• Open Access

Altered leaf colour is associated with increased superoxide-scavenging activity in aureusidin-producing transgenic plants


(Tel +1 208 258 6065; fax +1 208 258 6027; email crommens@simplot.com)


The health-promoting property of diets rich in fruits and vegetables is based, in part, on the additive and synergistic effects of multiple antioxidants. In an attempt to further enhance food quality, we introduced into crops the capability to synthesize a yellow antioxidant, aureusidin, that is normally produced only by some ornamental plants. For this purpose, the snapdragon (Antirrhinum majus) chalcone 4′-O-glucosyltransferase (Am4’CGT) and aureusidin synthase (AmAs1) genes, which catalyse the synthesis of aureusidin from chalcone, were expressed in tobacco (Nicotiana tabacum) and lettuce (Lactuca sativa) plants that displayed a functionally active chalcone/flavanone biosynthetic pathway. Leaves of the resulting transgenic plants developed a yellow hue and displayed higher superoxide dismutase (SOD) inhibiting and oxygen radical absorbance capacity (ORAC) activities than control leaves. Our results suggest that the nutritional qualities of leafy vegetables can be enhanced through the introduction of aurone biosynthetic pathways.


Aurones are flavonoids with a 5-membered C-ring that provide a bright yellow colour to the petals of some varieties of snapdragon (Antirrhinum), morning glory (Ipomoea), Dahlia and Coreopsis (Saito and Ine, 1990; Iwashina, 2000). An analysis of flower colour variation in natural populations of snapdragon suggests that aurones play a role in fertilization and seed set by attracting pollinators (Whibley et al., 2006). Indeed, the patterning of aurone pigmentation is thought to provide a nectar guide for pollinating bumblebees (Harborne and Smith, 1978; Lunau et al., 1996). In addition to this role in pigmentation, aurones have been described as phytoalexins that are used by the plant as defence agents against various pathogens; they were found to exhibit antiviral, antiparasitic and antifungal activities (Boumendjel, 2003). Previously, a two-step mechanism involving the oxidation of isoliquiritigenin by a hydrogen peroxide (H2O2)-dependent peroxidase (PRX), followed by dehydration of the intermediate compound to form aurone 4′,6-dihydroxyaurone, was proposed for aurone biosynthesis in soybean (Soja hispida) seedlings (Rathmell and Bendall, 1972; Wong, 1966). In snapdragon, the aurone aureusidin-6-O-glucoside (AOG) is produced by glucosylation of 2′,4′,6′,4-tetrahydroxychalcone (naringenin chalcone), which facilitates the transport of this compound from cytoplasm to vacuole (Ono et al., 2006), followed by cyclization of the carbon bridge. The proteins involved in these reactions are chalcone 4′-O-glucosyltransferase (Am4’CGT) and the copper-containing glycoprotein aureusidin synthase (AmAs1) (Nakayama et al., 2000), respectively. Ectopic expression of the Am4’CGT and AmAs1 genes in the related plant species Torenia hybrid resulted in the petal-specific formation of trace amounts of AOG (Ono et al., 2006). The simultaneous silencing of anthocyanin biosynthesis increased AOG formation to levels that are visible as a yellow hue (Ono et al., 2006).

Although commercial interests in aurones are currently limited to how these compounds affect flower colour, their antioxidant activities suggest future medicinal applications as well (Boumendjel, 2003; Detsi et al., 2009; Milovanovic et al., 2002). Indeed, the 3′,4′,6,7 tetrahydroxyaurone from Devil’s Beggarticks (Bidens frondosa) is more effective at scavenging free radicals than vitamin C, vitamin E and resveratrol (Venkateswarlu et al., 2004). The ability to produce aurones synthetically (Rathmell and Bendall, 1972; Wong, 1966) opens up the way to use them as dietary supplements. However, there is a preference to use naturally produced compounds, because supplement use has been linked to increased mortality (Bjelakovic and Gluud, 2007). Recent studies have shown that naturally produced compounds are more efficiently retained by the body than their synthetic counterparts.

The first report on aurone-containing foods relates to the aromatic herb bristly ox-tongue (Picris echoides), which is used in traditional medicine for the treatment of ingestion and intestinal parasites (Milovanovic et al., 2002). Studies on the therapeutic potential of aurones themselves were initiated only recently (Haudecoeur and Boumendjel, 2012; Zwergel et al., 2012). Given the outcome of this research, we felt urged to transfer the aurone biosynthetic pathway (Ono et al., 2006) from ornamental flowers to the leaves of crop plants. Our results demonstrate that overexpression of two aromatic biosynthesis genes, Am4’CGT and AmAs1, altered the colour of leaves and also enhanced their antioxidant activity.


