2024 Volume 72 Issue 5 Pages 507-511
Amaryllidaceae alkaloids are structurally diverse natural products with a wide range biological properties, and based on the partial identification of the biosynthetic enzymes, norbelladine would be a common intermediate in the biosynthetic pathways. Previous studies suggested that norbelladine synthase (NBS) catalyzed the condensation reaction of 3,4-dihydroxybenzaldehyde and tyramine to form norcraugsodine, and subsequently, noroxomaritidine/norcraugsodine reductase (NR) catalyzed the nicotinamide adenine dinucleotide phosphate (NADPH)-dependent reduction of norcraugsodine to generate norbelladine. However, recent studies have highlighted possible alternative Amaryllidaceae alkaloid biosynthetic pathways via the formation of isovanillin and vanillin from the 4-O- and 3-O-methylation reactions of 3,4-dihydroxybenzaldehyde, respectively. Herein, we focused on NpsNBS and NpsNR, which were initially identified from Narcissus pseudonarcissus, and explored their substrate recognition tolerance by performing condensation reactions of tyramine with various benzaldehyde derivatives, to shed light on the Amaryllidaceae alkaloid biosynthetic pathway from the viewpoint of the enzymatic properties. The assays revealed that both NpsNBS and NpsNR lacked the abilities to produce 4′-O- and 3′-O-methylnorbelladine from isovanillin and vanillin with tyramine, respectively. These observations thus suggested that Amaryllidaceae alkaloids are biosynthesized from norbelladine, formed through the condensation/reduction reaction of 3,4-dihydroxybenzaldehyde with tyramine.
Amaryllidaceae alkaloids are a large, structurally diverse group of plant metabolites with various biological properties, including antitumor, antibacterial, antimalarial, antiviral, antioxidant, and antiproliferative activities.1–3) They are biochemically classified into over ten types, including the biogenetically unexpectable “norbelladine (I),” “cherylline (II),” “galanthamine (III),” “lycorine (IV),” “homolycorine (V),” “crinine (VI),” “pretazettine (VII),” “montanine (VIII),” “narciclasine (IX), “pseudolycorine (X),” and other ring-types, based on biogenetic linkage and ring type.4,5) Due to its strong acetylcholine esterase inhibitory and nicotinic receptor binding properties, galanthamine is used as an Alzheimer’s disease drug to moderate the rate of cognitive decline.6) Several enzymes involved in the biosynthesis of Amaryllidaceae alkaloids have been identified.4,7)
Norbelladine synthase (NBS) and noroxomaritidine/norcraugsodine reductase (NR) are cooperative enzymes responsible for the efficient formation of norbelladine (4a), the common intermediate in the biosynthesis of Amaryllidaceae alkaloids. To date, NpsNBS/NpsNR, NpaNBS/NpaNR, and LaNBS/LaNR have been identified in Narcissus pseudonarcissus, Narcissus papyraceus, and Leucojum aestivum, respectively.8–10) In this cooperative catalysis, NBS accepts 3,4-dihydroxybenzaldehyde (1a) and tyramine (2) to produce a Schiff base-type of intermediate, norcraugsodine (3a) via their condensation, and then NR generates 4a from 3a by catalyzing the nicotinamide adenine dinucleotide phosphate (NADPH)-dependent reduction of 3a (Fig. 1). In addition, norbelladine 4′-O-methyltransferase (NpN4OMT), as well as its homologues, LaOMT, LrOMT, and LlOMT, have been identified in N. pseudonarcissus, L. aureus, L. radiata, and Lycoris longituba as a key enzyme with a para-O-methylation catalytic property for 4a to produce 4′-O-methylnorbelladine (4b), leading to the vast diversity of Amaryllidaceae alkaloids, such as the I−IX types.11–14) However, recent in vitro functional analyses of LrOMT revealed that it possesses a dual function to catalyze the para/meta-O-methylations of 4a and produce not only 4b, but also 3′-O-methylnorbelladine (4c), respectively.13) Furthermore, LrOMT exhibits substrate flexibility for 1a, the putative NBS substrate, and produces isovanillin (1b) and vanillin (1c), suggesting possible alternative Amaryllidaceae alkaloid biosynthetic pathways via the methylations of 1a by OMT, followed by condensation and reduction reactions by NBS and NR13) (Fig. 1). NpsNR also yielded 4b and 4c from 1b and 1c in combination with 2 (Fig. 1) via the formations of the corresponding Schiff base-type products 3b and 3c, respectively, even in the absence of NBS in the in vitro reaction, although the Schiff base formations by NpsNR are thought to be non-specific enzymatic reactions derived from the proximity of 2 with the aldehydes 1b/1c in the active site cavity.9) However, the substrate recognition scopes of NBS have not yet been definitively verified, complicating the interpretation of the Amaryllidaceae alkaloid biosynthetic pathways (Fig. 1).
