (*n = 2). Bottom panel: relative expression levels of introduced genes in transgenic and wild-type tomato fruit peel (Log scale; ND, not determined). Note different scales in flavonoid as well as expression levels.

Production of stilbenes

The introduction of a grape STS cDNA in tomato resulted in the accumulation of three stilbenes in fruit peel (Figure 3): the main stilbenes produced were resveratrol aglycon [20.3 ± 9.6 mg per kg fresh weight (FW) peel tissue; mean ± SD, n = 3 plants from best producing line] and its glucoside (piceid, 11.5 ± 4.2 mg per kg FW peel tissue). In addition, a relatively small peak of a second, more polar resveratrol-glycoside (not yet identified) was present in peel tissue of fruits of some plants (data not shown). Since the stilbene synthase introduced competes with endogenous chalcone synthase for their common substrate 4-coumaroyl CoA (Schröder and Schröder, 1990), STS overexpression may lead to a reduced level of flavonoids normally present. Although levels of naringenin chalcone were quite variable, they appeared to be lowered in STS fruit peel as compared to wild-type (WT) peel. This decrease was statistically significant (P = 0.024) when naringenin chalcone levels of more transgenic fruits were compared with data obtained from a larger set of wild-type measurements. In contrast to naringenin chalcone, the flavonol rutin was not significantly affected in STS fruit peel (Figure 3).

Whereas flavonoids normally do not accumulate in fruit flesh, significant levels of stilbenes were detected in this tissue as well (up to 9.3 mg/kg FW). On a whole fruit basis (n = 4 plants) 10.4 ± 1.4 mg/kg FW of total stilbenes were measured. The highest levels of stilbenes were found in peel of T1 STS plants (up to 37.5 mg/kg FW). Interestingly, these levels are significantly higher than those found in different red wines (ranging from 0.5 to 10 mg/L), the most common source of resveratrol (Celotti et al., 1996).

Production of deoxychalcones

Distribution of deoxychalcones in the plant kingdom is mainly restricted to the Leguminoseae, where the production of 6′-deoxychalcones results from a combination of chalcone reductase and chalcone synthase activity (Davies et al., 1998). Overexpression of both CHS and CHR in tomato indeed resulted in the accumulation of deoxychalcones (up to 265 mg total deoxychalcones per kg FW fruit peel). The main deoxychalcones were identified as butein (up to 89 mg/kg FW) and isoliquiritigenin (up to 176 mg/kg FW; Figure 3). The fact that other deoxychalcone-related flavonoid classes were not accumulating in these tomato indicates that the 6′-deoxychalcones synthesized in CHS/CHR tomato fruits were not incorporated into the subsequent 5-deoxy(iso)flavonoid pathway. This can be explained by the lack of endogenous type II CHI activity in tomato, necessary for 5-deoxy flavonoid biosynthesis, which is restricted to leguminous plants (Heller and Forkmann, 1993). Similarly, transgenic Petunia flowers overexpressing CHR accumulate deoxychalcones. Although the accumulation of deoxychalcones in our tomato was less than reported in CHR Petunia (Davies et al., 1998), in both plant species the same efficiency rate, i.e. ratio between hydroxy flavonoids and deoxychalcones (3 : 2), was obtained. The higher deoxychalcone accumulation in Petunia flowers (up to 22 g/kg dry weight) is likely due to a much (tenfold) higher total flavonoid background level of these petals compared to tomato peel. In agreement with results obtained in CHR-overexpressing Petunia and tobacco plants (Joung et al., 2003), overexpression of CHR in tomato resulted in strong competition for common substrates between the endogenous hydroxy flavonoid and the introduced deoxyflavonoid pathway. As a consequence, accumulation of deoxychalcones was accompanied by a clear loss of 6-hydroxyflavonoids: in CHR tomato fruit peel, average naringenin chalcone and rutin levels were reduced to approximately one-third of wild-type levels (Figure 3).