Constitutive expression of the snapdragon Am4’CGT and AmAs1 genes triggers flower-specific aureusidin formation in tobacco

The two snapdragon genes that catalyse aureusidin biosynthesis, Am4’CGT and AmAs1, were operably linked to the strong near-constitutive promoter of figwort mosaic virus (FMV). Insertion of the resulting expression cassettes into the T-DNA of a vector carrying the phosphomannose isomerase (pmi) gene yielded pSIM1251 (Figure 1a). Agrobacterium-mediated transformation of tobacco (Nicotiana tabacum) produced 25 mannose-resistant plants that, upon PCR-based confirmation of the presence of the three transgenes, were propagated to produce pSIM1251 lines. The original vector carrying only a marker gene was used to generate transgenic controls. One plant of each line was transferred to the glasshouse and allowed to mature at a constant temperature of 28 ± 2 °C. Experimental lines appeared phenotypically similar to their transgenic controls, except for flower colour. This new colour was unusual for tobacco but resembled that of flowers of the untransformed snapdragon variety ‘Rocket Yellow’ used as gene source (Figure 1b,c). High performance liquid chromatography (HPLC) analysis demonstrated that the yellow transgenic flowers contained a compound that is not naturally produced in tobacco (Figure 1d, peak 1’). This compound was confirmed to have the same retention time (RT) and mass as the predominant flavonoid of snapdragon flowers, which is AOG (also named 4,6,3′4′-tetrahydroxyaurone-6-O-glucoside, AOG) (Figure 1e, peak 1’, and Table 1). MS/MS analysis of peak 1’, which exhibited an [M + H]- ion at m/z 449, yielded MS2 fragmentation at m/z 287 because of the loss of 162 atomic mass units (amu), corresponding to one glucose moiety (Figure 1f). Peak 1 is unlikely due to the saturation of peak 1’ as the two peaks are well separated and not tailing. In snapdragon, a trace amount of molecular ion m/z 465 was revealed to co-elute with broad peak 1’, which was fragmented at m/z 287 (data not shown) and tentatively identified as bracteatin-6-O-glucoside. Additionally, the mass spectra and UV/Vis features of peak 1, a compound identified in snapdragon but not in the transgenic tobacco, were identical to those of peak 1’ (Figure S1a,b) and possibly corresponded to a structural isomer of AOG. Our results demonstrate that the ability to produce AOG can be transferred across family boundaries, from a scrophulariaceous to a solanaceous plant species, through heterologous expression of genes involved in the last two biosynthetic steps. The leaves of pSIM1251 tobacco plants did not contain detectable levels of AOG, indicating that the gene transfer had not broadened the tissue specificity of aurone formation beyond that of snapdragon.

Figure 1.

 Aurone formation in pSIM1251 tobacco. Diagram of the pSIM1251 transfer DNA. B = T-DNA border, P = promoter, T = terminator (a). Flower of transgenic tobacco (b) and untransformed snapdragon (c). HPLC chromatogram of pSIM1251 tobacco (d) and snapdragon flowers (e) showing aureusidin-6-O-glucoside (AOG) eluting as peaks 1 and 1’ at 400 nm. Mass spectra and MS-MS fragmentation of m/z 449 of AOG from snapdragon (f). mAU, milliabsorbance units.

Table 1.  Analyses of flavonoids and anthocyanins in aurone extracts of transgenic tobacco leaves
Transgenic lines and controlsInd.line no.HPLC chromatogram at different DAD wavelengths
Aurone at 400 nmAnthocyanin at 520 nmFlavonoids at 360 nm
Peak 1 2.6 minPeak 1’ 3.9 minPeak 2 2.6 minPeak 3 2.6 minPeak 4 5.1 minPeak 5 5.9 minPeak 6 6.2 minPeak 7 9.1 minPeak 8 9.7 minPeak 1 0.7 min
NC deriv
  1. UD, undetectable.

  2. Tentative peak identification: Peaks 1 and 1’, aureusidin-6-O-glucose (AOG); Peak 2, unidentified anthocyanin (NID); Peak 3, cyanidin-rutinose (Cyn-ru); Peak 4, pentahydroxy flavone-glucose (PHF-glu); Peak 5, naringenin chalcone (NC) derivative; Peak 6, naringenin chalcone diglucoside (NC-diglu); Peak 7, naringenin chalcone glucose (NC-glu); Peak 8, tetrahydroxy methoxy chalcone glucose (TMC-glu); Peak 9, naringenin chalcone (NC).

