• Open Access

Enhancing ascorbate in fruits and tubers through over-expression of the l-galactose pathway gene GDP-l-galactose phosphorylase

Authors


(Tel +64 9925 7254; fax +64 9925 8628; email william.laing@plantandfood.co.nz)

Summary

Ascorbate, or vitamin C, is obtained by humans mostly from plant sources. Various approaches have been made to increase ascorbate in plants by transgenic means. Most of these attempts have involved leaf material from model plants, with little success reported using genes from the generally accepted l-galactose pathway of ascorbate biosynthesis. We focused on increasing ascorbate in commercially significant edible plant organs using a gene, GDP-l-galactose phosphorylase (GGP or VTC2), that we had previously shown to increase ascorbate concentration in tobacco and Arabidopsis thaliana. The coding sequence of Actinidia chinensis GGP, under the control of the 35S promoter, was expressed in tomato and strawberry. Potato was transformed with potato or Arabidopsis GGP genes under the control of the 35S promoter or a polyubiquitin promoter (potato only). Five lines of tomato, up to nine lines of potato, and eight lines of strawberry were regenerated for each construct. Three lines of tomato had a threefold to sixfold increase in fruit ascorbate, and all lines of strawberry showed a twofold increase. All but one line of each potato construct also showed an increase in tuber ascorbate of up to threefold. Interestingly, in tomato fruit, increased ascorbate was associated with loss of seed and the jelly of locular tissue surrounding the seed which was not seen in strawberry. In both strawberry and tomato, an increase in polyphenolic content was associated with increased ascorbate. These results show that GGP can be used to raise significantly ascorbate concentration in commercially significant edible crops.

Introduction

Ascorbate or vitamin C is necessary for humans as our primate ancestors lost the ability to synthesise ascorbate about 60 million years ago, because of a mutation in l-gulono-γ-lactone oxidase the enzyme encoding the final step in the pathway (Lachapelle and Drouin, 2011). Consequently, humans obtain most of their vitamin C from plant sources. Current recommendations for vitamin C intake for humans are in the range of 40–90 mg per day, depending on the country, and it has been suggested this range could be higher because of differential accumulation in various body tissues (Levine et al., 1999; Vissers et al., 2011). Most food sources of vitamin C are relatively low in vitamin C (U.S. Department of Agriculture ARS, 2010), so it can be difficult to achieve these intakes without supplements. However, synthetic vitamin C supplements are likely to be in a less bioavailable form (Vissers et al., 2011). Consequently, there is a significant opportunity to improve human vitamin C intake by developing higher vitamin C plant-based foods, including through transgenic approaches.

Previous transgenic efforts to increase ascorbate have generally used model species such as Arabidopsis or tobacco, although some other commercially important species have been used. These include potato [up to twofold increase in tubers using the gene for galacturonate reductase (Hemavathi et al., 2010)], lettuce [up to sevenfold increase in leaves from a low baseline concentration using a rat gene for gulono lactone oxidase (Jain and Nessler, 2000)], maize [two to sevenfold increase in kernels from a very low baseline (Chen et al., 2003; Naqvi et al., 2009)] and tomato [1.2- to 1.6-fold increase in fruit using GDP-mannose epimerase (Zhang et al., 2010); for full details, Table S1].

Several routes to ascorbate production in plants have been proposed (Ishikawa et al., 2006), either through the l-galactose pathway (Wheeler et al., 1998) or through uronic acids probably derived from myoinositol or pectin (Ishikawa et al., 2006). In Arabidopsis, it is clear that the l-galactose pathway is normally the only significant pathway to producing ascorbate in seedlings (Dowdle et al., 2007), while in other species, there is evidence that the uronic acid precursors can be utilised as a substrate for ascorbate (Ishikawa et al., 2006). Published results on the over-expression of most of the enzymes in the l-galactose pathway have reported little effect on ascorbate concentration (Tabata et al., 2001; Gatzek et al., 2002; Tokunaga et al., 2005; Qian et al., 2007). Exceptions include GDP-mannose pyrophosphorylase where over-expression of the gene increased tobacco leaf ascorbate ∼twofold (Badejo et al., 2008) and our work with the gene for GDP-galactose phosphorylase (GGP, also known as VTC2), which increased Arabidopsis leaf ascorbate up to ∼threefold (Bulley et al., 2009). In addition, over-expression of the precursor gene to the pathway encoding the enzyme phosphomannomutase increased Nicotiana tabacum leaf ascorbate by up to 2.5-fold, although over-expression of the same enzyme in N. benthamiana increased leaf ascorbate by less than 50% and increased Arabidopsis leaf ascorbate by only 33% (Qian et al., 2007). Antisense, knockout and T-DNA insert approaches targeting the genes of the l-galactose pathway reduced leaf ascorbate in these model species (Tabata et al., 2001; Gatzek et al., 2002; Conklin et al., 2006; Qian et al., 2007; Maruta et al., 2008; Torabinejad et al., 2009), showing the importance of the l-galactose pathway in leaves of these species.