Production of flavones

To produce flavones, first a single gene construct encompassing the Gerbera FNS-II cDNA was introduced into tomato. Of the obtained transgenic plants, three lines accumulated small but significant quantities of flavones [luteolin and luteolin-7-glucoside: up to 6.7 ± 3.9 mg/kg FW, n = 5 plants (F1) from the best line] in their fruit peel. Rutin levels of these FNS-II tomato peel (23.2 ± 3.6 mg/kg FW, n = 5 plants) were approximately 30% of those found in wild-type, suggesting that flavone production was clearly competing with endogenous flavonol synthesis. To further increase the production of flavones, the FNS-II gene was expressed in combination with the Petunia CHI gene, to relieve the block in the pathway at CHI in tomato fruit peel (Muir et al., 2001). These tomatoes accumulated high levels of flavones in their peel, mainly present as luteolin aglycon (up to 340 mg/kg FW) and luteolin 7-glucoside (up to 150 mg/kg FW; Figure 3). In addition to flavones, several flavonols were increased in the peel of these transgenic fruits as compared to wild-type. Based on their retention time and PDA spectrum, these flavonols were identified as quercetin-3,7-trisaccharide, quercetin-3-trisaccharide, rutin, isoquercetin, quercetin-3-rhamnoside, kaempferol-3-rutinoside, kaempferol-3-glucoside (Muir et al., 2001) and quercetin aglycon. The accumulation of quercetin aglycon and isoquercetin (quercetin-3-glucoside), both precursors of rutin, suggests that both the glucosyl transferase and the subsequent rhamnosyl transferase required for the production of rutin (= quercetin-3-rutinoside) become rate limiting in the peel of CHI/FNS transgenic fruits.

Levels of the flavonols rutin and quercetin aglycon were quantified using chemical standards. Quercetin accumulated in peel of these tomatoes up to 34.1 ± 23.1 mg/kg FW (Figure 3). In addition, rutin levels in fruit peel of CHI/FNS-overexpressing tomatoes accumulated up to 900 mg/kg FW, i.e. 16-fold higher than those in wild-type (Figure 3). In analogy to plants overexpressing CHI alone (Muir et al., 2001; Verhoeyen et al., 2002), naringenin chalcone levels in fruit peel were strongly reduced in CHI/FNS plants, indicating that CHI/FNS overexpression leads to an increased flux through the pathway towards flavones and flavonols at the expense of the CHI substrate naringenin chalcone (Figure 3).

Although flavones are ubiquitous in plants, they are not commonly found in crops. On whole fruit basis, transgenic tomatoes reported here contained up to three times higher levels of luteolin than celery (13 mg/kg FW; USDA, 2003), a major natural food source rich in flavones (Haytowitz et al., 2003). Furthermore, the CHI/FNS fruits accumulated quercetin (mainly present as rutin) to levels as high as those found in onions (132 mg/kg FW; USDA, 2003), one of the richest dietary sources of the flavonol quercetin (Haytowitz et al., 2003).

In all transgenic tomatoes described above, the production of new tomato flavonoids in peel extracts corresponded with the presence of transgene mRNA (Figure 3, bottom panel) as determined by real-time RT-PCR analysis. However, no clear correlation could be found between metabolite levels and transgene expression. This indicates that other factors such as substrate availability, expression levels of endogenous genes and competition between enzymes for common substrates may also determine final product levels.

Novel flavonoid antioxidants in transgenic tomato fruit

Since at least part of the presumed health properties of flavonoids has been attributed to their antioxidant activity, we used an HPLC-coupled online antioxidant detection system (Beekwilder et al., 2005) to analyse the antioxidant activity of the individual compounds produced in our transgenic tomatoes. Wild-type and transgenic peel extracts (pooled material from > 50 fruits of three cuttings of the lines with the highest levels of stilbenes, deoxychalcones, or flavones and flavonols) were separated by HPLC and for each eluting metabolite, its PDA spectrum and antioxidant activity were online recorded sequentially (Figure 4). Subsequently, the relative contribution of each metabolite to the total antioxidant activity of the extract, which was calculated by summation of all integrated antioxidant peak areas, was determined and, if appropriate, trolox equivalent antioxidant capacity (TEAC) values were calculated (Table 1).

Figure 4.

Online antioxidant detection. HPLC analysis of peel extracts from wild-type, STS, CHR/CHS and CHI/FNS tomato fruits. Photodiode array detection indicating flavonoids, measured at 360 nm (black line) with corresponding antioxidant detection, visualized by a negative peak in absorption at 412 nm due to reduction of the ABTS radical (red line). Abbreviations: n.c., naringenin chalcone; isoliq., isoliquiritigenin; ND, not determined; q-3-g., quercetin-3-glucoside; K-rut., kaempferol-3-rutinoside; K-3-g, kaempferol-3-glucoside.

