Abstract
Quinine is a historically important natural product containing a methoxy group that is assumed to be incorporated at a late pathway stage. Here we show that the methoxy group in quinine and related Cinchona alkaloids is introduced onto the starting substrate tryptamine. Feeding studies with Cinchona plantlets definitively show that 5-methoxytryptamine is utilized as a quinine biosynthetic intermediate in planta. We discover the biosynthetic genes that encode the responsible oxidase and methyltransferase, and we use these genes to reconstitute the early steps of the Cinchona alkaloid biosynthetic pathway in Nicotiana benthamiana to produce a mixture of methoxylated and desmethoxylated Cinchona alkaloid intermediates. Importantly, we show that the co-occurrence of both tryptamine and 5-methoxytryptamine substrates, along with the substrate promiscuity of downstream pathway enzymes, enable parallel formation of both methoxylated and desmethoxylated alkaloids in Cinchona pubescens.
- Quinine
- Cinchona
- alkaloids
- biosynthesis
- genes discovery
Quinine (1), used for centuries as an antimalarial drug and as a bittering ingredient, is a member of the Cinchona alkaloids, a structurally diverse group of alkaloids produced primarily by Cinchona plants.[1] In addition to quinine (1), the quinine stereoisomer, quinidine (2), is commonly prescribed as an antiarrhythmic medication.[2] Desmethoxylated analogs (e.g. cinchonidine 3 and cinchonine 4) also have established pharmacological properties[1d, 3] and, moreover, are used in synthetic chemistry as chiral catalysts and additives.[4] Congeners with a single bond at C-10 – C-11 (dihydro alkaloids; for structures, see 5-8 in Scheme S1 or Figure S1) also occur in several Cinchona species and have similar properties as 1-4.[1d,3a] While total syntheses of many Cinchona alkaloids have been reported,[5] the biosynthesis of these compounds is still largely unelucidated.
Early investigations established that Cinchona alkaloids belong to the large family of monoterpenoid indole alkaloids.[6] These alkaloids are derived from strictosidine (12), which is formed from the Pictet-Spengler condensation of a tryptophan-derived moiety, tryptamine (10), and the iridoid-type monoterpene secologanin (11).[7] It has long been hypothesized that strictosidine is converted to cinchonidinone (15a) and cinchoninone (15b), compounds that are then methoxylated to form quinine (1) and quinidine (2), respectively (Scheme 1).[8] Here we show through feeding experiments that methoxylation instead occurs on the early precursor tryptamine (10). We show that both tryptamine and 5-methoxytryptamine (21) are carried through the downstream pathway, leading to mixtures of methoxy (e.g. 1, 2) and desmethoxy (e.g. 3, 4) Cinchona alkaloids in planta. Additionally, we report the discovery of the biosynthetic genes that are responsible for this methoxylation, allowing reconstruction of the early biosynthetic steps of desmethoxy and methoxy Cinchona alkaloid intermediates.
Methoxylation of cinchonidinone (15a) and cinchoninone (15b) was first proposed[8a] based on the incorporation of the radio-labeled compounds corynantheal (13) and cinchonidinone (15a) into quinine (1), though the incorporation was marginal (ca. 0.002%).[8b, 8c] Additionally, methoxylated ketone 17a has been detected in C. ledgeriana,[8c] and crude extracts of a cell culture derived from this species showed the presence of NADPH-dependent reduction activity that specifically reduced 17a/b to 1 and 2.[8f] Moreover, a recently discovered O-methyltransferase from C. pubescens can methylate 6′-hydroxycinchoninone (16a/16b).[8d] Together, these data are consistent with – though do not prove – a pathway order of hydroxylation of 15a/b, O-methylation, followed by keto-reduction to form 1 and 2 (Scheme 1). We set out to identify the missing oxidase that would hydroxylate cinchonidinone/cinchoninone (15a/b). In C. pubescens, 1 and 2, along with the C10 – C11 dihydro analogs (5 and 6), accumulate only in roots and stem,[8d] leading us to hypothesize that the 15a/b-hydroxylase would be specifically expressed in these two tissues. RNA-seq data from C. pubescens were therefore mined and a total of 150 putative oxidases (P450s, BBE, and PPOs) that were enriched in these tissues were selected, cloned into Agrobacterium tumefaciens and transiently expressed in N. benthamiana leaves. Infiltration of 15a/b into the treated N. benthamiana leaves, followed by LC-MS analysis of the crude transformed leaf extracts, indicated that none of the enzyme candidates could hydroxylate or otherwise metabolize 15a/b.
