Tobacco Ntomt2 Revisited: Multiple O-Methylation of Quercetin is Catalyzed by a Single Gene Product

Flavonoids are classified according to the oxidation level of the C-ring, and further enzymatic modifications such as methylation, glucosylation, sulfonation and (iso) prenylation that contribute to their diversity and biological activity. In plants, O-methylation of flavonoids is catalyzed by a family of S-adenosyl-L-methionine (AdoMet)dependent O-methyltransferases (OMTs) [1]. Generally, O-methylation reduces the reactivity of hydroxyl groups, modifies their solubility and increases their anti-microbial and anti-cancer properties [2,3]. So far, more than 100 plant OMTs have been characterized or annotated in GenBank, of which the caffeic acid OMTs (COMTs) are most abundant [4].


Introduction
Flavonoid compounds constitute one of the largest groups of natural products that originate from the shikimate and polyketide pathways. They exhibit a wide range of functions and play important roles in the biochemistry, physiology, and ecology of plants. In addition, they are considered as an important part of the human diet and act as active principles for many medicinal plants.
Flavonoids are classified according to the oxidation level of the C-ring, and further enzymatic modifications such as methylation, glucosylation, sulfonation and (iso) prenylation that contribute to their diversity and biological activity. In plants, O-methylation of flavonoids is catalyzed by a family of S-adenosyl-L-methionine (AdoMet)dependent O-methyltransferases (OMTs) [1]. Generally, O-methylation reduces the reactivity of hydroxyl groups, modifies their solubility and increases their anti-microbial and anti-cancer properties [2,3]. So far, more than 100 plant OMTs have been characterized or annotated in GenBank, of which the caffeic acid OMTs (COMTs) are most abundant [4].
The flavonol quercetin (3,5,7,3′,4′-pentahydroxyflavone; Figure 1) may be considered one of Nature's gifts to mankind, for its potential benefits to human health. It is of ubiquitous occurrence in plants and a common component of foods, especially fruits, vegetables and beverages such as tea, coffee, wine, beer and vinegar [5]. In addition, it is one of the most potent antioxidants derived from plant sources [6], and has been credited for its anti-cancer [7,8], anti-inflammatory [9], and anti-allergic [6,10] activities. Quercetin can be methylated to various methyl ether derivatives by plant OMTs, and several of these products were reported to occur naturally in the wild tobacco, Nicotiana attenuata [11]. Several flavonol-or quercetin-specific OMTs have been reported so far, based on enzyme essays of purified protein extracts, or the molecular cloning of their genes [12][13][14][15][16][17][18].
In tobacco (Nicotiana tabacum L. var. Samsun), three plant OMTs (NtOMT1, 2 and 3) have been reported [19]. NtOMT1 was mainly expressed in healthy tobacco leaves and shown to be involved in lignin biosynthesis; whereas NtOMT2 and its isoform, NtOMT3 were preferentially induced during the hypersensitive reaction to tobacco mosaic virus (TMV). However, NtOMT2 exhibited similar substrate preferences towards quercetin and caffeic acid [19]. The NtOMT2 gene was later cloned and characterized at the molecular level, and both NtOMT2 and NtOMT3 were considered to represent a new class

Abstract
Tobacco NtOMT2 was previously reported as an inducible, o-diphenol O-methyltransferase (OMT) that exhibits similar preferences towards quercetin and caffeic acid as substrates. In this report, we further examine the substrate specificity of NtOMT2 based on enzyme assays, kinetic data and phylogenetic analysis. We demonstrate that NtOMT2 is a flavonol OMT that catalyzes multiple O-methylation of the flavonol quercetin giving rise to its mono-, di-and trimethyl ether derivatives. Mono-methylation of quercetin on the 3-, 3′-/4′-and 7-hydroxyl groups occurs in a random fashion according to the physio-chemical properties of their hydroxyl groups, together with their similar binding energies in the NtOMT2 binding pocket. Homology modeling of NtOMT2 with the Medicago sativa COMT as a template indicates that the H-bonding network between the functional hydroxyl groups and the neighboring residues allowed NtOMT2 to establish a number of energetically similar binding arrangements with slightly different binding modes; thus resulting in a random multiple methylation. NtOMT2 functions as a proximity and orientation agent in using the general base catalysis for multiple methylation of quercetin. This is the first report to indicate that quercetin multiple methylation is catalyzed by a single gene product.  of OMT genes associated with TMV infection, or treatment with the porteinaceous elicitor, megaspermin [20].
Multiple O-methylation of quercetin was first reported more than three decades ago by enzyme extracts of Calamondin orange peel and tobacco cell cultures, and methylation at the 3-position was later shown to be the first step in the enzymatic process [12,21,22]. This phenomenon was attributed, at the time, to the presence of several OMT species in the partially purified protein preparations. Furthermore, multiple glucosylation of quercetin has also been reported by the recombinant proteins of Arabidopsis thaliana [23] and Medicago truncatulas [24], although such phenomenon has not been further investigated.
In this paper, we further characterize NtOMT2 enzyme activity, and examine the multiple methylation of quercetin. We report here that the multiple methylation of quercetin is catalzyed by a single gene product, and occurs in a random fashion according to the physico-chemical properties of its hydroxyl groups, together with their similar binding energies in the NtOMT2 binding pocket.

