Metabolic engineering of monoterpene biosynthesis: two-step production of (+)-trans-isopiperitenol by tobacco

Authors


For correspondence (fax +31 317418094; e-mail harrie.verhoeven@wur.nl).

Present address: Wine Research Centre, Faculty of Agricultural Sciences and The Michael Smith Laboratories, University of British Columbia, 216-2205 East Mall, Vancouver, BC, Canada V6T 1Z4.

Present address: Biomolekulare Lebensmitteltechnologie, TU München, Lise-Meitner-Strasse 34, 85354 Freising, Germany.

Summary

Monoterpenoid biosynthesis in tobacco was modified by introducing two subsequent enzymatic activities targeted to different cell compartments. A limonene-3-hydroxylase (lim3h) cDNA was isolated from Mentha spicata L. ‘Crispa’. This cDNA was used to re-transform a transgenic Nicotiana tabacum‘Petit Havana’ SR1 (tobacco) line expressing three Citrus limon L. Burm. f. (lemon) monoterpene synthases producing (+)-limonene, γ-terpinene and (−)-β-pinene as their main products. The targeting sequences of these synthases indicate that they are probably localized in the plastids, whereas the sequence information of the P450 hydroxylase indicates targeting to the endoplasmatic reticulum. Despite the different location of the enzymes, the introduced P450 hydroxylase proved to be functional in the transgenic plants as it hydroxylated (+)-limonene, resulting in the emission of (+)-trans-isopiperitenol. Some further modifications of the (+)-trans-isopiperitenol were also detected, resulting in the additional emission of 1,3,8-p-menthatriene, 1,5,8-p-menthatriene, p-cymene and isopiperitenone.

Introduction

Monoterpenes are secondary metabolites produced in plants and are involved in many ecological interactions with insects, pathogens and other plants (Harborne, 1991). Primary monoterpenes are formed from the general precursor geranyldiphosphate by the action of monoterpene synthases (Croteau, 1987). All monoterpene synthases cloned to date carry plastid-targeting signals (Haudenschild and Croteau, 1998), and some were shown to be localized in the plastids (Bouvier et al., 2000; Turner et al., 1999).

Further modification of monoterpenes is often initiated by cytochrome P450-catalysed hydroxylation at different positions of a monoterpene skeleton by highly specific enzymes (Bouwmeester et al., 1999; Duetz et al., 2003). As a result, one parent molecule can lead to the formation of many structurally related derivatives.

Most plant P450 systems are located on the endoplasmic reticulum (ER) that is the source of the microsomal fraction of tissue homogenates (Chapple, 1998; Schuler, 1996). Indeed, limonene hydroxylase (limh) enzyme activity in Mentha glandular trichomes was found to be present in the microsomal fraction of these plant organs (Mihaliak et al., 1993).

The different intracellular localization and the high temporal correlation of activity during development of monoterpene synthases and monoterpene hydroxylases in peppermint (McConkey et al., 2000) and caraway (Bouwmeester et al., 1998) implicate that primary monoterpenes are translocated from plastids to the ER.

Aspecific conversion of primary monoterpenes by endogenous enzymes was observed in tomato and Arabidopsis. When a Clarkia breweri S-linalool synthase was introduced into tomato plants, 8-hydroxy-linalool was detected as a product of further conversion (Lewinsohn et al., 2001). Transformation of Arabidopsis plants with a plastid-targeted strawberry dual S-linalool/nerolidol synthase resulted in the formation of several hydroxylated and glycosylated linalool derivatives because of endogenous P450 hydroxylase and glycosyl transferase activities (Aharoni et al., 2003). However, metabolic engineering of monoterpene biosynthesis in mint, petunia and carnation did not result in further detectable conversion (Diemer et al., 2001; Krasnyanski et al., 1999; Lavy et al., 2002; Lücker et al., 2001; Mahmoud and Croteau, 2001).

In experiments with Nicotiana tabacum‘Petit Havana’ SR1, new monoterpenes were formed in and emitted from transgenic plants in addition to the native linalool (Lücker et al., 2004). Both leaves and flowers of these engineered tobacco plants were able to produce and emit (−)-β-pinene, (+)-limonene and γ-terpinene when three Citrus limon monoterpene synthases were combined into a single plant line TERLIMPIN. No further conversion products of the primary monoterpenes were detected in these plants, except for a small amount of p-cymene, perhaps as a result of chemical conversion because of low intracellular pH (Lücker et al., 2004).

(−)-Limonene-3-hydroxylase (lim3h) and (−)-limonene-6-hydroxylase (lim6h) from mint species (Lupien et al., 1999) are suitable candidate monoterpene hydroxylases for the hydroxylation of limonene produced in the transgenic tobacco plants. Although both (−)-limhs were shown to be highly substrate- and regiospecific, they were able to convert (+)-limonene, albeit with somewhat lower efficiency (Karp et al., 1990; Wüst et al., 2001). Therefore, we assumed that these P450s could be used to test the conversion of (+)-limonene into a hydroxylated product in transgenic plants.

