FIELD OF THE INVENTION

The invention relates to methods and products for producing terpene indole alkaloid derivatives.

BACKGROUND OF THE INVENTION

Vinblastine, a potent anticancer drug, is produced by Madagascar periwinkle (Catharanthus roseus ) in small quantities. Heterologous reconstitution of vinblastine biosynthesis could greatly improve access to this drug. The complex chemistry underlying the extensive transformations observed in vinblastine synthesis makes identification of the biosynthetic genes challenging.

With reference to Scheme 1 (below), the biosynthesis of vincristine 5 and vinblastine 6, plant-derived anticancer drugs, has been studied for decades, but the genes that comprise this metabolic pathway have remained elusive. Specifically, it is unknown how the biosynthetic intermediate stemmadenine 1 is transformed into tabersonine 2 and catharanthine 3, the two scaffolds that are ultimately dimerized to yield vinblastine and vincristine (Scheme 1A). These chemical transformations are poorly understood, making it exceptionally challenging to identify the corresponding biosynthetic enzymes.

WO 2017/152273 and the related paper Qu et at. (2018,“Solution of the multistep pathway for assembly of corynanthean, strychnos, iboga, and aspldosperma monoterpenoid indole alkaloids from 19E-geissoschizine”, PNAS pii: 201719979) describe the enzymes which we refer to herein as Tabersonine Synthase (“TS”) and Catharanthine Synthase (“CS”) involved in the terpenoid indole alkaloid biochemical pathway. However, the substrates of the CS and TS enzymes were not identified In those reports and the catalytic functions of CS and TS remain cryptic.

Scheme 1. Vincristine and vinblastine biosynthesis. A. Vincristine 5 and vinblastine b are formed by dimerization from the monomers catharanthine 3 and vindo!ine 4. Gatbaranthine 3 and vindoiine 4 can be dimerized by a peroxidase or using chemical methods. The biosynthetic genes that convert tabersonine 2 to vindoiine 4 have been identified, but the biosynthesis of tabersonine 2 and catharanthine 3 from stemmadenine 1 is ascertained in this study. B. Representative additional bioactive Aspidosperma and iboga alkaloids.

Qureshi & Scott (1968; Chemical Communications 16: 948-950) suggests a pathway for alkaloid biosynthesis in V. rosea seeds. Stavrinides et a!. (2016; Nature Communications 7: 12116) describes 3 heteroyohimbine synthases isolated from C. roseus.

SUMMARY OF THE INVENTION

In a first aspect the invention relates to a method for producing a terpene indole alkaloid derivative, comprising the steps of:

(1 ) providing a terpene indole alkaloid;

(2) providing: (a) a first enzyme (also referred to herein as“Precondylocarplne Acetate Synthase”, or Its acronym“PAS”) having a first amino acid sequence comprising SEQ ID NO: 1 or a functional variant or homologue thereof; and/or

(b) a second enzyme (also referred to herein as“Dehydroprecondyiocarpine Acetate Synthase”, or its acronym“DPAS”) having a second amino acid sequence comprising SEQ ID NO: 2 or SEQ ID NO: 66, or a functional variant or homologue thereof;

(3) optionally also providing:

(c) a third enzyme (also referred to herein as“CS”) having an amino acid sequence comprising SEQ ID NO: 3 or a functional variant or homologue thereof; and/or

(d) a fourth enzyme (also referred to herein as“TS”) comprising an amino acid sequence comprising SEQ ID NO: 4 or a functional variant or homologue thereof; and

(4) contacting the terpene indole alkaloid with the first and/or second enzyme, and optionally also the third and/or fourth enzyme, under catalytic conditions to produce a terpene indole alkaloid derivative.

Guresbi & Scott (1968; supra ) suggests a pathway for alkaloid biosynthesis in i/ rosea seeds and Stavrinides et al. (2016; supra } describes 3 heteroyohimbine synthases isolated from C

roseus. However, the prior art does not describe the actual enzymes involved in the processes of producing a terpene indole alkaloid derivative. The invention Is based on the identification of missing steps In vinblastine biosynthesis, namely the Identification of an oxidase and reductase that isomerize stemmadenine acetate into dihydroprecondyiocarpine acetate, which is then

deacetoxyiated and cyciized to either catharanthine or tabersonine via two distinct hydrolases.

These enzymes provide insight into how plants create extensive chemical diversity and enable development of heterologous platforms for generation of many stemmadenine-derived bioactive compounds.

The inventors have identified two redox enzymes that convert stemmadenine acetate 7 into an unstable molecule, which although not !so!aiabie, is most likely dihydroprecondyiocarpine acetate 11 (see Scheme 2). Dihydroprecondyiocarpine acetate 11 acts as the substrate for two distinct hydrolases reported here that deacetoxylate this substrate and then catalyze one of two possible formal Die!s-A!der cyclizatlons to either tabersonine 2 or catharanthine 3. Notably, the catharanthine 3 scaffold can form spontaneously from substrate 11 under certain conditions, while formation of tabersonine 2 appears to be strictly dependent on enzymatic catalysis. Particularly, the biochemistry required for tabersonine 2 and catharanthine 3 formation from the known alkaloid intermediate stemmadenine 1 is disclosed.

The biochemistry required for tabersonine 2 and catharanthine 3 formation from the known alkaloid intermediate stemmadenine 1 is disclosed. The identification of these enzymes provides insight into the mechanisms that plants use to create extensive chemical diversity. Moreover, in addition to serving as the precursors for vincristine 5 and vinblastine b, tabersonine 2 and catharanthine 3 are also precursors for dozens of other biologically active alkaloids (Scheme 1 B). Therefore, the discovery of these enzymes and their function enables heterologous production or vincristine 5, vinblastine 6, and many other rare, high value alkaloids (Scheme 1 B).

Scheme 2. Biosynthesis of eatharanthine and tabersonine scaffolds. A. Initial hypothesis. Stemmadenine 1 undergoes a push-pull isomerization to generate a double bond at C20-21 (/so-stemmadenine 8). Collapse of the enamine and dehydration yields dehydrosecodine 9, which can then cyciize, possibly via a Die!s-A!der reaction, to form either eatharanthine 3 or tabersonine 2. B. Revised hypothesis. Stemmadenine acetate 7 (generated from

stemmadenine 1 using conditions i. (AC2O (excess) pyridine (excess) r.t., 4 h, >99%) undergoes an oxidation to form precondylocarpine acetate 10. This is catalyzed

enzymatically by the reticuline oxidase homologue PAS, or alternatively can be generated synthetically using conditions as reported by Scott and co-workers (Scott & Wei, 1972, J.

Am. Chem Soc. 94: 8264-8265) ii. (Pt (from 7.5 eq. PtC>2), EtOAc, O2 atm., r.t., 10 h, yields varied). Next, precondylocarpine acetate 10 is reduced by the alcohol dehydrogenase DPAS. This reduced intermediate could not be isolated due to its lability, but it is assumed to be dihydroprecondy!ocarpine acetate 11 based on the degradation product tubotaiwine 12. Dihydroprecondyiocarpine acetate 11 , in the open form, can form dehydrosecodine 9 either through the action of GS, TS or spontaneously. GS controls the cyclization to form eatharanthine 3, while TS controls the cyclization to form tabersonine 2. Gatharathine 3 can also form spontaneously.

BRIEF DESCRIPTION OF THE FIGURES

Fig. 1. Co-expression analysis for biosynthetic gene candidates.

Co-expression analysis for biosynthetic gene candidates. A. Expression profile of some of the genes involved in C. roseus MIA pathway and those investigated in this study. Data 'were extracted from the http://medicinalplantaenornics.msu.edu/index.shtml database. B. Snapshot of the RNA seq dataset showing the presence of several genes annotated as acetyl-transferases. C. Co-expression profile of TS and PAS. The FPKM values for PAS were multiplied by a factor 10 in order to be comparable to those or TS. TDC=Tryptophan decarboxylase; 8HGO=Geranio! 8- hydroxylase;

CrlSY=lridoid synthase; SLS=Seco!oganin synthase; STR=Strlctosldine synthase;

SGD=Strictosidine-0-beta-D-glueosidase; PAS=Precondyiocarpine acetate synthase;

DPAS=Dehydroprecondylocarpine acetate synthase; TS=Tabersonine synthase. TS was initially annotated in the transciptome dataset as 2-hydroxyisoflavanone dehydratase. Fig. 2. Virus-induced gene silencing of TS in C. roseus using a unique region of the gene. Virus-induced gene silencing of TS in C. roseus using a unique region of the gene. A. Fragment (grey box) of TS cDNA used to assemble the pTRV2 construct. The black box represents the coding region, whereas the black lines are the flanking untranslated regions. Arrows show the annealing sites of the primers used for qRT-PCR analysis (Table 1 ). B. Fold transcript change in TS silenced (pTRV2-TS) plants compared to TS control (pTRV2) plants. Values were calculated using 2 ^_DDa . Upper panel calculated using the EXP reference gene. Lower pane! calculated using the N2227 reference gene. C. Box plots of ACt values of 8 biological replicates for control (pTRV2) and CrTS silenced (pTRV2-TS) plants with median, min and max values indicated. Asterisks represent significant differences determined using an unpaired, two-tailed t test ( ^**** = p<0.Q001 ). Upper panel calculated using the EXP reference gene. Lower panel calculated using the N2227 reference gene. D, UPLC-MS analysis of TS silenced leaves showed a significant decrease of tabersonine and vindoline E. Data shown corresponds to average measurements of 12 plants. Error bars indicate standard error of the mean. Statistical significance calculated with Student’s t test (pTRV2 in comparison to pTRV2-TS) is indicated as followed: ^* = p<0 05 and ^** = p<0.005.

Fig. 3. Virus-induced gene silencing of CS in C. roseus using a unique region of the gene. Virus-induced gene silencing of CS in C. roseus using a unique region of the gene. A. Fragment (grey box) of CS eDNA used to assemble the pTRV2 construct. The black box represents the coding region, whereas the black lines are the flanking untranslated regions. Arrows show the annealing sites of the primers used for qRT-PCR analysis (see Table 1 }. B. Fold transcript change in CS silenced (pTRV2-CS) plants compared to CS control (pTRV2) plants. Values were calculated using 2 ^DD jjpp _er panel calculated using the EXP reference gene. Lower panel calculated using the N2227 reference gene. C, Box plots of AGt values of 8 biological replicates for control (pTRV2) and CS silenced (pTRV2-CS) plants with median, min and max values indicated. Asterisks represent significant differences determined using an unpaired, two-tailed t-test ( ^* = p<0.05). Upper panel calculated using the EXP reference gene. Lower panel calculated using the N2227 reference gene. D, UPLC-MS analysis of CS silenced leaves showed a significant decrease of catharanthine and increase of vindoline (E). Data shown corresponds to average measurements of 12 plants. Error bars indicate standard error of the mean. Statistical significance calculated with Student’s t test (pTRV2 In comparison to pTRV2-CS) is indicated as followed: ^* = p<0 05. Fig. 4. SDS-PAGE and proteomies anaiysis of the proteins used in in vitro activity assays. SDS-PAGE and proteomies were used to analyze the proteins purified for in vitro activity assays. A. SDS-PAGE of DPAS, CS and TS expressed in E. coii and purified by His-trap and gei-filtration. Lane 1 : protein molecular markers; lane 2: DPAS; lane 3: CS; lane 4: TS. B. Snapshot or the proteomies results (complete anaiysis can be found in dataset jlc002601 ) of NiNTA-purified PAS, DPAS and CS or PAS, DPAS and TS (pathway reconstitution in N. benthamiana) expressed in N. benthamiana leaves. C. SDS-PAGE of PAS expressed In P. pastohs. Lanes indicated with 5 were loaded with PAS enriched medium; lane 6: protein molecular markers. The box indicates the section of the gel that was excised and used for proteomies anaiysis. D. Snapshot of the proteomies results (complete anaiysis can be found in dataset jic002561 ) showing that PAS 'was indeed present in the sample and it 'was amongst the most abundant proteins.

