Localization of tabersonine 16-hydroxylase and 16-OH tabersonine-16-O-methyltransferase to leaf epidermal cells defines them as a major site of precursor biosynthesis in the vindoline pathway in Catharanthus roseus
Department of Biological Sciences, Brock University, 500 Glenridge Avenue, St Catharines, Ontario, L2S3A1 Canada
The Madagascar periwinkle (Catharanthus roseus) produces the well known and remarkably complex anticancer dimeric alkaloids vinblastine and vincristine, which are derived by the coupling of vindoline and catharanthine monomers. Recent data from in situ RNA hybridization and immunolocalization suggest that combinatorial cell factories within the leaf are involved in vindoline biosynthesis. In this study, the cell types responsible for vindoline biosynthesis were identified by laser-capture microdissection/RNA isolation/RT–PCR to show that geraniol hydroxylase, secologanin synthase, tryptophan decarboxylase, strictosidine synthase, strictosidine ß-glucosidase and tabersonine 16-hydroxylase can be detected preferentially in epidermal cells. A new and complementary application of the carborundum abrasion (CA) technique was developed to obtain epidermis-enriched leaf extracts that can be used to measure alkaloid metabolite levels, enzyme activities and gene expression. The CA technique showed that tabersonine and 16-methoxytabersonine, together with 16-hydroxytabersonine-16-O-methyltransferase, are found predominantly in Catharanthus leaf epidermis, in contrast to vindoline, catharanthine and later enzymatic steps in vindoline biosynthesis. The results show that leaf epidermal cells are biosynthetically competent to produce tryptamine and secologanin precursors that are converted via many enzymatic transformations to make 16-methoxytabersonine. This alkaloid or its 2,3 dihydro-derivative is then transported to cells (mesophyll/idioblast/laticifer) within Catharanthus leaves to complete the last three or four enzymatic transformations to make vindoline.
Catharanthus roseus (Madagascar periwinkle) produces a variety of monoterpenoid indole alkaloids (MIAs), some of which are particularly important because of their pharmacological and therapeutic applications. For example, vinblastine and vincristine, both of which are bisindole alkaloids derived from vindoline and catharanthine, have been used for cancer chemotherapy. The MIA biosynthetic pathway in C. roseus has been studied extensively in the past two decades, with the majority of studies focusing on either the biochemical or the molecular characterization and cloning of genes involved in vindoline biosynthesis.
The MIAs are composed of an indole moiety derived from the tryptophan decarboxylase (TDC)-mediated conversion of tryptophan to tryptamine (reviewed by De Luca, 1993) and the monoterpene secologanin, derived from the plastidic 2-C-methyl-d-erythritol 4-phosphate (MEP) pathway (Contin et al., 1998). Several genes in the MEP pathway have been cloned, including 1-deoxy-d-xylulose 5-phosphate synthase (DXS); 1-deoxy-d-xylulose 5-phosphate reductoisomerase (DXR); and 2-C-methyl-d-erythritol-2,4-cyclodiphosphate synthase (MECS) (Chahed et al., 2000; Veau et al., 2000). Geraniol 10-hydroxylase (G10H) and secologanin synthase (SLS), two membrane-associated cytochrome P450-dependent mono-oxygenases involved in late stages of secologanin biosynthesis, have also been characterized (Collu et al., 2001; Irmler et al., 2000; Yamamoto et al., 2000). A soluble vacuolar strictosidine synthase (STR) catalyses the stereospecific condensation reaction of tryptamine and secologanin into 3-α-(S)-strictosidine, a central precursor of several thousand MIAs (reviewed by De Luca, 1993). The structural diversity of MIAs comes first from the deglucosylation of strictosidine by strictosidine β-glucosidase (SGD), which may be associated with the endoplasmic reticulum (Geerlings et al., 2000). This unstable and versatile aglycone is biochemically converted by mostly uncharacterized enzyme reactions to the Corynanthe, Iboga and Aspidosperma classes of alkaloids such as ajmalicine, catharanthine and tabersonine, respectively.
The vindoline branch of the pathway involves hydroxylation of tabersonine by endoplasmic reticulum-associated cytochrome P450-dependent mono-oxygenase, tabersonine 16-hydroxylase (T16H) (Schröder et al., 1999; St-Pierre and De Luca, 1995), O-methylation via 16-hydroxytabersonine-16-O-methyltransferase (16-OH OMT), followed by an uncharacterized conversion to 2,3-dihydro-16-methoxytabersonine, and by a chloroplast thylakoid-associated 2,3-dihydro-16-methoxytabersonine N-methyltransferase (NMT) reaction to yield desacetoxyvindoline. Desacetoxyvindoline 4-hydroxylase (D4H) (Vazquez-Flota et al., 1997) and deacetylvindoline 4-O-acetyltransferase (DAT) (St-Pierre et al., 1998) catalyse the last two steps in vindoline biosynthesis. While several of these genes have been characterized, the three reaction steps between 16-hydroxytabersonine and desacetoxyvindoline have yet to be characterized in detail (van der Heijden et al., 2004).
Recent studies to locate the sites of vindoline biosynthesis through in situ RNA hybridization and immunolocalization experiments have revealed that multiple cell types in the leaf and in root tips of C. roseus are involved (Burlat et al., 2004; St-Pierre et al., 1999). All known genes related to the biosynthesis of the terpenoid part of MIA biosynthesis, DXS, DXR, MECS and G10H, were shown to be localized in vascular cells at the mRNA level, whereas another set of genes in the earlier steps of this pathway, TDC, STR and SLS, were expressed exclusively in epidermal cells. On the other hand, two of the gene products involved in the later steps of vindoline biosynthesis, D4H and, in particular, DAT, were expressed in specialized laticifer and idioblast leaf cells, whereas their expression was never detected in roots. In addition, TDC, STR, D4H and DAT genes are expressed at higher levels in the base part of the leaf than at the tip, at both mRNA and protein levels, suggesting that their expression is also developmentally regulated (St-Pierre et al., 1999). These and other findings suggest that expression of vindoline biosynthesis is strictly controlled at the cellular level; parts of the pathway are localized to particular compartments within cells; and one or more pathway intermediates should be translocated from one cell type to another (Burlat et al., 2004; St-Pierre et al., 1999). These results partially explain why cell-culture systems have not been used successfully to manufacture vindoline or the derivative dimeric alkaloids of commercial value.
