Application of carborundum abrasion for investigating the leaf epidermis: molecular cloning of Catharanthus roseus 16-hydroxytabersonine-16-O-methyltransferase

Authors


* (fax +1 905 688 5550; e-mail vdeluca@brocku.ca).

Summary

The Madagascar periwinkle (Catharanthus roseus) produces the well-known and remarkably complex anti-cancer dimeric alkaloids vinblastine and vincristine that are derived from the coupling of vindoline and catharanthine monomers. This study describes the novel application of a carborundum abrasion (CA) technique for large-scale isolation of leaf epidermis-enriched proteins in order to purify to apparent homogeneity 16-hydroxytabersonine-16-O-methyltransferase (16OMT), which catalyses the second of six steps in the conversion of tabersonine into vindoline, and to clone the gene. Functional expression and biochemical characterization of recombinant 16OMT demonstrated its very narrow substrate specificity and high affinity for 16-hydroxytabersonine. In addition to allowing the cloning of this gene, the CA technique clearly showed that 16OMT is predominantly expressed in Catharanthus leaf epidermis. The results provide compelling evidence that most of the pathway for vindoline biosynthesis, including the O-methylation of 16-hydroxytabersonine, occurs exclusively in the leaf epidermis, with subsequent steps occurring in other leaf cell types.

Introduction

Catharanthus roseus (Madagascar periwinkle) produces a variety of biologically active monoterpene indole alkaloids (MIAs), including the anti-hypertensive corynanthe alkaloid ajmalicine (Tikhomiroff and Jolicoeur, 2002), that accumulates to 0.57 mg g−1 DW in Catharanthus hairy root cultures (Vázquez-Flota et al., 1994a; b). In contrast, the dimeric bisindole alkaloids vinblastine (VBL) and vincristine (VCR), both of which are used to treat Hodgkin’s disease and childhood lymphomas, accumulated to 46.6 μg g−1 DW and only trace amounts, respectively, in above-ground parts of C. roseus (Choi et al., 2002). The economic and medical importance of VBL and VCR, as well as their chemical complexity, has prompted extensive research efforts to facilitate their production from the plant since their discovery over 45 years ago. In fact, the recently developed dimeric MIA drug known as vinflunine (Javlor) (Fahy et al., 1997) produced by Pierre Fabre Medicament (http://www.plantes-industrie.com) still requires the plant for supply of the MIA precursors vindoline and catharanthine for its production. Vinflunine is in phase III clinical trials due to its increased effectiveness and lower toxicity in the treatment of a broader range of human cancers.

VBL and VCR are produced as a result of a condensation of the Iboga alkaloid catharanthine and the aspidosperma alkaloid vindoline. While catharanthine-accumulating cell suspension cultures have been successfully produced (Moreno et al., 1995), cultures that make vindoline have never been reported. The complex regulation of vindoline biosynthesis, including the requirement for light and the involvement of several cell types in its production (St Pierre et al., 1999), could not be reproduced in cell culture systems. More recently, it has been suggested that the many of the genes responsible for MIA biosynthesis up to 16-methoxytabersonine (Figure 1) may in fact be expressed and localized to the leaf epidermis of Catharanthus (Murata and De Luca, 2005), whereas the last four steps required to make vindoline may occur in specialized idioblast and laticifer cell types within the leaf (St Pierre et al., 1999).

Figure 1.

 The biosynthesis of vindoline from tryptamine and secologanin.
Tryptamine is derived from tryptophan via the action of tryptophan decarboxylase (TDC), and strictosidine synthase (STR1) converts tryptamine and secologanin into the central intermediate strictosidine. Strictosidine is converted into 16-methoxytabersonine through a series of reactions, mostly uncharacterized but including tabersonine 16-hydroxylase (T16H) and 16-hydroxytabersonine-16-O-methyltransferase (16-OMT). The enzyme assay for T16H uses [14-CH3]-S-adenosyl-l-methionine ([14-CH3]-AdoMet), and radiolabeled 16-methoxytabersonine can be detected. The biosynthesis of vindoline involves three more enzymatic steps to form deacetylvindoline, which is converted to vindoline by the action of deacetylvindoline-4-O-acetyltransferase.

It is well known that most plant MIAs are derived from tryptamine and secologanin to produce the central intermediate, strictosidine, from which is derived tabersonine through a number of uncharacterized enzymatic steps (Figure 1). The enzyme tabersonine 16-hydroxylase converts tabersonine into 16-hydroxytabersonine, and the second of six steps involved in the conversion of tabersonine into vindoline is catalyzed by 16-hydroxytabersonine-16-O-methyltransferase (16OMT) (Figure 1). Significant efforts using homology-based methods (Cacacea et al., 2003; Schröder et al., 2002, 2004) have failed to clone the 16OMT, but several full-length flavonoid-O-methyltransferases (FOMTs) were successfully obtained, including a flavonoid/anthocyanin OMT that can perform sequential 3′,5′O-methylation of flavonoids and anthocyanins that accumulate in C. roseus. While additional efforts to clone this gene involved the successful purification to homogeneity of 16OMT from C. roseus cell cultures, the approach only succeeded in cloning yet another flavonoid 4′-OMT that was functionally characterized (Schröder et al., 2004). Both of these 16OMT cloning efforts used plant cell cultures activated for MIA biosynthesis as a source of enzymes. Although such induced cell cultures may be a cost-effective and convenient source of enzymes, these results suggest that they are not always suitable for selective purification of enzymes normally expressed in plant leaves.

The present study describes the convenient use of a carborundum abrasion (CA) technique (Murata and De Luca, 2005) as a tool for large-scale isolation of leaf epidermis-enriched proteins and for purification of 16OMT to apparent homogeneity. This versatile tool was also used to harvest leaf epidermis-enriched mRNAs, which facilitated the molecular cloning of 16OMT. Additionally, the functional expression and biochemical characterization of the recombinant 16OMT are reported, as well as its preferred localization and expression in the leaf epidermis compared to other OMTs expressed in other Catharanthus tissues.