Chemicals
Quercetin was purchased from Sigma-Aldrich Canada Ltd (Oakville, Ontario). S-Adenosyl-L-[ 3 H] methionine (80 Ci/mmol) was obtained from American Radio labeled Chemicals (St. Louis, MO). Most of the methylated derivatives used in this study were from our laboratory collection, or as generous gift from Professor Eckhard Wollenweber, Darmstadt, Germany. Unless otherwise specified, all other chemicals were of analytical grade.

Expression and purification of NtOMT2 in E. coli
After sequence confirmation, the ORF of NtOMT2 cDNA was amplified and subcloned into the expression vector pET200/D-TOPO (Invitrogen) for in vitro protein expression. The primers used for PCR were: NtOMT2 F: 5'-CACCATGGAATCCTCAACCAAAAGCC; NtOMT2 R: 5'-CTACTTGCAGAATTCCAT. PCR was conducted following the method described earlier [24]. The sequence and orientation of NtOMT2 ORF in the expression vector were confirmed before chemically transformed into E. coli BL21 (DE3) cells (EMD, Darmstadt, Germany) for protein expression and purification as described before [24].

Enzyme assay and identification of reaction products
The standard enzyme assays were performed as previously described [24] using 50 µM of quercetin, its methylated derivatives or selected phenolic substrates, 50 µM AdoMet containing 25 nCi of the 3 H-label, and 0.1 to 2.0 µg of the affinity-purified recombinant protein. To study the influence of pH on NtOMT2, Bis-Tris buffer (pH 6.5), Tris-Cl buffer (pH 7.5) and Bicine buffer (pH 8.5) were used in the enzyme assays.
Characterization of the enzyme reaction products was carried out using TLC, HPLC and LC-MS, as well as comparisons with reference compounds. TLC and HPLC were performed as described earlier [25], except that the solvents used for HPLC were acetonitrile and 10 mM H 3 PO 4 . HPLC was performed in a linear gradient consisting of 50% of each solvent for 30 min. Identity of the reaction products was confirmed by comparing their retention times (R t ) and their UV absorption maxima with those of the reference compounds.
LC-MS analyses were carried using an Agilent 1200 HPLC system with a binary pump, in-line degasser, high performance auto-sampler and thermo-stated column division, using a linear gradient of MeOH-0.1% HCOOH in H 2 O (40:60, v/v) for 30 min with a flow rate of 0.35 ml.min -1 on an Agilent SB-C18 column (2.1×30 mm; particle size, 3.5 µm), and a column temperature of 25°C. The system was optimized for UV detection at 254 nm and 340 nm as previously described [26]. Reference compounds were used to compare the corresponding retention times and mass spectral profiles of the methylated quercetin derivatives.
Kinetic analyses were performed using 10 µg of the affinity-purified NtOMT2 proteins with a saturating concentration (100 µM) of AdoMet, containing 25 nCi of radioactivity, and varied concentrations (5 µM to 80 µM) of the substrates. The assays were performed in triplicates and were repeated twice. Lineweaver-Burk plots [27] were applied for the determination of K m and V max values.

Phylogenetic analysis
To study the phylogenetic relationship among NtOMT2 and other plant OMTs, a number of closely related plant OMT amino acid sequences were aligned with NtOMT2 using Clustal W [28]. Phylogenetic analysis was conducted using MEGA3 [29]. The neighbor-Joining method was employed to generate the tree, and bootstrap analysis (100 replicates) was applied to evaluate the reliability of the tree.