These observations triggered the present studies to characterize each alkaloid cell factory, to acquire a comprehensive road map of vindoline biosynthesis by metabolite profiling, enzyme activity assays and RNA expression profiling at the cellular level. This report highlights the application of the laser-capture microdissection (LCM) and carborundum abrasion (CA) techniques to localize MIA metabolites, enzyme activities and gene expression of vindoline biosynthetic enzymes at the cellular level. The results showed that G10H and SLS involved in secologanin biosynthesis are also expressed in epidermal and laticifer cells, as well as in the vascular cells of leaves. In addition, the preferential expression of T16H and SGD in epidermal cells, and the enzymatic activities of 16-OH OMT in epidermal cells and of NMT in non-epidermal cells inside the leaf, suggest that the whole pathway from primary metabolism to 16-methoxytabersonine may be expressed in leaf epidermis, whereas at least the last three steps in vindoline biosynthesis are probably expressed in separate cells (mesophyll/idioblast/laticifer) within the Catharanthus leaf.
Laser-capture microdissection to harvest single cells
In order to extract mRNA from specific Catharanthus cells, four cell types and one tissue were harvested by LCM from the base part of young leaves, including epidermal cells (EP), palisade mesophyll cells (PM), palisade-assisted idioblast cells (PI), cross-connected laticifer cells (CL) and vascular cells (VS), where several genes associated with vindoline biosynthesis have been already localized. Tissues were prepared for histology to make longitudinal and cross-sections for the isolation of EP and VS, respectively, whereas paradermal sections were used to isolate PM, PI and CL. The typical longitudinal section is 150–300 μm thick, containing seven to 10 cell layers composed of the multiple cell types of a typical Catharanthus leaf. Cell-capturing procedures were monitored visually under the microscope, where each cell type could be precisely targeted, shot with a laser and captured on a special plastic film (Figure 1a–p). The halos that appeared in and around the targeted cells after laser capture showed the actual targeted area, and ensured that only the target cells were captured (Figure 1d,h,i,p). Each of the targeted cell types (Figure 1: EP, a–d; CL, e–h; PI, i–l; VS, m–p) could be distinguished morphologically from adjacent cells, and this facilitated their isolation. For example, CL was easily targeted by its unique canal-like structure (Figure 1e); whereas PI cells could be identified by their rounder shape, apparently larger size compared with adjacent palisade mesophyll cells, and their organized distribution where the average distance between two PI cells was approximately 50 μm (Figure 1i). The VS was also simple to identify as it has a characteristic organization and appeared darker than other cells under the microscope, due to tighter packing into smaller areas (Figure 1m). Depending on the cell type, repeated laser shots enabled the harvest of up to 2000 cells onto a single plastic film and, under optimized laser pulse parameters, only targeted cells were harvested without disrupting any surrounding cells. Varying numbers of cells (2500–5000) were captured from each cell type to yield between 6 and 12 ng total RNA that could be used for T7-based RNA amplification.
Expression analysis of MIA pathway genes in LCM captured cells
As the amount of RNA isolated from extraction of LCM-dissected cells was not sufficient for direct RT–PCR analysis, it was amplified by T7-based RNA amplification prior to PCR, as described previously (Nakazono et al., 2003). The quality of cDNA produced by this method was analysed by performing RT–PCR with primers based on nine genes that are involved in MIA biosynthesis, including the epidermis-specific markers TDC and STR as well as the idioblast/laticifer-specific markers D4H and DAT, and Actin as a control marker (Figure 2). Results for TDC, STR, D4H and DAT showed expression profiles consistent with those obtained by in situ RNA hybridization (St-Pierre et al., 1999), although the presence of DAT in idioblasts could not be verified by this technique. Diagnostic sequencing of the DAT PCR product matched exactly the published sequence for this gene (St-Pierre et al., 1998). Together, these results validated that the RNA isolated from these cell types was specific to the leaf epidermis, idioblast or laticifer, respectively, and that LCM combined with this type of RNA analysis could be useful to localize the cell-specific expression of other enzymes in MIA biosynthesis. The results clearly corroborate previous studies (St-Pierre et al., 1999) showing that TDC and STR are expressed preferentially in epidermal cells; D4H appears in epidermal cells, idioblast cells and laticifer cells; whereas DAT shows preferential expression in laticifer cells. It should be mentioned, however, that two of three independent experiments showed DAT expression could also be found partly in epidermal cells. In addition to these results, expression of T16H and SGD was also preferentially localized to the epidermis, whereas octadecanoid-derivative responsive Catharanthus AP2-domain3 (ORCA3), an AP2/ERF type of transcription factor that regulates expression of multiple MIA pathway genes, including TDC, STR and D4H (van der Fits and Memelink, 2000), appeared in all cell types tested. non-e of these transcripts, except D4H, ORCA3 and Actin, could be detected in LCM-captured mesophyll cells. While previous studies showed that expression of G10H and SLS were detected in vascular cells (Burlat et al., 2004), RT–PCR with LCM isolated materials suggested that epidermal and laticifer cells may also express this gene (Figure 2).
Development of carborundum abrasion technique for mRNA, protein and alkaloid extraction from epidermal cells
The LCM results suggest that epidermal cells of Catharanthus may be the major factory for MIA biosynthesis in leaves. This prompted the development of the CA technique, which abrades the leaf surface in a uniform manner in order to differentially extract indole alkaloids; enzymes active in MIA biosynthesis; and RNA from the leaf epidermis. As shown by scanning electron microscopy (SEM), no obvious damage to the leaf surface (Figure 3c) was observed in a leaf abraded on the abaxial surface, compared with an unabraded control (Figure 3a) or a leaf dipped in extraction buffer without abrasion (Figure 3b), except that a few trichomes were bent visibly on abrasion (Figure 3c). The lack of visible effects seen by SEM contrasted with the clear damage observed in abraded leaf paraffin sections of abaxial and adaxial epidermis, but not in palisade and spongy mesophyll cells (Figure 3f,g) compared with the lack of damage seen in unabraded controls (Figure 3e). When the leaf surface was abraded more severely, epidermal cells were heavily disrupted, and damage to the secondary cell layer could be observed both with SEM (Figure 3d) and in paraffin sections (Figure 3h). Visual inspection of paraffin sections (Figure 3h) revealed that 30–50% of epidermal leaf cells were damaged by the severe leaf abrasion treatment.