Fig. 5. NMR spectra for stemmadenine 1.

NMR spectra for stemmadenine 1 (MeOD, 300 K, 400 MHz) A. Skeletal formula of stemmadenine. B, Proton: 64 scans; C. COSY: 32 scans, D. NOESY: 32 scans; E. Carbon; F. HSQC: 32 scans; G, HMBC: 32 scans.

Fig. 6. NMR spectra for stemmadenine acetate 7.

NMR spectra for stemmadenine acetate 7 (MeOD, 300 K, 400 MHz) A. Skeletal formula of stemmadenine acetate B. Proton: 16 scans; G. COSY: 4 scans; D. TOCSY: 4 scans; E. NOESY: 4 scans; F. Carbon: 4096 scans; G. HSQC: 4 scans; H, HMBC: 8 scans.

Fig. 7. Activity guided fractionation of CS/TS substrate from Tabernaemontana divaricate ieaves.

Activity guided fractionation of CS/TS substrate from Tabernaemontana divaricata leaves. Fractions collected during preparative HPLC were assayed for the presence of the substrate using CS and IS. A. A fraction reacted with CS showed formation of catharanthine 3 after UPLG/QqQ-MS analysis. Catharanth!ne 3 was not formed In the control samples (no enzyme). B. The same fraction reacted with TS showed formation of tabersonine 2 after UPLC/QqG-MS analysis. Tabersonine 2 was not formed in the control samples (no enzyme). Very small peaks of endogenous catharanthine and tabersonine co-purifying with the CS/TS substrate were present in the control samples. Fig. 8. N R data of decomposed substrate isolated from plants, tubotaiwine 12.

NMR data of decomposed CS/TS substrate isolated from plants, tubotaiwine 12 (CDCI3, 300 K, 400 MHz); see also Table 7. A. Skeletal formula 0? tubotaiwine. B. Proton: 512 scans; G. COSY: 18 scans; D. NOESY: 24 scans; E. HSGC: 32 scans; F. HMBC: 500 scans.

Fig. 9. Virus-induced gene silencing of PAS in C. roseus using a unique region of the gene. Virus-induced gene silencing of PAS in C. roseus using a unique region of the gene. A. Fragment (grey box) of PAS cDNA used to assemble the pTRV2 construct. The black box represents the coding region, whereas the black lines are the flanking untranslated regions. Arrows show the annealing sites of the primers used for qRT-PGR analysis (Table 1 ). B. Fold transcript change in PAS silenced (pTRV2-PAS) plants compared to PAS control (pTRV2) plants. Values were calculated using 2 ^DDa . Upper panel calculated using the EXP reference gene. Lower panel calculated using the N2227 reference gene. C. Box plots of AGt values of 8 biological replicates for control (pTRV2) and PAS silenced (pTRV2-PAS) plants with median, min and max values indicated. Asterisks represent significant differences determined using an unpaired, two-tailed t-test ( ^** = p<0.01 ). Upper panel calculated using the EXP reference gene. Lower panel calculated using the N2227 reference gene. D. UPLC-MS analysis of PAS silenced leaves showed accumulation of siemmadenlne acetate 7 (for identification see Fig. 10 and 1 1 ). Data shown corresponds to average measurements of 12 plants. Error bars indicate standard error of the mean. Statistical significance calculated with Student’s t test (pTRV2 in comparison to pTRV2-PAS) is indicated as ^**** = p<0 0001 . E. UPLG/MS analysis of VIGS extracts showed the appearance of a new peak assigned as stemmadenine acetate 7 (see Fig. 10 and 11 ) in the pTRV2-PAS silenced plants that was not present in the pTRV2 empty vector controls.

Flu. 10. UPLC/MS analysis of VIGS PAS extracts.

UPLG/MS analysis of ViGS plant extracts. The analysis showed that the new peak at m/z 397.19 and RT 3.81 in the pTRV2-PAS silenced plants (A) co-eluted with a semi-synthetic standard or stemmadenine acetate 7 (B). The insets show a comparison between the MS and MS/MS spectra of the two chemical species. Fig. 11. ¹ H N R comparison between synthetic ste madenine acetate 7 and that obtained from extraction of PAS-siienced plants.

¹ H NMR comparison between synthetic stemmadenine acetate 7 and that obtained by partial purification from PAS-siienced plants.

Fig. 12. Virus-induced gene silencing of DPAS in C. roseus using a unique region of the gene. Virus-induced gene silencing of DPAS in C. roseus using a unique region of the gene. A. Fragment (grey box) of DPAS cDNA used to assemble the pTRV2 construct. The black box represents the coding region, whereas the black lines are the flanking untranslated regions. Arrows show the annealing sites of the primers used for qRT-PGR analysis (Table 1 ). B. Fold transcript change DPAS silenced (pTRV2-DPAS) plants compared to DPAS control (pTRV2) plants. Values were calculated using 2 ^DDa . Upper panel calculated using the EXP reference gene. Lower panel calculated using the N2227 reference gene. C. Box plots of AGt values of 8 biological replicates for control (pTRV2) and DPAS silenced (pTRV2-DPAS) plants with median, min and max values Indicated. Asterisks represent significant differences determined using an unpaired, two-tailed t-test ( ^*** = p<0.001 , ^**** = p<0.QQQ1 ). Upper panel calculated using the EXP reference gene. Lower panel calculated using the N2227 reference gene. D, UPLC-MS analysis of DPAS silenced leaves showed accumulation of precondy!ocarpine acetate 10 (identification see Figs 13 and 14). Data shown corresponds to average measurements of 12 plants. Error bars indicate standard error of the mean. Statistical significance calculated with Student’s t test (pTRV2 in comparison to pTRV2-DPAS) is indicated as ^** = p<0 01. E. UPLG/MS analysis of VIGS extracts showed a marked increase In a peak with m/z 395.19 In pTRV2-DPAS silenced plants that was not very abundant in the pTRV2 empty vector controls (see Fig. 13 and 14).

Fig. 13. UPLC/ S analysis of VIGS DPAS extracts.

UPLG/MS analysis of ViGS DPAS extracts. The analysis showed that the new peak at m/z 395.19 and RT 3.27 In the pTRV2-DPAS silenced plants (A) co-eluted with a semi-synthetic standard or precondyiocarpine acetate 10 (B). The insets show a comparison between the MS and MS/MS spectra of the two chemical species. Fig. 14. HMR data of synthetic precondylocarpine acetate 10.

NMR data of synthetic precondylocarpine acetate 10 (CD ₃ CN, 300 K, 400 MHz); see also Table 8. A. Skeletal formula of precondylocarpine acetate. B. Proton: 256 scans; C. COSY: 32 scans; D. NOESY: 16 scene; E . TOOSYi 24 scans; F HSGOi 72 scans, 0 HMBOi 425 scans (during HMBO acquisition degradation occurred, so not aii cross peaks in HMBC match 1 H NMR sprectrum).

Fig. 15. Pathway reconstitution in vitro using PAS expressed in N. benthamiana leaves.

Pathway reconstitution in vitro using PAS expressed in A/ benthamiana ieaves. The figure shows the extracted ion chromatograms for ions m/z 397.19 (stemmadenine acetate, the starting material), m/z 395.19 (precondylocarpine acetate) and /T?/Z 337.1 9 (catharanthine at RT 4.0 and tabersonine at RT 4.4). Heterologously expressed and purified proteins were used to reconstitute the biosynthetic pathway from stemmadenine acetate 7 to catharanthine 3 and tabersonine 2. Extracted ion chromatograms for each compound are shown. When no enzymes 'were present, very small amounts of precondylocarpine acetate 10 were observed in the reaction, likely due to spontaneous oxidation. After addition of PAS, most of the substrate was converted to precondylocarpine acetate 10. Addition of DPAS resulted in complete consumption of precondylocarpine acetate 10 and catharanthine 3 started to appear in the samples. However, when CS was present, ail the initial substrate (stemmadenine acetate 7) was converted to catharanthine 3. When PAS, DPAS and IS were combined together in the reaction, all initial substrate was converted to tabersonine 2.

Formation of catharanthine 3 and tabersonine 2 was validated by co-eiution with commercial standards. Formation of precondylocarpine acetate 10 was validated by co-eiution with the semi synthetic compound.

Fig. 16. Pathway reconstitution in vitro using PAS expressed in P, pastoris.

Pathway reconstitution in vitro using PAS expressed in P. pastoris. A. Total ion chromatograms for MRMs of catharanthine 3 (RT=10.90) for the reaction of PAS, DPAS and CS with stemmadenine acetate 7, compared to a commercial standard of catharanthine 3. B, Total ion chromatograms for MRMs of tabersonine 2 (RT=13.80) for the reaction of PAS, DPAS and TS with stemmadenine acetate 7, compared to a commercial standard of tabersonine 2. Fig. 17. in vitro pathway reconstitution from synthetic precondyiocarpine acetate 10, in vitro pathway reconstitution from synthetic precondyiocarpine acetate 10. The figure shows the extracted ion chromatograms for ions m/z 397.19 (stemmadenine acetate), m/z 395.19

(precondyiocarpine acetate 10) and m/z 337 19 (catharanthine 3 at RT 4.0 and tabersonine 2 at RT 4.4). Addition of DPAS to the reaction mixture resulted in complete consumption of

precondyiocarpine acetate and appearance of a small amount of catharanthine 3 but no reduced product (dihydroprecondyiocarpine acetate 11 ) at m/z 397.19 was observed. When DPAS and CS were present, all substrate was converted into catharanthine 3. When DPAS and TS were incubated with precondyiocarpine acetate 10, tabersonine 2 was formed.

Fig. 18. in vitro reaction of crude dihydroprecondyiocarpine acetate with CS and TS.

In vitro reaction of crude dihydroprecondyiocarpine acetate 11 with CS and TS. A. Total ion chromatograms for MRMs of catbaranthine 3 (RT=10.9G) for the reaction of CS with

dihydroprecondyiocarpine acetate 11 , compared to a control reaction without CS. B. Total ion chromatograms for MRMs of tabersonine 2 (RT=13.82) for the reaction of TS with

dihydroprecondyiocarpine acetate 11 , compared to a control reaction without TS.

Fig. 19. in vitro activity of PAS purified from P, pastoris culture medium,

in vitro activity or PAS purified from P. pastoris culture medium. The figure shows the extracted ion chromatograms for ions m/z 397.19 (stemmadenine acetate 7) and m/z395.19 (precondyiocarpine acetate 10). Addition of PAS to the reaction mixture resulted in complete conversion of

stemmadenine acetate 7 to precondyiocarpine acetate 10 However, heat inactivated enzyme (20 min in boiling 'water) was still able to consume some of the substrate, suggesting that this enzyme is quite resilient to heat inactivation.

Fig. 20. Assay of PAS wifh stemmadenine 1.

Assay of PAS with stemmadenine 1. UPLC/MS analysis of reactions in which stemmadenine 1 was used as substrate for PAS showed neither consumption of substrate or formation of n ew products. Extracted ion chromatograms for m/z 355.2 (stemmadenine) and m/z 353.2 (mass of the expected oxidation product) are shown. The peak at m/z 353.2 present in both samples was not a product of PAS activity. Extracted ion chromatogram for m/z 323.17 showed that no condyiocarpine 13 was formed during the reaction.

Fig. 21. S S spectra of condyiocarpine 13 and tubofaiwine 12,

MSMS spectra of condyiocarpine 13 and tubotaiwine 12. A, MSMS spectrum of condyiocarpine 13 (precursor ion m/z 323.17) at high energy B, MSMS spectrum of tubotaiwine 12 (precursor ion m/z 325.19) at high energy.

