Note: The nucleotide sequence for columbamine O-methyltransferase has been deposited in the GenBank database under GenBank Accession Number AB073908.
F. Sato, Division of Integrated Life Sciences, Graduate School of Biostudies, Kyoto University, Kyoto, 606-8502, Japan. Fax: + 81 75 753 6398, Tel.: + 81 75 753 6381, E-mail: email@example.com
To identify all of the O-methyltransferase genes involved in isoquinoline alkaloid biosynthesis in Coptis japonica cells, we sequenced 1014 cDNA clones isolated from high-alkaloid-producing cultured cells of C. japonica. Among them, we found all three reported O-methyltransferases and an O-methyltransferase-like cDNA clone (CJEST64). This cDNA was quite similar to S-adenosyl-l-methionine:coclaurine 6-O-methyltransferase and S-adenosyl-l-methionine:isoflavone 7-O-methyltransferase. As S-adenosyl-l-methionine:columbamine O-methyltransferase, which catalyzes the conversion of columbamine to palmatine, is one of the remaining unelucidated components in isoquinoline alkaloid biosynthesis in C. japonica, we heterologously expressed the protein in Escherichia coli and examined the activity of columbamine O-methyltransferase. The recombinant protein clearly showed O-methylation activity using columbamine, as well as (S)-tetrahydrocolumbamine, (S)-, (R,S)-scoulerine and (R,S)-2,3,9,10-tetrahydroxyprotoberberine as substrates. This result clearly indicated that EST analysis was useful for isolating the candidate gene in a relatively well-characterized biosynthetic pathway. The relationship between the structure and substrate recognition of the O-methyltransferases involved in isoquinoline alkaloid biosynthesis, and a reconsideration of the biosynthetic pathway to palmatine are discussed.
Higher plants produce a wide variety of chemicals, including more than 25 000 terpenoids, approximately 12 000 alkaloids and approximately 8000 phenolic compounds . Thus, secondary metabolism in plant cells could be a useful source of these chemicals, especially alkaloids, many of which have high biological activities and are used as medicines.
Isoquinoline alkaloid biosynthesis is the most well-characterized pathway in plant cells and some biosynthetic genes have been isolated [2,3 and references cited therein]. However, the conventional method for isolating the corresponding biosynthetic genes based on the purified enzymes is a rate-limiting step for characterization, even though this approach can be very useful when we have no information about the desired enzyme(s). Although a useful approach is to isolate a group of genes based on their structural similarity, such as O-methyltransferase (OMT) or P450 motifs, this approach faces the problem of identifying biological activity due to the high redundancy of related genes.
On the other hand, recent scientific and technological advances now enable the high-through-put sequencing of cDNA clones, and we can analyze these expressed sequence tag (EST) using on-line databases and search tools. These ESTs are also useful for evaluating mRNA transcripts to estimate their biological function.
To evaluate the effectiveness of this technology, we sequenced the cDNA clones isolated from high-alkaloid-producing cultured cells of Coptis japonica. As we previously characterized some of the OMTs in isoquinoline alkaloid biosynthesis in Coptis cells, we focused on the isolation of the another unelucidated OMT in this biosynthetic pathway; i.e. S-adenosyl-l-methionine:columbamine O-methyltransferase (CoOMT), which catalyzes the transfer of the S-methyl group of S-adenosyl-l-methionine (AdoMet) to the 2-hydroxyl group of columbamine to form palmatine (Fig. 1).
Methyltransferases are essential for directing intermediates to specific biosynthetic pathways . Each methyltransferase in the biosynthesis of palmatine (i.e. S-adenosyl-l-methionine:norcoclaurine 6-O-methyltransferase (6OMT) [6–10]; S-adenosyl-l-methionine:coclaurine N-methyltransferase (CNMT) [11–14]; S-adenosyl-l-methionine:3′-hydroxy-N-methylcoclaurine 4′-O-methyltransferase (4′OMT) [10,15]; S-adenosyl-l-methionine:scoulerine 9-O-methyltransferase (SMT) [16–18]; and CoOMT ) has strict substrate specificity, despite the structural similarities of the various substrates. Due to their importance in the production of pharmaceutically important alkaloids and their strict substrate recognition, these methyltransferases, especially OMTs, have been well characterized [6–17] and their cDNA has been isolated [10,14,18]. However, CoOMT was characterized using only partially purified enzymes  and its cDNA has not yet been isolated.
