Spike lavender (Lavandula latifolia) essential oil is widely used in the perfume, cosmetic, flavouring and pharmaceutical industries. Thus, modifications of yield and composition of this essential oil by genetic engineering should have important scientific and commercial applications. We generated transgenic spike lavender plants expressing the Arabidopsis thaliana HMG1 cDNA, encoding the catalytic domain of 3-hydroxy-3-methylglutaryl CoA reductase (HMGR1S), a key enzyme of the mevalonic acid (MVA) pathway. Transgenic T0 plants accumulated significantly more essential oil constituents as compared to controls (up to 2.1- and 1.8-fold in leaves and flowers, respectively). Enhanced expression of HMGR1S also increased the amount of the end-product sterols, β-sitosterol and stigmasterol (average differences of 1.8- and 1.9-fold, respectively), but did not affect the accumulation of carotenoids or chlorophylls. We also analysed T1 plants derived from self-pollinated seeds of T0 lines that flowered after growing for 2 years in the greenhouse. The increased levels of essential oil and sterols observed in the transgenic T0 plants were maintained in the progeny that inherited the HMG1 transgene. Our results demonstrate that genetic manipulation of the MVA pathway increases essential oil yield in spike lavender, suggesting a contribution for this cytosolic pathway to monoterpene and sesquiterpene biosynthesis in leaves and flowers of the species.
The aromatic shrub spike lavender (Lavandula latifolia Medicus), a member of the Lamiaceae family, is cultivated worldwide for its essential oil, which is used in perfumery, cosmetics, food processing and medicine (Miralles, 1998). Due to their volatility, flavour/aroma and toxicity, essential oils also play important roles in plant defense, plant-to-plant communication and pollinator attraction (Pichersky and Gershenzon, 2002). The composition of spike lavender oil is determined mainly by the genetic make-up of each cultivar, being mono- and sesquiterpenes (C10 and C15 isoprenoids, respectively) the major fractions of this oil (Harborne and Williams, 2002). As in other Lamiaceae, these isoprenoids seem to be synthesized and accumulated in the peltate glandular trichomes found on the aerial parts of the species (Hallahan, 2000).
Plant isoprenoids are synthesized through condensations of the universal five-carbon precursors: isopentenyl diphosphate (IPP) and its allylic isomer dimethylallyl diphosphate (DMAPP), which are derived from two alternative pathways (Figure 1). In the cytosol, IPP is synthesized from three molecules of acetyl-CoA by the classical acid mevalonic (MVA) pathway, while in plastids, it is derived from pyruvate and glyceraldehyde-3-phosphate (G3P) via the methyl-D-erythritol 4-phosphate (MEP) pathway (Rodriguez-Concepción and Boronat, 2002; Dudareva et al., 2006). The main MVA-derived isoprenoid end-products are certain sesquiterpenes, sterols and the side chain of mitochondrial ubiquinones, whereas monoterpenes, certain sesquiterpenes and photosynthesis-related isoprenoids (carotenoids and the side chain of chlorophylls and plastoquinones), are derived from the MEP pathway (Lichtenthaler, 1999). Although this subcellular compartmentalization of the MVA and MEP pathways allows them to operate independently, metabolic exchange between these two pathways is documented, and the extent of this crosstalk depends on the species, tissues or/and the physiological conditions (Bick and Lange, 2003; Hemmerlin et al., 2003; Laule et al., 2003; Schuhr et al., 2003; Dudareva et al., 2005; Hampel et al., 2005, 2006; Cusidóet al., 2007). Nevertheless, very little is known about the regulation of the exchange mechanism and the crosstalk between the MEP and MVA pathways (Eisenreich et al., 2004; Rodriguez-Concepción et al., 2004; Bouvier et al., 2005).
HMG-CoA (3-hydroxy-3-methylglutaryl-coenzyme A) reductase (HMGR, EC 184.108.40.206) catalyses the first committed step of the MVA pathway by converting HMG-CoA to MVA using NADPH as cofactor (Campos and Boronat, 1995). MVA is converted into IPP by the sequential action of MVA kinase, phosphomevalonate kinase and pyrophosphomevalonate decarboxylase (McGarvey and Croteau, 1995). Farnesyl diphosphate (C15, FPP) is formed from IPP by FPP synthase. Subsequently, squalene synthase catalyses the head-to-head condensation of two FPP molecules to generate squalene, the first committed step in sterol biosynthesis; end-product sterols (β-sitosterol and stigmasterol, Figure 1) are formed from squalene by a series of cyclization, methylation and desaturation reactions (Schaller, 2004).
All plants examined contain several isoforms of HMGR. In Arabidopsis thaliana, some HMGR is located in the membrane system of the endoplasmic reticulum (Leivar et al., 2005), but a predominant portion is found within spherical vesicular structures that are located in the cytoplasm and within the central vacuole. These vesicular structures appear to be derived from subdomains of the endoplasmic reticulum (Surpin and Raikhel, 2004; Jørgensen et al., 2005; Leivar et al., 2005). The different types of HMGR-enriched vesicles might carry different sets of enzymatic activities, providing the physical basis for metabolic channels that each catalyse the synthesis of specific isoprenoid classes (Leivar et al., 2005).