Localization of PAS.

PAS is targeted to endoplasmic reticulum and progressively secreted to vacuole by ER-derived vesicles. C roseus cells were transiently co-transformed with plasmids expressing the 40-first (sp40; A) or 60-first (sp6G; E, I, K, M) PAS residues fused to YFP and the endoplasmic (ER)-CFP marker (B, F) or vacuole (vac-) CFP marker (N). Localization was investigated during 96 h and

representative photos were taken at 24, 48, 72 and 96 h post-transformation to highlight permanent ER localization or sp40-YFP during 96 h (A) and the progressive translocation of sp80-YFP from ER at 24 h(E), vesicles at 48 h (I), to vacuole at 72 h and 96 h (K, M) Co-localization of the fluorescence signals appears when merging the two individual false colour images (C, G, O). Cell morphology Is observed with differential interference contrast (DIG) (D, H, J, L, P). Scale bars, 10 pm.

Fig. 23. Localization of OPAS,

DPAS displays a cytosolic localization and homodimerizes. C. roseus cells were transiently co transformed with plasmids expressing DPAS-YFP (A) or YFP-DPAS (E) and the plasmid encoding the nuclear (nuc)-CFP marker (B) or the nucieocytosolic marker CFP (F). DPAS dimerization was analyzed by b!mo!ecuiar fluorescence complementation (BiFC) assays through transient co expression of DPAS-YFPN and YFPC-DPAS (I) or YFPN-DPAS and YFPC-DPAS (K, M) with the nue-CFP marker (N). Co-localization of the fluorescence signals appears when merging the two individual false colour images (C, G, O). Ceil morphology is observed with differential interference contrast (DIG) (D, H, J, L, P). Scale bars, 10 pm. Fig. 24. Localization of CS and TS.

CS and TS both display nucleocytosolic localization. C. roseus cells were transiently co-transformed with plasmids expressing either CS-YFP (A), YFP-CS (E), TS-YFP (I) or YFP-TS (M) and the nuclear (nuc)-GFP marker (B, J) or the cytosolic CFP marker (F, N) Co-localization of the fluorescence signals appears when merging the two individual false colour images (C, G, K, O). Cell morphology is observed with differential interference contrast (DIG) (D, H, L, P). Scale bars, 10 pm.

Fig. 25. Interaction of DPAS with TS and CS.

Interaction of DPAS with TS and CS DPAS/CS and DPAS/TS Interactions were analyzed by BIFC in C. roseus cells transiently transformed by distinct combinations of plasmids encoding fusions with the two split YFP fragments, YFPN (YN) and YFPC (YC) as Indicated on each pictures of the first row (A, D, G, J, M). Identification of transiently transformed cells was achieved by co-transformation with the nucleus (nuc)-CFP markerfB, E, H, K, N). Efficient (TS-YN/YC-DPAS) and weak (GS- YN/YC-DPAS) reconstitutions of BiFC complexes revealed by fluorescence intensity reflect corresponding Interaction levels. Interactions of DPAS and TS with loganic acid methy!transferase (LAMT)-YN and LAMT-YC were also studied to evaluate the specificity of DPAS/TS interactions. Cell morphology Is observed with differential interference contrast (DIG) (C, F, I, L, O). Scale bars, 10 pm.

Amino acid sequence aiignment of PAS with other funetionaiiy characterized enzymes

Amino acid sequence alignment (over parts A [aiignment residues 1 -303] and B [ alignment residues 304-571] of PAS with other functionally characterised berberine bridge enzymes and PAS-iike proteins identified from other aspidosperma and iboga alkaloids producing plants. Boxes highlight the residues (His and Cys) Involved in b!cova!ent attachment of the FAD in berberine bridge enzymes involved in benzylisoquinoline alkaloids that are mutated in PAS and other PAS-like enzymes identified in aspidosperma and Iboga alkaloids producing plants Aiignment was performed using MUSCLE algorithm characterized berberine bridge enzymes and PAS-like proteins Identified from other aspidosperma and iboga alkaloids producing plants. 1. SEQ ID NO: 58; 2. SEQ ID NO: 59; 3. SEQ ID NO: 60; 4. SEQ ID NO: 61 ; 5. SEQ ID NO: 62; 6. SEQ ID NO: 63; 7. SEQ ID NO: 64; 8 SEQ ID NO: 1 ; 9. SEQ ID NO: 65. Amino acid sequence aiignment of CS and TS.

Amino acid sequence aiignment of CS (SEQ ID NO: 3) and TS (SEQ ID NO: 4). Alignment was performed using MUSCLE algorithm.

Fig. 28. Biosynthesis of tabersonine 2 and catharanthine 3 from ste madenine acetate 7 starting substrate.

A. Reconstitution in N. benthamiana from stemmadenine acetate 7. Extracted ion chromatograms for ions m/z 397 19 (stemmadenine acetate 7), m/z 395.19 (precondyiocarpine acetate 10) and m/z 337.19 (catharanthine at RT 4.0 and tabersonine at RT 4.4 min) are shown. Plants infiltrated with the empty vector (EV) and PAS were able to convert 7 into 10 but no 2 or 3 were detected. Co expression of PAS, DPAS and CS resulted in formation of 3. When PAS, DPAS and TS 'were combined together in the reaction, all Initial substrate was converted to 2. Formation of 3 and 2 was validated by co-e!ution with commercial standards. Formation of 10 was validated by co-elution with the semi-synthetic compound. B. Interaction of CS and TS with DPAS by biomoiecuiar fluorescence complementation (BiFC) in C. roseus ceils. Efficiency of BiFC complex reformation reflected by YFP fluorescence intensity highlighted that CS and DPAS performed weak interactions (i-iii) while TS and DPAS strongly interacted (Iv-vi). No interactions with loganic acid methy!transferase (LAMT) were observed (vii-ix) C. Phylogenetic relationship of PAS with other functionally characterized berberine bridge enzymes PAS and PAS-like homologues identified in other Aspidosperma and iboga alkaloids producing plants form a separate cluster.

DETAILED DESCRIPTION OF THE INVENTION

In a first aspect the invention relates to a method for producing a terpene indole alkaloid derivative, comprising the steps of:

(1 ) providing a terpene indole alkaloid;

(2) providing:

(a) a first enzyme having a first amino acid sequence comprising SEQ ID NO: 1 or a functional variant or homologue thereof; and/or

(b) a second enzyme having a second amino acid sequence comprising SEQ ID NO: 2 or SEQ ID NO: 66, or a functional variant or homologue thereof;

(3) optionally also providing:

(c) a third enzyme having an amino acid sequence comprising SEQ ID NO: 3 or a functional variant or homologue thereof; and/or

(d) a fourth enzyme comprising an amino acid sequence comprising SEQ ID NO: 4 or a functional variant or homologue thereof; and

(4) contacting the terpene indole alkaloid with the first and/or second enzyme, and optionally also the third and/or fourth enzyme, under catalytic conditions to produce a terpene indole alkaloid derivative.

See Tables 1 and 2, Fig. 26 and the co-filed SEQUENCE LISTING for full sequence information.

The term“terpene indole alkaloid” encompasses a terpene indole alkaloid which have halogens (for example, fluorine) or other substituents on the Indole ring.

The terpene indole alkaloid may be stemmadenine acetate.

The stemmadenine acetate may be contacted with the first enzyme and the terpene Indole alkaloid derivative may be precondylocarpine acetate (which may be open or closed tautomer form). The stemmadenine acetate may be contacted with the first and second enzymes to produce

dihydroprecondy!ocarpine acetate and/or catharanthine. Alternatively, the stemmadenine acetate may be contacted with the first, second and third enzymes and the terpene indole alkaloid derivative produced may be catharanthine. Alternatively, the stemmadenine acetate may be contacted with the first, second and fourth enzymes and the terpene indole alkaloid derivative produced may be tabersonine. The terpene Indole alkaloid may be precondy!ocarplne acetate. The precondy!oearplne acetate may be contacted with the second enzyme and the terpene indole alkaloid derivative(s) produced may be dihydroprecondylocarpine acetate and/or catharanthine. Alternatively, the precondylocarpine acetate may be contacted with the second and third enzymes and the terpene Indole alkaloid derivative produced may be catharanthine. Alternatively, the precondylocarpine acetate may be contacted with the first, second and fourth enzymes and the terpene indole alkaloid derivative produced may be tabersonlne.

The terpenoid indole alkaloid may be dihydroprecondylocarpine acetate. The

dihydroprecondylocarpine acetate may be contacted with the third enzyme and the terpene indole alkaloid derivative may be catharanthine. Alternatively, the dihydroprecondylocarpine acetate may be contacted with the fourth enzyme and the terpene indole alkaloid derivative may be tabersonine. Alternatively, the dihydroprecondylocarpine acetate may be contacted with the third and fourth enzymes and the terpene indole alkaloid derivatives may be tabersonine and catharanthine.

The first enzyme may be encoded by a first nucleotide comprising SEQ ID NO: 5, or a functional variant or homologue thereof. The first nucleotide may comprise the nucleotide sequence SEQ ID NO: 6.

The second enzyme may be encoded by a second nucleotide comprising SEQ ID NO: 7 or SEQ ID NO: 67, or a functional variant or homologue thereof.

The third enzyme may be encoded by a third nucleotide comprising SEQ ID NO: 9, or a functional variant or homologue thereof.

The fourth enzyme may be encoded by a fourth nucleotide comprising SEQ ID NO: 10, or a functional variant or homologue thereof.

The nucleotide(s) used in the method of the invention may be plant sequences. The plant sequences may be from an ibogaine-produeing plant (for example, Tabernanthe iboga, also known as“iboga”) and/or aspidosperma-producing plant, for example Catharanthus roseus.

The method of the invention may be performed in vivo, for example in pianta. The terpene Indole alkaloid may be provided by injection info the plant. The plant may be Nicotiana benthamiana.

The enzyme or enzymes used in the method of the invention may be provided by expression in vivo, for example heterologous expression. The method of the invention may be performed in vitro, for example in an isolated plant ceil. The plant cell may be a Nicotiana benthamiana ceil. The method may alternatively be performed in yeast, for example Pichia pastohs or Saccharomyces cerevisiae. The method of the invention may alternatively be performed in bacteria, for example E. coli.

The enzyme or enzymes may be provided by expression, for example heterologous expression, in the yeast, bacteria or plant cell.

In another aspect, the invention relates to a method for producing a biologically active composition, comprising the steps of:

(1 ) providing a terpene indole alkaloid;

(2) providing:

(a) a first enzyme having a first amino acid sequence comprising SEQ ID NO: 1 or a functional variant or homologue thereof: and/or

(b) a second enzyme having a second amino acid sequence comprising SEQ ID NO: 2, or SEQ ID NO: 66 or a functional variant or homologue thereof;

(3) optionally also providing:

(c) a third enzyme having an amino acid sequence comprising SEQ ID NO: 3 or a functional variant or homologue thereof; and/or

(d) a fourth enzyme comprising an amino acid sequence comprising SEQ ID NO: 4 or a functional variant or homologue thereof;

(4) contacting the terpene indole alkaloid with the first and/or second enzyme, and optionally also the third and/or fourth enzyme, under catalytic conditions to produce a terpene indole alkaloid derivative; and

(5) converting the terpene Indole alkaloid derivative into a biologically active composition.

In step (5), the converting may be enzymatic or synthetic.

The composition produced by the method of the invention may have anti-cancer activity, and may be, for example, vinblastine.

The methods of the invention may exclude naturally occurring processes, i.e. processes in which one or more of the recited enzymes of the invention are not provided.

