Petunia hybrida W115 was transformed with a Clarkia breweri S-linalool synthase cDNA (lis). Lis was expressed in all tissues analysed, and linalool was detected in leaves, sepals, corolla, stem and ovary, but not in nectaries, roots, pollen and style. However, the S-linalool produced by the plant in the various tissues is not present as free linalool, but was efficiently converted to non-volatile S-linalyl-β-d-glucopyranoside by the action of endogenous glucosyltransferase. The results presented demonstrate that monoterpene production can be altered by genetic modification, and that the compounds produced can be converted by endogenous enzymatic activity.
Isoprenoids are the largest family of natural compounds, and are based on varying numbers of C5 isoprene units. In plants they play important roles: for example, carotenoids are involved in photosynthesis, and mono- and sesquiterpenoids are important in cell-to-cell interactions and in communication between organisms. Monoterpenes, the C10 branch of the isoprenoid family, were first investigated for their economic importance as flavour and fragrance additives in foods and cosmetics. Anticarcinogenic effects (Crowell and Gould, 1994; Mills et al., 1995; Yu et al., 1995); antimicrobial properties (Belaiche et al., 1995); and antifungal properties (Vaughn and Spencer, 1991) have also been demonstrated for some monoterpenes. They have also been shown to be of ecological significance (Harborne, 1991), for instance in the interactions between plants and insects, and between plants and micro-organisms. Commonly used monoterpenes are often produced by chemical synthesis, but at present interest in naturally produced monoterpenes is increasing. Plants producing monoterpenes have been investigated in more detail, and this has resulted in a better understanding of the biochemical pathways leading to the formation of monoterpenes and their derivatives.
Linalool is an acyclic monoterpene alcohol that has a peculiar creamy, floral, sweet taste (Arctander, 1969). In Clarkia breweri linalool, among other compounds, is responsible for attracting pollinating moths (Raguso and Pichersky, 1995). Linalool can also act as a repellent against aphids, as shown in a study with Myzus persicae (Hori, 1998); it is also one of the volatile compounds released as a semiochemical after herbivore attack in some plant species (Pare and Tumlinson, 1997; Rose et al., 1996; Weissbecker et al., 1997), and as such may attract predators of the herbivores.
Many monoterpenes have been reported to be detrimental to biological structures (Weidenhamer et al., 1993). Linalool, for example, softens the structure of potato tubers incubated under a continuous flow of the compound (Vaughn and Spencer, 1991). When not emitted, linalool, like other terpenols, can be detoxified in many plant species by conjugation, for example as a glycoside of β-d-glucose or as a disaccharide. In this form it could also be transported via the phloem to other plant tissues (Raguso and Pichersky, 1999). In a later stage, bound linalool could still be emitted by the action of glycosidase enzymes (Raguso and Pichersky, 1999).
Increasing knowledge about isoprenoid biosynthesis has led to various ideas on the genetic modification of monoterpenoid composition and production in plants. Now there are opportunities to alter essential oil characteristics, floral scent profiles, ecological interactions with the environment of a plant, and production of larger quantities of highly valuable monoterpenes (Haudenschild and Croteau, 1998; Lange and Croteau, 1999; McCaskill and Croteau, 1997; McCaskill and Croteau, 1998). Recently, peppermint plants have been transformed with limonene synthase, but no drastic alterations were found in the terpenic profile (Krasnyanski et al., 1999). We are still unable to predict interactions between an inserted gene and the host genome and its subsequent influence on host metabolism (Buiatti and Bogani, 1995). Therefore we investigated monoterpene production in a model system, for use as a proof of concept of metabolic modification. Petunia hybrida W115 is an easily transformable plant (Horsch et al., 1985) that contains virtually no monoterpenes in floral tissues or any other organs (C.H.R. de Vos, unpublished results), as determined by gas chromatography–mass spectrometry (GC–MS) studies. This means that after the introduction of genes encoding enzymes from the monoterpene biosynthetic pathway, all monoterpenes detected will be the result of the genetic modification. This study shows that P. hybrida W115 transformed with the C. breweri S-linalool synthase cDNA under control of a constitutive double-enhanced CaMV 35S promoter produces S-linalool. However, the plant does not store or release the S-linalool but converts it into a non-volatile form, namely S-linalyl-β-d-glucopyranoside.
