The seco-iridoid pathway from Catharanthus roseus

Abstract

The (seco)iridoids and their derivatives, the monoterpenoid indole alkaloids (MIAs), form two large families of plant-derived bioactive compounds with a wide spectrum of high-value pharmacological and insect-repellent activities. Vinblastine and vincristine, MIAs used as anticancer drugs, are produced by Catharanthus roseus in extremely low levels, leading to high market prices and poor availability. Their biotechnological production is hampered by the fragmentary knowledge of their biosynthesis. Here we report the discovery of the last four missing steps of the (seco)iridoid biosynthesis pathway. Expression of the eight genes encoding this pathway, together with two genes boosting precursor formation and two downstream alkaloid biosynthesis genes, in an alternative plant host, allows the heterologous production of the complex MIA strictosidine. This confirms the functionality of all enzymes of the pathway and highlights their utility for synthetic biology programmes towards a sustainable biotechnological production of valuable (seco)iridoids and alkaloids with pharmaceutical and agricultural applications.

ARTICLE

Received 23 Aug 2013 | Accepted 10 Mar 2014 | Published 7 Apr 2014

DOI: 10.1038/ncomms4606 OPEN

The seco-iridoid pathway from Catharanthus roseus

Karel Miettinen1,*, Lemeng Dong2,*, Nicolas Navrot3,*, Thomas Schneider4,w, Vincent Burlat5,Jacob Pollier6, Lotte Woittiez4,w, Sander van der Krol2, Raphal Lugan3, Tina Ilc3, Robert Verpoorte1, Kirsi-Marja Oksman-Caldentey7, Enrico Martinoia4, Harro Bouwmeester2, Alain Goossens6,Johan Memelink1 & Danile Werck-Reichhart3

The (seco)iridoids and their derivatives, the monoterpenoid indole alkaloids (MIAs), form two large families of plant-derived bioactive compounds with a wide spectrum of high-value pharmacological and insect-repellent activities. Vinblastine and vincristine, MIAs used as anticancer drugs, are produced by Catharanthus roseus in extremely low levels, leading to high market prices and poor availability. Their biotechnological production is hampered by the fragmentary knowledge of their biosynthesis. Here we report the discovery of the last four missing steps of the (seco)iridoid biosynthesis pathway. Expression of the eight genes encoding this pathway, together with two genes boosting precursor formation and two downstream alkaloid biosynthesis genes, in an alternative plant host, allows the heterologous production of the complex MIA strictosidine. This conrms the functionality of all enzymes of the pathway and highlights their utility for synthetic biology programmes towards a sustainable biotechnological production of valuable (seco)iridoids and alkaloids with pharmaceutical and agricultural applications.

1 Sylvius Laboratory, Institute of Biology Leiden, Leiden University, Sylviusweg 72, PO Box 9505, Leiden 2300 RA, The Netherlands. 2 Laboratory of Plant Physiology, Wageningen University, Droevendaalsesteeg 1, Wageningen 6708 PB, The Netherlands. 3 Institut de Biologie Molculaire des Plantes, Unit Propre de Recherche 2357 du Centre National de la Recherche Scientique, Universit de Strasbourg, 28 rue Goethe, Strasbourg 67000, France. 4 Institute of Plant Biology, University Zurich, Zollikerstrasse 107, Zurich CH-8008, Switzerland. 5 CNRS; UMR 5546, Universit de Toulouse; UPS; UMR 5546, Laboratoire de Recherche en Sciences Vgtales, BP 42617 Auzeville, Castanet-Tolosan F-31326, France. 6 Department of Plant Systems Biology, VIB and Department of Plant Biotechnology and Bioinformatics, Ghent University, Technologiepark 927, Gent B-9052, Belgium. 7 Industrial Biotechnology, VTT Technical Research Centre of Finland, P.O. Box 1000, FI-02044 VTT (Espoo), Finland. * These authors contributed equally to this work. w Present addresses: Philip Morris

International R&D, Philip Morris Products S.A., Quai Jeanrenaud 5, Neuchatel 2000, Switzerland (T.S.); Plant Production Systems Group, Wageningen University, P.O. Box 430, Wageningen 6700 AK, The Netherlands (L.W.). Correspondence and requests for materials should be addressed to D.W.-R. (email: mailto:[email protected]

Web End [email protected] ) or to J.M. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606

Monoterpenoid indole alkaloids (MIAs) are a large group of plant-derived natural products with a range of pharmacological properties. Examples of MIAs are

camptothecinused to treat cancerand quininethe antimalarial drug of choice till the mid of the last century. Madagascar periwinkle, Catharanthus roseus, the best-characterized MIA-producing plant species, is the source of the valuable MIAs vincristine and vinblastine, which are used directly or as derivatives for the treatment of several cancer types. Owing to the extremely low concentrations (0.0002% fresh weight), production of vincristine and vinblastine is expensive and availability of the drug is sensitive to environmental and political instability in the production countries. Therefore, biotechnology-based production of MIAs in microorganisms or alternative plant hosts has been proposed as a sustainable substitute; however, progress has been hampered by the lack of knowledge of the enzymes responsible for MIA biosynthesis, particularly in the secologanin pathway (Fig. 1). Secologanin is the monoterpenoid (also called iridoid or secoiridoid) branch end point and is coupled to tryptamine by strictosidine synthase (STR) to form strictosidine, the universal MIA precursor in plants. The secologanin pathway has broad importance as many plant species accumulate iridoids and secoiridoids (including secologanin) as end products without incorporating them into complex alkaloids. Many (seco)iridoids are bioactive themselves, with among others anticancer, antimicrobial and anti-inammatory activities14. Iridoids are also pheromones in some insect species, and as such can be employed for pest management in agriculture and the control of insect-related disease vectors5,6.