Although epidermal guard cells do contain chloroplasts, the amount of chlorophyll per cell is 25- to 100-fold lower than in mesophyll cells (Willmer and Fricker, 1996). This information was useful as the degree of contamination in extracts with mesophyll cells was measured conveniently via chlorophyll analysis to show that epidermal cell extracts were not contaminated with palisade or spongy mesophyll cells (Table 1). The occurrence of guard cells in the epidermal cell layer was estimated as 5% of the total epidermal surface area (or 10% or the total number of epidermal cells) as calculated visually from SEM analysis. The ratio of the numbers of both palisade and spongy mesophyll cells against epidermal cells was approximately 4:1, as observed in paraffin sections, suggesting that approximately 1% of leaf cells are guard cells. Taken together, the predicted ratio of chlorophyll content in epidermal cells compared with mesophyll cells should be 1% of the total, and the epidermal chlorophyll content obtained in this study (0.5%, Table 1) reflects the low level of mesophyll contamination obtained by the CA technique. These observations helped to establish the precise CA method for harvesting the maximum amount of epidermis extract without increasing the background contributions of other cells within the leaf (see Experimental procedures).
Table 1. Relative levels of MIAs and chlorophyll in whole leaf, leaf surface and in Ca treated leaves
The values expressed in the table are in μg g−1 FW and in % (within parenthesis) related to whole leaf.
Chlorophyll and alkaloids accumulate differentially in leaf epidermal extracts obtained by the CA technique
To further verify the specificity of the CA technique for extraction of epidermal cells, chlorophyll content in the epidermal cell extract was compared with that of whole leaf or whole abraded leaf. The epidermal cell extracts contain very low levels of chlorophyll (0.5% of total) compared with levels found in whole leaves or in whole abraded leaves (Table 1). As the ratios of chlorophyll content in the epidermis were about 12-fold lower than those for tabersonine, chlorophyll qualifies as mesophyll-specific marker, and the results suggest that the CA technique extracted mostly epidermal materials.
As metabolite analysis, in addition to the gene-expression profile, is an important parameter for understanding the biochemical involvement of leaf epidermis in MIA biosynthesis, indole alkaloids were successfully extracted from epidermal cells by the CA technique and metabolites were analysed using high-performance liquid chromatography (HPLC) (Table 1). The vindoline and catharanthine levels in leaf epidermal cell extracts were very low (0.16–1.8% of total) compared with the whole leaf. These results suggest that vindoline and catharanthine are located in the central part of the leaf body, which includes palisade mesophyll cells, idioblasts, laticifers and vasculature. In contrast, the levels of tabersonine and 16-methoxytabersonine found in epidermal cells were five and 11 times higher, respectively, compared with the whole leaf than those of vindoline and catharanthine, despite the fact that Catharanthus leaves contain 2000- to 3000-fold more vindoline/catharanthine than tabersonine (Table 1). This differential distribution of tabersonine within the epidermis is consistent with the LCM results suggesting that the epidermis is the main site of leaf tabersonine biosynthesis.
Localization of MIA-pathway enzyme activities of CA-extracted epidermal cells and in leaves
The successful differential extraction of indole alkaloids by the CA technique was followed by the development of protein-extraction methods to show that extracts could be assayed for different enzyme activities. The 16-OH OMT and NMT, both involved in the conversion of tabersonine to vindoline, were studied using [14C-Methyl]-S-adenosylmethionine (AdoMet) as a methyl group donor, together with their respective indole alkaloid substrates. Activity of TDC and DAT was monitored as reference activity for epidermal and non-epidermal cells, respectively. The radiolabelled alkaloid products from each enzyme assay were analysed by thin-layer chromatography (TLC) to identify the predominant sites of different reactions in the pathway.
Activity of TDC was preferentially expressed in epidermal cells, while DAT activity was found only in the whole leaf (Figure 4), consistent with previous observations from in situ RNA hybridization and immunolocalization studies which located expression of these genes in leaf epidermis and in leaf idioblasts/laticifers, respectively (St-Pierre et al., 1999). An even higher ratio of 16-OH OMT activity was detected within abaxial and adaxial epidermal cells than in whole leaf (Figure 4), compared with the results obtained with TDC. The leaf epidermal location of this 16-OH OMT contrasted strongly with the lack of NMT activity observed in this cell type. Enzyme assays with whole-leaf extracts strongly suggested that NMT is expressed in cells within the Catharanthus leaf, a fact consistent with the biochemical localization of this enzyme within chloroplast thylakoids (Dethier and De Luca, 1993), raising the question of whether leaf mesophyll, idioblast or laticifer cells are involved this step of vindoline biosynthesis.
Flavonoid O-methyltransferase (FOMT), a soluble cytosolic enzyme, was also assayed to determine its cellular localization compared with 16-OH OMT and NMT. Catharanthus roseus has a few cloned and partially characterized FOMTs that catalyse O-methylation of flavonoids at different positions on the B-ring (Cacace et al., 2003; Schröder et al., 2004). FOMT assays showed that, while whole-leaf extracts actively O-methylated quercetin (Figure 4), this activity was not detected in epidermal cell extracts. The results show that while both enzymes are soluble OMTs, 16-OH OMT is expressed preferentially in the epidermis and FOMT, like the thylakoid membrane-associated NMT, is expressed within the Catharanthus leaf. Large-scale extractions of Catharanthus leaf epidermis using the CA technique, combined with various column chromatography techniques, have been used successfully in our laboratory to purify 16-OH OMT to homogeneity, free of contaminating FOMT (D.E. Levac and co-workers, unpublished data).