Introduction of linalool synthase cDNA (lis) into P. hybrida W115
Wild-type petunia W115 plants were transformed with a cDNA encoding linalool synthase under control of the CaMV d35S promoter. All regenerated PCR-positive plants were transferred to greenhouse conditions for further examination. Transgenic plants exhibited normal development compared with non-transformed and empty vector control plants that went through the same regeneration process.
Molecular and genetic analysis of transgenic plants
The number of inserts of lis was determined by Southern blotting. No signal was detected in the control plants. Of the lis-transformed plants, lines 2, 3, 17, 19, 22 and 24 showed a single gene insert (data not shown).
In a seed-plating experiment, after self-pollination of the primary transformed plants and the control, all seeds of the non-transformed control were sensitive to kanamycin. The seeds of transformed plant lines 17 and 24 exhibited the expected mendelian 1 : 3 segregation for a single gene insertion, assuming that the introduced kanamycin resistance gene, included on the T-DNA of the binary vector used for transformation, is co-localized with the lis gene in the genome.
Total RNA was extracted from young leaves of two non-transformed plants, 10 plants transformed with an empty binary vector, and 21 plants transformed with lis. Figure 1 shows that linalool synthase expression in a subset of five transformed plants varied strongly between the various independent transformants. The presence of the lis transcript was also detected by Northern blotting of the progeny of plant lines 17 and 24, showing that the gene was stably integrated (data not shown).
In order to determine whether the linalool synthase gene was properly transcribed and processed to a mature protein, crude enzyme extracts were prepared from the corolla and leaves of control and transgenic petunia plants. Linalool synthase activity was detected in corolla (1 day after anthesis) and young leaves (leaves 1–4 from the top of non-flowering plants) of transformed petunia plants, at levels of 2.3 and 4.9 nmol g−1 FW h−1, respectively, while no activity was found in extracts from control plants (Figure 2). Despite the presence of the inhibitor sodium orthovanadate, considerable phosphohydrolase activity still remained, as is clear from the presence of geraniol in all enzyme assays, resulting in a non-linear assay. In a crude extract from older leaves (mature leaves on branches of a flowering plant), no linalool synthase activity could be detected in either control or lis transformants. It is not clear whether this is due to a lower linalool synthase activity, or is the result of an even higher phosphohydrolase activity (data not shown).
Expression of the lis gene and volatile production
Gas chromatography–mass spectrometry (GC–MS) analysis of the headspace of the flowers and leaves on intact transgenic petunia plants using solid phase micro-extraction (SPME) resulted in detection of a trace level of linalool in the flowers. No linalool was detected in the leaf tissue (data not shown). However, linalool could be detected in the transgenic leaf tissue by homogenizing frozen tissue with a saturated calcium chloride (CaCl2) solution in water and subsequent GC–MS analysis of the headspace above the mixture using SPME. A typical chromatogram of young leaves of transgenic plant line 17 and a non-transformed control plant C, using the CaCl2 extraction method, is shown in Figure 3. In the control plant only a trace of monoterpenes could be detected. In the transgenic plant 17 linalool is detected (peak 1), together with α-terpineol (peak 2), at similar levels.
From a non-transformed control plant and plant 17, roots, stem, leaf, flower bud, open flower, corolla, pollen, styles, sepals, ovary and nectaries were isolated for both RNA expression and SPME/GC–MS analysis, as described under Experimental procedures. In the control plant, no expression of lis and only trace amounts of monoterpene products could be detected. Figure 4 shows that in the transgenic plant line 17 lis was expressed in all tissues analysed, although the level of expression varied strongly between the tissues. The monoterpenes linalool and α-terpineol were detected in stem, leaf, flower bud, open flower, corolla, sepals and ovary, but not in roots, pollen, styles and nectaries.
In a similar experiment, expression levels in leaves of different transformed plant lines were compared with levels of detected monoterpenes. The monoterpene levels detected in the leaves of the transgenic lines were high, between 1 and 5 µg g−1 FW (8–40 µm), but showed large variation between samples from the different plants, and did not seem to correlate with the expression levels (data not shown).