Altering the alternative biosynthetic pathways has had mixed success in changing ascorbate concentrations. In the galacturonate pathway, increasing the gene expression for galacturonate reductase increased potato tuber ascorbate by up to twofold (Hemavathi et al., 2009), while the same gene in Arabidopsis was only effective when the leaves were fed galacturonate (Agius et al., 2003). Reducing pectate production through antisense reduction of pectate lyase somewhat reduced strawberry ascorbate, supporting these authors’ contention that galacturonate was a source of ascorbate (Agius et al., 2003). Increasing myoinositol production has also shown varied results, with both increased (Lorence et al., 2004) and unaffected (Endres and Tenhaken, 2009) leaf ascorbate being reported. Related to this pathway, transformation of lettuce, tobacco and Arabidopsis with a rat gene encoding gulono lactone oxidase increased leaf ascorbate from two to sevenfold (Jain and Nessler, 2000; Radzio et al., 2003) although the increase observed in potato tubers was only 40% (Hemavathi et al., 2010). It is still unknown how significant the parallel route to the l-galactose pathways starting at GDP-mannose through GDP-gulose to gulono lactone is in plants. Generally, the overarching pattern seen following the modification of the alternative pathways is that in leaves, the lower the base amount of ascorbate is, the greater the increase.

Endogenous concentrations of leaf ascorbate vary strongly in plants from 25 mg/100 g fresh weight (FW) in glasshouse-grown tobacco species (Bulley et al., 2009) to over 500 mg/100 g FW in field-grown apple and kiwifruit (Bulley et al., 2009). Concentrations also vary between plants grown in light and darkness (Yabuta et al., 2007) and according to light quality (Bartoli et al., 2009). The high concentrations of leaf vitamin C have been suggested to protect the photosynthetic apparatus from strong oxidative pressure and stress (Foyer and Noctor, 2011). Fruit ascorbate concentrations vary from low [e.g. apple < 5 mg/100 g FW in ‘Royal Gala’ apple flesh (Davey et al., 2006)] to over 800 mg/100 g FW in some kiwifruit species (e.g. Actinidia eriantha) (Ferguson and MacRae, 1992; Bulley et al., 2009) and even higher in West Indian cherry (also known as Acerola, Malpighia emarginata) fruit (Badejo et al., 2007). Tubers such as potato (Shakya and Navarre, 2006; Hemavathi et al., 2009, 2010) and fruits such as tomato (Alhagdow et al., 2007; Zhang et al., 2010) have low to moderate ascorbate content of less than 20 mg/100 g FW, while strawberry fruits are reported to contain between 60 and 70 mg/100 g FW (Atkinson et al., 2006). These commercially valuable fruits and tubers would not be expected to experience high oxidative stress as leaves, and it is possible that the high concentrations of ascorbate reflect a loss of control mutation in ascorbate biosynthesis.

GDP-l-galactose phosphorylase was the last enzyme in the l-galactose pathway to be identified (Dowdle et al., 2007; Laing et al., 2007; Linster et al., 2007, 2008). We have shown in Arabidopsis that over-expression of this gene from kiwifruit leads to increased leaf ascorbate (Bulley et al., 2009). Here we show that this finding can be transferred to commercial crops, to establish that this gene from three sources increases fruit and tuber ascorbate concentrations. These crops represent two widely separated plant families and two different edible organ types, and the genes derive from kiwifruit, potato and Arabidopsis. We show that we can increase organ ascorbate concentration significantly in all three target food species.

