A Novel Mg2+-dependent O-Methyltransferase in the Phenylpropanoid Metabolism of Mesembryanthemum crystallinum*

Upon irradiation with elevated light intensities, the ice plant (Mesembryanthemum crystallinum) accumulates a complex pattern of methylated and glycosylated flavonol conjugates in the upper epidermal layer. Identification of a flavonol methylating activity, partial purification of the enzyme, and sequencing of the corresponding peptide fragments revealed a novel S-adenosyl-l-methionine-dependent O-methyltransferase that was specific for flavonoids and caffeoyl-CoA. Cloning and functional expression of the corresponding cDNA verified that the new methyltransferase is a multifunctional 26.6-kDa Mg2+-dependent enzyme, which shows a significant sequence similarity to the cluster of caffeoyl coenzyme A-methylating enzymes. Functional analysis of highly homologous members from chickweed (Stellaria longipes), Arabidopsis thaliana, and tobacco (Nicotiana tabacum) demonstrated that the enzymes from the ice plant, chickweed, and A. thaliana possess a broader substrate specificity toward o-hydroquinone-like structures than previously anticipated for Mg2+-dependent O-methyltransferases, and are distinctly different from the tobacco enzyme. Besides caffeoyl-CoA and flavonols, a high specificity was also observed for caffeoylglucose, a compound never before reported to be methylated by any plant O-methyltransferase. Based on phylogenetic analysis of the amino acid sequence and differences in acceptor specificities among both animal and plant O-methyltransferases, we propose that the enzymes from the Centrospermae, along with the predicted gene product from A. thaliana, form a novel subclass within the caffeoyl coenzyme A-dependent O-methyltransferases, with potential divergent functions not restricted to lignin monomer biosynthesis.

ification in natural product biosynthesis. Site-specific O-methylation modulates the physiological properties of such compounds and reduces the chemical reactivity of phenolic hydroxyl groups (1). In plants, O-methylation is also required for monolignol biosynthesis and accounts for the structural differences and properties of lignin, which next to cellulose is the most prominent polymer on earth (2).
Plant OMTs can be categorized into two major classes (3). Class I includes a group of low molecular weight (23,000 to 27,000) and Mg 2ϩ -dependent OMTs, whereas class II consists of higher molecular weight OMTs of about 38,000 to 43,000 that do not require Mg 2ϩ for catalytic activity. Prominent class II members include caffeic acid, flavonoid, coumarin, and alkaloid OMTs (4 -6). Within the class I OMTs, a group of small caffeoyl coenzyme A OMTs (CCoAOMTs) have been suggested to be key enzymes in the biosynthesis of monolignols, the precursors of gymnosperm and angiosperm lignins (7,8). In angiosperms, this task may also be performed by class II OMTs that are specific for caffeic acid, caffeyl aldehyde, or caffeyl alcohol (COMT) (9 -11). This apparent redundancy results in a cell-and tissue-specific metabolic grid that is essential for plants to regulate lignin composition and structure (12)(13)(14)(15). In gymnosperms, the situation could be more complex, with a bifunctional caffeic acid/caffeoyl-CoA methylating activity (AEOMT) and an additional specific caffeoyl-CoA activity found in loblolly pine (Pinus taeda; Refs. 16 and 17). However, studies on transcriptional profiling of monolignol biosynthesis in P. taeda cell cultures question any such difference of methylating enzymes between angiosperms and gymnosperms, and confirm that methylation of guaiacyl residues is the exclusive role of class I CCoAOMTs (18). Other phenylpropanoid esters or flavonoids have never been described as substrates for any of these enzymes (16).
In contrast to plants, animals methylate endogenous metabolites like dopamine or xenobiotics by a soluble and membranebound enzyme derived from a single Mg 2ϩ -dependent catechol OMT gene (19,20). Crystal structures and inhibitor studies of the rat and human catechol OMTs have probed the precise mechanism for substrate binding and methyl transfer with the aim to design pharmacologically effective inhibitors (21,22). Although plant class I OMTs share only 15% sequence identity with their animal counterparts, the cation dependence, important catalytic residues, and topology are all conserved between the plant and animal enzymes (23). The structural similarities of plant class I OMTs with animal catechol OMTs are in contrast to the apparent differences between class I and class II OMTs of plants, as recently elaborated by crystal structures of two alfalfa class II OMTs, which are involved in methylation of flavonoids and caffeic acid (24,25).
Mesembryanthemum crystallinum, the common ice plant, is a member of the herbaceous family Aizoaceae of the Caryophyl-lales. This taxon is characterized by its unique feature with regard to the phenylpropanoid pattern, in which betacyanins have replaced the anthocyanins as chromogenic fruit and flower pigments (26). The ice plant has been intensively used as a model system to analyze abiotic stress (27). Upon irradiation with elevated light intensities, and UV light, the plant accumulates red betacyanins accompanied by elevated levels of methylated and glycosylated flavonol conjugates and feruloylglucose in the upper epidermal tissue (28,29).
In this report, we describe the identification, purification, and functional analysis of a novel Mg 2ϩ -dependent O-methyltransferase, designated PFOMT, from leaves of the ice plant M. crystallinum. In contrast to other plant CCoAOMTs involved in monolignol biosynthesis, PFOMT accepts as substrates a broad spectrum of compounds with a vicinal dihydroxyl structure, and shows a preference for phenylpropanoids like caffeic acid esters and flavonols (Fig. 1). Both are characteristic constituents of chromogenic leaf and fruit conjugates of the ice plant. Characterized by its unique substrate specificity and distinct N-and C-terminal sequence motifs, PFOMT defines a novel subset of class I OMTs, which can be clearly distinguished from other CCoAOMTs to be involved in lignin biosynthesis. This subclass of CCoAOMT-like proteins is apparently not restricted to the Caryophyllales, but, as demonstrated by the cloning and functional expression of a novel Arabidopsis OMT, is also present in other angiosperms.

EXPERIMENTAL PROCEDURES
Plant Material-M. crystallinum L. (ice plant) was grown from seeds for 10 weeks in the greenhouse at a light intensity of 150 mol m Ϫ2 s Ϫ1 for a 16 h/8 h, 25-30°C/21-22°C day/night cycle. Treatment of the plants with light at an intensity of 1000 mol m Ϫ2 s Ϫ1 for 2 days was performed according to Ref. 29.
