Stilbene synthase gene transfer caused alterations in the phenylpropanoid metabolism of transgenic strawberry (Fragaria x ananassa).

The gene encoding stilbene synthase is frequently used to modify plant secondary metabolism with the aim of producing the self-defence phytoalexin resveratrol. In this study, strawberry (Fragaria x ananassa) was transformed with the NS-Vitis3 gene encoding stilbene synthase from frost grape (Vitis riparia) under the control of the cauliflower mosaic virus 35S and the floral filament-specific fil1 promoters. Changes in leaf metabolites were investigated with UPLC-qTOF-MS (ultra performance liquid chromatography-quadrupole time of flight mass spectrometry) profiling, and increased accumulation of cinnamate, coumarate, and ferulate derivatives concomitantly with a decrease in the levels of flavonols was observed, while the anticipated resveratrol or its derivatives were not detected. The changed metabolite profile suggested that chalcone synthase was down-regulated by the genetic modification; this was verified by decreased chalcone synthase transcript levels. Changes in the levels of phenolic compounds led to increased susceptibility of the transgenic strawberry to grey mould fungus.

Introduction

Polyketide synthases are important enzymes in the synthesis of plant natural products (secondary metabolites) such as phytoalexins and flavonoid precursors. The type III polyketide synthase superfamily encompasses a structurally and functionally related, yet highly versatile group of enzymes including chalcone synthase (CHS; Ferrer ), stilbene synthase (STS; Austin ), 2-pyrone synthase (Eckermann ), bibenzyl synthase (Preizig-Müller ), and acridone synthase (Junghanns ). All polyketide synthases have similar reaction mechanisms involving elongation of the CoA ester derivative of the phenylpropanoid substrate by condensation reactions with acetate units from malonyl-CoA, and folding and aromatization of the polyketide intermediate. Consequently, various phenolic compounds such as naringenin chalcone and stilbene are formed through reactions catalysed by CHS and STS, respectively (Austin and Noel, 2003; Watanabe ). The wide variety of end-products catalysed by type III polyketide synthases is based on their different preference for phenylpropanoid substrates, the number of condensation reactions performed, and differences in the folding mechanism (Morita ).

The CHS reaction in the entry point to the flavonoid pathway has been the subject of numerous investigations. One of the first reports showed the unexpected silencing of the homologous endogenous gene by an additional CHS copy, resulting in the loss of flower colour (Napoli ). Since then, several studies on the influence of CHS silencing on flower pigmentation have been published (Metzlaff ; Que ; Fukusaki ). CHS-deficient plants have also revealed the importance of flavonoids in pollen development and plant reproduction (Mo ; Ylstra ; Schijlen ). The effects of CHS silencing on the metabolites in strawberry fruit have been reported recently (Hoffmann ; Lunkenbein ). Both studies demonstrated that silencing of CHS results in increased levels of metabolites upstream of the CHS step and reduction in the levels of downstream metabolites.

Unlike CHS which is present in virtually all higher plants, STS is restricted to a relatively few species such as grapevine (Vitaceae), peanut (Cyperaceous), and pine (Pinus), although closely related enzymes are being characterized and the number of plant species found to contain stilbene-related compounds is steadily increasing (Morita ; Eckermann ). Elicitation of the plant defence response by, for example, UV-irradiation or pathogens triggers the synthesis of the stilbene phytoalexin, resveratrol, in a reaction catalysed by STS. Since STS and CHS use the same precursor metabolites (Fig. 1), the introduction of STS for the purpose of metabolic engineering of resveratrol in plants should be a relatively straightforward approach. Following the landmark study of Hain and co-workers (1993), where the production of resveratrol was demonstrated in tobacco as a result of STS gene transfer, a similar approach has been used for a range of plant species, excluding strawberry. Various sources of STS have been used, including STS from Vitis vinifera (most commonly used), Vitis labruska, Vitis riparia (Kobayashi ), Parthenocissus henryana (Liu ), and PINOSYLVIN SYNTHASE from Pinus sylvestris (Seppänen ). Promoters used to express STS include the native STS promoter, the PR10 promoter (Coutos-Thévenot ), as well as other inducible or constitutive promoters [cauliflower mosaic virus (CaMV) 35S and UBIQUITIN; Fettig and Hess, 1999). The outcomes of these experiments vary in terms of metabolic changes and biological effects, and include enhanced resistance against the fungus Botrytis cinerea that causes the grey mould disease in crop plants such as strawberry (Schwekendiek ).

Fig. 1.

Metabolite flow from the shikimate pathway via the central phenylpropanoid pathway to flavonoid metabolism. Key intermediates (lower case) and enzymes (upper case) are indicated. Novel information obtained from strawberry metabolism in this study is included (phenylpropanoid glycosides). PAL, phenylalanine-ammonia lyase; STS, stilbene synthase; CHS, chalcone synthase; CoA, coenzyme A; Phe, phenylalanine; HHDP, hexa-hydroxy-di-phenyl.

