• Open Access

Enhancement of seed phytosterol levels by expression of an N-terminal truncated Hevea brasiliensis (rubber tree) 3-hydroxy-3-methylglutaryl-CoA reductase

Authors


Correspondence (fax +44 1234 222552; e-mail dick.safford@unilever.com)

Summary

Dietary intake of phytosterols (plant sterols) has been shown to be effective in reducing blood cholesterol levels, thereby reducing the risk of cardiovascular disease. Phytosterols are most commonly sourced from vegetable oils, where they are present as minor components. We report here the generation of transgenic tobacco seeds substantially enhanced in phytosterol content by the expression of a modified form of one of the key sterol biosynthetic enzymes, 3-hydroxy-3-methylglutaryl-CoA reductase (HMGR). The constitutive expression of an N-terminal truncated Hevea brasiliensis HMGR (t-HMGR), lacking the membrane binding domain, enhanced seed HMGR activities by 11-fold, leading to increases in total seed sterol of 2.4-fold. Seed-specific expression of t-HMGR enhanced total seed sterol levels by 3.2-fold, to 1.36% dry weight or 3.25% of oil. 4-desmethylsterols were increased by 2.2-fold, whilst certain sterol biosynthetic intermediates, in particular cycloartenol and 24-ethylidene lophenol, also accumulated. The additional sterol in seed tissue was present in the form of fatty acid esters. Constitutive expression of t-HMGR increased leaf phytosterol sterol levels by 10-fold, representing 1.8% dry weight, and the sterol was sequestered, in acyl ester form, as cytoplasmic ‘oil droplets’. These studies establish HMGR as a key enzyme controlling overall flux into the sterol biosynthesis pathway in seed tissue, but the accumulation of certain intermediates suggests additional slow steps in the pathway. The expression of an N-truncated HMGR activity has generated novel phytosterol-enriched raw materials that may provide the basis of new sourcing opportunities for this important class of cholesterol-lowering actives.

Introduction

The dietary intake of phytosterols has been shown to lower serum cholesterol levels in humans, thereby reducing the risk of cardiovascular disease (Westrate and Meijer, 1998). Phytosterols are natural dietary components primarily sourced from vegetable oils, where they are present as minor components. Phytosterols can be classified into three groups based on methylation levels at C4 as follows: 4-desmethylsterols, 4-monomethylsterols and 4,4-dimethylsterols. 4-desmethylsterols are the predominant naturally occurring class and these phytosterols are the most efficacious in terms of their cholesterol-lowering properties (Westrate and Meijer, 1998). β-Sitosterol, campesterol and stigmasterol are the most common 4-desmethylsterols and they are structurally related to cholesterol, differing primarily in the presence of an alkyl group at C-24 of the side chain. It is this subtle structural difference that is thought to result in the competitive inhibition of cholesterol uptake from the small intestine (Westrate and Meijer, 1998). Most of the phytosterols occur either in the free form or esterified with fatty acids. Phytosterols, chemically esterified with fatty acids to yield phytosterol esters, have been incorporated into fat-based products such as margarine as a functional ingredient with cholesterol-lowering properties (e.g. Westrate and Meijer, 1998). Our interest lies in understanding the key regulatory steps controlling sterol biosynthesis in seed tissue, with the ultimate aim of generating nutritionally enriched vegetable oils with an elevated phytosterol content. This study aims to enhance seed phytosterol levels by increasing carbon flow into sterol biosynthesis by the expression of a modified form of a key ‘early’ enzyme in the metabolic pathway, 3-hydroxy-3-methylglutaryl-CoA reductase (HMGR).

HMGR catalyses the irreversible conversion of 3-hydroxy-3-methylglutaryl-CoA to mevalonate, a precursor for the biosynthesis of sterols (Figure 1). A number of studies have highlighted the important role of HMGR in regulating sterol biosynthesis in leaf tissue, although no equivalent studies have been undertaken with seed. Mevinolin, a highly specific competitive inhibitor of HMGR, has been shown to severely inhibit plant growth and development, by limiting the availability of mevalonate for de novo sterol biosynthesis (Bach and Lichtenthaler, 1983). Characterization of a sterol-overproducing tobacco mutant (Gondet et al., 1992) showed a threefold increase in HMGR activity in leaf tissue compared to wild-type tissue.

Figure 1.

Schematic representation of the sterol biosynthesis pathway in plants. Solid lines represent single enzymatic reactions and dashed lines represent several enzymatic reactions.