Results

Two genes for GGP have been identified in Arabidopsis, VTC2 and VTC5, (Dowdle et al., 2007) which are 76% identical in an amino acid alignment. In potato, we identified two variants, StVTC2A and StVTC2B, which are only 62% identical (Table S2) and show a similar degree of identity to the two Arabidopsis genes. This suggests the Arabidopsis and potato paralogs arose independently. Among the Actinidia species A. deliciosa, A. arguta, A. eriantha and A. chinensis, only one member of the gene family has been identified from a large transcript database (Crowhurst et al., 2008), now with over 1.2 million kiwifruit transcripts, or from a partial genomic sequence of A. chinensis (unpublished). Each kiwifruit protein sequence shows at least 95% identity to the others, suggesting that they represent alleles of a single gene. The kiwifruit gene used was selected based on previous work showing the gene functioned well in Arabidopsis to increase leaf ascorbate (Bulley et al., 2009). The kiwifruit genes all show 69%–70% predicted amino acid identity to VTC2 and 73%–76% identity to VTC5 in making it difficult to identify the ancestral gene. Alignment of the five protein sequences used in this work is also shown in Figure 1. The 50 C terminal residues show the greatest differences. The Arabidopsis (Dowdle et al., 2007; Linster et al., 2007, 2008) and Actinidia (Laing et al., 2007) genes have been functionally verified in other studies, while the potato genes were identified by homology searches and were tested in this study.

Figure 1.

 Alignment of the protein sequences of the GDP-l-galactose phosphorylase genes used in transformation. AtVTC2 refers to the Arabidopsis thaliana gene At4g26850, while AtVTC5 refers to its paralog AT5G55120. StVTC2A and StVTC2B refer to the two potato genes used in this study. AcFG528585 is the Actinidia chinensis gene.

Generation of transgenic plants expressing heterogonous GGP genes

In potato, two variants of the GGP gene from each of Arabidopsis and potato were transformed into potato using two different promoters (35S or a polyubiquitin promoter, PAT), allowing comparison of genes from different species and different promoters. Northerns were run on the various potato transgenics to show that all plants transformed with potato GGP genes expressed the transgene significantly higher than the endogenous genes in the control plants, and the Arabidopsis GGP genes were also expressed (Figure S1). Up to nine lines of each construct of transgenic potatoes were grown through to tuber production, with three replicate plants being measured for each line. The transgenic plants appeared normal and produced similar yields of tubers as their controls (Table 1). However, only the PAT-StVTC2A (3.0-fold increase), 35S-StVTC2B (2.4-fold) and PAT-StVTC2B (3.1-fold) constructs had significantly higher ascorbate than the controls as means over all lines (Table 1); individual lines within a construct varied up to 2.7-fold higher (35S-AtVTC2) to 4.7-fold higher (PAT-StVTCA) than the empty vector average (Table S2). Thus, several lines within each promoter-gene construct of potato could be generated with ascorbate concentrations over twice the control. However, it is clear that the 35S promoter is less effective in potato than the PAT promoter and that the Arabidopsis genes for VTC2 and VTC5 were less effective than the potato genes when driven by the 35S promoter. For the two potato genes, the StVTC2A and StVTC2B were similar in effectiveness.

Table 1.   Ascorbate concentration and tuber yield per plant in transgenic potato tubers
 Ascorbate (mg/g DW)NSDTuber weight (g)NSD
  1. The ascorbate values are means and SD of N independent lines in mg/g dry weight (DW). Potato tubers have a DW/FW ratio of ∼0.2, so multiplying the mean by 20 would convert mg/g DW into mg/100 g FW as is used in other tables. Each line was measured on tubers from three plants to give a mean value for that line. As the experiments were carried out at two separate dates, ANOVA within an experiment was used to establish least significant differences. PAT is the polyubiquitin promoter and SD is the standard deviation of the mean. VTC2A, VTC2B, AtVTC2 and AtVTC5 are described in Figure 1. The letters at the end of each row represent whether means differ at the 5% significance level. Means with different letters within an experiment differ significantly. Full ascorbate data on individual lines are presented in Table S1.