Metabolomics approaches are increasingly applied for the determination of changes resulting from genetic modification of plants (Roessner ; Mattoo ; Bovy ; Morant ). Metabolite analysis is an essential tool for studying metabolically engineered plants, as the correct prediction of the changes is not always possible due to the lack of a full understanding of the regulation of the pathways as well as transport and accumulation of the (novel) product(s). Non-targeted metabolite profiling allows the detection of the overall consequences, including unpredictable changes, and may help in the interpretation of the effects of the gene transfer and, furthermore, shed light on the unknown steps and regulation in the engineered pathway. Here the characterization of genetically modified strawberry plants expressing the STS gene (NS-Vitis3) from frost grape (V. riparia) is reported. The genetic engineering approach was taken in order to control the B. cinerea infection that is believed to occur at the flowering stage of strawberry via the filaments of the stamen (Powelson, 1960). In order to achieve production of resveratrol in the infection route of the grey mould fungus, a flower-specific (fil1) promoter in addition to the general (CaMV 35S) promoter was used. While targeted metabolite analysis revealed no resveratrol or related compounds anticipated from the genetic engineering approach, non-targeted metabolite profiling with UPLC-qTOF-MS (ultra performance liquid chromatography-quadrupole time of flight mass spectrometry) did reveal changes in phenylpropanoid metabolism. The altered metabolite profile suggested down-regulation of CHS, which was corroborated at the mRNA level by quantitative reverse transcription-PCR (qRT-PCR). The biological effect of the modified metabolite profile was seen as increased sensitivity of the transgenic plants to grey mould infection.

Materials and methods

Plant material and growth conditions

Strawberry (Fragaria×ananassa, cv. Jonsok) plants were grown in the greenhouse under the following conditions: daylight 16 h, temperature 18–20 °C (night–day), and relative humidity 60–70%. To induce flowering, a 5–6 week short daylight period (13 h) was applied. Plants were grown in 12 cm pots in a peat–sand mixture (3:1) and fertilized weekly with Superex-9 (N 19%, P 5%, K 20%), supplied with micronutrients (Kekkilä, Finland). The 35S:NS-Vitis3 strawberry line was generated essentially as described by Schaart by transformation with the pCAMBIA2301 vector in which GUS (β-glucuronidase) was replaced by the V. riparia STS gene [NS-Vitis3, GenBank accession no. AF128861, Goodwin ; cloned by oligonucleotides E176 and E177 (see Supplementary Table S1 available at JXB online) with kanamycin as the selectable marker under the control of the CaMV 35S promoter]. Young, folded strawberry leaves were surface-sterilized and chopped into small pieces. Inoculation with Agrobacterium tumefaciens LBA4404 was carried out overnight on agar plates. Regeneration without selective antibiotics on MS medium supplemented with thidiazuron (TDZ; 2.0 mg l−1), indole butyric acid (IBA; 0.5 mg l−1), and 250 mg/l cefotaxime for elimination of Agrobacterium was allowed to proceed for 2 weeks, after which the selective antibiotic kanamycin (50 mg l−1) was added. The transgenic lines J47/1 and J47/2 contain the NS-Vitis3 gene under the control of the upstream region of the FIL1 gene (for a detailed description of the generation of the transgenics, see Hanhineva and Kärenlampi, 2007). For the construction of fil1:NS-Vitis3 binary vectors, the GUS marker gene in the T-DNA region of the pCAMBIA1391Z vector was replaced by NS-Vitis3 at the NcoI and BstEII restriction sites (oligonucleotides E176 and E177, Supplementary Table S1). An upstream fragment of the FIL1 gene (X57296) from Antirrhinum majus was amplified using oligonucleotides D917 and D918 (Supplementary Table S1) from the pBR322/FIL1 vector (Nacken ). The FIL1 upstream region was inserted in a multiple cloning site (BamHI/SmaI) in front of the STS gene in the pCAMBIA1391Z vector that contained hygromycin resistance as the selectable marker under the control of the CaMV 35S promoter. Runner propagation was done for all three transgenic lines together with the parental cultivar. The presence of STS was checked by PCR after each runner cycle in all of the transgenic lines.

Sample collection

For the qRT-PCR analysis, mature, fully opened strawberry flowers were collected from three transgenic lines and the parental cultivar, and the stamen (anther and filament separated), pistil, petal, sepal, and receptacle (containing the immature achenes) of 15–20 flowers on average were separated, pooled, and frozen immediately in liquid nitrogen. Leaf samples were collected from young, pale green leaves, and mature, fully expanded dark green leaves as pools of three individual plants in three replicates (nine plants divided into three groups). The leaves were frozen in liquid nitrogen and stored at –80 °C. Similar leaf samples were used for metabolite analysis, qRT-PCR, and western analysis.

Gene expression analysis

Total RNA was extracted according to Bowtell and Sambrook (2003) and cDNA was synthesized from DNase I-treated total RNA by using M-MuLV reverse transcription reagents (Fermentas UAB, Vilnius, Lithuania). Oligonucleotides for STS gene expression studies were designed to the regions with no homology to the strawberry CHS, covering an exon–intron junction (qSTS_IS_F and qSTS_IS_R, Supplementary Table S1 at JXB online). The PCR product was cloned and sequenced to verify correct amplification. The endogenous CHS levels were measured by oligonucleotides based on the Fragaria vesca CHS (AY017485) genomic sequence (Fa_chs_IS_F and Fa_chs_IS_R, Supplementary Table S1). The expression levels of STS and CHS were measured by comparing the Ct values of the target gene with the endogenous control gene DBP (Fa_Dbp_F and Fa_Dbp_R, Supplementary Table S1). Real-time SYBR Green quantitative PCR was set up in three 20 μl replicates containing: cDNA from 10 ng of total RNA, 1 μM forward oligonucleotide (Supplementary Table S1), 1 μM reverse oligonucleotide (Supplementary Table S1), and 2× DyNAmo HS SYBR Green qPCR kit F-410 (containing a hot-start version of the modified Thermus brockianus DNA polymerase, SYBR Green I dye, 5 mM MgCl2, dNTP mix including dUTP; Finnzymes, Espoo, Finland). Cycling parameters for the PCR were 95 °C for 16 min to activate the hot-start polymerase, followed by 40 cycles of 94 °C for 3 s, 59.4 °C for 15 s, and 72 °C for 20 s in the iCycler iQ instrument (Bio-Rad). For quantification of the relative gene expression levels, the 2–ΔΔCt method was used (Livak and Schmittgen, 2001). The expression level of STS and CHS was calculated from the 2–ΔΔCt equation, where ΔΔCt is the ΔCttarget–ΔCtlowest Ct value. The expression of each target gene was thus compared with the lowest expression level (set to 1.00) in the sample series. The PCR efficiency of the oligonucleotides was analysed by using a 5-fold dilution series of strawberry cDNA, with a slope value of the linear regression curve set up with the least squares method. A slope value of –3.3 indicated the best PCR efficiency. The PCR conditions were optimized to achieve the same PCR efficiency with both oligonucleotide pairs in the same cycling conditions. The threshold position was set in every analysis to level 5000 in order to achieve comparable data.