The (seco)iridoid pathway is still largely unresolved. It starts with geraniol and is thought to comprise B10 enzymes catalysing successive oxidation, reduction, glycosylation and methylation reactions (Fig. 1). Although the pathway has been investigated for decades7,8, only the rst step (geraniol 10-hydroxylase/8-oxidase, G8O)9 and the two last steps (loganic acid O-methyltransferase, LAMT; and secologanin synthase, SLS)10,11 are well established. Only recently, two additional enzymes were identied, iridoid synthase (IS) responsible for the reductive cyclization step12 and geraniol synthase (GES)13. A complicating factor for gene discovery as well as biotechnological production is that the MIA pathway in C. roseus is organized in a complex manner, with the enzymes localized in different cell types and subcellular compartments (Supplementary Fig. 1)14,15.

Here we report the characterization of the last missing steps of the C. roseus secoiridoid pathway. We use an integrated transcriptomics and proteomics approach for gene discovery, followed by biochemical characterization of the isolated candidates. Furthermore, we reconstitute the entire MIA pathway up to strictosidine in the plant host Nicotiana benthamiana, by heterologous expression of the newly identied genes in combination with the previously known biosynthesis genes. This work provides essential tools that will allow development of synthetic biology platforms for the production of bioactive iridoids, secoiridoids and complex MIAs with a wide range of agricultural and pharmaceutical applications, including the treatment of cancer.

ResultsIridoid gene discovery. We previously reported the assembly of CathaCyc, a C. roseus metabolic pathway database based on Illumina HiSeq2000 RNA sequencing data16. The data set was derived from C. roseus suspension cells and shoots treated with the plant hormone methyl jasmonate16. Here we complemented this data set with RNA-Seq data from cell suspensions overexpressing either ORCA2 or ORCA3, transcription factors

that regulate the expression of LAMT, SLS and several other genes in the MIA biosynthesis pathway17, but not GES and G8O (Fig. 2a). These four and all other known MIA genes are induced by MeJA both in cell suspension cultures and whole C. roseus plants, although GES/G8O and LAMT/SLS show different induction characteristics (Fig. 2b). Furthermore, GES, G8O and IS are expressed in the internal phloem-associated parenchyma (IPAP) cells, whereas LAMT and SLS are expressed in the leaf epidermis1013,18,19. The differential induction and in situ expression data suggested that the rst part of the pathway (possibly up to 7-deoxyloganic acid) corresponds to one transcriptional regulon, whereas the subsequent steps up to the synthesis of secologanin would comprise a second regulon.

On the basis of the hypothetical pathway and predicted enzyme activities catalysing the hitherto missing steps (Fig. 1), we screened our data set for genes encoding NAD(P)-binding Rossmann fold domain-type oxidoreductases, cytochrome P450 monooxygenases (P450s) and UDP-glycosyltransferases (UGTs) that display co-expression with GES/G8O, and for P450s that show co-expression with LAMT/SLS. Three genes encoding putative oxidoreductases showed a high degree of co-expression with GES/G8O (Fig. 2a). The rst (accession number Caros008267 in the ORCAE database: http://bioinformatics.psb.ugent.be/orcae

Web End =http://bioinformatics.psb. http://bioinformatics.psb.ugent.be/orcae

Web End =ugent.be/orcae 16) was annotated as a progesterone reductase, the second (Caros003452) as an aldehyde dehydrogenase and the third (Caros009903) as a 12-oxophytodienoate reductase. We also identied four P450s (Caros020058, Caros001222, Caros018961 and Caros005234) and one UGT (Caros009839) that showed close co-expression with GES/G8O (Fig. 2a). The rst oxidoreductase (Caros008267) was found to encode the recently described iridoid synthase12, thus conrming the validity of our screening strategy. The others were selected for further functional analysis. We also veried the co-expression of these candidate genes in the publicly available data set from the Medicinal Plant Genomics Resource consortium (http://medicinalplantgenomics.msu.edu

Web End =http://medicinalplantgenomics. http://medicinalplantgenomics.msu.edu

Web End =msu.edu ), which has been integrated into the ORCAE website (http://bioinformatics.psb.ugent.be/orcae

Web End =http://bioinformatics.psb.ugent.be/orcae 16). This analysis strongly supported our selection of candidate genes (Fig. 2b).

The tissue localization of the enzymes in the leaf was used as a second criterion to pinpoint the most promising candidate genes. This was investigated using proteomics on epidermal and mesophyll protoplasts isolated from C. roseus leaves. The proteomics analysis resulted in the identication of 2,200 proteins. Three P450s (Caros005234, Caros006766 and Caros020058) and one UGT (Caros020739) were enriched in the mesophyll fraction, whereas one P450 (Caros007986), one UGT (Caros004449) and nine oxidoreductases (Caros022489, Caros002459, Caros017236, Caros002170, Caros012730, Caros006689, Caros007544, Caros021570 and Caros003491) were enriched in the epidermal fraction. One P450 (Caros003164) was present in both tissues (Fig. 2c and Supplementary Table 1).

Caros003452 is the missing 8-hydroxygeraniol oxidoreductase. The three oxidoreductases (Caros008267, Caros003452 and Caros009903) were produced in Escherichia coli and puried for in vitro enzyme assays. Conrming a previous report12, the putative progesterone reductase (Caros008267) was shown to possess IS activity in the presence of 8-oxogeranial, yielding a mixture of cis-trans-iridodial and cis-trans-nepetalactol. Caros003452 was active with the substrates 8-OH-geraniol, 8-OH-geranial and 8-oxogeraniol in the presence of NAD , yielding mixtures of the three compounds and 8-oxogeranial in varying relative amounts depending on the combination and the incubation time (Fig. 3). The enzyme was therefore coined 8-hydroxygeraniol oxidoreductase (8-HGO). With the cofactor

2 NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606 ARTICLE

NAD it did not convert 8-oxogeranial (Fig. 3), and it was not active with any of the substrates listed above in the presence of

NADP /NADPH. Relatively high activity was observed with some other primary alcohols such as geraniol (Supplementary

Table 2). Given the complex kinetics, with four interconvertible compounds and eight possible reactions, the reaction constants could not be determined.

