Elucidation of tropane alkaloid biosynthesis in Erythroxylum coca using a microbial pathway discovery platform
Edited by Ian Baldwin, Max-Planck-Institut fur chemische Okologie Abteilung fur Molekulare Okologie, Jena, Germany; received September 14, 2022; accepted October 25, 2022
Significance
We used a yeast platform to identify all unaccounted enzymatic steps of the tropane alkaloid biosynthetic pathway in Erythroxylaceae. We provide multiple lines of evidence that biosynthesis of tropane alkaloids independently arose at least twice during the evolution of land plants. The coca pathway includes an alternative route for the biosynthesis of critical polyamines, as well as the missing enzymes responsible for the second ring closure and formation of the 2-carbomethoxy moiety, which is exclusive to Erythroxylaceae and responsible in part for the euphorigenic properties of its alkaloids. Our work demonstrates unique enzyme families were independently recruited at nearly every step of the tropane alkaloid pathway, providing insights into the development of biosynthetic diversity in angiosperms.
Abstract
Tropane alkaloids (TAs) are heterocyclic nitrogenous metabolites found across seven orders of angiosperms, including Malpighiales (Erythroxylaceae) and Solanales (Solanaceae). Despite the well-established euphorigenic properties of Erythroxylaceae TAs like cocaine, their biosynthetic pathway remains incomplete. Using yeast as a screening platform, we identified and characterized the missing steps of TA biosynthesis in Erythroxylum coca. We first characterize putative E. coca polyamine synthase- and amine oxidase-like enzymes in vitro, in yeast, and in planta to show that the first tropane ring closure in Erythroxylaceae occurs via bifunctional spermidine synthase/N-methyltransferases and both flavin- and copper-dependent amine oxidases. We next identify a SABATH family methyltransferase responsible for the 2-carbomethoxy moiety characteristic of Erythroxylaceae TAs and demonstrate that its coexpression with methylecgonone reductase in yeast engineered to express the Solanaceae TA pathway enables the production of a hybrid TA with structural features of both lineages. Finally, we use clustering analysis of Erythroxylum transcriptome datasets to discover a cytochrome P450 of the CYP81A family responsible for the second tropane ring closure in Erythroxylaceae, and demonstrate the function of the core coca TA pathway in vivo via reconstruction and de novo biosynthesis of methylecgonine in yeast. Collectively, our results provide strong evidence that TA biosynthesis in Erythroxylaceae and Solanaceae is polyphyletic and that independent recruitment of unique biosynthetic mechanisms and enzyme classes occurred at nearly every step in the evolution of this pathway.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Tropane alkaloids (TAs) are a diverse class of plant-specialized metabolites (1). These heterocyclic nitrogenous compounds are identified by their N-methyl-8-azabicyclo[3.2.1]-octane core structure, found in more than 200 unique alkaloids (1–3). TAs are synthesized in seven plant orders with scattered distribution across angiosperms, including Malpighiales (Erythroxylaceae) and Solanales (Solanaceae) (4, 5). TAs possess varied pharmaceutical properties and are often used as starter molecules for drug development (2). The methylated nitrogen of the TA core is analogous to acetylcholine and can competitively inhibit the muscarinic acetylcholine receptors (6). One of the most infamous TAs is cocaine, exclusively found in the leaves of Erythroxylum coca and Erythroxylum novogranatense (7–9). TAs of the Erythroxylaceae, including cocaine, are characterized by a 3β stereospecific conformation and binding of the aromatic ring at the 3β position to specific receptor sites blocks reuptake of norepinephrine, serotonin, and dopamine, explaining their anesthetic and euphorigenic properties (10–12). In contrast, two of the most pharmacologically important TAs of the Solanaceae, atropine and scopolamine, possess an aromatic ring with a 3α configuration (13). The World Health Organization considers these TAs as essential medicines as they and their semi-synthetic derivatives are used to treat a range of diseases, including post-operative nausea, tremors, motion sickness, and chronic obstructive pulmonary disease (14).
TA biosynthesis occurs in three stages: 1) the first nitrogen-containing heterocyclic ring closure, 2) the second ring closure forming the bicyclic TA core scaffold, and 3) modifications via the addition of functional groups to the bicyclic core. Although TA biosynthesis has predominantly been studied in Solanaceae, multiple lines of evidence point towards an independent evolution of TA biosynthesis in Erythroxylaceae (2). Examples of this independent evolution are found in the enzyme classes and catalytic mechanisms of characterized steps in TA biosynthesis (Fig. 1). Methylecgonone reductase (EcMecgoR) belongs to the aldo-keto reductase family of enzymes instead of the short-chain dehydrogenases/reductases, which include the tropinone reductases (TRs) found in Solanaceae (1). Erythroxylaceae 3-oxoglutaric acid synthase (EcOGAS, previously referred to as pyrrolidine ketide synthase or PKS/PYKS based on its Solanaceae analogs) does not directly synthesize 4-(1-methyl-2-pyrrolidinyl)-3-oxobutanic acid (MPOB) from the N-methyl-Δ1-pyrrolinium cation (NMPy) and malonyl-CoA but instead forms 3-oxoglutaric acid from two units of malonyl-CoA (Fig. 1). While similar type III polyketide synthases responsible for this step are found in Solanaceae, recent evidence indicates these enzymes are the products of repeated evolution and have independently evolved within their respective lineages (15, 16). Erythroxylaceae and Solanaceae have also recruited different acyltransferase families for the esterification of the tropane ring. In E. coca, the last step of cocaine biosynthesis is catalyzed by a BAHD acyltransferase called cocaine synthase (EcCS), which condenses coenzyme A (CoA) thioesters like benzoyl-CoA with methylecgonine (17, 18). In contrast, Solanaceae such as Atropa belladonna have evolved a serine carboxypeptidase-like acyltransferase called littorine synthase (AbLS) which utilizes 1-O-β-phenyllactoylglucose formed by a uridine diphosphate (UDP)-dependent glucosyltransferase (UGT) as an acyl donor for tropine to produce littorine (18, 19). The recruitment of two disparate acyltransferase classes for TA biosynthesis in Erythroxylaceae and Solanaceae provides further evidence that these pathways evolved independently between the two plant families. The elucidation of the remaining missing steps in these pathways may reveal insights into the convergent evolution of TA biosynthesis and enable the development of TA drugs and derivatives with novel bioactivity.
Fig. 1.
TA biosynthesis in Solanaceae has been well characterized owing to well-established tools for pathway manipulation via Agrobacterium-mediated transformation (20, 21), RNAi/VIGS knockdown (19), and CRISPR-Cas9 gene knockout techniques (22, 23). Lack of comparable tools for pathway elucidation and tissue culture techniques, the recruitment of novel (or nonhomologous) enzymes compared to those in Solanaceae, and the absence of publicly available Erythroxylum genomes complicates the elucidation of the complete Erythroxylaceae TA pathway using traditional approaches.
Brewer’s yeast (Saccharomyces cerevisiae) is an established host for reconstructing biosynthetic pathways for complex plant alkaloids, including morphinan alkaloids like thebaine and hydrocodone from Papaveraceae (10), monoterpene indole alkaloids like vindoline and catharanthine from Apocynaceae (11), and TAs like hyoscyamine, scopolamine, and their derivatives from Solanaceae (3, 12). In addition to hosting much of the requisite cellular structures (e.g., endoplasmic reticulum) and machinery (e.g., enzyme N-glycosylation) to support the heterologous expression of plant-derived enzymes, yeast can be manipulated for high-throughput gene candidate screening and genomic expression of full pathways via well-characterized toolkits (e.g., CRISPR-Cas9) with far shorter design-build-test cycle times relative to conventional plant expression systems (13, 14). Our previous demonstration of solanaceous TA biosynthesis in yeast (3, 12) indicates that it can support the production and analysis of TA intermediates and derivatives.
Here, we develop a yeast platform for the efficient discovery of TA biosynthetic genes from E. coca transcriptomes and use this platform to demonstrate that Erythroxylaceae not only recruited different enzymes compared to those in the Solanaceae at several points in the TA pathway, but also evolved novel biosynthetic routes to catalyze the two cyclization steps in the tropane ring formation. We first use metabolomic analysis to show that polyamine levels in E. coca leaf extracts are inconsistent with the canonical route to pyrrolidine ring formation via putrescine N-methyltransferase (PMT), and characterize the activity of putative E. coca polyamine genes in vitro, in planta, and in the yeast discovery platform to demonstrate that the first ring closure in E. coca instead occurs through the concerted action of spermidine N-methyltransferases (SMTs) and both flavin-and copper-dependent amine oxidases. We then identify a SABATH family methyltransferase with activity on MPOB and a novel CYP81-family monooxygenase that catalyzes the second ring closure to form methylecgonone from E. coca transcriptomes, which are collectively responsible for the retention of the carbomethoxy (CMO) group that differentiates Erythroxylaceae TAs from those of Solanaceae. Finally, we show that the newly identified E. coca genes complete the biosynthetic pathway for Erythroxylaceae TAs by demonstrating the de novo production of methylecgonine in our engineered yeast platform. Our work provides insights into the diverse biocatalytic strategies that evolved in these plant families for the production of the same class of alkaloids, and may facilitate the development of TA derivatives with novel bioactivities.