Experiment 1
 35S-AtVT20.9380.27ab2092368a
 35S-StVTC2A1.0080.34ab2111645a
 PAT-StVTC2A1.6590.47b2212142a
 Empty vector0.5560.16a218693a
 WT0.5040.20a264646a
Experiment 2
 35S-AtVT50.8680.28a2252153a
 35S-StVTC2B1.4480.21b2232449a
 PAT-StVTC2B1.8380.24c2412050a
 Empty vector0.5960.22a210351a
 WT0.6760.18a228727a

Five independent lines of tomato were generated expressing the kiwifruit GGP, under the control of the 35S promoter, and grown in a glasshouse through to fruiting. Of the five lines, two displayed no change in ascorbate concentration and had normal fruit shape and seed development. The other three lines had significantly raised ascorbate concentration (Table 2), but the fruit showed a fasciated-like shape and the fruit had no seed or nonviable seed (Figure 2) plus the mucilage typically surrounding the seed (jelly of locular tissue) was also absent. Line two had a sixfold increase in vitamin C, and line 3 (3.5-fold) had no seed at all, while line 4 (2.6-fold) had small nonviable seed (data not shown). These three lines had significantly smaller fruit than the controls and had very much higher levels of GGP gene and enzyme activity than the controls (Table 2). Measurements of leaf ascorbate concentration showed no significant difference between lines. The polyphenolic content of these transgenic tomatoes were measured, and more than a doubling of the concentration of a range of polyphenolics was observed for lines with enhanced ascorbate (Table S4). Carotenoid concentration was also measured in these fruit and showed no significant differences from controls, showing this was not just an effect of fruit size (Table S4).

Table 2.   Ascorbate concentration in fruit and leaves and fruit size, phosphorylase activity and relative gene expression of five independent tomato lines
LineFruit ascorbate (mg/100 g FW)NSDLeaf ascorbate (mg/100 g FW)NSDFruit weight (g)NSDGGP activity (nm/min/mg protein) Gene expression
(Relative)
  1. The mean ascorbate values are means of N fruit harvests (multiple fruit per harvest) or leaves measured from the same plant over several dates in mg/100 g FW. The letters at the end of each block represent whether means differ at the 5% significance level. Means with different letters within an experiment differ significantly. nm, not measured. Gene expression is relative to the tomato UBQ9 housekeeping gene level.

Untransformed1543anm  5258anm nm
Empty vector18154a75711anm  0.002 0.2
117112a1011a731 nm 1.9
2111823b150366a1672b16.5 43.2
363614c101345a1174b14.0 70.7
446720c106359a3548c20.2 24.0
51733anm  nm  nm 0.1
Figure 2.

 Photographs of transgenic tomato fruit showing internal and external appearance. All fruit from lines 2 and 3 showed no seed and had high ascorbate concentration.

Seven independent lines of strawberry were generated expressing the kiwifruit GGP, under the control of the 35S promoter, and grown to fruiting and ascorbate concentration measured. The plants appeared normal and all transgenic lines had approximately a twofold increase in fruit ascorbate concentration compared to the control lines (Table 3). The transgenic lines were all very similar to each other in both fruit weight (except line six that produced very small fruit, 17 fruit in all, in three harvests) and in ascorbate content. Gene expression of the transgene was strong, as was phosphorylase activity compared with the background levels measured in the control plants. Leaf ascorbate was also up to twice the concentration of the controls (Table 3) but showed a poor correlation with fruit ascorbate concentration (r2 = 0.55 including the control plant, no correlation among transgenic plants). The polyphenolic content of these transgenic strawberries were measured, and about a 50% enhancement of the concentration of a range of polyphenolics was observed for lines with enhanced ascorbate (Table S4) although p-coumaroyl glucosides were more than doubled in concentration.