Conjugation of linalool
Storage of linalool as a conjugate could explain the absence of linalool emission from the transgenic plants. To investigate the presence of conjugates, leaf tissue was incubated with several hydrolysing enzymes. Free linalool, and no α-terpineol, was detected after incubation with an almond β-glucosidase enzyme and also with Rohapect 7104, but at a level about fivefold lower. The Rohapect enzyme mixture mainly contains β-glucosidase activity, hence the results indicate the presence of linalool as a glucoside in transgenic petunia. In control tissue, release of linalool could not be detected with any enzyme treatment. Addition of 5 m CaCl2 solution to the β-glucosidase assay on the transgenic tissue resulted in additional release of linalool, but also α-terpineol. Apparently the β-glucosidase enzyme did not release all the linalool from the transgenic tissue after overnight incubation (data not shown).
Identification of the glucoside in petunia leaf tissue
Linalyl-β-d-glucopyranoside was synthesized in order to verify the identity of the putative glycoside present in the transgenic petunia tissue. Subsequent HPLC–MS/MS analysis on control and transgenic petunia tissue, as shown in Figure 5, revealed that the m/z 375 ion trace of the compound detected in the transgenic petunia tissue (Figure 5b) had the same retention time as one of the two diastereomers of the synthesized reference (R,S)-linalyl-β-d-glucoside; these diastereomers are slightly resolved in Figure 5(a). Figure 5 shows that also the product ion spectrum of the synthesized reference compound (Figure 5d) is the same as the spectrum of the peak detected in the transgenic petunia tissue (Figure 5e). The control petunia tissue ion trace m/z 375 (Figure 5c) showed only a minor peak at the retention time of the linalyl-β-d-glucoside, indicating that there might also be a trace of linalyl-β-d-glucoside in the control plants. However, this putative peak could not be identified, as it was below the limit of detection of 0.5 µg g−1. In the transgenic petunia leaf tissue the concentration of the linalyl-β-d-glucoside was estimated to be around 5–10 µg g−1 (12–27 µm).
Chiral-phase multidimensional gas chromatography–mass spectrometry (MDGC–MS) analysis, after enzymatic hydrolysis of the highly concentrated glycoside fraction of leaf tissue (Figure 6), revealed that the transgenic petunia leaf contains highly enriched S-linalyl-β-d-glucoside, shown by the vast abundance of S-linalool detected in the m/z 93 spectrum. The hydrolysed glycoside fraction of the control plant appeared to contain low levels of slightly more R-linalool as opposed to S-linalool.
Finally, we also checked incubation of the synthesized linalyl-β-d-glucopyranoside with a 5 m CaCl2 solution. This resulted in the release of linalool and α-terpineol, showing that the CaCl2 solution is indeed responsible for the release of α-terpineol from the linalyl-glycoside present in transgenic tissues.
Genetic modification of P. hybrida W115 with the S-linalool synthase cDNA from C. breweri (Dudareva et al., 1996) resulted in the expression of a functional enzyme. The linalool produced by this enzyme in the transgenic plants was verified to be the S-enantiomer and was predominantly glucosylated, as established by the molecular and biochemical analysis. Instead of emitting the produced volatile S-linalool, the plants are storing it as S-linalyl-β-d-glucoside. Differences between organs in the production of linalool or its glycoside (Figure 4) suggest that biosynthesis depends more on the availability of the substrate GDP than on expression of the lis gene. However, the expression of lis under control of the CaMV-35S promoter can also be highly variable within a tissue of the transgenic plant, as has been shown previously for petunia plants with this promoter under control of a luciferase reporter system (van Leeuwen et al., 2000). The lack of correlation between expression of lis and levels of monoterpenes detected with SPME/GC–MS in leaf tissue between different plant lines might be explained in part by differences in environmental conditions, but is most probably due to the continuous production of S-linalool and the highly efficient conversion into the non-volatile S-linalyl-β-d-glucoside. The level at which the glucoside is accumulated in the leaf is around 5–10 µg g−1 (12–27 µm), which is more than 1000-fold higher than that detected when a fungal sesquiterpene cyclase was expressed in transgenic tobacco (Hohn and Ohlrogge, 1991). Apart from biosynthesis of GDP and linalool synthase, the tissue content of linalool depends on emission, storage in tissues, conjugation reactions and possible degradation or even transport.