In this paper, we describe the isolation of a lim3h cDNA from Mentha spicata L. ‘Crispa’ and subsequent introduction into transgenic tobacco plants containing three constitutively expressed monoterpene synthases. The GC profile of the plants revealed the presence of (+)-trans-isopiperitenol and some minor products, indicating the successful introduction of a functional P450.

Results

cDNA isolation of the M. spicata lim3h

A (−)-lim3h homologous cDNA, limh, was isolated from M. spicata L. ‘Crispa’ based on published sequence information (Lupien et al., 1999). The P450 had a higher percentage of identity, based on amino acid sequence, to the two published lim3hs from M. × piperita (88 and 92% identity) than to the lim6h from M. spicata (71% identity, Figure 1; Lupien et al., 1999). Typical P450 motifs, conserved for most cytochrome P450 enzymes are present in limh (Chapple, 1998). The ipsort program indicated that the putative N-terminal ER-targeting signal peptide was 30 amino acids long (Figure 1). Functional heterologous expression of limh in Saccharomyces cerevisiae resulted in the identification of a lim3h function (data not shown).

Figure 1.

Sequence alignment of mint limhs.

The M.×piperita (−)-lim3hs Mxplim3hPM17 (AF124816) and Mxplim3hPM2 (AF124817), and the M. spicata (−)-lim6hs Mslim6h (AF124815) were aligned to the isolated cytochrome P450 from M. spicata‘Crispa’, limh. Functional domains common to all P450 enzymes, the proline-rich region, the molecular oxygen-binding pocket, the haeme-binding region (Chapple, 1998) and a limh-specific domain (Schalk and Croteau, 2000) are indicated in the alignment and explained in Discussion. The psort- and ipsort-predicted non-cleavable N-terminal signal for targeting to the ER is underlined. The alignment was created with the clustalx program and standard settings. Shading indicates conserved identity for the aligned amino acids: black background shading indicates 100% conservation, dark grey shading indicates 75% conservation and light grey shading indicates 50% conservation.

N. tabacum transformation and expression analysis of the resulting transgenics

  • Nicotiana tabacum plants were transformed with limh under the regulation of the constitutive CaMV d35S promoter. Plants rooting on hygromycin-containing medium were transferred to the greenhouse. Ten independent transgenic plant lines were obtained by transformation of plant line TERLIMPIN, possessing three monoterpene synthases (Lücker et al., 2004) and were coded TERLIMPINLIMH. From the control plant line SR1 transformed with the limh construct, 11 independent plant lines were obtained and coded LIMH. The starting material of the plant lines was also transformed with the empty binary vector pCambia1300+ as a control.

Introduction of limh and its expression was analysed by RNA gel blotting of leaf total RNA (Figure 2a). Expression of limh was detected in all the transformed plants, although expression was very low in TERLIMPINLIMH-5 and TERLIMPINLIMH-6 (Figure 2a). Expression of limh was highest in plant line TERLIMPINLIMH-8. The RNA gel blot was subsequently hybridized with the C. limon (+)-limonene synthase ((+)-lim) cDNA as a probe (Figure 2b) to verify the (+)-lim cDNA expression. The TERLIMPIN and TERLIMPIN with empty vector control plants showed no expression of the P450 (Figure 2a). Also, wild-type SR1 plants transformed with the P450 construct (LIMH) were analysed on the same membrane and hybridized at the same time. This showed that the expression level of the introduced P450 in the LIMH plants was on average higher than the expression level detected in the TERLIMPINLIMH plants (data not shown). The hybridization of the membrane with the 25S ribosomal DNA (Figure 2c) showed that the apparent lower expression of limh and lim in the TERLIMPINLIMH-4 plant line was mainly because of a lower amount of total RNA loaded.

Figure 2.

RNA gel blot analysis of leaves of transgenic tobacco plant lines.

(a) limh cDNA from M. spicata‘Crispa’.

(b) (+)-lim cDNA from C. limon.

(c) 25S, 25S ribosomal cDNA from potato as control for loading differences as described previously by Lücker et al. (2001).

The RNA gel blot was subsequently hybridized with three different cDNA probes. Transformed plant lines tested: TERLIMPIN is expressing three different lemon monoterpene synthases (γTER (+)-LIM, and (−)-βPIN). TERLIMPINCam is a TERLIMPIN plant that was re-transformed with an empty pCambia1300+ vector. TERLIMPINLIMH are TERLIMPIN plants, re-transformed with the pCambia expression vector containing the isolated lim3h cDNA, limh.

GC–MS analysis using solid-phase micro-extraction (SPME)

The TERLIMPINLIMH-8 plant line was identified as having the highest limh expression. To assess whether this correlated with higher monoterpene alcohol levels, detached leaves of this plant were analysed by headspace SPME and compared with the controls. Wild-type SR1 and the empty binary vector control showed only traces of α-pinene, β-pinene and linalool, as described previously by Lücker et al. (2004). The LIMH plants showed the same profile as wild-type plants, indicating that merely the introduction of the P450 by itself did not result in a new phenotype (data not shown). A typical SPME GC–MS chromatogram obtained from the headspace of detached leaves of the TERLIMPIN and TERLIMPINLIMH plants is shown in Figure 3. The TERLIMPIN leaves mainly emitted β-pinene, limonene and γ-terpinene and some traces of other monoterpenes at minor levels (Figure 3a). The chromatogram obtained from the headspace of TERLIMPINLIMH-8 (Figure 3b) shows an abundant peak at a retention time of 15.95 min (peak #12), not present in the TERLIMPIN plant (Figure 3a). The mass spectrum of this peak was identical to the mass spectrum of trans-isopiperitenol described in the literature by Lupien et al. (1999).