Recently, we demonstrated that up-regulation of 1-deoxy-D-xylulose-5-P (DXP) synthase (DXS), which catalyses the first step of the MEP pathway, in transgenic spike lavender plants, resulted in elevated levels of the monoterpenes responsible for the characteristic flavour of this aromatic species (Muñoz-Bertomeu et al., 2006). These results also support the involvement of the MEP pathway in the biosynthesis of essential oils in aromatic plants, as previously suggested (Mahmoud and Croteau, 2002). However, our results did not exclude the possibility that the MVA pathway could provide precursors for monoterpene and sesquiterpene biosynthesis in spike lavender, as has been reported in strawberry fruits and foliage (Hampel et al., 2006). In order to study the possible contribution of the MVA pathway in the biosynthesis of essential oils, we targeted the first step in this pathway by over-expressing the catalytic domain of the HMGR1S enzyme from Arabidopsis in spike lavender. We demonstrate that up-regulation of this enzyme increases mono- and sesquiterpenes as well as end-product phytosterols, such as β-sitosterol and stigmasterol, but does not affect the accumulation of carotenoids and chlorophylls in spike lavender. These results indicate that the MVA pathway can contribute to essential oil production in spike lavender. To the best of our knowledge, this is the first report on this subject in an aromatic plant that biosynthesizes and accumulates mono- and sesquiterpenes in highly specialized, non-photosynthetic glandular structures.
Generation and evaluation of transgenic spike lavender
The plasmid pBICD1, encoding a truncated A. thaliana HMGR1S lacking the N-terminal membrane-binding domain of the mature enzyme, was used to obtain transgenic spike lavender plants through an Agrobacterium-based leaf culture protocol set up in our laboratory (Nebauer et al., 2000). Resistance to kanamycin was used to select putative transformants.
The integration of the HMG1 and nptII transgenes in the kanamycin-resistant plantlets was first confirmed by PCR. The PCR products showed clear bands for the HMG1 (1430 bp) and nptII (526 bp) genes, but no bands were observed in the control plants (data not shown). All nptII+/HMG1+ plants were cloned, acclimatized to ex vitro conditions and transferred to the greenhouse for further analyses. Six independent primary transformants (T0), designated as HMGR1 to HMGR6, were obtained. No observable morphological differences were evident among the HMGR transgenics and the controls.
The number of transgene inserts in T0 plant was determined by Southern blotting using both nptII and HMG1 probes (Figure 2a). Two out of the six independent transgenic lines (HMGR3 and HMGR6) presented single copy insertions while the rest of the lines presented from two to ten copies of the HMG1. Different hybridization patterns were obtained for the nptII transgene in HMGR1 and HMGR5, corroborating the insertion of the incomplete T-DNA, while in the other lines (HMGR2, HMGR3, HMGR4 and HMGR6) similar hybridization patterns corroborate the insertion of the complete T-DNA (Figure 2a).
To correlate phenotypes of the HMGR lines with gene expression, mRNA levels of HMG1 transgene were determined for transgenic T0 plants. Total RNA was isolated from leaves and flowers, and expression of the transgenic HMG1 mRNA was determined by Northern blot analyses. Figure 2b,c shows that there are large differences in expression levels of the HMG1 cDNA among the independent transgenic lines. In leaves (Figure 2b), all lines showed detectable levels of HMG1 mRNA, with lines HMGR1 and HMGR3 showing the highest levels. Lines HMGR2 and HMGR4 showed moderate levels of transgene expression, and lines HMGR5 and HMGR6 showed minimal levels of transgene expression. Transgene expression in flowers (Figure 2c) was also line-dependent, with lines HMGR1, HMGR2, HMGR4 and HMGR5 showing the highest levels. Under the conditions used in our experiments, the HMG1 probe gave no signal when hybridized with RNA from untransformed plants (Figure 2b,c). Thus, the hybridization signal obtained with the transgenic spike lavender plants is directly related to the expression of the transgene.
To characterize integration of the transforming cDNA into the spike lavender genome and its inheritance, we analysed T1 plants derived from the selfing T0 transgenic lines that flowered after growing for 2 years in the greenhouse and produced enough seeds to perform the analysis (HMGR2 and HMGR4). Inheritance of the HMG1 and nptII transgenes was determined by PCR. Line HMGR2, which according to the Southern blot analysis had two HMG1 copies (Figure 2a), showed a 3 : 1 segregation (χ2 = 0.53, P = 0.05), suggesting that in this plant both copies were integrated at the same locus and inherited as a single linkage block. Line HMGR4, which had four copies according to the Southern analysis (Figure 2a), failed to segregate any HMG1-negative individual out of the 18 T1 plants tested. Segregation of the nptII gene was concordant with the HMG1 gene.