Table 1.  Relative contribution of individual flavonoids to the total antioxidant activity of tomato fruit peel extracts
rt (min)CompoundWtSTSCHR/CHSCHI/FNSTEAC value
  1. For each compound eluting from the column, antioxidant activity was quantified by calculating its peak area at 412 nm and expressed relative to the sum of all antioxidant peaks (in percentage). The antioxidant peak at rt 37.4 min comprises of both quercetin and Luteolin (2.9 ± 0.3% and 18.8 ± 1.4% of total antioxidant activity, respectively).

  2. TEAC values of flavonoids were calculated by expressing antioxidant activity as TROLOX equivalents/mole. All values are mean values ± SD; n = 3 extractions (*n = 12, †quercetin TEAC value based on fruit peel extracts from plants (n = 5) over-expressing CHI only, n = 9).

2.5–3.2injectionpeak44.4 ± 0.745.8 ± 0.3 49.5 ± 2.015.8 ± 0.7 
4.7unknown    1.0 ± 0.0  
16.8unknown    1.2 ± 0.1 
17.9unknown 1.6 ± 0.2    
23.4unknown 1.8 ± 0.1 1.1 ± 0.0   
24.3unknown  1.2 ± 0.0   
25.2rutin 5.2 ± 0.1 3.0 ± 0.2  2.7 ± 0.127.9 ± 0.81.83 ± 0.18*
26quercitrin   28.5 ± 0.7 
28.1Kaempferol-rutinoside    3.4 ± 0.1 
29.1Kaempferol-3-glucoside    1.9 ± 0.1 
32.4resveratrol  6.1 ± 0.3  2.57 ± 0.12
37.4Quercetin and Luteolin    4.91 ± 1.82
     21.7 ± 1.71.84 ± 0.12
40.7unknown 2.0 ± 0.4 1.1 ± 0.1   
40.7butein   10.3 ± 1.0 2.01 ± 0.22
41.2naringenin chalcone44.0 ± 2.438.3 ± 0.1 17.0 ± 1.6 0.84 ± 0.11
42.8isoliquiritigenin   12.6 ± 1.0 0.98 ± 0.09
42.3unknown    3.9 ± 0.4  
total98.5 ± 0.495.8 ± 0.896.7 ± 0.4 99.3 ± 0.6  
Total TEAC (µmol/g FW)36.5 ± 4.338.8 ± 0.639.2 ± 1.5124.7 ± 12.7  

In all chromatograms a large polar antioxidant peak was visible at the injection peak (rt ∼3 min). This antioxidant peak comprised polar antioxidant compounds such as vitamin C, glutathione and cysteine and had a similar response area in all chromatograms, suggesting that the levels of these polar antioxidants were not altered in the transgenic plants (Figure 4). In wild-type fruit peel this injection peak contributed to nearly half of the total antioxidant activity, while naringenin chalcone and rutin were the major flavonoid antioxidants contributing to the other half of the antioxidant activity (Figure 4 and Table 1). Compared to the wild-type, STS-overexpressing tomato peel revealed the presence of one small additional antioxidant peak, which could be attributed to resveratrol aglycon. The antioxidant activity of the resveratrol glycosides remained below detection limit (i.e. less than 1% of the total antioxidant activity).