Results

MIA pathway enzyme activity profiling throughout seedling development, and in leaves, flowers, stems and roots of C. roseus

Seedlings were prepared, harvested and assayed for tryptophan decarboxylase (TDC), 16OMT and deacetylvindoline 4-O-acetyltransferase (DAT) activity, as previously reported (De Luca et al., 1988). TDC enzyme activity increased in 4-day-old dark-grown seedlings until 8 days of growth, and decreased thereafter (Figure 2a). Light treatment did not affect the TDC profile, but DAT activity was increased several-fold by light treatment compared to dark-grown control seedlings (Figure 2a,b), confirming previous results describing the light induction of DAT activity in etiolated seedlings (De Luca et al., 1988; St Pierre et al., 1998). 16OMT activity first appeared in 2-day-old dark-grown seedlings and increased to a maximum after 8 days of growth (Figure 2a). Unlike the effect of light on DAT induction, 16OMT activity was only increased by 30% compared to dark-grown controls (Figure 2a,b), displaying a very similar intermediate expression pattern to that observed for an N-methyltransferase responsible for the 3rd to last step in vindoline biosynthesis (Dethier and De Luca, 1993; De Luca and Cutler, 1987; De Luca et al., 1988, 1987).

Figure 2.

 Distribution of 16OMT enzyme activities in Catharanthus roseus seedlings, organs and cells.
(a, b) Regulated appearance of TDC, 16OMT and DAT enzyme activities during 10 days of growth of germinating seedlings in the dark (a) or the light (b).
(c) Comparison of 16OMT activities in leaves of various ages from the youngest 1st leaf pair to the oldest 3rd leaf pair.
(d) Comparison of 16OMT and FOMT enzyme activities in various plant organs.

The distribution of 16OMT enzyme activities was determined in the 1st, 2nd and 3rd pair of leaves from the leaf apical meristem, which represent various stages of leaf development. The specific activity of 16OMT was highest in crude extracts of the youngest leaf pair, and this activity diminished with developmental age (Figure 2c). The distribution of 16OMT and FOMT enzyme activities was also determined in crude extracts of leaves, flowers, stems and roots (Figure 2d). While 16OMT enzyme activities were highest in flowers and leaves, those of FOMT were highest in flowers and roots. The relative 16OMT and FOMT enzyme activities in various Catharanthus tissues may reflect the various biochemical roles played by these enzymes in each tissue, as well as their relative tissue-specific distribution.

Purification of C. roseus 16OMT to homogeneity, and molecular cloning from a cDNA library prepared from leaf epidermis-enriched mRNA

Young Catharanthus leaves (128 g) were extracted by the CA technique (Supplementary Figure S1) to selectively extract approximately 4 mg of crude leaf epidermis-enriched protein, compared to the much higher levels of protein (640 mg) that would be obtained from extraction of 128 g of whole leaves (Table 1). After concentrating the extract by ultrafiltration, it was subjected to Sephadex G150 size-exclusion chromatography as described in Experimental procedures. Active fractions with 16OMT activity were not active against FOMT substrate, and were further purified by adenosine agarose affinity chromatography and Mono Q anion exchange column chromatography (Table 1). In contrast, whole-leaf extracts contained residual FOMT activity after size exclusion and affinity chromatography, but it completely disappeared during the Mono Q anion exchange chromatography step (Table 1). The Mono Q step completed the purification of CA-extracted protein to produce a 1.29 × 104-fold enrichment of 16OMT compared to crude extracts [Supplementary Figure S2, see SDS–PAGE protein profiles of whole-leaf crude extracts (W) and carborundum leaf epidermis-enriched extracts (CA)], to yield a total of 4 μg of a purified 40 kDa protein (Table 1 and Supplementary Figure S2), as determined by SDS–PAGE and silver staining (Supplementary Figure S2, fractions 21–23). In contrast, purification of this enzyme from whole leaves produced a partially purified preparation containing several proteins, which did not include the predominant 40 kDa protein purified from CA leaf epidermis-enriched extracts (data not shown). Fractions containing the purified 16OMT were pooled, concentrated and subjected to SDS–PAGE to harvest a 40 kDa protein that was then submitted for sequencing.

Table 1.   Purification of 16OMT compared to FOMT using the carborundum abrasion technique (CA) compared to extraction of enzymes from whole leaves*
 StepProtein (mg)Total activity (pkat)Specific activity (pkat/mg protein)Ratio (16OMT/FOMT)Purification (fold)Yield (%)
(16OMT)FOMT(16OMT)FOMT
  1. *CA epidermis 16OMT was purified from 128 g FW C. roseus leaf, and whole-leaf 16OMT was purified from 6 g FW C. roseus leaf.

  2. ND, not detected; NA, not applicable.

CAConcentrated crude extract3.93.960.781.10.26.81100
Sephadex G1501.14.4ND4.0NDNA3.6112
Adenosine–agarose0.032.41ND80NDNA7360
Mono Q0.00045.56ND1.41 E4NDNA1.29 E4140
Whole leafConcentrated crude extract29.80.891.190.030.040.841100
Sephadex G1505.41.080.160.200.0366.7121
Adenosine–agarose2.013.020.121.50.062750339
Mono Q0.00120.26ND216.2NDNA7.2 E329

Seven peptides with 100% sequence identity to various published Catharanthus OMTs were obtained, and protein BLAST analysis of the sequenced peptides demonstrated that two peptides (peptide 1, WILHDWNDEDCVK; peptide 2, GIVLTMLDPAELK) had 100% sequence identity to CrOMT6 (AAR02420), four peptides (peptide 3, MVLHDWNDEDCVK; peptide 4, NEDGTAFETAHGK; peptide 5, IPPAHVVFLK; peptide 6, EAGFSSYK) had 100% sequence identity to CrOMT7 (AAR02421), and one peptide (peptide 7, CTVFDLPHVVANLESK) had 100% sequence identity to CROMT2 (AAM09497), CrOMT5 (AAR02417) and CrOMT6.