Results
The Committed Step of TA Biosynthesis in E. coca Proceeds Via Spermidine N-methylation.
Metabolic profiling of E. coca leaf extracts via liquid chromatography and tandem mass spectrometry (LC-MS/MS) revealed the presence of putrescine, spermidine, spermine, and N-methylspermidine but not N-methylputrescine (NMPUT) in buds, L1 leaves, L2 leaves, and L3 leaves (Fig. 2A and SI Appendix, Fig. S1)—in contrast to the abundant NMPUT found in TA biosynthetically active tissues of Solanaceae (24). A homology-based search of Erythroxylaceae species for PMT-like sequences failed to yield any close matches, and enzymatic characterization of E. coca tissue lysates failed to show PMT activity. Collectively, these metabolite and enzyme activity observations are inconsistent with a biosynthetic route from polyamines to the first ring closure via direct putrescine N-methylation as in Solanaceae and—given its abundance in E. coca leaves—instead suggest an alternate route in Erythroxylaceae via spermidine. In vitro enzyme assays using crude E. coca tissue extracts with spermidine as the substrate revealed S-adenosylmethionine (SAM)-dependent conversion to N-methylspermidine in L3 and L1 leaves (SI Appendix, Figs. S2 and S3), indicating the presence of one or more spermidine N-methyltransferases (SMTs). As no SMTs have been reported, we instead used sequences of biochemically characterized solanaceous PMTs, which we hypothesized may be the closest sequence match for SMT, as queries for a tBLASTn search of an in-house transcriptome database of TA biosynthetically active E. coca leaf tissues. We identified three full-length protein sequences, which more closely resembled spermidine synthases (SPDS) than PMTs and were thus denoted EcSPDS1-3 (Dataset S1). We cloned EcSPDS1-3 into expression vectors for heterologous expression and purification from Komagataella phaffii. Following affinity column purification, proteins were analyzed by SDS-PAGE (SI Appendix, Fig. S4–S6), which indicated molecular weights of 39.0, 36.9, and 38.5 kDa for EcSPDS1, 2, and 3, respectively.
Fig. 2.
Consistent with our observations from crude lysate assays, enzyme assays with purified recombinant EcSPDS1-3 failed to show PMT activity but instead revealed SPDS and SMT activities (Fig. 2 B–D). Both EcSPDS1 and EcSPDS2 showed SMT activity in which spermidine is methylated at the (internal) N4 position using the co-substrate SAM (Fig. 2 C and D and SI Appendix, Fig. S2), whereas EcSPDS2 and EcSPDS3 exhibited SPDS activity using decarboxylated SAM (dcSAM) as a co-substrate (Fig. 2 B and D). In particular, EcSPDS2 alone converted putrescine to N-methylspermidine when supplied with both dcSAM and SAM (Fig. 2D). Based on these observations, we have renamed EcSPDS1, EcSPDS2, and EcSPDS3 to EcSMT, EcSPMT, and EcSPDS, respectively. To elucidate phylogenetic relationships between these enzymes, we performed sequence alignments against biochemically characterized SPDS, spermine synthase (SPS), and PMT sequences and evaluated clustering patterns in the resulting dendrogram (Fig. 2E). All characterized PMTs cluster into a separate clade from SPDSs and SPSs, consistent with the proposed early divergence of PMT from the ancestral SPDS family (25, 26). In contrast, there appears to be no clear clustering between characterized SPDSs and SPSs, and all three E. coca proteins of interest are dispersed amongst the SPDSs and SPSs, suggesting a more recent evolution in substrate and co-substrate binding motifs.
As TA biosynthesis in E. coca is localized to buds, L1 and L2 leaves, we evaluated whether the tissue distribution of the three newly identified E. coca polyamine genes is consistent with a role in TA biosynthesis. We measured the relative expression of SAM decarboxylase (EcSAMDC, which supports SPDS activity), EcSMT, EcSPMT, and EcSPDS across E. coca tissues by qPCR (SI Appendix, Table S1). Expression profiles of EcSAMDC, EcSPMT, and particularly EcSMT are upregulated in buds, L1 and L2 leaves, whereas EcSPDS showed a basal expression level comparable to that of the Ec10131 and Ec6409 reference genes (27) (SI Appendix, Fig. S7). EcSMT and EcSPMT co-expressed strongly across tissues with the established TA genes ornithine decarboxylase (EcODC) and EcMecgoR (1, 28) (Pearson’s correlation coefficient: EcSMT and EcODC, r = 0.84, P < 0.05; EcSPMT and EcODC, r = 0.81, P < 0.10; EcSMT and EcMecgoR, r = 0.84, P < 0.05; EcSPMT and EcMecgoR, r = 0.74, P < 0.10), whereas no correlation was detected for EcSPDS. These observations support a specific role for EcSMT and EcSPMT, but not EcSPDS, in TA biosynthesis and provide further evidence of an alternative biosynthetic route to TAs via spermidine N-methylation in E. coca.
We developed a testing platform for in vivo characterization of the early E. coca TA pathway genes by designing a yeast strain engineered for the overproduction of spermidine. As in E. coca, in S. cerevisiae spermidine is biosynthesized from the four-carbon diamine putrescine by SPDS (Spe3p), whose aminopropyl donor group is supplied by SAM decarboxylase (Spe2p). To determine whether increased dcSAM availability is needed to support the testing of SPDS in yeast, we overexpressed SPE2, SPE3, both SPE2 and SPE3, or a blue fluorescent protein (BFP) negative control from low-copy plasmids in a yeast strain (CSY1242) previously engineered for increased accumulation of putrescine via deletion of the polyamine regulatory genes S-methyl-5′-thioadenosine phosphorylase (Meu1p) and ODC antizyme-1 (Oaz1p) coupled with overexpression of glutamate N-acetyltransferase (Arg2p), arginase (Car1p), polyamine oxidase (Fms1p), ODC (Spe1p), Avena sativa arginine decarboxylase (AsADC), and Escherichia coli agmatinase (speB) (29). LC-MS/MS analysis of culture supernatants following 72 h growth of transformed strains in selective media revealed a synergistic effect of overexpressing both SPE2 and SPE3, resulting in a 36-fold increase in spermidine levels relative to the BFP control and 3.9- or 12-fold increases relative to overexpression of SPE2 or SPE3 alone, respectively (SI Appendix, Fig. S8A).
We then characterized the activities of the putative SPDSs from E. coca (EcSPDS, EcSPMT) in the yeast platform. As SPE2 potentiates SPDS activity, we co-expressed either EcSPDS or EcSPMT, native yeast SPE3, or a BFP control with SPE2 from low-copy plasmids in CSY1242. LC-MS/MS analysis of culture supernatants after 72 h growth of transformed strains in selective media indicated that although EcSPMT, like native Spe3p, can function as a SPDS when paired with Spe2p in yeast, EcSPDS does not retain its activity in this context (SI Appendix, Fig. S8B). We constructed a spermidine overproduction strain (CSY1340) by integrating expression cassettes for SPE2, SPE3, and EcSPMT and disrupting the endogenous SPS gene SPE4 (which consumes spermidine and whose product spermine is not essential for yeast (15)), in the genome of the previously engineered putrescine overproduction strain CSY1242. We then demonstrated that unlike our previous observations in vitro, EcSMT but not EcSPMT exhibits spermidine N-methyltransferase activity when expressed from a low-copy plasmid in CSY1340, as indicated by LC-MS/MS analysis of culture supernatants following 72 h growth of transformed strains in selective media (SI Appendix, Fig. S9).