Southern blot analysis of the HMG1 transgene in self-pollinated progeny showed that all of the HMGR2-T1-positive plants inherited the two copies of the HMG1 transgene (Figure 3a), corroborating the 3 : 1 segregation obtained from PCR analyses. Five out of the six progenies of the four-copy HMGR4 parent line showed the four copies in the corresponding Southern blot, whereas one progeny had one copy (Figure 3d), suggesting that the four copies were not integrated at the same locus. Southern and Northern analyses showed complete concordance between the presence of the transgene and the corresponding transcripts, whereas material from either HMGR2 progenies that did not inherit the transgene (HMGR2-5 and HMGR2-6) or control plants contained no detectable signal (Figure 3b,d).
None of the HMGR-T1 plants displayed visual phenotypes that differed from the untransformed controls.
Over-expression of the HMG1 gene increases essential oil yield in spike lavender plants
To assess whether over-expression of the Arabidopsis HMG1 gene in spike lavender was correlated with essential oil yield, air-dried leaves and flowers from transgenic T0 plants were hydrodistilled, analysed by GC/MS, and compared with controls. Twenty-eight compounds were identified, accounting for 86.1% to 94.6% of the whole essential oils. In both transgenic and control plants, monoterpenes were the main essential oil constituents (96.7–99.0% and 91.4–98.6% in leaves and flowers, respectively), followed by sesquiterpenes (0.6–3.2% and 1.2–8.3% in leaves and flowers, respectively). The high percentage of monoterpenes was largely due to cineole and camphor in oils from leaves (36.6–45.3% and 32.6–44.4%, respectively) and cineole, camphor and linalool in oils from flowers (20.9–37.2%, 20.3–34.8% and 15.1–40.1%, respectively).
Essential oil yield (mg/g dried tissue) in leaves and flowers of spike lavender was markedly increased in most of the transgenic T0 plants that expressed the HMG1 cDNA (Table 1). In leaves, five out of six transgenic lines accumulated significantly more essential oils than controls (from 1.2- to 2.1-fold in HMGR5 and HMGR2, respectively). In transgenic as well as control plants, essential oil production in flowers was higher than in leaves. Flowers from four transgenic lines produced significantly more essential oils than the controls (from 1.4- to 1.8-fold in HMGR1 and HMGR2, respectively). Both monoterpenes and sesquiterpenes contributed to the increased oil yield of transgenic plants, but the latter showed the highest increases in relation to the controls (increase averages of 1.5- vs. 3.0-fold and 1.4- vs. 1.8-fold in leaves and flowers, respectively).
Table 1. Essential oil yield, monoterpene and sesquiterpene production (milligram per gram dried weight) in leaves (L) and flowers (F) of control and transgenic T0 spike lavender plants transformed with the Arabidopsis HMG1 gene Reported values for transgenic plants represent the means ± SD of four measurements. The control value represents the average of four wild-type plants with four measurements each. Within each column, values followed by different letters are significantly different according to Tukey's test at P ≤ 0.05
13.55 ± 1.94 a
34.60 ± 3.65 a
13.43 ± 2.08 a
33.85 ± 3.68 a
0.07 ± 0.03 a
0.50 ± 0.22 a
18.38 ± 0.03 c
48.03 ± 1.55 b
18.18 ± 0.03 c
46.35 ± 1.48 b
0.13 ± 0.00 c
1.18 ± 0.06 c
28.05 ± 0.10 d
62.01 ± 1.26 d
27.61 ± 0.12 d
61.09 ± 1.16 d
0.16 ± 0.01 d
0.56 ± 0.04 ab
18.87 ± 1.06 c
57.00 ± 3.12 c
18.40 ± 1.21 c
55.98 ± 3.01 c
0.38 ± 0.00 f
0.70 ± 0.10 b
27.34 ± 0.73 d
49.25 ± 1.58 b
26.89 ± 0.80 d
48.54 ± 1.57 b
0.17 ± 0.02 d
0.42 ± 0.01 a
15.57 ± 0.49 b
35.09 ± 4.41 a
15.41 ± 0.55 b
33.89 ± 4.29 a
0.09 ± 0.01 b
0.68 ± 0.10 b
14.11 ± 1.31 ab
32.88 ± 0.73 a
13.78 ± 1.50 ab
30.57 ± 0.55 a
0.32 ± 0.01 e
1.83 ± 0.17 d
In some transgenic T0 lines, the contribution of the HMG1 transgene to the accumulation of mono- and sesquiterpenes seems to be selective. It is well known that the variability in the spike lavender oil composition is primarily genotype dependent (Harborne and Williams, 2002; Muñoz-Bertomeu et al., 2007); thus, the variability in the mono- and sesquiterpene content among the HMGR lines may also reflect the bulked seed origin of the plants used as a source of explants for transformation.