We report here the first isolation of CoOMT from C. japonica ESTs, the expression of functionally active recombinant enzyme, and its characterization. Characterization of its specificity showed that Coptis CoOMT could methylate tetrahydrocolumbamine as well as columbamine, and suggests that the biosynthetic pathway to palmatine should be reconsidered.
The original cultured cells were induced from rootlets of C. japonica Makino var. dissecta (Yatabe) Nakai. A cell line (156–1) that produces large amounts of alkaloids was subcultured as described elsewhere . Ten-day-old-cultured cells were harvested and used for the extraction of mRNA.
(R,S)-Norlaudanosoline and (R,S)-laudanosoline were purchased from Aldrich, and berberine was from Wako Pure Chemical Industries, Ltd. (R)- and (S)-coclaurine were the gift of Dr Nagakura of Kobe Women's College of Pharmacy, and (S)-reticuline was a gift from Dr Facchini of the University of Calgary. The other alkaloids were generous gifts from Mitsui Petrochemical Industries, Ltd.
Chemical synthesis of columbamine
Columbamine was synthesized and purified from berberine by two-step reactions according to the method of Cava and Reed [20,21].
Construction and sequencing of cDNA library of C. japonica
Poly(A)+RNA was isolated, and a cDNA library was constructed as described elsewhere . The cDNA fragments were ligated into pDR196 vector . Sequencing of the cDNA library was performed using a MegaBACE 1000 DNA Sequencing System (Amersham Pharmacia Biotech) in accordance with the manufacturer's instructions. Overall, 1014 ESTs were obtained in sufficient length and these sequences were annotated using a blast search (http://www.ncbi.nlm.nih.gov/blast/) (E. Dubouzet et al. in preparation). During the blast search, we found an unidentified OMT cDNA (CJEST64) in addition to the sequences of reported Coptis OMTs. To obtain more information about the sequence of this CJEST64, the complete cDNA sequence was determined with deletion clones of CJEST64 and a DSQ-2000 L DNA sequencer (Shimadzu) with fluorescein isothiocyanate-labelled primers.
Isolation of full-length cDNA
To isolate full-length cDNA of the unknown OMT (CJEST64), rapid amplification of the 5′ end of cDNA (5′RACE) was carried out using a Marathon cDNA Amplification Kit (Clontech). Gene-specific antisense primer (5′-CTCCAAACTGAGAACTCTTCCG-3′) was designed based on the nucleotide sequence of CJEST64. PCR fragments were isolated and subcloned into pT7Blue vector (Novagen), and their nucleotide sequences were determined.
Construction of an expression vector for full-length CJEST64
An expression vector was constructed for full-length CJEST64 cDNA without the fused peptide derived from the vector sequence in pET-21d vector (Novagen). The full-length cDNA was first prepared by PCR with a Marathon adaptor primer (Clontech) and a forward primer 5′-TTGTTCTAAGGCCATGGTATCTCCG-3′ to introduce a NcoI site (CCATGG) at the start codon. A restriction site was introduced by changing the underlined base. pET-21d was digested with NcoI and XhoI, and the PCR product was ligated into the vector. This construct was completely sequenced to confirm that no changes were introduced by the subcloning process.
Heterologous expression of full-length CJEST64 in Escherichia coli
The expression vector for full-length CJEST64 cDNA was introduced into Escherichia. coli BL21 (DE3). After induction with 1 mm isopropylthiogalactoside, E. coli cells were incubated at 25 °C for 5 h, and then harvested and extracted in extraction buffer (0.1 m Tris/HCl, pH 8.0, containing 10% glycerol, 5 mm EDTA and 10 mm 2-mercaptoethanol). After centrifugation at 12 000 g for 10 min, the supernatant was desalted through an NAP-5 column (Amersham Pharmacia Biotech) and used to measure CoOMT activity.