Homology Modeling and Molecular Docking
Homology modeling of NtOMT2 was carried out using MsCOMT [30] as a template, as previously described [31]. Docking calculations were carried out using Docking Server (http:// www.dockingserver. com). The MMFF94 force field [32] was used for a preliminary energy minimization of the ligand molecule (quercetin) using Docking Server. PM3 semi empirical method was used in a second step to optimize the geometry of ligand molecule. PM3 semi empirical charges calculated by MOPAC2009 [33] were added to the ligand atoms. Non-polar hydrogen atoms were merged, and rotatable bonds were defined. Docking calculations were carried out on protein model. Essential hydrogen atoms, Kollman united atom type charges, and solvation parameters were added with the aid of AutoDock tools [34]. Affinity (grid) maps of 20×20×20 Å grid points and 0.375 Å spacing were generated using the Auto grid program [34]. Auto Dock parameter set-and distancedependent dielectric functions were used in the calculation of the van der Waals and the electrostatic terms, respectively. Docking simulations were performed using the Lamarckian genetic algorithm (LGA) and the Solis & Wets local search method [35]. Initial position, orientation, and torsions of the ligand molecules were set randomly. Each docking experiment was derived from 100 different runs that were set to terminate after a maximum of 2500000 energy evaluations. The population size was set to 150. During the search, a translational step of 0.2 Å, and quaternion and torsion steps of 5 were applied. Finally, molecular visualization and analysis were made using PyMOL 0.99 [36].

Site-directed mutagenesis
The mutants were prepared using Quick Change site-directed mutagenesis kit (Stratagene, CA), as described before [31]. The wild type NtOMT2 cDNA in vector pET200/D-TOPO was used as a template for PCR. Primers used for generating mutants are shown in Table 1.

an open access journal
NtOMT1 is involved in lignin biosynthesis; whereas NtOMT2 and its isoform, NtOMT3 are induced by tobacco mosaic virus, indicating that they may have different functions. NtOMT2 cDNA contains an ORF of 1098 bp that encodes a 365 amino acid polypeptide with a calculated molecular mass of 40.5 kDa and a theoretical isoelectric point of 5.27. Sequence analysis at the amino acid level showed that NtOMT1 and NtOMT2 share 53% identity, and exhibit similar relative enzyme activities towards quercetin and caffeic acid [19].
In order to compare NtOMT2 with NtOMT1 and other plant OMTs, we generated an amino acid sequence-based neighbor-joining tree. In this tree (Figure 2), NtOMT2 appeared more phylogenetically distant from NtOMT1 and other COMTs, especially MsCOMT and PtCOMT which utilize both caffeic acid and 5-HFA as substrates, and are involved in lignin biosynthesis. This suggests that NtOMT2 belongs to a different group of OMTs. It is also different from the wheat (Triticum aestivum) TaOMT2, a recently characterized flavone-specific OMT involved in tricin biosynthesis [25].
To re-examine the substrate preference of NtOMT2, we expressed its recombinant protein with a His-tag in vitro and affinity-purified it to near homogeneity ( Figure 3). It was then assayed against a number of potential methyl acceptors, including caffeic acid, 5-HFA and a few flavonoid compounds, based on our phylogenetic analysis and previous reports. The results ( Table 2) show that NtOMT2 had the highest preference for quercetin as substrate, although it also utilized caffeic acid, luteolin (a flavone), myricetin (a flavonol) and 5HFA, albeit to a much lesser extent. It is interesting to note that, compared with quercetin, myricetin was a very poor methyl acceptor, possible because of the steric hindrance caused by an extra hydroxyl group on the B-ring.
To determine the best substrate for NtOMT2, kinetic analysis was performed using quercetin and caffeic acid as substrates. The results indicated that quercetin was the preferred substrate for NtOMT2 based on its K m , V max and Kcat/ Km values ( Table 2). The fact that 5HFA, a good lignin precursor, was the least effective methyl acceptor indicates that NtOMT2 is not involved in lignin biosynthesis. Taken together, these results are in an agreement with our phylogenetic analysis and its physiogical role in the tobacco plant [19].