In CHS/CHR-overexpressing fruits, novel deoxychalcones were produced at the expense of naringenin chalcone and, to a lesser extent, of rutin. As these novel deoxychalcones were good antioxidants, their accumulation clearly led to an altered antioxidant profile. Although the levels of butein in these peel extracts were 60% lower than those of isoliquiritigenin (Figure 3), the relative contribution of both compounds to the total antioxidant activity of CHS/CHR peel extracts was more or less equal (Table 1). This result corresponds with the fact that butein is a better antioxidant than isoliquiritigenin (TEAC value of 2.01 vs. 0.98; Table 1). The higher antioxidant capacity of butein compared to isoliquiritigenin is apparently due to the second hydroxylgroup at position 2 in the B-ring of butein (3′,4′,2,4-tetrahydroxychalcone). This is comparable to the difference in antioxidant activity between the flavonols quercetin (2 hydroxyl groups in the B-ring) and kaempferol (1 hydroxyl group in the B-ring), and the flavones luteolin and apigenin, respectively (Rice-Evans et al., 1996). In CHI/FNS-overexpressing fruit peel, both flavonols and flavones were found to be the major antioxidants. In these fruits, rutin, isoquercetin, several kaempferol derivatives, quercetin and luteolin contributed for 83.4% of total antioxidant activity (Table 1). Likewise, polar compounds eluting in the injection peak, though not changed in absolute amount compared to WT, contributed only 15.8% to the total antioxidant activity. Flavones and flavonols are well-known antioxidants and share several structural features which contribute to their antioxidant activity: both flavonoid classes have a phenolic B-ring which can be hydroxylated at the 3′, 4′ and/or 5′ position and a C2-C3 double bond. In addition, flavonol aglycons contain a hydroxyl group at the C3 position, which makes flavonols even better antioxidants than flavones. (Table 1; Rice-Evans et al., 1996).

The total antioxidant activity of the transgenic peel extracts was determined in two ways: (i) by calculating the sum of all individual antioxidant peaks eluting from the HPLC column and (ii) by analysing the same crude extracts using the TEAC assay in 96-wells plates. As shown in Figure 5, both approaches produced comparable results and indicated that the total aqueous-methanol soluble antioxidant activity was statistically significantly increased in CHI/FNS fruit peel (up to 3.5-fold). The difference in increased antioxidant activity between both methods could be a result of the presence of other antioxidant compounds, such as polar carotenoids, within the abstract. Within the HPLC-based antioxidant detection system, improper elution of these compounds could result in higher relative differences between the samples. Despite the very good antioxidant activity of resveratrol (TEAC value is 2.57, Table 1), the level of this compound in our transgenic STS plants was still too low to have a significant impact on the total antioxidant activity. Recently, Giovinazzo et al. (2005) used a similar approach to constitutively express the grape STS gene in tomato cultivar Moneymaker. Their best performing STS plants contained levels of stilbenes that were up to fivefold higher than in our study. A significant increase in total antioxidant activity (measured as ABTS decoloration by hydrophilic and lipophilic antioxidants in aqueous and organic phase separately) of the transgenic fruits could be observed in their highest expressing lines.

Figure 5.

Total antioxidant activity. Total antioxidant activity of wild-type, STS, CHS/CHR and CHI/FNS overexpressing tomato fruit peel. Total antioxidant activity was measured in two ways: using the TEAC assay with peel extracts in 96-well plates (striped bars) and by calculating the sum of all antioxidant peaks eluting from the HPLC column after separation of the same peel extracts (solid bars).

In our CHS/CHR plants high levels of the deoxychalcones butein and isoliquiritigenin were produced. Although these deoxychalcones appeared to be good antioxidants, their accumulation was clearly at the expense of hydroxy flavonoids resulting in a decrease in both naringenin chalcone and rutin. As a consequence, the total antioxidant activity of fruit peel of CHS/CHR tomato lines was equal to that of wild-type plants.

Fruit phenotype and inheritance of transgenic plants

After determination of the transgene copy number, the offspring of single copy plants was tested for the inheritance of the transgene and the stability of the ‘novel flavonoid’ trait. In subsequent generations tested (F1 and F2), the ‘novel flavonoid’ phenotype segregated with the corresponding transgene. All transgenes, except STS, were inherited according to Mendelian law: 75% of the F1 offspring contained the transgene. Inheritance of the STS trait, however, occurred at a 1 : 1 ratio (P-value < 0.05, n = 20), which is less than the expected ratio of 3 : 1. This suggests that homozygote STS offspring may be lethal.