Gaining insight into the substrate flexibilities of NpsNBS and NpsNR would enhance our comprehension of the Amaryllidaceae alkaloid biosynthetic pathway, beyond the fundamental functional analyses of the enzymes. Herein, we report the substrate flexibilities of NpsNBS and NpsNR for seven 1a-analogs, including the 4- and 3-O-methylated 1a-analogs 1b and 1c, as well as 3,4-dimethoxy- (1d), 4- and 3-hydroxy- (1e and 1f), and 4- and 3-methoxy- (1g and 1h) benzaldehydes, in in vitro enzyme reactions. The in vitro assays revealed that NpsNBS showed high specificity for only 1a and 1f, and the productions of the corresponding norbelladine analogues were significantly accelerated by the combination of NpsNBS and NpsNR.
To shed light on the Amaryllidaceae alkaloid biosynthetic pathway from the viewpoint of the enzymatic properties, we investigated the substrate flexibility of NBS for the benzaldehyde derivatives 1b–h in combination with NR. All structures of tested substrates and related products are listed in Table 1. To this end, the sole incubation of NpsNBS (solo NpsNBS assay), the sole incubation of NpsNR (solo NpsNR assay), a one-step co-incubation of NpsNBS and NpsNR (one-step assay), and a step-by-step incubation of NpsNBS and NpsNR (two-step assay) were performed to test the substrate specificities of NBS and NR against the aldehyde substrates, since the accelerated formation of 4a was previously observed in the two-step assay.10) Due to the spontaneous conversion of 3a into 4a in the in vitro environment, the enzymatic formation of 3a and its analogs in the solo NpsNBS assays were confirmed by detecting 4a–h. The presence of 4a–h in all reactions was identified by comparing their retention times and MS spectra with those of chemically synthesized standards and their yield % with respect to 1a and its analogs were calculated, based on their extracted ion chromatogram peak areas. Papaverine was also used as an internal standard in all assays to compare the production abilities of 4a and its analogues with those reported previously.10)
 | |||||||
|---|---|---|---|---|---|---|---|
| Compound | R1 | R2 | Yields of 4a and its analogues (%)a) | ||||
| NpsNBS | NpsNR | NpsNBS + NpsNR (one-step) | NpsNBS + NpsNR (two-step) | ||||
| 1a | 4a | OH | OH | 0.93 ± 0.03 | 3.99 ± 0.29 | 4.48 ± 0.37 | 10.21 ± 0.15 |
| 1b | 4b | OCH3 | OH | — | — | — | — |
| 1c | 4c | OH | OCH3 | — | — | — | — |
| 1d | 4d | OCH3 | OCH3 | — | — | — | — |
| 1e | 4e | OH | H | — | — | — | — |
| 1f | 4f | H | OH | 0.05 ± 0.01 | 1.15 ± 0.14 | 0.95 ± 0.07 | 1.76 ± 0.12 |
| 1g | 4g | OCH3 | H | — | — | — | — |
| 1h | 4h | H | OCH3 | — | 0.22 ± 0.00 | 0.15 ± 0.01 | 0.03 ± 0.00b) |
a) The percentage of the yields of 4a and its analogues with respective to 1a and its analogues are calculated based on their extracted ion chromatogram peak areas. Mean average ± standard deviation (S.D.) of three independent experiments. b) Decreased production level due to boiling after the NpsNBS reaction for a two-step assay.