E. coca Synthesizes N-methyl-Δ1-pyrrolinium Cation (NMPy) via Spermidine N-methylation and Dual Amine Oxidation.
As Erythroxylaceae appear to have evolved one or more N-methyltransferases capable of using spermidine instead of putrescine (as in Solanaceae), a novel enzymatic mechanism is required for conversion of N-methylspermidine to NMPy, which condenses with 3-oxoglutarate produced by type III polyketide synthases (PYKS or OGAS) to form MPOB in both families (15, 16). If NMPy formation in Erythroxylaceae, like in Solanaceae, proceeds via non-enzymatic cyclization of a methylaminoaldehyde intermediate (e.g., 4-methylaminobutanal), we hypothesized that one or more amine oxidases are required for its biosynthesis from N-methylspermidine. Two classes of amine oxidases are found in plants: flavin-dependent amine oxidases (AOFs) catalyze polyamine-shortening (aminopropyl removal) reactions primarily in the cytosol, whereas copper-dependent amine oxidases (AOCs) catalyze the oxidation of primary amine groups to aldehydes in peroxisomes (30). We identified three AOF (EcAOF1-3) and six AOC (EcAOC1-6) candidates from the transcriptome of E. coca. We considered two hypotheses for amine oxidase-catalyzed NMPy formation from N-methylspermidine: (i) oxidation of both primary amines to aldehydes to form 4-(methyl(3-oxopropyl)amino)butanal, followed by non-enzymatic intramolecular cyclization to NMPy via retro-Michael addition; or (ii) oxidative shortening of N-methylspermidine to N-methylputrescine by an AOF, followed by oxidation to 4-methylaminobutanal by an AOC and non-enzymatic cyclization to NMPy (as in Solanaceae).
Both hypotheses for amine oxidase-catalyzed NMPy formation from N-methylspermidine were initially tested in the yeast platform. We tested hypothesis (i) by co-expressing each E. coca AOF and AOC candidate with EcSMT from low-copy plasmids in the spermidine overproduction yeast strain CSY1340, but observed accumulation of neither 4-(methyl(3-oxopropyl)amino)butanal nor NMPy by LC-MS/MS after 72 h growth of transformed strains. To evaluate hypothesis (ii), we first co-expressed each of EcAOF1-3 or a BFP control with EcSMT in CSY1340 and observed accumulation of N-methylputrescine in the strain expressing EcAOF1 (Fig. 2F), consistent with a mechanism involving sequential activity of an AOF and AOC. We then screened all six E. coca AOC candidates (EcAOC1-6), a copper-dependent N-methylputrescine oxidase (MPO) from Datura metel previously engineered for improved activity in the yeast peroxisome (denoted DmMPO1) as a positive control, or BFP and the remaining AOF candidates (EcAOF2, EcAOF3) as negative controls via co-expression with EcSMT and EcAOF1 from low-copy plasmids in CSY1340. LC-MS/MS analysis of culture supernatants after 72 h growth in selective media revealed the accumulation of hygrine only in the positive control and in strains expressing either EcAOC1 or EcAOC2 (Fig. 2G). Hygrine is a major side product of the first tropane ring formation that can be formed via spontaneous condensation of NMPy with endogenous yeast keto-metabolites (e.g., acetate and acetoacetyl-CoA). We have previously established that because hygrine only accumulates in yeast engineered with the requisite enzymes for NMPy biosynthesis, it can be used as a proxy for NMPy due to its vastly improved MS/MS detectability (31). This result indicated that both EcAOC1 and EcAOC2 are capable of N-methylputrescine oxidation to 4-methylaminobutanal and NMPy, supporting hypothesis (ii) that NMPy biosynthesis in E. coca proceeds via sequential activity of an AOF (EcAOF1) and an AOC (EcAOC1/2).
Subcellular localization can provide insight into enzymes’ functions, and by extension, phylogenetic origins. As plant AOFs and AOCs differ in their canonical subcellular location (30), we verified the localization of the newly identified E. coca amine oxidases via fluorescence microscopy of wild-type yeast (CEN.PK2) co-expressing N- or C-terminal GFP fusions of EcAOF1, EcAOC1, or EcAOC2 with a peroxisomal marker (mCherry-Pex3p) from low-copy plasmids. Consistent with its lack of detectable localization signals, EcAOF1-GFP appears to form cytosolic inclusion bodies while GFP-EcAOF1 is soluble and cytosolic (SI Appendix, Fig. S10). Consistent with the peroxisomal localization of AOCs like the analogous MPOs in Solanaceae (32), and with the presence of C-terminal peroxisome targeting sequences (PTSs), both GFP-EcAOC1 and GFP-EcAOC2 co-localize with mCherry-Pex3p to yeast peroxisomes, whereas EcAOC1-GFP and EcAOC2-GFP with masked C-terminal PTSs remain in the cytosol (SI Appendix, Fig. S10). These results suggest that the newly identified E. coca amine oxidases follow the expected pattern of subcellular localization for their associated enzyme families.
To further investigate phylogenetic relationships between the two classes of amine oxidases recruited for TA biosynthesis in Erythroxylaceae, we performed sequence alignments of EcAOF1 and EcAOC1/2 against known AOFs and AOCs, respectively, and evaluated clustering patterns in the resulting dendrograms (SI Appendix, Fig. S11). The N-methylspermidine oxidase EcAOF1 clusters with other known plant polyamine oxidases, including spermine oxidases isolated from Arabidopsis thaliana (48% identity to Q9FNA2) and Oryza sativa (50% identity to A0A0P0XM10, 51% identity to Q0J290) and a spermidine oxidase from Zea mays (51% identity to O64411) (SI Appendix, Fig. S11A). The MPOs EcAOC1 and EcAOC2 cluster closely with known copper-dependent amine oxidases (SI Appendix, Fig. S11B). In particular, EcAOC1 clusters more closely with AOCs from Malus domestica (84.35% AA identity to AIS23644.1), Vitis vinifera (86.53% AA identity to XP_002273532.2), Ricinus communis (85.55% AA identity to XP_002527922.1), Glycine max (84.65% AA identity to XP_003551224.1), and Medicago truncatula (86.24% AA identity to XP_003613133.2) than with known Solanaceae N-methylputrescine or primary amine oxidases from A. belladonna (AbMPO2), Nicotiana sylvestris (XP_009778427.1), and Nicotiana tabacum (BAF49520.1) (SI Appendix, Fig. S11B). EcAOC2 clusters closely with AOCs from M. domestica (83.59% AA identity to AIS23648.1, 84.64% AA identity to AIS23647.1) and R. communis (84.63% AA identity to XP_002511334.1). Collectively, these observations are consistent with independent recruitment of both AOF and AOC families for TA biosynthesis in Erythroxylaceae.
To validate the concerted action of previously and newly identified E. coca genes implicated in the first tropane ring closure, as well as to verify the observed in vitro dual function of EcSPMT in the context of a plant host, we reconstructed the Erythroxylaceae TA biosynthetic pathway up to NMPy, MPOB, and hygrine via transient co-expression of E. coca ODC (EcODC), EcSPMT, EcSMT, EcAOF1, EcAOC1/2, and/or EcOGAS1/2 in Nicotiana benthamiana (Fig. 2H). As endogenous putrescine levels in N. benthamiana are sufficient to support heterologous TA biosynthesis (33) and ODC overexpression can significantly increase the putrescine pool in transgenic plants (34), we used EcODC as a co-expression partner gene for all transient transformations. Co-expression of EcODC and EcSPMT alone was sufficient to enable comparable accumulation of N-methylspermidine as when EcSMT was additionally co-expressed. Taken together with our observations in vitro and in yeast, these data confirm that EcSPMT can function as both SPDS and SMT in E. coca TA biosynthesis. As in yeast, expression of EcAOF1 enables the conversion of N-methylspermidine to N-methylputrescine, which is subsequently converted to N-methylpyrrolinium and MPOB (as indicated by the product of its spontaneous decarboxylation, hygrine) upon expression of EcAOC1/2 and EcOGAS1/2. Collectively, our results show that the first tropane ring closure in E. coca occurs via an alternative biosynthetic route involving bifunctional SPDS/N-methyltransferases and a flavin-dependent polyamine oxidase compared to the canonical route via PMT observed in Solanaceae.
A SABATH Methyltransferase Enables 2-CMO Group Retention in Erythroxylaceae TAs.
Tian et al. (15) recently reported the characterization of two type III polyketide synthases from E. novogranatense (EnPKS1, EnPKS2) that catalyze the condensation of two malonyl-CoA units to form 3-oxoglutarate for MPOB biosynthesis. In Solanaceae, CYP82M3-mediated cyclization of MPOB to tropinone coincides with spontaneous decarboxylation of the 2-position. In contrast, retention of the carboxyl group as a 2-carboxymethyl ester in Erythroxylaceae TAs like cocaine implicates a missing methyltransferase—likely of the SABATH family (35)—in the second tropane ring formation step. We searched for SABATH methyltransferases in the transcriptome of E. coca using a tBLASTn search with a salicylic acid methyltransferase from Clarkia breweri (CbSAMT) and a jasmonic acid methyltransferase from A. thaliana (AtJAMT) as sequence queries, and identified three putative candidates (denoted EcSABATH1-3). Sequence alignments against known SABATH methyltransferases and dendrogram analysis showed that EcSABATH1 clusters closely with previously characterized salicylic acid methyltransferases from multiple species as well as the aforementioned AtJAMT (46.98% AA identity). EcSABATH2 and EcSABATH3 cluster with biochemically characterized paraxanthine methyltransferases from A. thaliana (~37% AA identity) (Fig. 3A).