CYP76A26 is the missing iridoid oxidase. Six candidate P450 genes (CYP76A26, CYP81Z1, CYP81Q32, CYP72A224, CYP71AY1 and CYP71AY2) were transferred to a yeast expression vector and co-expressed in Saccharomyces cerevisiae together with the P450 reductase ATR1 from Arabidopsis thaliana20 (Supplementary Fig. 2). Functional screening was carried out as described21 with geraniol, 8-hydroxygeraniol, 8-hydroxygeranial, 8-oxogeraniol, 8-oxogeranial, iridodial, iridotrial, 7-deoxyloganic acid and 7-deoxyloganetic acid as potential substrates. CYP76A26 converted both iridodial and iridotrial into 7-deoxyloganetic acid. The cis-iridodial and trans-iridodial freely interconverted with cis-trans-nepetalactol12, and although CYP76A26 seemed to use the bicyclic nepetalactol as the preferred substrate, the monocyclic cis- and trans-iridodials were also utilized, possibly after spontaneous conversion into nepetalactol (Fig. 4). The interconversion and sequential metabolism of nepetalactol and the iridodials in aqueous solution prevented reliable evaluation of the catalytic parameters with these substrates. Iridotrial was previously proposed as an intermediate of the secologanin pathway8. Whereas we never detected iridotrial as an intermediate in the iridodial to 7-deoxyloganetic acid conversion under conditions more likely to capture early reaction products such as low substrate concentrations, short incubations or low temperature, the latter was very efciently converted into 7-deoxyloganetic acid with a Kmapp of 25 mM

and kcat of 5.2 s 1 (Supplementary Fig. 3). This suggests that CYP76A26 is a multifunctional P450 enzyme catalysing successive hydroxylation/dehydration steps via a mechanism similar to that recently described for CYP701A3, which catalyses the conversion of kaurene to kaurenoic acid in the biosynthesis of gibberellins22.

As the related G8O (CYP76B6) was previously shown to oxidize several (other) monoterpene alcohols in addition to geraniol9, these compounds were also tested as potential substrates for CYP76A26. The enzyme converted 8-oxogeraniol into an unidentied product (Supplementary Fig. 4), albeit with a low efciency, and catalysed the hydroxylation of linalool, nerol, citronellol and lavandulol, but not geraniol (Supplementary Fig. 4; Supplementary Table 3). CYP76A26 thus catalyses the conversion of iridodial into 7-deoxyloganetic acid and was consequently named iridoid oxidase (IO; Fig. 1).

UGT709C2 is 7-deoxyloganetic acid glucosyl transferase. The UGT (Caros009839 or UGT709C2), produced in E. coli, catalysed

IPP

+

DMAPP

GPPS

Geranyl diphosphate

Geraniol

GES

OH

G8O

G8O

OH

8-Hydroxygeraniol

CHO

OH

8-HGO

1

2

8-Oxogeraniol

CHO

OH 8-Hydroxygeranial

OH

8-HGO

1

2

CHO

CHO

8-Oxogeranial

IS

cistrans-iridodial

CHO

H

H

H

H

CHO

cistrans-nepetalactol

O

OH

IO

1 2

H

H

H

H

H

H

H

H

H

H

CHO

OH

O Iridotrial

IO

1 2

COOH

7-Deoxyloganetic acid

O

COOH

OGlc

OH

7-DLGT

3

7-Deoxyloganic acid

O

7-DLH

COOH

OGlc

Figure 1 | The secologaninstrictosidine pathway. Genes indicated in boxes were published before (black background) or during (white background) the present study, or are reported here (yellow background). Frames indicate mRNA localization in the leaf IPAP (pink) or epidermis (blue). Numbers indicate predicted enzyme classes in the initial gene discovery strategy. 1: oxidoreductase, 2: cytochrome P450, 3: UGT. IPP, isopentenyl pyrophosphate; DMAPP, dimethylallyl pyrophosphate; Glc, glucose; GPPS, geranyl diphosphate synthase, GES, geraniol synthase; G8O, geraniol 8-oxidase; 8-HGO, 8-hydroxygeraniol oxidoreductase; IS, iridoid synthase; IO, iridoid oxidase; 7-DLGT, 7-deoxyloganetic acid glucosyl transferase; 7-DLH, 7-deoxyloganic acid hydroxylase; LAMT, loganic acid O-methyltransferase; SLS, secologanin synthase; STR, strictosidine synthase; TDC, tryptophan decarboxylase. Iridotrial indicated in brackets was a previously proposed intermediate that we did not detect in vitro or in vivo.

Loganic acid

Loganin

Secologanin

O

L-tryptophan

Tryptamine

N H

OH

TDC

LAMT

NH

COOCH

OGlc

N H

O

OH

SLS

NH

COOCH

OGlc

OHC

H

H

COOCH

OGlc

H

STR

O

O

Strictosidine

NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606

the glucosylation of 7-deoxyloganetic acid to form 7-deoxyloganic acid using UDP-glucose as the sugar donor (Fig. 5). The enzyme had a Kmapp of 9.8 mM and a kcat of 1.25 s 1 (Supplementary

Fig. 5). The enzyme was inactive with loganetic acid, loganetin, iridodial, iridotrial, 8-OH-geraniol, jasmonic acid, gibberellic acid, indole acetic acid, salicylic acid, abscisic acid, zeatin and luteolin. It thus behaved as a selective 7-deoxyloganetic acid glucosyl transferase (7-DLGT; Fig. 1).