Metabolite analysis by UPLC-qTOF-MS

Frozen plant material (young and fully expanded leaves, and whole flowers) was ground under liquid nitrogen, and 3 ml of 80% methanol was added per 1 g of powder. The suspension was sonicated for 20 min at room temperature, occasionally vortexed, centrifuged, filtered (0.22 μm PTFE filter, Acrodisc, PALL), and stored at –20 °C. Metabolite analysis was performed by UPLC-UV(PDA)-qTOF-MS (Waters Premier qTOF, Milford, MA, USA) as described previously (Hanhineva ). Peak picking and data processing were performed by MarkerLynx 4.1 software (Waters, Inc.) with the following parameters: retention time (Rt) range 1–23.5 min; mass tolerance 0.01 Da; peak width and baseline noise automatically calculated by the program; mass window at 0.05 Da; Rt window at 0.2 min. Automatic smoothing was applied and isotopic peaks were removed from the data. The collected peak lists with Rt, m/z, and peak area intensities were further processed with Excel software, and the statistically significant differences in individual markers between the wild type and each of the transgenic lines were demonstrated by pairwise t-test (two-tailed, two-sample unequal variance). The data were analysed by principal component analysis (PCA) within the Marker Lynx software on the mean centre of the peak area intensities with pareto scaling.

Two-dimensional western analysis

Proteins from three biological replicates of fully expanded leaves (1 g) of wild-type and 35S:NS-Vitis3 plants were extracted as described by Koistinen . The proteins were dissolved in two-dimensional electrophoresis (2-DE) sample buffer: 7 M urea, 2 M thiourea, 4% CHAPS, 1% dithiothreitol (DTT), and 2% (v/v) Bio-Lyte 3/10 ampholyte (Bio-Rad). The total protein concentration was analysed with the Bio-Rad Protein Assay Dye reagent, and 100 μg of protein was used for the 2-DE analysis performed as described by Lehesranta , except that the first dimension isoelectric focusing was performed using 7 cm immobilized pH gradient (4–7) strips (Amersham Biosciences) with the step-and-hold focusing program: 500 V for 30 min, 1000 V for 30 min, 5000 V for 1.5 h. The second dimension was performed using 12% SDS–PAGE (Bio-Rad Minigel II apparatus). The proteins were transferred onto a PVDF membrane (Immobilon P; Millipore) and visualized by SYPRO Ruby protein blot stain (Bio-Rad) according to the manufacturer's instructions. Gel images were acquired with the FLA-3000 fluorescent image analyser (Fuji Photo Film) using excitation and emission wavelengths of 473 nm and 580 nm, respectively. After staining, the same membranes were used for western analysis (Koistinen ) using a rabbit polyclonal antiserum raised against the recombinant STS protein (Giovinazzo ) as the primary antibody (1:1000 dilution), and alkaline phosphatase-conjugated anti-rabbit immunoglobulin (Zymed) as the secondary antibody.

Fluorescence microscopy

A fresh 1×1 cm disc of strawberry leaf was mounted with glycerol on a glass slide and enclosed with a coverslip. The sample was squashed between the glasses with an apparatus occupying two parallel metal discs, and immediately viewed with a Nikon Microphot-FXA microscope equipped with a fluorescence filter allowing excitation at 330–380 nm and emission above 420 nm. The autofluorescent images were captured with an RS Photometrix CoolSNAP digital camera and processed with CoolSNAP software.

Botrytis inoculation

Botrytis cinerea Pers. Fr strain B.05.10 (Quidde ) was maintained on potato dextrose agar containing a 250 g l−1 homogenate of the leaves of Phaseolus vulgaris, and the spores were harvested in the inoculation medium (Gamborg's medium B5, 10 mM sucrose, 10 mM KH2PO4, 0.05% Tween-80) as described by Muckenschabel et al. (2002). The inoculum concentration was adjusted to 1×105 conidia ml−1. Detached mature leaves (seven leaves of each genotype) were placed on a Petri dish containing water agar (0.8%), and cut petioles were immersed in the agar. The leaf surface was wounded in three places by punching with a syringe needle, and a 3 μl droplet of the conidial suspension was placed immediately on the wound. The plates were sealed with parafilm and kept at room temperature in natural light. The leaves were photographed and the lesion area was measured by the ImageJ program (Rasband, 1997–2007).

Results and discussion

Expression of a stilbene synthase-encoding transgene in strawberry

The V. riparia gene encoding STS (NS-Vitis3) was introduced in the Norwegian strawberry cv. Jonsok under the control of CaMV 35S promoter and a flower-specific promoter fil1 originating from A. majus (Nacken ). The resulting genetically modified (GM) strawberry lines are designated as 35S:NS-Vitis3 (35S promoter), and the lines expressing NS-Vitis3 under the fil1 promoter are referred to as J47/1 and J47/2. Quantitative PCR analysis indicated accumulation of NS-Vitis3 mRNA in different parts of all transgenic lines but not in the parental strawberry (Fig. 2). The highest expression levels were observed in the young and fully expanded leaves of the transgenic strawberries, being highest in the line that contained the transgene under the control of the 35S promoter. The fil1 promoter gives filament-specific expression in A. majus (Nacken ) but in strawberry the promoter was leaky and expression was also found in the leaves of the lines J47/1 and J47/2, albeit at lower levels than with the 35S promoter (Fig. 2). A possible explanation for the non-specific expression of the transgene might be that the pCAMBIA vectors used in the transformation have a strong 35S promoter driving the expression of the hptII selectable marker gene, and this promoter has been shown to interfere with the expression of other genes in the construct (http://www.cambia.org/daisy/bios/585.html).