As the LCM-based RT–PCR results suggested that T16H was also preferentially expressed in leaf epidermis, a coupled assay described by St-Pierre and De Luca (1995) was performed to localize this enzyme activity. While T16H activity was detected in whole-leaf extracts, the extracts from epidermal cells failed to produce detectable reaction products. However, when assays were also performed with extracts from abraded leaves, only 23% of this activity could be recovered. In contrast, levels of NMT activity in abraded leaf extracts were unaffected by the treatment, compared with control leaf extracts. These results, in addition to the LCM studies (Figure 2), provide indirect evidence that T16H activity occurs in the leaf epidermis, and that the abrasive treatment may contribute to enzyme inactivation resulting in the lack of activity observed within the leaf epidermal extracts.
Expression analysis of MIA pathway genes in epidermis-enriched cell extracts and in whole leaves
After CA treatment, epidermal cells were extracted for total RNA together with whole leaves. These RNA extracts were subjected to RT–PCR, confirming that the abrasion procedure could also be used for RNA analysis of Catharanthus leaf epidermis (Figure 5). The results showed that the abaxial and adaxial epidermis of Catharanthus leaves did contain mRNAs for TDC, G10H, SLS, STR, SGD, T16H, D4H, DAT, ORCA3 and Actin (Figure 5), as RT–PCR products of the expected size were produced with each gene-specific pair of primers used. However, there was little or no indication that epidermis contained mRNA for the small subunit of Ribulose 1,5-bisphosphate carboxylase/oxygenase (RBCS), a marker for the mesophyll layer, as only the whole-leaf extract produced a strongly detectable RT–PCR product of RBCS. This result provided useful information about the specificity of the CA-extraction protocol used to isolate epidermis-specific RNA. Although the results obtained were more variable than those documenting the enzyme activities for various steps in MIA biosynthesis shown in Figure 4, there was consistency with those obtained with LCM extracts (Figure 2). These results provide important insights about the use of these tools for performing transcriptome analysis of the Catharanthus leaf epidermis, and suggest that the CA technique on live tissue, and LCM on fixed tissue, provide information different from, and complementary to, results obtained with in situ hybridization studies (Burlat et al., 2004; St-Pierre et al., 1999).
LCM-assisted RT–PCR suggests leaf epidermis is the preferred site of T16H and SGD expression
LCM is a technology recently developed to isolate single cells from tissues, and has been applied to plant cells in the past few years (Asano et al., 2002; Casson et al., 2005; Emmert-Buck et al., 1996; Kerk et al., 2003; Nakazono et al., 2003). The use of LCM has been limited to DNA and RNA isolation from particular cells and even more limited protein analyses (Martinet et al., 2004). Practical limitations of the technique have prevented its use for obtaining active enzymes, or for isolation of metabolites from particular cells. LCM has several advantages over conventional micropipetting methods (Kehr, 2003; Nakazono et al., 2003) that use glass microcapillaries to penetrate one cell type, such as epidermal cells, in order to reach cells in subepidermal layers, and that could make it quite difficult to eliminate contamination from non-targeted cells. In addition, micropipetting techniques could also cause unexpected wounding-related effects by damaging the tissue, and the slowness of the technique does not make it applicable to high-throughput processes. As shown here, as well as in the studies described above, the LCM technique has enabled the harvest of particular cells in any part of a tissue with much higher reproducibility, speed and accuracy. For the harvest of epidermal cells and entire layers of cells, 1000 cells were captured within 1 h, while for all other cell types 1000 cells were harvested in 2 h.
The RT–PCR results for monitoring the presence and cell type-specific location of TDC, STR, D4H and DAT (Figure 2) using RNA isolated from LCM-captured cells generally showed profiles consistent with those obtained using in situ RNA hybridization and immunoblots of histological leaf sections of Catharathus leaves (St-Pierre et al., 1999). That localization patterns for expression of these four markers are relatively consistent with previous results suggests that LCM-assisted RT–PCR can be a valuable tool for studies of the transcriptome of uncharacterized genes at the cellular level, especially in the case of C. roseus, where microarray analysis is not available. Moreover, RT–PCR products can be sequenced to verify the specificity of the system, which is an important advantage over in situ RNA hybridization in paraffin sections where this is not possible. Diagnostic sequencing of various PCR products for G10H, SGD, T16H and DAT exactly matched those of the published sequences for these genes. In this study, the novel detection of T16H mRNA expression in epidermal cells by LCM-assisted RT–PCR (Figure 2) was confirmed by sequencing of the PCR product that proved to be identical to T16H over 200 bp. This sequence information is important confirmatory evidence that authentic T16H was amplified, as this gene belongs to a very large superfamily of P450-dependent mono-oxygenases, as exemplified by Arabidopsis thaliana which is estimated to have 256 members. Assuming that Catharanthus cytochrome P450s are as abundant, it is easy to see how false PCR products could be produced. The problem of cross-reactivity did arise when in situ RNA hybridization was used to localize T16H in paraffin sections (Irmler et al., 2000; unpublished data) that detected expression in multiple cell types in Catharanthus leaves. Although qualitative results obtained by RT–PCR should be considered carefully when it comes to exact quantities of mRNA found, they do show that LCM-assisted RT–PCR is a powerful alternative tool to localize mRNAs that can complement in situ RNA hybridization techniques.
In addition to T16H, expression of both ORCA3 and SGD was also localized at the cellular level. The expression of ORCA3 mRNA was detected in all cell types tested in the Catharanthus leaf (Figure 2), including epidermal cells where TDC and STR mRNA are localized, as well as in laticifer and idioblast cells where D4H is expressed (Figure 2). This result is consistent with observations that ORCA3 may be partly responsible for the activation of TDC, STR and D4H mRNA expression (van der Fits and Memelink, 2000). The mRNA for SGD was also preferentially expressed in epidermal cells (Figure 2), suggesting that both strictosidine formation and the removal of the sugar moiety to form the aglycone precursor of all Catharanthus MIAs occur in the same cells.