Enzyme Purification and Amino Acid Sequence Analysis-The flavonol and caffeoyl ester-specific OMT (PFOMT) was extracted from 6 kg of M. crystallinum leaves that were grown at 150 mol m Ϫ2 s Ϫ1 in a greenhouse, crushed in liquid nitrogen, and stored at Ϫ80°C. All extraction and purification steps were performed at 4°C. Extraction buffers contained 5% Polyclar AT (Serva, Heidelberg, Germany), 1 mM 2-mercaptoethanol, 0.4% ascorbate, and 10% glycerol. The crude plant protein extract was passed over Sephadex G-25 (Amersham Biosciences) to remove excess KP i and residual impurities. Forty-eight grams of crude protein were applied to Q-Sepharose (Amersham Biosciences) equilibrated with 10 mM KP i (pH 7.5). Proteins were eluted with a stepwise KP i gradient (10, 50, 150, and 300 mM KP i , pH 7.5) with monitoring the enzyme activity. Active fractions were adjusted to 1 M ammonium sulfate and separated by hydrophobic interaction chromatography on phenyl-Sepharose CL-4B (Amersham Biosciences), equilibrated with 1 M ammonium sulfate in 50 mM KP i (pH 7.5). Proteins were eluted by decreasing concentrations of ammonium sulfate. Active fractions were combined, diluted to 10 mM KP i (pH 7.5), and directly applied to a Mono-Q (HR 10/10) (Amersham Biosciences) column equilibrated with 10 mM KP i (pH 7.5), and then eluted with a linear KP i gradient from 10 to 300 mM KP i . Active fractions were adjusted to 1 M ammonium sulfate and separated by hydrophobic interaction chromatography on analytical high performance phenyl-Superose (Amersham Biosciences). Highly purified enzyme preparations were concentrated (30). Samples of the enzyme preparations were subjected to SDS-PAGE. One fraction of the major Coomassie Brilliant Blue R-250-visualized band representing the methyltransferase was blotted onto a polyvinylidene difluoride membrane (Bio-Rad) for N-terminal sequencing. The major fraction (ϳ2 g) was digested in situ with endopeptidase trypsin. The peptide mixture was separated by reversed phase liquid chromatography on a Phenomenex 5-m C18 column (15 ϫ 0.3 cm). Peptide fragments were separated using a linear gradient from 2 to 45% acetonitrile in 0.06% trifluoroacetic acid at a flow rate of 150 l/min. Amino acid sequence analyses of individual peptides were carried out on a Procise 492 (Applied Biosystems, Weiterstadt, Germany). The digestion, peptide separation, and amino acid sequence analysis was performed by TopLab (Martinsried, Germany). Determination of the molecular weight of partially purified recombinant PFOMT was performed on a High Load Superdex 200 16/60 (Amersham Biosciences) in 150 mM NaCl, 10% glycerol, 100 mM KP i (pH 7.0) at a flow rate of 1 ml/min using bovine serum albumin (67 kDa), ovalbumin (43 kDa), and chymotrysinogen (27 kDa) (Serva, Heidelberg, Germany) as reference proteins.
Enzyme Activity Measurements-Optimized enzyme assays were measured in a buffer containing 100 mM KP i (pH 7.5), 10% glycerol, 5 mM sodium ascorbate, 1 mM 2-mercaptoethanol, and 150 M MgCl 2 , with 10 M substrate (dissolved in 50% Me 2 SO), 0.5-50 g of total protein, and 100 M AdoMet in a total volume of 60 l. The assay was incubated at 37°C for 1 to 15 min (dependent on the amount of protein and substrate tested), after which the reaction mixture was terminated by the addition of 20 l of 3.5% trichloroacetic acid in 50% acetonitrile. The reaction products were analyzed by reversed phase liquid chromatography on a Nucleosil 5-m C18 column (5 ϫ 4 mm inner diameter; Macherey & Nagel, Dü ren, Germany), as described previously (28). Compounds were analyzed with a linear gradient from 10% B (100% acetonitrile) in A (1.5% aqueous phosphoric acid) to 70% B in A (for phenolics), from 5% B to 50% B in A (for free acids and CoA esters), from 5% B to 30% B (for glucose esters), and from 20% B in A to 80% B in A in 4 min (for flavonoids) at a flow rate of 1 ml min Ϫ1 . Detection of flavonoids, catechol, coumarins, and hydroxycinnamic acid esters was performed between 260 and 400 nm. Fluorescent flavonoid derivatives were identified online by a Waters 474 fluorescence detector (Waters, Eschborn, Germany) with an excitation at 370 nm and emission at 520 nm. Identification and quantification was achieved with reference compounds from our Institute collection or from external sources. Various 6-O-and 8-O-substituted quercetin and kaempferol derivatives were provided by E. Wollenweber (Technical University, Darmstadt, Germany) and K. Stich (Technical University, Vienna, Austria). 5-OH ferulic acid was obtained from V. Chiang (North Carolina State University). Quercetagetin, syringetin, eriodictyol, and naringenin were purchased from Extrasynthese (Genay, France). Caffeoyl-and 5-OH feruloylglucose were prepared from caffeic acid, 5-OH ferulic acid, and UDP-glucose with the purified recombinant sinapic acid glucosyltransferase (SGT) from Brassica napus (31) as follows. To 10 ml of 1 mM caffeic acid in KP i buffer (pH 7.0), 10 g of highly purified SGT preparation and UDP-glucose (10 mM final concentration) were added. After incubation for 30 min the reaction was stopped by adding acetonitrile to a final concentration of 20% and the mixture was concentrated by RP 18 solid-phase cartridge extraction (Waters). The concentrated mixture of caffeic acid and caffeoylglucose was further purified by a semipreparative reverse phase-HPLC on Nucleosil-RP 18 (10-mm inner diameter, 25 cm length) (Machery & Nagel) with a 30-min gradient of increasing acetonitrile concentration in 1% acetic acid (5-50% in 30 min, flow rate 2 ml min Ϫ1 ) and evaporated to dryness. Concentrations of caffeoylglucose and 5-OH feruloylglucose were calculated based on a standard of sinapoyl glucose (provided by A. Baumert, IPB) and calculated with an extinction coefficient of ⑀ 330 nm of 23,000. The standard curve for both esters by HPLC detection was linear at least up to 500 M of the corresponding ester. Caffeoyl-CoA, feruloyl-CoA, sinapoyl-CoA, and 5-OH feruloyl-CoA were prepared according to Ref. 32   calculated from the Lineweaver-Burk plots. All enzyme assays were recorded in triplicates. For determination of kinetic data, we adjusted the total protein quantity to 2 g corresponding to ϳ1 g of the corresponding OMT in case of the highly purified PFOMT and AraOMT1. Up to 30 g were used for the native PFOMT and the less purified enzymes from tobacco and chickweed. In the case of the recombinant PFOMT, two (kinetic data) and three (substrate specificities) independent measurements were performed. Standard deviations are shown for the substrate specificity of the PFOMT enzyme only, otherwise the data display the average of the two measurements.