Plants regulate HMGR activity at the transcriptional and post-translational level, the latter including reversible kinase-mediated phosphorylation (Russell et al., 1985) and regulated proteolytic degradation (Korth et al., 2000). Plant HMGR activity has been shown to respond to developmental and environmental signals such as cell division, light and pathogen infection (Stermer et al., 1994). The over-expression of Arabidopsis thaliana HMGR in Arabidopsis resulted in substantially enhanced levels of HMGR mRNA, but only modest increases in HMGR activity and no change in leaf sterol levels (Re et al., 1995), suggesting that HMGR activity is regulated at multiple levels in plant systems. Expression in tobacco of a laticifer-specific HMGR gene from Hevea brasiliensis (rubber tree), a high isoprenoid producing plant, has been reported to increase sterol levels in leaf (Schaller et al., 1995).

Studies in animal systems have shown that one mechanism for the post-transcriptional regulation of sterol biosynthesis is the accelerated degradation of HMGR triggered by end product sterols (Goldstein and Brown, 1990). Degradation is mediated via the N-terminal region of the HMGR enzyme that traverses the endoplasmic reticulum membrane. Deletion of the N-terminal membrane-binding domain of hamster HMGR was found to abolish the sterol-mediated turnover of the enzyme (Jingami et al., 1987). In plants, end product sterols have been shown to feedback-inhibit HMGR activity (Russell and Davidson, 1982), but the mechanism for this is unknown. The expression of an N-terminal truncated hamster HMGR in tobacco resulted in increased sterol levels in leaf tissue (Chappell et al., 1995). Plant HMGR enzymes differ significantly in the organization of their N-terminal membrane-binding domains compared to their animal counterparts, possessing only two membrane spanning sequences (Campos and Boronat, 1995; Denbow et al., 1996), compared to eight in animal HMGR (Roitelman et al., 1992). The consequences of this structural difference for plant HMGR in terms of the regulation of enzyme activity are unknown.

In this study we enhanced carbon flux into the sterol biosynthesis pathway in seed tissue by the expression of a modified HMGR activity. Specifically, we expressed in tobacco a laticifer-specific form of Hevea brasiliensis HMGR, from which the N-terminal membrane-binding domain had been removed. We describe the consequences of expression of this modified HMGR activity on sterol accumulation in seed but, moreover, highlight the differential effects observed in seed and leaf tissue.

Results

Constitutive expression of HMGR genes in tobacco

Our first experiments to enhance seed phytosterol levels employed the constitutive expression of a full length H. brasiliensis HMGR (pHEV36) in tobacco. Analysis of 30 transgenic plants revealed relatively modest increases, of up to 20%, in seed sterol levels (Figure 2a). Assays of developing seed from the highest sterol containing lines showed increases in HMGR activity of up to 1.8-fold (data not shown).

Figure 2.

Sterol analysis of T0 HEV36 and MH5 seed. (a) Total seed sterol levels of individual T0 tobacco lines (20 of 30 lines) transformed with the full-length H. brasiliensis HMGR under control of the 2 × 35S CaMV promoter (pHEV36). The solid line represents the average value of nine wild-type controls and the dashed lines represent the standard deviation. (b) Total seed sterol levels of individual T0 tobacco lines (20 of 30 lines) transformed with the N-terminal truncated H. brasiliensis HMGR under control of the 2 × 35S CaMV promoter (MH5). The solid line represents the average value of 10 wild-type controls and the dashed lines represent the standard deviation. (c) Seed sterol composition of high sterol MH5 lines as compared to the average value of 10 wild-type controls. (d) The free and esterified sterol content of intermediate and end product sterols in seed from wild-type and MH5 tobacco. The letter C refers to the wild-type control and T to transgenic line MH5#33. Sterol abbreviations: cycloart = cycloartenol, 24mca = 24-methylene cycloartanol, 24mloph = 24-methylene lophenol, 24eloph = 24-ethylidene lophenol, D7-avena = D7-avenasterol, isofuc = isofucosterol, sito = sitosterol, stig = stigmasterol, camp = campesterol, chol = cholesterol.

In an effort to further enhance flux into the sterol biosynthesis pathway, a construct (MH5) encoding a truncated H. brasiliensis HMGR, lacking the N-terminal membrane-binding domain (amino acids 1–153 of the mature HMGR), was transformed into tobacco under control of the 2 × 35S CaMV promoter.