Table 3.   Strawberry fruit and leaf ascorbate concentration and other fruit properties
LineFruit ascorbate (mg/100 g FW)NSDLeaf ascorbateNSDFruit sizeNSDGGP activity (nm/min/mg protein)SE of the meanGene expressionSE
  1. ANOVA was carried out only on the ascorbate concentration and fruit size. N refers to the number of separate samples of fruit (between 1 and 7 fruit per sample) measured per line and the SD to the standard deviation of the mean. The letters refer to whether to values in a column are significantly different at the 5% level. The wt plants were two independent lines that were combined for statistical analysis, as they were not significantly different. Gene expression is normalised to the strawberry actin housekeeping gene level. See methods in Experimental procedures for details. nm, not measured.

wt622311c281844c3.3181.5a0.60.20.420.10
1011678ab513314b2.471.3ab6.71.5484
31281517ab422534a2.4151.2a3.91.1365
3b128646ab466339ab2.840.6ab3.60.6719
51311010b45635ab2.9103.1ab4.92.8248
6131314abnmnmnm0.530.01a10.72.8nmnm
91092113a406645a5.0212.4b6.01.9337
9a1251020ab366320a5.693.9+b3.90.95713

Discussion

We have shown that a gene in the l-galactose pathway of ascorbate biosynthesis can function in three different commercial food plant species to raise ascorbate in the fruit or tubers significantly. This GGP gene encodes an enzyme that catalyses the first committed step in ascorbate biosynthesis in plants, and we have identified GGP as rate limiting in Arabidopsis and tobacco leaves (Laing et al., 2007; Bulley et al., 2009). The target species represent two unrelated families of plants (Solanaceae and Rosaceae) and two different types of edible organs. The tomato and potato are commercially very important food crops, while strawberry is a major dessert and processing fruit.

While there is much debate about satisfactory vitamin C intake in humans, recommended daily allowances range from 40 to 80 mg per day. Given concentrations of ascorbate in transgenic potatoes of up to 45 mg/100 g FW, this would ensure that the daily requirement is met with only 200-g serving of raw potato or about 300 g of cooked potato (U.S. Department of Agriculture ARS, 2010). Strawberry, with over 130 mg/100 g FW would require less than a 100 g serving. In tomato, fruit size was significantly reduced probably because of the effect of high ascorbate on seed production, and consequently, the highest line would increase vitamin C intake only by 50% compared with the control when expressed as a per fruit basis. However, when eating 100 g fruit, the transgenic tomato would provide over 100 mg vitamin C compared with only 17 in the control. Thus, eating these transgenic fruit and tubers should significantly increase vitamin C intake in humans. Bioavailability should also be enhanced by the vitamin C being in a fruit environment with an enhanced polyphenolic concentration, which has been shown to enhance vitamin C uptake in mice and guinea pigs (Cotereau et al., 1948; Vissers et al., 2011).

In the case of tomato and strawberry, while leaf ascorbate concentration in the transgenics was raised above the controls, it was quite variable depending on light exposure. Consequently, no significant difference in leaf ascorbate concentration was noted in tomato, while in strawberry, the increase was not well correlated with the concentration of fruit ascorbate. This suggests in plants where leaf ascorbate is already high, screening on the basis of leaf ascorbate would not be very discriminatory.

The fact that the gene GGP encoding a step in the l-galactose pathway of ascorbate biosynthesis can increase vitamin C so markedly in a wide variety of crops strongly suggests that this l-galactose pathway is significant in ascorbate biosynthesis. Others (Agius et al., 2003; Hemavathi et al., 2009) have shown that a uronic acid pathway gene can also increase ascorbate concentration in Arabidopsis leaves and potato tubers. One interpretation of this is that while the uronic acid pathway is a viable source of ascorbate, it only functions when a supply of substrates is present and is not necessarily a usual route to ascorbate in vivo. This is supported by the observation that feeding galacturonic acids to wild-type Arabidopsis leaves had no effect but stimulated ascorbate to leaves transformed with galacturonate reductase (Agius et al., 2003).