The high amino acid sequence identities between various Catharanthus OMTs made it difficult to design 16OMT gene-specific cloning primers (Supplementary Figure S3), and this may explain one reason for the failure of previous efforts to clone this gene (Cacacea et al., 2003; Schröder et al., 2004). This issue was solved by designing three sets of primers based on peptides (Figure S3) derived from in CrOMT6 (primers F1 and R1), CrOMT7 alone (primers F2 and R2) or based on peptides derived from CrOMT6 (primer F3) and CrOMT7 (primers R3) (Figure 3). Both the CrOMT6 and CrOMT7 sets of primers only produced PCR products in preparations from whole leaves but not in epidermis-enriched preparations (Figure 2, inset). In contrast, PCR with a gene-specific forward primer based on CrOMT7 (Supplementary Figure S2, peptide 4) and a reverse primer based on CrOMT6 (Supplementary Figure S2, peptide 1) produced a unique 361 bp product (Figure 3, inset) from cDNA libraries derived from mRNA isolated from whole leaves and from CA leaf epidermis-enriched extracts, which corresponded to a putative 16OMT gene. Identical PCR products were obtained from both cDNA libraries as shown by DNA sequencing (data not shown), and they were 87% identical to a putative uncharacterized CrOMT5, and 83% identical to CrOMT6, which appears to catalyze the 4′-O-methylation of the flavanone eriotictyol, of the flavonols isorhamnetin, quercetin and kaempferol, and of the flavone chrysoeriol (Schröder et al., 2004). New unique sequence-specific primers based on this novel sequence were designed to amplify the remainder of the clone by 5′ and 3′ RACE using a leaf epidermis-enriched cDNA library (see Experimental procedures), producing a 1323 bp fragment encoding a 1068 bp ORF for a putative 355 amino acid OMT with a theoretical mass of 39.8 kDa (Supplementary Figure S3) (Genbank accession number EF444544). This novel protein, with significant sequence identity to CrOMT6 (64%) and CrOMT5 (66%) (Supplementary Table S1), possesses a highly conserved C-terminal S-adenosyl-l-methionine (AdoMet) catalytic binding domain characterized by a central α/β Rossman fold, as well as an N-terminal domain responsible for dimerization and for formation of the back wall of the catalytic site that is required for substrate recognition (Zubieta et al., 2001).

Figure 3.

 Alignment of peptide sequences 1–7 in relation to the putative open reading frame of 16OMT, and identification of a unique PCR product that is preferentially expressed in leaf epidermal cells compared to whole leaf.
The following primer sets CrOMT6 (515-735 bp) (F2, 5′-CCATGGCTAATGACTCTG-3′; R2, 5′-GTCTCCTCCAACAAACTC-3′); CrOMT7 (348-716 bp) (F1, 5′-TCCAATGCTAGATCCACTTC-3′; R1, 5′-CCACCAACAAACTCCAAGT-3′), and CrOMT7/CrOMT6 (425-786 bp of 16OMT clone) (F3, 5′-CTGCTTTTGAAACAGCTCATGG-3′; R3, 5′-CAGTCATGGAGAATCCACTT-3′) were used to produce the PCR products shown in the inset using cDNA libraries derived from mRNA isolated from CA leaf epidermal cells and from whole leaves.

Expression and functional characterization of recombinant 16OMT (r16OMT)

Biotransformation experiments using Escherichia coli transformed with an r16OMT-containing vector completely converted exogenously supplied 16-hydroxytabersonine into the expected 16-O-methylated derivative, whereas induction with IPTG prevented this conversion from occurring (data not shown). Control E. coli cultures expressing the empty vector were unable to catalyze the biotransformation as only the original 16-hydroxytabersonine was recovered.

The ability of E. coli cultures expressing r16OMT to produce the 16-O-methylated product prompted efforts to directly characterize the enzyme. Cell-free extracts from whole C. roseus leaves (Figure 4a, lane 1, and Figure 4b, lane 2), as well as from E. coli transformed with r16OMT-containing vector (Figure 4a, lane 2, and Figure 4b, lane 2), catalyzed the enzymatic O-methylation of 16-hydroxytabersonine in the presence of an AdoMet methyl group donor. In contrast, IPTG-treated soluble (Figure 4a, lane 3, and Figure 4b, lane 3) and inclusion body (Figure 4a, lane 4, and Figure 4b, lane 4) extracts from E. coli expressing r16OMT were not active, nor were extracts from untransformed E. coli cells before (Figure 4a, lane 5, and Figure 4b, lane 5) or after (Figure 4a, lane 6, and Figure 4b, lane 6) IPTG induction. Inclusion bodies harvested from a 50 ml E. coli culture were dissolved in urea containing buffer and protein was renatured by dialysis (see Experimental procedures) to produce approximately 250–400 mg of highly purified r16OMT (Figure 4b, lane 4*).

Figure 4.

 Representation of SDS-PAGE protein profiles and 160MT activity in different plant and E. coli protein extracts. In the SDS-PAGE gel lanes represent standard protein molecular weight markers (lane M); whole-leaf extract (lane 1); soluble proteins from E. coli expressing 160MT before adding IPTG (lane 2); soluble proteins from E. coli expressing 160MT after adding IPTG (lane 3); inclusion body proteins found in the pellet after centrifugation from E. coli expressing 160MT after adding IPTG (lane 4); soluble proteins from E. coli expressing empty vector before adding IPTG (lane 5); soluble proteins from E. coli expressing empty vector after adding IPTG (lane 6). Enzyme assays were performed in the presence of each protein extract (represented by lanes 1 to 6 well as inclusion bodies after solubilization (4*) as described in Materials and methods), S-(14CH3)-adenosyl-L-methionine and with 16 hydroxytabersonine substrate (+S) or without it (−S). In addition, the same enzyme assays were performed with each boiled (B) protein extract with (+) the (14CH3)-adenosyl-L-methionine and 16-hydroxytabersonine substrate. The 16 [14C]-methoxytabersonine reaction product was visualized by thin layer chromatography and autoradiography. Reaction products were only observed in the presence of 16-hydroxytabersonine (+) with extracts 1, 2 and 5, while the other extracts were not active (−).