Figure 3.

Figure  Headspace SPME measurements on detached leaves of tobacco transgenic plant lines TERLIMPIN (a) and TERLIMPINLIMH-8 (b).

The chromatogram shows the combined m/z 68 + 69 + 82 + 84 + 91 + 93 + 119. 1, α-thujene; 2, α-pinene; 3, sabinene; 4, unidentified monoterpene; 5, β-pinene; 6, tentatively identified as 1,5,8-p-menthatriene (KI = 1006); 7, p-cymene; 8, limonene; 9, γ-terpinene; 10, linalool; 11, tentatively identified as 1,3,8-p-menthatriene (KI = 1115); 12, trans-isopiperitenol; 13, cis-isopiperitenol; and 14, isopiperitenone.

In addition to the trans-isopiperitenol peak, four other monoterpenoids were detected as minor peaks in the chromatogram of TERLIMPINLIMH-8 that could not be detected in the TERLIMPIN control. Peak #13 was identified as cis-isopiperitenol. The headspace of flowers of the TERLIMPINLIMH-8 plant was sampled by SPME and also contained trans-isopiperitenol (data not shown).

Based on Kovats indices (KI) and mass spectra, peak #6 (Figure 3b) was tentatively identified as 1,5,8-p-menthatriene (KI = 1006) and peak #11 (Figure 3b) as 1,3,8-p-menthatriene (KI = 1115) (Adams, 1995). Peak #14 was identified as isopiperitenone by comparison with an authentic standard.

The levels of the main products of the introduced β-pinene synthase (βPIN) and γ-terpinene synthase (γTER) were lower in the TERLIMPINLIMH-8 plant than in the TERLIMPIN plant. In contrast, the level of p-cymene was elevated in the TERLIMPINLIMH-8 plant, and in all other P450-transformed lines that produced trans-isopiperitenol.

Assessment of the configuration of biosynthesized isopiperitenol

All four isomers of isopiperitenol synthesized from (±)-limonene (Guillon et al., 2000) were separated and identified by analysis on a multidimensional tandem GC–MS comprising a DB-wax column coupled to an enantioselective column (Figure 4a). These products were identified as (+)- or (−)-isomers based on comparison to partially separated (+)-isopiperitenol, chemically synthesized from (+)-limonene (peaks #3 and #4, Figure 4b). The synthesized mixed isomer isopiperitenol sample was first run on a DB-wax column in the first GC. The first eluted peak containing the cis-isomer, as previously determined for this column (Werkhoff et al., 1998), was separated on an enantioselective column in the second GC (Figure 4c); peak #1 was identified as (−)-cis-isopiperitenol and peak #4 as (+)-cis-isopiperitenol. Another verification of the correct identification results from the complete separation of the (−)-isopiperitenol isomers (Figure 4a), as the 40 : 60 ratio of (−)-cis-isopiperitenol (#1) and (−)-trans-isopiperitenol (#2), which was found, is consistent with the values reported in the literature by Guillon et al. (2000). The isopiperitenol formed in the transgenic plants was also analysed by multidimensional tandem GC–MS technique. In flowers of plant TERLIMPINLIMH-8, the GC–MS analysis showed 99% excess of (+)-trans-isopiperitenol over its enantiomer and (+)- and (−)-cis-diastereoisomers (Figure 4d). The level of emitted (+)-trans-isopiperitenol from flowers, sampled by Tenax trapping, was estimated to be around 400 ng g−1 h−1. The isopiperitenol emitted by the leaves of plant line TERLIMPINLIMH-8 was also predominantly (+)-trans-isopiperitenol (data not shown). The (+)-trans-isopiperitenol in flowers of plant line TERLIMPINLIMH-4 was at approximately 95% excess over its enantiomer and (+)- and (−)-cis-diastereoisomers (data not shown).

Figure 4.

Assessment of the configuration of the major new monoterpene emitted in the P450 transformed TERLIMPINLIMH-8 plant as (+)-trans-isopiperitenol by MDGC–MS.

(a) All four chemically synthesized isomers of isopiperitenol separated by MDGC–MS.

(b) The isopiperitenol chemically synthesized from (+)-limonene separated by the enantioselective column in (+)-trans-isopiperitenol (peak #3) and (+)-cis-isopiperitenol (peak #4).

(c) The first peak of the racemic isopiperitenol on the DB-wax column representing the cis-isomers (Werkhoff et al., 1998) was transferred to the enantioselective column and separated in (−)-cis-isopiperitenol (peak #1) and (+)-cis-isopiperitenol (peak #4).