Analysis of MH5 T0 transgenic seed

Analysis of mature seed from 30 MH5 T0 transformed plants showed increases in total sterol contents of up to 2.4-fold compared to the SR1 control mean (Figure 2b). Four plants had sterol contents at least twofold higher than the control mean and a further 15 plants had 1.5-fold higher levels. The maximum enhanced sterol values equated to 0.9% dry weight, or approximately 2.1% of oil content. In terms of sterol composition, the high sterol lines contained substantial increases, up to 1.7-fold, in 4-desmethylsterols and also elevated levels of sterol biosynthetic intermediates, in particular cycloartenol which represented up to 30% of total sterol (Figure 2c). The major seed sterol, β-sitosterol, was enhanced by up to 1.5-fold, isofucosterol, the second most abundant sterol, by up to 2.1-fold and campesterol by up to 1.8-fold. Interestingly, despite the overall increase in the characteristic ‘phyto’ 4-desmethylsterols, the levels of cholesterol, also a 4-desmethylsterol, were relatively unchanged.

Further analysis of high sterol samples was carried out to determine the degree of sterol esterification (Figure 2d). Control seed was found to contain, on average, 48% of total sterol as free sterol and the remainder in the acyl ester form. Seed from the high sterol transgenic lines showed only marginally higher levels of free sterol than control seed and essentially all the ‘additional’ sterol was present in the esterified form (representing up to 74% of total sterol). Interestingly, both sterol biosynthetic ‘intermediates’ and end product sterols were esterified to a high degree. The exception was stigmasterol, where the free sterol form was dominant over the ester form, reflecting the high proportion of free stigmasterol present in wild-type seed. Cycloartenol had the highest proportion of sterol in the ester form, greater than 90% in some lines.

Analysis of MH5 T1 transgenic lines

Seed from a number of primary transgenic (T0) MH5 plants, containing single HMGR transgene inserts, was germinated on kanamycin and second generation (T1) plants were grown. Sterol analysis of mature T1 seed showed that the enhanced sterol content observed in the first generation was stable in the subsequent generation (Figure 3a). In some cases, higher sterol contents were observed than in corresponding T0 seed, with up to 2.7-fold increases over control levels. The sterol compositions of the T1 samples were similar to the T0 counterparts (data not shown).

Figure 3.

Sterol analysis of T1 MH5 seed. (a) Total seed sterol levels in T1 lines transformed with the N-terminal truncated H. brasiliensis HMGR under control of the 2 × 35S CaMV promoter (MH5). T0 seed germinated on kanamycin and T1 plants grown and mature seed harvested. The solid line represents the average value of 10 wild-type controls and the dashed lines represent the standard deviation. (b) The developmental profile of HMGR activity in T1 seed of MH5 and wild-type tobacco. Symbols •, ○, ∇ and ▾ represent 4 MH5 : 33 lines whilst ▪ represents the average of three wild-type controls.

HMGR assays were carried out on developing seed from a number of T1 lines derived from the high sterol MH5#33 primary transgenic plant. The seed was harvested from each line at five developmental stages and assayed along with corresponding samples from SR1 control plants. Figure 3b shows substantial increases in the HMGR activity profiles of the MH5 lines compared to the control seeds, with increases of up to 11-fold in peak activities. This would suggest that the enhanced sterol levels are a result of increased HMGR activity.

Seed specific expression of N-truncated HMGR

In an effort to further increase phytosterol levels specifically in seed tissue, tobacco was transformed with a binary vector (pNH61) containing the truncated H. brasiliensis HMGR gene linked to a seed-specific acyl carrier protein (ACP) promoter isolated from Brassica napus (de Silva et al., 1992). Mature seed from 22 independent T0 NH61 transgenic plants was analysed for sterol levels, and increases of up to 3.2-fold over the SR1 control mean were obtained (Figure 4a). Seventeen of the 22 transgenic plants had at least twice the sterol content of the control mean. The highest sterol values are equivalent to 1.36% dry weight or 3.24% oil content. The levels of sterol increase are clearly greater than those obtained with the 2 × 35S CaMV constitutive promoter constructs in seed tissue. 4-desmethylsterols were increased by up to 2.2-fold over the SR1 control mean and, in general, the proportion of 4-desmethylsterols as a function of total sterols is similar to that seen when the t-HMGR is driven by the 2 × 35S CaMV promoter. However, interestingly, some differences in the composition of the sterol biosynthetic intermediates that accumulate were observed. Thus, in contrast to the situation where the t-HMGR was driven by the 2 × 35S CaMV promoter, where cycoartenol was the major intermediate accumulating, in this case both cycloartenol and the 4-monomethylsterol 24-ethylidene lophenol accumulated to approximately 15% of total sterols (Figure 4b). As with the MH5 transgenic seeds, the majority of the ‘additional’ sterol in the NH61 seeds was present in the esterified form (data not shown).