Recombinant 16OMT is a highly specific O-methyltransferase

The substrate specificity of r16OMT was assayed by determining its ability to use a range of substrates including MIAs, flavonoids and aromatic amines. When assayed between pH 5.0–9.0, 16-hydroxytabersonine, but not tabersonine (Figure 5), was converted to the 16-O-methylated product, with a pH optimum of 7.0–7.5 (data not shown). Aromatic and indole amines such as 3-hydroxytyramine, 4-hydroxytyramine and 5-hydroxytryptamine (5HT) were also assayed at various pHs, but no reaction products were obtained. In addition, r16OMT activity was assayed in vitro against various alkaloids (2,3-dihydro-3-hydroxytabersonine, lochnericine, hörhammericine) and their respective 16-hydroxylated products (Figure 5), but none were accepted as substrates for O-methylation. Finally, r16OMT activity was assayed in vitro against quercetin, kaempferol and caffeic acid, and none were accepted as substrates. The results show that, despite its high similarity to other functionally characterized FOMTs (76–78% nucleotide sequence identity, 60–65% amino acid identity (Supplementary Figure S3 and Supplementary Table S1), 16OMT is a highly specific enzyme for its alkaloid substrate.

Figure 5.

 Substrate specificity of r16OMT for its methyl acceptor.
The only methyl acceptor for r16OMT was 16-hydroxytabersonine (100% = specific activity: 6.95 ± 0.8 pkat/mg protein, = 6 replicates). The substrates 16-hydroxy-2,3 dihydrotabersonine, 16-hydroxylochnericine and 16-hydroxyhörhammericine were produced by biotransformation of 2,3 dihydrotabersonine, lochnericine and hörhammericine using Escherichia coli expressing a recombinant tabersonine-16-hydroxylase:NADPH cytochrome P450 reductase fusion protein (Schröder et al., 1999).

Kinetic analysis of the recombinant 16OMT

Substrate saturation kinetics revealed that r16OMT displayed a high affinity for 16-hydroxytabersonine and AdoMet, with Km values of 2.6 and 21.7 μm, respectively (Table 2). In addition, S-adenosyl-l-homocysteine (AdoCys), a well known competitive inhibitor of OMTs, had an inhibitory constant (Ki) of 600 nm (Table 2), whereas 16-methoxytabersonine was not an inhibitor at the concentrations tested (5–50 μm, data not shown).

Table 2.   Kinetic data for 16-hydroxytabersonine and S-adenosyl-l-methionine as substrates and S-adenosyl-l-homocysteine as an inhibitor for r16OMT
 Kmm)Vmax (mm sec−1)Kcat (sec−1)Kcat/Vmax (m−1 sec−1)Kim)
  1. The data represent mean values (± SD) from three independent experiments.

16-hydroxytabersonine2.6 ± 0.060.58 ± 0.08615445.2236.7
S-adenosyl-l-methionine21.7 ± 11.60.79 ± 0.47647278.647.2
S-adenosyl-l-homocysteine6 ± 0.5

Leaf epidermal cells are 900-fold enriched in 16OMT activity compared to whole leaves, and sixfold enriched in 16OMT transcripts compared to LCM-captured whole leaves

Crude epidermis-enriched and total leaf extracts were prepared by the CA technique using the small-scale protocol described in Experimental procedures. Extracts were desalted to remove small molecular weight molecules, and these were assayed for various OMT activities using 16-hydroxytabersonine and quercetin as substrates, as well as for NMT (2,3-dihydro-16-methoxytabersonine N-methyltransferase) and DAT activities, which are not expressed in leaf epidermis (Murata and De Luca, 2005). Inspection of 16OMT reaction products showed that CA treatment was required to obtain leaf epidermis-enriched 16OMT extracts, as simply dipping leaves in extraction buffer without carborundum treatment did not produce any activity (data not shown). Extractions were performed in triplicate, and the specific activities of whole-leaf protein extracts were 0.03 and 0.04 pkat mg−1 protein for 16-hydroxytabersonine and quercetin, respectively (Figure 6a). In contrast, leaf epidermis-enriched extracts had 300-fold (9.43 pkat mg−1) and 30-fold (1.37 pkat mg−1) higher 16OMT and FOMT activities, respectively, compared to those of whole leaves (Figure 6a). To control for possible extraction of cells within the leaf mesophyll, the leaf epidermis-enriched extracts were shown to have no NMT or DAT activity compared with whole-leaf extracts (Figure 6a).

Figure 6.

 Preferential expression of 16OMT in Catharanthus leaf epidermis as determined by biochemical assay and real-time PCR. (a) Quantification of 16OMT, FOMT, NMT and DAT enzyme activities in leaf epidermis-enriched extracts (CA) compared to whole-leaf extracts (WL).The specific activities in pkat/mg protein were obtained in triplicate for each enzyme, and the mean values are shown (means ± SE for CA are 16OMT, 9.43 ± 0.74; FOMT, 1.37 ± 0.122; NMT, 0.00; DAT, 0.00; those for WL are 16OMT, 0.03 ± 0.0015; FOMT, 0.04 ± 0.0060; NMT, 0.06 ± 0.0050; DAT, 0.05 ± 0.0080).
(b) Real-time PCR quantification of 16OMT in various cell types: whole leaves (W); leaf epidermis (E); palisade mesophyll cells (M); palisade-assisted idioblast cells (I); cross-connected laticifer cells (L); vascular cells (V). Actin is used to calibrate the reaction.

The levels of 16OMT transcripts were quantified by real-time PCR in tissues obtained by laser capture microdissection (LCM) (whole leaves, epidermal cells, palisade mesophyll cells, palisade-assisted idioblast cells, cross-connected laticifer cells and vascular cells). As described previously (Murata and De Luca, 2005), the amount of RNA isolated from extraction of LCM-dissected cells was not sufficient for direct RT-PCR analysis, so it was amplified by T7-based RNA amplification prior to PCR as previously described (Nakazono et al., 2003) to produce cDNA that was useful for detecting expression of various genes in various cell types by real-time or by RT-PCR. Real-time PCR analysis revealed that the leaf epidermis of C. roseus contained at least six times more 16OMT transcript than whole leaves or any of the other cell types (Figure 6b). Together, the very high leaf epidermis enrichment ratio of 16OMT activity and the preferential detection of 16OMT transcript in LCM-captured leaf epidermal cells strongly suggests that this reaction occurs in leaf epidermis, in contrast to the later stages in vindoline biosynthesis (NMT and DAT) (Figure 6a,b).

Further RT-PCR analysis showed that all CrOMTs, with the exception of CrOMT5, could be detected in all LCM-captured cells. It is interesting to note that CrOMT6, whose gene product catalyzes the 4′-O-methylation of flavonoids (Schröder et al., 2004), showed a strong signal in Catharanthus roots, whereas CrOMT5, of unknown biochemical function, was strongly detected in flowers and stems (data not shown). This result also suggests flowers are a promising tissue to elucidate the function of CrOMT5, as it may have a specific biological role in flower biochemistry.