The terpene Indole alkaloid derivative may be tabersonine, wherein converting the tabersonine Into vinblastine comprises the steps of:

(1 ) converting the tabersonine to vindoline; and (2) synthetically and/or enzymatically coupling the vindoline with catharanthine to produce vinblastine.

The composition may have anti-diabetic activity, and may be, for example, conophyiline.

The composition may be a vasodilator, and may be, for example, vinca ine.

The composition may have anti-addiction activity, and may be, for example, ibogaine.

In another aspect, the invention relates to a kit comprising:

(1 } a first enzyme having a first amino acid sequence comprising SEQ ID NO: 1 or a functional variant or homo!ogue thereof;

and optionally also comprising:

(2) a second enzyme having a second amino acid sequence comprising SEQ ID NO: 2, or SEQ ID NO: 66 or a functional variant or homo!ogue thereof;

(3) a third enzyme having an amino acid sequence comprising SEQ ID NO: 3 or a functional variant or homologue thereof; and/or

(4) a fourth enzyme comprising an amino acid sequence comprising SEQ ID NO: 4, or a functional variant or homologue thereof.

In another aspect, the invention relates to an isolated enzyme having an amino acid sequence comprising SEQ ID NO: 1 , or a functional variant or homologue thereof

In another aspect, the invention relates to an isolated nucleic acid having a nucleotide sequence comprising SEQ ID NO: 5, or a functional variant or homologue thereof.

In another aspect, the invention relates to an expression vector encoding an enzyme having an amino acid sequence comprising SEQ ID NO: 1 , or a functional variant or homologue thereof, and optionally also encoding: a second enzyme having an amino acid sequence comprising SEQ ID NO: 2 or SEQ ID NO: 66, or a functional variant or homologue thereof; and/or a third enzyme having an amino acid sequence comprising SEQ ID NO: 3 or a functional variant or homologue thereof; and/or a fourth enzyme comprising an amino acid sequence comprising SEQ ID NO: 4 or a functional variant or homologue thereof.

The expression vector may include an artificial regulatory sequence. in another aspect, the invention relates to a host ceil comprising a nucleic acid having a nucleotide sequence comprising SEQ ID NO: 5, or a functional variant or homo!ogue thereof, and/or an expression vector according to the invention.

In another aspect, the invention relates to a host cell which has been genetically modified to express the enzyme of the Invention and optionally also to express: a second enzyme having an amino acid sequence comprising SEQ ID NO: 2 or SEQ ID NO: 88, or a functional variant or homo!ogue thereof; and/or a third enzyme having an amino acid sequence comprising SEQ ID NO: 3 or a functional variant or homologue thereof; and/or a fourth enzyme comprising an amino acid sequence comprising SEQ ID NO: 4 or a functional variant or homologue thereof.

The host cell may be a yeast cell such as a Pichia pastohs cell, or a plant cell, for example a Nicotiana benthamiana cell.

In another aspect, the invention relates to a genetically modified plant comprising the nucleic acid of the invention and/or the expression vector of the invention.

In another aspect, the invention relates to a plant which has been gene edited to express the enzyme of the invention, and optionally also gene edited to express: a second enzyme having an amino acid sequence comprising SEQ ID NO: 2 or SEQ ID NO: 66, or a variant, homologue, or functional variant thereof; and/or a third enzyme having an amino acid sequence comprising SEQ ID NO: 3 or a functional variant or homologue thereof; and/or a fourth enzyme comprising an amino acid sequence comprising SEQ ID NO: 4 or a functional variant or homologue thereof. The plant may be Nicotiana benthamiana.

The present invention also encompasses Identification or selection of organisms such as plants with modified activity of one or more of the enzymes described herein. Markers for such identification or selection may be developed using methods known in the art from the nucleic acid sequences described herein.

As used herein, a“functional variant or homologue” is defined as a polypeptide or nucleotide with at least 50% sequence identity, for example at least 55% sequence identity, at least 60% sequence identity, at least 65% sequence identity, at least 70% sequence identity, at least 75% sequence identity, at least 80% sequence Identity, at least 85% sequence identity, at least 90% sequence identity, at least 95% sequence identity, at least 96% sequence identify, at least 97% sequence identity, at least 98% sequence Identity, or at least 99% sequence identity with the reference sequence. Sequence identity between nucleotide or amino acid sequences can be determined by comparing an alignment of the sequences. When an equivalent position in the compared sequences is occupied by the same base or amino acid, then the molecules are identical at that position. Scoring an alignment as a percentage of identity is a function of the number of identical amino acids or bases at positions shared by the compared sequences. When comparing sequences, optimal alignments may require gaps to be introduced into one or more of the sequences to take into consideration possible insertions and deletions in the sequences. Sequence comparison methods may employ gap penalties so that, for the same number of Identical molecules in sequences being compared, a sequence alignment 'with as few gaps as possible, reflecting higher relatedness between the two compared sequences, will achieve a higher score than one with many gaps. Calculation of maximum percent Identity involves the production of an optimal alignment, faking Into consideration gap penalties.

Suitable computer programs for carrying out sequence comparisons are widely available in the commercial and public sector. Examples Include MatGat (Campaneila et a!., 2003, BMC

Bioinformatics 4: 29; program available from http://bitincka.com/ledion/matgat), Gap (Needleman & Wunsch, 1970, J. Mol. Biol. 48: 443-453), PASTA (Altschul et a!., 1990, J. Mol. Biol. 215: 403-410; program available from http://www.ebi.ac.uk/fasta), Ciusta! W 2.0 and X 2.0 (Larkin et al., 2007, Bioinformatics 23: 2947-2948; program available from http://www.ebi.ac.uk/tools/clustalw2) and EMBOSS Pairwise Alignment Algorithms (Needleman & Wunsch, 1970, supra ; Kruskai, 1983, In: Time warps, string edits and macromolecules: the theory and practice of sequence comparison, Sankoff & Kruskai (eds), pp 1 -44, Addison Wesley; programs available from

http://www.ebi.ac.uk/tools/emboss/align). All programs may be run using default parameters.

For example, sequence comparisons may be undertaken using the“Needle” method of the

EMBOSS Pairwise Alignment Algorithms, which determines an optimum alignment (including gaps) of two sequences when considered over their entire length and provides a percentage identity score. Default parameters for amino acid sequence comparisons (“Protein Molecule” option) may be Gap Extend penalty: 0.5, Gap Open penalty: 10 0, Matrix: B!osum 62 Default parameters for nucleotide sequence comparisons (“DNA Molecule” option) may be Gap Extend penalty: 0.5, Gap Open penalty: 10.0, Matrix: DNAtu!l.

In one aspect of the Invention, the sequence comparison may be performed over the full length of the reference sequence. Particular non-limiting embodiments of the present invention will now be described in detail.

Example 1 :

Introduction

The biochemistry required for tabersonine 2 and catharantbine 3 formation from the known alkaloid intermediate stemmadenine 1 is described below.

It has been hypothesized that catharanthine 3 (iboga-type alkaloid) and taberson!ne 2

(aspidosperma-type) scaffolds are generated by dehydration of the known biosynthetic intermediate stemmadenine 1 to dehydrosecodine 9, which can then cyclize to either catharanthine 3 or taberson!ne 2 via a net [4+2] cycloaddition reaction (Scheme 2A). Since the dehydration product dehydrosecodine 9 is highly unstable, we though that it would be too reactive to diffuse from an enzyme active site without decomposing. Therefore, we concluded that the dehydration and cyc!ization reactions would be catalyzed by a single enzyme.

Since the biosynthetic genes for vincristine 5 and vinblastine b are not clustered in the plant genome, RNA-seq data was searched for gene candidates In RNA-seq data from the

vincristine/vinblastine producing plant Catharanthus roseus. These data revealed that a gene annotated as 2-bydroxyisof!avanone dehydratase shared a similar co-expression profile with previously identified vinblastine pathway genes (Fig. 1 ). In addition, further analysis revealed the existence of a close homologue (80% amino acid identity) that has low expression levels in all the tissues included in the RNA-seq experiment. These dehydratase genes were selected as candidates to be tested using virus-induced gene silencing (VIGS) in C. roseus. Silencing of the highly expressed gene, herein named Tabersonine Synthase (TS), caused a substantial decrease in the accumulation of tabersonine 2 (p=0.0048) and vindoline 4 (p=0.02) (Scheme 1 , Fig. 2), the major taberson!ne-der!ved alkaloid in C. roseus leaves, while silencing of the gene expressed at lower levels, herein named Gatharanthine Synthase (CS), resulted in a statistically significant decrease in catharanthine (p=0.01 ) and in an increase of vindoline (p=0.01 ) (Fig. 3).

These silencing experiments therefore strongly support the involvement of CS and TS In catharanthine 3 and tabersonine 2 biosynthesis in C. roseus. However, when CS and TS 'were heteroiogously expressed in E. coil (Fig 4A) and tested for reactivity with stemmadenine 1 , no reaction occurred. Previous work showed that certain monoterpene indole alkaloids with similar scaffolds have a propensity to spontaneously derormy!ate. It was speculated, particularly given the abundance of acetyl-transferases in the RNA-seq dataset (Fig 1 B), that an acetylated version of stemmadenine 1 , in which deformylation would be hindered, is the actual substrate for CS and/or TS (Scheme 2B) This compound could be readily synthesized from stemmadenine 1 (Figs 5 and 6, Tables 4 and 5), but stemmadenine acetate 7 was also not turned over by CS or TS. Frequently, when a gene is silenced in plants, the substrate or the corresponding enzyme will accumulate. However, no new compound that accumulated in the tissue of the silenced plants was detected, so in this case, gene silencing provided no clues to the identity of the CS/TS substrate(s).

An attempt was made to isolate the active substrate for the TS and CS enzymes from various aspidosperma- and iboga-alkaloid producing plants using enzyme-assay guided fractionation. Tabemaemontana plants were used, as they are known to accumulate more stemmadenine 1 intermediate relative to the downstream alkaloids, in the hope that these plants would accumulate more of the TS/CS substrate for purification. These experiments demonstrated that TS and CS were always active with the same fractions (Fig. 7), consistent with the hypotheses that both enzymes utilize the same substrate. However, attempts to structurally characterize the substrate were complicated by the rapid decomposition of the molecule, and the deformylated product tubotaiwine 12 was the major compound detected by NMR (Scheme 2B, Fig 8, Table 7). Given the propensity for deformylation in these structural systems, it was rationalized that tubotaiwine 12 could be the result of the actual substrate, which would correspond to /sostemmadenine 8

(d!hydropreeondy!ocarpine) or Its protected form (dihydroprecondylocarpine acetate 11) (Scheme 2B). Both dihydroprecondylocarpine 8 and dihydroprecondylocarpine acetate 11 are susceptible to fragmentation, and would not be expected to accumulate in TS/GS-siienced tissues, consistent with our VIGS experiments. However, with a lead for the identity of the TS/CS substrate in hand, it was surmised that a coupled oxidation-reduction cascade could perform a net isomerization to generate dihydroprecondylocarpine 8 (or dihydroprecondylocarpine acetate 11 ) from stemmadenine 1 (or stemmadenine acetate 7) The RNA-seq dataset was examined for two redox enzymes that could convert stemmadenine acetate 7 to dihydroprecondylocarpine acetate 11.