The vegetative phenotype of all transgenic plants was indistinguishable from the parental variety. In fruits, however, some phenotypic changes were observed. In STS-overexpressing plants, reduced seed set and occasionally parthenocarpic fruits with reduced fruit size were observed in lines with highest stilbene levels. This is in line with the observation of Giovinazzo et al. (2005) that all fruits of their STS-overexpressing tomatoes, which accumulated up to fivefold higher stilbene levels than our STS tomatoes, were seedless. The reduced fertility of STS-overexpressing plants may be due to a, yet unknown, biological effect of stilbenes, or, alternatively, may be caused by reduced flavonoid levels as a result of substrate competition between endogenous chalcone synthases and the introduced stilbene synthase. Indeed, reduced fertility due to a reduction in flavonoid levels has been reported in Petunia plants in which flavonoid biosynthesis was blocked through antisense suppression of chalcone synthase (Van der Meer et al., 1992; Ylstra et al., 1992) as well as in tobacco plants overexpressing stilbene synthase (Fischer et al., 1997). In the latter case, fertility could be restored by exogenous application of flavonoids to the stigma, suggesting that a lack of endogenous flavonoids rather than the accumulation of stilbenes lead to reduced fertility. Clearly additional research is needed to unravel the mechanisms underlying the reduced fertility and seedless phenotype of STS-overexpressing tomato plants.

Occasionally, in some CHS/CHR- and CHI/FNS-overexpressing lines, the colour of red ripe fruits was changed from orange-red to more pink-red (result not shown). This colour change is likely due to the decrease in levels of the yellow-coloured flavonoid naringenin chalcone, which normally accumulates in high levels in the epidermal layers of wild-type tomato fruits (Figure 3; Hunt and Baker, 1980) resulting from the increased flux towards deoxychalcones in case of CHS/CHR, and towards flavones and flavonols in case of CHI/FNS. A similar pink-red phenotype has been described in CHI-overexpressing tomato lines accumulating high levels of flavonols but lacking naringenin chalcone (Muir et al., 2001).

In relation to its potential value for other phenotypic characteristics, also the presence of different flavonoids in other tissues would be of potential interest for future research, for example, in view of putative increased resistance.

Conclusions

We explored and demonstrated the possibilities to target the flavonoid pathway in tomato towards classes of flavonoids or stilbenes that are normally not present in tomato fruit. In this report we focused on producing phenolic compounds that have potential health effects but are not very abundant in the Western diet, e.g. stilbenes, deoxychalcones and flavones. We successfully produced significant levels of these target compounds in tomato peel. These novel flavonoids are good antioxidants and peel extracts of the obtained transgenic fruits showed altered antioxidant profiles. In addition these flavonoids may show different bioavailability and or bioactivity as compared to endogenous flavonoids normally found within most cultivated tomatoes. In case of CHI/FNS, total antioxidant activity was increased up to threefold. To our knowledge, this is the first report describing the production of deoxychalcones and flavones in an edible crop plant. These genetically engineered tomatoes may be very useful to study the potential health benefits of specific natural flavonoids present in the same tomato-based food matrix.

Experimental procedures

Gene cloning

A full length cDNA encoding stilbene synthase (STS) (X76892; Sparvoli et al., 1994) was isolated from ripe grape skin tissue Vitis vinifera cv. Lavelee by RT-PCR using the following oligonucleotides: 3′ end oligo: 5′-AATACCTTACTCCTATTCAACA-3′; 5′ end oligo: 5′-GATCAATGGCTTCAGTCGAG-3′. The amplified cDNA fragment was subcloned into pGEM-T-Easy Vector (Promega, Madison, WI) according to standard protocols. Upstream of the STS cDNA, a BamHI restriction site was introduced by PCR, followed by subcloning of the STS coding region as an NcoI/SalI fragment into the pMOGEN18 (Sijmons et al., 1990) derivative pHB003. The resulting plasmid pHB022 encompasses a fusion of the double CaMV 35S promoter (Pd35S), an Alfalfa Mosaic Virus translation enhancing 5′UTR, the inserted STS cDNA sequence, and the Agrobacterium tumefaciens nopaline synthase terminator (Tnos).