 Wild type1UDUDUDUD1.6UDUD0.005UD0.0006
9UDTraceUDUD1.49 UD0.005Trace0.0015
11UDTraceUDUD0.89 UD0.002Trace0.006

Chalcone accumulation promotes aureusidin formation in the leaves and stems of transgenic tobacco plants

A modified strategy was employed to overcome the flower-limited formation of AOG in tobacco. As a first step, to promote the formation of flavonoid AOG precursors, wild-type tobacco plants were transformed with pSIM646. This vector contains the potato (Solanum tuberosum) transcription factor StMtf1M gene fused to the strong promoter of the potato Ubi7 gene (Rommens et al., 2008). The resulting overexpression of the anthocyanin-associated StMtf1M gene produced deep purple transgenic plants (646 tobacco; Figure 2a), which were demonstrated by LC/MS to contain large amounts of anthocyanins. Two compounds were not fully separated by LC (Figure 2b,c, peaks 2 and 3) and had absorption maxima at 518 nm. Using UV/Vis spectra and MS fragmentation, peak 2 was tentatively identified as pelargonidin aglycone (molecular ion at m/z 271, see Figure S2a), and peak 3 was identified as cyanidin-3-O-rutinose (molecular ion at m/z 595), which could be fragmented to m/z 499 (loss of a rhamnose moiety, 146 amu) and 287 (loss of rutinose, 308 amu) (Table 1 and Figure S2b). Furthermore, concentrations of a pentahydroxy flavone-glucose, tentatively identified as quercetin-3-O-glucoside, were higher in the StMtf1M plants than in their transgenic controls (Figure 2d, peak 4; Table 1). The mass spectra of peak 4 at 5.1 min showed a molecular ion at m/z 465 and the MS/MS fragment at m/z 272.9 (data not shown).

Figure 2.

 Overexpression of StMtf1m in potato. Typical phenotype of 646 tobacco leaves (top) and flowers (bottom) (a). Extracts used to generate HPLC chromatograms were from leaves of untransformed tobacco plants recorded at 520 nm (b) and of 646 plants recorded at 520 nm for anthocyanins (c) and 360 nm for flavonoids (d). Peaks: (2) unidentified anthocyanin at 2.4 min, (3) cyanidin-3-O-glucoside at 2.6 min, (4) pentahydroxy flavone-glucoside at 5.1 min. For quantitative analysis, see Table 1. RT, retention time.

To partially suppress anthocyanin formation and, instead, promote the accumulation of flavonoid intermediates, pSIM646 plants were retransformed with pSIM1252 T-DNA, which carries a silencing cassette targeting the chalcone isomerase (Chi) gene, together with the hygromycin selection marker gene hpt. This second modification altered plant colour from deep purple to green with a slight purple hue. The 646/1252 plants accumulated naringenin chalcone and several glycosylated naringenin chalcones in all tested (5) independent lines. The HPLC chromatograms of anthocyanin and flavonoid profiles are shown in Figure 3a–c, and the quantitative amounts from three representative lines are presented in Table 1. The presence of naringenin chalcone and its glycosylated derivatives was also investigated by MS/MS analysis. The positive ion electrospray product ion tandem mass spectra of m/z 435, 465 and 272.9 are shown in Figure S3a–c. Peak 5, eluting at 5.9 min, showed a molecular ion at m/z 272.9, which corresponds to aglycone naringenin chalcone, but no confirmed parent molecular ion was detected. Peak 6, eluting at 6.2 min, was tentatively identified as naringenin chalcone diglucoside with m/z = 597 and MS2 ion at 272.9 owing to the loss of 324 amu, corresponding to two glucose moieties. Peak 7, eluting at 9.1 min with a [M+] peak at 435 and a fragment of 272.9 obtained after the loss of 162 amu (hexose moiety), was identified as naringenin chalcone glucoside. Peak 8, eluting at 9.7 min with an m/z of 465 and MS2 fragment of 303 because of the loss of glucose (−162 amu), was attributed to tetrahydroxy methoxychalcone glucoside. Peak 9, eluting at 10.7 min, was identified as naringenin chalcone, according to the mass spectrum with an m/z of 272.9 and UV absorption maximum (Figure 3d).

Figure 3.

 HPLC chromatograms of 646/1252 tobacco plants. Extracts were obtained from the leaves of 646/1252 plants recorded at 520 nm for anthocyanin (a), untransformed plants at 360 nm for flavonoids (b) and 646/1252 plants at 360 nm (c). The UV spectrum of peak 9 is shown in (d). Peaks: (4) pentahydroxy flavone-glucose, (5) naringenin chalcone derivative, (6) naringenin chalcone-diglucose, (7) naringenin chalcone glucose, (8) tetrahydroxy methoxychalcone glucose and (9) naringenin chalcone. Quantitative analyses are summarized in Table 1.