Fig. 2.

Quantitative PCR analysis of the STS expression levels in the genetically modified strawberry lines. The results shown are means (with the standard error) of biological samples in triplicate values based on the 2–ΔΔCt method. 35S:NS-Vitis3, n=3; J47/1 floral organs, n=1; fruits and leaves, n=3; J47/2 floral organs, n=2; fruits and leaves, n=3. No expression was detected in the wild-type controls.

In order to examine the presence of NS-Vitis3 protein in the 35S:NS-Vitis3 line, two-dimensional western analysis with rabbit antiserum raised against recombinant STS protein from V. vinifera was performed (Giovinazzo ). Several cross-reacting spots were detected in the protein extract prepared from the fully expanded leaves of the 35S:NS-Vitis3 plants (Fig. 3). This was not unexpected, as non-specific cross-reaction has also been observed in STS-modified white poplar using the same antibody (Giorcelli ). The NS-Vitis3 protein (AF128861) has a calculated molecular weight of 42.72 kDa and pI of 6.23 (ExPASy Proteomics server). These values match rather well with the only clearly differently expressed spot in the 2D western blot (Fig. 3, arrow). However, since the amount of protein in the spot was too low to permit its identification by mass spectrometry, the expression of NS-Vitis3 at the protein level could not be verified.

Fig. 3.

Leaf proteome of wild-type (WT) and transgenic 35S:NS-Vitis3 strawberry. (A) Sypro Ruby staining of the proteome. (B) Western blot with anti-STS as the primary antibody. The area marked by a square on the protein blot is shown in the western blot. A different spot detected in the 35S:NS-Vitis3 line is indicated by an arrow.

Altered metabolite pattern in the transgenic strawberries

NS-Vitis3, as well as other STS-encoding genes, catalyses the synthesis of resveratrol in several plant genera. In a number of cases, transformation of plants with STS has led to the production of resveratrol (Hain ; Leckband and Lörtz, 1998; Coutos-Thévenot ). Methanolic extracts of the flowers and leaves of the transgenic strawberry lines were screened by UPLC-qTOF-MS using commercial trans-resveratrol as well as cis-resveratrol (obtained by exposing trans-resveratrol standard solution to UV/visible light) as reference compounds. In addition, the most common derivatives (i.e. piceid and viniferins) were searched for by using mass spectral information available in the literature (Püssa ). Extensive analysis revealed no resveratrol or related metabolites in any of the three lines or the parental strawberry line. The NS-Vitis3 gene used in this study originates from V. riparia (Goodwin et al., 2005) whereas the STS gene most often used in genetic modification is VST1 from V. vinifera. However, another V. riparia STS gene (pBSRIP) has been shown to give rise to glucosylated resveratrol in the kiwifruit (Kobayashi ). The amino acid alignment of four different STS sequences shows that there are changes in 11–17 amino acids in the protein used in this study (AAF00586) when compared with other sequences including V. riparia (BAB20978, Kobayashi ) and V. vinifera (P28343, Coutos-Thevenot et al., 2001; Giorcelli ; Zhu ) that have been successfully used in gene transfer experiments. However, none of the changes is located at the active site of the enzyme that is conserved in all of the sequences compared here (prosite pattern PS0041, http://au.expasy.org) (Supplementary Figure S2). Based on these considerations, it is unlikely that the choice of STS gene was the reason for the absence of resveratrol. In the course of this study, resveratrol was found from the achenes and receptacle of several conventional strawberry cultivars (not including cv. Jonsok) after multistep purification (Wang ). Although the direct methanol extraction used in the present study has been proven to be suitable for the extraction of resveratrol from several transgenic plants (Szankowski ; Zhu ; Rühmann ), resveratrol has not been found in methanol or acetone extracts of strawberry (Aharoni ; Määttä-Riihinen ; Aaby ). This suggests that resveratrol is not present in all strawberry cultivars or, more probably, that it is present at very low levels compared with the overall phenolic content of strawberry and can be detected only upon elicitation or after targeted purification.

While resveratrol or its derivatives could not be detected in the targeted metabolite analysis, a clear patterning of metabolite markers was observed in PCA (Fig. 4) when non-targeted metabolite profiling was carried out on the line having the strongest transgene expression (fully expanded leaves of the 35S:NS-Vitis3 line, see Fig. 2). This suggested that the genetic modification gave rise to alterations in the metabolite content other than the expected production of resveratrol. The most prominent differences in the fully expanded leaves of the 35S:NS-Vitis3 line and parental plants were already detectable in the total ion chromatograms (TICs; Fig. 5A, B). The altered metabolites were identified as described in related work on strawberry flower (Hanhineva ). Data analysis revealed a total of 22 metabolites with statistically significantly altered amounts (base peak response areas) between the 35S:NS-Vitis3 line and parental plants (Table 1). In the young leaves of the 35S:NS-Vitis3 line, four and nine metabolites were down- and up-regulated, respectively. Sixteen metabolites were statistically significantly altered in the fully expanded leaves, 11 being down-regulated and five up-regulated in the 35S:NS-Vitis3 line. Apart from phenylalanine, the significantly differing metabolites were either flavonol or phenolic acid derivatives. The major flavonols in strawberry leaves were kaempferol (six derivatives) and quercetin (two derivatives), all of which were significantly decreased in the fully expanded leaves of the 35S:NS-Vitis3 line, the acylated kaempferol-coumaroyl-glucosides also being decreased in the young leaves of the 35S:NS-Vitis3 line. Two isomers of an unidentified compound with the m/z value of 523.22 [ES(–)] eluting at two retention times were strongly down-regulated and virtually undetectable in the 35S:NS-Vitis3 line (Table 1). In contrast to the flavonols, phenolic acid derivatives were present at higher levels in the leaves of the 35S:NS-Vitis3 line compared with the parent (nine and five derivatives up-regulated in young and fully expanded leaves of the transgenic plants, respectively).