CYP72A224 is the 7-deoxyloganic acid hydroxylase. Despite its poor expression in yeast, CYP72A224 catalysed the conversion of 7-deoxyloganic acid into loganic acid in yeast microsomes (Fig. 6) with a Kmapp of 400 mM and a Vmax of 0.01 pmol s 1 per mg microsomal protein (Supplementary Fig. 6). kcat could not be determined due to low expression in yeast preventing precise quantication of the enzyme concentration. Owing to the low expression of CYP72A224 in yeast, a N. benthamiana leaf-disc

CellSus

Sdlg

MeJA6

Not

MeJA

O3

O2

Not

MeJA24

Caros010005-STR1 Caros003452-ADH10 Caros006766-G8O Caros003727-GES Caros005234-CYP72A224 (7-DLH) Caros001222-CYP81Z1 Caros009839-UGT709C2 (7-DLGT) Caros009903-OPR-like Caros020058-CYP76A26 (IO) Caros018961-CYP71AY1 Caros008267-IS Caros020659-SLS Caros011578-STR2 Caros002904-LAMT Caros020166-STR3 Caros009426-SGD Caros014930-TDC

CellSus

WtMeJA

WtCon

HairRt

Sdlg

Con24

Not

Fl

mL

St

Ro

iL

MeJA

Con

Not

YE6

YE12

YE24

MeJA6

MeJA12

MeJA24

WtNot

TdNot

TdMeJA

TdCon

RebH

Caros020166-STR3 Caros010005-STR1 Caros003452-ADH10 Caros003727-GES Caros020058-CYP76A26 (IO) Caros006766-G8O Caros008267-IS Caros018961-CYP71AY1 Caros014930-TDC Caros009426-SGD Caros002904-LAMT Caros020659-SLS Caros011578-STR2 Caros005234-CYP72A224 (7-DLH) Caros009839-UGT709C2 Caros009903-OPR-like Caros001222-CYP81Z1

Epidermis

Mesophyll

Hit numbers

14.79.9

Caros006766

Caros005234

Caros003676

Caros007686

Caros003164

CYP76A26 (IO)

CYP76B6 (G8O)

CYP72A224 (7-DLH)

1.01.0

0.61.2 11.08.4

0.00.0 2.32.5

8.02.6 3.02.6

6.34.2 7.77.2

CYP71AY2

CYP81Q32

Figure 2 | Gene discovery strategy. (a,b) Complete-linkage hierarchical clustering of early MIA pathway gene expression in C. roseus based on our data (a) or the Medicinal Plant Genomics Resource consortium (http://medicinalplantgenomics.msu.edu) (b). Colours indicate transcriptional activation (blue) or repression (yellow) relative to untreated samples. Tissues: Fl, ower; mL, mature leaves; iL, immature leaves; St, stem; Ro, root; Sdlg, seedling. Suspension cells (CellSus): Wt, wild-type; O2, ORCA2; O3, ORCA3. Hairy roots (HairRt): Wt, wild-type; Td, TDCi; RebH, RebH_F. Treatments: Not, no treatment; MeJA, methyl jasmonate (6, 12 or 24 h); Con, mock; YE, yeast extract. (c) Candidate P450 protein hits in the epidermis and mesophyll samples from proteomics analysiss.d. (n 3). GES, geraniol synthase; G8O, geraniol 8-oxidase; 8-HGO, 8-hydroxygeraniol oxidoreductase; IS, iridoid

synthase; IO, iridoid oxidase; 7-DLGT, 7-deoxyloganetic acid glucosyl transferase; 7-DLH, 7-deoxyloganic acid hydroxylase; LAMT, loganic acid O-methyltransferase; SGD, strictosidine b-D-glucosidase; SLS, secologanin synthase; STR, strictosidine synthase (13: three related genes); TDC, tryptophan decarboxylase.

4 NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606 ARTICLE

a

b

84

Peak 1 Peak 3

55

300

150

55

135 121

107 150

8-Oxogeraniol

8-OH-geraniol

8-Oxogeranial

67 83

109

121 148

150

0

300

0

84 55

67

8-Oxogeraniol 8-Oxogeraniol

55

135 121 107

Total Ion count (thousands)

83 109

121 148

0

150

0

150

Relative intensity

75

68

Peak 2

84

55 91 121

Peak 4

200

75

84

8-OH-geranial

100

55 91 119

135

150

1

8-OH-geraniol +8HGO +NAD+

225

8-OH-geraniol

8-OH-geranial

150

68

3

84

55 91 121135150

2

4

84 55

0

91 119

Mass to charge ratio (m/z)

17 18 19 20 21 22 Retention time (min)

Figure 3 | Functional characterization of recombinant 8-HGO. Afnity-puried enzyme expressed in E. coli was incubated with 8-OH-geraniol and NAD. (a) GCMS prole of the reaction extract compared with authentic standards. (b) The identity of peaks 14 was conrmed by comparisonof MS spectra with authentic standards. 8-HGO catalyses the stepwise conversion of 8-OH-geraniol into 8-oxogeraniol or 8-OH-geranial and then into 8-oxogeranial.

cistrans-nepetalactol

1

2

3

7-Deoxyloganetic acid standard67

136

2

2

2

2

2

1

81

0 7-Deoxyloganic acid

1

Relative intensity

0

Intensity (mV)

EV + cistrans-nepetalactol + NADPH

IO + cistrans-nepetalactol + NADPH

93

121

1

0

m/z

IO + cistrans-nepetalactol NADPH

67

1

81

IO product

0

1

Relative intensity

0

9.0 9.5 10.0 10.5 Retention time (min)

93

121

136

m/z

Figure 4 | Functional characterization of IO. The cytochrome P450 enzyme was expressed in yeast and assayed as a microsomal preparation by incubation with cis-trans-nepetalactol and NADPH. (a) GCMS prole of the reaction extract compared with the negative control and authentic standards. (b) Product identity was conrmed by comparison of the MS spectrum with the authentic standard. Cis-trans-nepetalactol is the peak 3 (ref. 13), in the substrate standard, peaks 1 and 2 are its open dialdehyde forms (cis and trans iridodial). These compounds are in equilibrium in water12. Peak 3 is the rst converted by IO, but eventually, possibly after interconversion, all three compounds are metabolized (not shown).

assay was also carried out to further conrm its role in loganic acid formation (Supplementary Fig. 7). The aglycone derivative of 7-deoxyloganic acid (7-deoxyloganetic acid) was not a substrate for CYP72A224, conrming that glycosylation precedes hydro-xylation of the cyclopentane ring. CYP72A224 was thus named 7-deoxyloganic acid hydroxylase (7-DLH).