We first investigated the substrate preferences of NBS and NR for 1a and its dimethyl analog 1d in the presence of 2. The LC-electrospray ionization (ESI)-MS analysis revealed that all incubation assays using 1a and 2 as substrates showed a cationized molecular ion peak at m/z 260 [M + H]+, corresponding to the chemically synthesized 4a, in yields of 0.93% for the solo NpsNBS assay, 3.99% for the solo NpsNR assay, 4.48% for the one-step assay, and 10.21% for the two-step assay (Table 1, Figs. 2a, d). These yields were comparable with relative peak area ratios of 1.49% for the solo NpsNBS assay, 6.27% for the solo NpsNR assay, 10.66% for the one-step assay, and 23.17% for the two-step assay with the internal standard, papaverine, which are almost in good agreement with previously reported data observed with NpaNBS/NpaNR and LaNBS/LaNR.10) The production of 4a was not observed in reactions without NpsNBS and/or NpsNR. Furthermore, the two-step assay led to the increased yield of 4a by 11.0, 2.6, and 2.3-fold over those in the solo NpsNBS, solo NpsNR, and one-step assays, respectively, which are also almost consistent with previously reported data10) (Table 1, Figs. 2a, d). In contrast, when 1d was used as the aldehyde substrate with 2, none of the incubation assays showed any production of 4d or the other compounds associated with enzymatic formation (Table 1, Supplementary Fig. 1d).
(a–c) The productions of 4a (a), 4f (b), and 4h (c) in NpsNBS (solo NpsNBS assay), NpsNR (solo NpsNR assay), NpsNBS + NpsNR (one-step assay), and NpsNBS + NpsNR (two-step assay). The cationized molecular ion peaks at m/z 260, 244, and 258 [M + H]+ were monitored for 4a, 4f, and 4h, respectively. The reactions without enzyme were used as negative controls (No enzyme). (d–f) The productions of 4a (d), 4f (e), and 4h (f) in NpsNBS (solo NpsNBS assay), NpsNR (solo NpsNR assay), NpsNBS + NpsNR (one-step assay), and NpsNBS + NpsNR (two-step assay). The decreased production level due to boiling after the NpsNBS reaction in the two-step assay is marked with ◆.
Next, the catalytic activities of NpsNBS and NpsNR were tested against 1b and 1c, corresponding to the 4- and 3-O-monomethylated 1a-analogs, respectively. No enzymatic products corresponding to 4b and 4c (m/z 274 [M + H]+) were detected in the solo NpsNBS assays (Table 1 and Supplementary Figs. 1b, c), suggesting the inability of NpsNBS to accept 1b and 1c. Furthermore, in contrast to the previous report stating that NpsNR produced 4b from 1b and 2 in a production ratio 25.1% lower than that of 4a,9) no enzymatic products including 4b were detected in all other assays (Table 1, Supplementary Figs. 1b, c). Considering the lower 4b-forming activity of NpsNR, this may be below the detection limit of our assay system. In addition, the assays using NpsNR with/without NpsNBS did not show any production of 4c, for the same reasons as the cases of 4b. Nevertheless, the NpsNR activities to form 4b and 4c would be significantly lower than that for 4a. A comprehensive assessment of the substrate specificities observed for NpsNR, and especially for NpsNBS, suggests that the Amaryllidaceae alkaloids would be biosynthesized via the condensation reaction of 2 with 1a by NBS and the reduction reaction of 3a by NR, which precedes the methylation reaction by OMT, as previously proposed.4)
Finally, the 4-hydroxy- and 3-hydroxybenzaldehydes 1e and 1f, and their methylated analogs 1g and 1h were assessed to investigate the regiospecific preferences of NpsNBS and NpsNR for the hydroxyl groups at these positions. As a result, the solo NpsNBS, solo NpsNBS, and one-step assays were capable of generating 4f from 1f and 2 in yields of 0.05, 1.15, and 0.95%, respectively, and its production (1.76% yield) was further accelerated by a step-by-step NpsNBS and NpsNR reaction with 35.2-, 1.5-, and 1.9-fold increases compared with those of the solo NpsNBS, solo NpsNBS, and one-step assays, respectively (Table 1, Figs. 2b, e). However, none of the enzymatic products derived from 1e and 2, including 4e, were detected in all tested reactions (Table 1, Supplementary Fig. 1e). Furthermore, as in the cases of 1b and 1c, NpsNBS did not accept 1g and 1h to yield 4g and 4h (Table 1, Supplementary Figs. 1g, h). Nevertheless, the 4a-forming activity of NpsNBS (0.93% yield) was 18.6-fold increases compared with that of 4f (0.05% yield), implying that the disubstitutions at C-3 and C-4 with the hydroxy groups are required for the proper catalytic reaction of NpsNBS.