Discussion

The carborundum abrasion technique is a cost-effective, robust method to study leaf epidermis biology

The carborundum abrasion technique has been used successfully to differentially extract indole alkaloids, enzymes active in MIA biosynthesis, and mRNA from the leaf epidermis of Catharanthus (Murata and De Luca, 2005). Combined use of this technique with enzyme assays suggested that leaf epidermis preferentially expresses 16OMT compared to other leaf cell types (Murata and De Luca, 2005). These results were used to suggest that leaf epidermal cells are biosynthetically competent to produce tryptamine and secologanin precursors, which are converted via many enzymatic transformations to make 16-methoxytabersonine (Murata and De Luca, 2005), while the remainder of the pathway leading to vindoline biosynthesis appears to occur within specialized cells in the leaf mesophyll (St Pierre et al., 1999).

The present study shows that the CA technique could be scaled up to useful levels (Supplementary Figure S1) for purifying leaf epidermis-localized proteins such as 16OMT to homogeneity (Table 1). Using the scaled-up approach, the specific activity of 16OMT was 36 times higher in concentrated crude CA extracts compared to that of whole-leaf extracts (Table 1). This enrichment of 16OMT by CA extraction was essential to purify this enzyme to homogeneity (Supplementary Figure S2) and to obtain peptide sequences that led to the successful cloning (Figure 3 and Supplementary Figure S3) of 16OMT from a cDNA library produced from leaf epidermis-enriched mRNA. In contrast, previous efforts using plant cell suspension cultures led to the isolation of a number of novel contaminating OMTs (CrOMT2, 4, 5, 6 and 7) (Cacacea et al., 2003; Schröder et al., 2004).

The usefulness of the CA technique as a tool to enrich for leaf epidermis-localized biological processes has also recently been validated by random sequencing of cDNA libraries produced from leaf epidermis-enriched mRNA (Murata et al., unpublished data). The sequencing of almost 10 000 random clones from this library produced four independent sequences identical to various parts of 16OMT (Supplementary Figure S4), but this clone was not represented in EST sequencing of cDNA libraries from mRNA isolated from whole leaves or from root tips (Murata et al., 2006). This study clearly shows that the CA technique is a cost-effective, robust means by which to enrich for epidermal components of plant tissues, and can be selectively used to dissect various metabolic pathways, to purify the proteins involved or to generate useful cDNA libraries that represent the biological activities of the leaf epidermis. It is clear that combining the CA technique with traditional protein purification or with modern proteomics tools significantly improves our capability to study epidermis-localized biological processes. The versatility of epidermis-enriched extracts for investigating cellular specialization is surprisingly similar to that found in studies performed with isolated glandular trichomes for investigating their highly specialized gene expression and chemistry (Lange et al., 2000; Gang et al., 2001; Wagner et al., 2004; Fridman et al., 2005). Clearly this approach may be of great use to study the comparative biology of Catharanthus leaf epidermis during growth and development, or it could be used to compare the leaf epidermis biology between plant species.

Properties of r16OMT

r16OMT showed very high substrate specificity for 16-hydroxytabersonine, as slight modifications of this substrate eliminated enzyme activity (Figure 5). The enzyme did not accept selected aromatic amines or the flavonoid quercetin that is a substrate for other related Catharanthus OMTs (Cacacea et al., 2003; Schröder et al., 2004). The inability of the enzyme to O-methylate 16-hydroxy-2,3-dihydrotabersonine suggests that this reaction occurs prior to further substitution of 16-hydroxytabersonine (Figure 1), and supports its order in the pathway of vindoline biosynthesis. Both r16OMT and the enzyme from leaf extracts showed characteristic features of OMTs (Koch et al., 2003), such as strong inhibition by low concentrations of S-adenosyl-l-homocysteine, but neither was affected by the divalent cations (Mn2+, Mg2+, Zn2+, Ca2+ and Cu2+) that modulate the activities of some OMTs (Lin et al., 2006; Schubert et al., 2003) (data not shown). Together, these data suggest that 16OMT is the gene responsible for in vivo 16OMT activity in Catharanthus.

Leaf epidermis expression of 16OMT activity in Catharanthus

The enrichment of 16OMT activity (Figure 6a) and 16OMT transcript (Figure 6b) in leaf epidermis compared to FOMT (Figure 6a), and the greater FOMT activity found in most Catharanthus organs tested (leaf, flower, stem and root) (Figure 2d) provides very strong evidence for preferential expression of 16OMT within leaf epidermal cells, which have been shown to be specialized for much of the MIA pathway (Murata and De Luca, 2005; St Pierre et al., 1999). While combined use of carborundum abrasion and laser capture microdissection techniques has enabled targeted analysis of the specialized biochemical function of different cell types, the expression profiles of all currently published CrOMTs were very similar in most Catharanthus tissue types, with the exception of CrOMT5, which was preferentially expressed in flowers and stems, and CrOMT6, which was preferentially expressed in roots (data not shown). In addition, neither idioblast nor mesophyll cells appear to express any of the tested OMT transcripts at high levels.

While 16OMT activity is present in Catharanthus seedlings growing in the dark or in the light, its activity profile is similar to that of 2,3-dihydro-3-hydroxytabersonine N-methyltransferase (De Luca and Cutler, 1987; De Luca et al., 1988). Additionally, 16OMT responds to light in a similar way as the downstream N-methyltransferase, whereby both enzyme activities are increased slightly when etiolated seedlings are treated with light. In contrast, both deacetylvindoline-4-hydroxylase (Vázquez-Flota and De Luca, 1998) and DAT (Figure 2b) (De Luca et al., 1988; St Pierre et al., 1998) activities were induced many fold when etiolated seedlings were submitted to light treatment.

