Isolation and characterization of two O-methyltransferases involved in benzylisoquinoline alkaloid biosynthesis in sacred lotus (Nelumbo nucifera)

Benzylisoquinoline alkaloids (BIAs) are a major class of plant metabolites with many pharmacological benefits. Sacred lotus (Nelumbo nucifera) is an ancient aquatic plant of medicinal value because of antiviral and immunomodulatory activities linked to its constituent BIAs. Although more than 30 BIAs belonging to the 1-benzylisoquinoline, aporphine, and bisbenzylisoquinoline structural subclasses and displaying a predominant R-enantiomeric conformation have been isolated from N. nucifera, its BIA biosynthetic genes and enzymes remain unknown. Herein, we report the isolation and biochemical characterization of two O-methyltransferases (OMTs) involved in BIA biosynthesis in sacred lotus. Five homologous genes, designated NnOMT1–5 and encoding polypeptides sharing >40% amino acid sequence identity, were expressed in Escherichia coli. Functional characterization of the purified recombinant proteins revealed that NnOMT1 is a regiospecific 1-benzylisoquinoline 6-O-methyltransferase (6OMT) accepting both R- and S-substrates, whereas NnOMT5 is mainly a 7-O-methyltransferase (7OMT), with relatively minor 6OMT activity and a strong stereospecific preference for S-enantiomers. Available aporphines were not accepted as substrates by either enzyme, suggesting that O-methylation precedes BIA formation from 1-benzylisoquinoline intermediates. Km values for NnOMT1 and NnOMT5 were 20 and 13 μm for (R,S)-norcoclaurine and (S)-N-methylcoclaurine, respectively, similar to those for OMTs from other BIA-producing plants. Organ-based correlations of alkaloid content, OMT activity in crude extracts, and OMT gene expression supported physiological roles for NnOMT1 and NnOMT5 in BIA metabolism, occurring primarily in young leaves and embryos of sacred lotus. In summary, our work identifies two OMTs involved in BIA metabolism in the medicinal plant N. nucifera.

The natural occurrence of BIAs is restricted to a relatively small number of plant families primarily in the order Ranunculales, which includes the opium poppy (Papaver somniferum), an extensively investigated model species that remains the sole commercial source for the narcotic analgesic morphine and the antitussive and potential anticancer drug noscapine (3). BIAs have also been detected in other plant orders including the Cornales, Laurales, Magnoliales, Piperales, Sapindales, and Proteales, although BIA metabolism in members of these taxa has not yet been empirically determined (4). Sacred lotus (Nelumbo nucifera Gaertn.; Fig. 2a) is an aquatic basal eudicot in the order Proteales that has been cultivated as an edible crop for nearly seven millennia, and is highly valued in both Chinese and Ayurvedic traditional medicine for its healing properties (5). The pharmacological properties of sacred lotus are largely the result of the BIA constituents in the plant. For instance, 1-benzylisoquinolines such as norcoclaurine and its 6-O-methylated derivative coclaurine have exhibited anti-HIV activity (6), whereas the 6,7-O,O-and N-methylated derivative armepavine has displayed promising immunomodulatory effects for the treatment of autoimmune diseases such as systemic lupus erythematosus (7). In addition, major aporphines (e.g. nuciferine) and bisbenzylisoquinolines (e.g. neferine) have shown potential for the treatment of lung, liver, ovarian, and breast cancers (8 -12). However, none of the biosynthetic genes and enzymes responsible for the BIA biosynthesis in sacred lotus have been characterized (13).
Based on the reported alkaloid profile in sacred lotus and well-characterized pathways in opium poppy, BIA metabolism is predicted to begin with the formation of norcoclaurine and proceed via a limited number of enzyme types including one or more O-methyltransferases (OMTs), N-methyltransferases, and cytochrome P450 monooxygenases potentially belonging to CYP80 and CYP719 subfamilies (13). Specifically, the formation of known 1-benzylisoquinoline alkaloids in sacred lotus would involve only O-methylations and N-methylations of norcoclaurine (Fig. 1). Because O-methylation contributes substantially to the diversity of BIAs in sacred lotus, we targeted the isolation and functional characterization of OMTs.
Methylations play a central role in the functionalization of specialized metabolites (14). In particular, OMTs involved in BIA metabolism catalyze methyl transfer to the hydroxyl group of an alkaloid substrate, using S-adenosyl-L-methionine (SAM) as the methyl donor group and proceeding via an S N 2-like nucleophilic attack yielding a methylated alkaloid product and S-adenosyl-L-homocysteine (SAH) (15). OMTs are typically homodimeric or, as recently discovered, heterodimeric enzymes (16,17). Within each monomer, the C-terminal domain adopts a canonicalRossmann-fold(i.e.a␤-strandsandwichedbytwo␣-helices) and plays a fundamental role in substrate binding and catalysis, whereas the N-terminal domain is of central importance in dimerization (15, 18 -20).
Sacred lotus accumulates 1-benzylisoquinolines, and derived aporphines (via C8-C2Ј intramolecular coupling) and bisbenzylisoquinolines (via C8-C3Ј/5Ј and C7-O-C3Ј/5Ј intermolecular coupling) (5), most of which are O-methylated at C6, C7, and/or C4Ј (13). The availability of a sacred lotus draft genome (30,31) greatly facilitates gene mining for candidate OMTs based on amino acid sequence similarity with respect to characterized O-methyltransferases (32)(33)(34). However, OMT candidates have so far been investigated only in terms of gene expression, with no functional characterization of the encoded proteins. In addition, enzyme stereospecificity is an intriguing feature of BIA metabolism in sacred lotus, because norcoclaurine has been isolated from the plant as both R-and S-enantiomers, and other BIAs, including coclaurine, norarmepavine, and N-methylcoclaurine have been exclusively reported as the R-conformer (13), in contrast with the predominantly S-stereochemistry of most BIAs in opium poppy and related plants (3). Herein we report the identification and characterization of O-methyltransferases involved in BIA metabolism in sacred lotus.