It was noted that a gene annotated as reticuiine oxidase had low absolute expression levels, but a similar tissue expression pattern to the TS gene (Fig. 1 C), identified by using a self-organizing map algorithm. The chemistry of previously reported reticuiine oxidase enzymes, such as berberine bridge enzyme and dlhydrobenzophenanthrldine oxidase, suggests that these enzymes are capable of C-N bond oxidation, which is what would be required in this reaction sequence (Scheme 2B). When this oxidase gene was silenced in C. roseus, a compound with a mass and ¹ H NMR spectrum corresponding to semi-synthetica!!y prepared siemmadenlne acetate 7 (the proposed oxidase substrate) accumulated, suggesting that this gene encoded the correct oxidase, and was thus named precondylocarpine acetate synthase (PAS) (Figs 9-1 1 ). Similarly, silencing of a medium chain alcohol dehydrogenase, as part of an ongoing screen of alcohol dehydrogenase function In C roseus metabolism, resulted in accumulation of a compound with a mass, retention time and fragmentation pattern consistent with a partially characterized synthetic standard of

precondylocarpine acetate 10 (the proposed reductase substrate) (Figs 12-14; Table 6). This standard could be synthesized from stemmadenine acetate 7 using Pt and O2, yields were low and variable, and the product decomposed during characterization. However, the limited 2D NMR data set was consistent with an assignment of precondylocarpine acetate 10. Thus, this alcohol dehydrogenase was renamed dihydroprecondy!ocarpine acetate synthase (DPAS). Collectively, these silencing data strongly suggest that PAS and DPAS act in concert with CS or TS to generate catharanthine 3 and tabersonine 2.

To validate whether these enzymes produce catharanthine 3 and tabersonine 2, PAS, DPAS and CS or TS were transiently co-expressed in the presence of stemmadenine acetate 7 in Nicotiana benthamiana, a convenient plant expression host. These experiments illustrated the sequential activity of the newly discovered enzymes, whereby the formation of catharanthine 3 with plant tissue expressing PAS/DPAS/CS was observed, as well as tabersonine 2 'with experiments containing PAS/DPAS/TS, when the leaf was also co-infi!trated with stemmadenine acetate 7 (Fig. 28A). The presence of all proteins was vaiidated by proteomics analysis (Fig. 4B). Formation of

precondylocarpine acetate 10 was observed when stemmadenine acetate 7 was infiltrated into A/. benthamiana in the absence of any heterologous enzymes (Fig. 28A), suggesting that an endogenous redox enzyme(s) of A/ benthamiana can oxidize siemmadenlne acetate 7.

In addition to serving as the precursors for vincristine 5 and vinblastine 6, tabersonine 2 and catharanthine 3 are precursors for dozens of other biologically active alkaloids (Scheme 1 B).

Purified, heterologous protein was required to validate the biochemical steps of this reaction sequence in vitro. While CS, TS and DPAS ail expressed in soluble form in E. coii (Fig. 4A), the flavin-dependent enzyme PAS failed to express In standard expression hosts such as E. coii or S. cerevisiae. To overcome this obstacle, the native full-length PAS was expressed in N. benthamiana plants using a transient expression system (Fig. 4B) and a chimeric version of the protein, in which the AZ-termina! signal peptide was replaced with a yeast secretory signal sequence, in Pichia pastohs (Fig. 4G and D). In both cases, the presence of PAS was vaiidated by proteomic data. When the PAS proteins, along with stemmadenine acetate 7, were combined with heterologous DPAS and CS, catharanthine 3 was formed, and when combined with DPAS and TS, tabersonine 2 was observed (Fig. 15 and 16). Moreover, reaction of PAS with stemmadenine acetate 7 produced a compound that had an identical mass and retention time to our semi-synthetic standard of precondy!ocarpine acetate 10 (Fig. 15 and 16). Semi-synthetic precondy!ocarp!ne acetate 10 could be reacted with DPAS and TS/CS to yield tabersonine 2 and catharanthine 3, respectively (Fig. 17). In addition, a crude preparation of dihydroprecondyiocarpine acetate 11 was converted to catharanthine 3 and tabersonine 2 by the action of CS and TS, respectively (Fig. 18). The enzymatic assays with PAS protein derived from P. pastoris ensure that formation of the expected products is not the result of a protein contaminant found in the plant-expressed PAS proteins. Although the yeast-produced PAS was less active than the plant-produced enzyme, formation of a compound corresponding to precondylocarpine acetate 10 was observed when PAS (P. pastoris) 'was reacted with

stemmadenine acetate 7 (Fig. 19), along with formation of catharanthine 3 or tabersonine 2 when DPAS and CS or TS were also included (Fig. 16).

Reaction of PAS (purified from N. benthamiana) and DPAS with stemmadenine acetate 7 yielded catharanthine 3, suggesting that cyclization to catharanthine 3 can occur spontaneously under these reaction conditions (Fig. 15). Tabersonine 2 'was never observed in the absence of TS, suggesting that, at least under the aqueous (pH 8.5) conditions used in the enzyme assays, dehydrosecodine 9 does not spontaneously form the aspidosperma-type scaffold. As observed during attempts to purify the CS/TS substrate from Tabernaemontana plants, dihydroprecondyiocarpine acetate 11 can also deformylate to form tubotaiwine 12. It is likely that the solvent and reaction conditions play a significant role in how the highly reactive dihydroprecondyiocarpine acetate 11 decomposes.

Notably, PAS failed to react with stemmadenine 1 , indicating that the acetyl group is an important recognition group for this enzyme (Fig. 20). It was noted that upon oxidation of stemmadenine 1 , a compound with a mass consistent with that of the shunt product condy!ocarpine 13 resulted; this 'was additionally supported by comparison of the MS/MS spectrum to the closely related compound tubotaiwine 12 (Fig. 21 ). It is hypothesized that the acetylation of stemmadenine 1 is necessary to slow spontaneous deformyiation after oxidation, while also serving as a good leaving group to allow formation or dehydrosecodine 9. Acetylation has also served a role as a protecting group in the biosynthesis of noscapine in opium poppy.

The reactivity of the intermediates involved in the transformation of stemmadenine acetate 7 to catharanthine 3 or tabersonine 2 suggests that PAS, DPAS and CS/TS should be eo-!oca!!zed, since the unstable post precondylocarpine acetate 10 intermediates may not survive transport between ceil types or compartments. Using YFP-tagged proteins in C. roseus cell suspension culture, the inventors showed that PAS is targeted to the vacuole through small vesicles budding from the endoplasmic reticulum (ER), as was previously observed for the PAS homologue, berberine bridge enzyme (Fig. 22). This expected localization indicates that stemmadenine acetate 7 oxidation occurs in the ER-!umen, ER-derived vacuole-targeted vesicles and/or vacuole. In contrast, co-!oca!!zation of DPAS, CS and IS were confirmed in the cytosol (Figs 23 and 24). Furthermore, bimo!ecular fluorescence complementation showed strong Interactions between DPAS and TS, along with much 'weaker interactions between DPAS and CS (Fig. 28R, Fig. 25). Strong DPAS/TS interactions may not only prevent undesired reactions on the reactive dihydroprecondyocarpine acetate 11 intermediate but may also play a role in controlling the flux of 11 into tabersonine 2.

Homologues of PAS are utilized widely throughout benzy!!soquino!!ne and pyridine alkaloid biosynthesis, but there are key sequence mutations in PAS that appear to be unique to the enzymes found in aspidosperma and iboga alkaloid producing plant c!ades (Fig 28C). For instance, PAS lacks the His and Cys residues Involved In covalent binding of the FAD cofactor (Fig. 28). It is anticipated that these aspidosperma-assoclated PAS homologues will be a rich source of biosynthetic enzymes for the wide range of aspidosperma alkaloids found in nature. DPAS is a member of the medium chain alcohol dehydrogenase family, an enzyme class that has been previously shown to play numerous roles in monoterpene indole alkaloid biosynthesis CS and TS represent the first hydrolase-type enzymes implicated in monoterpene indole alkaloid biosynthesis. It was hypothesized that CS and TS retain the hydrolysis function of the putative ancestor hydrolase enzyme to allow formation of dehydrosecodine 9 from dlhydroprecondylocarpine acetate 11

However, the non-enzymafic formation of cafharanthine 3 suggests that formation of

dehydrosecodine 9 via deacetoxylation can also occur spontaneously. In principle, the formation of tabersonine 2 and cafharanthine 3 Is formed via two different modes of cyc!izat!on, and it has been noted that dehydrosecodine 9 can undergo two distinct Die!s-A!der reactions to form either cafharanthine 3 or tabersonine 2 (Scheme 2A). The handful of putative Diels-Alderases have been isolated from microbes; CS and TS would constitute the first examples of Diels-Alderases from plants. Through detailed analysis of the structure and active site composition of TS and CS (amino acid identity of CS and TS is 80%, Fig. 27), it will be revealed whether this cyclization is actually concerted. Active site features that control the cyclization modes can be determined. We propose that TS/GS could be engineered to produce other naturally occurring dehydrosecodine-derived alkaloid scaffolds such as enf-catharanthlne (Beatty & Stephenson, 2014, J. Am. Ghem. Soc. 136: 10270-10273), /so-catharanthine (Yun et a! , 2016, Chem. Sci. 7: 5530-5536), a//o-catharanthlne (Szantay et al., 1990, Tetrahedron 46: 171 1 -1732) or pseudotabersonine. Discussion

Herein is disclosed four enzymes that convert stemmadenine acetate 7 to tabersonine 2 and catharanthine 3. This completes the biosynthetic pathway for vindoline 4 and catharanthine 3, compounds that can be used to semi-syntbetica!iy prepare vinblastine. Heterologous production of catharanthine 3 and vindoline 4, which is of high interest to synthetic biologists, is now possible. These discoveries are the first step to enhanced production of catharanthine 3 and vindoline 4, as well as the many other aspidosperma and iboga alkaloids that also use these enzymes in their biosynthesis.

Materials and Methods

1.1. Chemicals and molecular biology kits

Ail solvents used tor extractions, chemical synthesis and preparative HPLC were of HPLG grade, whilst solvents for UPLC/MS analysis were of MS grade. All were purchased from Fisher Scientific. Catharanthine 3 was purchased from Sigma Aldrich, whilst tabersonine 2 was obtained from Ava Chem Scientific. Stemmadenine was. Kanamycin sulfate, carbenicililn and gentamycin were from Formedium, whilst r!famp!cin was from Sigma Aldrich. All gene and fragment amplifications were performed using Platinum Superfi polymerase (Thermo Fisher) whilst colony PCRs were performed using Phlre II master mix (Thermo Fisher). PGR product purifications were performed using the Maeherey-Nage! PGR clean-up kit. Plasmids purifications were performed using Promega Wizard minipreps. cDNA 'was prepared using Superscript IV VILO master mix and Turbo DNAse (Thermo Fisher). qPGR was performed using Sensi-FAST Sybr No-ROX kit (Bioline). All restriction enzymes and ligase were from NEB.

1.2. RNA-seq data and analysis for biosynthetic gene candidates

Analysis of the C. roseus gene expression profile data was performed on the transcriptome dataset available from the Medicinal Plant Genomics Resource website

(http://medicinalplantaenomics.msu.edu/final version release info.shtml). Co-expression analysis by hierarchical clustering was performed on the FPKM matrix using the algorithms embedded in the Multi Experiment Viewer (MeV v.4.8), whilst self-organizing maps analysis was performed. A cluster of ca. 3600 co-reguiated contigs was identified. This cluster contained ail the known genes Involved in the C. roseus MIA biosynthetic pathway. After further analysis based on functional annotation and gene onthology, a list of ca. 300 genes of interest was compiled, from which the inventors selected candidates for V!GS analysis based on putative function. 1.3. Virus induced gene silencing , metabolite analysis and qPCR

Fragments for CS and IS silencing were selected on the 3’-UTR regions, due to the high sequence similarity between the two genes, whilst fragments for PAS and DPAS silencing were designed using sequence from the ORF regions. Primers are shown in Table 1 . A BLAST search against the transcriptome for each of the regions suggested that the VIGS fragments selected did not contain regions of homology that have significant overlap to other genes in the plant that could cause potential cross-silencing. The silencing fragments for CS, TS and DPAS were amplified from cDNA using the primers listed in Table 1 , treated 'with restriction enzymes BamHl and Xho\ and ligated into pTRV2 vector using T4 ligase. The fragment for PAS silencing was amplified from cDNA using the set of primers in Table 1 and cloned into the USER compatible VIGS plasmid pTRV2u as previously described (Geu-Flores et ai, 2012, Nature 492: 138-142).