MS Analysis of Quercetagetin Derivatives-The positive and negative ion electrospray (ES) mass spectra and the ES-MS/MS measurements were obtained from a Finnigan MAT TSQ 7000 instrument (electrospray voltage positive ions: 4.5 kV, negative ions: 4.0 kV; heated capillary temperature 220°C; sheath gas nitrogen) coupled with a Surveyor HPLC (ThermoFinnigan) system equipped with a RP18 column (5 m, 1 ϫ 100-mm, SEPSERV). The following HPLC gradient system was Isolation of Full-length cDNAs-Full-length cDNAs encoding PFOMT from M. crystallinum were isolated by screening 500,000 plaques of a cDNA library prepared in -ZAP III (Stratagene, Heidelberg, Germany) using [␣-32 P]dATP-labeled oligodeoxynucleotides. The full-length CCoAOMT cDNA from Stellaria longipes (34) was used as a heterologous probe. Several positive signals were obtained. In vivo excision and transformation of the resulting plasmids in Escherichia coli XL-1 blue MRFЈ according to the manufacturer's instructions revealed several candidate colonies. Plasmids were isolated, digested with restriction enzymes, and those with inserts of the expected size sequenced on an Applied Biosystems sequencer. Two analyzed clones that contained the deduced amino acid sequence of the native, purified and sequenced O-methyltransferase were completely sequenced in both directions using the T3 and T7 primers of the plasmids.
Heterologous Expression of O-Methyltransferases-The full-length cDNA isolated by library screening was digested from the pBluescript SK(ϩ/Ϫ) (Stratagene) with BamHI/KpnI, ligated into BamHI/KpnIdigested expression vector pQE 32 (Qiagen, Hilden, Germany), and sequenced using the pQE promotor primer. The expression constructs were introduced into E. coli strain M15 (pREP4) (Qiagen). A 3-ml preculture was grown overnight at 37°C in LB medium containing 100 g/ml ampicillin (amp). This culture was used to inoculate 400 ml of fresh medium to which 1 mM isopropyl-1-thio-␤-D-galactopyranoside was added after 3 h to induce protein expression. Cells were further grown for 12 h at 30°C. After centrifugation for 10 min at 11,000 ϫ g, the bacteria were resuspended in 50 mM KP i buffer, containing 10% glycerol and 1 mM 2-mercaptoethanol (pH 7.5). Cells were lysed by a combination of a 20-min lysozyme treatment (1 mg/400 ml of culture) and subsequent ultrasonication. The supernatant containing the recombinant PFOMT was adjusted to 1 M ammonium sulfate in 50 mM KP i (pH 7.5), and purified on Phenyl-Sepharose (Amersham Biosciences) with a stepwise gradient of decreasing salt concentration, concentrated by ultrafiltration on PM 50 membranes (Waters), followed by metal affinity chromatography on Talon (BD Biosciences) with a stepwise gradient of increasing imidazole concentrations. Partially purified protein was used for enzyme assays, SDS-PAGE, and Western blots. Western blots were performed on polyvinylidene difluoride membranes with a monoclonal anti-His antibody coupled to alkaline phosphatase (Novagen, Madison, WI) and detected by the chromogenic bromochloroindolyl phosphate/nitro blue tetrazolium complex. The S. longipes cDNA was cloned in pBluescript (34,35), and protein expression in E. coli XL-1 blue was induced with 2 mM isopropyl-1-thio-␤-D-galactopyranoside at 30°C overnight. The tobacco CCoAOMT cDNA kindly provided by U. Matern (University of Marburg, Germany) was cloned into the vector pET21 (Novagen), and protein expression was induced in E. coli BL21(DE3) (Stratagene) with 0.8 mM isopropyl-1-thio-␤-D-galactopyranoside at 30°C overnight. The phenyl-Sepharose CL-4B and Mono-Q anion exchange purified protein fractions were used for the analyses of kinetic data and substrate specificity of these recombinant enzymes.
Cloning and Expression of the Arabidopsis PFOMT-like Gene-Based on sequence information obtained from GenBank TM two specific primers were designed corresponding to the 5Ј-end (5Ј-CGGGATCCATG-GCTAAGGATGAAGCCAAG-3Ј with a BamHI restriction site) and 3Јend (5Ј-CTCAAGCTTGTGTCAATATAACCTCCTGCAAATAG-3Ј with a KpnI site) of the A. thaliana nucleotide sequence. RNA from the whole plants of A. thaliana Columbia-ecotype Col-0 1092 (obtained from the Arabidopsis Stock Center, Nottingham, United Kingdom) was isolated based on a kit provided by Qiagen. 5 g of total RNA was reversed transcribed with Moloney murine leukemia virus-RNase H(Ϫ) reverse transcriptase (Promega) and the corresponding cDNA was amplified to yield a 700-bp fragment, which after digestion with BamHI and KpnI was ligated in-frame into the vector pQE 30 (Qiagen) and transformed into E. coli M15rep 4 cells. Several amp (ϩ) -resistant colonies were induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside at 37°C for 4 h. Bacteria were lysed and soluble protein was assayed for OMT activity and in parallel by SDS-PAGE. Bacteria from the most active preparation were grown on a 1-liter scale, bacteria lysed, and the His-tagged protein was partially purified by Talon affinity chromatography according to the manufacturers instructions (Clontech). 2 g of partially purified protein per assay were used for determination of kinetic parameters. The correct sequence of the corresponding gene was confirmed by sequencing using a pQE promotor primer.
Northern Blot Analysis-Total RNA was isolated from 200 mg of leaf tips and epidermis cells from plants grown in normal or high light conditions, using Qiagen RNeasy mini kit. Northern blots were done according to standard procedures using PFOMT cDNA as a probe and the Mega-Prime TM labeling kit (Amersham Biosciences). Hybridization was performed under stringent conditions with a buffer containing 7% SDS, 250 mM NaCl adjusted to pH 7.0 with 250 mM KP i .
Sequence Analysis and Alignments-Sequence analysis was performed with National Center for Biotechnology Information (NCBI) BLAST Search programs. Sequence alignments were performed with the Lasergene software program (DNASTAR, Madison, WI).