Most of the HMGR2 and HMGR4 progenies analysed maintained their elevated essential oil phenotype; average oil yield increases in HMGR2 and HMGR4 progenies that inherited the transgene were 1.6- and 1.5-fold with respect to their T1 counterparts without the transgene or control plants of the same age, respectively (Table 2). Note, however, that the contribution of sesquiterpenes to the increased oil yield of the progenies was lower than that previously observed in transgenic T0 lines (Table 1); thus, the increase averages for mono- and sesquiterpenes in HMGR2-T1 were 1.6- vs. 1.8-fold as compared to their T1 counterparts without the transgene, whereas in HMGR4-T1 were 1.5- vs. 1.2-fold when compared to control plants of the same age (Table 2).
Table 2. Essential oil yield, monoterpene and sesquiterpene production (milligram per gram dried weight) from leaves of representative transgenic T1 spike lavender plants obtained from controlled self-pollination of T0 transgenic HMGR2 and HMGR4 lines Reported values for each T1 plant represent the mean ± SD of four measurements. Within each column, values followed by different letters are significantly different according to Tukey's test at P ≤ 0.05. *T1 plants that did not inherit the HMG1 transgene
Supplementary Tables S1 and S2 summarize the percentages of the most common mono- and sesquiterpenes found in the oils from transgenic (T0 and T1) and control spike lavender plants. As previously reported for transgenic spike lavender over-expressing the DXS transgene, these percentages were within the ranges observed in other studies on chemical composition of spike lavender oils (Muñoz-Bertomeu et al., 2006, references therein). Thus, despite their increased monoterpene- and sesquiterpene-phenotype, transgenic T0 and T1 plants showed the same essential oil profile as control plants, indicating that over-expression of the HMG1 gene did not result in new chemotypes. Representative GC essential oil profiles of control and transgenic HMGR2 and HMGR4 lines are shown in Figures 4 and 5.
Over-expression of the HMG1 gene increases end-product sterols in spike lavender plants
Incorporation studies with [13C]-labelled glucose show that sterols are synthesized predominantly via the MVA pathway-derived precursors in all plants analysed (Eisenreich et al., 2004, references therein). Thus, we quantified end-product phytosterols in our transgenic plants. The levels of β-sitosterol and stigmasterol, the most common 4-desmethylsterols in plants, were analysed by GC/MS in transgenic and control spike lavender plants. Analysis of leaves from six independent transgenic T0 lines showed increase averages in β-sitosterol and stigmasterol of up to 1.8- and 1.9-fold, respectively, compared to the control mean (Table 3). The highest increases in β-sitosterol were found in lines HMGR3 and -4 (2.5- and 2.7-fold, respectively), whereas lines HMGR3 and -5 showed a higher stigmasterol content (2.2-fold with respect to the controls).
Table 3. Sterol content (microgram per gram dried weight) in leaves of transgenic T0 spike lavender plants transformed with the Arabidopsis HMG1 gene Reported values for each T0 plant represent the mean ± SD of four measurements. The control value represents the average of four wild-type plants with four measurements each. Within each column, values followed by different letters are significantly different according to Tukey's test at P ≤ 0.05
246.14 ± 41.49 a
13.14 ± 2.47 a
430.01 ± 33.31 d
22.34 ± 1.85 b
304.41 ± 22.91 b
24.13 ± 3.30 b
612.00 ± 32.12 e
29.35 ± 1.94 c
661.45 ± 41.79 e
23.08 ± 1.33 b
372.42 ± 62.06 c
28.62 ± 2.14 c
282.21 ± 20.40 ab
23.48 ± 1.68 b
Sterol analyses in spike lavender leaves from T1 progenies of HMGR2 and HMGR4 lines showed that the enhanced sterol content observed in transgenic T0 plants was stable in the subsequent generation (Table 4). Interestingly, stigmasterol content in most of the HMGR2 progenies was largely unchanged, but levels of β-sitosterol, the immediate precursor of stigmasterol, were significantly increased in most of the T1 plants that inherited the transgene (increase average of 1.7-fold with respect to their T1 counterparts without the transgene). This differential effect was more evident in transgenic HMGR4 progenies, where the increase average of β-sitosterol reached 9.9-fold when compared to the control means (Table 4).
Table 4. Sterol content (microgram per gram dried weight) from leaves of representative transgenic T1 spike lavender plants obtained from controlled self-pollination of T0 transgenic HMGR2 and HMGR4 lines Reported values for each T1 plant represent the mean ± SD of four measurements. Within each column, values followed by different letters are significantly different according to Tukey's test at P ≤ 0.05. *T1 plants that did not inherit the HMG1 transgene
Over-expression of the HMG1 gene does not increase chlorophyll and carotenoid content of the spike lavender leaves
Experimental evidence has clearly demonstrated that carotenoids and the phytol side chain of chlorophylls are some of the major products derived from the MEP pathway (Lichtenthaler, 1999). Because of this, we quantified chlorophylls and carotenoids in spike lavender leaves from transgenic T0 and control plants. Chlorophyll and carotenoid content in transgenic lines depended on the plant analysed, showing either no variation or a slight, but significant, increase or decrease as compared to the controls (Table 5). Note, however, that a clear trend towards increase or decrease in the chlorophyll and carotenoid content could not be inferred in the HMGR transgenic lines as a whole. Thus, the average amount (mg/g fresh weight) of the two photosynthetic pigments in transgenic and control plants was similar (1.72 ± 0.25 vs. 1.72 ± 0.23 and 0.40 ± 0.06 vs. 0.38 ± 0.07 for chlorophylls and carotenoids, respectively).