Characterization of quercetin reaction products
To examine the pattern of quercetin (Q) methylation products, [ 3 H]-labeled enzyme assays were analyzed by TLC and autoradiography (Figure 4), as well as HPLC of preparative scale, non-labeled assays ( Figure 5). In a 10-min incubation, the products of Q methylation were characterized on TLC as 3-MeQ, 7-MeQ, 3′-MeQ, a diMeQ and a triMeQ (presumably 3,7,3′-triMeQ), albeit with different label intensities ( Figure 4). The same pattern of methylated products was also observed on HPLC, except that 3-MeQ was the major compound. These products co-migrated on TLC and HPLC with authentic samples of Q mono-, di-and trimethyl ether derivatives. A comparison of the relative intensity of the [ 3 H]-label in 3-MeQ and tri-MeQ ( Figure  4) and their amounts ( Figure 5) seems to reflect their respective low and high specific activities, and suggests that Q methylation by NtOMT2 proceeds towards the formation of trimethyl derivatives utilizing mono-and dimethyl intermediates, as is evident in a 60-min, steady-state incubation (Figure 4). In fact, when 3-MeQ was used as a substrate, 3,3′/3,4′-diMeQ was identified as the product, whereas 7-MeQ gave rise to 7,3′/7,4′-di-MeQ. However, no product was formed when 4′-MeQ was used as a substrate (data not shown). It should be noted that 3′-MeQ and 4′-MeQ, as well as 3,3′-and 3,4′-di-MeQ are difficult to separate, either on TLC or HPLC, mainly because of their similar physico-chemical properties [11].

Effect of pH on quercetin methylation
The fact that Q methylation profiles differed with changes in pH under standard enzyme assay prompted us to investigate the    phenomenon of multiple methylation in relation to the free energies of deprotonation/free energies of methylation of quercetin hydroxyl groups [37]. Therefore, NtOMT2 enzyme assays were performed in three different pHs (6.5, 7.5 and 8.5) and the reaction products analyzed by LC-MS (Table 3). At low pH 6.5 the major product of methylation was identified as 3′-MeQ, together with a trace amount of 7-MeQ. When the pH was increased to 7.5, the pattern of methylation changed to 3-MeQ> 3,3′-di-MeQ >>7,3′-di-MeQ. At pH 8.5, the major product of methylation was 3′-MeQ, with a trace amount of 7,3′-di-Me derivative (Table 3). These results are in agreement with the reported O-H bond dissociation enthalpy (BDE) where 3-OH> 3′-OH> 7-OH [37]. Therefore, under physiological conditions (pH 7.5) the 3-and 3′-hydroxyls are the more preferred sites of methylation, whereas at lower or higher pHs only the 3′-and 7-hydroxyls are methylated, with the latter being much less favored.

Substrate binding and catalysis
The fact that NtOMT2 catalyzed the multiple methylation of quercetin, prompted us to further investigate this phenomenon at the structural level. NtOMT2, which exhibits 56.1% identity to the crystallized MsCOMT (PDB, 1KYZ) [30], was used in a homology modeling approach to study its structural basis for substrate binding and catalysis. After modeling, quercetin and AdoMet were docked into the substrate binding pocket (Figure 7) using Docking Server. Superimposition of the modeled NtOMT2 with that of MsCOMT revealed that most of the highly conserved residues of NtOMT2 are located in the C-terminal portion of the protein that includes the signatures of most plant Class II OMTs [38,39].
NtOMT2 transfers the methyl group of AdoMet to the hydroxyl groups of quercetin, giving rise to various methylated derivatives and AdoHcy as products. In NtOMT2, AdoMet is bound in a similar manner as in MsCOMT (Table 4). Based on docking results, several H-bonds were observed extending from the ligands and the functional groups of their neighboring residues. The H-bonding network allowed NtOMT2 to establish a number of energetically similar binding arrangements with slightly different binding modes.
In order to determine the critical residues involved in the phenolic substrate binding, different residues such as Leu131, Leu132 and Cys318 were chosen for site-directed mutagenesis based on comparison of the residues surrounding the binding site (Table 5), and the mutant proteins analyzed for enzyme activity. Only mutant protein C318V showed difference with NtOMT2 wild protein when assayed with Q. The HPLC (Figure 6) shows that such mutation resulted in the loss of multiple O-methylation, and the formation of only one product, 3-MeQ, albeit in a small amount. This mutation experiment confirmed the crucial role of Cys318 in the sequential methylation of quercetin by NtOMT2, and conferred credibility to the homology model and docking procedure used.
In most structurally characterized MTs, the transmethylation reaction was reported to be mediated by a general acid-base catalysis using a histidine residue [30,31,40]. In the case of NtOMT2, we propose His271 (equivalent to His269 in MsCOMT), as the deprotonating residue of the substrate (Figure 7). Mutation of His271 to Val completely abolished the catalytic activity of the mutant protein (data not shown), indicating the importance of this residue in methylation. As was with other small-molecule OMTs, methyl transfer in NtOMT2 is proposed to be catalyzed by a nucleophilic attack of the resulting phenolate anion on the reactive methyl group of AdoMet. It is proposed that the methylation sequence proceeds through proximity and orientation effects [41], as