Fig. 3.
To test EcSABATH1-3 for MPOB carboxyl methyltransferase activity, we used a yeast strain (CSY1246) previously engineered to facilitate the screening of gene candidates for the second tropane ring formation step using solanaceous TA enzymes (29). CSY1246 harbors three metabolic modules to support this function (SI Appendix, Table S2): Module I increases the accumulation of putrescine via overexpression of glutamate N-acetyltransferase (Arg2p), arginase (Car1p), polyamine oxidase (Fms1p), ODC (Spe1p), A. sativa arginine decarboxylase (AsADC), and E. coli agmatinase (speB); Module II-a enables conversion of putrescine to NMPy via A. belladonna PMT 1 (AbPMT1) and D. metel MPO 1 engineered for improved activity in the yeast peroxisome (DmMPO1); and Module II-b comprises TR I from D. stramonium (DsTR1) for conversion of tropinone to tropine, whose detection can be used as an unambiguous positive test if both ring formation steps (e.g., PYKS/OGAS + CYP82M3) are functional. We first evaluated the heterologous activities of PYKS and OGAS enzymes from Solanaceae (AbPYKS) and Erythroxylaceae (EcOGAS1 and EcOGAS2, E. coca homologs of EnPKS1 and EnPKS2 (15)) via co-expression with A. belladonna tropinone synthase (AbCYP82M3) from low-copy plasmids in CSY1246. LC-MS/MS analysis of culture supernatants following 72 h growth of transformed strains indicated that while all three PYKS/OGAS enzymes are functional in yeast, AbPYKS enables increased flux through MPOB to tropine and is thus a better candidate to support SABATH methyltransferase screening (SI Appendix, Fig. S12). We next co-expressed each of the three SABATH methyltransferase candidates or a BFP control with AbPYKS from low-copy plasmids in CSY1246 and analyzed the accumulation of NMPy, MPOB, and MPOB methyl ester (MPMOB) in the culture medium via LC-MS/MS after 72 h growth of transformed strains. Only EcSABATH1 resulted in the production of a molecule whose mass transition ([M+H]+ m/z 200 → 84) and retention time were consistent with that of MPMOB (Fig. 3B). Consistent with a role in TA biosynthesis, EcSABATH1 is overexpressed in E. coca L1 leaf tissues relative to L2 and L3 leaves (SI Appendix, Fig. S13). Thus, we designated EcSABATH1 as the missing MPOB methyltransferase (denoted EcMPOBMT) supporting TA biosynthesis in E. coca.
As the tropane moiety of bioactive Solanaceae and Erythroxylaceae TAs differ primarily in the presence of the CMO group, we sought to determine whether expression of EcMPOBMT can convert a biosynthetic pathway for solanaceous TAs into one for Erythroxylaceae-like CMO-TAs. To evaluate the heterologous activity of remaining TA pathway enzyme candidates, we first constructed a MPMOB-producing yeast strain (CSY1341) by integrating expression cassettes for AbPYKS, EcOGAS1, and EcMPOBMT into the genome of the solanaceous NMPy-producing yeast strain CSY1246. We observed that A. belladonna tropinone synthase (AbCYP82M3) is tolerant of MPOB methylation and can convert both MPOB to tropinone and MPMOB to methylecgonone when co-expressed with A. thaliana NADPH:cytochrome P450 reductase (AtATR1) in CSY1341 (SI Appendix, Fig. S14). Accordingly, co-expression of these two enzymes with E. coca methylecgonone reductase (EcMecgoR) (1) resulted in de novo accumulation of methylecgonine after 72 h growth of transformed strains in selective media (SI Appendix, Fig. S14). As DsTR1 and other TRs do not appear to reduce methylecgonone (SI Appendix, Fig. S14) (1), we suspected that MPOBMT and MecgoR activities are together sufficient to convert a solanaceous TA pathway into one producing CMO-TAs, although the production of specific CMO-TA esters is dependent on the substrate tolerance of downstream solanaceous TA enzymes (Fig. 3C).
We examined the ability of MPOBMT and MecgoR activities combined with solanaceous TA pathway activities to enable CMO-TA production in the yeast platform. We co-expressed EcMPOBMT and EcMecgoR or a BFP control from low-copy plasmids in CSY1297, a yeast strain we previously engineered (31) for de novo biosynthesis of the solanaceous TAs hyoscyamine and scopolamine via three additional metabolic modules: Module III, which enables the production of the solanaceous TA acyl donor phenyllactic acid glucoside from phenylalanine via Wickerhamia fluorescens phenylpyruvate reductase (WfPPR) and A. belladonna UDP-glucosyltransferase (AbUGT); Module IV, which enables downstream modifications to the TA ester scaffold for conversion of littorine to scopolamine via A. belladonna littorine mutase (AbCYP80F1), D. stramonium hyoscyamine dehydrogenase (DsHDH), and D. stramonium hyoscyamine 6β-hydroxylase/dioxygenase (DsH6H); and Module V, which comprises the acyltransferase littorine synthase from A. belladonna engineered for activity in the yeast vacuole via N-terminal fusion of the red fluorescent protein DsRed. Following 96 h growth of the transformed strains in selective media, LC-MS/MS analysis of culture supernatant revealed decreased accumulation of solanaceous TAs and intermediates and substantial accumulation of methylecgonine in CSY1297 expressing both EcMPOBMT and EcMecgoR (Fig. 3D). We also observed accumulation of a molecule whose primary mass transition ([M+H]+ m/z 348 → 182) is consistent with that of littorine and/or hyoscyamine ([M+H]+ m/z 290 → 124) with an added CMO group (H [mass = 1] replaced by CMO [mass = 59]) (Fig. 3D), suggesting that littorine synthase, littorine mutase and/or hyoscyamine dehydrogenase may have some tolerance for CMO-TA intermediates. Collectively, these data confirm that MPOBMT and MecgoR activities are sufficient to convert a solanaceous TA pathway into one producing Erythroxylaceae-like TAs.
A CYP81-Family Monooxygenase Catalyzes Methylecgonone Ring Closure in E. coca.
In Solanaceae, cyclization of MPOB to form tropinone requires the cytochrome P450 tropinone synthase (CYP82M3), which is thought to catalyze hydroxylation of the 1 position on MPOB to enable dehydrative iminium formation and intramolecular nucleophilic attack by the β-keto enolate (Fig. 1) (33). We suspected that cyclization of the β-keto acid methyl ester, MPMOB, to methylecgonone in E. coca may proceed via a similar mechanism (Fig. 4A). Although tropinone synthase also exhibits methylecgonone synthase activity (SI Appendix, Fig. S14), no methylecgonone synthase has yet been identified in Erythroxylaceae, and the independent evolution of TA biosynthesis in these families raised the possibility that coca may have recruited an alternate enzyme class for this task.
Fig. 4.
We performed a homology-based search for AbCYP82M3-like sequences in a 454 transcriptome of E. coca and identified five candidate CYP450s (denoted EcCYP1-5) (SI Appendix, Fig. S15). As hydroxylation of secondary metabolites in plants may also be performed by 2-oxoglutarate-dependent dioxygenases (2ODDs) (36), we next expanded our search to include more diverse CYP450 families as well as 2ODDs. Using a hierarchical clustering approach, we searched the transcriptomes of three E. coca plants (denoted 48, 113, and 124), two E. novogranatense plants (denoted 209 and 228), and a low-cocaine-producing Erythroxylum species, Erythroxylum hondense (37), for CYP450 and 2ODD candidates exhibiting similar patterns of expression across the three species as previously identified Erythroxylaceae TA genes (see Methods). This co-expression analysis identified five 2ODD candidates (denoted EcODD1-5) and 12 additional CYP450 candidates (denoted EcCYP6-17) for testing (Fig. 4B). We also included the two putative proline 4-hydroxylases found in E. coca (EcP4H1-2) whose enzyme family’s canonical substrate and activity resemble that of MPMOB hydroxylation (38) for testing. Formation of olivetolic acid via intramolecular cyclization in Cannabis sativa requires the concerted activity of a type III polyketide synthase and a small, polyketide cyclase (PKC)-like dimeric α + β barrel (DABB) protein for carboxyl retention (39, 40), suggesting the involvement of a similar PKC in methylecgonone formation in E. coca. Thus, we also selected the two PKC-like protein candidates containing DABB-like domains found in the E. coca transcriptomes (EcPKC1-2) for investigation.