Fig. 4.

Principal component analysis of the mass signals in ES(–) in leaves of the 35S:NS-Vitis3 and wild-type (WT) plants.

Fig. 5.

Total ion chromatograms (TICs) of (A) wild-type (WT) and (B) 35S:NS-Vitis3 leaf extracts in ES(–); the main differences between the WT and 35S:NS-Vitis3 TICs are indicated by arrows. Reconstructed ion chromatograms of the m/z 147.04 ion representing the coumaroyl moiety in ES(+) in (C) WT and (D) 35S:NS-Vitis3. (This figure is available in colour at JXB online.)

Table 1.

Relative amounts of metabolites identified in strawberry leaves by the UPLC-qTOF-MS analysis

			young leaves				fully expanded leaves
Metabolite	Rt	m/z ES(–)	WT (mean ±SE)	35S:NS-Vitis3 (mean ±SE)	Ratio	P-value	WT (mean ±SE)	35S:NS-Vitis3 (mean ±SE)	Ratio	P-value
Flavonol derivatives
Kaempferol hexose derivative	9.24	447.056	87.5±7.7	114.4±9.6	↑1.31	0.0981	228.9±2.9	187.3±6.5	↓1.22	0.0120
Kaempferol hexose glucuronide	9.24	623.124	69.4±4.5	54.3±10.1	↓1.28	0.2714	218.3±23.1	78.4±9.1	↓2.79	0.0159
Kaempferol pentose glucuronide	10.49	593.114	2853.8±101.5	2665.6±248.3	↓1.07	0.5396	4268.7±166.7	2538.7±162.2	↓1.68	0.0018
Kaempferol glucuronide	12.00	461.072	366.4±60.6	246.5±43.0	↓1.49	0.1893	772.9±119.0	265.3±69.7	↓2.91	0.0307
Kaempferol coumaroyl glucoside	17.48	593.129	2033.1±180.0	836.9±208.1	↓2.43	0.0127	1209.3±202.8	301.7±47.3	↓4.01	0.0404
Kaempferol coumaroyl glucoside	18.29	593.131	288.7±4.1	130.3±19.9	↓2.22	0.0127	257.4±39.8	100.9±8.7	↓2.55	0.0532
Quercetin glucuronide	10.27	477.066	1460.8±179.9	1364.4±306.8	1.07	0.8028	1036.3±84.2	630.1±117.4	↓1.64	0.0540
Quercetin pentose glucuronide	8.89	609.108	3187.5±164.5	2943.4±93.6	↓1.08	0.2832	3128.4±71.6	1866.6±80.6	↓1.68	0.0003
Phenolic acid derivatives
Coumaric acid hexose	5.24	325.092	18.1±9.2	102.7±42.9	↑5.67	0.1827	152.2±44.3	1190.5±425.5	↑7.82	0.1333
Coumaric acid hexose	5.63	325.092	17.9±8.6	56.4±12.4	↑3.16	0.0709	34.7±5.3	141.1±36.3	↑4.06	0.0964
Coumaric acid hexose	5.88	325.091	1.7±0.5	31.9±4.6	↑18.23	0.0213	51.8±17.2	328.9±118.5	↑6.34	0.1417
Coumaroyl quinic acid	6.55	337.091	196.7±36.1	363.8±58.5	↑1.85	0.0847	192.9±20.3	447.5±31.3	↑2.32	0.0041
Chlorogenic acid	6.60	353.086	85.2±17.7	155.8±16.5	↑1.83	0.0436	133.3±6.4	225.5±17.1	↑1.69	0.0219
Eutigoside A	15.04	445.149	243.2±107.2	1175.5±182.2	↑4.83	0.0185	1539.2±251.5	4174.5±432.4	↑2.71	0.0112
Eutigoside A isomer	16.30	445.149	20.9±5.2	155.6±27.7	↑7.44	0.0360	296.1±46.4	1112.3±291.7	↑3.76	0.1040
Hydroxyethylphenyl-coumaroyl-glucopyranoside	13.72	445.149	9.7±5.1	40.4±13.4	↑4.15	0.1376	186.1±6.1	746.3±129.6	↑4.01	0.0493
Grayanoside A	15.70	475.160	146.2±13.9	321.3±64.7	↑2.20	0.1078	598.9±67.6	992.5±210.2	↑1.66	0.1946
Hydroxyphenylethyl-caffeoyl-glucopyranoside	13.06	461.145	184.3±33.5	699.9±133.5	↑3.80	0.0535	931.9±98.9	1732.0±317.9	↑1.86	0.1178
Hydroxyphenylethyl-caffeoyl-glucopyranoside isomer	11.70	461.144	2.0±1.0	7.0±1.0	↑3.96	0.0228	48.0±6.0	96.0±18.0	↑1.97	0.1051
Di-coumaroyl-glucopyranoside	14.31	471.127	0.2±0.2	1.7±0.5	↑8.47	0.0682	20.2±5.7	104.9±41.4	↑5.19	0.1748
Di-coumaroyl-glucopyranoside	17.33	471.128	0.1±0.1	2.2±1.0	↑18.78	0.1707	16.3±6.8	285.0±146.6	↑17.48	0.2081
Coumaroyl-galloyl-glucopyranoside	11.21	477.102	126.9±28.3	445.2±58.0	↑3.51	0.0172	44.0±17.7	151.7±50.0	↑3.45	0.1533
Unidentified coumaroyl-phenylpropanoid	14.02	459.129	11.1±5.5	40.8±7.8	↑3.68	0.0418	143.6±9.8	420.9±90.6	↑2.93	0.0906
Unidentified coumaroyl-phenylpropanoid	16.65	445.112	30.7±9.1	97.3±7.0	↑3.18	0.0054	57.2±9.0	95.7±2.7	↑1.67	0.0409
Gallo- and ellagitannins
Galloylquinic acid	1.42	343.065	3843.3±869.2	2584.2±1468.3	↓1.49	0.5103	3166.7±227.0	2612.5±414.6	↓1.21	0.3232
Galloyl hexose	1.22	331.065	2146.2±67.2	2864.4±270.1	↑1.33	0.1095	244.7±16.7	221.4±37.4	↓1.11	0.6129
Bis (HHDP) glucose	2.57	783.068	2032.2±201.8	2535.1±150.8	↑1.23	0.1223	1784.1±21.7	1811.6±230.1	↑1.01	0.9162
Galloyl (HHDP) glucose	5.29	633.072	2182.8±285.5	1882.8±256.2	↓1.16	0.4784	254.0±31.6	197.8±22.4	↓1.28	0.2287
Galloyl bis (HHDP) glucose	8.07	935.080	1529.2±78.1	1325.9±59.2	↓1.15	0.1118	102.5±11.1	96.0±13.5	↓1.07	0.7291
Penta galloyl glucose	11.67	939.114	189.3±12.1	162.6±41.6	↓1.16	0.5927	4.6±2.3	1.0±1.0	↓4.40	0.2642
Tri galloyl glucose	6.93	635.087	965.2±218.1	1191.2±26.7	↑1.23	0.4092	98.8±12.0	160.4±39.3	↑1.62	0.2535
Tri galloyl (HHDP) glucose	9.29	937.096	446.9±38.3	350.9±14.8	↑1.27	0.1157	78.0±14.1	82.2±29.5	↑1.05	0.9058
Proanthocyanidins
Catechin	4.88	289.070	2118.0±654.0	2700.0±408.0	↑1.26	0.4991	3810.0±228.0	3756.0±178.0	↑1.14	0.7915
Procyanidin dimer	4.40	577.133	1525.0±535.0	1902.0±399.0	↑1.25	0.6043	2597.0±27.0	2975.0±114.0	↑1.15	0.0729
Procyanidin trimer	4.70	865.201	314.0±140.0	310.0±99.0	↓1.01	0.9812	828.0±69.0	821.0±46.0	↓1.01	0.9378
Other metabolites
Phenylalanine ES(+)	2.18	166.086	127.2±4.3	129.5±15.6	↑1.01	0.8962	143.3±8.8	95.4±5.6	↓1.51	0.0150
Unknown	11.92	477.101	414.3±59.0	380.1±72.0	↑1.09	0.7312	116.7±10.0	145.7±13.0	↑1.25	0.1575
Unknown	12.80	523.218	127.1±16.8	0.3±0.2	↓387	0.0172	141.2±17.4	0.0±0.0	↓na	0.0149
Unknown	12.45	523.218	252.0±40.4	0.3±0.1	↓874	0.0249	231.8±21.8	0.1±0.1	↓1997	0.0087