Sacred lotus BIA profile
Alkaloids corresponding to the leaf (folded and unfolded developmental stages), rhizome, root, and embryo (lotus plumule) of two N. nucifera cultivars, referred as Pink and White because of their distinct floral coloration, were extracted and analyzed by mass spectrometry (MS) (Fig. 2, b and c, Table S1). Fifteen alkaloids (i.e. six 1-benzylisoquinolines, four aporphines, one pro-aporphine, and four bisbenzylisoquinolines) were identified based on their retention times and collisioninduced dissociation (CID) spectra compared with available authentic standards or previously published data. 1-Benzylisoquinoline and aporphine alkaloids were prevalent in the leaves, whereas bisbenzylisoquinolines were most abundant in the embryos. In general, the rhizome and roots contained only low BIA levels, although no bisbenzylisoquinolines were detected. Notably, the structural isomers N-methylisococlaurine and norarmepavine were exclusively detected in the Pink and White varieties, respectively. The bisbenzylisoquinolines liensinine and neferine were found only in the White variety, whereas nelumboferine and isoliensinine were the only bisbenzylisoquinolines identified in Pink and accumulated at 20-(in folded leaves) and 4-fold (in embryos) higher levels compared with the White variety. In addition, alkaloids were detected in the leaves (norarmepavine, nelumboferine, and isoliensinine), rhizome (norcoclaurine, coclaurine, N-methylcoclaurine, norarmepavine, N-methylisococlaurine, anonaine, roemerine, N-nornuciferine, O-nornuciferine, nuciferine and pronuciferine), roots (coclaurine, N-methylcoclaurine, norarmepavine, N-methylisococlaurine, armepavine, anonaine, roemerine, N-nornuciferine, and O-nornuciferine), and embryo (norarmepavine and O-nornuciferine) that had not been previously reported (5,13). Overall, the alkaloid profile found for both varieties supports the occurrence of 6-O-, 7-O-, and 4Ј-O-methyltransferases in sacred lotus.
Amino acid sequence alignment of the functionally characterized Tf6OMT and sacred lotus OMT candidates showed a high conservation of key amino acid residues, including the catalytic determinants His 256 , Asp 257 , and Glu 315 (18), except for NnOMT4, in which case Asp 257 was substituted with a Tyr residue (Fig. 4). Amino acids implicated in BIA substrate binding (Gly 165 , Asp 169 , Cys 253 , and Asp 306 ) were also conserved among sacred lotus OMT candidates except for G165V and C253W substitutions in NnOMT2, G165A in NnOMT3, G165S in NnOMT4, and G165A, D169H, and C253S in NnOMT5. The main residues involved in SAM binding including Thr 170 , Gly 195 (motif I), Asp 218 (motif II), Asp 238 (motif III), and Lys 252 (motif IV) were conserved across sacred lotus OMT candidates, excluding Asp 218 (substituted by Gly in NnOMT4) and Asp 238 (substituted by Asn in NnOMT3). Two methionine residues, Met 166 and Met 307 , assisting in the isoquinoline placing through sulfur-aromatic interactions were also conserved in lotus OMT candidates. The aromatic residues Trp 149 and Phe 162 , which interact with both SAM and BIA substrates, were mutated, Trp 149 to Cys, Ser, and Ile in NnOMT3, NnOMT4, and NnOMT5, respectively, and Phe 162 to Leu in NnOMT5. The conserved GXGXGX sequence in motif I, a hallmark of SAM-dependent methyltransferases (14), was fully maintained

O-Methyltransferases from sacred lotus
in all the candidates except NnOMT5, in which the middle Gly residue was mutated to Cys.

NnOMT purification and characterization
Sacred lotus recombinant His 6 -tagged OMTs produced in Escherichia coli were purified using cobalt-affinity chromatography and detected on an immunoblot using an anti-His 6 antibody (Fig. S1). The molecular weights of the purified proteins and the predicted translation products were similar. Recombinant proteins were initially screened for enzymatic activity using 15 potential BIA substrates representing the 1-benzylisoquinoline, aporphine, protoberberine, and morphinan structural subclasses. The potential 1-benzylisoquinoline substrates differentially contained free hydroxyl groups at C6, C7, C3Ј, and/or C4Ј, and both R-and S-enantiomers were in some cases available (Table S4). Reaction products showed an increase of 14 Da with respect to the corresponding substrate and were detected in reactions containing NnOMT1 or NnOMT5 incubated with various 1-benzylisoquinoline substrates ( Reaction products corresponding to the 6-O-methylation of norcoclaurine and norlaudanosoline (coclaurine (m/z 286) and 6-O-methylnorlaudanosoline (m/z 302), respectively) and the 7-O-methylation of N-methylcoclaurine (armepavine (m/z 314)), were unambiguously identified by comparison with available authentic standards. Enzyme assays with coclaurine and reticuline were subjected to positive-mode electrospray ionization (ESI(ϩ)) high performance LC (HPLC)-tandem mass spectrometry (MS/MS) for the accurate characterization of reaction products by CID fragmentation analysis to determine whether O-methylation was associated with the isoquinoline (C7) or benzyl moiety of coclaurine (C4Ј) and reticuline (C3Ј) (Fig. S3). ESI(ϩ)-CID of 1-benzylisoquinolines at low ionization energy yielded isoquinoline and/or benzyl moieties as major ion fragments; thus, the addition of 14 Da to either fragment with respect to the parent ion revealed the regiospecificity of the reaction. In this regard, coclaurine showed two main fragments at m/z 178 and 107, corresponding to the isoquinoline and benzyl moieties, respectively. After incubation with NnOMT5, the m/z 107 fragment was retained in the reaction product, indicating that the methyl group was added to the isoquinoline moiety (i.e. 7-O-methylation) resulting in the formation of norarmepavine, an alkaloid found in sacred lotus (the other possible reaction product, 4Ј-O-methylcoclaurine, has not been reported in this plant). Similarly, the reaction product of NnOMT5 incubated with (S)-reticuline showed a mass increase of the isoquinoline moiety from m/z 192 to 206, corresponding to the 7-O-methylated product laudanine, whereas the mass of the benzyl moiety (m/z 137) remained unchanged. It is notable that neither reticuline nor laudanine have been Values were normalized to the sample with the highest level for each compound. (c) Structures corresponding to detected aporphine, pro-aporphine, and bisbenzylisoquinoline alkaloids isolated from sacred lotus (for 1-benzylisoquinoline alkaloid structures refer to Fig. 1). FL, folded leaf; UL, unfolded leaf; Rh, rhizome; Ro, roots; E, embryos.

