Biosynthesis of strychnine

Strychnine, a complex monoterpene indole alkaloid isolated in 1818 from Strychnos nux-vomica, is a potent neurotoxin used as a rodenticide via high-affinity antagonism of the glycine receptor. Its structural elucidation and numerous total syntheses profoundly influenced organic chemistry. Despite this history, the enzymatic pathway by which plants assemble strychnine’s polycyclic scaffold remained unknown. Prior isotopic feeding studies suggested origins from tryptophan and geranyl pyrophosphate, passing through geissoschizine and an intermediate termed the Wieland–Gumlich aldehyde. This study aims to identify and functionally validate the enzymes that convert geissoschizine to strychnine, brucine, and diaboline, and to reconstitute these pathways heterologously in Nicotiana benthamiana.

Early biogenetic hypotheses (1948) and radioisotope feeding in S. nux-vomica established that strychnine derives from tryptophan and geranyl pyrophosphate via geissoschizine and a proposed Wieland–Gumlich aldehyde intermediate, with acetate incorporation forming the piperidone ring. The upstream conversion to geissoschizine is fully elucidated in Catharanthus roseus, which provided homologous genes (e.g., geissoschizine oxidase, CrGO). Additional prior work characterized α/β-hydrolases and cytochrome P450 roles in alkaloid diversification, and MDR enzymes in heteroyohimbine biosynthesis, informing candidate selection. Historical reports described a heat-labile ‘prestrychnine’ intermediate convertible to strychnine upon warming acid extracts, consistent with a non-enzymatic step. The literature thus provided biochemical logic and candidate enzyme families but lacked a complete, experimentally validated route to strychnine and derivatives.

• Species and omics: Generated tissue-specific RNA-seq from a strychnine producer (S. nux-vomica) and a non-producer (Strychnos sp. accumulating diaboline). Identified homologs to the known geissoschizine pathway and filtered candidates by high root expression (FPKM ≥ 20), co-expression with upstream genes, and plausible catalytic functions.
• Candidate discovery: BLAST with C. roseus geissoschizine oxidase (CrGO) identified SnvGO (CYP71AY6). Co-expression networks (Pearson r ≥ 0.95) nominated α/β-hydrolases, P450s, and MDRs for downstream steps. Parallel discovery in Strychnos sp. identified orthologs (SpGO, SpNS1/2, SpNO, SpWS) and BAHD acyltransferases (SpAT and S. nux-vomica ortholog SnvAT).
• Heterologous expression: Transient expression in N. benthamiana via Agrobacterium, with infiltration of synthetic standards (e.g., geissoschizine 1; later strychnine for tailoring steps). Products analyzed by LC–MS/LC–MS–MS; retention times confirmed with synthetic authentic standards.
• Stepwise pathway reconstitution and validation: • SnvGO converts geissoschizine 1 to akuammicine 3 (deformylation product of dehydropreakuammicine 2). • α/β-hydrolases SnvNS1/SnvNS2 divert dehydropreakuammicine to norfluorocurarine 4; observed decreased akuammicine when co-expressed. • P450 SnvNO (CYP71A144) hydroxylates 4 to 18-OH norfluorocurarine 5; MDR SnvWS (cluster 4032.5004) reduces the 2,16 double bond in 4 or 5 to yield Wieland–Gumlich aldehyde 6 after spontaneous cyclization. • Kinetic assays showed higher catalytic efficiency of SnvWS with 5 versus 4 (kcat/Km ≈ 0.297 vs 0.068 min−1 μM−1); docking suggested Thr95/Ser309 hydrogen bonds with 18-OH enhance catalysis. • In Strychnos sp., SpGO/SpNS1 or SpNS2/SpNO/SpWS reconstituted 6. A BAHD acyltransferase SpAT acetylated 6 to diaboline 8. • In S. nux-vomica, SnvAT preferentially malonylated 6 to N-malonyl WG aldehyde 9; co-expression with AAE13 (A. thaliana malonyl-CoA synthetase) and co-infiltration of disodium malonate boosted 9 formation. Chemical derivatization (TMS-diazomethane methylation, NaBH4 reduction) supported structure assignment. • Mutational analysis and homology modeling identified a single residue switch (SnvAT R424F / SpAT F421R) modulating malonyl- vs acetyl-CoA selectivity via a bidentate salt bridge to malonyl-CoA. • Conversion of 9 to strychnine 10 and isostrychnine 11 occurred slowly non-enzymatically in planta and in extracts; accelerated by heating leaves at 60 °C for 2 h. • Tailoring to brucine: Identified Snv10H (CYP82D367) as strychnine-10-hydroxylase forming 10-OH strychnine 12; SnvOMT methyltransferase yielded β-colubrine 13; Snv11H (CYP71AH44) hydroxylated 13 to 11-deMe brucine 14, which SnvOMT further methylated to brucine 15. SnvOMT could also methylate alternative hydroxylated intermediates with lower efficiency.
• In planta reconstitution: Expressed SnvGO, SnvNS1, SnvNO, SnvWS, SnvAT, AAE13, Snv10H, SnvOMT, Snv11H with geissoschizine and malonate, leading over time (up to 4 weeks) to accumulation of strychnine 10, isostrychnine 11, β-colubrine 13, and brucine 15; extracted-ion chromatograms confirmed products.
• In planta feeding: Hydroponic feeding of deuterium-labeled 6 to S. nux-vomica roots showed labeled 9 at 3 days; trace 10/11 at 7 days, consistent with slow, likely non-enzymatic conversion.
• Additional analyses: Phylogenetics of BAHDs; structural modeling of SnvWS and ATs; position-selectivity observations for 17-O-acylation in vitro versus N-acylation in planta.