(d) The isopiperitenol emitted by flowers of the TERLIMPINLIMH-8 transgenic plant line could be identified as (+)-trans-isopiperitenol (peak #3).

(e) Mass spectrum of the synthesized standard (+)-trans-isopiperitenol.

(f) Mass spectrum of the major new product emitted by flowers of the TERLIMPINLIMH-8 plant.

Flower headspace analysis of transgenic plant lines using Tenax trapping

Volatile compound emission of detached open flowers of the same developmental stage was compared for different transgenic lines (Figure 5). The flowers of the TERLIMPIN plant line did not emit isopiperitenol (Figure 5a). Several transgenic lines, transformed with the P450 cDNA, showed emission of the trans-isopiperitenol (peak #9) as a major product in the GC profile of trapped volatiles (Figure 5b–d).

Figure 5.

Volatile emission profiles (TIC) of detached flowers of transgenic plant lines sampled for 24 h using Tenax trapping.

(a) Plant line TERLIMPIN.

(b) Plant line TERLIMPINLIMH-8.

(c) Plant line TERLIMPINLIMH-2.

(d) Plant line TERLIMPINLIMH-4.

Each time four open flowers were sampled from stages 11 and 12 (Goldberg, 1988; Koltunow et al., 1990). Monoterpene and sesquiterpene peaks are numbered: 1, sabinene; 2, β-pinene; 3, tentatively identified as 1,5,8-p-menthatriene (KI = 1006); 4, p-cymene; 5, limonene; 6, γ-terpinene; 7, linalool; 8, tentatively identified as 1,3,8-p-menthatriene (KI = 1115); 9, trans-isopiperitenol; 10, isopiperitenone; and 11, β-caryophyllene. is, internal standard (212 ng (+)-3-carene). Unmarked peaks are no mono- or sesquiterpenes; the peak at 9.98 min in (a) is derived from decanal, a natural constituent of tobacco eluting at the same time as isopiperitenol.

As observed in leaves, flowers of some TERLIMPINLIMH plants showed a clear decrease in the emission of the main products of some of the introduced monoterpene synthases compared to the TERLIMPIN plant line. In the case of TERLIMPINLIMH-3, limonene, γ-terpinene and, interestingly, isopiperitenol could not be detected (data not shown). Although plant line TERLIMPINLIMH-7 also showed very low emission of limonene and γ-terpinene, a trace level of isopiperitenol was detectable (data not shown). In the transgenic plant line TERLIMPINLIMH-8 that showed an abundant (+)-trans-isopiperitenol peak, limonene and β-pinene levels are still intact, whereas the level of γ-terpinene was 66% lower than that in the TERLIMPIN plant (Figure 5b).

The emission of p-cymene from flowers of the TERLIMPINLIMH plants emitting isopiperitenol was much higher than that from the TERLIMPIN flowers in the plant lines TERLIMPINLIMH-2 (Figure 5c) and TERLIMPINLIMH-4 (Figure 5d). The p-cymene level was frequently as high as the limonene level (peak #5) in headspace samplings of earlier stages of flower development in the plants TERLIMPINLIMH-2 and TERLIMPINLIMH-4. On the other hand, the level of p-cymene was never higher than around one-third of the emission of either one of the main products of the heterologous monoterpene synthases in the TERLIMPIN plant (Lücker et al., 2004). This increase in p-cymene levels was higher in flowers than in leaves of the same transgenic plant line (data not shown).

Similar to what we observed in leaves of the TERLIMPINLIMH-8 plant (Figure 3b), there were two additional small peaks in the chromatogram of the flowers of the TERLIMPINLIMH-4 plant that were tentatively identified as 1,5,8-p-menthatriene (peak #3, Figure 5d) and 1,3,8-p-menthatriene (peak #8, Figure 5d). These two peaks were also observed after concentration of the flower headspace sample of TERLIMPINLIMH-8 and TERLIMPINLIMH-2 lines, but never detectable in the TERLIMPIN plant line (data not shown).

Isopiperitenone was detected in the headspace of leaves as well as in the headspace of flowers from TERLIMPINLIMH-4 (Figure 5d) and in TERLIMPINLIMH-8 and TERLIMPINLIMH-2 after concentration of the headspace samples. This ketone was never detected in the TERLIMPIN plant line (data not shown).

Discussion

Metabolic engineering of a hydroxylation step in addition to a new monoterpene biosynthetic route in tobacco plants was shown to be functional and resulted in the efficient conversion of (+)-limonene into (+)-trans-isopiperitenol. The (+)-trans-isopiperitenol was emitted as a major component of the volatile spectrum of the plants possessing four integrated cDNAs involved in monoterpene biosynthesis. Isopiperitenol is an uncommon compound in the plant kingdom and mostly present as an intermediate, which is converted into other derivatives with the same oxygenation pattern (Guillon et al., 2000). However, in our tobacco plants, further modification of the monoterpenol occurred only to a small extent (to isopiperitenone).

cDNA isolation and characterization

Volatiles isolated from M. spicata‘Crispa’ were found to contain carvone, carveol and isopiperitenol, indicating that both regioselective (6- and 3-) hydroxylase activities must be present in this plant (data not shown). Therefore, this plant was used as a source to isolate a limh. The gene-specific reverse primer, based on the M. spicata (−)-lim6h sequence (AF124815), was designed to amplify a 6-hydroxylase or the highly homologous 3-hydroxylase. The last six bases of this primer (5′-TTAAGGACTTTTATAGAGTGTGG-3′) were identical to the cDNA sequences of the two described M.×piperita (−)-lim3h isoforms (AF124816 and AF124817). The isolated limh cDNA is probably an isozyme of the previously isolated (−)-lim3hs (Figure 1; Lupien et al., 1999).