Conclusions

Using C. roseus as a model system, a simple, general and highly useful method for harvesting proteins and mRNA by the carborundum abrasion technique is reported. This method was essential for the successful molecular cloning, functional characterization and expression profiling of 16OMT (Figure 1), which is responsible for the 5th to last step in vindoline biosynthesis that is localized in the leaf epidermis. This versatile method could be applied for studying the changing biochemistry of specialized leaf epidermis during growth and development, as has already been done with harvested glandular trichomes (Lange et al., 2000; Gang et al., 2001;Wagner et al., 2004; Fridman et al., 2005). The specialized nature of the leaf epidermis suggests that CA combined with methods for studying the complement of genes expressed in this cell type should yield remarkable new information, as illustrated in this study.

Experimental procedures

Plant material

Madagascar periwinkle [C. roseus (L.) G. Don, Little Delicata] was used throughout the study. Seeds were germinated in moist artificial soil (Sun Gro, Horticulture Canada Ltd; http://www.sungro.com) and grown under controlled greenhouse conditions (25°C, 57% humidity, 16 h photoperiod). The plants were watered once a week, and once every three months they were provided with all-purpose fertilizer (Miracle Gro, 12% nitrogen, 4% phosphate, 8% potassium) (Scotts Canada; http://www.scottscanada.ca).

Large-scale CA extraction method to harvest leaf epidermis protein: purification of 16OMT and protein sequencing

Young leaves (128 g, 1.5 cm long) harvested near the apical meristem of shoots were abraded with carborundum number F, grit size 320 Grade (Fisher Scientific; http://www.fishersci.ca) using a vortex to selectively extract the leaf epidermis (Figure S1), according to a protocol modified from that described by Murata and De Luca (2005). The length and intensities of vortex treatment were optimized to ensure underlying cell layers were largely left undisturbed, which was determined by monitoring for the absence of idioblast- and laticifer-localized DAT enzyme activity. The optimized protocol used a 1:1 w/w ratio of FW of young leaves to carborundum, and a 1:12 w/v ratio of FW of young leaves to protein extraction buffer (100 mm Tris–HCl, pH 7.5, 13 mmβ-mercaptoethanol) at 4°C to perform CA. Leaf epidermal proteins were extracted at maximum vortex intensity over 10 min. To keep the extraction buffer cool, the extract was placed on ice for 2 min after the first 5 min of CA. During CA, the extract was vigorously shaken approximately every 30 sec to suspend any sedimented carborundum. To ensure that all leaf material was abraded as evenly as possible, the angle at which the vessel rested on the vortex during abrasion was approximately 45°. The same extraction buffer was used up to four times by removing abraded leaf material and carborundum by filtration through a nylon mesh and adding fresh leaf material and fresh carborundum (1:1 w/w ratio), as well as fresh extraction buffer to compensate for the loss in volume.

The combined extract was filtered sequentially through a 20 mm nylon mesh, then by suction through 0.45 μm Filtropur® vacuum filtration unit (Sarstedt; http://www.sarstedt.com), and the filtrate was centrifuged at 500 g for 5 min at 4°C to remove remaining carborundum particles. Depending on the volume, extracts were concentrated to 2.5 ml using Centricon Plus-70® 10 kDa and Amicon Ultra® 10 kDa centrifugation concentrators (Fisher Scientific; http://www.fishersci.ca), and the concentrates were desalted by PD-10 column chromatography (GE Healthcare; http://www.gehealthcare.com/ca).

The concentrated desalted protein extract (3.5 ml) was diluted to 5 ml and was submitted to Sephadex G150 (Sigma-Aldrich, http://www.sigmaaldrich.com/) column chromatography [125 ml bed volume, equilibrated in buffer A (100 mm Tris–HCl, pH 7.5, 13 mmβ-mercaptoethanol), flow rate of 0.3 ml min–1]. Eluted fractions were collected in 4 ml volumes, and fractions were assayed for 16OMT and FOMT to harvest peak 16OMT fractions that were concentrated by an Amicon Ultra® 10 kDa centrifugation concentrator (Fisher Scientific).

The concentrated sample (2 ml) was incubated with 1 ml activated adenosine–agarose affinity resin [prepared from 5′-AMP-agarose (Sigma-Aldrich) by dephosphorylation with bovine alkaline phosphatase (James et al., 1995)] overnight at 4°C with constant gentle mixing. The mixture was centrifuged at 10 000 g for 10 sec in a benchtop microcentrifuge. The resin was washed three times with 2.5 ml buffer A, and bound proteins were eluted three times with 2.5 ml buffer A containing 500 μm AdoMet (Sigma-Aldrich). After desalting, active fractions were pooled and applied to a Mono Q™ anion exchange column (2 ml bed volume) (GE Healthcare) equilibrated in buffer A at 4°C, and, after washing the column, 16OMT activity was eluted in 0.5 ml fractions using a 20 ml 0–50% gradient of 2 m NaCl in buffer A. Active fractions were concentrated, subjected to SDS–PAGE, and stained with colloidal Coomassie blue (Invitrogen, http://www.invitrogen.com/). A 40 kDa protein band (Supplementary Figure S2) that positively correlated with 16OMT activity was excised from the gel and forwarded for sequence analysis, performed by Harvard Microchemistry Facility (Harvard University) by microcapillary reverse-phase HPLC nano-electrospray tandem mass spectrometry (μLC/MS/MS) on a Finnigan LCQ DECA XP Plus quadrupole ion trap mass spectrometer (Thermo; http://www.thermo.com). Seven peptide sequences were obtained from this protein (peptide 1, WILHDWNDEDCVK; peptide 2, GIVLTMLDPAELK; peptide 3, MVLHDWNDEDCVK; peptide 4, NEDGTAFETAHGK; peptide 5, IPPAHVVFLK; peptide 6, EAGFSSYK; peptide 7, CTVFDLPHVVANLESK).

Small-scale CA extraction of leaf epidermis-enriched protein compared to whole leaf extraction for estimating 16OMT activity in the epidermis

Young leaves (2 g FW, 1.5 cm long) were harvested and processed as described for the large-scale method with the following modifications. Fresh leaves, carborundum and 24 ml of extraction buffer were placed in a 50 ml PP tube (Sarstedt) and shaken vigorously by hand for 1 min. The extraction buffer was then filtered through P5 filter paper (Fisher Scientific) to remove residual carborundum. The filtrate was further filtered using a Luer-Lok™ syringe and 25 mm syringe filter (pore size 0.22 μm) (Fisher Scientific). The second filtrate was concentrated to 2.5 ml using Amicon Ultra centrifugal filter units (10 kDa molecular weight cut-off) (Fisher Scientific). Concentrated epidermis-enriched protein extract was desalted using PD-10 desalting columns (GE Healthcare), and then used for enzyme assay.