Measurement of O-methylation activity
The standard CoOMT reaction mixture (50 µL) consisted of 100 mm potassium phosphate (pH 7.8), 25 mm sodium ascorbate, 1 mm columbamine, 3 mm AdoMet and the enzyme preparation (c. 100 µg protein). The assay mixture was incubated at 30 °C for 1 h, after which the reaction was terminated by the addition of methanol. After protein precipitation, the reaction product was detected by reversed-phase HPLC (mobile phase, 35% acetonitrile/H2O containing 1% acetic acid; column, TSKgel ODS-80Tm (4.6 ∞ 250 mm; TOSOH); flow rate, 0.5 mL·min−1; detection, absorbance measurement at 280 nm. Mass spectra were obtained with an LCMS-2010 (Shimadzu).
Tetrahydrocolumbamine was prepared from scoulerine via a scoulerine 9-O-methyltransferase (SMT) reaction. The SMT reaction was carried out as described previously  with a modification. In brief, the expression of SMT in E. coli was induced using a pET expression system and the E. coli lysate was prepared. The SMT reaction mixture consisted of 100 mm Tris/HCl (pH 8.0), 25 mm sodium ascorbate, 1 mm (S)-scoulerine, 5 mm AdoMet and the enzyme preparation. The reaction mixture was incubated at 30 °C for 3 h. In this reaction, scoulerine was converted into tetrahydrocolumbamine. After the enzyme reaction was terminated by the addition of 0.5 m sodium carbonate, the reaction product was extracted twice with ethyl acetate. The extracted product was dried and used for further characterization. HPLC analysis showed that tetrahydrocolumbamine prepared was 80% pure and contained unconverted scoulerine. Because scoulerine was a poor substrate for CoOMT, we analyzed substrate affinity of CoOMT for tetrahydrocolumbamine using this 80% pure tetrahydrocolumbamine.
To quantify the enzymatic activity of CoOMT, the transfer of the 3H-labeled methyl group of S-adenosyl-l-[methyl-3H]methionine (0.5 MBq·µmol−1) (PerkinElmer Life Sciences) to the product was measured as described elsewhere . The kinetic constants of the crude enzyme were determined by varying the concentration of alkaloid substrates, but keeping the concentration of 3H-labeled AdoMet fixed at 1 mm then apparent Km and Vmax values were calculated.
The subunit molecular mass of the enzyme was determined by SDS/PAGE (12.5% polyacrylamide). Protein concentration was determined according to Bradford  with bovine serum albumin as a standard.
Sequence analysis of the OMT-like clone
A cDNA library was prepared from C. japonica cells that produce large amounts of alkaloids, and its 1014 ESTs were obtained (E. Dubouzet et al. in preparation). A blast search (http://www.ncbi.nlm.nih.gov/blast/) showed that these sequenced clones included 4 clones of 4′OMT, 2 clones of 6OMT and 4 clones of SMT. This result indicated that this cDNA library was highly enriched with biosynthetic genes involved in isoquinoline alkaloid biosynthesis. Thus, we speculated that this library also included the gene for the additional unelucidated OMT in isoquinoline alkaloid biosynthesis in C. japonica, i.e. CoOMT. In fact, a blast search identified a clone (CJEST64) which was quite similar to isoflavone 7-O-methyltransferase (IOMT) and 6OMT.
Whereas CoOMT is found in the last step of palmatine biosynthesis and 6OMT is found at an early step, we examined the possibility that this CJEST64 was CoOMT. As only a partial sequence was determined for the EST project, full-length of CJEST64 was determined. The sequence obtained further confirmed that CJEST64 was similar to 6OMT, 4′OMT and IOMT (data not shown), although the sequence was not full-length based on sequence alignment. Thus, the 5′-fragment of CJEST64 was obtained by 5′RACE and sequenced. Finally, full-length cDNA was re-cloned from the cDNA library and the expression vector was constructed as described below.