This failure to identify 15a/b-hydroxylase, along with previously reported unsuccessful attempts,[8d] prompted us to consider that methoxylation may occur at an earlier stage of the biosynthetic pathway. Strictosidine (12), dihydrocorynantheal (18), and cinchonamine (19) were tested as substrates with the oxidase gene candidates as described above, but no activity could be detected. Therefore, to pinpoint the stage at which methoxylation occurs, we performed untargeted and targeted metabolomics analyses of young tissue from C. pubescens leaves, stem, and root. These analyses revealed the presence of the putative intermediates tryptamine (10), strictosidine (12), corynantheal (13), dihydrocorynantheal (18), cinchoni(di)none (15a/b) (along with cinchonamine (19) and tetrahydroalstonine (20), which are not predicted to be on pathway for 1 and 2) (Figures S2-S7). Notably, for each of these metabolites a compound corresponding to the methoxylated analog was also detected (Figures S2-S7). These observations clearly indicated that two parallel pathways, one with methoxy and one with desmethoxy substrates, operate in C. pubescens (Scheme 2). The detection of 5-methoxytryptamine (21) in C. pubescens suggested that the aromatic hydroxylation occurs either on tryptamine (10) or tryptophan (9). However, the hydroxylated or methoxylated derivatives of tryptophan (9) were not detected in these metabolomic analyses.[9] Additionally, no orthologs of any tryptophan-5-hydroxylase could be detected in the C. pubescens transcriptome. Therefore, we speculated that 5-methoxytryptamine (21) derives from hydroxylation of tryptamine (10) to form serotonin (27), which is then O-methylated to form 21 (Scheme 2).
Currently accepted hypothetical biosynthetic pathway to quinine (1) and congeners 2-4. Highlighted in red are the putative steps installing the methoxy group. The asterisk designates previously reported proteins that have been biochemically characterized from crude Cinchona tissues, but the corresponding gene remains undiscovered. The dashed double bond at C18-C19 in 13 indicates that this compound also occurs in the dihydro form (i.e., with a single bond, see 18).
Revised biosynthetic pathway for Cinchona alkaloids. The key steps and involved enzymes that introduce the aromatic methoxy group and leading to parallel pathways are highlighted in red. For a detailed scheme see Scheme S1.
To identify C. pubescens enzyme candidates that catalyze formation of 5-hydroxytryptamine (27) from tryptamine (10), we BLAST-searched for orthologs of CYP71P1 from Oryza sativa, the only reported functionally characterized tryptamine-5-hydroxylase.[10] We transiently expressed the 15 top candidates in N. benthamiana leaves and infiltrated tryptamine substrate 10 into these infected leaves. LC–MS analysis of the resulting leaf extracts indicated the formation of 27 (m/z 177.102 [M+H]+) only in N. benthamiana leaves expressing either of two highly similar candidate enzymes (97.9% identical amino acids, Figure S8), which we thus named CpT5H1 and CpT5H2 (Figure 1a). Notably, CpT5H1 and CpT5H2 have complementary gene expression profiles (Figure 1b), consistent with the presence of different methoxylated alkaloid types (quinoline-, indole-, yohimbine-, and corynanthe-type) across different C. pubescens tissues (Figures S2-S7, and S9).