CYP72A224 belongs to the same P450 subfamily as secologanin synthase (SLS; CYP72A1), which catalyses the conversion of loganin into secologanin10. Both of these P450s use glycosylated substrates and have similar regiospecicities, suggesting that SLS evolved from 7-DLH thus extending the iridoid pathway to the

secoiridoids. The overlap in their substrate specicities was therefore investigated (Supplementary Fig. 8). As SLS showed no 7-DLH activity and vice versa, our proposed evolutionary process appears to have yielded specic and mutually exclusive catalytic activities for each enzyme.

Tissue-specic and subcellular localization of the pathway. Efcient pathway engineering requires precise knowledge of the cellular and subcellular organization of different pathway components. IO, 7-DLH, 8-HGO and 7-DLGT were therefore expressed as green uorescent protein fusions in C. roseus cells

NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606

a b

199

7-DLGT product

Relative intensity Relative intensity

7-Deoxyloganic acid

500

250

0

Absorbance at 235 nm (mAU)

0

400

200

378

361 343

181

7-Deoxyloganetic acid

Retention time (min)

163

m/z

m/z

199

7-Deoxyloganic acid standard

225

150

50

7-DLGT + 7-Deoxyloganetic acid

181

163

378

361 343

0

10 15

14

13

12

11 16

Figure 5 | Functional characterization of 7-DLGT. Afnity-puried enzyme expressed in E. coli was incubated with 7-deoxyloganetic acid and UDP-glucose. (a) HPLC-DAD prole of the reaction extract compared with authentic standards. (b) Product identity was conrmed by comparison of the mass spectrum with the authentic standard.

7-Deoxyloganic acid + 7DLH + NADPH

a b

7-Deoxyloganic acid

0.2

215 7-DLH product

0.1

0.2

0 Loganic acid

Absorbance at 236.5 nm (AU)

Relative intensity

0.2

0.1

0

197

395

378

179

0.2

EV

7-Deoxyloganic acid + 7DLH NADPH

8 9 10 11 12 13 14 Retention time (min)

m/z

0.1

0

215

Loganic acid standard

0.2

0.1

197

0

Relative intensity

179

395 378

0.1

0

m/z

Figure 6 | Functional characterization of 7-DLH. The cytochrome P450 enzyme was expressed in yeast and assayed as a microsomal preparation by incubation with 7-deoxyloganic acid and NADPH. (a) HPLC-DAD prole of the reaction extract compared with the negative control and authentic standards. (b) Product identity was conrmed by comparison of the mass spectrum with the authentic standard.

together with mCherry markers for nucleus, cytosol, endoplasmic reticulum (ER), plastids, mitochondria or peroxisomes. This revealed that IO and 7-DLH are ER-associated as predicted, whereas 8-HGO and 7-DLGT are soluble proteins found in the cytosol and the nucleus (Supplementary Fig. 9). Previous in situ transcript analysis18,19,23 suggested that a pathway intermediate is translocated from IPAP cells (the site of the early biosynthetic steps) to the epidermis, where the nal steps of the pathway occur. In situ hybridization showed that 8-HGO, IO, 7-DLGT and 7-DLH are expressed in the same tissue as G8O (Fig. 7). The separate clustering of the expression of these genes (Fig. 2a) and the similar tissue-specic transcript localization suggest that these genes constitute a transcriptional regulon for the production of loganic acid in IPAP cells. The epidermis-specic positive control SGD conrms the localization of the next transcriptional regulon, consisting of LAMT, SLS, STR and TDC, in the epidermis. The existence of this regulon is supported by the highly similar co-expression proles of these genes as shown in Fig. 2a.

Reconstitution of strictosidine synthesis in N. benthamiana. The identication of the four enzymes described above allowed us to propose a complete secologanin pathway (Fig. 1), which we tested by stepwise combinatorial transient expression of the corresponding genes in N. benthamiana (Fig. 8). To boost substrate availability for the pathway, a geranyl diphosphate synthase from Picea abies (PaGPPS)24 and a geraniol synthase from Valeriana ofcinalis (VoGES)25 were used because the C. roseus orthologues were not available at the onset of these studies. The transient expression of PaGPPS VoGES in

N. benthamiana resulted in the formation of geraniol, but also several additional oxidized and glycosylated derivatives were detected, presumably produced by endogenous N. benthamiana enzymes as previously reported25. The stepwise addition of G8O and IS resulted in the generation of new compounds, including new derivatives of the anticipated pathway intermediates (combinations 1, 2 and 4 in Fig. 8a,b; Supplementary Fig. 10). In contrast, the addition of 8-HGO to the combination

6 NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606 ARTICLE

ipap

ipap

ipap

ipap

ipap

ep

G8O AS

8-HGO AS

IO AS

7-DLGT AS

7-DLH AS 7-DLH S

SGD AS

8-HGO S

IO S

7-DLGT S

Figure 7 | Expression of the transcriptional regulon required for loganic acid biosynthesis in the IPAP cells. In situ hybridization on serial longitudinal sections of young developing leaves was carried out with antisense (AS) probes and sense (S) probes as controls. G8O and strictosidine b-D-glucosidase (SGD) AS probes were used as IPAP and epidermis markers, respectively. Sense probe controls gave no signals (not shown). 8-HGO, 8-hydroxygeraniol oxidoreductase; IO, iridoid oxidase; 7-DLGT, 7-deoxyloganetic acid glucosyl transferase; 7-DLH, 7-deoxyloganic acid hydroxylase; IPAP, internal phloem-associated parenchyma;ep, epidermis. Scale bar 100 mm.