VIGS experiments were performed using the C. roseus Little Bright Eye variety grown in a growth chamber at 25 °C with a 12 h dark/12 h light regime. Briefly, each construct was infiltrated Into 10 to 12 C. roseus seedlings (8 weeks old). Additionally, eight seedlings were infiltrated with pTRV2 lacking an insert (empty vector negative control) and four plants were infiltrated with a vector containing a fragment of the protoporphyrin IX magnesium cheiatase gene (Gh!H), which provided a visual marker (bleaching) to act as a positive control. After 21 days, seedlings infiltrated with the pTRV-ChIH vector displayed substantial yellowing of leaves; the last leaf pair to emerge above the inoculation site was harvested, frozen In liquid nitrogen and homogenized using a cryo bead mill. A portion of each sample (10-20 mg) was used for metabolite analysis, whilst the remaining or the samples were used for RNA extraction.

Samples for metabolite analysis were extracted with 1 mL of MeOH containing 1 pg/mL ajmaline as internal standard, filtered and diluted 1 :4 with MeOH before LC/MS analysis using the method described in the UPLC/MS section.

Relative transcript abundance was determined by qRT-PCR on a BioRad CFX96 G-PCR instrument using cDNA synthesized from isolated total RNA and the primers listed in Table 1. Eight biological replicates and three technical replicates were analyzed for each gene using two reference genes: Expressed protein, EXP, and N2227-like family protein, N2227. Efficiencies for all primer sets were approximately equal and always >90%. The entire VIGS experiment was performed in triplicate with essentially Identical results. 1.4. Expression and purification of proteins

CS, TS and DP AS expression in E. coii

The iuii-iengih sequences of CS, TS and DPAS were amplified from C. roseus cDNA using ihe primers listed in Table 1. The PCR products were purified from agarose gel, ligated into the BamH I and Kpn\ restriction sites of the pOPINF vector using the In-Fusion kit (Clontech Takara) and transformed into chemically competent E. coii Stellar cells. Recombinant colonies were selected on LB agar plates supplemented with carbenici!!in (100 pg/mL) Positive clones were Identified by colony PCR using T7_Fwd and pOPIN_Rev primers (see Table 1 ). Plasmids were isolated from positive colonies grown overnight. Identities of the inserted sequences were confirmed by Sanger sequencing.

Chemically competent So!uBL21 E. coii cells (Amsbio) were transformed by heat shock at 42 ^° C. Transformed cells were selected on LB agar plates supplemented with carbenicillin (100 pg/mL). Single colonies were used to inoculate starter cultures In 50 mL of 2 x YT medium supplemented with carbenicillin (100 pg/mL) that were grown overnight at 37 ^° C. Starter culture (10 mL) was used to inoculate 1 L of 2 x YT medium containing the antibiotic. The cultures 'were incubated at 37 ^° C until OD600 reached 0.6 and then transferred to 16 ^° C for 30 min before induction of protein expression by addition of IPTG (0.2 mM). Protein expression was carried out for 16 h. Cells were harvested by centrifugation and re-suspended In 50 mL or Buffer A (50 mM Tris-HCI pH 8, 50 mM glycine, 500 mM NaCI, 5% glycerol, 20 mM Imidazole,) with EDTA-free protease Inhibitors (Roche Diagnostics Ltd.). Ceils were lysed by sonication for 4 minutes on ice. Cell debris was pelleted by centrifugation at 35,000 g for 20 min.

Hise-tagged enzymes were purified on an AKTA Pure system (GE Healthcare) using a HisTrap HP 5 mL column (GE Healthcare) equilibrated with Buffer A. Samples were loaded at a flow rate of 2 mL/minute and step-eluted using Buffer B (50 mM Tris-HC! pH 8, 50 mM glycine, 500 mM NaCI, 5% glycerol, 500 mM imidazole). Eluted proteins were subjected to further purification on a Superdex Hiload 16/60 S200 gel filtration column (GE Healthcare) at a flow rate of 1 mL/m!nute using Buffer C (20 mM HEPES pH 7.5, 150 mM NaCI) and collected in 1.5 mL fractions.

Transient expression of proteins in N. benthamiana

CS, TS and DPAS full-length sequences were cloned into a modified TRBO vector (Lindbo, 2007, Plant Physiol. 145: 1232-1240) in which the cloning cassette of the pOPINF vector was inserted in the Noi\ restriction site. This allowed the vector to be compatible with the PCR products generated for cloning into pOPINF vector and to obtain N-termina! Hise-tagged recombinant proteins. Cloning was performed using the In-Fusion kit PAS full-length, instead, was cloned into pDONR207 (Thermo Fisher) using primers with attB1 and attB2 overhangs (see Table 1 ) via BP Clonase reaction, then recombined into pEAG-HT-DEST3 vector for transient expression with a C-terminai Hise-tag using LR Clonase reaction.

The constructs were used for E. coii Stellar cells transformation by heat shock and recombinant colonies were selected on LB + Kanamycin (100 pg/mL). Positive colonies were also screened by colony PCR using the primers listed In Table 1 and sequenced. The constructs were then used to transform eiectrocompetent A. tumefaciens strain GV31 G1 by electroporation. Recombinant colonies were selected on LB agar containing rifampicin (100 pg/mL), gentamyc!n (50 pg/mL) and kanamycin (100 pg/mL). Single colonies 'were grown in 10 mL of LB with antibiotics for 48 h at 28 ^" C, then the ceils were collected by centrifugation and re-suspended in 10 mL infiltration buffer (10 mM NaG!,

1 .75 mM CaCh and 100 pM acetosyringone). After Incubation at room temperature for 2 h, the cell cultures were diluted to OD600 0.1 and used to Infiltrate N. benthamiana leaves. When multiple constructs were infiltrated simultaneously, the corresponding A. tumefaciens ceil cultures were mixed so that the final OD600 of each would be 0.1 . Infiltration was performed using a syringe without needle on leaves of 3-4 weeks old plants. Leaves 'were harvested 5 days post-infiltration.

Protein purification for proteomics analysis was performed by extraction of 2 g of pulverized frozen tissue in 10 mL of cold Tris-HCI buffer (50 mM, pH 8.0) containing EDTA-free protease inhibitors. After incubation on ice for 1 h and vortexing, the samples were centrifuged for 20 min at 35,000 g. The supernatants were collected and Incubated with 300 pL of Ni-NTA slurry for 1 h. The slurry was then collected by centrifugation at 1 ,000 gr for 1 min and washed 3 times 'with 10 mL of Tris-HCI buffer. Proteins were eluted by washing the slurry with 600 pL of Tris-HCI buffer containing 500 mM imidazole.

To purify PAS for in vitro enzyme activity assays, 300 g of fresh N. benthamiana leaves that had been Infiltrated with the PAS expression construct were homogenized in 600 mL of Tris-HCI buffer (50 mM, pH 8.0) containing EDTA-free protease inhibitors and 1 % insoluble polyvinylpolypyrrolidone (PVPP) using a blender. The homogenate was filtered through two layers of miracloth and then centrifuged at 3,500 g for 10 min to remove the insoluble PVPP and tissue debris. The supernatant was further clarified by centrifugation at 35,000 g for 20 min. PAS was then bound to a 5 mL Ni-NTA column and eluted with 20 mL of Trls-HGI buffer containing 500 mM imidazole. The eluted protein was dialyzed in ConA binding buffer (20 mM Tris-HCI buffer pH 7.4, 500 mM NaCI, 1 mM MnCL and 1 mM CaCE) and manually applied to a ConA HiTrap 1 mL column (GE Healthcare) at a low flow rate using a syringe. After loading of the protein, the column was washed with 10 mL of binding buffer and then eluted with 10 mL of ConA elution buffer (20 mM Tris-HCI buffer pH 7.4, 500 mM NaCI and 300 mM methy!-D-g!ucaside) The protein was then dialyzed into Tris-HCI buffer (50 mM, pH 8.0), concentrated and stored at -20 ^° C.

Expression of PAS in Pichia pastoris

A truncated version of PAS, lacking the Initial 23 amino acids, was generated by PGR (Table 1 ) and cloned into the pPINK-HC vector (Thermo Fisher) according to the manufacturer’s instructions. In this way, the native plant N-termina! signal peptide was replaced with the yeast a-mat!ng sequence for extracellular secretion. The pPINK-HC::FAS construct was transformed into P. pastoris by electroporation In accordance with the P!ehiaPink™ Expression System protocol.

A single colony of a P. pastoris transformant was inoculated in 10 mL BMGY medium in a 250-mL Erlenmeyer flask and grown for 24 hours at 250 rpm and 30 ^° G. The inoculation was then transferred to 100 mL BMGY medium In a 500-mL baffled flask and cultured in the same conditions. After another 24 hours, this culture was transferred to 2 L BMGY medium, split equally into three 2-L baffled flasks, and grown at 220 rpm, 30 ^° C for approximately two days until GDeoo reached 2 to 3. The cells were then collected by centrifugation at 5,000 g for 5 minutes, re-suspended in 700 mL BMMY medium (containing 0.5 % methanol) for protein expression, split equally into two 2-L non- baffied flasks. The culture was allowed to grow at 28 ^° C and 250 rpm. After 120 hours, the culture was centrifuged at 10,000 s for 10 minutes.

The medium containing secreted proteins (supernatant) was concentrated using 30,000 MWGO concentrators (Merck Millipore) to approximately 20 mL, dialyzed into 50 mM HEPES buffer pH 7.0 using PD-10 desalting columns (GE Healthcare), and further concentrated to 3 mL. The sample was subjected to IEX chromatography on a HiTrap Q HP 1 mL column (GE Healthcare) to enrich the PAS protein. After loading of the sample, the column was washed with 10 mL of 50 mM HEPES buffer pH 7.0 before elution of the protein 'with 50 mM HEPES buffer pH 7.0 containing 500 mM NaCI. The sample was dialyzed in 50 mM Tris-HC! pH 8.5 and concentrated to a final volume of 1 mL. BMGY and BMMY media were prepared in accordance with the PichiaPink™ Expression System protocol.

Concentrated P. pastoris culture medium containing secreted proteins was analyzed on SDS-PAGE with Goomassie staining. A band corresponding to the size of PAS (ca. 57 kDa) was visible. To confirm the identity of the protein, the band was excised, de-stained and subjected to trypsin digestion and LG/MS/MS analysis on a nanoLC-orbitrap (Thermo Fisher Scientific). 1.5. Pathway reconstitution in N. benthamiana

Leaves of 3-4 weeks old N. benthamiana plants 'were Infiltrated with A. tumefaciens GV3101 cultures harbouring the transient protein expression constructs, as described above. After 4 days, each leaf was infiltrated with 1 mL of 50 mM stemmadenine acetate dissolved in Infiltration buffer. After 24 h, the infiltrated leaves were harvested and flash-frozen in liquid nitrogen. After grinding in liquid nitrogen, 200-300 mg of pulverised tissue were extracted with 1 volume (w/v) of MeOH containing ajma!!ne as internal standard. After Incubation at room temperature for 1 h and vortexing, the samples were centrifuged at 17,000 c? for 10 min, filtered and analyzed by UPLC/MS using the same method described in the UPLC/MS section.

1.6. in vitro enzyme assays

in vitro assays of PAS alone 'were performed In 50 mM Tris-HCI buffer pH 9.0, whilst those in which PAS was coupled with the other enzymes were performed in 50 mM Tris-HCI pH 8.5. In ail cases 20 mM FAD was added as co-factor. Each assay contained 50 mM stemmadenine acetate (substrate) delivered in methanol (not exceeding 5% of the reaction volume). Due to the very low amounts of PAS purified, the amount of protein in the assays was not accurately determined. However, the amount of enzyme added to each reaction was consistent throughout each set of experiments.