O-Methyltransferases from sacred lotus
reported in sacred lotus. 4ЈOMT activity was not detected with any combination of tested substrates and OMT candidates.
NnOMT1 displayed strict regiospecificity with respect to the C6 hydroxyl and showed similar turnover using racemic norcoclaurine and individual enantiomers of norlaudanosoline, the 3Ј-hydroxylated derivative of norcoclaurine (Fig. 5). NnOMT5 nonpreferentially methylated the C7 hydroxyl of N-methylated (N-methylcoclaurine and reticuline) or N-demethylated (coclaurine) substrates, and also acted on the C6 hydroxyl of (R,S)-norcoclaurine and (S)-norlaudanosoline. The 6-O-methylation activity of NnOMT5 was not accompanied by secondary methylations (an increase of 28 Da). Contrary to NnOMT1, NnOMT5 showed a strict stereospecificity for the substrates in the S-conformation.
Both enzymes showed a pH optimum of 8.0 using (R,S)-norcoclaurine (NnOMT1) or (S)-N-methylcoclaurine (NnOMT5) as substrates and maximum catalytic activity at 30 and 37°C, respectively (Fig. S4). NnOMT1 and NnOMT5 displayed K m values of 20 and 152 M for (R,S)-norcoclaurine, respectively, whereas NnOMT5 exhibited a K m value of 13 M and a K i of 870 M for (S)-N-methylcoclaurine (Table 1, Fig. S5). Catalytic efficiencies (k cat /K m ) were similar for the 6OMT activity of NnOMT1 and the 7OMT activity of NnOMT5. The 7OMT activity of NnOMT5 was 50-fold higher than the corresponding 6OMT activity using (R,S)-norcoclaurine as the substrate.

NnOMT and BIA correlations in the plant
Alkaloid content, O-methyltransferase activity, and OMT gene expression were measured in leaves, rhizome, roots, and embryos for two sacred lotus varieties. Alkaloid content was determined for norcoclaurine, coclaurine, N-methylcoclaurine, and armepavine, for all of which authentic standards were available (Fig. S6). In both the Pink and White varieties, 1-benzylisoquinoline intermediates were predominantly detected in the leaves (Fig. 6a, Table S5). In the White variety, norcoclaurine was the most abundant alkaloid, with folded leaves accumulating 7-, 70-, and 1400-fold higher levels compared with the embryo, rhizome, and root, respectively. The Pink cultivar was richer in N-methylcoclaurine, with the folded leaves containing 3-, 40-, and 90-fold higher levels than the rhizome, root, and embryo, respectively.
Crude plant protein extracts incubated with (R,S)-norcoclaurine, (S)-N-methylcoclaurine, or (R)-armepavine revealed similarities between the Pink and White varieties (Fig. 6b, Table  S6). 6OMT activity was most abundant in folded leaves and embryos, whereas 7OMT activity was predominant in the unfolded leaves. 6OMT activity was 5-and 4-fold higher than 7OMT activity in folded leaves and embryos, respectively. 7OMT activity was 8-fold higher than 6OMT in the White variety, whereas 6OMT activity was not detected in unfolded leaves of the Pink variety. Rhizomes and roots showed only low levels of 6OMT and 7OMT activities. 4ЈOMT activity was not detected.
Gene expression profiles were similar for both the Pink and White varieties (Fig. 6c, Table S7). NnOMT1 and NnOMT5 transcripts were most abundant in leaves and were 3-to 30-fold higher compared with other OMT genes. The levels of all five NnOMT transcripts were similar in the embryos of both varieties.