PaGPPS VoGES G8O only modied the existing product

prole quantitatively but did not generate any new compounds (combination 3 in Fig. 8a,b; Supplementary Fig. 10). This is probably due to the fact that endogenous N. benthamiana enzymes have similar activity as 8-HGO (Supplementary Fig. 11). When IO was co-expressed with PaGPPS VoGES G8O 8-

HGO IS (combination 5), the metabolic prole did not change,

and 7-deoxyloganetic acid and its derivatives were not detected (Fig. 8a,b). However, reconstitution of the pathway up to 7-DLGT (combination 6) was successful and resulted in the production of

7-deoxyloganic acid and putative acetylated 7-deoxyloganic acid (Fig 8a,c). Without IO, these products were not detected (Supplementary Fig. 12), indicating that IO is functional inN. benthamiana and is an essential part of the biosynthesis pathway. These ndings also illustrate the importance of full pathway coverage for functional analysis of individual enzymes.

Subsequently, the entire postulated secologanin pathway (PaGPPS to SLS) was introduced by agroinltration; however, this only yielded products up to 7-deoxyloganic acid. Therefore, we increased the input halfway into the pathway by inltrating the intermediates iridodial, iridotrial or 7-deoxyloganic acid in combination with the pathway genes. In all cases this resulted in the production of secologanin, indicating that the second half of the pathway is also functional (Fig. 8d). When 7-DLH, LAMT or SLS were omitted from combination 9with inltration of the intermediates iridodial, iridotrial or 7-deoxyloganic acidno secologanin was detected showing that all conversions require the inltrated C. roseus genes (Supplementary Fig. 13). Finally, we tested whether the biosynthesis pathway up to secologanin can be functionally combined with the tryptamine branch of the MIA pathway. When the secologanin biosynthesis pathway genes were co-inltrated with the tryptophan decarboxylase (TDC) and STR genes, and the ux through the pathway was boosted by co-inltration of the intermediates iridodial, iridotrial or 7-deoxyloganic acid, strictosidine was indeed produced (combination 10 in Fig. 8d).

DiscussionWhereas feeding experiments clearly indicated that secologanin is derived from geraniol26, the exact sequence of intermediates and enzymes leading to the formation of its penultimate precursor loganic acid was still obscure. We report here four novel enzymes that together with two previously reported enzymes12,13 ll the existing gaps and thereby provide a full description of the core (seco)iridoid pathway. Geraniol is converted to secologanin by the sequential action of four different cytochrome P450 enzymes, two different oxidoreductases, one glucosyltransferase and one methyltransferase. The missing enzymes and corresponding genes were identied by a combination of transcriptomic and proteomic approaches exploiting the current knowledge of the spatiotemporal regulation of the secoiridoid pathway.

Our results also address longstanding questions in the eld. First, as previously proposed for oxidoreductase proteins puried from Rauwola serpentina and Nepeta racemosa27,28, 8-HGO catalyses two successive and reversible oxidation steps for the formation of 8-oxogeranial. Therefore, two enzymes can contribute to the formation of the intermediate 8-oxogeraniol. G8O was recently shown to also produce 8-oxogeraniol from geraniol21, thus G8O and 8-HGO appear to catalyse partially overlapping (and in the case of G8O, monodirectional) oxidation reactions that result in the production of 8-oxogeraniol from 8-hydroxygeraniol (Fig. 1). Second, a single cytochrome P450 enzyme IO/CYP76A26 can convert cis-trans-iridodial and cis-trans-nepetalactol into 7-deoxyloganetic acid without the release of an iridotrial intermediate. Third, our data demonstrate that 7-deoxyloganetic acid glycosylation precedes its further oxygenation by 7-DLH. This answers the longstanding question which intermediate is transferred from IPAP cells to the epidermis. The expression of 7-DLH in IPAP cells indicates that loganic acid is the mobile intermediate transferred to the epidermis, and hence that glycosylation by 7-DLGT is not sufcient for mobility but that further hydroxylation by 7-DLH is also required. The tissue-specic expression of pathway sections may increase the ux through the pathway by alleviating feedback inhibition by intermediates and products, and/or it may segregate

NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606

200

100

100

Gene combination1 PaGPPS+VoGES2 PaGPPS+VoGES+G8O3 PaGPPS+VoGES+G8O+8-HGO4 PaGPPS+VoGES+G8O+8-HGO+IS5 PaGPPS+VoGES+G8O+8-HGO+IS+IO6 PaGPPS+VoGES+G8O+8-HGO+IS+IO+7-DLGT7 PaGPPS+G8O+8-HGO+IS+IO+7-DLGT (no VoGES)8 Empty vector9 IO+7-DLGT+7DLH+LAMT+SLS10 IO+7-DLGT+7-DLH+LAMT+SLS+TDC+STR

6 m/z=401

EV + 7-DLA m/z=401

7 m/z=401

6 m/z=359

EV + 7-DLA m/z=359

EV + CGA m/z=359

3 2

1

4 5

6

0

8

7

150 150

50

0

50

Peak intensity (CPS)

200

100

0

1,600

800

0

200

100

0 3,200

1,600

0 3,200

1,600

0 3,200

1,600

Acetyl-7-DLA

Acetyl-7-DLA

Hex-CGA

Strictosidine

10 m/z=575

10 + iridodial* m/z=575

EV + strictosidine m/z=575

EV + secologanin m/z=433

Peak intensity (CPS)

2,400

1,200

0

2,400

1,200

0

2,400

1,200

0

2,000

1,000

0

2,000

1,000

0

2,000

1,000

Strictosidine

10 + iridodial* m/z=433

9 + iridodial* m/z=433

hex-CGA

7-DLA

7-DLA

0 10 18

16

14

12 22

20 Time (min)

Hex-CGA

Secologanin

Secologanin

0 10 18

16

14

12 22

20 Time (min)

Figure 8 | Reconstitution of the strictosidine pathway in N. benthamiana. (a) Gene combinations inltrated in leaves in triplicate. (b) Principal component analysis. PC1 and PC2 describe 36.2 and 31.1% of the total mass variation, respectively. (c) LC-MS analysis showing selected masses 401 and 359 representing (acetylated) 7-deoxyloganic acid (7-DLA) from inltrations with 8-carboxygeranic acid (CGA), 7-DLA or gene combinations 6 or7 (negative control). The two peaks likely represent 7-DLA acetylated at two different positions in the glucose moiety. (d) LC-MS analysis showing selected masses 433 (formic acid adduct of secologanin) and 575 (formic acid adduct of strictosidine) from inltrations with secologanin or strictosidine, or with gene combinations 9 or 10, with or without iridodial. *Identical proles with iridotrial or 7-DLH. Hex hexosyl; CPS counts per second.

the iridoid and secoiridoid/MIA pathways, allowing them to fulll tissue-specic functions.