• Identified and functionally validated nine enzymes converting geissoschizine 1 to diaboline 8, strychnine 10, and brucine 15. • Early steps: SnvGO (CYP71AY6), α/β-hydrolases SnvNS1/SnvNS2 (to norfluorocurarine 4), P450 SnvNO (CYP71A144) to 18-OH norfluorocurarine 5, and MDR SnvWS (cluster 4032.5004) reducing the 2,16 double bond to form Wieland–Gumlich aldehyde 6. • SnvWS shows higher catalytic efficiency with 5 than 4 (kcat/Km ≈ 0.297 vs 0.068 min−1 μM−1); modeling implicates Thr95/Ser309 in recognizing the C18-OH.
• Divergent BAHD acyltransferase activities determine pathway branching: • In Strychnos sp., SpAT acetylates 6 to diaboline 8. • In S. nux-vomica, SnvAT preferentially malonylates 6 to prestrychnine (N-malonyl WG aldehyde) 9; co-expression of AAE13 and malonate feeding boosts 9 production ~10-fold. • A single residue switch (Arg vs Phe: SnvAT R424 / SpAT F421) governs malonyl-CoA vs acetyl-CoA selectivity via salt-bridge interactions.
• Prestrychnine 9 non-enzymatically decarboxylates and cyclizes to strychnine 10 and isostrychnine 11 in vitro and in planta; heating at 60 °C for 2 h accelerates the conversion. Hydroponic feeding corroborates slow formation in S. nux-vomica (9 detected by 3 days; 10/11 by 7 days).
• Brucine tailoring pathway established: Snv10H (CYP82D367) hydroxylates strychnine to 10-OH strychnine 12; SnvOMT methylates to β-colubrine 13; Snv11H (CYP71AH44) hydroxylates 13 to 11-deMe brucine 14; SnvOMT completes methylation to brucine 15.
• Complete heterologous reconstitution in N. benthamiana from geissoschizine 1 yields strychnine 10, isostrychnine 11, β-colubrine 13, and brucine 15, demonstrating feasibility of metabolic engineering of strychnos alkaloids.

The study resolves the long-standing question of how plants assemble the strychnine scaffold by delineating each enzymatic step from geissoschizine to the Wieland–Gumlich aldehyde and revealing the branching reactions that yield diaboline or, via a malonylated prestrychnine, strychnine. The key insight that prestrychnine formation is enzymatic (SnvAT-mediated malonylation) while its conversion to strychnine proceeds slowly and largely non-enzymatically reconciles historical observations of heat-induced conversion. The discovery that closely related BAHD acyltransferases in different Strychnos species have diverged (malonyl- versus acetyl-transfer) explains species-specific alkaloid profiles and provides a molecular handle (a single residue determinant) to reprogram acyl-donor specificity. Identification of late tailoring enzymes to brucine establishes the complete route to a major natural derivative and shows methylation/hydroxylation order. Heterologous pathway reconstitution in N. benthamiana demonstrates practical access to these complex alkaloids and intermediates, enabling synthetic biology approaches for diversification and potential drug discovery.

This work elucidates the biosynthetic pathways to strychnine, diaboline, and brucine, identifies nine key enzymes, and reveals that the final conversion of prestrychnine to strychnine can occur non-enzymatically. It explains interspecies metabolic divergence via BAHD acyltransferase specificity and provides genetic parts to reconstitute and engineer strychnos alkaloid biosynthesis in a heterologous plant host. Future work should optimize flux and compartmentalization in hosts to increase yields, explore enzyme engineering (e.g., BAHDs and MDR) to modulate selectivity and efficiency, search for potential physiological catalysts of the prestrychnine decarboxylation, and leverage the pathway to generate novel analogues for biological testing.

• The conversion of prestrychnine 9 to strychnine 10 and isostrychnine 11 appears slow and largely non-enzymatic; no enzyme was found that markedly accelerates this step, leaving uncertainty about any modest enzymatic contribution in planta.
• Heterologous production in N. benthamiana required prolonged incubation and/or heating to accumulate final products, indicating suboptimal flux and potential compartmentalization or cofactor limitations.
• In vitro acylation showed predominant 17-O-acylation at physiological pH, differing from N-acylation observed in planta, highlighting context-dependent selectivity.
• One intermediate of the brucine pathway (11-deMe brucine 14) was not detected in S. nux-vomica roots, possibly due to rapid turnover below detection limits.
• Reported kinetic parameters are limited to SnvWS; comprehensive enzyme kinetics and structural validations for all enzymes remain to be completed.