(a) Identification and functional characterization of 5-methoxytryptamine (21) biosynthetic genes; (b) expression profiles of identified genes in C. pubescens (* indicates previously reported genes; # denotes previously reported but here functionally repurposed gene); (c) reconstruction of the pathways to early Cinchona alkaloids in N. benthamiana (for further details see Figure S16); (d) peak areas of key metabolites, strictosidines and corynantheal-type alkaloids, produced when only biosynthetic genes are co-expressed in N. benthamiana leaves; and (e) accumulation of the same key metabolites when the transporter gene CpSTTr is co-expressed with the biosynthetic genes in N. benthamiana leaves.
Subsequent O-methylation of serotonin (27) is required to install the methoxy group on 5-methoxytryptamine (21). To identify candidate enzymes that methylate the hydroxyl function of 27, we searched for transcripts annotated as hydroxyindole-O-methyltransferase genes in the C. pubescens transcriptome, focusing on those that were expressed in all tissues (FPKM > 4) and having a similar co-expression profile with either CpT5H1 or CpT5H2 (Pearson correlation coefficient > 0.5). From this process, seven candidates were identified, expressed in Escherichia coli, purified, and assayed in vitro using 27 as a substrate. Among these candidates was the previously[8d] identified 6′-hydroxycinchoninone-O-methyltransferase (CpOMT1) mentioned above. Surprisingly, only CpOMT1 showed methyltransferase activity with 27, and the methylated product was identical to an authentic standard of 5-methoxytryptamine (21) (Figure 1a). This enzyme more efficiently consumed 5-hydroxytryptamine (27) than 6′-hydroxycinchoninone (16a/16b) (substrate conversion after 60 min: > 80% and < 20%, respectively, Figure S10). This finding was corroborated by assays using transient expression in N. benthamiana, where 27 was almost fully converted to 21 after 3h, while no methylation activity on 6′-hydroxycinchoninone (16a/b) could be detected, even after 2d. Together, these data suggest that 27 is the native substrate for CpOMT1. We further tested in planta the catalytic activity of CpOMT1 on two other hydroxy-indole substrates, 5-hydroxytryptophan and N-acetyl-serotonin. We also tested the phenolic substrates quercetin-3-β-glucoside, quercitrin and caffeic acid, since C. pubescens also produces a wide range of phenolic natural products (Figure S11). Only N-acetyl-serotonin was methylated by CpOMT1 to yield melatonin (Figure S12a), though since this compound was not detected in C. pubescens, this enzyme activity may not be physiologically relevant. Moreover, in substrate competition assays, CpOMT1 preferred 27 over N-acetyl-serotonin (65% and 40% of conversion, respectively, Figure S12b). Finally, CpOMT1 appears to be evolutionarily related to aromatic O-methyltransferases from other monoterpenoid indole alkaloid pathways (Figure S13). We henceforth reclassify CpOMT1 as a serotonin-O-methyltransferase.
We then characterized how CpT5H1/2 and CpOMT1 impact the ratio of methoxylated to non-methoxylated products when reconstituted with previously reported downstream Cinchona pathway enzymes in planta. Specifically, we intended to determine if downstream enzymes CpDCS and CpDCE, which produce intermediate dihydrocorynantheal (18) from strictosidine aglycone,[8d] could produce both methoxy and desmethoxy products. However, while previous feeding studies have shown that many downstream monoterpene indole alkaloid pathway enzymes turn over substituted tryptamine analogs,[11] known strictosidine synthase (STR) enzymes do not turn over tryptamine substrates with substituents at the C-5 position.[12] Rationalizing that the C. pubescens strictosidine synthase must be able to turn over both tryptamine and 5-methoxytryptamine, we identified and functionally characterized strictosidine synthase from C. pubescens, (CpSTR) (Figure S14). We also identified two off-pathway enzymes (CpTHAS1 and CpTHAS2) that reduce the reactive strictosidine aglycone into tetrahydroalstonine 20 (Figure S15).