The naringenin chalcone-rich plants were transformed a third time with the T-DNA of pSIM1257, which carry the aurone biosynthetic genes (similar to pSIM1251, except that pmi was replaced with the hygromycin phosphotransferase selectable marker gene, hpt). Transformed cells proliferated only on tissue culture media supplemented with naringenin chalcone and developed bright yellow calli, suggesting an effective conversion of the plant-produced compound to AOG. Subsequent regeneration produced 20 yellow-green shoots that were markedly different from the green-purple shoots of parental lines. Upon planting in soil, all these shoots started to accumulate some purple pigments, indicative a lingering Chi activity, so that leaves of triply transformed plants appeared bronze-green (Figure 4a–c). Unlike the pink or purple flowers of control and parental lines (Figure 4d,f), these plants produced yellow-orange coloured flowers (Figure 4g). The bronze-green leaves of 646/1252/1257 lacked detectable amounts of the anthocyanins and flavanones that were abundant in 646/1252 lines expressing the StMtf1M gene and partially silenced for Chi (Figure 4h,i). Confirming our earlier assumption, these compounds were converted into AOG. The yellow aurone compound had accumulated in leaves to levels nearly two-thirds those in snapdragon flowers (Figure 4j, peak 1 and 1’ and Table 1). The parental 646/1252 line lacked these peaks (Figure 4k–m). Interestingly, overexpressing the Am4CGT and AmAs1 genes in deep purple plants (646/1257) without silencing Chi produced only trace amounts of AOG (Figure 4n). The MS/MS fragmentation (Figure 4o) of peak 1' at a RT of 3.9 min and UV-diode array detection (UV-DAD) profile of AOG of 646/1252/1257 (Figure 4p,q) were identical to both snapdragon AOG and commercially available maritimein (3′,4′,6,7-tetrahydroxy-6-O-glycosylaurone or maritimetin-6-O-glucoside). The trace amount of compound with molecular ion at m/z 465 was co-eluting with AOG, peak 1’ in both 646/1252/1257 leaves and snapdragon flower which has the same UV maximum as that of AOG tentatively and identified as bracteatin-6-glucose.

Figure 4.

 Aurone formation in transgenic tobacco. Phenotype of a glasshouse-grown 646/1252/1251 plant (a) and individual leaves of controls (b and c, left) and 646/1252/1251 plants (b and c, right). Flowers are shown for control (d), 646 (e), 646/1252 (f) and 646/1252/1257 (g) plants. An HPLC chromatogram of 646/1252/1257 recorded at 520 nm for anthocyanin (h), 360 nm for flavonoids (i) and 400 nm for aurone (j). Compounds eluting at 2.5 and 3.9 min and denoted as peaks 1 and 1’, respectively, were both identified as aureusidin-6-O-glucoside (AOG) and compared to wild-type, 646, 646/1252 and 646/1257 plants (k–n). Mass spectra and MS/MS fragmentation of AOG (m/z = 449) in the positive mode (o). Comparison of UV spectra of AOG from snapdragon flower (p) and the 646/1252/1257 plant (q).

Aureusidin formation in lettuce plants expressing the Am4’CGT and AmAs1 genes

The lettuce Lactuca sativa cultivar ‘Eruption’ produces purple leaves. Plants of this variety were transformed to express the Am4’CGT and AmAs1 genes (pSIM1610). Upon transfer to the glasshouse, leaf colour turned bronze-green (Figure 5a,b). The leaves of these transgenic plants (1610) were demonstrated by LC/MS to contain a large amount of AOG, which is not present in the untransformed control (Figure 5c,d, peak 2). The associated peak co-eluted with an anthocyanin compound that also accumulated in untransformed plants, as shown in the HPLC chromatogram in Figure S4a,d and the UV spectrum in Figure S4e. LC/MS-MS detection in positive ionization modes was used to obtain more information on compound structure. The co-eluted compound in peak 2 was attributed to cyanidin-3-(6′-malonyl) glucoside, based on MS/MS fragmentation (m/z 535, MS2 fragments, 449 and 287 corresponding to loss of first 86 amu, i.e., the malonyl moiety, and then 162 amu, i.e., the hexose moiety) (Figure S5a) and the UV spectrum (Figure S5b). AOG and cyanidin-3-(6′-malonyl)-glucoside were quantified as shown in Table 2. Silencing of the Dfr (dihydroflavonol 4-reductase) gene (pSIM1618) in wild-type lettuce almost completely blocked the formation of this anthocyanin. As expected, retransformation of the Dfr-silenced plants (1618) with the Am4’CGT and AmAs1 genes (pSIM1610) resulted in AOG formation. The amount of AOG was slightly higher in 1610/1618 lettuce plants than in plants that were not silenced for Dfr (Figure 5e,f). LC/MS and MS-MS data tentatively identified three main flavonoids (denoted as peaks 3, 4 and 5) in wild-type lettuce as a quercetin derivative (m/z 479, product ion 303), kaempferol glucoside (m/z 463, product ions 463 & 287) and quercetin-3-(6′-malonyl) glucoside (m/z 551, product ions 465 & 303), respectively. The amount of flavonoids did not change significantly upon the overexpression of Am4’CGT and AmAs1, regardless of whether or not Dfr was silenced. The HPLC chromatograms are illustrated in Figure 6a–d, and the UV spectrum showed the absorption maximum of major flavonoid peak 5 at 255 and 351 nm (Figure 6e).

Figure 5.