Due to the absence of the three-dimensional structure of NBS, we constructed a NpsNBS model structure complexed with 1a and 2. The model structure predicted that NpsNBS possessed an active site cavity to accommodate both 1a and 2. Furthermore, the model structure suggested that it locked the C-3 and C-4 hydroxy groups of 1a in a narrow pocket between Pro139 and Glu71, with a polar interaction of Glu71 with the C-3 hydroxy group (2.4 Å), to locate the aldehyde moiety in close distances with the putative catalytic center Lys83 (3.3 Å) and the amine moiety of 2 at the deeper inside in the active site cavity. Thus, the active site cavity was filled by both 1a and 2, suggesting that 1 was no longer able to accept the methoxylated 1-analogues at C-3 and C-4, 1b–d, 1g, and 1h. However, the model structure did not predict any significant interaction between the C-4 hydroxy group and residues lining the active site cavity. It is thus also speculated that NpsNBS could not accept 1e without the C-3 hydroxy group to produce 4e, since it could not lock 1e in the proper binding site at the active site to facilitate the NBS catalytic reaction (Fig. 3, Supplementary Fig. 3). These pairs of Pro and Glu are also conserved in NBSs (Supplementary Figs. 4, 5), indicating that NpaNBS and LaNBS would have similar substrate and product specificities.
Compounds 1a and 2 are shown with deep blue and deep purple stick models. The surface model represents the cavity of the active site. The black arrow represents the entrance of the cavity.
In contrast, the solo NpsNR, one-step, and two-step incubations with 1h and 2 were capable of producing 4h, with yields of 0.22, 0.15, and 0.03%, respectively, despite the finding that NpsNR lacked the ability to produce 4g from 1g and 2 (Table 1, Figs. 2c, f, Supplementary Fig. 1g). We also found that the additions of NpsNR and NADPH to a pre-boiled 1h-containing reaction solution led to almost the same level of 4h production as that in the two-step assay (Supplementary Fig. 2). Since the boiling process was conducted before the NpsNR reaction in the two-step assay, the observed lower yield of 4h in the two-step assay might be due to the experimental methodology, rather than the interaction of the substrate with NpsNR (Supplementary Fig. 2). Similar cases were found with NpsNR, which can produce not only 4a and 4b as mentioned above, but also piperonal and oxomaritinamine, as previously reported.9) Thus, NpsNR is considered to be much more substrate tolerant than NpsNBS.
In this study, the substrate flexibilities of NBS for 1b–h and NR for 1d–h were for the first time investigated. An in-depth analysis of the substrate scope of NpsNBS demonstrated that this enzyme has comparatively stringent specificity, in contrast to LrOMT and NpsNR with relatively tolerant substrate specificities in the biosynthesis of Amaryllidaceae alkaloids.9,13) Remarkably, our study newly revealed that NpsNBS has an ability to accept 1f with the hydroxy group at C-3 as a substrate, in addition to 1a. Furthermore, NpsNR was capable of forming not only 4a and 4f, but also 4h with the methoxy group on C-3. Unlike the previous report for NpsNR,9) our analysis also showed that NpsNR may lack the ability to produce 4b and 4c. In contrast, our study revealed that NpsNBS lacked the ability to generate C-3 and/or C-4 methyoxylated 4-analogues, 4b–d, 4g, and 4h. Hence, in agreement with previous studies, the main biosynthetic pathway of Amaryllidaceae alkaloids in plants would share 4a as the intermediate by the sequential functions of NpsNBS and NpsNR. However, our results do not rule out the possibility that similar or novel enzymes, instead of NpsNBS, may be involved in an alternative biosynthetic pathway of Amaryllidaceae alkaloids. This might be solved by genetic investigations of the plants producing Amaryllidaceae alkaloids.
All materials and methods used in this study are included in Supplementary Materials.
This study was supported in part by Grants-in-Aid for Scientific Research from JSPS (JP22H02777 to H.M. and JP22K15303 to Y.N.).
S.Y.Y.H. synthesized compounds and performed assays. Y.N. constructed plasmids and prepared enzymes. S.Y.Y.H. and Y.N. wrote this manuscript draft. H.M. designed this study and wrote this manuscript. All authors commented on the manuscript and approved the final version.
The authors declare no conflict of interest.
The LC-ESI-MS data generated in this study are provided in the Source Data file. Any additional data required will be made available upon request.
This article contains supplementary materials.