This full-length cDNA carried 1378 nucleotides, with an open reading frame that encoded 351 amino acids (Fig. 2). The deduced amino acid sequences had conserved putative AdoMet binding domains at the C-terminal end (Fig. 2, motifs A–C), like other OMTs . This polypeptide had a high degree of similarity (approximately 40% identity) to Coptis 6OMT, Coptis 4′OMT, IOMT of alfalfa , and 6a-hydroxymaackiain 3-O-methyltransferase of pea . The deduced polypeptide showed a rather low identity (21%) to SMT. Phylogenetic analysis clearly indicated that this polypeptide belongs to the same branch as 6OMT and 4′OMT (Fig. 3).
Expression of the full-length cDNA in E. coli
We constructed the expression vector to produce recombinant proteins in E coli to examine the activity of CoOMT. We introduced an NcoI site into the cDNA to fit the initiation codon in the E. coli expression vector pET-21d to produce nontagged polypeptide. This construct was then introduced into E. coli cells, and production of the recombinant protein was induced. The crude E. coli lysate was used to detect CoOMT activity. SDS/PAGE analysis clearly showed that transgene expression was successfully induced and the subunit molecular mass was estimated to be approximately 40 kDa (Fig. 4).
HPLC analysis of the reaction mixture clearly showed that recombinant E. coli lysate had CoOMT activity (data not shown). The control E. coli lysate with the pET-21d vector showed no enzymatic activity. LC-MS analysis confirmed that palmatine was produced from columbamine by recombinant E. coli lysate (Fig. 5), and that the CJEST64 product was CoOMT.
Characterization of recombinant CoOMT
As CoOMT activity was identified, we further characterized the enzyme properties of CoOMT using this crude recombinant enzyme. First, we optimized the pH conditions for the CoOMT reaction using columbamine as a substrate. Enzyme assays at various pH values indicated that the optimum pH for the methylation of columbamine was approximately 8.4. Whereas product inhibition of 6OMT and SMT has been reported [8,17], berberine, one of the end-products of isoquinoline alkaloid biosynthesis in Coptis cells, did not inhibit CoOMT activity in the assay mixture at 2.5 mm.
Next, we determined the substrate specificity of CoOMT using the incorporation of radioactivity from S-adenosyl-l-[methyl-3H]methionine into the products as an index (Table 1). When columbamine was used as the control substrate (i.e. relative incorporation 100%), the respective relative activities with (S)- and (R,S)-scoulerine and 2,3,9,10-tetrahydroxyprotoberberine were 31, 22 and 14%, whereas no significant methylation was found for other substrates. The methylation of scoulerine and 2,3,9,10-tetrahydroxyprotoberberine by CoOMT was further confirmed by LC-MS (Fig. 6). Interestingly, mono- and di-methylated products were formed from 2,3,9,10-tetrahydroxyprotoberberine by reacting with CoOMT. This result indicated that the regio-specificity of CoOMT was rather low.
Table 1. Transfer of [3H]-methyl group of S-adnosyl-l-methionine into different substrates by columbamine OMT. Values indicate the incorporation of radioactivity from S-adenosyl-l-[methyl-3H]methionine into the product relative to columbamine. Other assay conditions are indicated in Materials and methods. ND, not detected.
Above result suggested that tetrahydroprotoberberine should also be a good substrate for CoOMT. To confirm this idea, tetrahydrocolumbamine was produced from scoulerine by an SMT reaction, and the reaction product, tetrahydrocolumbamine, was used in the successive CoOMT reaction. Figure 6E,F show that a new reaction product with a 14-m/z greater mass was detected.
To compare the substrate specificity, we determined the kinetic constants of CoOMT for columbamine, tetrahydrocolumbamine and (S)-scoulerine. Substrate-saturation kinetics of the recombinant CoOMT prepared from E. coli for these alkaloids were the typical Michaelis–Menten type. Then, we calculated apparent Km and Vmax values by varying the concentration of these alkaloids in the presence of 1 mm AdoMet. The respective Km values of CoOMT for columbamine, tetrahydrocolumbamine and (S)-scoulerine were 66 ± 18, 35 ± 18 and 173 ± 51 µm, and respective Vmax were 125 ± 12, 21 ± 7, and 6.9 ± 1.3 pkat·mg protein−1.