The candidate hydroxylase and cyclase enzymes were screened in the yeast platform. We co-expressed each of the 24 hydroxylase candidates (EcODD1-5, EcCYP1-17, EcP4H1-2) or a BFP control with the NADPH:cytochrome P450 reductase AtATR1 and both EcPKC candidates via three low-copy plasmids (Fig. 4C) in the MPMOB-producing yeast strain CSY1341 and evaluated accumulation of the expected product, methylecgonone, via LC-MS/MS analysis of culture supernatants after 72 h growth of transformed strains in selective media. One of the 24 candidates, EcCYP9 (denoted methylecgonone synthase), resulted in accumulation of a molecule whose mass transition ([M+H]+ m/z 198 → 166) corresponds to that of methylecgonone (1) (Fig. 4D), and whose retention time matches that of the same mass transition peak produced via expression of tropinone synthase (AbCYP82M3) in the same strain (Fig. 4E). Unlike with olivetolic acid biosynthesis in Cannabis, methylecgonone biosynthesis via EcCYP9 does not appear to require either of the two putative PKCs, as EcCYP9 expression without EcPKC1 and EcPKC2 enabled comparable methylecgonone accumulation (Fig. 4E). We analyzed the phylogenetic relationship between the newly identified methylecgonone synthase (EcCYP9), tropinone synthase (CYP82M3), four previously studied Erythroxylum oxygenases of the CYP79D family (41), and a panel of sequenced plant CYP450s (Fig. 4F). Methylecgonone synthase—which we have designated as EcCYP81AN15 based on its clustering in the CYP81 family—is most closely related to flavonoid/isoflavone 2′-hydroxylases from Jatropha curcas (60.4% identity to XP_012079324), Manihot esculenta (60% identity to XP_021631824 and 59% identity to XP_021632242), and Prunus species (59%, 58%, and 60% identity to XP_007206553, XP_034218022, and PQQ16710, respectively). In contrast, EcCYP81AN15 clusters separately from the four previously characterized Erythroxylum CYP79D-family aldoxime synthases (CYP79D60-63) and, consistent with independent recruitment for TA biosynthesis, is only distantly related to A. belladonna CYP82M3.
To verify the function of the newly identified E. coca genes in their full biosynthetic context, we reconstructed the complete TA pathway for de novo production of methylecgonine in yeast using only Erythroxylaceae genes. We constructed yeast strain CSY1343 by incorporating E. coca genes encoding two additional metabolic modules into CSY1340, in which SPE2, SPE3, and EcSPMT are overexpressed and SPE4 is disrupted to support spermidine overproduction: Module A, enabling conversion of spermidine to MPOB via chromosomally integrated expression cassettes for EcSMT, EcAOF1, EcAOC2, and EcOGAS1; and Module B, enabling conversion of MPOB to methylecgonine via low-copy plasmid expression of EcMPOBMT, EcCYP81AN15, AtATR1, and EcMecgoR. CSY1343 and two variant strains in which EcMPOBMT and/or EcCYP81AN15 activities are absent were cultured for 96 h in selective media and the culture supernatants were analyzed by LC-MS/MS. We observed de novo production of methylecgonine in the full pathway strain CSY1343 only, validating the function of the core E. coca TA pathway in a single host (Fig. 4G).
Discussion
This study identifies the remaining TA biosynthetic genes in Erythroxylaceae to reveal that the entire TA pathway is polyphyletic in origin and that its sources in Solanaceae and Erythroxylaceae are independent of one another. Although convergent evolution in plant specialized metabolism is well-documented (42), in most reported cases independent recruitment of non-orthologous genes into analogous pathways occurs with those containing a limited number of enzymatic steps. The more steps necessary for producing a core structure, the greater the requirement for the concomitant selection of duplicated genes for neofunctionalization, and potentially, the greater the time needed for this evolution to occur. One example of the convergent evolution of specialized metabolites in plants is illustrated by the purine alkaloids. Methylation of the purine xanthine to produce caffeine and theobromine is catalyzed by several distantly related SABATH methyltransferases whose recruitment occurred multiple times in different plant lineages, explaining the scattered distribution of these compounds among the angiosperms (43). Another well-documented example of plant alkaloid convergent evolution is found in the biosynthesis of pyrrolizidine alkaloids, for which the enzyme homospermidine synthase was independently recruited at least four times during the evolution of land plants (44). The scattered distribution of TAs across the angiosperms suggests these pathways must have emerged after the development of Eudicots 120 Mya (5). Although the capacity for TA biosynthesis appears to be shared across at least seven plant orders, the phylogenetic relationships between the respective TA pathways outside Solanaceae and Erythroxylaceae are unclear.
Our discovery of EcSMT and EcSPMT provided evidence of the recruitment of SPDS-like enzymes in the formation of N-methylputrescine in Erythroxylaceae through an alternative route that differs from the solanaceous TA pathway (Fig. 2 B and D). In vitro assays of EcSPDS and EcSPMT demonstrated SPDS activity, whereas further evaluation in S. cerevisiae revealed that EcSPDS caused depletion of spermidine while EcSPMT maintained SPDS activity. Although the reason for EcSPDS’s depletion of spermidine in yeast is unknown, phylogenetic analysis shows that it clusters with Arabidopsis thermospermine synthase (AtACL5), suggesting a potential dual role which may explain its spermidine depletion in vivo. EcSPDS and EcSPMT do not cluster together despite sharing SPDS activity, consistent with the possibility that EcSPDS may be involved in primary metabolism while EcSPMT has undergone neofunctionalization for recruitment into the TA pathway. In vitro assays of EcSMT and EcSPMT demonstrated spermidine N-methyltransferase activity, whereas in yeast EcSPMT lost SMT activity and only functioned as an SPDS. The exact reason for this discrepancy is unknown but may be related to differences in substrate/cofactor availability and/or in the cellular environment (e.g., post-translational modifications). Subsequent in planta expression of EcSPMT validated its dual SPDS/SMT activity (Fig. 2H). Although EcSMT and EcSPMT share SMT activity, EcSMT does not appear to cluster with EcSPMT (49.8% AA identity). EcSPMT’s usage of both SAM and dcSAM as cofactors is novel and has not yet been reported for any enzyme in the family. Prior work demonstrated that SPDS1 from A. thaliana could gain PMT activity while maintaining SPDS activity by changing two to three amino acids in the active site (26). That mutagenesis can allow an SPDS to bind dcSAM and SAM provides evidence that neofunctionalization and recruitment of novel enzymatic functions are possible with this family of enzymes.
Our observation of SMT activity in E. coca is consistent with earlier feeding studies. Radiolabeled feeding studies of E. coca leaves using [5-14C]ornithine demonstrated incorporation through a symmetrical intermediate like putrescine (45). In the context of the newly identified alternate polyamine route, the radiolabeled ornithine is converted to putrescine, incorporated into spermidine, and methylated to form N-methylspermidine. The N-methylputrescine produced following oxidation in the alternate pathway would therefore exhibit similar equal incorporation patterns. A separate study postulated that spermidine could be a potential intermediate in the production of NMPy (46). Spermidine was labeled asymmetrically in its four-carbon moiety ([6-14C]-1,5,10-triazadecane) and used in feeding studies with Nicotiana glutinosa. The degradation analysis revealed the label was incorporated equally at C-2′ and C-5′ of the pyrrolidine rings in nicotine and nornicotine. These results provide evidence that while unusual, spermidine can act as an intermediate in the production of NMPy in both families, and suggest that even within the same family, alternate options for critical early stage intermediates in a pathway can exist.
In contrast to the single amine oxidase involved in Solanaceae TA biosynthesis, our work implicates two amine oxidase families in Erythroxylaceae, providing additional evidence for the convergent evolution of these enzymes. Two classes of amine oxidases exist in plants: those requiring a metal ion cofactor (typically copper) for catalysis and those requiring a covalently bound flavin adenine dinucleotide. Copper-dependent amine oxidases oxidize the carbon adjacent to primary amines resulting in the reduction of molecular oxygen to hydrogen peroxide and producing the corresponding aldehyde and ammonia (47). In Solanaceae, a family-specific subgroup of MPO has evolved from the more general copper-dependent diamine oxidases, important for the formation of lignin and cell wall constituents (31). These MPOs localize to the peroxisome, potentially implicating a transporter to export the N-methyl-Δ1-pyrrolinium cation for further utilization by cytosolic TA enzymes. In contrast, flavin-dependent polyamine oxidases (PAOs) oxidize secondary amines (30). In the case of PAOs localized to the cytosol, the oxidized carbon is located on the exo-side of the N4 atom of spermidine or spermine to produce putrescine or spermidine, respectively, as well as hydrogen peroxide and 3-aminopropanal. EcAOF1 is novel in that it acts on a tertiary amine by oxidizing the carbon on the exo-side of the methylated N4 atom to produce N-methylputrescine and 3-aminopropanal. Not only was a novel flavin-dependent oxidase recruited for E. coca TA biosynthesis, but a second oxidase was required to convert the newly formed NMPUT to NMPy. EcAOC1 and EcAOC2 were recruited from copper-dependent oxidases within the Erythroxylaceae to catalyze this reaction. EcAOC’s closest related family members are all outside the Solanaceae (SI Appendix, Fig. S11B), providing further evidence for the independent evolution of methylputrescine oxidase activity in angiosperms.