The linalool synthase gene has been described to contain a putative plastid-targeting signal (Cseke et al., 1998) that was included in the construct transformed into petunia. Although correct targeting would result in the localization of the active linalool synthase enzyme in the plastids, which are the most likely source for monoterpene precursors in the form of GDP, mainly synthesized via 2-C-methyl-d-erythritol-4-phosphate in the MEP pathway (Eisenreich et al., 1997), this has not been determined in the transgenic petunia. The presence of trace amounts of linalyl-glycoside in the control petunia plant indicates that GDP is available and that there is a very low level of linalool synthase activity in the control plants (Figures 5 and 6).
The emission rate of linalool from the transgenic plants is expected to be extremely low, as only traces of linalool were detectable when the headspace of the transformed plants was analysed (data not shown). Volatilization occurred only from the flowers and not from leaves.
Storage of linalool may occur in a hydrophobic cell compartment, as in Mentha, where the terpenes are secreted to the extracellular space between the cell membrane of the secretory cells and the cuticle (Gershenzon et al., 1989). Intracellular storage may occur in lipophilic vesicles, or the component may be chemically modified to become hydrophilic, e.g. by glycosylation, in order to prevent cell membrane damage by the highly reactive monoterpenes (Stahl Biskup et al., 1993). In petunia tissue the linalool might also cause cell membrane damage if it was present in a free form. As no tissue-specific promoter for expression was used, the enzyme can be formed in all plant organs and will give a product in all cells where GDP is present. By the action of an endogenous glycosyltransferase that is able to efficiently conjugate the S-linalool produced by the transgenic plants to S-linalyl-β-d-glucoside, cellular damage can be prevented. Such a highly active glycosyltransferase was also reported in transgenic kiwi fruit expressing stilbene synthase, leading to the accumulation of picied, a resveratrol-glucoside in stead of resveratrol (Kobayashi et al., 2000). It is not clear whether the enzymatic activity in petunia responsible for the glycosylation of linalool represents a constitutive activity or an activity that is induced in transgenic plants by the presence of phytotoxic linalool.
Introduction of the linalool synthase in petunia may result in a redirection of the flux of the isoprenoid precursors dimethylallyldiphosphate and isopentenyldiphosphate towards GDP, thus competing with the formation of FDP and GGDP which are required for the production of sesquiterpenes, sterols, diterpenes, carotenoids, gibberellins and higher terpenes such as ubiquitin. A depletion in some of these isoprenoids could have a dramatic effect on the plant phenotype, as was the case when a phytoene synthase gene was overexpressed in tomato and depletion of the gibberellin pathway occurred, resulting in a dwarf phenotype (Fray et al., 1995). In our plants, however, no obvious phenotypic changes were visible, suggesting that the production of linalool did not result in drastic changes in the flux to other vital pathways.
In the transgenic petunia tissue samples that were measured using a saturated CaCl2 solution, α-terpineol was always detected along with the expected linalool. Incubation of the synthesized linalyl-glucoside standard with a CaCl2 solution also resulted in the formation of linalool and α-terpineol in a similar ratio as with the transgenic plant extracts. So the formation of α-terpineol is merely a result of hydrolysis of the linalyl-β-d-glucopyranoside in the CaCl2 solution, and is not caused by any other factor. Figure 7 explains how linalool and α-terpineol could be formed during incubation in CaCl2 solution via a putative carbocation intermediate. Measurements using saturated CaCl2 solution during volatile detection appeared to be an efficient way to release glycosidally bound volatiles.
We have shown convincingly that petunia transformed with S-linalool synthase produces S-linalool, and that most of it is directly conjugated to linalyl-β-d-glucopyranoside. There are indications that this efficient glycosylation of the monoterpenol is not common to all plant species, and this question is currently under investigation. The combination of endogenous glycosyltransferase activity and the introduction of the lis gene leads to exciting new opportunities, not only to produce terpenoids, but also for accumulation and non-toxic storage of the components produced in suitable plants.