Figure 4.

Sterol analysis of T0 NH61 seed. (a) Total seed sterol levels of individual T0 tobacco lines transformed with the N-terminal truncated H. brasiliensis HMGR under control of the ACP promoter (NH61). The solid line represents the average value of 10 wild-type controls and the dashed lines represent the standard deviation. (b) Seed sterol composition of high sterol NH61 lines as compared to the average value of 10 wild-type controls. Sterol abbreviations as in Figure 2.

Analysis of MH5 transgenic leaf tissue

Sterol analysis of mature leaf samples from the MH5-transformed plants revealed dramatic increases in sterol content, up to 10-fold higher than the SR1 control mean of 0.18% dry weight (Figure 5a). Some 60% of the transgenic samples had sterol contents fivefold greater than the control mean. The compositional changes in the high sterol leaf were marked, with cycloartenol levels enhanced up to 100-fold, accounting for some 40–50% of total sterol content (Figure 5b). Levels of the 4-desmethylsterol class of sterols were increased by up to 3.4-fold compared to the control mean. Interestingly, levels of the major leaf 4-desmethylsterol, stigmasterol, were largely unchanged, although levels of β-sitosterol, the immediate precursor of stigmasterol, were increased by up to sevenfold. As in seed, virtually all the additional sterol in the high sterol leaf was in the ester form, representing up to 80% of total sterol (data not shown).

Figure 5.

Sterol analysis of T0 MH5 leaf. (a) Total leaf sterol levels in T0 tobacco lines transformed with the N-terminal truncated H. brasiliensis HMGR under control of the 2 × 35S CaMV promoter (MH5). The solid line represents the average value of 10 wild-type controls and the dashed lines represent the standard deviation. (b) Leaf sterol composition of high sterol MH5 lines as compared to the average value of 10 wild-type controls. Sterol abbreviations as in Figure 2. (c) Fluorescence microscopy of the epidermis layer of wild-type and MH5 tobacco stained with Nile blue. Bar = 15 µm.

Analysis of leaf tissue from eight individual MH5#33 T1 plants showed HMGR activities ranging from five to eightfold higher than the control average (data not shown). The sterol levels and composition of the samples were similar to the T0 counterparts (data not shown).

To investigate the intracellular location of the ‘additional’ sterol in the high sterol lines, leaf epidermal strips from a high sterol line were stained with the lipid-specific dye Nile Blue. Light microscopy revealed the presence of numerous ‘oil droplets’ within the cytoplasm, ranging in size from 1 to 5 µm (Figure 5c). The oil droplets were essentially absent from control tissue (Figure 5c), strongly suggesting these structures represent sterol-containing ‘storage bodies’. Confocal Raman spectroscopy was used to probe, in situ, the composition of droplets in MH5 leaf tissue. Comparison of the spectra obtained from individual droplets with those of ‘free’ sterol (sitosterol, stigmasterol) and sterol ester (cholesterol oleate) standards confirmed that the droplets contained high levels of sterol and that the sterol was esterified to a high degree (data not shown).

The transgenic plants expressing truncated HMGR, either in a constitutive or a seed-specific manner, showed no obvious morphological differences from the wild-type, and they flowered and set seed normally.

Discussion

We report here, for the first time, the generation of transgenic seeds, substantially enhanced in phytosterol content by expression of an N-terminal truncated H. brasiliensis HMGR. Previous studies employing constitutive expression of a full length H. brasiliensis HMGR in tobacco reported enhancement of sterols in leaf tissue (Schaller et al., 1995). In our studies, the constitutive expression of this gene produced only modest increases in seed sterol levels. In this respect our data is more consistent with findings obtained with over-expression of full length Arabidopsis HMGR in Arabidopsis thaliana (Re et al., 1995). These authors reported high levels of HMGR mRNA, but only modest increases in enzyme activity and no changes in levels of sterols or sterol precursors, suggesting that HMGR expression in plants is controlled at multiple levels.