 Aurone production in transgenic lettuce. Leaves of control (left) and two 1610 plants (right) (a–b). HPLC chromatograms of lettuce leaf extracts detected at 400 nm for aurone. Extracts were obtained from leaves of untransformed (c), 1610 (d), 1618 (e) and 1610/1618 (f) lettuce plants. Peak 2, which eluted at 4.2 min, was identified as aureusidin-6-O-glucoside. For quantitative data, see Table 2.

Table 2.  Analyses of flavonoids and anthocyanins in aurone extracts of transgenic lettuce leaves
Transgenic lines and controlsIndependent lines no.HPLC chromatogram at different DAD wavelengths
Peak 2 at 400 nm 4.3 minPeak 2 at 520 nm 4.4 minPeak 5 at 360 nm 8.3 min
Aureusidin-6-O-glucoside mg/g DWCyanidin-3-(6′-malonyl) glucoside mg/g DWQuercetin-3-(6′-malonyl) glucoside mg/g DW
  1. UD, undetectable.

 Wild type10.0000.0126.073
 Transgenic controlA0.0000.01422.609
 Snapdragon petals1UD3.76UD
Figure 6.

 HPLC chromatograms detected at 360 nm for flavonoids. Extracts used were obtained from the leaves of untransformed lettuce (a) and the 1610 (b), 1618 (c) and 1610/1618 (d) plants. The UV spectrum of peak 5 (e) indicates the typical flavonoid λmax. Peaks 3, 4 and 5 represent quercetin derivative, kaempferol glucoside and quercetin-3-(6′-malonyl) glucoside, respectively. For a detailed quantitative analysis, see Table 2.

Aureusidin formation is linked to enhanced dismutase activity

The peroxyl radical–scavenging capacity of transgenic control plants was 12 μmole equivalents of the vitamin E analog Trolox (TE)/g. This value is similar to those of most vegetables (Song et al., 2010). As shown in Table 3, activation of the anthocyanin biosynthetic pathway in ANT1 transformants (T0) resulted in a 2.5-fold increase in oxygen radical absorbance capacity (ORAC) value (to 29 μmoles TE/g), to levels that are typical for common fruits, such as orange and grape (Wolfe et al., 2008). Interestingly, the almost complete conversion of anthocyanins to aurones that was accomplished in 646/1252/1257 plants resulted in a much greater increase in ORAC values, to an average of 78 μmoles TE/g. These levels resembled those of various berries, such as blueberry, blackberry and raspberry, that provide the highest known antioxidant activities of any edible food (Wolfe et al., 2008; and Wu et al., 2004). Similar results were obtained for heterozygote plants from the next two selfed generations T1 and T2 (Table 3).

Table 3.  Gene type and oxygen radical absorbance capacity (ORAC) of transgenic tobacco leaves and their control
LinesGenesORAC assay (μmoles TE/g)
StMtf1 M Chi Am4CGT AmAs1 T0T1T2
  1. ORAC data expressed as micromoles of Trolox equivalent per gram (μmoles of TE/g). The presence or absence of a gene is indicated by ‘+’ and ‘−’, respectively.

Transgenic control++12.220.813.6
646/1252++  113.3
646/1252/1257-12,14,19++++7854.2 ± 5.083.5 ± 9
646/1252/1257-26,27,31++++  103.3 ± 57

Self-fertilization of the triply transformed T0 plants produced segregating T1 families with various seedling colours (Figure 7). Seedlings with a bronze-green colour, confirmed to contain at least one copy of each of the three constructs used for transformation, were allowed to develop into mature plants in the glasshouse. ORAC analysis confirmed unusually high antioxidant activities of, on average, 54.2 μm TE/g in leaves of randomly selected T1 plants. Similar results were obtained for homozygous T2 plants (Table 3).

Figure 7.

 T1 seedling of triply transformed (646/1252/1257) tobacco. Colour of 1-week-old seedlings (above) is correlated with the presence or absence of transgenes, as determined by PCR (below).

Because superoxide free radicals are at least as important in triggering oxidative stress as peroxyl radicals, we employed a xanthine–xanthine oxidase system with a tetrazolium salt as reducing agent to assess the capacity of plants to scavenge such O2 anions. As shown in Table 4, leaf extracts of transgenic T0 and T1 control plants inhibited superoxide dismutase (SOD) by 27% and 25.5%, respectively. This inhibitory activity increased slightly, to 36%, when extracts of the anthocyanin-rich T1 leaves of StMtf1M plants (ANT1) were used, whereas no increase in inhibitory activity was found in T0 leaves. However, the conversion of most of the anthocyanins to aurones resulted in superior SOD inhibiting activities of up to 90% in T0 and 50%–60% in T1 plants of two 646/1252/1257 lines (Table 4). Homozygous AOG-producing T2 plants continued to display high SOD inhibiting activities (62%–77%) compared to their transgenic controls (24%).