The mRNAs for both G10H- and SLS-expressing enzymes involved in the biosynthesis of secologanin, the terpenoid moiety of MIAs, were also detected within epidermal and laticifer cells for the first time, as well as in the vascular cells (Figures 2 and 5) where this was shown for G10H previously by in situ RNA hybridization (Burlat et al., 2004). Epidermal cells may therefore also express the secologanin biosynthesis pathway in addition to cells in the vasculature (Burlat et al., 2004), to directly supply the isoprenoid precursors required for indole alkaloid biosynthesis, possibly making it unnecessary to import these precursors from the vasculature pool. It remains to be determined if the pool of secologanin being produced in the spatially separate vascular cells actually accumulate as a separate terpenoid pool. It is known that Catharanthus will also accumulate these secoiridoids (Hallard et al., 1998; Palacios-Rojas and Leech, 2004) in addition to monoterpenoid indole alkaloids, but little is known about how this is accomplished within the organism.
The CA technique provides metabolite, enzymatic and RNA-based evidence that epidermal cells are the major factory for MIA biosynthesis in leaves
The leaf-surface CA method is a classical technique that has been used to infect plants with viruses. The physical damage imposed on the leaf surface by carborundum particles is enough to help virus infect the tissue, while maintaining the rest of the leaf tissue intact. The CA technique has now been successfully applied for the selective extraction of alkaloids, active enzymes and mRNA from epidermal cells of fresh Catharanthus leaves.
It is well known that tabersonine and its biosynthesis have been associated mainly with cortical cells within the Catharanthus root tip (Laflamme et al., 2001), to supply precursors for the biosynthesis of aspidosperma alkaloids such as hörhammericine (Laflamme et al., 2001; Rodriguez et al., 2003). Alkaloid analysis of epidermal extracts showed that the level of tabersonine is about five times higher than that of vindoline or catharanthine (Table 1). The enrichment of tabersonine and 16-methoxytabersonine in leaf epidermal cells, together with the preferential expression of TDC, STRSGD and T16H mRNA in the leaf epidermis (Figures 2 and 5) and the detection of 16-OH OMT enzyme activity only in epidermal cells (Figure 4), is consistent with the suggestion that the pathway leading to and including 16-methoxytabersonine biosynthesis also occurs within the Catharanthus leaf epidermis, in addition to the presence of the tabersonine pathway in the root tip. The low level of catharanthine found in the leaf epidermis suggests that this tissue may be a site of neither its biosynthesis nor its accumulation. The ability of cell and root cultures to accumulate catharanthine is well documented, and raises the possibility that the vindoline and catharanthine components of dimeric MIAs are synthesized in separate cell types.
While the enzymatic reaction for converting 16-methoxytabersonine into 2,3-dihydro-16-methoxytabersonine remains to be characterized, biochemical assay of NMT clearly showed that this reaction occurs within the leaf, rather than in the epidermis of Catharanthus (Figure 4). This suggests that either 16-methoxytabersonine or its hydrated product is the alkaloid being transported from epidermis to cells (mesophyll, idioblast or laticifers) within the leaf for subsequent enzymatic modifications to produce vindoline (Figure 6). While the possibility remains that tabersonine made in the root tip is transported to leaves for conversion into vindoline, the data presented here suggest that the leaf epidermis must be involved to make part of the enzymatic conversion possible.
While the tabersonine pathway is present in the epidermis of Catharanthus leaves and in the tips of Catharanthus roots, the internal phloem of Catharanthus also appears to participate in the biosynthesis of a number of secondary metabolites including secoiridoids (Burlat et al., 2004); flavonoids (Kaltenbach et al., 1999); and toxic phytosterols (Fraenkel, 1959). Studies to localize flavonoid biosynthesis in young Catharanthus leaves (Kaltenbach et al., 1999) showed that the early part of the pathway (chalcone synthase) occurred in the epidermis, whereas a later step (flavonoid 3′,5′ hydroxylase, F3′5′H) was localized in the internal phloem. The presence of FOMT activity within Catharanthus leaves rather than in epidermal cells (Figure 4) is supported by the finding that F3′5′H was localized within the phloem of leaves (Kaltenbach et al., 1999). Although this does not provide clear evidence that both F3′5′H and FOMT occur within the same cell to catalyse sequential reactions, the correlation supports the suggestion that O-methylation of flavonoids probably does not take place in young Catharanthus epidermal leaf cells. Most importantly, this experiment clearly shows that 16-OH OMT and FOMT, both of which are soluble O-methyltransferases, were differentially extracted from Catharanthus leaves using the CA technique.
Proposed model for MIA and terpenoid biosynthesis involving different leaf cell types
The involvement of different cell types in the expression of alkaloid biosynthesis and accumulation observed here is not restricted to the C. roseus model system (De Luca and St-Pierre, 2000; Kutchan, 2005). The benzylisoquinoline alkaloids (BIAs) of opium poppy accumulate in laticifers, whereas the pathway for BIA biosynthesis also takes place in a different cell type associated with the vascular cells (Bird et al., 2003; Facchini and De Luca, 1995;Weid et al., 2004). In the case of tropane alkaloids, they are manufactured in more than one cell type in the root and then transported to the leaf for accumulation and storage (Nakajima and Hashimoto, 1999). The experimental model described in Figure 6 combines our results with previous studies to suggest that C. roseus leaves have separate cellular sites for the biosynthesis and accumulation of MIAs and terpenoids, respectively. This colour-enhanced paradermal section of a young Catharanthus leaf (Figure 6) shows that the adaxial and abaxial epidermal (blue) and vascular (pink) cells are spatially separated and there is no direct physical contact between them. Laticifer cells are associated with leaf tracheids, while palisade-assisted idioblast cells are closely associated with abaxial epidermal cells (St-Pierre et al., 1999) and are coloured yellow for visualization. The spongy mesophyll-assisted idioblast cells could not be identified in paradermal sections as they are not distinguishable from adjacent mesophyll cells. The model suggests that vascular cells are specialized for terpene biosynthesis where genes encoding enzymes involved in the MEP pathway, together with the monoterpene pathway, produce secologanin, other terpenoids and perhaps even O-methylated flavonoids that accumulate in Catharanthus leaves. It is not clear, however, if secologanin made in the vascular cells can be mobilized to the leaf epidermal MIA cell factories for incorporation into MIAs. The abaxial and adaxial leaf epidermal cells make tryptamine from tryptophan, in addition to expressing the entire pathway for secologanin biosynthesis for incorporation into 16-methoxytabersonine, as the rest of this pathway is also expressed. An intermediate alkaloid (16-methoxytabersonine or the 2,3-dihydro derivative) is transported from epidermal cells into the leaf where the last three reactions would take place in chloroplast thylakoids (Dethier and De Luca, 1993) and/or idioblast/laticifers (St-Pierre et al., 1999) to elaborate the vindoline molecule. Biochemical characterization of the enzyme reaction that converts 16-methoxytabersonine into the 2,3-dihydro derivative will be required to determine which cell type makes this intermediate, in order to identify correctly which of these two MIAs is transported from the leaf epidermis for further elaboration within the leaf to complete the vindoline pathway.