Pathway reconstitution experiments in a heterologous host validated the enzyme sequence leading to the formation of strictosidine (Fig. 1). We however observed a low ux through the pathway in the leaves of N. benthamiana. Previous experiments carried out with the upstream segment of the pathway indicated substantial conversion of intermediate products to other oxidized, acetylated, malonylated and glycosylated derivatives by endogenous N. benthamiana enzymes21. Despite downstream channeling of intermediates upon reconstruction of the full pathway, competing conversions are still observed and are probably responsible for the low ux. This would, for example, explain why we cannot see the product of IO upon inltration of pathway combination 5 (Fig. 8; Supplementary Fig. 12). Nevertheless, we know that IO is functional because inltration of combination 9 or 10 with iridodial did result in production of secologanin or strictosidine, respectively (Fig. 8d). Side reactions could possibly be eliminated by further metabolic engineering. The side products obtained in our experiments may also provide valuable information on the enzyme networks naturally present inN. benthamiana and could be leads for combinatorial biochemistry programmes to produce chemical diversity and novel bioactive compounds.

In conclusion, we have identied the four missing enzymes in the secologanin pathway in C. roseus. In combination with the genes previously identied, the genes encoding these four enzymes are sufcient to engineer secologanin production and, together with TDC and STR, biosynthesis of the complex alkaloid strictosidine in the heterologous plant N. benthamiana. Although different segments of the strictosidine pathway are localized in different cell types in C. roseus, our results show that the entire pathway can be successfully reconstituted in a singleN. benthamiana organ. This paves the way for the biotechnological production of valuable (seco)iridoids and derived compounds, such as the MIAs vincristine and vinblastine, making these important anticancer drugs available to more people and at a lower price.

Methods

Chemicals. The substrates 8-carboxygeranial, 8-carboxygeranic acid, 8-oxogeranic acid and 8-hydroxygeranic acid were synthesized on order by Synthelor (Vandoeuvre-Ls-Nancy, France), whereas 8-OH-geraniol, 8-oxogeraniol, 8-OH-geranial and 8-oxogeranial were synthesized by Chiralix B.V. (Nijmegen, Netherlands). Iridodial-glucoside, iridotrial-glucoside and 7-deoxyloganic acid were synthesized from aucubin extracted from Aucuba japonica leaves by Chiralix B.V. as described29,30. The aglucone iridoid pathway intermediates were produced by incubation with almond b-glucosidase (Sigma-Aldrich) in 50 mM acetate buffer (pH 5). Loganetic acid and loganetin were produced by the deglucosylation of

8 NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606 ARTICLE

loganic acid and loganin (Extrasynthese). Aglycones were extracted with diethyl ether, evaporated under N2 and quantied by 1H-NMR (1H-nuclear magnetic resonance).

Transcriptomic analysis. Transgenic derivatives of C. roseus cell line MP183L (overexpressing the ORCA transcription factors) were generated by particle bombardment31 with derivatives of the pER8 plasmid32 carrying either the ORCA2 or ORCA3 open reading frames (ORFs). Selected transgenic lines were treated for 24 h with 10 mM estradiol, and RNA was isolated as described16. Illumina

HiSeq2000 RNA sequencing, assembly, annotation and mapping of the RNA-Seq reads on the reference transcriptome was carried out as described16. Complete linkage hierarchical cluster analysis was achieved using the CLUSTER and TREEVIEW software33 and the log10 transformed values of the normalized FPKM values were used as input for CLUSTER.

Proteomic analysis. For the plant material of C. roseus var. little Bright Eyes seeds were sown in sterilized soil and covered with transparent plastic until germination. The soil was kept humid. The plants were fertilized weekly with 0.2% liquid Wuxal (29 g l 1 nitrate-N; 46 g l 1 ammonium-N; 25 g l 1 carbamide-N; 100 g l 1 P2O5 total phosphate; 75 g l 1 K2O; 124 mg l 1 B; 50 mg l 1 Cu; 248 mg l 1 Fe; Cu, Fe,

Mn and Zn as EDTA-chelate). The plants were re-potted twice during further growth. For the isolation of epidermal protoplasts, young, light-green C. roseus leaves (length 47 cm) from side branches (without buds or owers) of 8- to 11-week-old plants were harvested for protoplast isolation. The mid-vein was removed and the leaves were cut into 1- to 2-mm strips. Protoplasts were isolated as described34. To collect epidermal protoplasts, a layer of 1-ml betaine solution(0.5 M betaine, 1 mM CaCl2, 10 mM MES, pH 5.6 with KOH) was added on top of the tubes containing the protoplast mix. After centrifugation for 7 min at 1,500 r.p.m. and 4 C, epidermal protoplasts were collected from the upper interphase. The suspension was mixed with 4-ml protoplast solution and 25% Percoll (pH 6), and overlaid with 1.5 ml of betaine solution for a second purication step. The tubes were centrifuged at 700 r.p.m. for 30 min and the epidermal protoplasts were again collected from the upper interphase. Protoplasts were pelleted in the betaine solution. Isolation of mesophyll protoplasts was performed as described for the epidermal protoplasts with replacement of the MCP solution with TEX solution (3.2 g l 1 Gamborgs B5 medium, 3.1 mM NH4NO3,5.1 mM CaCl2, 2.6 mM MES, 0.4 M sucrose, pH 5.65.8 with 0.5 M KOH), and the betaine solution was replaced with mannitol/W5 (1 mM D-glucose, 30 mM NaCl, 25 mM CaCl2, 1 mM KCl, 0.3 mM MES, 320 mM a-mannitol, pH 5.65.8 with