Reactions involving DPAS, CS and TS were performed in 50 mM HEPES buffer pH 7.5. DPAS requires NADPH for activity, therefore 100 mM NADPH was added to each assay. Precondylocarpine acetate (DPAS substrate) (50 mM) delivered in MeOH (not exceeding 5 % of the reaction volume) was added to each reaction 10-20 pg of enzymes were used in the assays. All reactions were performed at 37 °C for 1 h. For controls, protein sample was replaced with boiled protein sample or sample from culture of non-transformed yeast. After incubation, the reactions were quenched by addition of 1 volume of methanol, filtered through 0.22-pm nylon Spin-X centrifuge filters (Corning) and analyzed by LC/MS as described in the UPLC/MS and UPLC/GqG-MS sections.

1.7. Purification of CS/TS substrate from plant material

Alkaloids from fully expanded leaves of Tabernaemontana divaricata“Flore p!eno” (50 g) were extracted with acetonitrile (300 mL x 3). The extract was concentrated in vacuo and reconstituted in acetonitriie:water (70:30). The hydrophobic components were removed by passing through a Cis SPE cartridge (10 g). The flow-through was collected, dried and dissolved in 20 mL of 50 mM phosphate buffer pH 7.0. 5 mL of solution was sequentially applied to 500 mg WCX OASIS cartridges (Waters) equilibrated with phosphate buffer. Each cartridge was washed 'with buffer (6 mL), acetonitrile (6 mL), acetonitriie:water (50:50) solutions before elution 'with 100 mM CaCL in acetonitrile:water (6 mL, 50:50). Fractions were !yophilized and reconstituted in acetonitri!e:wa!er (2 mL, 50:50) before subjection to reverse-phase preparative HPLC.

Preparative HPLG was performed on a Thermo Dionex Ultimate 3000 chromatography apparatus using a Phenomenex Luna C18 (5 pm, 30 x 250 mm) column. The solvents used were 0.01% acetic acid, solvent A, and acetonitrile, solvent B. A linear gradient from 5% B to 50% B over 25 min was used to separate the alkaloids. Chromatography was performed at a flow rate or 30 mL/min and monitored with a UV detector at 254 nm. Fractions (30 mL) were collected and assayed for CS and TS activity as follows. A 200 pL aliquot of each fraction was dried in vacuo and re-dissolved in 50 pL of 50 mM HEPES buffer pH 7.5. CS or TS enzyme (1 pg) was added to each sample and the reactions were incubated at 37 °C for 1 h. Control reactions 'without enzyme were also prepared. After incubation, the reactions were quenched by addition of 50 pL of MeOH, filtered and analyzed by UPLC/QqQ-MS as described above.

1.8. Purification of stemmadenine acetate from C. roseus leaves.

Leaves of plants in which PAS was silenced by V!GS were harvested 21 days post infection and frozen In liquid nitrogen. 20-50 g of frozen leaves were ground and extracted 3 times with acetonitrile (300 mL x 3). The extract was concentrated in vacuo and reconstituted In 100 mL acetonitriie:water (70:30). The hydrophobic components were removed by passing the sample through a Cia SPE cartridge (10 g). The flow-through was collected, dried and dissolved In 10 mL of acetonitri!e:wafer (50:50).

Stemmadenine acetate 7 was purified by semi preparative HPLC on a Thermo Dionex Ultimate 3000 chromatography apparatus using a Waters Xbridge BEH C18 (5 pm, 10 x 250 mm) column. Mobile phase A was water containing 0.1 % formic acid; mobile phase B was acetonitrile. The flow rate was 5 mL/min, and the gradient profile was 0 min, 10% B; from 0 to 30 min, linear gradient to 35% B; from 30 to 35 min, isocratic 35% B; from 35.1 to 37 min, wash at 95% B then back to the initial conditions of 10% B for 3 min. The injection volume was 500 pL. Elutions of the compounds was monitored at 268 nm. 5 mL fractions were collected throughout the purification and tested for the presence of the compounds by direct Injection on an Advlon express-ion Compact Mass

Spectrometer in ESL mode. Fractions containing the compound of Interest were lyophiilzed

1.9. Liquid-chromatography mass spectrometry analysis

UPLC/MS

This method was applied to the analysis of VIGS leaf extract, in vitro enzyme assays and synthetic products, unless otherwise indicated in the description of the experiments. UPLC/MS analysis was performed on a Shimadzu LCMS-IT-TOF Mass Spectrometer coupled to a Nexera 2 chromatographic system. Chromatographic separation 'was carried out on a Phenomenex Kinetex column 2.6 pm XB-C18 (100 c 2.10 m), and the binary solvent system consisted of solvent A, H ₂ 0 + 0.1 % formic acid, and solvent B, acetonitrile. Flow rate was 600 pL7min. A linear gradient from 10% to 30% solvent B in 5 min, allowed the separation of the alkaloids of interest. The column was then washed at 100% B for 1 .5 min and re-equilibrated to 10% B. Injection volume was 1 pL.

Mass spectrometry was performed in positive ion mode 'with scanning over the m/z range from 150- 1 ,200. The source settings were the following: heat block temperature 300 °C, nebulizing gas flow 1 4 L/mln, CDL temperature 250 °C, detector voltage 1.6 kV. Data analysis was performed using the Shimadzu Profiling Solution software.

UPL C/QqQ- S

This method was used to analyse samples from the in vitro pathway reconstitution using PAS expressed in P. pastoris and CS/TS reactions of alkaloid fractions purified from Tabernaemontana divaricata“Flore Pieno” leaves. UPLC/QqQ-MS analysis was carried out on a UPLC (Waters) equipped with an Acquity BEH C18 1.7 pm (2.1 x 50 mm) column connected to Xevo TQS (Waters). Chromatographic separation was performed using 0.1 % NH ₄ OH as mobile phase A and acetonitrile as mobile phase B. A linear gradient from 0 to 65% B in 17.5 min was applied for separation of the compounds followed by an increase to 100% B at 18 min, a 2-min wash step and a re-equilibration at 0% B for 3 min before the next Injection. The column was kept at 60 °C throughout the analysis and the flow rate was 0.6 mL/min

MS detection was performed in positive ESI. Capillary voltage was 3.0 kV; the source was kept at 150 °C; desoivation temperature was 500 °C; cone gas flow, 50 L7h; and desolvation gas flow, 800 L/h. Unit resolution was applied to each quadrupole. The MRM transitions used to monitor the elution of the alkaloids of interest are reported in Table 3.

HR-MS

For high resolution MS analysis, compounds were infused at 5-10 pL/min using a Harvard Apparatus syringe pump onto a Synapt G2 HDMS mass spectrometer (Waters) calibrated using a sodium formate solution. Samples were analyzed for 1 minute with a scan time 1 sec in the mass range of 50-1200 m/z. Capillary voltage was 3.5 V, cone voltage 40 V, source temperature 120 ^° C, desolvation temperature 350 ^” C, desolvation gas flow 800 L/h. Leu-enkepha!ine peptide (1 ng/pL) was used to generate a dual lock-mass calibration with [M+H] ⁺ = 558.2766 and m/z = 278.1 135 measured every 10 see. Spectra were generated in MassLynx 4 1 by combining a number of scans, and peaks were centred using automatic peak detection with lock mass correction.

1.10. Proteo ic analysis

PAS, DPAS, CS and TS transiently expressed in N. benthamiana leaves and pre-purified on NiNTA resin were precipitated 'with chioroform/metbanoi and dissolved in 0.2 M TEAB/1% sodium deoxycholate (SDC), whilst PAS expressed in P. pastohs was extracted from SDS-PAGE. Protein concentration was determined using the Direct Detect™ Assay (Merck). 10 pg of protein 'was treated with DTT and iodoacetamide to reduce and alkylate cysteine residues and digested with 1 pg of trypsin (Promega) at 50 °C for 8 h. Approx. 0.5 pg of the digested protein was used for data dependent LC-MS/MS analysis on an Orbitrap-Fusion™ mass spectrometer (Thermo Fisher, Fieme! Hempstead, UK) equipped with an U!iiMaie™ 3000 RSLCnano System (Thermo Fisher) using a nanoEase M/Z HSS C18 T3 1 .8 pm, 150 pm x 100 mm, (Waters). The samples were loaded and trapped using a pre-column which was then switched in-line to the analytical column for separation. Peptides were eluted with a gradient of acetonitrile in water/0.1 % formic acid (main step from 11 - 30.5% at a rate of 0.19% min ¹ ). The column was connected to a 10 pm Si!icaTip™ nanospray emitter (New Objective, Woburn, MA, USA) for infusion into the mass spectrometer. Data dependent analysis was performed using an HCD fragmentation method with the following parameters: positive ion mode, orbitrap MS resolution = 60k, mass range (quadrupole) = 300-1800 m/z, MS2 in ion trap, threshold 2e ⁴ , Isolation window 1.6 Da, charge states 2-5, MS2 top20, AGC target 1.9e ⁴ , max inject time 35 ms, dynamic exclusion 1 count, 15 s exclusion, exclusion mass window ± 5 ppm. MS scans were saved in profile mode 'while MS2 scans were saved In centroid mode.

Raw files were processed with MaxGuant (version 1.6.1.0) (http://maxauant.ora). The searches were performed using the Andromeda search engine in MaxQuant on a custom database of the N.

benthamiana sequences available from Uniprot to 'which the protein sequences of interest were added using trypsin/P with 2 missed cleavages, carbamidomethylatlon (G) as fixed and oxidation (M), acetylation (protein N-terminus), and deamidation (N,Q) as variable modifications. Mass tolerances 'were 4.5 ppm for precursor ions and 0.5 Da for fragment ions.

1.11. Synthesis procedures

Genera ! and NMR analysis

Progress of the reactions was monitored by direct injection on an Advion express-ion Compact Mass Spectrometer in ESP mode. The mobile phase was 0.1 % formic acid in watermethano! (10:90). LC/MS analysis was performed using the method described in the UPLC/MS section, unless otherwise stated. Higb-reso!uiion mass spectrometry was performed as described in the HR-MS section. NMR spectra (1 D and 2D NMR) were acquired using a Bruker Avance ill 400 NMR spectrometer equipped with a BBFO plus 5 mm probe, unless stated otherwise. The residual ¹ H and ¹³ C NMR signals of GD ₃ OD (6 3.31 and 49.0, respectively), CD ₃ GN (d 1.94 and 1.32, respectively) and GDCI ₃ (d 7.26 and 77.16, respectively) were used for calibration. The number of scans depended on sample concentration and are indicated in Figures and Tables accordingly.

Synthesis of stemmadenine acetate (7)

To a 1 .5-mL FIPLC vial containing stemmadenine 1 (10.2 mg, 0 0289 mmol), pyridine (400 pL) was added. The mixture was sonicated and stirred at r.t until complete dissolution was achieved. Acetic anhydride (50 m!_, 0.529 mmol) was subsequently added to the reaction vessel and the reaction was allowed to stir at r.t for 4 h. Reaction progress was monitored by MS direct- inject! on of 1 pL of the reaction mixture in 100 pL of methanol. The reaction was quenched with methanol (1 mL) once peak for SM at m/z 355 was no longer observed. The reaction was then concentrated in vacuo at 30 °G. Toluene (350 pL) and methanol (150 uL) was added to reaction vial, sonicated to homogeneity and concentrated in vacuo at 30 °C (repeated x4) to afford 33 (ca. 1 1 .4 mg, 0.0287 mmol, 99%) as a dark brown solid.