Another line of evidence for the role of repeated evolution in TA biosynthesis was presented in a recent study implicating two type III polyketide synthases (EnPKS1/2) in the second ring formation step in E. novogranatense (15). The equivalent PYKS enzymes in Solanaceae from Anisodus acutangulus, D. stramonium, A. belladonna, and Solanum lycopersicum (15, 16) belong to a separate clade from those of the Erythroxylaceae, although both clades perform decarboxylative condensation of two malonyl-CoA subunits to produce 3-oxoglutaric acid. Furthermore, the comparison of the enzyme active site residues reveals significant differences that are consistent with their repeated evolution from ancestral enzymes such as chalcone synthases or stilbene synthases present in their respective lineages. Our data on EcOGAS1/2, the orthologous enzymes to EnPKS1/2 in E. coca (100% AA identity), are also consistent with those reported by Tian et al. (15) and Kim (48).
A unique and definitive feature of Erythroxylaceae TAs is the presence of a 2-CMO group on the tropane ring. Previous chemical analog studies coupled with kinetic binding measurements revealed that the 2-carboymethoxy group is critical for cocaine binding to the dopamine transporter and thus for the subsequent reuptake inhibition of dopamine (49, 50). The absence of MPOB methyltransferase activity in solanaceous plants results in spontaneous decarboxylation of this group during ring closure due to the unstable nature of β-keto acids (51). The resulting TAs have anticholinergic properties but are unable to interact with the dopamine receptor. The recruitment of MPOBMT in coca but not in nightshades may reveal novel insights into the ecological relationships (e.g., plant–herbivore interactions) that shaped the evolution of chemical diversity in these species (52). We postulate that the ancestral methyltransferase for EcMPOBMT was most likely involved in the production of methylsalicylate, an abundant compound in coca leaves (53), as EcMPOBMT clusters with known salicylic acid methyltransferases in the SABATH family (~60% AA identity).
The formation of the second ring in TA biosynthesis most likely proceeds via the formation of an imminium cation similar to NMPy (54). The concomitant oxidation reaction was shown to be catalyzed by a CYP82M3 enzyme (tropinone synthase) in A. belladonna (32). We have established that a similar reaction occurs in E. coca, although the cytochrome P450 enzyme belongs to the CYP81A family. CYP81A members from Z. mays are implicated in the modifications of zealexins (sesquiterpene acids) (55) and other roles for CYP81 members include herbicide resistance in cereals and grasses (56). To date, the only characterized CYP81A members were thought to be lineage-specific, belonging to the monocotyledonous grasses (57). However, our phylogenetic analysis reveals at least several other dicotyledonous families included in this clade. Nonetheless, it is clear that methylecgonone synthase and tropinone synthase have polyphyletic origins.
Consistent with a polyphyletic origin of TA biosynthesis in Erythroxylaceae and Solanaceae, previous work demonstrated that different enzyme classes were also independently recruited to catalyze the terminal esterification steps in these families. Solanaceae evolved a UDP-dependent glucosyltransferase (AbUGT1/UGT84A27) likely recruited from flavonoid metabolism for phenyllactate glucosylation and an acyltransferase recruited from the serine carboxypeptidase-like family (littorine synthase) for condensation of 1-O-β-phenyllactoylglucose with tropine to form littorine (19). In contrast, Erythroxylaceae evolved one or more CoA ligases likely recruited from monolignol metabolism (e.g., benzoyl/cinnamoyl/coumaroyl-CoA ligase) for activation of the acyl donor moiety (benzoate/cinnamate) and a BAHD acyltransferase (cocaine synthase) for condensation of the CoA thioester with methylecgonine to form cocaine (18). Further investigation may reveal novel insights into the structural evolution that occurred in these enzymes to enable their recruitment for TA biosynthesis.
Our work highlights the utility of yeast as a discovery platform for high-throughput gene candidate screening, efficient in vivo enzyme characterization, and rapid reconstruction of complex natural product pathways. In addition to advantages in throughput and speed over plant hosts, a yeast platform offers reduced metabolic “cross-talk”: because fungi are a more distant biological context for plant enzymes, there is a lower likelihood of native enzymes or metabolites interacting with and confounding measurements of heterologous enzyme activities, enabling greater confidence in the positive identification and classification of new genes relative to in conventional hosts like Arabidopsis or N. benthamiana. For example, the initial screening of E. coca enzymes involved in the first tropane ring closure in N. benthamiana may have been confounded by its native NMPy/nicotine pathway. Heterologous enzyme testing in yeast may also reveal alternate activity profiles for heterologous enzymes (as was observed for EcSPDS/SPMT/SMT), providing clues for the elucidation of their sequence-structure-function relationships. However, such discrepancies in activity across different hosts also suggest that candidate screening in yeast should be paired with in planta characterization (e.g., N. benthamiana or RNAi/VIGS) to validate the function of new enzymes within their native biochemical context. Genomic integration in yeast enables stable gene expression in contrast to transient expression in plants. This provides a potential advantage for reconstructing long plant pathways as blocks of genes that have been discovered, tested, and validated can be integrated, facilitating the screening and testing of the next batch. In contrast, evaluation of a long heterologous pathway in a plant host may require co-infiltration and transient co-expression of more than 10 genes, which can pose practical challenges for consistent pathway expression. With advancements in multiplexed Cas9-mediated genome modification tools, we expect that yeast platforms can complement traditional plant-based approaches for natural product pathway discovery.
Synthetic biology offers a unique opportunity to erase the phylogenetic divisions between biosynthetic pathways across disparate species. By co-expressing key Solanaceae and Erythroxylaceae TA enzymes in a single engineered yeast host, we coupled two analogous pathways that evolved independently over 120 My (1)—and which had never before coexisted in a single organism—resulting in the formation of new-to-nature TAs (Fig. 3C, D). Future work may reveal that such hybrid natural products share not only structural features of both lineages, but also pharmacological features, enabling access to an unexplored repertoire of medicinal and bioactive compounds using a genetic mix-and-match approach across independently evolved biosynthetic pathways. We expect that leveraging functional genomics, plant-specialized metabolism, and synthetic biology platforms can expand opportunities for the development of new chemical targets across drug discovery, nutrition, and agricultural applications.
Materials and Methods
Detailed materials and methods for all experiments described in this study are provided in SI Appendix, Materials and Methods.
Plant material.
N. benthamiana plants (4–6 weeks old) were grown in a greenhouse at 55% humidity, 22°C, and 16-h-light/8-h-dark photoperiod. E. coca v. coca plants were grown at 22°C under a 12-h-light/12-h-dark photoperiod, with humidity of 65% and 70%, respectively, and were fertilized weekly with Ferty 3 (15-10-15) and Wuxal Top N (Planta). A voucher specimen was deposited at the Herbarium Haussknecht (JE) at Friedrich Schiller University in Jena, Germany.
De Novo Transcriptome Assembly.
Mature leaf transcript profiling on Erythroxylum plants grown under greenhouse conditions (identical to those described above) at the United States Department of Agriculture in Beltsville Maryland was constructed by extracting mRNA from E. hondense, E. coca cv coca, and E. novogranatense cv novogranatense. Following library preparation, sequencing was performed using Illumina sequencing and CLC workbench for the assembly of contigs.
Hierarchical Clustering Analysis of Erythroxylum Transcriptomes.
Hierarchical clustering of methylecgonone synthase candidates from the Erythroxylum transcriptome datasets (generated as described in the previous section) was performed using a custom R script (coca_analysis_v1.R), which will be made available on GitHub at the time of publication; the full computational pipeline is described in the SI Appendix, Materials and Methods.
Phylogenetic Analyses and Dendrograms.
Amino acid sequences for all proteins selected for analysis were first aligned using CLUSTAL X2 with default alignment settings and the following clustering parameters: neighbor joining, n = 1000 bootstrap trials, random number generator seed = 111. The resulting Phylip format (.phb) tree was transferred to Fig Tree v1.4.4 software for processing.
Statistics.
For yeast experiments, biological replicates are defined as independent cultures inoculated from separate yeast colonies or streaks and cultivated in separate containers. For tobacco experiments, one biological replicate is defined as all infiltrated leaves from a single plant. For each transient expression attempt, four biological replicates were used for each gene stacking that was performed.