Table 5. Total chlorophyll and carotenoids content (milligram per gram fresh weight) in leaves of transgenic T0 spike lavender plants transformed with the Arabidopsis HMG1 gene Reported values for each T0 plant represent the mean ± SD of at least three measurements. The control value represents the average of four wild-type plants with at least three measurements each. Within each column, values followed by different letters are significantly different according to Tukey's test at P ≤ 0.05
Chlorophylls (a + b)
1.72 ± 0.23 bcd
0.38 ± 0.07 ab
1.87 ± 0.08 de
0.42 ± 0.02 bc
2.06 ± 0.09 e
0.48 ± 0.01 c
1.85 ± 0.11 cde
0.44 ± 0.02 c
1.56 ± 0.09 abc
0.36 ± 0.03 ab
1.49 ± 0.14 ab
0.35 ± 0.03 ab
1.46 ± 0.09 a
0.32 ± 0.01 a
In a previous paper, we reported that up-regulation of DXS, the first enzyme of the MEP pathway, leads to a significant increase in essential oil yield of spike lavender (from 2.0- to 4.6-fold and from 1.2- to 1.7-fold in leaves and flowers, respectively, as compared to controls), which suggests that this pathway contributes to mono- and sesquiterpene biosynthesis in this species (Muñoz-Bertomeu et al., 2006). In the present study, we found that up-regulation of HMGR1S, a key enzyme of the cytosolic MVA pathway, also increased essential oil yield in spike lavender (up to 2.1- and 1.8-fold in leaves and flowers, respectively, as compared to controls). Furthermore, progeny that inherited the HMG1 transgene had significantly enhanced essential oil yield. All these results seem to support the involvement of the MVA pathway in the biosynthesis of essential oils in this aromatic plant. Although monoterpenes always constituted the main contribution to the increased oil yield in transgenic spike lavender plants, sesquiterpenes showed the highest increases in relation to the controls (increase averages of 1.5- vs. 3.0-fold and 1.4- vs. 1.8-fold in leaves and flowers of transgenic T0 plants, respectively). This differential behaviour could be explained considering that sesquiterpenes are mainly formed in the cytoplasm (McCaskill and Croteau, 1995; Dudareva et al., 2005); thus, an increase in MVA-derived precursors could facilitate sesquiterpene biosynthesis in this compartment.
Both feeding studies with labelled compounds and functional evaluation of expressed sequence tags (McCaskill and Croteau, 1995; Lange et al., 2000), indicated that the MVA pathway is blocked at HMGR in the non-photosynthetic secretory cells of glandular trichomes of peppermint (Mentha × piperita). Nevertheless, the MVA pathway in these cells appears to be latent, since the exogenous [14C]-MVA is metabolized by the isolated secretory cells of peppermint, producing detectable levels of mevalonate phosphate, mevalonate diphosphate, IPP, DMAPP, and FPP (McCaskill and Croteau, 1995). The blockage of MVA pathway at HMGR has also been demonstrated in snapdragon (Antirrhinum majus) petals, which form volatile terpenes only in the non-photosynthetic epidermal cells (Dudareva et al., 2004).
Since the glandular trichomes of a variety of aromatic species undergo similar ontogeny to that observed in peppermint (McCaskill and Croteau, 1995), we can speculate that, under normal physiological conditions, mono- and sesquiterpenes of the spike lavender oils are also exclusively derived from the MEP pathway. However, it can be assumed that the constitutive expression of the Arabidopsis HMG1 cDNA in spike lavender glandular trichomes allowed the activation of the MVA pathway, resulting in an increased production of MVA-derived precursors that were channeled to both cytosolic sesquiterpene and plastidial monoterpene biosynthesis. In aromatic plants, monoterpene production is limited by precursor availability from the MEP pathway (Mahmoud and Croteau, 2002); thus, the uptake of MVA-derived isoprenoid precursors from the cytosol should overcome this limitation, improving monoterpene yield in the plastids of secretory cells of spike lavender glandular trichomes. Supporting this idea, up-regulation of HMGR in the Arabidopsis rim1 (resistant to inhibition with mevinolin 1) mutant increased production of MVA-derived precursors that were channeled to the biosynthesis of plastidial isoprenoids (Rodriguez-Concepción et al., 2004). In addition, Wu et al. (2006) recently engineered high level terpene production by diverting carbon flux from cytosolic or plastidic IPP through over-expression in either compartment of a farnesyl diphosphate synthase and an appropriate terpene synthase. Although this engineered metabolism was not confounded by crosstalk between the two pathways, such high level terpene accumulation was made possible, in part, because of simultaneous over-expression of a truncated HMGR (Wu et al., 2006).