An alfalfa (Medicago sativa) cDNA sequence encoding chalcone reductase (CHR; Ballance and Dixon, 1995) was obtained as original clone. Cloning of the full length cDNA sequence encoding the Petunia chalcone synthase (CHS; Koes et al., 1989) was described earlier (Colliver et al., 2002). In order to create the double gene construct Pd35S-CHS-Tnos//Pd35S-CHR-Tnos, three initial single gene constructs had to be made. First, the plasmid FLAP600 was digested by BamHI followed by self ligation of the 5119-bp fragment resulting in a plasmid containing Pd35S-CHS-Tnos, called pHEAP28. A second gene construct including Pd35S-CHS-Tnos (pHEAP29) was obtained after exchanging the e8 promoter of plasmid pFlap600 by the d35S promoter derived from EcoRI/BamHI digestion of pFlap50 (Colliver et al., 2002). Third, a single gene construct encompassing Pd35S-CHR-Tnos was obtained after PCR amplification of the original CHR cDNA using oligonucleotides creating a 5′BamHI and a 3′SalI restriction site. Subsequently, the CHS cDNA of pHEAP28 was replaced by the BamHI/SalI CHR cDNA fragment resulting in plasmid pHEAP33. Finally, the double gene construct containing Pd35S-CHS-Tnos//Pd35S-CHR-Tnos was created by ligating the Pd35S-CHS-Tnos from pHEAP29 as a NotI/AscI fragment into plasmid pHEAP33. The resulting double gene construct was called pHEAP34.

Cloning of the Petunia chalcone isomerase (Chi-a; Van Tunen et al., 1988) has been described previously (Muir et al., 2001). The Gerbera hybrida cDNA clone encoding flavone synthase-II (FNS-II; Martens and Forkmann, 1999) was obtained as original gene construct. After introduction of proper restriction sites by PCR, the FNS-II encoding cDNA was cloned as BamHI/SalI fragment into pFLAP50 (Muir et al., 2001). The resulting gene construct, pFLAPFNS encompass fusions of the Pd35S promoter, followed by the cDNA sequence and the Agrobacterium tumefaciens nopaline synthase terminator (Tnos) terminator. To create a double gene construct Pd35S-CHI-Tnos//Pd35S-FNSII-Tnos, the EcoRI/BamHI fragment of the former construct, encompassing 35S-FNSII-Tnos, was ligated into the 4577-bp EcoRI/BamHI fragment derived from FLAP600 (Colliver et al., 2002).

Finally, single and double gene constructs were transferred as PacI/AscI fragment into the binary vector pBBC50 (Muir et al., 2001). The resulting gene constructs are shown in Figure 2.

Plant transformation and growth

Binary plasmids were transferred into Agrobacterium tumefaciens strain COR308 (Hamilton et al., 1995) by a freeze-thaw method (Höfgen and Willmitzer, 1988).

To obtain transgenic tomato plants (Lycopersicon esculentum cv. Moneymaker), 8-day-old hypocotyls were used for Agrobacterium-mediated transformation according to Fillatti et al. (1987). Kanamycin-resistant shoots were grown from March 2003 through October 2004 under controlled glasshouse conditions in Wageningen, the Netherlands. Plants were grown on rock-wool plugs connected to an automatic irrigation system comparable with standard commercial cultivation conditions with a minimum temperature set point of 19 °C during the day and 17 °C at night. To compensate for the lack of sunlight, between autumn and spring, supplementary high pressure sodium light was provided with a minimum light intensity of, on average, 17 W/m2 at a photoperiod of 16 h light: 8 h dark.

All plants were self-pollinated to produce fruits and offspring. The transgenic status of tomato plants was confirmed by PCR analysis on young leaf material with gene specific-primers according to manufacturers protocol (X-amp PCR, Sigma-Aldrich, Zwijndrecht, Netherlands).

High-molecular-weight genomic DNA was isolated from young leaves of tomato, as described by Dellaporta et al. (1983). Insert copy number of transgenic plants was determined by Southern blot hybridization, according to manufacturer's protocol (DIG labelling, Roche, Mannhein, Germany) using 10 µg Bgl II-digested genomic DNA and npt-II as probe.

For further biochemical and gene expression analyses, fruits were harvested when visually ripe. From each plant at least three fruits were pooled for extraction to minimize sample variation. Fruit peel (approximately 2 mm thick, consisting of the cuticula, epidermis and subepidermis) and fruit flesh (i.e. columella; jelly parenchym; seeds excluded) were collected separately, frozen in liquid nitrogen and stored at −80 °C until further analysis.