Discussion
We have shown that two cDNAs from sacred lotus efficiently catalyze the regiospecific 6-O-methylation of norcoclaurine to coclaurine (NnOMT1) and 7-O-methylation of coclaurine and N-methylcoclaurine to norarmepavine and armepavine, respectively (NnOMT5). NnOMT1 was not stereospecific, whereas NnOMT5 accepted only S-enantiomers. Correlation between transcript levels, enzyme activity, and alkaloid content supported a physiological role for NnOMT1 and NnOMT5 in BIA metabolism.
Despite sharing an apparently common biosynthetic pathway to N-methylcoclaurine (Fig. 1), sacred lotus and opium poppy exhibit substantial differences in their alkaloid profiles. In this regard, whereas reticuline constitutes a key branch point intermediate in the biosynthesis of most BIAs in opium poppy (3), sacred lotus does not accumulate reticuline and other 1-benzylisoquinolines containing a C3Ј functional group, or reticuline-derived alkaloids including phthalideisoquinolines, benzo[c]phenanthridines, protoberberines, and morphinans, which occur in members of the Ranunculales (13). In contrast, aporphines and bisbenzylisoquinolines are the major BIAs in sacred lotus, although alkaloid profiles vary among nearly 600 known varieties, and across different plant organs and develop-
The sacred lotus BIA profile showed a preferential accumulation of 1-benzylisoquinolines and aporphines in leaves, and bisbenzylisoquinolines in embryos (Fig. 2b), in agreement with previous investigations (36,37,40). Alkaloids containing 6-O-, 7-O-, and/or 4Ј-O-methylations were detected in Pink and White varieties, supporting the search for specific O-methyltransferase activities in the plant. OMT gene candidates have been predicted in sacred lotus based on sequence similarity to functionally characterized OMTs from BIA-accumulating species (32-34), but the purported catalytic activities have not been assessed. We show that two of these candidates, NnOMT1 and NnOMT5, catalyze the O-methylation of 1-benzylisoquinolines.
Fifteen BIA substrates possessing diverse structural backbones were screened, including 1-benzylisoquinolines with a free hydroxyl at C6, C7, and C4Ј. As described for several 6OMTs from the Ranunculales (41,42), NnOMT1 displayed strict C6 regiospecificity and accepted both (R)-and (S)-1-benzylisoquinoline conformers (Fig. 5). NnOMT1 contains all conserved catalytic and substrate-binding residues occurring in Tf6OMT (Fig. 4). In Tf6OMT, the O-methyltransfer reaction occurs via His 256 base-assisted deprotonation of the substrate C6 hydroxyl, followed by a nucleophilic attack on the SAM methyl group. Interactions of His 256 and Asp 257 with the C6 hydroxyl group provide proper orientation and Glu 315 contributes to the strong basicity in the His 256 NE2 required for catal-

O-Methyltransferases from sacred lotus
ysis (18). Because NnOMT1 and Tf6OMT possess equivalent key catalytic residues (His 256 , Asp 257 , and Glu 315 ) and share ϳ70% amino acid sequence identity, a similar catalytic mechanism is expected for NnOMT1.
Regardless of a nomenclature suggestive of regiospecific functions, OMTs involved in BIA metabolism are generally promiscuous and exhibit apparent redundancy in most plants (22, 26 -28, 43, 44). In Tf6OMT co-crystalized with (S)-norlaudanosoline the C7 hydroxyl group of the BIA substrate was found in close proximity to the catalytic His 256 , suggesting that minor adjustments in substrate positioning could lead to productive 7-O-methyltransfer (18) and the observed dual 6-Oand 7-O-methyltransferase activity of some OMTs. Based on the available crystal structure for Tf6OMT (Protein Data Bank 5ICE), we constructed a homology model for NnOMT5 (Fig.  S7). In the resulting structure, the key catalytic histidine (His 251 in NnOMT5) involved in hydrogen bonding with the target C6-OH (3.0 Å), is also near the C7-OH (3.1 Å). Analysis of residues likely involved in BIA substrate binding showed that G165A and C253S mutations did not have a significant impact (i.e. the main chain carbonyl involved in hydrogen bond formation with the BIA substrate C3Ј and C6 hydroxyl groups, respectively, remained in a similar position) in agreement with the observed variability of these residues in other 7OMTs (i.e. G165A in Ps7OMT, PsN7OMT and Ct7OMT, C253W in Ps7OMT and Ct7OMT, C253N in PsN7OMT, and C253G in

O-Methyltransferases from sacred lotus
Ec7OMT). Conversely, the D169H mutation disrupted a hydrogen bond with the C4Ј-OH (2.6 Å), which stabilizes the substrate in the active site. Interestingly, whereas Asp 169 is a conserved residue across all 6OMTs, or is replaced by Glu in 4ЈOMTs, no consensus was found in functionally characterized 7OMTs (i.e. Asp in PsN7OMT, Ser in Ps7OMT and Ec7OMT, and Thr in Ct7OMT); thus, this residue might not play a significant role in 7OMT regiospecificity. When both C6 and C7 hydroxyl groups were available, NnOMT5 exclusively catalyzed 6-O-methylation (Fig. S2). Similar results were reported for E. californica and Coptis japonica 4ЈOMTs, which catalyze norcoclaurine 6-O-methyltransfer, although with lower efficiency than 4Ј-O-methylation on already 6-O-methylated substrates (45), in support of the importance of 6-O-methylation as a purported rate-limiting step in BIA metabolism (21). However, because NnOMT5 catalyzed 7-O-methylation of (S)-N-methylcoclaurine 50-fold more efficiently than 6-O-methylation of (R,S)-norcoclaurine, the physiological relevance of the latter activity is questionable compared with NnOMT1 (Table 1). Although coclaurine and N-methylcoclaurine are almost certainly physiological substrates of NnOMT5 (Fig. 1), the occurrence of 1-benzylisoquinolines such as N-methylisococlaurine and lotusine, which contain 7-O-methyl and C6 hydroxyl moieties, does not correlate with the inability of NnOMT5 to 7-O-methylate norcoclaurine in vitro. In this regard, it is feasible that norcoclaurine N-methylation should occur first for NnOMT5 to catalyze 7-O-rather than 6-O-methylation, however, the required substrate to test this possibility was not available.
NnOMT5 also preferred norcoclaurine over norlaudanosoline (Fig. 5), whereas NnOMT1 showed no substrate preference. Nonetheless, the main catalytic activity detected for NnOMT5 was the 7-O-methylation of several N-methylated and N-demethylated substrates (Fig. 5). Although our understanding of 7OMT substrate recognition and catalysis is limited, it has been suggested that the conserved residue Asp 306 in PsN7OMT is able to hydrogen bond with the nitrogen of its only substrate norreticuline, whereas Ps7OMT and Ec7OMT contain smaller uncharged residues at equivalent positions presumably to ease steric hindrance with (R,S)-reticuline and other N-methylated substrates (22,23,44), and Ct7OMT, with a Asp 306 to Ser substitution, is able to 7-O-methylate (S)-norcoclaurine (43). Notably, NnOMT5 contains a corresponding Asp 306 residue (Figs. S7 and S8), but catalyzes similarly efficient 7-O-methylation of N-methylated and N-demethyl substrates (Fig. 5).