The limh contained several signature sequences common to limhs: the PPXP motif (occurring as PPIP in limh) pertinent to hydroxylase regioselectivity (Schalk and Croteau, 2000); the conserved (A/G)GX(D/E)T(T/S) (occurring as AGXETS in limh), proposed to form the threonine-containing binding pocket for the oxygen molecule required for catalysis (Von Wachenfeldt and Johnson, 1995); the FXXGXXXCXG sequence (as FGAGRRICPG in limh), in mint P450s proposed to be the haeme-binding domain (Von Wachenfeldt and Johnson, 1995); and the (P/I)PGPX(G/P)XP motif (PPXPPKLP in limh), the proline-rich region that conceivably acts as a hinge to confer optimal orientation of the enzyme in the ER membrane (Yamazaki et al., 1993). Microsomal P450s, like the limhs (Mihaliak et al., 1993), contain an N-terminal hydrophobic signal sequence for transfer into the ER, which is not cleaved but anchored in the membrane, and the catalytic domain of the protein is exposed on the cytosolic side of the membrane (Schuler, 1996). Sequence analysis of the PCR-isolated limh P450 cDNA with the psort and ipsort prediction programs indicated with a certainty level of 0.82 that limh is likely targeted to the ER just as most plant P450s (Chapple, 1998; Schuler, 1996), and that it has a non-cleaved N-terminal signal peptide.

Function of the cytochrome P450 enzyme

Limonene is a minor product in peppermint (M. × piperita), and (−)-limonene predominates (+)-limonene by approximately 80%. At saturation, the (−)-isomer is preferred twofold over the (+)-isomer as substrate for (−)-lim3h of this species (Karp et al., 1990). (+)-Limonene is very abundant in our transgenic tobacco plants transformed with (+)-LIM, with only a low level of (−)-limonene produced as a side product of the βPIN and γTER activities (Lücker et al., 2004). The major product in the TERLIMPINLIMH-8 plant line expressing limh is (+)-trans-isopiperitenol with a high excess of 99% over its enantiomer and (+)- and (−)-cis-diastereoisomers. The lack of (−)-trans-isopiperitenol can be explained by the low activity of the monoterpene synthases that produce β-pinene and γ-terpinene as a main product and (−)-limonene as a side product in the TERLIMPINLIMH-8 plant line. Congruently, the TERLIMPINLIMH-4 plant line that showed a higher level of β-pinene and γ-terpinene emission in the flowers than the TERLIMPINLIMH-8 plant line did indeed form a detectable, although minute, amount of (−)-trans-isopiperitenol (data not shown). This suggests that (−)-limonene, a side product of the βPIN and γTER in this transgenic tobacco line, is also converted by the P450 hydroxylase enzyme.

Targeting of the introduced enzymes

Monoterpene synthases in general have N-terminal plastid-targeting peptides, and leucoplast localization was confirmed for a peppermint (−)-limonene synthase ((−)-lim) in secretory cells (Haudenschild and Croteau, 1998; Turner et al., 1999), but these synthases are probably also localized in chloroplasts of leaf parenchyma cells (Bouvier et al., 2000). A plastid-targeting signal was found in all monoterpene synthases isolated from lemon, including (+)-lim (Lücker et al., 2002). The predicted plastid-targeting sequence and part of the N-terminal region of the native enzyme expressed from this lim cDNA were demonstrated as requirements for chloroplast localization in tobacco protoplasts (A. Aharoni, unpublished results).

Based on the N-terminal targeting signal sequence, the lim3h isolated from M. spicata like other plant P450s is probably associated with the ER. P450 enzymes typically function as a complex with an NADPH cytochrome P450 reductase (Schuler, 1996), of which two distinct isoforms from poplar were shown to be localized in the ER (Ro et al., 2002). These reductases appear to be non-specific electron donors to cytochrome P450s (Donaldson and Luster, 1991). The N. tabacum plants in this study were not co-transformed with a reductase to complement the function of the introduced P450, thus implicating that there was sufficient endogenous, native cytochrome P450 reductase activity present in the transgenic tobacco. Although the heterologously expressed lim and the limh are probably localized to different cell compartments in the tobacco plants, the detection of the product (+)-trans-isopiperitenol emitted from the plants shows that the introduced second step in the monoterpene biosynthetic pathway was successful. If the localization in different compartments of these enzymes is true, a previously hypothesized transport mechanism (active or passive) for monoterpenes to the ER is likely to be functioning in tobacco and might be common in plants (Bosabalidis, 1996; Bouwmeester et al., 1998).