Purification of Native PFOMT and Amino Acid Sequence
Analysis-A crude enzyme preparation from irradiated leaves showed only slightly higher enzyme activities as compared with non-irradiated tissues (data not shown). The amount of leaf tissue needed for purification of the enzyme and limitations in time and space to grow a large number of plants under elevated light conditions prompted us to purify the protein from the plants grown in the greenhouse under normal light conditions (150 mol m Ϫ1 s Ϫ1 ). PFOMT was purified by a combination of ammonium sulfate precipitation, anion exchange, and hydrophobic interaction chromatography (Table I) with a total recovery of less than 1%. The final purified and concentrated protein preparation, which corresponded to the activity profile, yielded two bands at ϳ29 and 48 kDa, as revealed by SDS-PAGE (Fig. 2a). N-terminal sequence analysis revealed that the 29-kDa protein was similar to several plant OMTs (data not shown). Additional sequence data from three internal peptides from the 29-kDa protein showed sequence similarities to the CCoAOMTs of class I OMTs. The highest identity was to an OMT from the taxonomically closely related chickweed, S. longipes, which also belongs to the Caryophyllales (34,35). It seemed possible that the 48-kDa band detected by the SDS-PAGE (Fig. 2a) might be a flavonoid OMT because of its similar size to the average relative monomer molecular mass of 38 to 43 kDa for class II OMTs (3). To verify this possibility, the 48-kDa protein was end-sequenced. Sequence comparisons, however, identified it as a transketolase-like protein.
Cloning and Expression of the PFOMT cDNA-A cDNA library was constructed from total RNA from leaves irradiated with elevated light intensities. Based on the peptide amino acid sequences and various combinations of the corresponding degenerate primers used in reverse transcriptase-PCR, we tried to obtain the cDNA encoding PFOMT. This approach was not successful. Therefore, the CCoAOMT cDNA from S. longipes (34) was used to screen this cDNA library. Several full-length clones with identical open reading frames of 711 bp were obtained and sequenced. All the sequenced peptide fragments from the native plant OMT could be located in the deduced protein sequence of the cDNA clones (Fig. 3). Therefore, the gene for the PFOMT cDNA indeed encodes the PFOMT protein that was extracted and purified from the ice plant.
The predicted PFOMT protein sequence consists of 237 amino acids, with a calculated molecular mass of 26.6 kDa, which is similar to the size (29 kDa) of the native protein estimated by SDS-PAGE (Fig. 2a), and a theoretical isoelectric point of 4.88. The PFOMT cDNA was expressed in E. coli. The recombinant protein was purified with an affinity column and visualized by SDS-PAGE (Fig. 2b). A major band with a molecular weight of 32,000 was observed and was recognized by an anti-His antibody (Fig. 2b). Its larger size (32 kDa) compared with the native form (29 kDa, Fig. 2a) is because of the presence of 50 additional N-terminal amino acids, including the His tag present only in the recombinant PFOMT, amino acids derived from the vector sequence, and 11 N-terminal amino acids, which are absent in the native protein. Additional bands of smaller sizes that were not recognized by the anti-His tag antibody on Western blots were also revealed by SDS-PAGE (Fig. 2b). To determine whether these bands represent truncated versions of the PFOMT or contaminating proteins, two of these bands (a 30 and 27 kDa proteins, Fig. 2b) were N-terminal sequenced. The data suggest that the 30-kDa band is a  14-amino acid shorter, and His tag-free version of the recombinant PFOMT. The corresponding N-terminal sequence reads Ala-Ala-Gly-Ile-Arg-His-Glu-Ala, is derived from the pQE32 vector, and is located 5 amino acids downstream from the His tag. The other band (27 kDa) was also sequenced and showed 100% homology to the N terminus of E. coli tagatose-bisphosphate aldolase.
The protein activity could be concentrated by ultrafiltration through a 50-kDa membrane. Further gel filtration analysis using bovine serum albumin, ovalbumin, and chymotrypsinogen as standard molecular mass markers indicated a molecular mass of 54 Ϯ 3 kDa for the recombinant PFOMT protein, which is consistent with a dimeric structure, as proposed previously for CCoAOMTs from other plant species (36). The PFOMT amino acid sequence did not contain any known protein cleavage sites, nor signaling peptides as predicted by the program Signal P (37). Protein sequencing showed that not only was the N-terminal methionine (Met) absent in the native PFOMT protein, but that the N terminus of the native form started with a threonine, corresponding to number 12 from the amino acid sequence (Fig. 3). In contrast to the lack of any cleavage or signaling sites, several significant serine and tyrosine phosphorylation sites could be detected with the NetPhos 2 software package (38). Specifically, Ser 32 , Ser 50 , and Ser 54 are likely to be phosphorylated in vivo (Fig. 3). Phosphorylation may not be essential for PFOMT activity, because expression of the PFOMT in E. coli resulted in functionally active and soluble proteins.
Functional Characterization of the Native and Recombinant PFOMT-All enzyme characteristics were measured with partially purified native (anion exchange and phenyl-Sepharose) or metal affinity purified recombinant PFOMT. The results obtained were comparable between the native and recombinant PFOMT protein, except for some minor differences in cation preference, regioselectivity, and substrate preference.
The native PFOMT showed an apparent temperature maximum at 37 to 40°C, a maximum activity around 150 M Mg 2ϩ , and a broad pH optimum between pH 7 and 9. The temperature optimum of the recombinant enzyme was determined at 45 to 50°C. As expected, incubation with 1 mM EDTA dramatically reduced both enzyme activities (data not shown). When different bivalent cations were used with the recombinant enzyme, up to 1.5-fold stimulation of enzyme activities, as compared with Mg 2ϩ , was observed with Ca 2ϩ , Co 2ϩ , and Zn 2ϩ at concentrations from 100 M to 1.5 mM. If Mn 2ϩ instead of Mg 2ϩ was included, the enzyme activity did not change significantly. This is different from the results obtained with native PFOMT (data not shown) and the native tobacco CCoAOMT, where Co 2ϩ , Ca 2ϩ , and Mn 2ϩ were shown to reduce the activity, in the case of Co 2ϩ as much as 50% (36). Both native and recombinant PFOMT activities were irreversibly inhibited in the presence of 20 M Cu 2ϩ and could not be restored by subsequent incubation with an excess of other cations (data not shown). In agreement with previous reports on the Mg 2ϩ -dependent CCoAOMTs, we carried out all measurements of substrate specificity and kinetics with Mg 2ϩ as the bivalent cation. It is worth noting that the term "Mg 2ϩ -dependent" may be misleading, at least in the case of the recombinant PFOMT, because Mg 2ϩ is not the sole bivalent cation able to promote enzyme activity.