Plant material and transformation
Petunia hybrida variety W115 was used for all experiments. Plants were grown in a greenhouse at 18°C under 18/6 h light/dark conditions, propagated using cuttings, and selfed for progeny analysis by controlled pollination.
A Clarkia breweri (Gray) Greene (Onagraceae) pBluescript cDNA clone, lis, encoding linalool synthase, was kindly provided by Dr E. Pichersky and was used for all experiments. The linalool synthase BamHI, SalI cDNA insert was ligated (Sambrook et al., 1989) with T4-DNA ligase (Life Technologies B.V., Breda, the Netherlands) between the BamHI site of an enhanced CaMV-35S promoter and the SalI site before a Nos terminator sequence of a pFlap10 vector (kindly provided by Dr A. Bovy, Plant Research International). The ligation product was transformed to Escherichia coli DH5a competent cells, and transformed colonies were grown for 16 h at 37°C and 250 r.p.m. on a rotary shaker. The expression cassette was removed from the resulting vector by using PacI and AscI restriction enzymes (NEB, Hitchen, Herts., UK) and ligated into the binary vector pBin+ (Engelen et al., 1995) containing a kanamycin resistance selection marker (nptII), after digestion with PacI and AscI. Colonies were checked after transformation by back-transformation to E. coli DH5a competent cells. Leaf cuttings of petunia W115 were transformed with Agrobacterium tumefaciens strain LBA4404 using a standard plant-transformation protocol (Horsch et al., 1985). Regeneration of the transformants was performed as described previously (van Tunen et al., 1990).
As control, leaf cuttings were also transformed with LBA4404 containing the original pBin+ vector. Some non-transformed leaf cuttings were also taken through the regeneration process. Rooting plants, arising from the lis Agrobacterium transformation, were tested with PCR for the presence of the gene. Positive plants, pBin+ transformed and non-transgenic plants were transferred to the greenhouse.
Northern and genetic analysis
Total RNA was isolated from petunia leaf and several other tissues as described previously (Verwoerd et al., 1989). For RNA gel blot analysis, 10 µg total RNA was denatured for 1 h at 50°C in 15 mm sodium phosphate buffer pH 6.5, 1.7 m glyoxal (Sigma-Aldrich, Chemie B.V., Zwijndrecht, the Netherlands) and 50% dimethylsulfoxide (DMSO) prior to electrophoresis. RNA was separated on 1.3% agarose gels in 15 mm sodium phosphate buffer pH 6.5. The gels were blotted with 25 mm sodium phosphate buffer onto Hybond-N+ nylon membranes (Amersham Pharmacia Biotech Benelux, Roosendaal, the Netherlands). Labelling of cDNA was done using the RadPrime DNA labelling system (Life Technologies). RNA gel blots were hybridized with an [α-32P]ATP-labelled BamHI–SalI fragment of 2.7 kb of the lis cDNA insert. Equal loading of RNA in the gel slots was verified by hybridization with an [α-32P]ATP-labelled 1.2 kb fragment of a 25S ribosomal cDNA from potato (kindly provided by Dr D. Florack, Plant Research International). Hybridization was performed for 16–20 h at 65°C in 1 m NaCl, 1% (w/v) SDS and 10% dextran sulphate. The membranes were washed twice for 30 min at 65°C in 0.1 × SSC, 0.1% (w/v) SDS and exposed to an autoradiographic film (Fuji, Japan) at −80°C.
Quantification of mRNA abundance of the lis gene on the Northerns was done using bio-imaging plates that were developed in a FUJIX BAS2000 bio-imaging analyser (Fuji) and analysed using tina software (Raytest, Straubenhardt, Germany). The results were standardized using the 25S ribosomal transcript levels.
Plants were selfed and backcrossed with the wild-type W115 plant. Seeds were stored dry for 1 month before sowing. 100 seeds of non-transgenic control plant, and two transgenic plants with one insertion of the lis cDNA, were plated on solid MS medium with 10 g l−1 sucrose including 100 mg l−1 kanamycin, to select for transgenics.