The metabolite values are averages of the relative peak response area in three replicates with standard error values. The up- (↑) or down- (↓) regulation ratio is shown and is in bold when there is a statistically significant difference (P <0.05).

The characteristic fragment of the coumaric acid moiety after loss of water in ES(+) is m/z 147.04. The reconstructed ion chromatograms of the m/z 147.04 ion demonstrated a clear difference between the 35S:NS-Vitis3 line and parental leaf samples in the relative abundance of coumaric acid derivatives not only in the early region of the chromatogram, where the phenolic acids eluted (Rt 3–7 min), but also later (Rt 13–18 min) (Fig. 5C, D). The late-eluting metabolites could not be assigned directly by their UPLC-MS characteristics, and were consequently elucidated by nuclear magnetic resonance (NMR) spectroscopy to contain phenylethyl derivatives of phenylpropanoid glucoside compounds, that have not been previously characterized in strawberry (Hanhineva ). In addition, in the chromatogram region where the phenolic acid derivatives eluted (Rt 3–7 min; Fig. 5C, D) additional similar compounds were detected. As this region was extremely dense in metabolites and would have required optimized conditions for better separation, these metabolites remained uncharacterized in the current study. Interestingly, the only coumaroyl-containing metabolites that were present at lower levels in the 35S:NS-Vitis3 line were the two acylated kaempferol hexose molecules (Table 1), in agreement with the decrease in the levels of other flavonols in the 35S:NS-Vitis3 line. The majority of the compounds that were unchanged between the 35S:NS-Vitis3 line and parental plants were either proanthocyanidins or derivatives of ellagitannin metabolism (Table 1; Fig. 1).