We observed that the fruit of three lines of tomato containing significantly higher ascorbate content were either parthenocarpic (seedless) or had nonviable seed. We hypothesise that either seed formation is directly inhibited by ascorbate (e.g. its pro-oxidant activity), or ascorbate regulates pollen viability, pollination, fertilisation, ovule development or embryo development. A possibly related phenomenon is the inhibition of seed germination by treatment with relatively high concentrations of ascorbate (Tewary and Mookerjee, 1982; Cano et al., 1997; Ishibashi and Iwaya-Inoue, 2006), which we have confirmed in a wide variety of plant species (S. Bulley, unpublished data). Additionally, quiescent cells in meristems and mature desiccated seeds (including embryos) contain little or no reduced ascorbate and very low concentrations of oxidised dehydroascorbate (DeGara et al., 1997; Tommasi et al., 2001; De Tullio and Arrigoni, 2003). While mature seeds are not equivalent to tomato flower ovules, these observations suggest that there are particular types of cells in which ascorbate concentrations are kept low by the cell as they may be sensitive to ascorbate concentration, disrupting normal development processes. Alternatively, depletion of GDP-mannose and its precursors through over-expression of GGP may specifically affect seed cell wall biosynthesis (Gilbert et al., 2009). GGP expression was driven by the constitutive 35S promoter, suggesting that the effect of higher ascorbate may have exerted itself very early in fruit development, possibly during or just after pollination before early embryo development. We did not pollinate the flowers, relying on self-pollination, and we did not notice any effect on the ability of our tomato lines to set fruit compared with control plants. We have no data on the viability of the transgenic lines pollen.

The high ascorbate seedless tomato fruit also had no mucilage around the seed, and it has been suggested that the mucilage is produced by the seeds (Hayward, 1938). Our results support this as transgenic lines of tomato with no seeds had no mucilage. As it stands, our ‘Roma’ type tomato lines demonstrate a potentially useful biotechnological innovation for producers of processed tomato products. This is because in many cases, the seeds and mucilage from tomatoes are often separated away from the useful tomato pulp during processing. Therefore, not only would GGP-transformed varieties be nutritionally higher in ascorbate but they would also require less-energy-intensive postharvest processing. Interestingly, we observed no difference in seed in strawberry where the seeds are located effectively externally to the compound fleshy fruit, and thus they would possibly be less exposed to high ascorbate concentrations.

In conclusion, we have significantly raised the ascorbate concentration of the edible organs of three important commercial crops through transgenic means using GGP, a gene encoding a rate-limiting step in the l-galactose pathway of ascorbate biosynthesis. In doing so, we have also discovered a fascinating effect of ascorbate in tomato fruit development that requires further investigation and an associated increase in fruit polyphenolic content.

Experimental procedures

Growth and other physiological measurements of transgenic plants

Plants were grown in containment in a glasshouse. Tomato plants (Solanum lycopersicum‘UC82B’) were grown at a minimum temperature of 25 °C and a maximum of 30 °C, while strawberry plants (Fragaria x ananassa‘Camrosa’) were grown at 20–25 °C, both under natural day lengths. At least two independent lines transformed with the empty vector were included along with the high expressing lines. Potato plants (Solanum tuberosum‘Ranger Russet’) were grown for 3 months, and tubers were then harvested. All transgenic plants were T0 generation (i.e. regenerated after transformation from tissue culture) and heterozygous for the transformed gene. Control and transgenic plants were grown and harvested together, although in the case of fruits, harvests were continuous as fruits matured. Usually more than one fruit made up a harvest, but these fruit were combined and constitute a replicate measurement in the tables.

GDP-l-galactose phosphorylase cloning and transformation

To generate a vector for the over-expression of the Actinidia chinensis phosphorylase gene, a full-length GGP cDNA transcript (GenBank accession FG528585) was PCR amplified. The resulting product was recombined using Gateway reactions to form pHEX2_319998 (Hellens et al., 2005). All Gateway reactions were carried out according to the manufacturer’s (Invitrogen, Mulgrave, Victoria, Australia) instructions.

Full-length cDNAs for the StVTC2A (JN000934) and STVTC2B (JN000933) phosphorylase genes were amplified from leaf RNA of potato; RNA of Arabidopsis thaliana was used to amplify cDNAs for the AtVTC2 (At4g26850) and AtVTC5 (At5g55120) phosphorylase genes. The various cDNAs were fused to either the 35S promoter of cauliflower mosaic virus (potato and Arabidopsis genes) or the strong constitutive PAT promoter of potato (HM439286) (potato genes) and positioned within binary vectors derived from pSIM401 (Rommens et al., 2004). These genes were used to transform potato.

Strawberry, potato and tomato plants were transformed with Agrobacterium tumefaciens LBA 4404 containing the relevant gene using published methods [(Fillatti et al., 1987; Rommens et al., 2004; Lin-Wang et al., 2010) respectively]. In all cases, kanamycin-resistant plants were propagated. The 35S promoter driving the kiwifruit gene was used for the fruit species, and either the 35S promoter or the polyubiquitin promoter (PAT) driving the potato or Arabidopsis genes were used for potato.