Substrate Specificities of Native and Recombinant PFOMT-Up to now, all class I OMTs were reported to methylate only caffeoyl-CoA. However, our preliminary study showed that the partially purified PFOMT methylated both caffeoyl-CoA, and the flavonols 6-OH kaempferol and quercetagetin in a noncompetitive manner. This initially implied the presence of two individual enzymes, which is consistent with the accepted classification of caffeoyl-CoA methylating enzymes (class I) and flavonoid methyltransferases (class II) (3). Surprisingly, further purification of the enzyme revealed that the caffeic acid ester and the flavonol activity always co-purified. The very low amounts of protein obtained after the final purification step were only sufficient to obtain enough material for sequencing and therefore no competitive data could be obtained. N-terminal sequencing clearly indicated that only one protein could be responsible for methylating both substrates. To unequivocally clarify this puzzle, we cloned and expressed the respective enzyme and proved the multifunctional properties of the ice plant PFOMT in vitro. A wide array of potential methyl acceptors were tested to determine substrate preference and apparent kinetic data (Table III). With AdoMet as the methyl group donor, only phenylpropanoids and esters with aromatic vicinal dihydroxy groups acted as methyl acceptors by PFOMT (Table  II). Flavonoids such as quercetin, the corresponding 3-O-glucoside, and 6-hydroxykaempferol were all methylated, as were flavones and flavanones with a vicinal dihydroxy system including luteolin and eriodictyol. No products were observed with que-4Ј-O-methyl ether, quercetin-4Ј-O-glucoside, luteolin-4-O-methyl ether or kaempferol, nor with syringetin, apigenin, or naringenin, which are all characterized by the lack of vicinal dihydroxy groups (Fig. 4). With quercetin, only the meta posi- tion of the B-ring was methylated and hence the 3Ј-O-methyl ether (isorhamnetin) was formed, as identified by co-chromatography with an authentic standard. Concentrations higher than 20 M quercetin in the assay were inhibitory.
The major flavonol aglycone found in the ice plant after induction with high light intensities was identified as 6-hydroxyquercetin (quercetagetin), which contains a total of six hydroxyl groups. With the native PFOMT, we identified only two products corresponding to the 6-mono-and 6,3Ј-dimethyl ether of quercetagetin, with retention times of 2.6 and 3.1 min, respectively. In addition to co-chromatography with authentic standards and the UV spectra, structures of these products were verified by positive and negative ES-MS as well as LC-ES-MS/MS measurements. The molecular masses of the substrate quercetagetin (peak 2) and two corresponding products at 2.6 and 3.1 min were identified at m/z 318 (S), 332 (peak 5), and 346 (peak 6), and are consistent with the subsequent addition of one or two methyl groups to the quercetagetin aglycone (Fig. 4). Peak 5 could be correlated to the 6-O-methylated quercetagetin standard, based on retention times and its ES-MS/MS data both under positive and negative ionization. The dimethylated product ([M ϩ H] ϩ at m/z 347) gave identical retention times on HPLC as compared with the previously characterized, acid-hydrolyzed 6,3Ј-O-dimethylated and glycosylated quercetagetin conjugate isolated from the ice plant (28). In addition to the 6-mono-and 3Ј,6-dimethylated derivatives, very minor amounts of the 3Ј-O-methyl ether (retention time at 2.5 min) were formed in vitro by the native enzyme (tested at pH 6 to 8) and identified by co-chromatography with an authentic standard.
When quercetagetin was used as a substrate for the recombinant enzyme, a total of five products were identified (Fig. 5). In addition to the 6-O-, the 3Ј-O-(now also identified by MS/MS data), and the 3Ј,6-di-O-methyl ether, two compounds not previously detected were observed, with retention times of 1.8 and 2.3 min, respectively. Positive and negative ESI-MS data suggested the first to be another monomethylated (m/z 332), and the second to be a second dimethylated quercetagetin derivative (m/z 346). In contrast to all other quercetagetin derivatives obtained for the native PFOMT, both compounds displayed a striking yellow fluorescence (emission max 520 nm) when irradiated with long wavelength UV/blue light (340 to 380 nm) (Fig. 5). This yellow fluorescence is characteristic of 3-hydroxy-5-O-methylated flavonols (39). The novel monomethylated compound was therefore identified as quercetagetin-5-O-methyl ether, the dimethylated compound as 5,3Ј-di-O-methyl ether, based on identical retention times with the fluorescent, enzymatically prepared methylated product of quercetagetin-3Ј-Omethyl ether. It is noteworthy that upon 5-O-methylation, we never observed the formation of any trimethylated product that TABLE II Structures and HPLC retention times (substrates, S and products, P) on C 18 Nucleosil (5 cm ϫ 4 mm, flow rate 1 ml/min) and relative reaction rates with native and recombinant PFOMT for O-methyltransferase activity Quercetin (Que) at 10 M was set to 100%. Retention times of structures 1 (except for the glucosides) were determined with an acetonitrile gradient from 5 to 50% (caffeic acid and CoA esters), from 5 to 30% (glucose esters), and from 10 to 70% (all others) in 1% H 3 PO 4 ( max at 260 to 380 nm). R 1-11 residues as per Fig. 4.

Substrate
Structure, (Fig. 1 Table II. would be possible by additional methylation of the 7-hydroxyl group. Concentrations higher than 15 M quercetagetin caused a marked loss of enzyme activity, because unspecified denaturing effects of the polyhydroxylated, tannin-like quercetagetin may lead to irreversible protein damage, impeding correct docking and turnover of any substrates (40).
The recombinant PFOMT activity in vitro apparently showed a broader regiospecificity toward quercetagetin than did the native protein (Table II and Table III). This may be caused by the presence of the extra residues in the recombinant protein (including the His tag), but may also be because of the shortened N terminus of the native form as compared with the recombinant protein. We noticed some proteolysis and degradation of the His tag during purification of the PFOMT protein. Activity tests of the His tag-free protein fraction (30 kDa), which did not bind to the affinity column, revealed the same substrate and regiospecificity with quercetagetin as the Histagged PFOMT, suggesting that the six histidine residues may not cause the difference in regiospecificity. Alternatively, phosphorylation of the native protein may be responsible for the observed difference in regiospecificity, although to the best of our knowledge this has never been described before for any plant protein involved in natural product biosynthesis. In a preliminary experiment, a 15-min preincubation of 25 g of a crude protein preparation from M. crystallinum with 2 g of recombinant PFOMT and ATP did not alter the observed regiospecificity of the enzyme assay (data not shown).