Isolation of linalool synthase activity and product identification
During enzyme isolation and preparation of the assays, all operations were carried out on ice or at 4°C. Frozen tissues were ground in a pre-chilled mortar and pestle: leaves (1.5 g) in 7 ml buffer; flowers (2.4 g) in 9 ml buffer containing 50 mm Mopso (pH 6.8), 20% (v/v) glycerol, 50 mm sodium ascorbate, 50 mm NaHSO3, 10 mm MgCl2 and 5 mm DTT slurried with polyvinylpolypyrrolidone (PVPP; 0.5 g g−1 tissue) and a spatula tip of purified sea sand. Polystyrene resin (0.5 g g−1 tissue, Amberlite XAD-4, Serva Electrophoresis GmbH, Heidelberg, Germany) was added, and the slurry was stirred carefully for 2 min, then filtered through cheesecloth. The filtrate was centrifuged at 20 000 g for 20 min (pellet discarded), and the supernatant centrifuged at 100 000 g for 90 min. Then 3 ml of the 100 000 g supernatant was desalted using a desalting column to a buffer containing 15 mm Mopso pH 7.2, 10% (v/v) glycerol, 5 mm sodium orthovanadate, 1 mm sodium ascorbate, 1 mm MnCl2 and 2 mm DTT. To 1 ml of these enzyme preparations, 10 µm[3H]GDP at 0.1 Ci mmol−1 was added ([3H]GDP from American Radio labeled Chemicals Inc., St Louis, MO, USA; unlabelled GDP from Sigma after a buffer change as described for FDP; de Kraker et al., 1998). After addition of a 1 ml redistilled pentane overlay, the assays were incubated for 60 min at 30°C. After the assay, the tubes were vortexed and the pentane layer removed and passed over a short column of silica overlaid with anhydrous MgSO4. The aqueous phase was extracted again with 1 ml redistilled diethyl ether, which was also passed over the silica column, and the column was washed with 1.5 ml of diethyl ether. The total volume of the pentane/diethyl ether extract was determined and 100 µl of the extract was removed for liquid-scintillation counting in 4.5 ml Ultima Gold cocktail (Packard Bioscience, Groningen, The Netherlands). Before radio-GLC analysis, the unlabelled reference compounds linalool and geraniol were added to the extract, which was then slowly concentrated under a stream of N2.
Radio-GLC was performed on a Carlo-Erba 4160 Series gas chromatograph (Carlo-Erba, Milano, Italy) equipped with a RAGA-90 radioactivity detector (Raytest, Straubenhardt, Germany) essentially as described previously (Bouwmeester et al., 1999). The GC was equipped with an EconoCap EC-WAX column (30 m × 0.32 mm i.d., 0.25 µm d.f.; Alltech Associates, Breda, The Netherlands) and the oven temperature was programmed to 70°C for 5 min, followed by a ramp of 5°min−1 to 210°C and a final time of 5 min.
Samples were also analysed by GC–MS using an HP 5890 series II gas chromatograph and an HP 5972A mass selective detector essentially as described previously (Bouwmeester et al., 1999). The GC was equipped with an HP-5MS column (30 m × 0.25 mm i.d., 0.25 µm d.f.) and programmed at an initial temperature of 45°C for 1 min, with a ramp of 5°min−1 to 120°C, a ramp of 20°min−1 to 270°C, and a final time of 5 min.
Analysis of monoterpenes produced by petunia plants
To measure the release of volatiles by the petunia plants, the headspace of leaves and flowers was measured both on the plant in the greenhouse and detached from the plant. A glass cylinder was used to capture the plant part to be analysed, and aluminium foil was used as seal. A 100 µm polydimethylsiloxane (PDMS)-coated solid phase micro-extraction (SPME) fibre (Supelco, Belfonte, PA, USA) was used in order to capture all volatiles released from the intact plant tissue. Measurements were done for 0.5–24 h, and repeated several times.
The tissues to be analysed for volatile contents and gene expression were collected in the greenhouse and frozen in liquid nitrogen. After grinding, half of the sample was stored for Northern and one for GC–MS analysis. For the latter, in routine assays 200 mg frozen material was homogenized in a mortar containing 1.5 ml 5 m calciumchloride (CaCl2) solution, to inhibit enzymatic conversions, with a small amount of purified sea sand. For some tissues smaller samples were used: pollen, 30 mg; styles, 20 mg; ovaries, 25 mg and 12 single nectaries. These samples were mixed with 0.75 ml 5 m CaCl2 solution. The material was ground rapidly and thoroughly with a pestle, and 0.75 ml of the sample was introduced into a 1.8 ml GC vial containing a small magnetic stirrer. The vial was then closed with an aluminium cap with a PTFE/butylrubber septum, and incubated for 20 min in a water bath at 50°C under continuous stirring. The headspace was sampled for 30 min with a 100 µm PDMS SPME fibre.