Extraction of whole leaves for comparative enzyme assays to the CA-based extraction method

Whole young leaves (2.0–6 g FW) were homogenized in 3–9 ml protein extraction buffer using a mortar and pestle. The sample was centrifuged at 500 g for 5 min at 4°C to pellet cell debris, and the supernatant was processed as described for leaf epidermal extracts for direct assay of crude extracts or for protein purification. The protein concentration of extracts was determined using a protein assay kit (Bio-Rad, http://www.bio-rad.com/).

CA extraction to harvest leaf epidermis enriched mRNA

Young C. roseus leaves (2.5 g FW) were harvested and combined with 8 ml Trizol® reagent (http://www.invitrogen.com) and 2.5 g carborundum (Fisher Scientific) in a conical 50 ml tube (Sarstedt) (Figure S1). The tube containing leaf material, carborundum and Trizol was vortexed at maximum intensity for 1 min, ensuring that most of the leaves in the tube were well abraded. The tube was then placed at room temperature for 5 min to allow extraction buffer to settle to the bottom. Abraded leaves were removed with forceps, and the process was repeated with another 2.5 g of fresh leaf tissue. The extract was centrifuged at 2600 g for 5 min, and the supernatant (5 ml) was harvested and mixed by vortex sequentially with 1 ml 5 m NaCl, followed by 3 ml chloroform. The mixture was centrifuged at 2600 g for 30 min, and aliquots of the aqueous phase were transferred to sterile 1.5 ml tubes together with 0.9 vol isopropanol with vortex mixing, and the mixture was incubated at −20°C for 1 h or overnight in the presence of linear acrylamide to maximize the amount of mRNA isolated. Samples were centrifuged at 21 000 g for 30 min, the supernatant was removed, and the precipitated pellet was washed with 70% ethanol, followed by centrifugation at 21 000 g for 5 min. The pellet, containing mRNA, was resuspended in 20–50 μl RNase-free water. RNA quality and quantity were measured by spectrophotometry, and approximately 50 μg of total RNA was obtained from 5 g FW of leaves extracted by CA.

Construction and random sequencing of leaf epidermis-specific cDNA library

A Catharanthus leaf epidermis-specific cDNA library was constructed using a SMART cDNA library construction kit (Clontech, http://www.clontech.com/) according to the manufacturer’s instructions. After producing cDNA from leaf epidermis enriched mRNA, it was amplified by PCR prior to packaging with Gigapack III gold packaging extract (Stratagene, http://www.stratagene.com/). The primary library (1.0 × 106 pfu) was directly converted to plasmids by in vivo excision, and the obtained colonies of E. coli were randomly selected for single path sequencing using primers from the 5′ end of the inserts. The sequencing reactions were performed using a Templiphi DNA sequencing template preparation kit (GE Healthcare), and the resulting DNA templates were sequenced using ABI Prism Big Dye terminator sequencing kits (Applied Biosystems, http://www.appliedbiosystems.com/) and an ABI 3730 genetic analyzer (Applied Biosystems).

Identification of potential 16OMT expressed sequences

The sequence files in ABI format were analyzed using the BLASTX algorithm (Altschul et al., 1997). Multiple clones with overlapping areas of identical sequences were clustered and classified as ‘Clustered’, while sequences that appeared only once in the EST were classified as ‘Singletons’. The threshold of sequence similarity was set as an E-value of 10−6 and lower, and sequences that did not show significant homology were termed ‘No hits’. The sequences were archived in the fiesta software package (http://bioinfo.pbi.nrc.ca/napgen.beta//login.html) at the Plant Biotechnology Institute of the National Research Council of Canada (NRC/PBI). The functional categorization was first performed automatically by the putative annotations, followed by manual inspection to verify the annotation.

Molecular cloning of 16OMT

Gene-specific forward primers were designed to the mRNA domains of CrOMT7 (AY343492) corresponding to peptide 4. The associated forward primer corresponding to peptide 4 was 5′-CTGCTTTTGAAACAGCTCATGG-3′ (F1), and the reverse primer designed to the mRNA domain of CrOMT6 (AY343490) corresponding to the N-terminus of peptide 1 was 5′-CAGTCATGGAGAATCCACTT-3′ (R1). PCR using F1, R1 and 2 μl of epidermal-enriched cDNA in lambda phage was performed for 25 cycles. The amplified reaction was diluted 1000-fold, and a 2 μl aliquot was used as template for a second PCR amplification with F1 and R1 to yield a novel 361 bp putative OMT sequence. The PCR product was sequenced at Robarts Research Institute (London, Ontario, Canada), and specific primers corresponding to the novel putative OMT were then designed to amplify the 5′ and 3′ ends by RACE using lambda phage primers flanking the cDNA insert. This sequence was used to produce a contig with Vector nti software version 6.0.0.0 (InforMax Inc., http://www.invitrogen.com/). The full-length gene (1065 bp, encoding the full-length ORF) was amplified using forward primer 5′-CACCATGGATGTTCAATCTGAGGA-3′ and reverse primer 5′-TCAAGGATAAACCTCAATGAGACTCC-3′, which are compatible with the Gateway® pENTR directional vectors (Invitrogen), and directionally ligated into pENTR and subsequently mobilized to pDEST-17 of the Gateway® expression system (Invitrogen), which harbors six histidine residues at the N-terminus. This was performed according to the manufacturer’s specifications. pDEST-17 vectors containing the full-length amplified product–His fusion were transformed into E. coli (DE3)pLysS cells.

Expression, refolding and purification of r16OMT

A 3 ml culture containing 2 × YT medium with 50 μg ml–1 ampicillin was inoculated with E. coli expressing r16OMT, and grown overnight at 37°C to produce a saturated culture. A 500 μl aliquot of saturated culture was then used to inoculate 50 ml 2 × YT medium containing 50 μg ml–1 ampicillin. Cultures were allowed to grow at 37°C to an OD600 of 0.8, induced with IPTG (2 mm final concentration) and grown at 16°C for 24 h. The culture was then centrifuged at 5000 g for 10 min to harvest cells, the supernatant was removed, and cell pellets were stored at −20°C until processing for extraction.