PFOMT also catalyzed the formation of two products with the flavonol myricetin, suggesting a 3Ј-monomethylation and a 3Ј,5Ј-dimethylation (Table II). The putative dimethylated compound showed identical retention times and spectra to the commercially available, myricetin-3,5-di-O-methyl ether syringetin (data not shown). The formation of the dimethyl ether increased with prolonged reaction time until myricetin and the monomethylated product were nearly completely converted to the dimethylated derivative. The same holds true for gossypetin with two vicinal dihydroxy systems (Fig. 2, Table II), where three different products were detected: gossypetin-8-O-methyl ether (as the major product) and 8,3Ј-di-O-methyl ether, both of which could be identified based on spectral data and co-chromatography with an authentic standard. The third minor product might be the expected gossypetin-3Ј-O-methyl ether. Dual methylation of a single compound apparently is possible, as long as the acceptor has one available dihydroxyl moiety. Similar dual methylation specificity observed for the PFOMT has only recently been described for some class II OMTs involved in monolignol or flavonoid biosynthesis (41,42). This dual methylation of the same substrate by a single enzyme is surprising, but may be because of a large and flexible active site that accepts molecules of different size and polarity. Such a flexible active site has been described previously for the crystal structures of the class II lignin monomer OMT (25). Formation of dimethylated compounds in case of the PFOMT was time dependent, suggesting that after an initial methylation step, a second step may occur, if the compound still fits the active site. It is interesting to note that methylated substrates with dihydroxy systems, like quercetin-7-O-methyl ether, gossypetin-7,8-di-O-methyl ether, and 5-OH ferulic acid, appear to be even better substrates than their corresponding non-methylated derivatives quercetin, gossypetin, or caffeic acid, the latter lacking a complete methoxy group. However, the precise mechanism to explain this phenomenon awaits structural information about the active site.
Only after sequencing of the native PFOMT did the similarity to CCoAOMTs become evident. Consequently, the CoA and glucose ester of caffeic acid were used as substrates. Caffeoylglucose was not tested with the partially purified native PFOMT extracts. Both compounds were methylated by the recombinant enzyme with an efficiency similar to the flavonols (Table II). Consistent with previous data, feruloyl-CoA and feruloylglucose were not substrates of the PFOMT, but 5-hydroxyferuloyl-CoA and the feruloylglucose were methylated as expected.
Kinetic Properties of Native and Recombinant PFOMT-Apparent kinetic data on a selected set of substrates for both the native and recombinant PFOMT are summarized in Table III. It should be considered that these measurements were not performed with homogenously purified proteins. This also holds true for the data obtained with the Arabidopsis thaliana, the chickweed, and the tobacco proteins. Therefore, no K cat data were determined and the V max measurements were calculated as specific activities of the different proteins analyzed. As stated in the legends to Tables III and IV (see below), differences between individual proteins reflect differences in purification stage and not necessarily differences in the K cat .
The best in vitro substrates are the glucose ester of caffeic acid (only tested with the recombinant enzyme) and the flavonols quercetin and quercetagetin, which had similar V max /K m ratios for both the native and recombinant PFOMT. However, the recombinant PFOMT in vitro efficacy toward caffeoyl-CoA apparently is reduced as compared with the native PFOMT by a factor of 10. This discrepancy may be explained by the presence of an additional caffeoyl-CoA-specific OMT (CCoAOMT) in the partially purified native protein extracts of the ice plant, which were used for the determination of kinetic data. This additional activity, most likely highly specific for caffeoyl-CoA would lead to an overestimation of the affinity of native PFOMT for this substrate only. A potential candidate for this activity, a putative, corresponding CCoAOMT gene and protein sequence have already been deposited in the data bases (Gen- Bank TM accession number T12206). It shows a more than 90% amino acid sequence identity to the monolignol specific CCoAOMT from tobacco, but only a 69% identity to PFOMT. This CCoAOMT cDNA could be obtained from H. Bohnert (University of Illinois). Several attempts to functionally express the corresponding protein and confirm its specificity for caffeoyl-CoA in our laboratory were unsuccessful up to now. The recombinant protein was always found in inclusion bodies and activity could not be detected (data not shown). However, based on sequence identities it is different from the PFOMT reported here and likely belongs to the group of CCoAOMT-like proteins involved in lignin biosynthesis (included in the phylogenetic tree in Fig. 6). Therefore, the data obtained for the recombinant protein seems to better represent the affinity for this substrate, although one should keep in mind that the recombinant PFOMT is not identical to the native plant enzyme at the N terminus.
Although not a substrate for the native enzyme, caffeic acid and 5-OH ferulic acid are potential substrates of the recombinant PFOMT (Table II). Moreover, the kinetic data for caffeic acids indicate a lower in vitro reaction rate (Table III). As seen in Table II, the native enzyme apparently showed a somewhat more stringent structural requirement for substrates than the recombinant enzyme. Low molecular compounds with catechollike structures such as caffeic acid, the coumarin esculetin, or catechol itself were not accepted as substrates by the native PFOMT, even after prolonged incubations. In contrast, the recombinant enzyme methylated all three compounds above, albeit with a low reaction rate (Table II). The reason for this discrepancy is unknown, but the difference in the N-terminal region between the native and recombinant PFOMT or a putative phosphorylation or other modifications of the native form might be responsible for some of the observed functional variations. On the other hand, it is not unusual to observe differences in substrate affinities between native and recombinant forms of the sample protein (43).