For quantification of linalool and α-terpineol in transgenic petunia leaves, a range of known amounts of linalool or α-terpineol were added to non-transgenic petunia tissue. The headspace was analysed as described above, and the correlation between the amounts added and peak areas (m/z 93) was determined. The value m/z 93 is typical for many monoterpenes, including linalool and α-terpineol (Adams, 1989).
To check if linalool was present in the petunia in a bound form, different glycosidases and phosphatases were added to the petunia tissue. From a batch of frozen and ground transgenic leaf material, 60 mg tissue was taken to 0.5 ml buffer with an enzyme-specific pH. The following enzymes were added: 52 units alkaline phosphatase to phosphate buffer (10 mm) pH 8; eight units acid phosphatase to citrate buffer pH 4.5; nine units α-glucosidase to phosphate buffer pH 7; 140 units β-glucosidase; one unit α-mannosidase; 0.3 units β-mannosidase to citrate buffer pH 4.5 (all from Sigma); Rohapect D5S (10% w/v) and Rohapect 7104 (10% w/v) to citrate buffer pH 4.5 (Röhm, Darmstadt, Germany). The enzyme mixtures were incubated overnight at 25°C, and the headspace was sampled by SPME for 30 min and analysed as described above. After the first measurement, 0.5 ml 5 m CaCl2 solution was added to the same samples to check for possible additional linalool in the tissue, and again the headspace was sampled. The β-glucosidase enzyme was also used in a next experiment, where 100 mg of petunia leaves in triplo was used from control and transgenic plants to incubate overnight at 37°C (optimum temperature for the enzyme) at pH 5 in 1 ml of a 10 mm phosphate/citrate buffer. As a control, the same tissues were also incubated overnight without the enzyme added to the assay. The next day all samples were stored at −20°C to stop enzymatic reactions. After the first measurement of the 12 samples, 1 ml of 5 m CaCl2 solution pH 4.5 was added to the samples and they were measured again after 30 min incubation at 50°C. In the m/z 93 ion-count profile, the peak areas of linalool and α-terpineol were compared.
GC–MS analysis was performed using a Fisons (Manchester, UK) 8060 gas chromatograph coupled to an MD 800 mass spectrophotometer (Interscience, Breda, The Netherlands). An HP-5 column (50 m × 0.32 mm, film thickness 1.05 µm) was used with He (37 kPa) as carrier gas. The GC oven temperature was programmed as follows: 2 min 80°C, ramp to 250°C at 8°min−1 and 5 min at 250°C. Mass spectra in the electron impact mode were generated at 70 eV. Injection was performed by thermal desorption of the SPME fibre in the injector at 250°C during 1 min using the splitless injection mode, with the split valve being opened after 60 sec (Verhoeven et al., 1997). The compounds were identified by comparison of GC retention times and mass spectra with those of authentic reference compounds.
Identification of glucoside and analysis of petunia leaf samples
R,S-linalyl β-d-glucopyranoside was synthesized from R,S-linalool and 2,3,4,6-tetra-O-acetyl-α-d-glucopyranosyl bromide, according to a modified Koenigs–Knorr synthesis (Paulsen et al., 1985).
Petunia leaves (3–7 g) were homogenized in 50 ml of 80% methanol and centrifuged (2000 g for 5 min). The residue was washed with 50 ml 80% methanol and the supernatants were combined. Methanol was removed in vacuum and the remaining aqueous solution was extracted with 2 × 20 ml diethyl ether. The aqueous extract was subjected to XAD-2 (20 cm, 1 cm inner diameter) solid-phase extraction. The column was successively washed with 50 ml water and 50 ml diethyl ether. Glycosides were eluted with 80 ml methanol and the eluate was concentrated in vacuum. The residue was dissolved in 1 ml 50% methanol in water and analysed by high-performance liquid chromatography electrospray-ionization tandem mass spectrometry (HPLC–ESI–MS–MS).