Measurement of transgene expression and activity

Fruit and tuber samples were harvested when mature and immediately frozen in liquid nitrogen. The samples were then ground to a fine powder and stored frozen at −80 °C. Gene expression in the two fruits was measured by qPCR as previously described (Bulley et al., 2009). Gene expression in potato was measured by Northern using DIG-labelled probes derived from the introduced cDNAs according to the manufacturer’s recommendations (Roche, Indianapolis, IN). X-ray films were exposed to the hybridised filters, developed with Konica SRX-101A, and scanned to obtain electronic images.

GDP-l-Galactose phosphorylase activity was measured in extracts (Laing and Christeller, 2004) from finely powdered frozen fruit samples according to our published method (Laing et al., 2007). The substrate was generated from GDP-mannose using GDP-mannose epimerase, and the reaction was coupled to NAD reduction through l-galactose-1-P phosphatase and galactose dehydrogenase, all available as Escherichia coli expressed enzymes from kiwifruit.

Ascorbate, polyphenolic and carotenoid measurement

For fruit total ascorbate, approximately 100 mg of frozen powder was weighed out and ascorbate was extracted and measured as previously described by HPLC (Rassam and Laing, 2005) and expressed as mg ascorbate per 100 g FW. For potato, reduced ascorbate was also measured (Shakya and Navarre, 2006) by HPLC on freeze-dried tissue and expressed as milligram per gram DW. As potato is approximately 80% water, potato results can be converted to % FW values by dividing the DW by five and multiplying by 100.

For polyphenolic extraction, approximately 100 mg of powdered fresh material was mixed with 1 mL of ethanol/water (80 : 20, v/v). This mixture was then homogenised using a vortex for 30 s and shaken for 2 h at room temperature. After centrifugation at 10 000 g for 15 min, the supernatant was collected and filtered through a 0.45-μm filter prior to analysis.

Hydrophilic antioxidant capacity was evaluated using the Folin–Ciocalteu or total phenolics assay following the procedure described earlier (Andre et al., 2006). The assay measures the reducing capacity of the sample and is not specific to polyphenols (Huang et al., 2005). Carotenoids and ascorbic acid present in the extract will therefore also react in the assay. Total phenolics were expressed as microgram of chlorogenic acid equivalents per gram of fresh weight.

Identification of the phenolic compounds was performed using a Dionex Ultimate 3000 system (Sunnyvale, CA) equipped with a diode array detector. A 5-μL aliquot was injected onto a Dionex C18 Acclaim PolarAdvantage II column (150 × 2.1 mm i.d.; 3 μm particle size) (Sunnyvale, CA). The mobile phases were (A) water with 0.1% formic acid and (B) acetonitrile with 0.1% formic acid. The flow rate was 0.35 mL min-1, and the column temperature was 35 °C. The 42-min gradient was as follows: 0–5 min, 0%–8% B; 5–10 min, 8%–15% B; 10–20  min, 15%–20% B; 20–27  min, 20% B linear; 27–34  min, 27%–100% B; 34–36 min, 100% B linear; 36–42 min, 0% B, re-equilibration time. Simultaneous monitoring was set at 254 nm (ellagitannins, flavonols), 280 nm (flavanols), 320 nm (hydroxycinnamic acids) and 520 nm (anthocyanins) for quantification. Polyphenol compounds were identified by their retention time and spectral data as compared to authentic standards. Ellagitannins, flavonols, flavanols, hydroxycinnamic acids and anthocyanins were quantified as gallic acid, catechin, rutin, chlorogenic acid and cyaniding-3-galactoside equivalents, respectively. Polyphenols were expressed in microgram per gram of fresh weight.

Carotenoid content was determined spectrophotometrically. Carotenoids were extracted as described earlier (Andre et al., 2006), and total carotenoid concentrations were determined using hexane as a solvent (Scott, 2001).

Acknowledgements

Tim Holmes is thanked for photography and Roshani Shakya for experimental support. This work was funded by the Ministry for Science and Innovation contract number C11X1007.

Ancillary