Substrate Specificities of Other CCoAOMT-like Proteins-
Based on the novel specificities observed with the recombinant PFOMT, two other recombinant CCoAOMTs, the chickweed CCoAOMT (34,35) and one of the "typical" caffeoyl-CoA-specific CCoAOMTs from tobacco (44), were partially purified and assayed for substrate specificity. The specificity of the heterologously expressed OMT from chickweed was identical to the recombinant PFOMT. In addition to its known caffeoyl-CoAmethylating properties, the partially purified recombinant protein also methylated flavonoids, caffeic acid, and caffeoylglucose with the efficiency comparable with the PFOMT (Table  IV). The same five different reaction products (queg-5-Omethyl ether, queg-3Ј-O-methyl ether, queg-6-O-methlyether, queg-5,3Ј-di-O-methyl ether, and queg-6,3Ј-di-O-methyl ether) were also observed with quercetagetin as substrate (data not shown). This is not surprising because PFOMT and the chickweed OMT share the closest sequence similarity among plant class I OMTs. Substrate preference of the tobacco enzyme, however, was different. In contrast to the recombinant PFOMT from the ice plant and the enzyme from chickweed, the tobacco enzyme did not use caffeoylglucose or caffeic acid as substrates (Table IV). Unexpectedly, this enzyme methylated not only caffeoyl-CoA, but also quercetin although with a 10-fold lower affinity (Table IV). Other flavonoids with vicinal dihydroxyl groups, like quercetagetin, myricetin, or luteolin were also methylated by the tobacco OMT (data not shown). A difference in regiospecificity compared with the enzymes from the ice plant and chickweed is also seen for methylation of quercetagetin, because only the 3Ј-O-methyl ether was formed (data not shown). This flavonoid methylating activity has never been reported for cation-dependent class I OMT from plants. The same high specificity for caffeoyl-CoA was also observed for a "classical" CCoAOMT from grapevine (data not shown), which together with the sequence alignment data, led us to speculate that the second CCoAOMT-like protein from the ice plant (T12206), although not yet functionally analyzed, may also display a comparable specificity. Based on a sequence comparison and a bootstrapped phylo-   genetic tree (Fig. 6) we were able to identify, clone, and heterologously express the third potential candidate of the proposed novel subclass, a CCoAOMT-like protein from A. thaliana (GenBank TM accession number AAM64800). The corresponding protein, in stark contrast to the tobacco and grapevine OMT, displayed a substrate specificity very similar to the enzymes from chickweed and the ice plant, accepting caffeoylglucose, caffeoyl-CoA, caffeic acid, and quercetin as substrates (Table IV). The affinity for all substrates is very similar, but V max for caffeoylglucose and caffeoyl-CoA is higher than that for the flavonoid and caffeic acid. Quercetagetin is also accepted as a substrate, but in contrast to the enzymes from the ice plant and chickweed only two major products are observed, Phylogenetic Analysis of the OMTs-Using several Blast algorithms (45), the amino acid sequence of PFOMT was compared with those of other OMTs available in GenBank TM / EMBL. The results show that PFOMT shares the highest sequence similarity with the enzyme from chickweed and one previously uncharacterized gene product from A. thaliana with an identity of 71 and 53%, respectively (Fig. 6). Several other members of the O-methyltransferase family were chosen from the data base to construct the cladogram. Bootstrapping (1000 trials) of this unrooted tree (there is no clear indication of any common ancestor to all class I OMTs) demonstrated with a high probability that the three novel PFOMT-like sequences represent a small subset of proteins clustered in a separate branch of the otherwise highly homologous CCoAOMTs presumably involved in lignin monomer biosynthesis. The latter include the putative CCoAOMT from the ice plant (GenBank TM accession number T12206), one from A. thaliana (GenBank TM accession number AAL09793), the characterized enzymes from tobacco (44) and others that may all show a strong preference for caffeoyl-CoA (Fig. 6). Therefore, sequence analysis and the deduced phylogenetic results support the functional classification described in this report. Other class I, Mg 2ϩ -dependent OMTs from two different cyanobacteria, from rat, human, yeast, and several plant OMTs (class II) form distinct branches within the tree. The rat, human, and yeast catechol class I OMT show only 15% amino acid sequence identities to the PFOMT. The functionally not characterized putative CCoAMTs from the cyanobacteria Nostoc sp. and Synechocystis sp. display up to 38% identity to the PFOMT, and contain several highly conserved motifs (Fig. 2). A similar specificity for vicinal dihydroxy systems may be postulated for these enzymes.
Induction of PFOMT Gene Expression-To investigate the expression of the PFOMT gene in the ice plant, and to correlate these data with the observed accumulation of flavonol (quercetagetin) and betacyanin conjugates in the ice plant (28,29) we carried out Northern blot analysis. Consistent with low levels of flavonols and betacyanins, a low level of PFOMT mRNA in the leaf tissue of the ice plant grown under normal conditions was observed, suggesting a constitutive pattern of the PFOMT gene expression (Fig. 7). The low RNA level also correlates to the presence of low amounts of protein in tissue grown under greenhouse conditions, which prompted us to isolate the PFOMT protein from this leaf extract. However, when the plants were irradiated with light intensities up to 1500 M s Ϫ1 m Ϫ1 for 2 days (28), the level of leaf PFOMT mRNA increased significantly (Fig. 7). This indicates that the PFOMT gene is inducible by light. The same dramatic accumulation is seen for the potential PFOMT products, the 3Ј,6di-O-methylated quercetagetin derivatives, feruloylated betacyanin conjugates, and feruloylglucose in epidermal layers of  the ice plant within the first 4 days of high light exposure (28,29). This strongly suggests that the PFOMT may indeed be involved in the methylation of these compounds. Similar responses to biotic or abiotic stresses have been observed on the transcript level for several other CCoAOMT-like genes (36,44,46,47). DISCUSSION In contrast to animal cells, plants are reported to contain several classes of OMTs involved in modification of phenolic compounds, with a clear cut difference in substrate preference (4). In this report we present several lines of evidence that methylation of flavonoid structures is not restricted to class II enzymes (4,25,48). The cation-dependent class I enzymes, the so-called "caffeoyl CoA-specific enzymes," also methylate other o-hydroquinone structures. We could demonstrate that all CCoAOMTs, such as the enzyme from tobacco, in addition to their preference for caffeoyl-CoA, are also able to methylate flavonoids. However, the CCoAOMTs do not represent one uniform cluster of enzymes with identical substrate specificity. Among all CCoAOMTs, we propose a novel subgroup of ohydroquinone-specific enzymes of class I OMTs from plants. These enzymes can be distinguished from the highly homologous CCoAOMTs, presumably involved in lignin biosynthesis, by a lower sequence homology and a more promiscuous substrate and regiospecificity. The substrate preferences of this new subcluster of OMTs imply that they function not only in lignin biosynthesis, but may also be involved in the biosynthesis of soluble, esterified phenylpropanoid and flavonoid conjugates. So far, we have identified and characterized three members of the new subclass of OMTs, two enzymes from the Centrospermae and one from Arabidopsis, a member of the order Brassicales (Fig. 6). Further protein sequences can be identified in the data bases based on motifs close to N and C termini as illustrated in Fig. 3, which distinguishes them from the tobacco enzyme and other plant classical class I CCoAOMTs characterized so far (data not shown).