We transiently co-expressed CpT5H, CpOMT1, CpSTR, CpDCS, CpDCE, CpTHAS1, and Catharanthus roseus strictosidine glucosidase (CrSGD, previously[13] shown to accept 10-methoxy strictosidine) in N. benthamiana leaves along with co-infiltration of tryptamine (10) and secologanin (11). LC-MS analysis of the transformed leaf extracts showed the presence of 5-methoxytryptamine (21) in addition to the downstream products strictosidine (7), dihydrocorynantheal (18), tetrahydroalstonine (20), along with the respective methoxylated analogs (Figures 1c and S16). We also noticed that the aldehydes dihydrocorynantheal (18) and methoxy dihydrocorynantheal (24) were reduced into the corresponding alcohols 34 and 35 by endogenous tobacco reductases (Figure S17). While these results demonstrated that CpSTR and all downstream enzymes could turn over both methoxy and desmethoxy intermediates, we noted that all observed dihydrocorynantheal-type products (18, 24, 34, or 35) were produced in low amounts (Figure 2d). In contrast, 10-methoxystrictosidine (22) and strictosidine (12) accumulated to higher levels (Figure 1d). We hypothesized that 10-methoxystrictosidine (22) and strictosidine (12), which are likely produced in the vacuole,[14] are not efficiently exported from the vacuole of N. benthamiana, and thus cannot access the downstream cytosolic enzymes. Therefore, we searched for and identified an ortholog of a Catharanthus roseus strictosidine transporter[15] in the C. pubescens transcriptome (CpSTTr), which, when co-expressed in N. benthamiana leaves along with other biosynthetic enzymes, improved the metabolic flux of both strictosidine (12) and 10-methoxystrictosidine (22), as evidenced by the relatively higher levels of downstream products, most notably 10-methoxydihydrocorynantheol (35) (Figure 1e).
(a) synthesis of the labeled precursors tryptamine-(indole-d5) (10a) and 5-methoxytryptamine-(O-methyl-13C, d3) (21a), which were separately fed to the roots and leaves of C. pubescens; (b) key methoxylated Cinchona alkaloids 1 and 2, and extracted ion chromatograms of the corresponding isotope-labeled analogs, MS isotopic patterns, MS/MS spectrum (20.0-50.0 eV), and selected putative MS/MS fragments I-V, all evidencing the incorporation of 21a and 10a.
Methoxylated alkaloids appear to be present in larger quantities than desmethoxy alkaloids in Cinchona mature plants.[3a, 16] We also observed substantial overproduction of methoxylated products compared to desmethoxy products in N. benthamiana reconstitution, though only when the transporter CpSTTr was included. The observed ratios of combined production of methoxylated aldehyde 24 and corresponding alcohol 35 versus combined formation of their desmethoxylated counterparts 18 and 34 were 1.23 and 0.23 with and without the transporter, respectively. This indicates that CpSTR, CpDCS, and CpDCE are not specific for desmethoxylated substrates. This is particularly remarkable for CpSTR, since orthologs from C. roseus (CrSTR) and Rauvolfia serpentina (RsSTR) are highly specific for tryptamine.[17] We separately tested CpSTR in N. benthamiana with tryptamine (10), 5-methoxytryptamine (21) and 5-hydroxytryptamine (27) confirming that this enzyme turns over all of these substrates (Figures S18 and S19). This is consistent with early studies using enzyme activity purified from Cinchona robusta suspension culture.[18] Notably, it has been shown that a single point mutation enables CrSTR to accept C5-substituted tryptamine analogs,[12] but this mutation is not observed in CpSTR. Instead, we noted that CpSTR is phylogenetically distant to CrSTR and RsSTR (Figure S20), which are part of alkaloid pathways where the aromatic methoxylation occurs at a late biosynthetic stage, after formation of strictosidine.