Table 4.  Gene type and superoxide radical–scavenging capacity (superoxide dismutase (SOD) inhibition) of transgenic tobacco leaves and their control
LinesGenesSOD activity (inhibition rate %)
StMtf1 M Chi Am4CGT AmAs1 T0T1T2
  1. The presence or absence of a gene is indicated by ‘+’ and ‘−’, respectively.

Transgenic Control++2725.5 ± 0.529
646++1936 ± 5.055
646/1252++/−  60
646/1252/1257–12,14,19++/−++9160.3 ± 5.969
646/1252/1257–26,27,31++/−++ 50.0 ± 6.162

Antioxidant activities were also determined in aurone-overexpressing lettuce in the presence and absence of the Dfr gene. As shown in Table 5, SOD inhibition was threefold greater in T0 aurone-expressing lettuce (1610/1618) than in wild-type and transgenic lettuce controls (1610 and 1618). All T1 lettuce plants that overexpressed aurone (1610), were silenced for Dfr (1618) and both overexpressed aurone and were silenced for Dfr (1610/1618) showed a twofold inhibition of SOD inhibition compared to controls. Similar results were obtained with the ORAC assay performed on T1 transgenic lettuce leaves.

Table 5.  Gene type, oxygen radical absorbance capacity (ORAC) and superoxide radical–scavenging capacity (superoxide dismutase (SOD) inhibition) of transgenic lettuce leaves and their controls
LinesGenesSOD activity (inhibition rate %)ORAC assay (μmoles TE/g)
Dfr Am4CGT AmAs1 T0T1T1
  1. ND, not done.

  2. ORAC data expressed as micromoles of Trolox equivalent per gram (μmoles of TE/g). The presence or absence of a gene is indicated by ‘+’ and ‘−’, respectively.

Wild type28.534.211.4
Transgenic control31.735.014.1
1610 ++80.556.411.5


We have shown that the aurone biosynthetic pathway can be transferred from flowers of the ornamental plant snapdragon to the vegetative tissues of tobacco and lettuce. In addition to the expression of the snapdragon Am4’CGT and AmAs1 genes, aurone formation in tobacco required modifications that increased the accumulation of the flavonoid naringenin chalcone, which is the substrate for Am4’CGT. These modifications involved increasing StMtf1M gene expression and lowering the expression of the Chi gene. Although transformed cells produced large amounts of aurones in tissue culture, developing bright yellow calli, it was difficult to subsequently regenerate transgenic shoots. Indeed, aurone-producing tobacco plants were obtained only when tissue culture media were supplemented with naringenin chalcone. These results confirm the important role that flavonoids play in mediating auxin transport (Peer and Murphy, 2007). Chi gene silencing was unnecessary in the lettuce variety ‘Eruption’, which has a functionally active flavonoid biosynthetic pathway and naturally produces anthocyanins. However, aurone formation was effectively enhanced upon silencing of the alternative gene, Dfr. The presence of cyanidin-3-(6′-malonyl) glucoside in the doubly transformed 1610/1618 lines was attributable to the partial silencing of Dfr. These data were supported by the ammonia test (Lawrence, 1926), which detects anthocyanins in plant tissues (data not shown).

Under stress conditions, aurone-containing plants have even higher free radical–scavenging activity, because stress induced flavonoid biosynthesis in plants (Ebel, 1986 and Shirley, 2002). Our data support the notion that aureusidin-6-O-glucose formation is enhanced under conditions of nutrient limitation. All controls, aurone-overexpressing lines and Dfr-silenced lines were stunted in growth, displayed accelerated flowering and produced lower amounts of purple pigments during nutrient limitation than when normal amounts of fertilizer were applied. These changes had a negative effect on the antioxidant activities of transgenic controls and aurone lines (data not shown). However, the imposed abiotic stress was correlated with an increased formation of yellow pigment in double transformants. These plants displayed an increased capacity to scavenge peroxyl radicals and inhibit SOD.

Aurone formation has been reported for several edible Asteraceae and Amaranthaceae (Ferreira et al., 2004; General books LLC, 2010; Milovanovic et al., 2002). The ability of transgenic crops to produce aurones broadens the spectrum of dietary antioxidants available for consumption, which is likely to enhance their nutritional value. However, the US Food and Drug Administration (FDA) did not yet determine that aurones are ‘generally recognized as safe (GRAS)’, and more studies will need to be carried out to confirm the biosafety of new aurone-producing foods. Although aurones are simple flavonoid compounds, their biosynthesis is associated with a significant increase in total antioxidant power. Indeed, the novel strategy presented in this study increased the total antioxidant power by up to sevenfold. It may prove useful in continued efforts to optimize the health-promoting properties of vegetables and fruits. Indeed, efforts to increase the total antioxidant power of foods through genetic engineering are generally limited to less than a ∼2-fold increase (Aksamit-Stachurska et al., 2008; and Reddy et al., 2007).