As the above-mentioned metabolite classes were screened in the fil1 promoter-containing lines, parallel results were obtained (Fig. 6). The compound assigned as bis-HHDP-glucose, one of the ellagitannin compounds present at a rather constant level in the 35S:NS-Vitis3 line (Table 1), showed no differences between the three transgenic and parental lines (Fig. 6A). Similarly, both isomers of kaempferol coumaroyl glucoside exhibited decreased amounts when the metabolite levels in all three transgenic lines were compared (Fig. 6D). Additionally, as in the 35S:NS-Vitis3 line, signals for phenolic acid-containing metabolites were increased in both fil1 promoter-containing lines. The most prominent increases were found in coumaric acid hexose (Fig. 6B) and coumaroyl-galloyl-glucopyranoside (Fig. 6C). A slight difference between the transgenic lines was that quercetin pentose glucuronide (Fig. 6E) and kaempferol pentose glucuronide showed enhanced levels in the J47/1 and J47/2 lines, whereas a slight decrease was found in the 35S:NS-Vitis3 line. Overall, the results obtained with the two fil1 promoter-containing lines support the findings obtained with the 35S:NS-Vitis3 line.

Fig. 6.

Metabolite levels in the three transgenic and parental lines. Values are average areas of each metabolite base peak (molecular ion) in three biological replicates with standard error. The significance level on pairwise comparision against the wild type (WT) is marked as *P ≤0.1, **P ≤0.05, ***P ≤0.01.

Despite the fact that STS has been widely introduced in transgenic plants, very few papers report metabolites other than resveratrol or its derivatives in those plants. Since STS competes with CHS for the same precursors in the rate-limiting step of the flavonoid pathway (Muir ; Fig. 1), it could be expected that flavonol levels are decreased due to precursor competition. However, only one paper describes a slight decrease in flavonols in transgenic plants modified with STS (Rühmann ), whereas others conclude that substantial resveratrol production has no effect on the flavonol concentration (Giovinazzo ; Schwekendiek ). In tomato, the decrease in the precursors was suggested to be due to their enhanced use by both endogenous CHS and the introduced STS (Giovinazzo ).

Down-regulation of chalcone synthase in the transgenic strawberries

A logical conclusion from the metabolite analyses which indicated systematic changes in the levels of metabolites associated with the central phenylpropanoid and flavonoid pathways was that the endogenous CHS was down-regulated in the transgenic lines. Therefore, the endogenous CHS mRNA levels were analysed by quantitative PCR.

The CHS mRNA levels in the non-GM parental plants were, on average, 7-fold in the young leaves compared with the fully expanded leaves (Fig. 7). Developmental expression of CHS in strawberry leaves has not been reported previously. As the young, expanding leaves are more vulnerable than the fully expanded leaves, it is feasible that the young leaves have stronger CHS expression to ensure effective production of protective flavonoids (Lokvam and Kursar, 2005).

Fig. 7.

Quantitative PCR analysis of CHS expression levels in young and fully expanded strawberry leaves. The results shown are means (with the standard error) of biological samples in triplicate values based on the 2–ΔΔCt method. WT (wild-type) samples, n=4; 35S:NS-Vitis3 samples, n=4 (fully expanded leaves), n=3 (young leaves). The significance level is marked as *P ≤0.1, **P ≤0.05, ***P ≤0.01.

While in the young leaves of the 35S:NS-Vitis3 line the CHS mRNA level was only slightly lower than in the non-GM parent, it was dramatically diminished in the fully expanded leaves, being <8% of that in the parent (Fig. 7). In line with this, the CHS expression in the fully expanded leaves of J47/1 and J47/2 lines was also lower than in the parental strawberry (Fig. 8). The decrease in CHS mRNA levels in the J47/1 and J47/2 lines was not as dramatic as in the 35S:NS-Vitis3 line, which can be explained by the lower level of NS-Vitis3 mRNA in the leaves of these lines (see Fig. 2).

Fig. 8.

Quantitative PCR analysis of CHS expression levels in fully expanded strawberry leaves in the three transgenic and parental lines. The results shown are means (with the standard error) of three biological samples in triplicate values based on the 2–ΔΔCt method. The significance level in pairwise comparision against the wild type (WT) is marked as *P ≤0.1, **P ≤0.05, ***P ≤0.01.

These results suggested that the expression of the STS gene resulted in silencing of the endogenous CHS mRNA. One possible mechanism for the down-regulation is RNA interference (RNAi; Pickford and Cogoni, 2003; Kusaba 2004), in which the formation of double-stranded RNA triggers gene silencing via degradation of mRNAs derived both from the inserted gene and from the endogenous homologous gene. The sequences of strawberry CHS (Fragaria×ananassa ‘Nyoho’, AY997297) and NS-Vitis3 (V. riparia, AF128861) have 69% identity (Supplementary Fig. S1 at JXB online). In mammalian cells, as few as 11 contiguous identical nucleotides were found to be sufficient for silencing (Jackson ). Sequence comparison indicated that seven such stretches varying in length from 11 to 19 nucleotides occur between strawberry CHS and NS-Vitis3 (Supplementary Fig. S1). The recent observation of the presence of resveratrol in certain strawberry cultivars suggests that polyketide-type enzymes other than CHS are present in strawberry, but no STS genes have yet been characterized. As the STS genes can be highly homologous (>99%; Goodwin ) between different species, silencing of a putative endogenous strawberry STS gene would also be feasible. This could lead to the silencing of CHS via the phenomenon of spreading of RNA silencing (Van Houdt ). One explanation for the presence of NS-Vitis3 mRNA in the leaves of the transgenic lines is that the transgene is expressed at excess levels compared with CHS, and would still be detectable even though part of the mRNA is degraded. Different levels of silencing are characteristic for the RNAi phenomenon, a fact that is exploited, for example, in the studies of genes whose absence is lethal to the plant. An alternative explanation for the changes in the CHS expression levels may lay in the regulatory control of the phenylpropanoid pathway; the introduction of the transgene may have resulted in changes in the control of gene expression, shown as diminished CHS mRNA levels.