The expression of a truncated version of the HMGR enzyme, lacking the N-terminal membrane-binding domain, did, however, result in a more dramatic enhancement of sterol levels. Expression of t-HMGR resulted in elevated HMGR activity that, in turn, led to enhanced sterol levels. In animal systems, flux into the sterol biosynthesis pathway is controlled, in part, by the turnover of HMGR (Goldstein and Brown, 1990) and removal of the N-terminal membrane-binding domain of HMGR abolishes this sterol-mediated turnover of the enzyme (Jingami et al., 1987). In plants, controlled proteolytic degradation has been proposed as one post-translational mechanism for regulating the levels of HMGR in response to environmental stimuli in plants (Korth et al., 2000), but no information is available in respect of this mechanism. Expression of an N-truncated animal HMGR in tobacco has previously been shown to enhance sterol levels in leaf tissue (Chappell et al., 1995). Sequence comparisons of plant HMGR enzymes with those from mammalian systems show a high degree of sequence identity within the catalytic domains, but a significant divergence in the membrane-binding regions. Mammalian HMGRs contain eight membrane spanning domains, in contrast to plant HMGRs which only contain two transmembrane domains (Campos and Boronat, 1995; Denbow et al., 1996; Roitelman et al., 1992). The consequences of this difference in structural organization of the plant HMGRs, in relation to the regulation of enzyme activity, has hitherto been unknown. However, data from this study would suggest that plant HMGR activity is, controlled, at least in part, via the N-terminal membrane-binding domain. Thus expression of an N-truncated HMGR in tobacco led to increased HMGR activity compared to that obtained with expression of the full length HMGR. This increased HMGR activity presumably enhanced carbon flux into the isoprenoid pathway via an increased production of mevalonate, that, in turn, led to elevated levels of sterols.

A substantial proportion of the increases in total seed sterol in the transgenic plants was in the form of 4-desmethylsterols (the most efficacious cholesterol-lowering phytosterols, see Westrate and Meijer, 1998). Seeds also accumulated significant levels of certain sterol biosynthetic intermediates, in particular the 4,4-dimethylsterol cycloartenol and the 4-monomethylsterol 24-ethylidene lophenol. Interestingly, the ratio of these two intermediates varied depending on the nature of the promoter driving the t-HMGR transgene. The accumulation of high levels of cycloartenol has been observed in leaf tissue where HMGR activity has been increased either directly (Chappell et al., 1995; Schaller et al., 1995) or indirectly (Gondet et al., 1992). These findings led to the suggestion that the next step in the pathway, namely the conversion of cycloartenol to 24-methylene cycloartanol, catalysed by C-24 sterol methyl transferase1 (SMT1), is a ‘slow step’ in the synthesis of 4-desmethylsterols (Chappell et al., 1995; Schaller et al., 1995). It would appear from the current study that conversion of cycloartenol to 24-methylene cycloartanol is also a ‘slow step’ in seed tissue. Based on these findings it would be predicted that the co-expression of t-HMGR with SMT1 would increase the proportion of 4-desmethylsterols synthesized as a function of total sterols, and we are currently investigating the validity of this hypothesis in transgenic tobacco. Driving the t-HMGR transgene via the seed-specific ACP promoter resulted in a greater increase in total seed sterols than seen with the 2 × 35S CaMV promoter construct, but the proportion of 4-demethylsterols as a function of total sterols was broadly the same. The composition of the sterol biosynthetic intermediates from the two transgenic populations was, however, different. When the t-HMGR was driven by the seed specific promoter, in addition to accumulation of cycloartenol seen with the 2 × 35S CaMV promoter construct, the 4-monomethylsterol 24-ethylidene lophenol also accumulated, suggesting that the C4-demethylation of 24-ethylidene lophenol to avenasterol is a further ‘slow step’ in seed tissue. It is interesting to consider whether the accumulation of different sterol intermediates is a function of increased flux through the pathway, driven primarily by promoter strength, or whether it results from a different temporal regulation of the promoters during seed development. Interestingly, levels of cholesterol, a 4-desmethylsterol, were relatively unaltered in comparison to changes in levels of the ‘phyto’ 4-desmethylsterols. In plants, cholesterol biosynthesis branches from the pathway leading to the common end product sterols, such as β-sitosterol and stigmasterol, at cycloartenol and this result suggests that this part of the pathway contains additional control steps.