Since the biosynthetic origin of IPP and DMAPP has not been investigated yet in spike lavender glandular trichomes, we cannot exclude, however, that under normal physiological conditions, the MVA pathway could constitutively provide precursors for essential oils in spike lavender. In fact, certain terpenes are biosynthetic mosaics where one or several isoprenoid-building blocks are derived from the MEP pathway, whereas others are derived from the MVA pathway (Adam and Zapp, 1998; Itoh et al., 2000; Nagata et al., 2002; Yang and Orihara, 2002; Hampel et al., 2006). This observation implies that certain biosynthetic steps proceed in different compartments, and that specific intermediates traverse the plastid boundary (Eisenreich et al., 2004).
The over-expression of HMGR1S also led to an increased amount of β-sitosterol and stigmasterol in transgenic T0 spike lavender plants (increase averages of 1.8- and 1.9-fold, respectively). This increased sterol phenotype of transgenic T0 spike lavender plants was inherited by their progenies, demonstrating that the biosynthetic flux leading to β-sitosterol and stigmasterol was controlled by HMGR. Both sterol content and essential oil yield of spike lavender were increased in most transgenic plants that expressed the HMG1 cDNA. Nevertheless, a deviation of the flux to the biosynthesis of end-product sterols was also evident, particularly in HMGR4 progenies, where the increase average of β-sitosterol was 9.9-fold with respect to the controls. This high amount of sterols could explain the lower increase in sesquiterpenes found in HMGR4 progenies. As previously reported, sesquiterpenes are mainly formed in the cytosol, where the pool of FPP is also used for the production of sterols (Lichtenthaler, 1999). These results clearly reinforce the pivotal role of HMGR in the sterol metabolism of plant cells, as has been reported previously (reviewed by Schaller, 2004; Bouvier et al., 2005). In contrast, HMGR1S up-regulation had virtually no effect on the amount of carotenoids and chlorophylls of the species, corroborating previous work on the compartmentalization of MVA and MEP pathway in photosynthetic tissues (Lichtenthaler, 1999).
Taken together, these results suggest that there is a metabolic crosstalk between MVA and MEP pathways that is tissue/organ dependent in spike lavender. A strict compartmentalization of both pathways exists in photosynthetic tissues, where the over-expression of HMG1 transgene led to an increased amount of sterols without effects in carotenoids and chlorophylls. This does not hold true, however, in non-photosynthetic secretory cells of glandular trichomes, where the up-regulation of HMGR1S could provide MVA-derived precursors for both cytosolic sesquiterpene and plastidial monoterpene biosynthesis, leading to an increase in essential oil yield. Although direct measurements of HMGR activity in transgenic lines were not undertaken, we consider that the levels of essential oils and sterols in transgenic spike lavender plants are a good indication of the effect of the genetic modification.
In transgenic spike lavender plants (T0 and T1), there was no clear correlation between the number of transgene insertion loci or HMG1 expression and either essential oil or sterol content. It has been suggested that high copy number can be associated with co-suppression and gene silencing (Hobbs et al., 1990; Meyer and Saedler, 1996), but this does not appear to be the case for the spike lavender transgenic lines used in this work. In fact, our results demonstrate that the high-yielding essential oil and sterol phenotype was inherited in the progeny of both two- and four-copy lines of the HMG1 gene. On the other hand, the activity of HMGR is regulated at both transcriptional and post-transcriptional levels (Holmberg et al., 2003), and it has been suggested that the Arabidopsis HMG1 gene could be more susceptible to regulation than other isoforms (Godoy-Hernández et al., 1998). This could explain the lack of correlation between the levels of HMG1 transcripts and metabolite accumulation in some transgenic T0 and T1 plants. Essential oil composition could be also altered by apparent insertional effects of the HMG1 transgene. This phenomenon is not uncommon and has been observed previously in transgenic peppermint over-expressing the DXR gene, which encodes the enzyme that catalyses the second step of the MEP pathway (Mahmoud and Croteau, 2001).
In conclusion, the most relevant contribution of our work is the demonstration that over-expression of the HMG1 gene from Arabidopsis leads to an increased yield of essential oil in spike lavender. Nevertheless, further investigation is needed to actually know whether the increased essential oil yield resulted from either the induction of a latent MVA pathway blocked at HMGR or an up-regulation of an existing MVA pathway. The increase in essential oil yield is a biotechnologically relevant target of this study. Moreover, our results may contribute to a better understanding of the role that HMGR plays in isoprenoids biosynthesis in aromatic plants.