There is an inverse association between the total intake of fruits and vegetables and the risk of developing diseases such as cancer (Boffetta et al., 2010) and coronary heart disease (Dauchet et al., 2006). This health-promoting effect has been attributed to the additive and/or synergistic activity of mixtures of antioxidants (Liu, 2004; Messina et al., 2001; http://www.cnpp.usda.gov/ dietaryguidelines.htm). Until now, aurones have been considered only as a means of enhancing ornamental flower colour. Transferring the capacity to produce specific antioxidants across plant species through genetic engineering could compensate for the lack of diversity in many modern diets (DeWeerdt, 2011). This study therefore presents a strategy to introduce new antioxidants into the food supply.

Experimental procedures

Chemicals and standards

HPLC-grade acetonitrile, water and trifluoroacetic acid (TFA) and also naringenin and chalcone standards were purchased from Sigma (St. Louis, MO). Naringenin-7-O-rutinoside and cyanidin-3-O-glucoside were purchased from Indofine (Hillsborough, NJ). Maritimein (3′,4′,6,7-tetrahydroxyaurone or maritimetin-7-glucoside) was purchased from Chromadex (Irvine, CA). All standards were prepared as stock solutions at 10 mg/mL in methanol and diluted in water, except for chalcone, which was prepared in 50% methanol. UV external standard calibration was used to obtain calibration curves of cyanidin-3-O-glucose, naringenin-7-O-rutinoside and chalcone, which were used to quantify anthocyanins, flavones and chalcones, respectively. Both UV and mass spectrometry (MS) external calibration of maritimein were employed for the quantitation of aureusidin-6-O-glucose.

Genes and plasmid constructs

A full-length cDNA of the aureusidin synthase (AmAS1) gene was isolated from snapdragon (Antirrhinum majus‘Rocket Yellow’) flowers by reverse transcriptase (RT-)PCR using the primer set 5′-GGA TCC AAA TTA CAT TGC TTC CTT TGT CCC AC (forward) and 5′-AAG CTT CTC AAA AAG TAA TCC TTA TTT CAC (reverse). The product digested with BamHI and HindIII was fused to regulatory elements, the 35S promoter of FMV and the terminator of the potato ubiquitin-3 gene, and the resulting expression cassette was cloned into pBluescript (Agilent Technologies, Santa Clara, CA). The cytosolic chalcone 4′-O-glucosyltransferase (Am4’-Cgt) cDNA was also amplified from flower RNA, and the primer set used in this case was 5′-GGA TCC ATG GGA GAA GAA TAC AAG AAA ACA C (forward) and 5′-ACT AGT TTA ACG AGT GAC CGA GTT GAT G (reverse). The BamHI–HindIII fragment was linked to the FMV promoter and Ubi3 terminator and also inserted into pBluescript. The binary vector pSIM1251 contains both the AmAS1 and Am4’CGT gene expression cassettes and a cassette for the phosphomannose isomerase (pmi) selectable marker gene (Aswath et al., 2005). Vector pSIM1618 is similar to pSIM1251, but carries a neomycin phosphotransferase (nptII) selectable marker gene. Primers used to amplify a 0.6-kb fragment of the tobacco chalcone isomerase (Chi) gene (GenBank accession AB213651) had the sequences 5′AGA TCT CTA GAC TCC AAT TTC TGG AAT GGT AG (forward) and 5′-CTC GAG AGT GCT CTT CCT TTT CTC GCC GC (reverse) for the antisense fragment and 5′-CTC GAG GAG TCC ATT ACC ATT GAG AAT TAC G (forward) and 5′-CTC GAG GAG TCC ATT ACC ATT GAG AAT TAC G (reverse) for the sense counterpart. Vector pSIM1252 carries the inverted repeat of Chi gene fragments positioned between the FMV promoter and Ubi3 terminator. A silencing cassette targeting the dihydroflavonol 4-reductase (Dfr) gene from the lettuce variety Eruption (identical to GenBank CV700105) was generated using the primer pairs 5′-GGA TCC GCA GGT ACA ACT AGA CAC CG (forward) and 5′-CCA TGG ATT GGT GTT TAC ATC CTC TGC G (reverse) for a 708-bp sense fragment and 5′-ACT AGT GCA GGT ACA ACT AGA CAC CG (forward) and 5′-CCA TGG AGT CGT TGG TCC ATT CAT CA (reverse) for a 542-bp antisense fragment. The vector carrying the inverted repeat of Dfr fragments fused to regulatory elements and positioned within the T-DNA was named pSIM1610.