O-Methyltransferases from sacred lotus
On the other hand, the K m of NnOMT5 with (S)-N-methylcoclaurine (13 M) is substantially lower compared with that for (R,S)-norcoclaurine (152 M), but similar to that reported for Ps7OMT with (S)-reticuline (16 M) (22) and PsN7OMT with (S)-norreticuline (44 M) (23), suggesting that NnOMT5 acts primarily as a 7OMT in the plant.
Assayed aporphine, protoberberine, and morphinan alkaloids were not accepted as substrates (Table S4). OMTs shown to O-methylate 1-benzylisoquinolines and protoberberines occur in species accumulating both alkaloid types (26 -28), although specificity for 1-benzylisoquinolines (e.g. Cj6OMT and PsN7OMT) has also been reported (23,42). Although sacred lotus does not accumulate morphinans or protoberberines (13), aporphines are major alkaloids in the leaves (34,35). The lack of NnOMT activity on aporphines suggests that the C8 -C2Ј coupling yielding the aporphine scaffold occurs after the O-methylation of 1-benzylisoquinoline intermediates. However, Dactylicapnos scandens 7OMT was recently reported to O-methylate corytuberine, demonstrating the plausibility of aporphines as OMT substrates (47). It is notable that the aporphines and bisbenzylisoquinolines of sacred lotus were not available; thus, it remains possible that certain intermediates within these BIA structural types are NnOMT substrates.
The occurrence of most BIAs as R-conformers is another unique feature in sacred lotus compared with members of the Ranunculales (13). The importance of the stereochemistry of 1-benzylisoquinoline substrates is evident in opium poppy, in which reticuline conformation determines the formation of major end point alkaloids, with morphine derived from (R)reticuline, and other alkaloids such as noscapine and sanguinarine produced from (S)-reticuline. BIA stereochemistry is initially established by the stereoselectivity of NCS. Although all characterized NCS isoforms from the Ranunculales yield exclusively (S)-norcoclaurine (48 -50), norcoclaurine in sacred lotus has been detected as both S-and R-enantiomers (6,51,52) and consequently, the presence of both enantiomeric forms of norcoclaurine-derived alkaloids depends on the stereospecificity of downstream enzymes. We show that NnOMT1 catalyzed the 6-O-methylation of (R)-and (S)-norlaudanosoline with similar turnover efficiency, whereas NnOMT5 accepted only (S)-norlaudanosoline and (S)-reticuline (Fig. 5, Table S4). Previously characterized OMTs have either shown a lack of stereospecificity (e.g. Cj6OMT, Ps6OMT, and Ps7OMT) or strict S-stereospecificity (Bk4ЈOMT) (22,42,53). Our results support the formation of (R)-coclaurine and (S)-armepavine as previously reported in lotus (6,54). However, despite the fact that NnOMT5 showed strict stereospecificity toward S-enantiomers, neither norlaudanosoline nor reticuline are physiologically relevant substrates and it is possible that the enzyme catalyzes the turnover of (R)-coclaurine and (R)-N-methylcoclaurine in the formation of the reported compounds (R)norarmepavine and (R)-armepavine, respectively (55). Unfortunately, none of the relevant R-enantiomers were available as potential substrates.
Previous work on the BIA profile in sacred lotus focused on the quantification of aporphines and bisbenzylisoquinolines in leaves and embryos, respectively, but limited information has been provided for 1-benzylisquinoline intermediates (34,36). We quantified key alkaloids produced in the early steps of the pathway (i.e. norcoclaurine, coclaurine, N-methylcoclaurine, and armepavine) in different sacred lotus organs (Fig. 6a). Although 1-benzylisoquinolines were predominant in leaves, the detected compounds were likely intermediates leading to aporphine and bisbenzylisoquinoline end products. The low levels of 1-benzylisoquinolines in embryos could also reflect their transformation to predominant bisbenzylisoquinolines (36,37,40,56).
Assays containing crude plant extracts incubated with (R,S)norcoclaurine, (S)-N-methylcoclaurine, and (R)-armepavine showed 6-O-and 7-O-methyltransferase activity in all organs (Fig. 6b), but mostly in leaves and embryos. OMT transcripts were also expressed in all organs, although NnOMT1 and NnOMT5 showed higher expression levels, mainly in leaves (Fig. 6c). These results are consistent with a recent report on NnOMT expression at different leaf developmental stages, whereby NnOMT1 and NnOMT5 transcript levels were up to 10-fold higher compared with NnOMT2-NnOMT4 (34). A lag between NnOMT expression and aporphine alkaloid accumulation was also noted in this study, which is potentially associated with the involvement of OMTs early in BIA metabolism and a requirement for the expression of additional genes to complete aporphine and bisbenzylisoquinoline biosynthesis. It is notable that because this earlier study did not perform a functional characterization of candidate genes, NnOMT5 was reported as a 4ЈOMT (34). In agreement with the suggested order of O-methylations in other plants (21,45), NnOMT1 appears to act earlier than NnOMT5 in the leaves because 6OMT activity decreased in unfolded leaves, whereas 7OMT activity increased (Fig. 6b). Similarly, NnOMT1 transcript levels were higher in folded leaves than unfolded leaves, whereas NnOMT5 transcript levels remained constant (Fig. 6c), consistent with a previous report (34).
Although NnOMT1 and NnOMT5 transcript levels were 5to 15-fold lower in embryos compared with folded leaves, specific 6-and 7-O-methyltransferase activities were similar in both organs, likely because of the prevalence of photosynthetic leaf proteins. Interestingly, the levels of transcripts encoding O-Methyltransferases from sacred lotus CYP80G, purportedly involved in aporphine biosynthesis, in leaves were 5 times higher than levels of transcripts encoding CYP80A, potentially involved in bisbenzylisoquinoline biosynthesis, in embryos (33), suggesting that lower biosynthetic gene transcript levels in embryos might be sufficient in this organ. However, neither CYP80G nor CYP80A orthologs in sacred lotus were functionally characterized, so it is not certain that these transcripts are relevant to BIA metabolism. Bisbenzylisoquinoline biosynthesis has been proposed to occur in leaves, with alkaloids subsequently transported to embryos (36). NCS activity (4) and purported NnNCS transcripts (57) were detected in leaves, suggesting that BIA biosynthesis occurs in this organ. The detection of O-methyltransferase activity indicates that embryos are also involved in alkaloid biosynthesis.
From an evolutionary perspective, it is not known whether OMT recruitment occurred repeatedly or if an ancestrally promiscuous OMT gave rise to functionally distinct enzymes (20). The 6-O-methyltransferase activity of NnOMT1 was correctly predicted based on similarity to Tf6OMT and other 6OMTs. However, the 7-O-methyltransferase activity of NnOMT5 was unexpected because of its relatively low similarity to characterized 7OMTs from the Ranunculales (Fig. 3, Table S3), suggesting a possible independent recruitment in sacred lotus. The characterization of other enzymes (e.g. norcoclaurine synthase, coclaurine N-methyltransferase, and aporphine and bisbenzylisoquinoline synthases) will provide additional key insights into the evolutionary history of BIA biosynthesis in sacred lotus.