The ‘additional’ sterol in the high sterol transgenic seed tissue was present in the form of fatty acid esters, as has been observed in the leaf tissue of other sterol-overproducing plant systems (Gondet et al., 1992; Schaller et al., 1995). Due to the very low solubility of free sterols in fat, phytosterols are incorporated into fat-based cholesterol-lowering food products in the form of fatty acid esters. Currently, free sterols are distilled from vegetable oils and then chemically esterified with fatty acids for product formulation. The provision of oils containing high levels of phytosterol esters opens up the possibility of the direct use of oils, circumventing the need for chemical esterification.

The high degree of sterol esterification seen in the transgenic lines suggests that the activity of sterol acyl transferase (SAT), the enzyme responsible for sterol esterification, is not limiting sterol biosynthesis. Indeed the activity may be inducible in response to increased flux through the sterol biosyn-thesis pathway, as has been suggested to occur in a sterol-overproducing tobacco mutant (Bouvier-Nave and Benveniste, 1995). SAT is ostensibly regarded as the ‘terminal’ enzyme in the sterol biosynthesis pathway, functioning to sequester ‘excess’ end product sterols (4-desmethylsterols). However, the fact that nearly all the sterol biosynthetic ‘intermediates’ in the high sterol lines were esterified to a high degree demonstrates that esterification is not restricted to end-product sterols. This suggests that either the SAT enzyme has broad substrate specificity or that there are several SAT isoforms exhibiting differing substrate specificity. The finding also raises the intriguing question as to whether the esterified intermediates remain substrates for subsequent enzymatic reactions or whether, once esterified, they are ‘sequestered’ and unavailable for conversion to end-product sterols.

Expression of t-HMGR via the constitutive 2 × 35S CaMV promoter gave a much greater degree of sterol enhancement in leaf (10-fold) than seed (2.4-fold), resulting in a comparatively higher sterol content, on a dry weight basis, in the former tissue. In leaf the additional sterol is esterified and stored in the form of cytoplasmic ‘oil droplets’. This esterification and storage mechanism is presumably necessary to tightly regulate the pool of free sterol in order to maintain optimal functioning of the cell (Maillot-Vernier et al., 1991). Unsurprisingly, the accumulation of sterol biosynthesis intermediates was much greater in leaf than in seed, with cycloartenol representing > 40% of total sterols. Interestingly, levels of stigmasterol, the major leaf sterol, were not increased despite the fact that levels of β-sitosterol, the immediate precursor of stigmasterol, were substantially increased. This suggests that the C22-desaturase that converts β-sitosterol into stigmasterol is also a highly controlled step in this tissue. In the highest sterol containing lines sterol constituted 1.8% of leaf dry weight. Enhancement to such a level coupled with the concentrated, stable physical form (cytoplasmic droplets) of the sterol raises the intriguing question as to the feasibility of sourcing sterols from such a novel tissue.

In conclusion, we have demonstrated that expression of an N-terminal truncated H. brasiliensis HMGR leads to significant enhancement of sterol accumulation in tobacco. The results establish, for the first time, the key role of HMGR in controlling flux into the sterol biosynthetic pathway in seed tissue. Whilst HMGR is clearly shown to be a key ‘early’ enzyme regulating sterol biosynthesis in seed, the accumulation of significant levels of sterol intermediates in transgenic seed tissue does demonstrate that further critical slow steps exist in the pathway. These steps serve to highlight that the over-expression of several genes will be required in order to maximize the levels of 4-desmethysterols, the most efficacious cholesterol-lowering phytosterol class. To that end the consequences of co-expressing t-HMGR and SMT1 in tobacco will be reported in the near future. Expression of an N-truncated HMGR activity has generated phytosterol-enriched plant tissues, and these nutritionally enhanced raw materials may provide the basis for novel sourcing opportunities for this important class of cholesterol-lowering actives.