The present results indicate that high-through-put sequencing of a cDNA library of high-metabolite-producing cells and computer-assisted sequence analysis can be useful for isolating a desired but not yet identified gene(s). Using this strategy, we successfully identified cDNA of CoOMT without purification of the enzyme from plant material. Similar approaches have been reported to isolate enzyme in peppermint oil gland cells . Heterologously expressed enzyme clearly showed CoOMT activity and provided a sufficient amount of material for characterization, whereas such characterization was only partial due to substrate limitations. Further computer-assisted analyses for the hydropathy and localization of CoOMT using the signal p (http://www.cbs.dtu.dk/services/SignalP/) and psort (http://psort.nibb.ac.jp/) programs indicated that CoOMT was a cytosolic enzyme, whereas Berberis CoOMT has been reported to be vesicle-bound .
This CoOMT is the fourth OMT isolated from C. japonica cells to play a role in isoquinoline alkaloid biosynthesis. As each enzyme shows distinct substrate specificity, we compared the sequences of these OMTs. Surprisingly, CoOMT was much more similar to 6OMT and 4′OMT than SMT, even though SMT catalyzes the O-methylation of a protoberberine alkaloid, like CoOMT. CoOMT showed 40% identity to 4′OMT. However, we did not detect recombinant CoOMT in the crude extract of E. coli by an immunoblot analysis with anti4′OMT polyclonal antibodies. Multiple sequence alignment of CoOMT and the biosynthetic enzymes involved in benzylisoquinoline alkaloid biosynthesis, including berberine bridge enzyme [28–30] and (S)-N-methylcoclaurine 3′-hydroxylase (CYP80B1) , also failed to indicate any sequence homology (data not shown). A comparison of CoOMT and CNMT only showed limited sequence similarity in motif A, as seen between CNMT and other OMTs . It is not clear why the sequence of CoOMT is so different from that of SMT.
The three-dimensional structures of IOMT and chalcone OMT in flavonoid biosynthesis have recently been characterized . Multiple sequence alignment of Coptis OMTs, including CoOMT, with OMTs in flavonoid biosynthesis showed that the residues involved in AdoMet binding and the catalytic residues were highly conserved, and the methionine residues that interact with methyl group acceptor were also conserved. However, other substrate-binding residues were not conserved (data not shown). The sequence diversity in the substrate-binding site, which is expected to be located in the N-terminal end of Coptis OMTs (T. Morishige & F. Sato, unpublished data), is considerably high. The sequence of CoOMT may be useful for understanding substrate binding and the enzyme structures in isoquinoline alkaloid biosynthesis.
Our recombinant CoOMT also provides additional information about the branching point in the late steps in palmatine biosynthesis. Columbamine was once thought to be the main substrate for palmatine biosynthesis in Berberis wilsoniae and Berberis aggregata using partially purified enzyme, as Berberis CoOMT could not methylate tetrahydrocolumbamine . However, our findings clearly indicate that tetrahydrocolumbamine can be a substrate for the formation of tetrahydropalmatine (Fig. 6), which can be easily converted to palmatine by (S)-tetrahydroprotoberberine oxidase . Our preliminary analyses also showed that CoOMT had rather smaller Km value for tetrahydrocolumbamine than columbamine, while the Vmax value for tetrahydrocolumbamine was also smaller than that for columbamine. These data suggested that both pathways might operate in Coptis cells. Differences in substrate specificity have also been reported for the CNMT of Coptis and Berberis, i.e. Coptis enzyme could N-methylate norlaudanosoline, whereas Berberis enzyme could not [11,13]. These results suggest that the late biosynthetic pathway should be re-examined in various plant species, and that the pathway in secondary metabolism may vary depending on the enzyme(s) that each plant has acquired during its evolution.
We thank Dr N. Nagakura, Dr P. Facchini and Mitsui Petrochemical Industries Ltd. for their generous gifts of the alkaloids. We also thank Ms. L. Huang for her technical assistance with preparing columbamine. We thank Dr W. Frommer of the University of Tuebingen for the gift of pDR196 vector
This research was supported in part by a Research for the Future Program Grant (JSPS-RFTF00L01607) from the Japan Society for the Promotion of Science (to F. S), and a fellowship from the Japan Society for the Promotion of Science (to T. M).