Side products and conversions

Recently, (−)-lim6h has been shown to convert the unnatural substrate (+)-limonene into mainly (+)-cis-carveol along with smaller levels of (−)-trans-carveol, (+)-trans-carveol, (+)-trans-isopiperitenol and 1,2-cis-epoxide (Wüst and Croteau, 2002). Likewise in the leaves of the TERLIMPINLIMH-8 plant, the cis-isomer of isopiperitenol was detected, likely because of the fortuitous conversion of the unnatural, epimeric substrate (+)-limonene to (+)-cis-isopiperitenol.

In addition, the introduction of a new monoterpene alcohol in plants might trigger other secondary reactions. It has been shown previously that hydroxylated monoterpenes can be converted into their respective ketones by cultured suspension cells of N. tabacum, governed by an NAD+-dependent alcohol dehydrogenase (Suga and Hirata, 1990), although the equilibrium in this reversible reaction tends towards the alcohol. The trace of isopiperitenone detected in the headspace of the transgenic plants (Figures 3 and 5) indicates that such an endogenous alcohol dehydrogenase activity is also present in our tobacco plants.

While the small peak of p-cymene observed in the TERLIMPIN plants was probably because of oxidation of the terpenes by light, air or low intracellular pH (Lücker et al., 2004), the substantial increase of p-cymene observed after introduction of the cytochrome P450 lim3h cannot be explained in that way. However, it is possible that the p-cymene is derived from (+)-trans-isopiperitenol (Figure 6). After the formation of (+)-trans-isopiperitenol, water could be eliminated, for example by an endogenous dehydratase, traces of acid or heat (Kobold et al., 1987), causing the formation of 1,3,8-p-menthatriene, which can isomerize to the more stable p-cymene by the action of traces of acid or light by a suprafacial 1,7-hydrogen shift followed by a suprafacial 1,3-hydrogen shift (Smith and March, 2001). Another possibility could be that the heat of the GC-injection port induces these rearrangements (Kobold et al., 1987), but this is unlikely as GC–MS analysis of the samples with injection port temperature set at 150°C instead of the normal 250°C did not have any effect on p-cymene levels (data not shown). Dehydratase enzymes, which could catalyse such a conversion, have been reported, for example in the shikimate pathway in microorganisms, converting 3-dehydroquinate into 3-dehydroshikimate and water (Figure 7; Herrmann and Weaver, 1999). This enzyme is present in plants as a bifunctional 3-hydroquinate dehydratase-shikimate dehydrogenase (Herrmann and Weaver, 1999). A partial cDNA of the gene encoding this enzyme was also isolated from tobacco (Bonner and Jensen, 1994). In the TERLIMPINLIMH plants, two peaks were observed, which were tentatively identified as 1,5,8-p-menthatriene (KI = 1006) and 1,3,8-p-menthatriene (KI = 1115), which could indeed be intermediates in the supposed conversions. The observation that the increase in p-cymene is often more obvious in flowers than in the leaves of the transgenic TERLIMPINLIMH plants might indicate that the putative dehydratase enzyme is more active in the flowers than in leaves.

Figure 6.

Hypothetical scheme of enzymatic or non-enzymatic conversions occurring in the transgenic tobacco TERLIMPINLIMH plants.

Conversion of isopiperitenol to p-cymene is supposed to be triggered by an endogenous dehydratase and occurs via 1,3,8-p-menthatriene and possibly 1,5,8-p-menthatriene (not shown). Also heat, acid or light could result in these reactions.

Figure 7.

The reaction catalysed by 3-hydroquinate dehydratase in microorganisms (Herrmann and Weaver, 1999).

This dehydratase reaction, which is occurring in the shikimate pathway of microorganisms, shows a similar reaction as the suggested catalysis of isopiperitenol to 1,3,8-p-menthatriene and 1,5,8-p-menthatriene.

Gene silencing

Figures 3 and 5 show that TERLIMPINLIMH-8 emitted lower levels of β-pinene and γ-terpinene than the TERLIMPIN plant. Also, some of the other P450-transformed lines showed a similar decrease in some of the products of the monoterpene synthases, like TERLIMPINLIMH-3, which showed only β-pinene emission (data not shown). This might indicate that gene silencing occurs. All the TERLIMPIN plants were already expressing three genes regulated by the same CaMV 35S promoter. This promoter is known to cause silencing effects in transgenic plants when multiple copies are present (Dr J. P. Nap, personal communication). This might explain the higher average level of expression of limh observed in the LIMH plant lines, compared to the average expression of limh observed in the TERLIMPINLIMH plants (data not shown).

The present results show that the metabolic engineering of monoterpene biosynthesis involving enzymatic conversions in two separate cellular compartments can be successful. This creates opportunities to build a monoterpenoid biosynthetic pathway from scratch in any plant, in order to generate products of commercial value or ecological impact. This approach can, for instance, be used to modify the flower scent profile of commercially interesting cut flowers such as roses, or be used to modify the interaction between plants and their pests.