Tissue fixation and embedding
Young C. roseus (L.) G. Don. (Little Delicata) leaves (1.5 cm long) were processed for LCM as described by Kerk et al. (2003), with some modifications. The leaf were infiltrated in vacuo in 10 ml volume per 1 g fresh weight in Farmer's fixative (3:1 ethanol:acetic acid) for 30 min at room temperature (21°C). The leaves were then transferred to fresh fixative at 4°C for 16 h, after which they were dehydrated at room temperature by sequential transfer to a series of ethanol concentrations [3 h each (v/v) 75, 85, 100, 100 and 100%], followed by a series of ethanol plus xylene concentrations [3 h each (v/v) 75:25, 50:50, 25:75, 0:100, 0:100 and 0:100%]. Liquified Paraplast X-Tra (Fisher Scientific, Nepean, Canada) was added in the final step at room temperature, and the media exchanged twice at 58°C. Ten to 15 leaves were stacked together in the same orientation to achieve higher throughput, and positioned in liquified Paraplast-X-Tra solidified at 4°C. Solid blocks could be stored at 4°C for several months until they were required. Sections were cut on a rotary microtome (Reichert Jung, Depew, NY, USA) to 10 μm thickness, floated in a warm water bath at 42°C, stretched on glass slides, air-dried, and stored in darkness at 4°C under dehydrating conditions. Just before use for LCM, slides were deparaffinized with two xylene treatments for 5 min each, and air-dried.
Laser-capture microdissection and RNA extraction
The procedure used for LCM was basically as described by Kerk et al. (2003) with some modifications. Longitudinal leaf sections were used for harvesting epidermal cells, and cross-sections for targeting vascular cells. Paradermal sections were used to isolate idioblast, laticifer and mesophyll cells. In all cases, LCM was conducted using the Pixcell II system (Arcturus, Mountain View, CA, USA). For standard harvesting conditions, the spot size, power and duration of the laser shots for capturing cells were 7.5 μm, 55 mW and 800 μsec, respectively. Approximately 1000 cells could be harvested onto a single plastic cap from five to 10 slides that contained 30–45 histological sections each. In total, 2000–5000 cells from each cell type were collected and pooled together after non-specific tissue that adhered to the cap was selectively removed using a Post-It note (Kerk et al., 2003). RNA was extracted using the PicoPure RNA Isolation Kit (Arcturus) according to the manufacturer's protocol. The obtained RNA was treated with RNase-free Dnase I Set (Qiagen, Mississauga, Canada) prior to the elution step from the spin column to give 11 μl eluate. The RNA was quantified with RiboGreen RNA Quantitation Reagent (Invitrogen, Burlington, Canada) using a 1- μl sample from each cell type. The RNA samples were kept at −85°C until use.
RNA amplification was carried out according to Nakazono et al. (2003) with some modifications, as follows: Superscript III (200 U μl−1; Invitrogen) was used in place of Superscript II (200 U μl−1; Invitrogen), for reverse transcription. Linear acrylamide (Ambion, Austin, TX, USA) was used as a co-precipitant for ethanol precipitation of RNA. The synthesized double-stranded DNA was then used for in vitro transcription using the MEGAscript T7 kit (Ambion) following the manufacturer's protocol. The amplified RNA was purified using the RNeasy Mini Kit (Qiagen), following elution in 10 μl diethylpyrocarbonate DEPC-treated water. The second-round RNA amplification was performed basically as described above, except that 1 μg μl−1 of random hexamers (Amersham Biosciences, Piscataway, NJ, USA) was used for first-strand synthesis in place of T7-oligo(dT) primer.
mRNA levels of genes in vindoline biosynthesis were analysed by RT–PCR using gene-specific oligonucleotides as follows: CrG10H-RT01 5′-GGTAGCCTCACGATGGAGAA-3′, CrG10H-RT02 5′-CCTTGGCAGAATCCGAATAA-3′, CrSLS-RT01 5′-CTTTGAGGGTGCAAAATGGT-3′, CrSLS-RT02 5′-TGGGATCCTTGTTTTTCAGC-3′, CrTDC-RT01 5′-CGCCTGTATATGTCCCGAGT-3′, CrTDC-RT02 5′-GTTGCGATTTGCCAATTTTT-3′, CrSTR-RT01 5′-ACCATTGTGTGGGAGGACAT-3′, CrSTR-RT02 5′-CCATTTGAATGGCACTCCTT-3′, CrSGD-RT01 5′-ATTTGCACCAGGAAGAGGTG-3′, CrSGD-RT02 5′-TATGAACCATCCGAGCATGA-3′, CrT16H-RT01 5′-GCTTCATCCACCAGTTCCAT-3′, CrT16H-RT02 5′-CCGGACATATCCTTCTTCCA-3′, CrD4H-RT01 5′-TTGACATTTGGGACAAGCAA-3′, CrD4H-RT02 5′-CCAAAAGCAACAGCAACAGA-3′, CrDAT-RT01 5′-GTGCGTATCCGTTGGTTTCT-3′, CrDAT-RT02 5′-CGAACTCAATTCCATCGTCA-3′, ORCA3-RT01 5′-CGGGATCCGAAATACAGAA-3′, ORCA3-RT02 5′-GCCCTTATACCGGTTCCAAT-3′, CrRBCS-RT01 5′-TCTTCAATGATTTCCACGGC-3′, CrRBCS-RT02 5′-TTCTTCTTTCCCAATGGTGG-3′, CrActin-RT01 5′-GGCTGGATTTGCTGGAGATGAT-3′, CrActin-RT02 5′-TAGATCCTCCGATCCAGACACTG-3′. The oligonucleotides for RBCS and Actin of C. roseus were designed by the authors (unpublished data). Reverse transcription was performed using RNA PCR Kit (AMV) ver. 2.1 (Takara, Otsu, Japan) according to the manufacturer's protocol. The PCR reaction was carried out for 35 cycles of 15 sec at 94°C, 20 sec at 57°C and 30 sec at 72°C. ExTaq DNA polymerase (Takara) was used in place of rTaq, which was included in the kit. By testing every three cycles between 26 and 38 cycles, the program for PCR reaction was optimized considering the linearity of the amount of PCR products. Amplified cDNA fragments were run on 1.5% agarose gel and visualized by ethidium bromide staining. PCR products for T16H, SGD, G10H and DAT in Figure 2 were sublcloned to pGEM-T easy (Promega, Madison, WI, USA) and sequenced using T7 promoter primer at the John P. Robarts Research Institute DNA sequencing facility (London, Canada) for verification.