Plant material and bacterial strain
Bulked seeds of Lavandula latifolia Medical (spike lavender), from Spanish natural populations (Intersemillas SA, Valencia, Spain), were germinated under sterile conditions as described in Calvo and Segura (1988). The first two pairs of leaves from 40-day-old seedlings were used as primary explants for transformation.
Transgenic T0 lines refer to plants regenerated from explants originally infected with Agrobacterium tumefaciens. T1 plants (first generation) are seed-derived plants obtained from controlled self-pollination of T0 plants. Non-transgenic, wild-type spike lavender plants were grown under the same conditions as controls. Both flowers and either developing (first and second verticils) or fully expanded (fourth to tenth verticils) leaves were sampled for molecular and phenotypical analyses.
The Agrobacterium tumefaciens strain C58 containing the plasmid pBICD1, kindly provided by Dr Boronat (University of Barcelona, Spain), was used for transformation experiments. This binary vector contains a neomycin phosphotransferase II (nptII) marker gene, driven by the nopaline synthase (NOS) promoter and terminator, and a truncated form of the Arabidopsis (A. thaliana) HMG1 cDNA (AT1G76490), encoding the amino acids 165–592 that included the catalytic domain of the enzyme, under the control of the cauliflower mosaic virus 35S promoter and NOS terminator.
Spike lavender transformation
Agrobacterium-mediated transformation and regeneration of spike lavender was carried out according to Nebauer et al. (2000). Acclimatization and maintenance of regenerated plants under greenhouse conditions were accomplished, as previously described (Muñoz-Bertomeu et al., 2006). After 2 years, all transgenic T0 lines flowered. Progenies of some T0 lines were obtained by selfing (July–August) in controlled conditions. T1 seeds were harvested in October, germinated in vitro as described in Calvo and Segura (1988) and seedlings transplanted to pots and transferred to the greenhouse as previously described. Leaves of T1 seedlings were used to evaluate segregation of HMG1 and nptII genes by PCR analysis.
Polymerase chain reaction and Southern blotting
DNA was isolated from developing leaves (50–100 mg fresh weight) of in vitro-grown transgenic and wild-type plants using the CTAB (cetyl trimethyl ammonium bromide) procedure described by Doyle and Doyle (1990). For large-scale leaf preparations (2 g fresh weight of fully expanded leaves from greenhouse-grown plants), genomic DNA was extracted as described in Muñoz-Bertomeu et al. (2006).
PCRs were performed in 50 µL reaction volumes containing 75 mm Tris-HCl (pH 9.0), 50 mm KCl, 2.5 mm MgCl2, 20 mm (NH4)2SO4, 0.1 mm dNTP, 0.25 mm of each oligonucleotide primer, 100 ng DNA, and 4 units of Taq polymerase (Biotools, Madrid, Spain). Primer sets used were: 5′-GTCGCTTGGTCGGTCATTTCG-3′ and 5′-GTCATCTCACCTTGCTCCTGCC-3′ for the nptII gene, and 5′-GTTCATTTCATTTGGAGAGGAC-3′ and 5′-AGGGCAAACGCATACGCAC-3′ for the HMG1 gene. The predicted sizes of the amplified DNA fragments were 526 bp and 1430 bp for nptII and HMG1, respectively. Amplification parameters for nptII gene were 95 °C for 5 min followed by 40 cycles at 95 °C for 1 min and 65 °C for 2 min. A 65 °C incubation for 5 min as final step was included. Amplification parameters for HMG1 gene were 94 °C for 5 min followed by 35 cycles at 94 °C for 1 min, 55 °C for 1 min and 72 °C for 1.5 min with a final extension step at 72 °C for 7 min. Amplified products were detected by ultraviolet light fluorescence after electrophoresis on 1% (w/v) agarose TBE gels with 1.27 µm ethidium bromide.
The Southern blot analysis was performed with non-radioactive digoxigenin-11-dUTP-labelled probes as described in Muñoz-Bertomeu et al. (2006).
Total RNA was extracted from 0.8 g of fresh fully expanded leaves and 0.4 g flowers of transgenic and wild-type plants by using the Tripure Isolation Reagent (Roche Applied Science, Indianapolis, IN, USA) according to the supplier's protocol. RNA electrophoresis and subsequent transference on to hybond–N nylon membranes (Amershan Biosciences UK Limited, Little Chalfont, UK) were performed as described previously (Muñoz-Bertomeu et al., 2006). DNA probes, labelled with [α-32P]dCTP, were prepared by random priming of the fragments amplified by PCR. The HMG1 fragment (576 bp) was amplified from the plasmid with primers 5′-GCTGGCTCTGCTGTTGCAGGC-3′ and 5′-GGGCAAACGCATACGCACATGGC-3′, whereas the tubulin3 (TUB3) fragment (969 bp), used to verify the equal loading of RNA in the gel slots, was amplified from A. thaliana cDNA, with primers 5′-CCTGATAACTTCGTCTTTGGTCAATCC-3′ and 5′-GAACTCCATCTCGTCCATTCC-3′. For both genes, the amplification parameters were 94 °C for 3 min, followed by 30 cycles at 94 °C for 1 min, 60 °C for 2 min and 72 °C for 2 min. A 72 °C for 7 min as a final step was included. Blots were prehybridized in buffer [0.4 m NaH2PO4 (pH 7.2), 1 mm EDTA and 7% (w/v) SDS] at 65 °C or 55 °C for 20 min, and hybridized overnight at 65 °C or 55 °C in the same buffer with 5 ng/mL of the HMG1 or TUB3 probes, respectively. The membranes were washed twice for 10 min in 4× SSC, 0.1% (w/v) SDS at 65 °C, and twice for 5 min in 0.4× SSC, 0.1% (w/v) SDS at 65 °C and exposed to an auto-radiographic film at –80 °C.