The unique specificity of the newly characterized class I subgroup for caffeoylglucose, and the enhanced affinity for flavonoids and caffeic acid suggest several differences in the protein tertiary and quaternary structure and a more flexible substrate binding site, as compared with the highly conserved caffeoyl-CoA-specific enzymes. As expected, the PFOMT contains all amino acids that are shown to be essential to bind caffeoyl-CoA and existed in the tobacco class I OMT (23), including the highly conserved Arg 218 (Arg 220 in the tobacco sequence). Because of the high sequence identities, it is difficult to predict the reason for the observed discrepancies in substrate and regiospecificity between the two CCoAOMT clusters. It appears unlikely that the variable C-terminal motif of the new subclass is essential for substrate binding. Nevertheless, PFOMT-like enzymes lack two otherwise highly conserved Arg residues and one Lys in the proposed ␣-helix 8 close to the C terminus, which were initially suggested to be essential for caffeoyl-CoA binding, but later rejected based on deletion and modeling studies with the tobacco enzyme (23). In contrast, the more divergent N termini compared with the classical CCoAOMTs may be one of the reasons for the observed less stringent substrate specificity of the new subcluster. The differences in the quercetagetin methylation pattern between the native and the recombinant enzymes may be regarded as a first hint to its potential role in regiospecific turnover of selected substrates. Of course, single amino acid changes, as observed in two closely related class II phenylpropene OMTs, may determine the different specificities (49).
For the newly described specificity for flavonoids, the class I OMTs resemble the corresponding catechol OMTs from animals. For example, the animal liver COMTs in vitro display a thousand-fold higher affinity toward potentially mutagenic flavonols, like quercetin, than toward the endogenous catecholamines like dopamine (1). Although the PFOMT from the ice plant shares only 15% sequence identity with the rat catechol OMT, the overall topology, the catalytically active site, substrate binding, and orientation may be evolutionary conserved among plant and animal class I proteins (3,21,23). Pig catechol OMT even methylates caffeic ester derivatives, which are typical plant constituents (23). In addition, all class I proteins display a conserved position specificity toward the metaposition (1,50). These specificities may be related to several highly conserved amino acids suggested to be involved in substrate binding, including Asp 154 and the Lys 157 of PFOMT (Fig.  3), which are equivalent to Asp 141 and the catalytic Lys 144 in rat catechol OMT (21). These residues are also conserved in the two uncharacterized class I OMTs from cyanobacteria (Fig. 6). Unusually high sequence identities (90 to 95%) observed among the lignin monomer-specific CCoAOMTs may reflect the evolutionary need to methylate preferentially one substrate only, caffeoyl-CoA. A 50 to 70% identity, as observed in the novel PFOMT subcluster, appears to be more typical for enzymes with a broad substrate tolerance, such as glucosyl-or acyltransferases (51,52).
Our results clearly indicate the presence of two different class I CCoAOMT-like genes in both the ice plant and A. thaliana (Fig. 6). One member is part of the cluster of lignin monomer biosynthesis, and the other describes a novel cluster with an apparent broad substrate specificity. The question may arise as to why the plants should contain two similar enzymes to perform identical methylations. Considering the similar apparent kinetic constants of the purified and expressed PFOMT for o-dihydroxylated flavonols, caffeoylglucose, and caffeoyl-CoA (Tables II and III), it is plausible that all of these compounds are potential in vivo substrates. The complex flavonol conjugates, which accumulate in the epidermal layers of the ice plant after irradiation with high light intensities, are not only methylated at the quercetagetin structure, but also contain one or, in the case of betacyanin conjugates, several ferulic acid moieties attached to individual sugars of the glycosylated compounds (28). The activity of PFOMT, which may be epidermally localized, would theoretically be sufficient to methylate caffeoylglucose, caffeoyl-CoA, and the quercetagetin aglycone. The rapid induction of the PFOMT transcript after exposure to elevated light conditions and the simultaneous accumulation of several betacyanin and flavonol conjugates are consistent with the observed pattern of light inducible accumulations of feruloylated betacyanins and feruloylated and methylated flavonols (28). Glucose esters (such as feruloylglucose) are considered substrates for acyltransferases in betacyanin biosynthesis FIG. 7. Induction of the PFOMT expression in leaves by light intensity. Total RNA (10 g), extracted from leaves of the ice plant harvested before and after 2 days of treatment with light intensities of 1500 mol s Ϫ1 m Ϫ1 was blotted on nylon and probed with the full-length PFOMT cDNA. (53), whereas the hydroxycinnamoyl-CoA esters are generally involved in the biosynthesis of phenylpropanoid conjugates (54,55). The remarkably high level of feruloylglucose present in leaves of the ice plant after high light irradiation (29) correlates with our observation of a preferred in vitro affinity of PFOMT for caffeoylglucose and suggests the role of PFOMT is to deliver the methylated precursors in the formation of feruloylated betacyanin conjugates. Therefore, methylation apparently can take place at the level of the glucose or the CoA ester, and not only at the level of the free caffeic acid. Enzymes like PFOMT would provide feruloylglucose for the formation of betacyanins (specific for the Caryophyllales) and feruloyl-CoA to acylate flavonol glycosides, respectively, and therefore could play a key role in metabolic regulation.
How the PFOMT-and other CCoAOMT-like proteins contribute to the overall methylation pattern of plant natural products in the ice plant and other plants, like A. thaliana remains to be investigated. Their specificities, especially those of the new subcluster, overlap with class II OMTs. Recent studies described a multifunctional class II OMT from ripening Fragaria vesca as able to in vitro methylate volatiles as well as various catechol-like structures including caffeoyl-CoA (56), and a class II OMT capable of specifically methylating quercetin at the 3Ј-position (57). This in vitro functional redundancy of class I and II enzymes may be circumvented in vivo by a tightly controlled compartmentation and tissuespecific expression (11,13). Deciphering additional regulatory elements other than compartmentation may request a focus on internal regulatory mechanism and external biotic and abiotic stressors. If the expression of class II OMTs is generally regulated by developmental and tissue differentiation programs (14,58), class I OMTs seem to respond to external stressors (Ref. 43, this report). Currently, class I enzymes are considered as a primitive remnant from land plant evolution, supported by the lack of any class II OMTlike genes in cyanobacteria (19). Methylation of glucose esters and flavonoids by the novel class I OMTs may be regarded as an intriguing example of how ancestral enzymes have evolved and acquired new specificities during adaptive processes. The developmentally and spatially controlled expression profile of several genes and their products may dictate that lignin biosynthesis and accumulation of soluble catechol-like conjugates can be regulated independently in response to changing environmental conditions. Members of the class I OMT subgroup reported here should be present throughout the plant kingdom. Besides the newly annotated candidate from A. thaliana, similar sequences can be found in the data bases for monocots and dicots, like rice and tobacco. Identification and accurate annotation of these new subsets may help to establish the precise role of several classes of cation-dependent, catechol OMT-like enzymes in plants.