Scanning electron microscopy
Images were obtained using an AMRAY 1600 Turbo SEM. Specimens were mounted onto a carbon adhesive tab and silver paint was applied to the specimen edges to aid in sample conductivity. Using a secondary electron scintillation detector and 15 kV accelerating voltage, images were processed using orion digital image-grabbing software ver. 6.51. Calibrated μm-scaled bars are incorporated in the CRT output of the AMRAY SEM.
Chlorophyll extraction and quantification
Chlorophyll extraction was performed as follows: 200 μl of either leaf epidermal extract or whole-leaf extract was mixed with 1.8 ml acetone and incubated for 30 min at −20°C. The solution was centrifuged at 21000 g for 20 min, and the absorbance A663/A750 of the supernatant was determined by spectrophotometer.
Carborundum abrasion technique
The upper or lower epidermis from 2 g young 1.5-cm-long leaves was abraded with carborundum number F (Fisher Scientific) using a cotton swab to apply even pressure to damage the leaf surface. The epidermis was rubbed four times per leaf and then dipped in either 3.0 ml protein extraction buffer (100 mm Tris–HCl pH 8.0, 13 mmβ-mercaptoethanol) at 4°C, 3.0 ml RNA extraction buffer (100 mm Tris–HCl pH 8.0, 100 mmβ-mercaptoethanol) at 4°C, or 10 ml alkaloid extraction solution (50% MeOH) at room temperature, depending on the application. Each abraded leaf was gently agitated to produce the crude epidermal cell extract. Typically, 2.5 ml extract was obtained after 2 g abraded leaf was individually dipped. The degree of damage caused by the procedure was monitored mainly by measuring the chlorophyll content in the epidermal extract, but also by microscope analysis (Figure 3).
RNA extraction from leaf epidermis samples
RNA samples obtained by the CA technique were vortexed for 5 min together with carborundum particles. Samples were then centrifuged for 5 min at 1000 g in 1.5-ml tubes at room temperature to remove the carborundum. The supernatant was transferred to new 1.5-ml tubes for ethanol precipitation. Linear acrylamide was used as a co-precipitant. The pellet was dissolved in 50 μl 100 mm Tris–HCl pH 8.0, and the RNA was treated with 10 U DNase I FPLCpure (GE Healthcare, Piscataway, NJ, USA) for 30 min at 37°C followed by phenol/chloroform extraction and concentration by ethanol precipitation. RNA was quantified by spectrophotometry and used for RT–PCR. Typically, 2–3 μg RNA was obtained from the surface of 2 g fresh leaf compared with the 500 μg obtained from entire leaf extracts.
Preparation of crude protein extract
Leaf epidermis protein extract was centrifuged at 500 g for 5 min at 4°C and desalted by PD-10 column (GE Healthcare) according to the manufacturer's protocol. For whole-leaf protein extraction, 2.0 g fresh weight of young leaves was homogenized in 3 ml protein extraction buffer using a mortar and pestle. Typically, 8.75 mg and 175 μg protein was obtained from whole leaf and leaf surface, respectively, per g fresh weight leaf. The sample was filtered through 20 μm Nylon mesh, and the filtrate was centrifuged at 500 g for 5 min at 4°C. The protein was eluted with 3.5 ml protein extraction buffer and used directly for enzyme activity assays. The protein concentration was determined using a Protein assay kit (Bio-Rad, Hercules, CA, USA).
Leaf epidermis extract in 10 ml 50% methanol was incubated for 2 h at 50°C. Alternatively, for alkaloid extraction from whole leaf or abraded leaf body, 2 g tissue was homogenized in 10 ml 50% methanol using a mortar and pestle and incubated for 2 h at 50°C, followed by filtration through 20-μm Nylon mesh. The extract was centrifuged at 1000 g for 5 min to remove residual carborundum particles, and the supernatant was evaporated to remove methanol. The aqueous phase (5 ml) was acidified with 1.25 ml 10% sulphuric acid, and extracted three times with an equal volume of ethyl acetate that was discarded. The aqueous phase was then adjusted to pH 9–10 with 1.25 ml 10 m NaOH, and extracted three times, each with equal volume of ethyl acetate. The organic phase was evaporated to dryness using SPD SpeedVac (Thermo Savant, Holbrook, NY, USA) and the residue was resuspended in 0.2 ml MeOH.