Plant material
Seeds for two sacred lotus (N. nucifera) varieties, named Pink and White because of their distinct flower colors, were germinated in water according to the instructions of the vendor (Rarexoticseeds). After the emergence of embryogenic leaves, the seedlings were planted in pots containing heavy loam soil and submerged in an artificial pond. Plants were grown at 30°C and under high-intensity lighting with a 16-h photoperiod. Alternatively, seeds were carefully opened with a bench press and the embryos were isolated. All harvested plant samples were immediately frozen in liquid nitrogen and stored at Ϫ80°C until used.

Plant alkaloid extraction and LC-MS analysis
N. nucifera samples corresponding to folded and unfolded leaves, rhizomes, roots, and embryos were ground to a fine powder under liquid nitrogen using a mortar and pestle. Alkaloid extraction was performed using a modified version of a previously reported protocol for sacred lotus (36). Briefly, 1 ml of methanol was added to ϳ100 mg of plant material, and samples were sonicated in an ultrasonic bath for 30 min at room temperature, centrifuged at 14,000 ϫ g for 10 min, and the supernatant was filtered (0.2-m Acrodisc syringe filter, Pall Corporation, NY). The insoluble material was dried and weighed. Samples (5 l) were fractionated on a Zorbax C18 column, 2.1 ϫ 50 mm, 1.8 m (Agilent, CA) using a Dionex UltiMate 3000 HPLC system (Thermo Scientific) at a flow rate of 500 l/min. The mobile phase consisted of solvent A (10 mM ammonium acetate, pH 5.5, 5% (v/v) acetonitrile) and solvent B (100% acetonitrile), starting at 100% (v/v) solvent A, and ramping from 0 to 20% (v/v) solvent B over 5 min, 20 to 50% (v/v) solvent B over 3 min, 50 to 100% (v/v) solvent B over 3 min, remaining isocratic at 100% (v/v) solvent B over 2 min, ramping from 100 to 0% (v/v) solvent B over 0.1 min, and remaining isocratic at 100% (v/v) solvent A over 1.9 min. The total run time was 15 min and data were collected for the first 10 min. Heated ESI source and interface conditions were operated in positive ion mode as follows: vaporizer temperature 400°C, source voltage 3 kV, sheath gas 60 arbitrary units, auxiliary gas 20 arbitrary units, capillary temperature 380°C, capillary voltage 6 V, and tube lens 45 V. A LTQ-Orbitrap-XL (Thermo Fisher) mass spectrometer was operated using LTQ Tune Plus version 2.5.5 SP1 and Xcalibur software (version 2.1.0.1140), with additional analyses using the QualBrowser feature of Xcalibur. Internal and external calibration, tuning, and general operations were performed as previously reported (27,59). Error was maintained at Ͻ2 ppm. Exact masses, retention times, and CID spectra of available authentic standards, or previously published data, was used for alkaloid identification (Table S1, Fig. S6). Quantification of norcoclaurine, coclaurine, N-methylcoclaurine, and armepavine was performed using authentic standards.

NnOMT identification and phylogeny
A previously reported N. nucifera draft genome (30,31) was searched to identify sequences encoding potential O-methyltransferase involved in BIA biosynthesis using the tBLASTn algorithm and functionally characterized OMTs from P. somniferum as queries (Ps6OMT, Ps7OMT, PsN7OMT, and Ps4ЈOMT2). Five candidates (NnOMT1-NnOMT5) sharing Ͼ40% amino acid sequence identity with query sequences were isolated. The GenBank TM accession codes were as follows: XM_010245752, XM_010249599, XM_010249600, XM_010273389, and XM_010277761) (Table S2). Protein molecular weight and pI predictions were made using the

O-Methyltransferases from sacred lotus
Geneious software package (Biomatters, NJ). Amino acid sequence alignments were performed using the default parameters of the MUSCLE algorithm (60) implemented in MEGA X (61). Evolutionary history was inferred using the Maximum Likelihood method established on the JTT matrix-based model (62). The tree with the highest log likelihood was drawn, with branch lengths proportional to the number of substitutions per site, and nodes labeled according to the percentage of trees in which the associated taxa clustered together in the bootstrap test based on 1000 replicates. The amino acid percent identity matrix among N. nucifera (Nn) and other OMTs was performed using Clustal Omega (63) Table S9.