Experimental procedures

Plant material

As a source for the limh cDNA, young leaves from the top of non-flowering M. spicata L. ‘Crispa’ plants were obtained from the botanical garden of Wageningen University (Wageningen, the Netherlands).

Plant material for transformation was wild-type N. tabacum‘Petit Havana’ SR1, and a transformed plant line (TERLIMPIN) containing three different active lemon monoterpene synthases (γ-terpinene synthase (ClγTERS, or TER), limonene synthase (Cl(+)LIMS1, or LIM) and β-pinene synthase (Cl(–)βPINS or (PIN); Lücker et al., 2002, 2004). New plants were regenerated in vitro on MS20 medium supplemented with 1 mg l−1 BAP from cuttings of young leaves taken from plants in the greenhouse. The leaf was sterilized in 1.5% (w/v) sodium hypochlorite solution for 10 min and washed in sterilized water in three subsequent steps. The regenerated plantlets served as starting material for the transformation and as controls for the experiments. Plants in the greenhouse were grown under a 16-h photoperiod and standard greenhouse conditions.

RNA isolation and cloning of a P450 limh using RACE-PCR

Total RNA was isolated from M. spicata L. ‘Crispa’ leaves, by grinding 100 mg of tissue in liquid nitrogen and using a hot phenol method previously described by Lücker et al. (2004). The isolated RNA was dissolved at 60°C and treated for 10 min at this temperature with 1 vol. of cetyl trimethylammonium bromide (CTAB) buffer (0.2 m Tris, pH 7.5, 50 mm EDTA, 2 m NaCl and 2% CTAB) in order to remove undesired sugars. The sample was subsequently extracted with chloroform:isoamylalcohol (24 : 1 (v/v)) and precipitated using ethanol before dissolving in ddH2O.

In the initial attempt to amplify the M. spicata (−)-lim6h, first-strand cDNA synthesis was carried out using primer 209 (815SPIPstI) 5′-CGTACTCTGCAGTTAAGGACTTTTATAGAGTGTGG-3′. This primer anneals to the 3′ end of the open-reading frame (ORF) of the previously described cDNA (GenBank (http://www.ncbi.nlm.nih.gov/) Accession number: AF124815 (Lupien et al., 1999)), with a PstI restriction site overhang and another six randomly chosen bases added. SuperScript II RT enzyme was used for the first-strand cDNA synthesis according to the manufacturer's recommendations (Life Technologies, Breda, the Netherlands). By using primers specific for the (−)-lim6h based on the published sequence under various different PCR conditions, no full-length cDNA could be obtained. A cDNA fragment identical to the published cDNA was obtained using specific primers F2spic (5′-CATAAGGCAGGAGGAGATCG-3′), R2spic (5′-TCTCTTACCTCCGCCTGCAC-3′) for the middle of the described cDNA and the PCR programme: 94°C, 30 sec; 53°C, 30 sec; 72°C, 1.5 min; 40 cycles.

Secondly, a RACE-PCR approach was followed requiring only one specific primer on the 3′ end of the cDNA. RACE-PCR was performed according to the manufacturer's recommendations (Clontech Laboratories Inc., Palo Alto, CA, USA) and again 1 µg of total RNA with specific primer 209 was used to prepare the first strand. 5′ RACE-PCR was performed at 55 and 60°C annealing temperatures, 15 cycles using primer 219: 5′-TTAAGGACTTTTATAGAGTGTGG-3′, exactly annealing to the 3′ end of the ORF of the (−)-lim6h, together with the universal primer mix (UPM) from the RACE-PCR kit. The PCR resulted in several fragments, of which one, limh, was 1.6 kb. This fragment was ligated into the pGEM-T EASY vector system according to the manufacturer's recommendations (Promega Benelux b.v., Leiden, the Netherlands), and subsequent sequencing resulted in the identification of a full-length cytochrome P450.

Using targeting sequence prediction programmes available on the Internet, possible localization and targeting was analysed using psort (http://psort.nibb.ac.jp/form.html) and ipsort (http://hypothesiscreator.net/iPSORT/#predict).

Alignments were made using the clustalx 1.81 program (Thompson et al., 1997) using standard settings and genedoc software. Sequence homology levels were analysed using megalign of the DNASTAR software package (DNASTAR Inc., Madison, WI, USA).