Data, Materials, and Software Availability
Data supporting the findings of this work are available within the paper and/or SI Appendix . Full sequences of E. coca genes newly identified in this work are hosted in GenBank with the following accession numbers: EcSPMT, OP382839; EcSMT, OP382840; EcSPDS, OP382841; EcAOF1, OP382842; EcAOC1, OP382843; EcAOC2, OP382844; EcOGAS1, OP382845; EcOGAS2, OP382846; EcMPOBMT, OP382847; EcCYP81AN15, OP382848. The custom R scripts used for hierarchical clustering analysis of Erythroxylaceae transcriptome datasets will be available at the time of publication on the Smolke Laboratory GitHub repository (https://github.com/smolkelab/coca_coexpression_analysis).
Acknowledgments
We thank the greenhouse staff of the Max Planck Institute for Chemical Ecology and the staff at the Leibniz Institute for Plant Genetics and Crop Plant Research for the rearing and care of the plants involved in this study. We thank Dr. Mohammed-Reza Hajirezaei (IPK) for assistance with LC-MS measurements; Dr. A. Cravens (Stanford University) for the yeast multiplex CRISPR–Cas9/sgRNA plasmids used for yeast genomic modifications; and the Stanford Cell Sciences Imaging Facility for access to microscopy equipment and training. We thank Dr. Birgit Dräger for the generous donation of synthesized dcSAM for in vitro enzyme assays. We thank Kyle Glockzin for his help with data collection and discussion. This work was supported by an Alexander von Humboldt postdoctoral fellowship to J.C.D., the NSF (NSF-1714236 grant to J.C.D.), the NIH (AT007886 grant to C.D.S.), the Chan Zuckerberg Biohub (C.D.S. is a Chan Zuckerberg Biohub investigator), the Siebel Scholars Foundation (doctoral fellowship to P.S.) and the Natural Sciences and Engineering Research Council of Canada (doctoral postgraduate scholarship to P.S.).
Author contributions
B.G.C., P.S., C.D.S., and J.C.D. designed research; B.G.C., P.S., K.G., N.K., O.M.E., J.J., G.R., C.D.S., and J.C.D. performed research; B.G.C., P.S., K.G., N.K., O.M.E., J.J., G.R., J.S., L.M., C.D.S., and J.C.D. analyzed data; and B.G.C., P.S., K.G., N.K., O.M.E., J.J., G.R., J.S., L.M., C.D.S., and J.C.D. wrote the paper.
Competing interests
The authors declare no competing interest.
Supporting Information
Appendix 01 (PDF)
- Download
- 9.92 MB
Dataset S01 (XLSX)
- Download
- 45.28 KB
References
1
J. Jirschitzka et al., Plant tropane alkaloid biosynthesis evolved independently in the Solanaceae and Erythroxylaceae. Proc. Natl. Acad. Sci. U.S.A. 109, 10304–10309 (2012).
2
M. Lounasmaa, T. Tamminen, “The tropane alkaloids” in The Alkaloids: Chemistry and Pharmacology, G. A. Cordell, Ed. (Academic Press, 1993), pp. 1–114.
3
P.-A. Nocquet, T. Opatz, Total synthesis of (±)-scopolamine: challenges of the tropane ring. Eur. J. Org. Chem. 2016, 1156–1164 (2016).
4
P. C. Meltzer, Y. Anna, Bertha Liang, K. Madras, The discovery of an unusually selective and novel cocaine analog: Difluoropine. Synthesis and inhibition of binding at cocaine recognition sites. J. Med. Chem. 37, 2001–2010 (1994).
5
N. Kim, O. Estrada, B. Chavez, C. Stewart, J. C. D’Auria, Tropane and granatane alkaloid biosynthesis: A systematic analysis. Molecules 21, (2016).
6
T. Schmeller, F. Sporer, M. Sauerwein, M. Wink, Binding of tropane alkaloids to nicotinic and muscarinic acetylcholine receptors. Pharmazie 50, 493–495 (1995).
7
T. Plowman, The Ethnobotany of Coca (Erythroxylum spp., Erythroxylaceae). Adv. Econ. Bot. 1, 62–111 (1984).
8
T. Plowman, L. Rivier, Cocaine and cinnamoylcocaine content of erythroxylum Species. Ann. Bot. 51, 641–659 (1983).
9
T. Plowman, Amazonian coca. J. Ethnopharmacol. 3, 195–225 (1981).
10
F. I. Carroll et al., Probes for the cocaine receptor. Potentially irreversible ligands for the dopamine transporter. J. Med. Chem. 35, 1813–1817 (1992).
11
S. Kelkar et al., Synthesis, Cocaine receptor affinity, and dopamine uptake inhibition of several new 2.beta.-substituted 3.beta.-phenyltropanes. J. Med. Chem., (September 6, 2022).
12
F. I. Carroll, A. H. Lewin, J. W. Boja, M. J. Kuhar, Cocaine receptor: Biochemical characterization and structure-activity relationships of cocaine analogs at the dopamine transporter. J. Med. Chem. 35, 969–981 (1992).
13
F. Bai et al., Overexpression of the AbSAUR1 gene enhanced biomass production and alkaloid yield in Atropa belladonna. Ind. Crops Prod. 140, 111705. (2019).
14
World Health Organization, “World Health Organization Model List of Essential Medicines: 21st List 2019” (World Health Organization, 2019).
15
T. Tian et al., Catalytic innovation underlies independent recruitment of polyketide synthases in cocaine and hyoscyamine biosynthesis. Nat. Commun. 13, 4994 (2022).
16
J.-P. Huang et al., Tropane alkaloids biosynthesis involves an unusual type III polyketide synthase and non-enzymatic condensation. Nat. Commun. 10, 4036 (2019).
17
J. C. D’Auria, Acyltransferases in plants: A good time to be BAHD. Curr. Opin. Plant Biol. 9, 331–340 (2006).
18
G. W. Schmidt et al., The last step in cocaine biosynthesis is catalyzed by a BAHD acyltransferase. Plant Physiol. 167, 89–101 (2015).
19
F. Qiu et al., Functional genomics analysis reveals two novel genes required for littorine biosynthesis. New Phytol. 225, 1906–1914 (2020).
20
G. Song, A. Walworth, Agrobacterium tumefaciens-mediated transformation of Atropa belladonna. Plant Cell Tissue Organ Cult. PCTOC 115, 107–113 (2013).
21
A. C. Rajewski, K. B. Elkins, A. Henry, J. Van Eck, A. Litt, In vitro plant regeneration and Agrobacterium tumefaciens–mediated transformation of Datura stramonium (Solanaceae). Appl. Plant Sci. 7, e01220. (2019).
22
L. Zeng et al., Development of Atropa belladonna L. plants with high-yield hyoscyamine and without its derivatives using the CRISPR/Cas9 system. Int. J. Mol. Sci. 22, 1731 (2021).
23
F. Hasebe et al., CRISPR/Cas9-mediated disruption of the PYRROLIDINE KETIDE SYNTHASE gene reduces the accumulation of tropane alkaloids in Atropa belladonna hairy roots. Biosci. Biotechnol. Biochem. 85, 2404–2409 (2021).
24
T. Zhao et al., Ornithine decarboxylase regulates putrescine-related metabolism and pollen development in Atropa belladonna. Plant Physiol. Biochem. PPB 192, 110–119 (2022).
25
S. Biastoff, N. Reinhardt, V. Reva, W. Brandt, B. Drager, Evolution of putrescine N-methyltransferase from spermidine synthase demanded alterations in substrate binding. FEBS Lett. 583, 3367–74 (2009).
26
A. Junker, J. Fischer, Y. Sichhart, W. Brandt, B. Dräger, Evolution of the key alkaloid enzyme putrescine N-methyltransferase from spermidine synthase. Front. Plant Sci. 4, 260 (2013).
27
T. Docimo, G. W. Schmidt, K. Luck, S. K. Delaney, J. C. D’Auria, Selection and validation of reference genes for quantitative gene expression studies in Erythroxylum coca (2013), https://doi.org/10.12688/f1000research.2-37.v1 (September 7, 2022).
28
T. Docimo et al., The first step in the biosynthesis of cocaine in Erythroxylum coca: The characterization of arginine and ornithine decarboxylases. Plant Mol. Biol. 78, 599–615 (2012).
29
P. Srinivasan, C. D. Smolke, Engineering a microbial biosynthesis platform for de novo production of tropane alkaloids. Nat. Commun. 10, 1–15 (2019).
30
P. Tavladoraki, A. Cona, R. Angelini, Copper-containing amine oxidases and FAD-dependent polyamine oxidases are key players in plant tissue differentiation and organ development. Front. Plant Sci. 7, (2016).
31
M. Naconsie, K. Kato, T. Shoji, T. Hashimoto, Molecular evolution of N-methylputrescine oxidase in tobacco. Plant Cell Physiol. 55, 436–444 (2014).