NnOMT cDNA isolation and expression
Plant tissues were ground with a mortar and pestle to a fine powder under liquid nitrogen. Total RNA was extracted using the cetyltrimethylammonium bromide method, as previously reported for N. nucifera tissues (64). RNA quality was confirmed by A 260/280 and A 260/230 absorbance measurements using a NanoDrop ND-1000 (Thermo Fisher Scientific) spectrophotometer, and visualization by agarose gel electrophoresis. After genomic DNA removal (AccuRT Genomics DNA Removal kit, Applied Biological Materials, BC, Canada), firststrand cDNA synthesis was performed on 1 g of RNA using 5ϫ All-in-One RT MasterMix kit, according to the instructions of the manufacturer (Applied Biological Materials, BC, Canada). NnOMT open reading frames were amplified from cDNA using Q5 High Fidelity DNA polymerase (New England Biolabs, MA) and sequence-specific primers (Table S8a) under the following conditions: 98°C for 30 s; 35 cycles of 98°C for 10 s, 52-65°C (optimized for each NnOMT) for 30 s, and 72°C for 30 s; 72°C for 2 min. Amplicons were cloned in a pMiniT 2.0 vector, according to the instructions of the manufacturer (New England Biolabs, MA) and used to transform the E. coli TOP 10 strain for colony PCR screening. Plasmids from positive colonies were purified (GeneJet Plasmid Miniprep kit, Thermo-Fisher Scientific) and sequenced. PCR products were purified (GeneJet Gel Extraction Kit, Thermo-Fisher Scientific) and ligated into pRSET-A (Invitrogen) to construct the expression vectors using one-step sequence-and ligation-independent cloning (65). Full-length NnOMT coding regions were cloned in-frame with sequences encoding an N-terminal His 6 tag using specific primers (Table S8a).

Recombinant protein purification
The pRSET-A-NnOMT expression plasmids were used to transform the E. coli Rosetta (DE3) pLysS (EMD Millipore, MA) strain and single colonies were used to inoculate 50 ml of LB medium supplemented with 100 g/ml of ampicillin and 35 g/ml of chloramphenicol. Cultures were grown at 30°C with orbital shaking at 200 rpm for 16 h and used to inoculate 1 liter of LB media (100 g/ml of ampicillin and 35 g/ml of chloramphenicol) to a starting OD 600 of 0.1. Cultures were grown at 30°C until OD 600 ϳ0.6, cooled to 16°C and the production of recombinant proteins was induced by the addition of isopropyl ␤-D-thiogalactoside to a final concentration of 0.1 mM. Cultures were kept at 16°C with shaking at 200 rpm for 16 h and cells were harvested by centrifugation at 5,000 ϫ g for 20 min at 4°C. Cell pellets were resuspended in 40 ml of protein extraction buffer (100 mM Tris-HCl, pH 7.5, 300 mM NaCl, 10% (v/v) glycerol) supplemented with 1 mg/ml of lysozyme and sonicated on ice for 3 min (10-s on, 30-s off). The crude lysate was centrifuged at 16,000 ϫ g for 20 min at 4°C to remove cellular debris. The cleared supernatant was loaded onto 1 ml of equilibrated Talon cobalt affinity resin (Clontech) and incubated at 4°C for 30 min with gentle shaking. The resin was washed with 20 ml of protein extraction buffer, followed by 5 ml of protein extraction buffer containing 20 mM imidazole. Purified protein was eluted using 5 ml of protein extraction buffer containing 200 mM imidazole, and subsequently concentrated and desalted by repeated ultrafiltration on an Amicon Ultra 30K column (EMD Millipore, MA) in storage buffer (100 mM Tris-HCl, pH 7.5, 10% (v/v) glycerol, 1 mM ␤-mercaptoethanol). Purified His 6 -tagged protein concentration was determined using the Bradford reagent according to the instructions of the manufacturer (Thermo Fisher) and using BSA as the standard. Protein purity was assessed by SDS-PAGE using a 10% (w/v) gel, which was subsequently transferred to a nitrocellulose membrane. Protein blots were blocked overnight with 5% (w/v) skim milk and incubated for 1 h with a 1:1,000 dilution of His 6 antibody. After washing with TBST buffer (200 mM Tris-HCl, pH 7.6, 1.5 M NaCl, 0.1% (v/v) Tween 20), the blots were incubated for 1 h with a 1:10,000 dilution of goat anti-mouse IgG antibodies conjugated with horseradish peroxidize (Bio-Rad). Immunoblots were washed and developed using SuperSignal West Pico chemiluminescent substrate, according to the instructions of the manufacturer (Thermo Scientific) and the bands were visualized using an Amersham Biosciences Imager 600 (GE Healthcare).

Plant protein extraction
Plant tissues were ground to a fine powder under liquid nitrogen using a mortar and pestle. Total proteins were extracted by adding 5 ml of ice-cold extraction buffer A containing 100 mM Tris-HCl, pH 7.5, 10% (v/v) glycerol, 1%

O-Methyltransferases from sacred lotus
(w/v) polyvinylpyrrolidone 40, 5 mM dithiothreitol and 1 mM phenylmethylsulfonyl fluoride to ϳ1 g of plant material, and sonicated in an ultrasonic bath for 5 min at 4°C. Plant debris was eliminated by centrifugation at 14,000 ϫ g for 10 min, and the supernatant was desalted in a PD-10 column (GE Healthcare) and eluted in extraction buffer without dithiothreitol and phenylmethylsulfonyl fluoride. The solution was concentrated by repeated ultrafiltration using an Amicon Ultra 30K column (EMD Millipore) in storage buffer (100 mM Tris-HCl, pH 7.5, 10% glycerol, 1 mM ␤-mercaptoethanol). Crude plant protein concentration was determined using the Bradford reagent according to the instructions of the manufacturer (Thermo Fisher) and using BSA as the standard.