Synthesis of precondytocarpine acetate (18)

Synthesis of precondylocarpine acetate was performed as reported by Scott and co-workers (1972; see above) with some modifications. Briefly, Adam’s catalyst (> 75 m ² /g, 25 2 mg, 0.1 12 mmol, 7.5 equivalents) was reduced with H _s in ethyl acetate (1 mL) for 2 h. The apparatus 'was flushed with nitrogen for 5 min. The freshly prepared platinum in ethyl acetate was transferred to a 25-mL pear- shaped flask containing stemmadenine acetate 7 (6 mg, 0.0152 mmol) dissolved in ethyl acetate (1 mL) via glass pipette. An oxygen atmosphere was introduced by a balloon to the reaction vessel via Agani 1 .5-inch needle. The reaction mixture was allowed to stir vigorously at r.t. Reaction progress was monitored by dissolving 1 pL of the reaction mixture in 100 pi of methanol, filtered through a Fisherbrand™ porosity 0.2 pm PTFE syringe filter before injection on the Advion MS. Fresh batch of Pf catalyst was prepared and added as above when producLSM ratio appeared to stagnate. The reaction was stopped when amount of SM was <5% product (estimated via Advion MS) to prevent formation of by-product(s). After flushing the reaction vessel with N ₂ , gravity filtration with ethyl acetate (10 mL) gave a pale-yellow solution. The filtrate was concentrated in vacuo at 25 °G to afford a dark yellow solid (crude yield: 2.6 mg, -60% pure based on ¹ H NMR; see also Table 6). Synthesis of dihydroprecondyiocarpine acetate (11)

Synthesis of dihydroprecondyiocarpine acetate was performed. Briefly, Adam’s catalyst (> 75 m ² /g, 0.8 mg, 3.524 pmol) was reduced with H ₂ in ethanol (400 mϋ for 2 h. The apparatus was flushed with nitrogen for 5 min. The freshly prepared platinum in ethanol was transferred to a 1 5-mL HPLC vial containing crude precondy!ocarpine acetate 10 (ca. 200 pg, 0.508 pmol) dissolved in ethanol (100 mϋ via glass pipette. A hydrogen atmosphere was introduced by a balloon to the reaction vessel. The reaction mixture was allowed to stir vigorously at r.t. overnight. The reaction vessel 'was flushed 'with N ₂ , followed by filtration of the reaction mixture using Flsherbrand™ porosity 0.2 pm PTFE syringe filter, and washing of the filter with EtOH (200 pL x 2), and concentrating the filtrate in vacuo yielded a compound (yield undetermined due to small scale) which showed activity in enzymatic assays with CS, and TS, forming catharanth!ne 3, and tabersonlne 2, respectively. Due to the small scale and instability, the compound could not be further characterized though.

Synthesis of condyiocarpine (13)

Synthesis of condyiocarpine was performed analogously to that for precondy!ocaprine acetate 10. Adam’s catalyst (> 75 m2/g„ 2.30 mg, 10.1 pmol, 7.5 equivalents) was reduced with H ₂ in ethyl acetate (1 mL) until solution turned black. The apparatus was flushed with nitrogen for 5 min. The freshly prepared platinum in ethyl acetate was transferred to a 1 5-mL HPLC vial containing stemmadenine 1 (0.48 mg, 1.36 pmol). An oxygen atmosphere was Introduced by a balloon to the reaction vessel via Aganl 1.5-inch needle. Reaction progress was monitored by dissolving 1 pL of the reaction mixture In 100 pi of methanol, filtered through a FisherbrandTM porosity 0.2 pm PTFE syringe filter before Injection on the Advion MS in ESP mode. After 2h, conversion reached ca. 26%, which did not improve by 4 h. The reaction was stopped to prevent formation of by-product(s). After flushing the reaction vessel 'with N2, the reaction mixture 'was filtered as above and the catalyst was washed with ethyl acetate (6 L), and concentrating in vacuo yielded a yellow residue. The product was the Isolated by preparative HPLC on a Waters Xbridge BEH C18 (5 pm, 10 x 250 mm) column. Mobile phase A was 'water containing 0.1 % formic acid; mobile phase B was acetonitrile. The flow rate was 5 mL/min, and the gradient profile was 0 min, 10% B; from 0 to 30 min, linear gradient to 40% B; from 30 to 35 min, linear gradient to 95% B; from 35 to 39 min, wash at 95% B then back to the initial conditions of 10% B in 1 min for 5 min. The injection volume was 500 pL. Elution of the compounds was monitored at 290 nm. 2.5 L fractions were collected throughout the purification and tested for the presence of the compounds by direct injection on the Advion MS In ESI ⁺ mode. 1.12. Localization and Interaction studies

Subcei!u!ar localizations of DPAS, CS and IS were studied by creating fluorescent fusion proteins using the pSGA-cassette YFPi plasmid. Full-length open reading frames encoding each enzyme were amplified using specific primer couples (Table 1 ), which were designed to introduce the Spel restriction site at both cDNA extremities. PGR products were sequenced and cloned at the 5’-end of the yellow fluorescent protein (YFP) coding sequence, to generate the DPAS-, TS-, CS-YFP fusion proteins or at the 3’-end to express the YFP-DPAS, YFP-TS and YFP-CS fusions. The coding sequences of the first forty or sixty amino acids of PAS encompassing the ER-to-vacuoie targeting sequence were amplified by mixing PAS-YFP for with PAS40-YFPrev or PAS6G-YFPrev (Table 1 ) before cloning into the Spel restriction site of pSGA-cassette YFPi to express sp40-YFP and spoQ-YFP.

Interactions of DPAS with CS and TS were characterized by bimolecular fluorescence complementation (BiFC) assays using the previously amplified GS and TS PGR products cloned via Spel into the pSCA-SPYNE173 to express CS-YFPN and TS -YFPN; and using DPAS amp!icon cloned into pSCA-SPYCE (MR) to express YFPC-DPAS. Plasmids encoding LAMT-YFPN and LAMT- YFPC were used as controls.

Transient transformation of C. roseus ceils by particle bombardment and fluorescence imaging were performed. Briefly, C. roseus plated cells were bombarded with DNA-coated gold particles (1 pm) and 1 ,100 psi rupture disc at a stopping-screen-to-target distance of 6 cm, using the Bio-Rad PDS1000/He system. Cells were cultivated for 16 h to 38 h (DPAS, GS and TS) or up to 96 h for PAS before being harvested and observed. The subce!iu!ar localization was determined using an Olympus BX-51 epifiuorescence microscope equipped 'with an Olympus DP-71 digital camera and a combination of YFP and CFP filters. The pattern of localization presented in this work is representative of circa 100 observed ceils. Localizations of the different fusion proteins were confirmed by co-transformation experiments using a nuciear-GFP marker, an ER-GFP marker, vacuole-CFP marker or a nucleocytosollc-GFP marker. Such plasmid transformations were performed using 400 ng of each plasmid or 100 ng for BiFC assays. In these assays, transformed cells were Identified via co transformation with the nucleus GFP marker.

Although the present invention has been described with reference to preferred or exemplary embodiments, those skilled in the art will recognize that various modifications and variations to the same can be accomplished 'without departing from the spirit and scope of the present invention and that such modifications are dearly contemplated herein. No limitation with respect to the specific embodiments disclosed herein and set forth In the appended claims is intended nor should any be inferred.

A!! documents cited herein are incorporated by reference in their entirety.

TABLES

Table 1. Primer sequences used in this study, Cioning/restriction sites are underlined.

Table 2, Selected nucleotide and amino acid sequences of the enzymes used in this investigation (see also accompanying SEQUENCE LISTING).

T bJe_3J¾/fRiyi transitions used for metabolites detection with UPLC/QqQ-tVIS method.

Compound Parent ion Daughter ion Collision Energy (V)

Catharanthine 337.2 173.1 16

165.1 20

144.1 20

Tabersonine 337.2 305.2 22

228.2 22

168.1 36

Stemmadenine 397.2 337.1 18

acetate

228.1 24

168.0 40

Precondyiocarp!ne 395.2 234 0 38

acetate

228.1 22

196.1 32

Tabie 4. NMR data for stemmadenine 1. HRMS, ESS positive: m/z caicuiated for C _2I H ₂₇ N ₂ 0 ₃ ⁺ [M-t-H]*: 355.2016, observed:

355

o\

19 129.4 129.5 unclear ^d 5.56 (1 H, q, 6.9) 5.55 (1 H, q, 6.0) 5.45 (1 H, m) ^d

20 127.0 126.8 126.8 n/a n/a n/a

21 52.8 52.4 53.5 3.36 (1 H, m) 3.35 (1 H, m) unclear ^® O

2.51 (1 H, mj 2.50 (1 H, m) 2.5 (1 H, m) ^d o

COOMe 172.1 172.0 172.4 n/a n/a n/a

CGGMe 52.4 52.0 52.3 3.70 (3H, s) 3.70 (3H, s) 3.69 (3H, s) ^d

n/a n/a n/a 5.76 (1 H, t, 4.2) 5.79 (1 H, t, 5.0) 5.73 (1 H, br s) ®

³ recorded at 300 K, 400 MHz (100 MHz for ¹³ C) o b recorded at 298 K, 300 MHz (75 MHz for ¹³ C) Grover et at. (2002) Magn. Reson. Chem. 40: 474-476.

c temperature and frequency not reported. Feng et al. , 2010, J. Nat. Prod. 73: 22-26.

d No assignments or numerical data reported, just images of spectra. Assignments here are based on comparison with Grover et al. and

our data.

-Pi.

CD

O

H

O

ES

o o o

Table 5. NMR data for stemmadenine acetate 7 in comparison to the starting material stemmadenine 1. HRD/IS, ESI positive: m/z calculated for C23H29N2C [M÷H] ⁺ : 397.2122, observed: 397.21 2, Appm= 0.0,

Table 8. MR data and key correlations for precondyiocarpine acetate 10, HRDIS, ESI positive: m/z calculated for C23H27N204 [M÷H]÷: 395.1965, observed: 395,1967, Appm= 0.5.

,

1.84 (1 H, m)

7 n.d. ^b n/a

8 147.5 n/a

9 1 12.2 7.44 (1 H, m)

10 127.3 7.25 (1 H, did, 7.3, 7.3, 1.3)

1 1 128.4 7.31 (1 H, ddd, 7.5, 7.5, 1.4)

12 121.3 7.45 (1 H, m)

13 154.4 n/a

14 31.5 2 13 (1 H, m)

2.05 (1 H, m)

15 34.6 3.52 (1 H, m)

16 n.d. ^b n/a

17 67.8 5.01 (1 H, d, 1 1.2)

3.92 (1 H, d, 1 1.2)

18 13.3 1.57 (3H, d, 6.9)

19 124.2 5.35 (1 H, q, 6.9)

20 134.8 n/a

21 76.1 3.89 (1 H, d, 2.3)

CQQMe 172.5 n/a

GQQMe 52.9 3.71 (3H, s)

0C(0)Me 171.4 n/a

OG(Q) e 21.0 2.05 (3H, s]

a recorded at 300 K, 400 MHz (100 MHz for ^1J G)

b Not identified by HMBC due to instability and low scale TabJej^NWIR data for tubotaiwine 12 obtained as the degradation product of dihydroprecondyiocarpsne acetate 11 (580 pg crude mass. 1.79 pmol).

^¾ recorded at 400 MB?. {108 Hz for «C); temperature not reported

= recorded si 29S K. 600 MHz {125 MHz for «0}

sNu Kssigstmems reported; sigaais ssskj-red t!sre 6 on stertterify toYsmaudi sd ai sad our data

‘Assignment supported sy HM8C ccjrmiaisons {OCXs Me to CCOMe, B-8 to C-2)

< «sited 6y HM8C

s>No routSpSsdty reported