32
M. A. Bedewitz, A. D. Jones, J. C. D’Auria, C. S. Barry, Tropinone synthesis via an atypical polyketide synthase and P450-mediated cyclization. Nat. Commun. 9, 5281 (2018).
33
T. Zhao et al., Engineering tropane alkaloid production based on metabolic characterization of ornithine decarboxylase in Atropa belladonna. ACS Synth. Biol. 9, (2020).
34
J. C. D’Auria, F. Chen, E. Pichersky, “The SABATH family of MTs in Arabidopsis thaliana and other plant species” in Recent Advances in Phytochemistry, Integrative Phytochemistry: From Ethnobotany to Molecular Ecology, J. T. Romeo, Ed. (Elsevier, 2003), pp. 253–283.
35
P. Srinivasan, C. D. Smolke, Biosynthesis of medicinal tropane alkaloids in yeast. Nature 585, 614–619 (2020).
36
Y. Kawai, E. Ono, M. Mizutani, Evolution and diversity of the 2-oxoglutarate-dependent dioxygenase superfamily in plants. Plant J. Cell Mol. Biol. 78, 328–343 (2014).
37
S. Bieri, A. Brachet, J.-L. Veuthey, P. Christen, Cocaine distribution in wild Erythroxylum species. J. Ethnopharmacol. 103, 439–447 (2006).
38
K. L. Gorres, R. T. Raines, Prolyl 4-hydroxylase. Crit. Rev. Biochem. Mol. Biol. 45, 106–124 (2010).
39
S. J. Gagne et al., Identification of olivetolic acid cyclase from Cannabis sativa reveals a unique catalytic route to plant polyketides. Proc. Natl. Acad. Sci. U. S. A. 109, 12811–12816 (2012).
40
X. Yang et al., Structural basis for olivetolic acid formation by a polyketide cyclase from Cannabis sativa. FEBS J. 283, 1088–1106 (2016).
41
K. Luck et al., CYP79D enzymes contribute to jasmonic acid-induced formation of aldoximes and other nitrogenous volatiles in two Erythroxylum species. BMC Plant Biol. 16, 215 (2016).
42
E. Pichersky, E. Lewinsohn, Convergent evolution in plant specialized metabolism. Annu. Rev. Plant Biol. 62, 549–566 (2011).
43
R. Huang, A. J. O’Donnell, J. J. Barboline, T. J. Barkman, Convergent evolution of caffeine in plants by co-option of exapted ancestral enzymes. Proc. Natl. Acad. Sci. U.S.A. 113, 10613–10618 (2016).
44
D. Ober, E. Kaltenegger, Pyrrolizidine alkaloid biosynthesis, evolution of a pathway in plant secondary metabolism. Phytochemistry 70, 1687–1695 (2009).
45
E. Leete, Biosynthesis of the pyrrolidine rings of cocaine and cuscohygrine from [5-14C]-ornithine via a symmetrical intermediate. J. Am. Chem. Soc. 104, 1403–1408 (1982).
46
E. Leete, Spermidine: An indirect precursor of the pyrrolidine rings of nicotine and nornicotine in Nicotiana glutinosa. Phytochemistry 24, 957–960 (1985).
47
I. Fraudentali et al., The copper amine oxidase AtCuAOδ participates in abscisic acid-induced stomatal closure in Arabidopsis. Plants 8, 183 (2019).
48
N. Kim, “Tropane Alkaloid Biosynthesis in Erythroxylum coca Involves an Atypical Type III Polyketide Synthase” (Texas Tech University, Lubbock, TX, 2020).
49
K. C. Schmitt et al., Interaction of cocaine-, benztropine-, and GBR12909-like compounds with wild-type and mutant human dopamine transporters: Molecular features that differentially determine antagonist-binding properties. J. Neurochem. 107, 928–940 (2008).
50
J. D. Elsworth, J. R. Taylor, P. Jatlow, R. H. Roth, Serotonin involvement in cocaine sensitization: Clues from studies with cocaine analogs. Drug Dev. Res. 30, 189–200 (1993).
51
J. Lohman, L. Stunkard, A. Benjamin, T. Boram, Exploring enzymatic β-keto acid (de)carboxylation with malonyl-CoA analogs. FASEB J. 33, 633.17–633.17 (2019).
52
J. Jirschitzka, “The Final Steps of Cocaine Biosynthesis in the Erythroxylaceae Provide Insight into the Biochemistry, Physiology and Evolution of Tropane Alkaloid Biosynthesis” (Friedrich-Schiller-Universität Jena, Jena, Germany, 2014).
53
M. Novák, C. Salemink, The Essential Oil of Erythroxylum coca. Planta Med. 53, 113–113 (1987).
54
N. Kim, B. Chavez, C. Stewart, J. C. D’Auria, “Structure and function of enzymes involved in the biosynthesis of tropane alkaloids” in Tropane Alkaloids: Pathways, Potential and Biotechnological Applications, V. Srivastava, S. Mehrotra, S. Mishra, Eds. (Springer, 2021), pp. 21–50.
55
Y. Ding et al., Genetic elucidation of interconnected antibiotic pathways mediating maize innate immunity. Nat. Plants 6, 1375–1388 (2020).
56
C. C. Hansen, D. R. Nelson, B. L. Møller, D. Werck-Reichhart, Plant cytochrome P450 plasticity and evolution. Mol. Plant 14, 1244–1265 (2021).
57
S. Iwakami et al., Cytochrome P450 CYP81A12 and CYP81A21 are associated with resistance to two acetolactate synthase inhibitors in echinochloa phyllopogon. Plant Physiol. 165, 618–629 (2014).
Information & Authors
Information
Published in
Classifications
Copyright
Copyright © 2022 the Author(s). Published by PNAS. This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).
Data, Materials, and Software Availability
Data supporting the findings of this work are available within the paper and/or SI Appendix . Full sequences of E. coca genes newly identified in this work are hosted in GenBank with the following accession numbers: EcSPMT, OP382839; EcSMT, OP382840; EcSPDS, OP382841; EcAOF1, OP382842; EcAOC1, OP382843; EcAOC2, OP382844; EcOGAS1, OP382845; EcOGAS2, OP382846; EcMPOBMT, OP382847; EcCYP81AN15, OP382848. The custom R scripts used for hierarchical clustering analysis of Erythroxylaceae transcriptome datasets will be available at the time of publication on the Smolke Laboratory GitHub repository (https://github.com/smolkelab/coca_coexpression_analysis).
Submission history
Received: September 14, 2022
Accepted: October 25, 2022
Published online: November 28, 2022
Published in issue: December 6, 2022
Keywords
Acknowledgments
We thank the greenhouse staff of the Max Planck Institute for Chemical Ecology and the staff at the Leibniz Institute for Plant Genetics and Crop Plant Research for the rearing and care of the plants involved in this study. We thank Dr. Mohammed-Reza Hajirezaei (IPK) for assistance with LC-MS measurements; Dr. A. Cravens (Stanford University) for the yeast multiplex CRISPR–Cas9/sgRNA plasmids used for yeast genomic modifications; and the Stanford Cell Sciences Imaging Facility for access to microscopy equipment and training. We thank Dr. Birgit Dräger for the generous donation of synthesized dcSAM for in vitro enzyme assays. We thank Kyle Glockzin for his help with data collection and discussion. This work was supported by an Alexander von Humboldt postdoctoral fellowship to J.C.D., the NSF (NSF-1714236 grant to J.C.D.), the NIH (AT007886 grant to C.D.S.), the Chan Zuckerberg Biohub (C.D.S. is a Chan Zuckerberg Biohub investigator), the Siebel Scholars Foundation (doctoral fellowship to P.S.) and the Natural Sciences and Engineering Research Council of Canada (doctoral postgraduate scholarship to P.S.).
Author contributions
B.G.C., P.S., C.D.S., and J.C.D. designed research; B.G.C., P.S., K.G., N.K., O.M.E., J.J., G.R., C.D.S., and J.C.D. performed research; B.G.C., P.S., K.G., N.K., O.M.E., J.J., G.R., J.S., L.M., C.D.S., and J.C.D. analyzed data; and B.G.C., P.S., K.G., N.K., O.M.E., J.J., G.R., J.S., L.M., C.D.S., and J.C.D. wrote the paper.
Competing interests
The authors declare no competing interest.
Notes
This article is a PNAS Direct Submission.
Authors
Metrics & Citations
Metrics
Altmetrics
Citations
Cite this article
Elucidation of tropane alkaloid biosynthesis in Erythroxylum coca using a microbial pathway discovery platform, Proc. Natl. Acad. Sci. U.S.A.
119 (49) e2215372119,
https://doi.org/10.1073/pnas.2215372119
(2022).
Copied!
Copying failed.
Export the article citation data by selecting a format from the list below and clicking Export.
Cited by
Loading...
View Options
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDFLogin options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to access the full text.