Construction of the plant expression vector

For construction of a binary vector, the limh ORF had to be inserted in a pFLAP10 vector (kindly provided by Dr A. Bovy, Plant Research International) containing a CaMV-d35S promoter and a nopaline synthase (Nos)-terminator sequence. PCR primers were used for this step containing restriction site overhangs XbaI and NcoI. The forward primer used was 5′-TTATATTCTAGATGGAGCTCCAGATTTCGTCG-3′ and the reverse primer was 5′-TTATAACCATGGTTAAGGACTTTTATAGAGTGTGG-3′. PCR was performed using Pfu-turbo (Stratagene Europe, Amsterdam Zuidoost, the Netherlands) and the programme, 94°C for 30 sec, 55°C for 30 sec and 72°C for 2 min 30 sec, was repeated for 40 cycles. After purification of the amplified fragment from agarose gel, it was digested using XbaI and NcoI restriction enzymes and ligated to a pFLAP10 XbaI- and NcoI-digested vector fragment. The resulting vector pFLAPlimh was re-sequenced to check for correct PCR amplification, digested using restriction enzymes PacI and AscI (NEB, Hitchin, Hertfordshire, UK) and ligated to a PacI-, AscI-digested pCambia1300+ vector (CAMBIA, Canberra ACT, Australia), containing a hygromycin plant resistance gene (hptII) and added PacI and AscI restriction sites (kindly provided by Hans Janssen, Plant Research International), resulting in the binary vector for expression in plants. The hygromycin resistance gene was required as the transgenic plant material used for the subsequent genetic modification was already resistant to kanamycin, as a result of the selection marker gene nptII that was used for transformation with the monoterpene synthases (Lücker et al., 2004). Via a cold shock method, the limh containing plant expression vector was transformed to Agrobacterium tumefaciens strain LBA4404, and colonies were checked after transformation (Lücker et al., 2004). Leaf discs of in vitro grown plant material of tobacco were transformed using a protocol previously described by Horsch et al. (1985). Selection for transformants was carried out using 15 mg l−1 hygromycin.

RNA isolation for RNA gel blotting

RNA was isolated from the rooting in vitro plantlets. RNA gel blotting was performed with a 1.5-kb XbaI–NcoI-digested full-length ORF of the P450 cDNA (limh), a 1.8-kb NheI–SalI lim cDNA fragment and a 25S ribosomal cDNA from potato. Hybridization, washing and autoradiography using phosphoimager plates were carried out as described previously by Lücker et al. (2001).

Volatile analysis

Leaf material (0.5 g) from M. spicata was used for volatile analysis by GC–MS after hydro-distillation (Lücker et al., 2002). Transgenic plants were screened by SPME of fully stretched mature leaves (10–20 cm long) from transgenic and control tobacco plants, collected from the greenhouse. The same method was used for the analysis of the volatile emission of tobacco flowers at various developmental stages, but in this case, sampling was performed on the intact plant. All samples were analysed by GC–MS (Lücker et al., 2004). For trapping onto Tenax TA (20/35 mesh, Alltech, Breda, the Netherlands), detached flowers or leaves were used. Simultaneous analysis of emission by different plant lines (four flowers or five young leaves in each jar) was performed by headspace trapping for 24 h. Extraction of Tenax and GC–MS analysis were performed as previously described by Lücker et al. (2004).

Synthesis of isopiperitenol from isopiperitenone

Isopiperitenone standard (10.3 mg; kindly provided by Haarmann & Reimer, Holzminden, Germany) was added to 1.5 mg of LiAlH4 in 0.5 ml of re-distilled ether and vigorously mixed for 17.5 h. Three small portions (in total approximately 0.1 g) of Na2SO4·10H2O (Merck KgAA, Darmstadt, Germany) were carefully added. The mixture was stirred for another 30 min. Distilled water (1.5 ml) was added, and after mixing vigorously, the mixture was centrifuged at 1200 g for 5 min. The organic phase was subsequently loaded onto a short aluminium oxide (Al2O3) column, and the column was eluted with 1 ml of ether.

Synthesis of isopiperitenol from (+)- and (−)-limonene and multidimensional GC–MS (MDGC–MS) measurements

Chemical synthesis of the mixture of two sets of enantiomers of isopiperitenol from limonene was performed exactly as described previously and yielded a 40 : 60 (cis:trans) mixture of isopiperitenol (Guillon et al., 2000).

MDGC–MS analyses with a DB-wax column in GC1 and an enantioselective column in GC2 were carried out as described previously by Lücker et al. (2001), but with a different temperature programme. The DBwax column in GC1 was programmed at an initial temperature of 50°C, which was increased at a rate of 4°C min−1 up to 240°C and was kept at this temperature for 10 min. GC2 with the enantioselective column (2,3-di-O-ethyl-6-O-tert-butyl dimethylsilyl-β-cyclodextrin/PSO86, 25 m × 0.25 mm inner diameter, 0.15 µm film thickness) was programmed at an initial temperature of 60°C for 20 min and was then increased at a rate of 1°C min−1 up to 200°C.

Acknowledgements

The authors would like to thank the chemical company Haarmann & Reimer for kindly providing the isopiperitenone standard, and Prof. Dr Carlo Bicchi for his help in the identification of the (+)-trans-isopiperitenol.

Accession numbers: Mxplim3hPM17: AF124816, M.×piperita (−)-limonene-3-hydroxylase PM17; Mxplim3hPM2: AF124817, M.×piperita (−)-limonene-3-hydroxylase PM2; Mslim6h: AF124815, M. spicata (−)-limonene-6-hydroxylase; limh: to be submitted to GenBank, M. spicata‘Crispa’ limonene-3-hydroxylase.

GenBank Accession number for the sequence of the gene described in this paper is AY622319.

Ancillary