We next wanted to conclusively demonstrate that hydroxylation at this early biosynthetic stage actually occurs in C. pubescens, particularly since CpOMT1 can also methylate 6′-hydroxycinchoninone (16a/16b), albeit at low efficiency. Therefore, we synthesized isotopically labeled tryptamine-(indole-d5) (10a) and 5-methoxytryptamine-(O-methyl-13C, d3) (21a) and administrated these two compounds separately to roots and leaf tissues from in vitro generated C. pubescens plantlets (Figure 2a). Gratifyingly, both labeled substrates were incorporated into quinine (1) and quinidine (2) (Figure 2b) along with other related alkaloids (Figures S21-S29), as evidenced by LC-MS analysis. In particular, the incorporation of 5-methoxytryptamine (21) into 1 and 2, along with the detection of 5-methoxytryptamine-(indole-d4) 21b in the feeding experiments with labeled tryptamine (10a) (Figure S25), firmly demonstrated that 1 and related methoxylated metabolites derive from 5-methoxytryptamine (21), which, in turn, is formed from tryptamine (10) (Schemes 2 and S1).
Since the desmethoxylated alkaloid corynantheal (13) had been suggested to be an intermediate of the methoxylated metabolites 1 and 2,[8b] we also wanted to conclusively determine whether 13 would be incorporated using our established system for feeding studies. Since labeled 13 was not available, we enzymatically synthesized an aromatic ring-d4 labeled form of the dihydro analog 18 (i.e., 18a, Figure S30), to track its incorporation into the downstream dihydro quinine-like compounds present in Cinchona. Feeding tissues of Cinchona plantlets with 18a followed by LC-MS analyses of the resulting extracts revealed that 18a was incorporated into dihydrocinchonidine and dihydrocinchonine (7 and 8) (Figure S30). Importantly however, there was no observable conversion of 18a into the dihydro methoxylated analogs 5 and 6. Moreover, the labeled isotopic pattern of 18a was observed in a metabolite tentatively assigned as dihydrocinchonamine 32 (a mid-pathway biosynthetic shunt alkaloid, Figure S31). Hence, this experiment corroborated the finding that methoxylation occurs prior to the formation of 18 ((dihydro)corynantheal) and 32 ((dihydro)cinchonaminal)). In addition, these feeding studies also suggest that the biogenic relationship between 1-4 and the respective dihydro analogs 5-8 diverge early in the pathway, most likely by equilibration of the cyclized forms of strictosidine aglycone as previously[8d] proposed.
In conclusion, metabolomics, transcriptomics and feeding experiments using stable isotope-labeled precursors demonstrate that the biosynthesis of quinine (1) and related methoxylated Cinchona alkaloids are initiated by the hydroxylation of tryptamine, followed by O-methylation to produce 5-methoxytryptamine. The co-occurrence of both tryptamine and 5-methoxytryptamine in planta and the promiscuity of the pathway enzymes drive concomitant parallel formation of methoxylated and desmethoxylated alkaloids in C. pubescens by the same pathway enzymes. Along with the discovery of improved product formation gained by inclusion of a vacuolar transporter in reconstitution experiments, this work sets the stage for production of valuable Cinchona alkaloids via synthetic biology.
Supporting Information
The Supporting Information contains detailed methods and materials. Likewise contained are supplementary scheme, figures and data including chemical structures, LC-MS results, gene nucleotide sequences, phylogenetic analysis, and characterization data of synthesized labeled compounds. The authors have cited additional references within the Supporting Information.[19-35]
Acknowledgements
We are grateful to Dr. Maritta Kunert and Sarah Heinicke for assistance with mass spectrometry, to Jens Wurlitzer for help with some genes cloning, and to Katrin Luck for RNA isolation from roots. Dr. Allwin McDonald is acknowledged for providing E. coli mutant harboring RgnTDC. Eva Rothe and the greenhouse team are thanked for taking care of plants. This work was supported by the European Research Council (Grant 788301) and the Max Planck Society. The plant and tissues arts in Figures 1 and 2 were created with BioRender.com. The heatmap in Figure 1 was generated with Morpheus (https://software.broadinstitute.org/morpheus).
Footnotes
↵* E-mail: : oconnor{at}ice.mpg.de
References