Analysis of methanol extracts was performed on a triple-stage quadrupole TSQ 7000 LC–MS–MS system with an electrospray ionization (ESI) interface (Finnigan MAT, Bremen, Germany). The temperature of the heated capillary was 240°C. The ESI capillary voltage was set to 3.5 kV, resulting in a 3.4 µA current. Nitrogen served as both the sheath (70 psi) and auxiliary gas (10 l min−1). Data acquisition and evaluation were carried out on a Personal DECstation 5000/33 (Digital Equipment, Unterföhring, Germany) and ICIS 8.1 software (Finnigan MAT). HPLC separations were carried out on a Eurospher 100 C-18 column (100 × 2 mm, 5 µm, Knauer, Berlin, Germany) using a linear gradient with a flow rate of 200 µl min−1. Solvent A was 5 mm ammonium acetate in water; solvent B was 5 mm ammonium acetate in methanol. The gradient program was as follows: 0–30 min, 5–100% B. Mass spectra were acquired in the negative mode. Product ion spectra were available by collision-induced dissociation (CID) (1.5 mTorr Argon, −20 eV).
The synthesized linalyl-β-d-glucoside had a concentration of 1 mg ml−1. From the peak heights found in the HPLC–MS spectra, the concentration of linalyl glycoside in the transgenic petunia leaf was estimated to be in the range 5–10 µg g−1 (12–27 µm).
Multidimensional gas chromatography mass spectrometry (MDGC–MS)
The enantiomeric distribution of linalool produced by transgenic petunia was analysed using MDGC–MS. Enzymatic hydrolysis was performed by dissolving an aliquot of the methanol extract, as described above, in 2 ml 0.2 m phosphate buffer pH 5.5. Subsequently, 200 µl Rohapect D5L (Röhm), a pectinolytic enzyme preparation exhibiting glycosidase activity, was added. After an incubation period of 24 h at 37°C, the liberated aglycons were extracted twice by 1 ml of diethyl ether. The combined organic layers were dried over Na2SO4 and concentrated.
MDGC–MS analyses were performed with a Fisons 8160 GC connected to a Fisons 8130 GC and a Fisons MD 800 quadrupole mass spectrometer, using Fisons MassLab software (Version 1.3). The first GC was equipped with a split injector (1 : 10 at 230°C) and a flame ionization detector (at 250°C). This GC employed a 25 m × 0.25 mm i.d. fused silica capillary column coated with a 0.25 µm film of DB-Wax 20 m (J&W Scientific Inc., Folsom, CA, USA) for the pre-separation of the target molecule. Separation of enantiomers was achieved with the second GC using a 25 m × 0.25 mm i.d. fused silica capillary column coated with a 0.15 µm film of 2,3-di-O-ethyl-6-O-tert-butyl dimethylsilyl-β-cyclodextrin/PS086. A multicolumn switching system (Fisons) connected the column in GC1 to that in GC2. The retention time of the compound of interest was determined by GC separation while the column in GC1 was connected to the FID. Separation of the enantiomers was achieved in the second GC after transfer of the compound of interest from the capillary column in GC1 to the column in GC2 via the switching device. The fused silica capillary column in GC1 was maintained at 60°C, then programmed to 240°C at 10°C min−1 with He gas flow at 3 ml min−1. The fused silica capillary column in GC2 was maintained at 60°C (15 min), then programmed to 200°C at 2°C min−1 with He gas flow at 3 ml min−1. The compound of interest was transferred from GC1 to GC2 from 9.8 to 10.3 min. The MS operating parameters were ionization voltage, 70 eV (electron impact ionization); ion source and interface temperature, 230°C and 240°C, respectively.
We wish to thank Dr Eran Pichersky for providing the Clarkia breweri S-linalool synthase cDNA, C. Ruff and B. Weckerle for the MDGC–MS analyses, John Franken for help with the transformation, and Prof. Dr Arjen van Tunen and Dr Ric de Vos for helpful discussions. The support of the Joseph-Schormüller-Gedächtnisstiftung for W.S. is gratefully acknowledged.