HPLC analysis of flavonoids

Flavonoids were determined after extraction in 75% aqueous methanol with 15 min of sonication. Compounds were separated on a C18 reverse phase HPLC column (Luna C18(2), 3 µm, 150 × 4 mm, Phenomenex, Torrance, CA, USA) at 40 °C, and analysed by photodiode array detection (type 996, Waters, the Netherlands). A gradient of 5%−50% acetonitrile in 0.1% tri-fluoro acetic acid was used as mobile phase. Absorbance spectra (240–600 nm) and retention times of eluting compounds were used for identification by comparison with authentic flavonoid standards (Apin Chemicals, Abingdon, UK).

Antioxidant activity

Online HPLC detection of antioxidant activity of corresponding eluting compounds was based on the postcolumn antioxidant reaction system firstly described by Koleva et al. (2000) and subsequently modified by Beekwilder et al. (2005). Shortly, compounds eluted from the HPLC column and passed through the PDA detector were allowed to react for 30 s with a buffered solution of ABTS [2,2′-azinobis(3-ethylbenzothiazoline-6-sulphonic acid); Roche] cation radicals online, before passing through a second detector, monitoring the ABTS radicals at 412 nm. Antioxidant activity of compounds was expressed as trolox (a water-soluble vitamin E analogue) equivalents, using a calibration curve of trolox injected in the same HPLC system.

Total antioxidant activity of the same extracts was measured by mixing 10 µL of extract with 90 µL of ABTS reagent (pH 7.4, final concentration: 400 µM). Decrease of absorbance at 412 nm was immediately determined using a spectra fluor plate reader (Tecan, Grödig, Austria) and expressed as trolox equivalents.

RNA isolation

Total RNA was isolated from tomato fruit tissues by hot phenol extraction and lithium chloride precipitation (Pawlowski et al., 1994). After DNAse-I treatment (Boehringer Mannheim, Germany) followed by RNeasy column purification (Qiagen, Benelux, Venlo, Netherlands), total RNA yield was measured by absorption at 260 nm (nanodrop) and evaluated on a 1% TAE agarose gel.

RT-PCR gene expression analysis

Real-time quantitative (RT) PCR analysis was performed to test expression levels of introduced transgenes. TaqMan sequence detection primers (Applied Biosystems, Warrington, UK) were designed using SDS 1.9 software (Table 2). Two-microgram total RNA was used for cDNA synthesis by Superscript II reverse transcriptase (Invitrogen) in a 100-µL final volume according to standard protocol. Expression levels of each gene were measured as triplicate reactions, performed with the same cDNA pool, in presence of fluorescent dye (SYBR-Green) using an ABI 7700 sequence detection system. The constitutively expressed mRNA encoding tomato fruit abscisic stress ripening protein-1 (L08255) was used as internal reference. This gene was previously shown to be constitutively expressed during tomato fruit ripening (Bovy et al., 2002). Oligonucleotides used for detection of expression levels from transgenes are given in Table 2. Calculations of each sample were carried out according to the comparative Ct method (PE Applied Biosystems).

Table 2.  Oligo nucleotides used for TaqMan RT-PCR analysis
OligoSequence 5′–3′Gene IDSource
CHR-536Fttg cca ctg ttc ttc cag cchalcone reductase (CHR 7) U13925Medicago sativa
CHR-586Rgcc atg caa ggt tca ttt cc  
FNS-197Fcgc tct ctc cac tcg cta cgflavone synthase II (CYP93B2) AF156976Gerbera hybrida
FNS-247Ract gag ccg aga cgg agg t  
STS-530Fcct tcg aac cgc taa gga tctstilbene synthase (StSy) X76892Vitis vinifera
STS-582Rcaa gaa ctc gtg ctc ctg cat  
CHS-516Fggt tct tcg gtt agc caa ggachalcone synthase (CHS) AF233638Petunia hybrida
CHS-620Rgta tca ttt ggc cca cgg aa  
CHI-338Fggc acg acc ctc atc atc achalcone isomerase A (CHI A) AF233637Petunia hybrida
CHI-405Ragg tgg cgg aga att gtg tt  
ASRIP-88Fcct gtt cca cca caa gga caafruit abscisic stress ripening protein-1 (TOMASRIP) L08255Lycopersicon esculentum
ASRIP-186Rgtg cca agt tta ccg att tgc  

Acknowledgements

The authors wish to thank Dr Richard Dixon for providing the CHR cDNA. This research was supported by the EU as part of the PROFOOD project (QLK1-CT-2001-01080).

Ancillary