Cell pellets were resuspended in 5 ml 100 mm Tris–HCl, pH 7.5, 13 mmβ-mercaptoethanol, and lysed by sonication. After centrifugation at 21 000 g (4°C) for 20 min, the pellet with inclusion bodies was washed three times in 10 ml wash buffer (100 mm Tris–HCl, pH 7.5, 150 mm NaCl) containing 1 m urea per g pellet weight, 0.1% Triton X-100. After each wash, inclusion bodies were re-pelleted by centrifugation at 21 000 g for 10 min at 4°C. This was followed by a single wash step with wash buffer to remove residual Triton X-100. Purified inclusion bodies were then solubilized in solubilization buffer [100 mm Tris–HCl, pH 7.5, 8 m urea, 1 mm phenylmethylsulfonyl fluoride to a final protein concentration of ≤ 10 mg ml–1 (Kopito, 2000)]. Protein inclusion bodies were solubilized in a beaker with gentle stirring for 1 h at room temperature, and solubilized proteins were dialyzed (100 mm Tris, pH 7.5) overnight at 4°C using a 30 ml Slide-A-Lyzer dialysis cassette (10 kDa molecular weight cut-off, Fisher Scientific) according to the manufacturer’s instructions. Refolded proteins were used directly for enzyme assays.

Enzyme kinetic analysis of r16OMT

E. coli expressing recombinant tabersonine-16-hydroxylase:NADPH cytochrome P450 reductase protein (Schröder et al., 1999) was used to biotransform tabersonine into the 16-hydroxytabersonine used in r16OMT assays. For Km determinations, r16OMT enzyme assays were performed with 2, 6, 10, 14 and 18 μm 16-hydroxytabersonine at constant concentration of 2.08 μm (0.025 μCi) (methyl-14C)AdoMet, and with 1, 2, 5, 10 and 20 μm (methyl-14C)AdoMet at a constant concentration of 0.1 mm 16-hydroxytabersonine. For Ki determinations, r16OMT enzyme assays were performed, containing 1, 2, 5, 10 and 20 μm (methyl-14C)AdoMet at a constant concentration of 30 μm 16-hydroxytabersonine and varying AdoCys concentrations (0, 0.05 and 0.3 × the Km for AdoMet). All assays were performed in triplicate at 35°C for 30 min at pH 7.5 (100 mm Tris–Cl, 0.1%β-mercaptoethanol).

Preparation of cDNA from laser capture microdissected cells

cDNA was prepared from RNA that had been amplified after being isolated from epidermal, mesophyll, idioblast, laticifer, vascular and representative whole-leaf cells isolated by laser capture microdissection as previously described (Murata and De Luca, 2005).

Real-time PCR for quantification of 16OMT in various cell types obtained by laser capture microdissection

β-Actin and 16OMT sequences were aligned using clone manager professional suite (version 7.11, Scientific & Educational Software; http://www.scied.com) to determine possible regions for the best primer candidates. Initial primer design was performed using with the software primer 3 (Whitehead Institute, MIT, Cambridge, MA, USA), and further examined by integrated DNA technology (IDT) web-based software (http://scitools.idtdna.com/scitools/Applications/OligoAnalyzer/) to avoid hairpin, hetero- or self-dimer structures. The primer pairs Actin-F (5′-GGAGCTGAGAGATTCCGTTG-3′) and Actin-R (5′-GAATTCCTGCAGCTTCCATC-3′), and 16OMT-F (5′-CCTTACCCCATCAAGTCGAA-3′) and 16OMT-R (5′-ACAGGGTCAGCCATGGTAAG-3′) were used to generate 73 and 81 bp PCR products, respectively. The real-time quantitative PCR reaction was carried out in a final 20 μl containing 150 nm of each primer, 10 μl QuantiTect® SYBR Green PCR Master Mix (Qiagen, http://www.qiagen.com/), and 2 μl of cDNA (corresponding to approximately 216 ng cDNA). Real-time PCR conditions were as follow: 95°C for 15 min, then 45 cycles of 95°C for 10 sec, 57°C for 15 sec and 72°C for 30 sec. All real-time PCR experiments were run in triplicate for each biological replicate of cDNA produced from mRNA isolated from whole leaves and from cells from the leaf epidermis, mesophyll, idioblasts, laticifers and vasculature. The average threshold cycle (Ct) and relative quantities were calculated using sds software version 2.1 (Applied Biosystems). A Ct value from each sample was calculated from the amplification curves by selecting the optimal ΔRn (emission of reporter dye over starting background fluorescence) in the exponential portion of the amplification plot. β-Actin was used as an internal standard to calculate the relative fold difference based on the comparative Ct method. To determine relative fold differences for each sample in each experiment, the Ct value for the 16MOT gene was normalized to the Ct value for β-Actin, and was calculated relative to a calibrator (laticifer cells) using the formula inline image.

Expression of 16OMT, TDC and DAT enzyme activity in light- and dark-grown seedlings

Catharanthus roseus seeds were germinated and grown for various times in the absence and presence of light (De Luca et al., 1988). Enzyme assays for TDC, DAT (De Luca et al., 1988) and 16OMT (Fahn et al., 1985;Murata and De Luca, 2005) were performed as previously described.

Protein determinations

The protein concentration of extracts was determined using a protein assay kit (Bio-Rad).

Acknowledgements

This work was supported by a Discovery Grant to V.D.L and by a Joint Strategic Grant to Dr Peter Facchini and V.D.L. from the Natural Sciences and Engineering Research Council of Canada. V.D.L. holds a Canada Research Chair in Plant Biotechnology. We thank the Natural Products Genomics Resource Research Consortium (Plant Biotechnology Institute of National Research Council of Canada) for financing and performing the EST sequencing component of this work, the Harvard Microchemistry facility for protein sequencing, and Professor Joachim Schröder (University of Freiburg, Germany) for supplying us with a functional T16H:NADPH cytochrome P450 reductase fusion for preparing MIA substrates for 16OMT enzyme assays.

Genbank accession numbers EF444544, EF444545, EF444546, EF444547, EF444548.

Ancillary