Alkaloid extracts were analysed by HPLC basically as described by Tikhomiroff and Jolicoeur (2002). Briefly, the solvent system was composed of 5 mm sodium phosphate pH 6.0 (solvent A) and acetonitrile (solvent B), both of which were pre-filtered through a 0.45-μm White Nylon filter (Millipore, Billerica, MA, USA). The ratio of solvent A:B was: 0–20 min, linear gradient from 80:20 to 20:80; 20–25 min, isocratic with 20:80; 25–30 min, linear gradient from 20:80 to 80:20. Samples were run on an Inertsil ODS-3 (GL Sciences, Tokyo, Japan) reverse-phase C18 column (4 × 250 mm) with a 3 × 4-mm guard column (Phenomenex, Torrance, CA, USA) at a flow rate of 2 ml min−1 and analysed using a 2996 Photodiode Array Detector (Waters, Milford, MA, USA) and empower pro software (Waters) with UV detection of peaks at either 220 nm for vindoline and catharanthine or 330 nm for tabersonine. Pure vindoline, catharanthine and tabersonine extracted from Catharanthus leaves were available as references.
Production of 16-OH tabersonine
Prior to 16-OH OMT activity assay, we performed bioconversion of tabersonine into 16-OH tabersonine to obtain direct substrate for this enzyme. Escherichia coli strain RM82 containing plasmid pUBS520 was further transformed with the plasmid harbouring T16H/P450 reductase fusion protein (Schröder et al., 1999). Luria–Bertani media (4 × 50 ml) was inoculated with 0.5 ml saturated culture cells for each flask, and cells were grown for 3 h at 37°C. Tabersonine (0.1 mm) was added as a substrate, and fusion protein expression was induced by adding 3 mm IPTG for 16 h at 25°C. Cells were centrifuged at 3000 g for 10 min to produce a pellet, and the alkaloid extracted directly from the supernatant as described above, except that the Rotavapor R-205 (Büchi, Zurich, Switzerland) was used to concentrate samples in place of the SpeedVac system. The conversion rate of tabersonine to 16-OH tabersonine was close to 100%, which was estimated by the disappearance of tabersonine and appearance of 16-OH tabersonine by TLC. The spot for 16-OH tabersonine was scraped off from the TLC plate, and the silica powder was vortexed with 500 μl ethylacetate in a 1.5-ml tube. After centrifugation the solvent was transferred to a new 1.5-ml tube and dried to dryness using SPD SpeedVac (Thermo Savant). The residue was resuspended in 200 μl methanol. The authenticity of 16-OH tabersonine was confirmed by HPLC and mass spectrometry as described previously (Schröder et al., 1999).
Enzyme activity assays
TDC, T16H, 16-OH OMT, NMT and DAT enzyme activities were assayed basically as described previously (De Luca et al., 1986, 1987, 1989; St-Pierre and De Luca, 1995;St-Pierre et al., 1998). All enzyme assays were performed in 200 μl reaction volumes in protein-extraction buffer. Briefly, desalted crude protein extract from either leaf epidermal cells or whole leaf was incubated with various substrates and co-factors depending on the enzyme assay. The standard assay condition was 60 min incubation at 37°C, except for the DAT assay which was incubated for only 15 min to obtain quantitative results, as shown previously (St-Pierre et al., 1998). The substrates and co-factors used for each enzyme assay were as follows: TDC assay [20.8 μm (0.1 μCi) l-Tryptophan (side chain-3-14C), Moravek, Brea, CA, USA]; 16-OH OMT assay [approximately 30 μm 16-OH tabersonine and 8.33 μm (0.1 μCi) S-adenosyl-l-(methyl-14C)Met, GE Healthcare, Buckinghamshire, UK]; NMT assay [30 μm 2,3-dihydro-3-hydroxy-tabersonine and 8.33 μmS-adenosyl-l-(methyl-14C)Met]; DAT assay [30 μm deacetylvindoline and 16.6 μm (1-14C)Acetyl-coenzyme A, GE Healthcare]. In contrast to the others, T16H assay is a coupled assay that detects the production of 16-methoxy tabersonine from tabersonine by coupling it with endogenous 16-OH OMT activity, as described previously (St-Pierre et al., 1995). The crude protein extract was incubated with 30 μm tabersonine, 8.33 μmS-adenosyl-l-(methyl-14C)Met and 1 mm NADPH that would produce 16-methoxytabersonine. The FOMT assay was performed using 30 μm quercetin as a substrate and 18 μmS-adenosyl-l-(methyl-14C)Met as a methyl group donor. Each of the above reactions was terminated by adding 20 μl 10 m NaOH with the exception of the FOMT assay, which was acidified by 100 μl 2.5% sulphuric acid. The products were extracted three times with 400 μl ethyl acetate, and the organic phase was taken to dryness by vacuum centrifugation in the SpeedVac system. Each sample was dissolved in 5 μl methanol and applied to TLC analyses. Polygram Polyamid-6 (Macherey-Nagel, Düren, Germany) was used for FOMT activity assay, and Polygram Sil G/UV254 (Macherey-Nagel) was used for all the other enzyme activity assays. TLC plates were developed in various solvent systems depending on the assay; T16H and 16-OH OMT [diethylether–hexane (1:1, v/v)]; NMT and DAT [ethylacetate–methanol (9:1, v/v)]; TDC [chloroform–methanol–25% ammonium hydroxide (5:4:1, v/v)]; FOMT [toluene–methyl ethyl ketone–methanol (4:3:3, v/v)]. The radioactivity was visualized and quantified by exposure of TLC to a storage phosphor screen (GE Healthcare, Piscataway, NJ, USA) for 16 h and emissions detected using a Phosphorimager FLA-3000 (Fujifilm, Tokyo, Japan) and multi gauge ver. 3.0 (Fujifilm).
We would like to thank D. Bienzle for allowing us access to the LCM facility, and D. Hale, M. Harron, S. Lapos and S. Tatarski for training J.M. and their kind help in preparing histological sections at the University of Guelph, Canada. The plasmid carrying T16H/P450 Reductase fusion protein was a generous gift from J. Schröder (Universität Freiburg, Germany). We also would like to thank G. Hooper (Brock University, Canada) for her photos of SEM images. V.D.L. holds a Canada Research Chair in Plant Biotechnology. This work was supported by a grant from the Natural Sciences and Engineering Research Council of Canada to V.D.L.