Experimental procedures

Plasmids

The binary vector pHEV36, containing a 2.1 kb cDNA encoding Hevea brasiliensis hmg1 (GenBank accession no. X54659) linked to the 2 × 35S CaMV promoter, was obtained from N. Chua, Rockefeller University, New York. A truncated form of the H. brasiliensis hmg1 gene, encoding amino acids 153–575 of the full sequence, was cloned by PCR using primers based on the published sequence (Chye et al., 1991). The truncated gene was inserted between the 2 × 35S CaMV promoter and nos terminator of pNH4, a modified pUC 19 vector, to give plasmid pMH3. The chimaeric gene was digested from pMH3 and cloned into pSJ34, a modified version of pGPTV-Kan (Becker et al., 1992), to give the binary vector pMH5. For seed specific expression, the Brassica napus acyl carrier protein (ACP) promoter including the 5′-untranslated region was amplified by PCR from pTZ5BS vector (de Silva et al., 1992). The ACP promoter and the t-hmg1 fragments were inserted into a modified poly linker of pUC19 from which the resulting ACP-t-hmg1-nos expression cassette was cloned into pSJ34 to give the binary vector pNH61.

Generation of transgenic plants

Binary vectors were used to transform Nicotiana tabacum cv. SR1 as previously described (An et al., 1988). PCR positive plants were transferred into soil and grown under 24 °C day/18 °C night regime with a 16 h daylight period. Mature leaf and seed samples were collected, frozen in liquid nitrogen and stored at –80 °C until sterol analysis. Seed from high sterol primary transgenic lines was germinated on kanamycin-containing media and second-generation (T1) plants were grown and mature seed harvested for sterol analysis. Developing seed from MH5#33 T1 plants was collected at 10, 14, 18, 22 and 26 days after anthesis for assay of HMGR activity. Developing leaf tissue was assayed for HMGR activity.

Sterol analysis

Leaf and seed tissues were freeze-dried and extracted in chloroform/methanol 2 : 1 (v/v) at 80 °C. After filtration and removal of solvent, the lipid residue was dissolved in toluene followed by sodium methoxide to a concentration of 0.33 m. The mixture was heated for 30 min at 80 °C followed by a further 10 min at 80 °C in the presence of 5.6% boron trifluoride. Following diethyl ether extraction and washing with water, the ether was evaporated and the free sterol silylated by addition of trimethylchlorosilane: N,O-bis (trimethylsilyl) acetamide (5 : 95) and heating for 10 min at 50 °C. Gas chromatographic analysis was carried out using a Perkin-Elmer 8420 GC equipped with a BPX5 column. The temperature programme was 80–230 °C at 45 °C/min, followed by 230–280 °C at 4 °C/min and 355 °C for 6 min. Peak areas were calculated automatically using Turbochrom software. The identity of the sterols was confirmed by GC-MS, using a Hewlett Packard 5890 GC coupled to a Quadrapole 5972A MSD.

Free sterol and sterol ester fractions were separated by TLC, using petroleum ether/diethyl ether/acetic acid (80 : 20 : 2), and the fractions eluted with diethyl ether. The sterol ester fraction was heated for 2 h at 70 °C with 8% potassium hydroxide in 80% aqueous ethanol and the liberated sterols extracted with hexane. Free sterol and sterol ester fractions were silylated and analysed as above.

HMGR assay

Developing seeds and leaves were homogenized in the ratio 1 : 10 (w/v) tissue : homogenization buffer (0.2 m potassium phosphate, pH 7.5 containing 0.4 m sucrose, 10 mm EDTA, 5 mm MgCl2, 5 mm glutathione and 4 g/100 mL insoluble polyvinylpolypyrrolidone) using an Ultra-turrax. Extracts were centrifuged for 5 min at 1200 g and, for leaf samples, the supernatants were assayed immediately for HMGR activity (Chappell et al., 1995). For seed samples the supernatant was discarded and the lipid and pellet fractions were extracted together with homogenization buffer containing 2% Brij (w/v) and left on ice for 20 min. After centrifugation at 1200 g for 5 min, the supernatant and floating lipid layer were retained. The pellet was re-extracted as above and the supernatant/lipid fractions were pooled and assayed immediately for HMGR activity (Chappell et al., 1995). TLC analysis of products was carried out as described previously (Schaller et al., 1995).

Confocal Raman spectroscopy

A Kaiser holoprobe 5000r Raman spectroscope, with a 785 nm laser, coupled to an Olympus microscope was used to examine epidermal layers of MH5 : 55 leaf tissue. Lipid droplets were located via the microscope using bright field light illumination and then examined by Raman spectroscope. Spectra obtained were compared with those obtained from standards of sitosterol, stigmasterol and cholesterol oleate.

Acknowledgements

We would like to thank Mike Asquith for generation of the micrographs and Paul Pudney for provision of the in situ Raman spectroscopy data.

Ancillary