Plant transformation

Tobacco was transformed as described previously (Richael et al., 2008). For transformation of the lettuce variety Eruption, ∼250 seeds were transferred to a 1.7-mL Eppendorf tube, immersed for 1 min in 70% ethanol and for 15 min in 10% bleach with a trace of Tween and then triply rinsed with sterile water. Sterilized seeds were spread evenly over solidified medium consisting of half-strength MS with vitamins (M404, Phytotechnology) containing 10 g sucrose per litre and 2% Gelrite in Magenta boxes (30–40 seeds/box), and germinated at 24 °C under a 16-h day/8-h night regime. Agrobacterium was grown overnight from frozen glycerol stock (−80 °C) in a small volume of Luria broth with kanamycin (100 mg/L) and streptomycin (100 mg/L). Cotyledons from 4-day-old seedlings were wounded with a scalpel to give small cuts at right angles to the midvein and immersed in agrobacterium suspensions. After 10 min, the suspension was removed by aspiration and the explants were blotted on sterile filter paper. Explants were placed adaxially on co-culture medium that consisted of MS medium (pH 5.7) with vitamins (M404, Phytotechnology), 30 g sucrose per litre, 0.1 mg/L 6-benzylaminopurine (BAP) and 0.1 mg/L 1-naphthaleneacetic acid (NAA), solidified with 6 g/L agar. After 2 days, the explants were transferred to regeneration medium that consisted of MS medium (pH 5.7) with vitamins (M404), 30 g sucrose per litre, 0.1 mg/L BAP, 0.1 mg/L NAA, 6 g/L agar, 150 mg/L timentin and 100 mg/L kanamycin. Explants were transferred to fresh media at 2-week intervals. After 2–3 weeks, shoot buds were harvested and transferred to the same media. At least 25 shoots that elongated within the next 2–4 weeks were transferred to media lacking hormones, to promote root formation.

Sample preparation for biochemical analysis

Glasshouse-grown lettuce or tobacco leaves or flowers were harvested, immediately frozen in liquid N2 and then homogenized. The powder was then freeze-dried and stored at −80 °C until used. Samples were extracted as described by Ono et al., 2006, with modification. Briefly, about 150 mg freeze-dried ground leaves or flowers was placed in a 2-mL screwcap tube along with 50% acetonitrile/0.1% TFA and 500 mg of 1.0-mm glass beads. Tubes were shaken in a BeadBeater (Biospec, Bartelsville, OK) using a prechilled rack for 10 min at maximum speed and centrifuged for 5 min at 4 °C, and the supernatant was transferred to a clean tube. The remaining pellet was re-extracted with 1 mL of the same extraction solvent and centrifuged. The supernatants were combined and concentrated in a SpeedVac (Thermo Savant, Waltham, MO) prior to HPLC analysis.

In order to confirm anthocyanin, freeze-dried leaves were also extracted in acidified methanol (0.01% HCl) for anthocyanin and purified by solid-phase extraction using C-18 cartridge as described in Current Protocols in Food Analytical Chemistry (Rodriguez- Saona and Wrolstad, 2001).

LC/MS analysis

Aurone analyses were performed using an Agilent HPLC series 1200 equipped with ChemStation software, a degasser, quaternary pumps, autosampler with chiller, column oven and diode array detector. The separation was performed with an Agilent Zorbax Eclipse XDB-C18 (150 × 4.6 mm, 5-μm particle size) with a C18 guard column operated at a temperature of 35 °C. The mobile phase consisted of 0.1% TFA/water (eluent A) and 90% acetonitrile in water/0.1% TFA (eluent B) at a flow of 0.8 mL/min using the following gradient programme: 20% B (0–3 min); 20%–60% B (3–20 min); 60% B isocratic (20–27 min); 60%–90% B washing step (27–30 min); and equilibration for 10 min. The total run time was 40 min. The injection volume for all samples was 10 μL. Specific wavelengths were monitored separately at 400 nm for aurone and 360 nm for flavones. Additionally, UV/Vis spectra were recorded at 520 nm for anthocyanins. The HPLC system was coupled online to a Bruker (Bremen, Germany) ion trap mass spectrometer fitted with an ESI source. Data acquisition and processing were performed using Bruker software. The mass spectrometer was operated in positive ion mode and auto MSn with a scan range from m/z 100 to 1000. Nitrogen was used both as drying gas at a flow rate of 12 L/min and as nebulizer gas at a pressure of 45 psi. The nebulizer temperature was set at 350 °C.

Antioxidant capacity assays

The capacity to scavenge peroxyl and superoxide radicals was determined using 2,2′-azobis (2-amidino-propane) dihydrochloride (AAPH) (Haung et al., 2005; and Prior et al., 2003) and a Superoxide Dismutase Activity Assay Kit (BioVision Research Products, Mountain View, CA), according to the manufacturer’s recommendations. Inhibition of SOD was also assayed using the SOD Assay Kit from Cell Technology Company. The SOD activity (inhibition rate %) was determined by measuring absorbance at 440 nm. Samples were diluted 40 times so as to be in the linear range.


We thank Robert Chretien and Craig Richael for plant transformation and Michele Krucker for growing plants in the glasshouse. Rekha Chawla and Jeffrey Hein are acknowledged for excellent molecular biology support.