None of the previous studies on transgenic plants modified by introducing STS demonstrates down-regulation of endogenous genes, whereas the introduction of CHS has caused silencing of the transferred gene and the endogenous homologue (Napoli ). Similarly, down-regulation of a lignin biosynthesis enzyme cinnamoyl-CoA reductase resulted in a decrease of lignin synthesis and a concomitant increase in the precursor ferulic acid, observed as ester conjugates (Leplé ).

The metabolite profiles previously observed in the fruits of strawberries with silenced CHS (decrease in flavonols, increase in phenolic acid derivatives; Hoffmann ; Lunkenbein ) were similar to those observed in the present study. It was further shown that the amount of ellagitannin compounds remained at a rather constant level. This can be explained by the fact that ellagitannins are synthesized from gallic acid units which are produced directly from the shikimate pathway and should not be affected by the down-regulation of CHS. Interestingly, the proanthocyanidin levels did not change to the same extent as did the flavonols synthesized earlier in the same pathway (Fig. 1, Table 1). This suggests that there might be a flux control at the end of the pathway or possibly CHS-independent regulation of the different branches of the flavonoid pathway. In the fruits of CHS-silenced strawberry, the proanthocyanidins have been shown previously to be either decreased (Lunkenbein ) or not affected (Hoffmann ). The present results also suggested that flavonol synthesis was effective in the young leaves. The flavonol metabolites observed in the fully expanded leaves of transgenic lines might derive from the earlier developmental stage, and thus occur in the fully expanded leaves at higher levels than would be expected from the nearly complete absence of CHS mRNA from the transgenic lines at that developmental stage.

Phenolic acid derivatives exhibit blue-green fluorescence when excited by UV-A (Lichtenthaler and Schweiger, 1998; Liakopoulos ). All of the metabolites that showed increased accumulation in the leaves of the transgenic strawberries contained phenylpropanoids with a phenolic acid moiety. The difference in the phenolic content of the transgenic 35S:NS-Vitis3 line and the parent was large enough to be clearly visualized by epifluorescence microscopy. Excitation of the autofluorescent phenolic compounds resulted in a clear blue fluorescence in the fully expanded leaves of the 35S:NS-Vitis3 line while it was absent in the leaves of the parental plants (Fig. 9A).

Fig. 9.

(A) Autofluorescence observed in the leaf cells of the 35S:NS-Vitis3 strawberries by fluorescence microscopy with excitation at 330–380 nm and emission above 420 nm. (B) Botrytis cinerea inoculation of the leaves of the wild type (WT) and 35S:NS-Vitis3 photographed 5 d after inoculation. The lesion area is highlighted. The P-value of the pairwise t-test was calculated from the average lesion areas.

Susceptibility of the transgenic strawberry to grey mould infection

Significant alterations in the metabolite profile of the 35S:NS-Vitis3 strawberry line compared with the parental strawberry prompted a test of its susceptibility to fungal infection. A common strain of the grey mould-inducing fungus B. cinerea Pers.: Fr strain B.05.10 (Quidde ) was used to infect leaf discs of the transgenic and parental plants. Of the 21 conidia-containing droplets applied on seven leaves of both the parent and transgenic line, 15 and 17, respectively, developed clear lesions. Pair-wise comparison indicated that the lesion areas in the leaves of the transgenic line were statistically significantly larger (P=0.017) than those developed on the parental leaves (Fig. 9B). This suggests that the depletion of the flavonol content resulted in increased susceptibility to Botrytis infection, which is in agreement with the reports indicating that flavonols contribute to strawberry defence against fungi (Terry ; Halbwirth ; Hukkanen ).

Products of the flavonoid pathway are widely known to contribute to plant defence (Grotewold, 2006), also in strawberry (Terry ; Halbwirth ; Hukkanen ). In the present study, the compounds with decreased contents in the leaves of the 35S:NS-Vitis3 line were all representatives of the flavonol group. It was thus not surprising to discover that the leaves showed increased susceptibility to grey mould (Fig. 9B), which further supports the crucial role of flavonols in defence. There was one compound (with two isomers) almost completely missing from the leaves of the 35S:NS-Vitis3 line while being present in the parental strawberry (Table 1; Rt 12.45 and 12.80; m/z 523.22), and it cannot be ruled out that the absence of this compound led to the increased susceptibility. Proanthocyanidins known to contribute to the defence mechanisms of various plants (Dixon ) remained rather constant in the 35S:NS-Vitis3 line, suggesting that this group of compounds may not have a crucial role in strawberry defence against grey mould. In general, the mechanisms responsible for the biochemical responses of strawberry upon pathogen or other environmental stress challenge are rather poorly understood and the understanding of the defence response is far from clear. The findings in the present study thus open up interesting possibilities for further studies on the defence mechanism.

Conclusions

The introduction of the NS-Vitis3 gene into strawberry caused major changes in the biosynthesis of phenolic compounds. The results suggested that endogenous CHS was down-regulated, leading to the accumulation of metabolites of the central phenylpropanoid pathway. Although conclusive interpretation of the observed changes was not possible, the detailed analysis proved that non-targeted profiling rather than targeted analysis is required to detect non-predictable changes possibly related to genetic modification. Additionally, this study showed that the current knowledge of the structure and regulation of the phenylpropanoid pathway in strawberry is incomplete, as major unexplainable changes in gene expression were observed, among them increased synthesis of newly identified compounds.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Sequence homology between the coding sequences of Fragaria×ananassa CHS and Vitis riparia STS genes

Fig. S2. Alignment of the amino acid sequence of stilbene synthase proteins from Vitis species.

Table S1. Oligonucleotides used in this study