Enzyme assays
Substrate-acceptance assays were performed in 100 mM Tris-HCl, pH 7.5, using 100 M alkaloid, 200 M SAM, and 5 g of purified recombinant protein in a reaction volume of 50 l. Negative control assays were performed using purified recombinant protein denatured in boiling water for 10 min. Assays were incubated at 30°C and quenched after 4 h with 100 l of acetonitrile, centrifuged at 17,000 ϫ g for 40 min to precipitate proteins, and the supernatant analyzed by LC-MS/MS. Product formation, determined as an increase of 14 Da with respect to the substrate, was confirmed by comparison of retention times and CID spectra with respect to authentic standards, or data reported in the literature. Substrate conversion rates were calculated based on substrate loss. NnOMT1 and NnOMT5 pH optimum assays were conducted in 25 mM sodium citrate, pH 5.0 -6.0, 50 mM sodium phosphate, pH 6.0 -7.5, and 100 mM Tris-HCl, pH 7.5-9.0, using 100 M (R,S)-norcoclaurine or 100 M (S)-N-methylcoclaurine, respectively, 200 M SAM and 1 g of purified recombinant protein in a reaction volume of 50 l. Assays were incubated within the linear range for product formation at 30°C for 5 min and stopped with 100 l of acetonitrile. Temperature optima were determined under identical conditions except that 100 mM Tris-HCl, pH 8, and incubation temperatures between 4 and 60°C were used. Kinetic parameters were measured for 5 min under the same conditions except that NnOMT1 was incubated with (R,S)-norcoclaurine at 30°C and NnOMT5 with either (R,S)-norcoclaurine or (S)-N-methylcoclaurine at 37°C, using alkaloid concentrations from 5 to 400 M and a fixed SAM concentration of 400 M. Product formation for NnOMT1 and NnOMT5 was quantified by comparison with authentic standards of coclaurine and armepavine, respectively. All assays were performed in triplicates. Saturation curves and kinetic parameters were determined based on the Michaelis-Menten equation (NnOMT1 and NnOMT5 incubated with (R,S)-norcoclaurine) or a substrate-inhibition equation (NnOMT5 incubated with (S)-N-methylcoclaurine), using Prism 5 (GraphPad).
Plant protein assays were performed using 50 g of crude protein in 100 mM Tris-HCl, pH 7.5, using 200 M alkaloid and 400 M SAM in a 100-l reaction. Negative control assays were performed using crude plant protein extracts denatured in boiling water for 10 min. Reactions were incubated overnight at 30°C, quenched with 200 l of acetonitrile and analyzed by LC-MS/MS. Product formation, determined as an increase of 14 Da with respect to the substrate, was confirmed by comparison of retention times and CID spectra with respect to authentic standards.

HPLC-MS/MS
HPLC-MS/MS was performed using a 1200 HPLC instrument coupled to a 6410 triple quadrupole MS (Agilent, CA). Samples (5 l) were injected onto a Poroshell 120 SB-C18 HPLC column, 2.1 ϫ 50 mm, 2.7-m particle size (Agilent). Analytes were eluted using a mobile phase gradient of solvent A (10 mM ammonium acetate, pH 5.5, 5% (v/v) acetonitrile) and solvent B (100% acetonitrile) at a flow rate of 600 l/min. The gradient started at 100% (v/v) solvent A, ramped linearly to 60% (v/v) solvent B by 8 min, further increased linearly to 99% (v/v) solvent B over 2 min, remained isocratic at 99% (v/v) solvent B from 10 to 11 min, and returned to 100% (v/v) solvent A at 11.1 min for a 3-min re-equilibration period. Analytes were applied to the mass analyzer using an electrospray ionization probe operating in positive mode with the following conditions: capillary voltage 4 000 V, fragmentor voltage 110 V, source temperature 350°C, nebulizer pressure 50 p.s.i., gas flow 10 liters/min. For full-scan analysis, quadrupole 1 and 2 were set to radio frequency only, whereas the third quadrupole scanned from m/z 200 to 700. ESI(ϩ) and CID spectra were analyzed, the precursor m/z was selected in quadrupole 1 and collision energy of 25 eV and an argon collision gas pressure of 1.8 ϫ 10 Ϫ3 torr were applied in quadrupole 2. The resulting MS 2 fragments were resolved by quadrupole 3 scanning from 30 to 5 m/z greater than the precursor ion m/z. Compounds were identified based on retention times and ESI(ϩ)-CID spectra compared with authentic standards or published spectral data.

Quantitative real-time PCR
RNA extraction and cDNA synthesis were performed as described above. Quantitative RT-PCR analysis was performed using PowerUp SYBR Green Master Mix and a QuantStudio-3 Real-Time PCR System (Applied Biosystems). Reactions (10 l) contained diluted cDNA (ϳ8 ng), 5 l of SYBR Green Master Mix (2 times) (Applied Biosystems), and 500 nM of each primer (Table S8b). Thermal conditions were 50°C for 2 min, 95°C for 2 min, 40 cycles of 95°C for 1 s, and 60°C for 30 s. Amplification specificity was confirmed by melt-curve analysis, generated at a ramp rate of 1.6°C/s to 95°C maintained for 15 s, 1.6°C/s to 60°C maintained for 1 min, and 0.15°C/s to 95°C, which was maintained for 15 s. Primer efficiency was verified using Lin-RegPCR software (66) and samples with values between 1.8 and 2 were selected for analysis. Relative transcript abundance was calculated by the 2 Ϫ⌬⌬Ct method (67), using ␤-actin as the endogenous reference gene (68). Transcript levels were normalized with respect to the gene showing the lowest expression level.

Homology modeling
The previously reported crystal structure from T. flavum norcoclaurine 6-O-methyltransferase (Tf6OMT) in complex