OMT Residues
MsCOMT Met130 Asn131 Phe176 Met180 Ile316 Ile319 Met320 Asn324 NtOMT2 Leu131 Leu132 Phe178 Met182 Cys318 Ile321 Met322 Asn326 well as changes in the hydrogen bonding environment created by the amino acid residues neighboring the different ligands. This would also allow the sequential methyl transfers to take place without the need for transfer of the methylated intermediate to another catalytic site.
Docking experiments indicated that there was no strong, energetically-favorable position for all NtOMT2-quercetin conformers ( Table 6). This assertion is supported by the observed NtOMT2 enzyme activity with quercetin as the substrate (Figure 4 and 5). The sequence of multiple mono-methylation of quercetin is not critical. It is assumed that the methyl transfer reactions can begin at any OH group of quercetin at random, followed by further methylation to give rise to various di-and trimethyl derivatives.

Discussions
Multiple O-methylation of quercetin has been reported to be catalyzed by individual OMTs [12,[16][17][18]. We have demonstrated that NtOMT2, a single gene product, can catalyze the methylation of different hydroxyl groups of quercetin. NtOMT2 is a novel flavonol OMT. It is different from other reported flavonol OMTs, such as Arabidopsis AtOMT1 which catalyzes a single methylation of quercetin to its 3′-methyl derivative [15] or the tomato, trichome-specific OMTs which methylate myricetin at its 3′/5′-or 7/4′-positions [18]. NaOMT2 is quercetin-specific and methylates its 3-, 7-and 3′-hydroxyl groups, albeit with various degrees. The mono-methylated products are further methylated to di-and trimethylated products. While this phenomenon seems to occur in a random fashion, it is dependent on the environmental conditions, especially the pH of the assay buffer. This suggests that the mechanism involved in the multiple methylation is different from the sequential-type methylation that is catalyzed by the wheat flavone TaOMT2. In the latter case, tricetin (5,7,3' ,4' ,5'-pentahydroxyflavone) is methylated to its 3'-methyl-(selgin), 3' ,5'-dimethyl-(tricin) and 3' ,4' ,5'-trimethyl ether derivatives [23]. The acidity and activity of the different hydroxyl groups are dependent on the polarity of the solvent. In polar solvents, position 7 is the most acidic site, followed by position 3′/4′ and 3. This is consistent with the experimental pKa values of 7.7, 8.8 and 9.8 obtained for the three most acidic groups: 7-OH, 4′-OH and 3-OH respectively [42]. On the other hand, at low pH, none of these positions is deprotonated, so that the easiest site to be methylated is clearly the 3′/4′-OH, followed by the 7-OH, 3-OH groups and some di-methylation. When the pH increases, deprotonation occurs on three different sites, which allows the mono-methylated products to be further methylated to di-and even tri-methyl quercetin derivatives, as shown in Figure 4. It was also shown that the extent of flavonoid methylation correlated well with the negative electron densities of the different hydroxyl groups of quercetin as well as other hydroxyflavones [12], as calculated by the complete neglect of differential overlap (CNDO/2) theory [43].
The homology modeling approach used in this study demonstrated that the similar binding energies of the different OH groups in the NtOMT2 binding pocket (Table 6) seem to be responsible for the random fashion of mono-methylation, which is in agreement with our experimental data. Furthermore, the low level of mono-methylation and multiple methylations may be explained by a weak binding affinity of the enzyme.   Finally, methylated flavones have been reported to be most effective compounds for cancer prevention [44]. Among the various quercetin methylated derivatives, 3-MeQ has been shown to have many potential benefits for human beings. In this study, we demonstrated that Cys318 is the most critical residue involved in the multiple O-methylation of quercetin. It was possible to replace this amino acid with a different residue in favour of 3-MeQ production. We are pursuing this line of research with the view to increase the yield of this valuable metabolite.