Essential oil analysis
Air-dried (for 30 days), fully expanded leaves (1.5 g) or flowers (0.5 g of cymes with three to five open flowers) from each examined plant were treated separately for essential oil extraction and quantification by gas chromatography (GC) and GC/mass spectrometry (MS) analyses as described in Muñoz-Bertomeu et al. (2006).
Dried (60 °C for 7 days), fully expanded leaves from each transgenic or control plants were ground to a powder in a mortar and pestle to ensure sample uniformity. The powder (300 mg) was transferred into a pyrex tube with 5 mL of ethyl acetate, vortexed 10 s, sonicated for 15 min, incubated for 15 min and the supernatant transferred to a new tube. This step was repeated four times. Before extraction, 5α-cholestane was added to each sample as internal standard (0.1 mg). The pooled ethyl acetate phases were reduced to dryness under nitrogen. The residue was then recovered with 4 mL of 2 N KOH in 90% ethanol, and saponified at 90 °C for 1 h. After cooling, samples were acidified with 2 N HCl (5 mL), and then extracted with 6 mL of hexane for 1 h (three times). The hexane phases were pooled and reduced to dryness under nitrogen. Finally, dried residues were dissolved in 4 mL hexane and filtered through 0.22 µm PVDF Millipore membranes (Millipore Corporation, Bedford, MA, USA). Sterols were kept in air-tight glass containers at 4 °C until further use.
Quantitative sterol analysis was accomplished by GC and GC/MS. GC analyses were performed with a Focus GC (Thermo Finnigan, Milan, Italy) equipped with a flame ionization detector and fitted with a BPX5 capillary column (5% phenyl polysilphenylene-siloxane, 30 m × 0.25 mm × 0.25 µm film; SGE Europe Ltd, Villebon, France); carrier gas He at 4 mL/min; 3 µL was injected in ‘splitless mode’ (0.8 min) with an AI 3000 Autosampler (Thermo Finnigan, Milan, Italy). The normal oven temperature was programmed initially at 150 °C followed by a ramp of 20 °C/min to 300 °C, and finally held isothermal at 300 °C for 13 min. Temperatures of injector and detector were 250 °C and 310 °C, respectively. GC/MS analyses were carried out on an Agilent 6890 N gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) with the same column; the coupled mass spectrometer was an Agilent 5973 N with a quadrupole mass selective detector (Agilent Technologies, Santa Clara, CA, USA). All mass spectra were acquired in the electron impact mode at 70 eV. The mass spectrometer scanned in the range of 30–550 m/z at a rate of 5.36 scans per second. The other analytical conditions were the same as for GC analysis. Compounds were identified by comparison of retention indices and mass spectra to those of authentic standards, or by reference mass spectra in a computer library (Wiley7n, Wiley STM Databases, Chichester, West Sussex, UK). MS/GC was performed by the Central Service for the support to experimental research (SCIE, University of Valencia).
The products were quantified (µg/g dried tissue) by comparison of detector response with that of the internal standards, assuming equal response factors. All analyses were performed at least four times.
Chlorophyll and carotenoid content
Extraction and determination of total chlorophylls and carotenoids were conducted as described by Lichtenthaler (1987). Extracts were obtained in 100% acetone from 200 to 300 mg of fresh fully developed leaves from T0 spike lavender plants and controls. Spectrophotometric quantifications were carried out in a Shimadzu UV-1203 spectrophotometer (Shimadzu Corporation, Kyoto, Japan). All analyses were performed at least three times.
Significance of the variation in essential oil production, sterols and photosynthetic pigment content between transgenic and control plants was determined using analysis of variance (statgraphics plus for Windows version 2.1, Manugistics, Rockville, MD, USA), and mean comparisons using Tukey's (1953) procedure were carried out when appropriate. Inheritance observed data were compared to the expected ratios using a chi-squared analysis with Yates's correction (Zar, 1996).
This work was supported by DGICYT, Madrid, Spain (Project AGL2002-00977), and Generalitat Valenciana, Valencia, Spain (Projects GV2001-020 and Grupos 03/102). A FPU Research Fellowship (from the Spanish Ministerio de Educación y Cultura) to Jesus Muñoz-Bertomeu is acknowledged.