KOH). For the rst gradient, 510% Percoll (pH 6) was added instead of 35%. The protoplast pellet was resuspended in 24 ml buffer A (20 mM HEPES pH 7.2,1 tablet per 10 ml Roche Protease Inhibitor, 1 mM PMSF). The mixture was pressed 1020 times through a syringe with needle and centrifuged for 10 min at3,000 r.p.m. and 4 C. The samples were ultra-centrifuged for 1 h at 30,000 r.p.m. The membrane pellets were dissolved in 1 ml washing solution (0.3 M NaCI,20 mM HEPES-KOH pH 7.2, 1 tablet per 10 ml Roche Protease Inhibitor, 1 mM PMSF) by stirring with a small brush, and then vortexed for 30 s. The microsomes were frozen at 80 C for at least 1 h, thawed and then centrifuged for 1 h at 30,000 r.p.m. and 4 C. The pellets were dissolved in 50100 ml buffer A for further analysis. The protein concentration was determined using the Bio-Rad DC protein assay according to the manufacturers instructions.

For mass spectrometry analysis, B50 mg of protein was separated by one-dimensional gel electrophoresis35 in 10% polyacrylamide gels to reduce sample complexity. The gels were stained with Coomassie Brilliant Blue, the lanes were cut into 10 slices, the proteins were reduced with tris(carboxyethyl)phosphine hydrochloride and sulhydryl groups were blocked with iodoacetamide. In-gel digestion with sequencing-grade-modied trypsin (Promega V5111) was carried out overnight at 37 C36. The resulting peptides were recovered by adding 40 mM Tris-HCl (pH 8.0) and 50% acetonitrile/1% formic acid. The peptide mixtures were desalted by solid-phase extraction on C18 reversed-phase columns and analysed on an LTQ Orbitrap mass spectrometer (Thermo Fischer Scientic, Bremen, Germany) coupled to an Eksigent Nano HPLC system (Eksigent Technologies, Dublin, CA, USA) as previously described37. For protein identication, databases were searched using Mascot v2.3. Raw data were searched against a composite database consisting of all entries in the NCBI Viridiplantae database (released in November 2010), all publicly available C. roseus-expressed sequence tags (downloaded from NCBI in November 2010) and the reference transcriptome released on CathaCyc and ORCAE16 (database contained forward and reverse protein entries, total number of protein entries 1,166,013). The parameters for precursor and fragment ion mass tolerance were set to 5 p.p.m. and 0.8 Da, respectively. One missed trypsin cleavage was allowed. Carboxyamidomethylation of cysteine was specied as xed modication, and oxidation of methionine and pyroglutamate formation from glutamine were selected as variable modications. Scaffold v3.0 (Proteome Software, Portland, OR, USA) was used to validate and quantify MS/MS-based peptide and protein identications. Peptide identications were accepted if they were established at 495% probability as specied by the

Peptide Prophet algorithm38. Protein identications were accepted if they were established at 490% probability and at least one peptide was uniquely assigned to a corresponding protein in a minimum of two of our samples. Protein probability was assigned by the Protein Prophet algorithm39. Proteins that were identied with

the same set of peptides and could not be differentiated by the MS/MS analysis were grouped to protein clusters to satisfy the principles of parsimony. Peptide and protein false-discovery rates (FDR) were determined by the Scaffold software.

A peptide FDR of 0.01% and a protein FDR of 0.2% were computed. The Scaffold software was also used to determine protein abundance in the mesophyll and epidermal factions based on the number of spectra assigned to each protein.

An F-test was applied to assess signicant differences in protein abundances.

Gene isolation. ORFs were amplied by PCR from a pACT2 cDNA library of aC. roseus cell culture elicited with yeast extract40 using the primers listed in Supplementary Table 4. For expression in plants, the ORFs were transferred to vector pRT101 (ref. 41) to bring them under the control of the Cauliower Mosaic Virus 35S promoter, and the expression cassettes were then transferred to the binary vector pCAMBIA1300 (Cambia). For expression in E. coli, the ORFs were transferred to vector pASK-IBA45plus (IBA) and/or pET16-H (Novagen pET-16b derivative). Probes for in situ hybridization were prepared from the same ORFs cloned in pBluescript II SK . For localization analysis, the ORFs were transferred

to vector pTH2 (refs 42,43) and/or pTH2BN (a derivative of pTH2). The marker for nucleocytosolic localization (pRT101-mCherry) was prepared by amplifying the mCherry ORF from plasmid ER-rk44 (The Arabidopsis Information Resources, TAIR, clone CD3-959).

Isolation of His-tagged recombinant proteins. Recombinant proteins carrying a His6 tag were expressed using plasmid pASK-IBA45plus and/or pET16-H in E. coli strain BL21 (DE3) pLysS and were puried using Ni-NTA agarose chromatography (Qiagen). For quality analysis, the recombinant proteins were separated by 12.5% (w/v) SDSPAGE, transferred to Protran nitrocellulose membranes (Whatman) by semidry electroblotting, and western blots were probed with mouse monoclonal anti-His horseradish peroxidase-conjugated antibodies (5Prime). Antibody binding was detected by incubation in 250 mM sodium luminol, 0.1 M TrisHCl (pH 8.6), 3 mM H2O2 and 67 mM p-coumaric acid, followed by exposure to an X-ray lm.

Enzymatic assays of UGT and oxidoreductases. UGT activity was detected in0.1-ml reaction buffer containing 50 mM potassium phosphate (pH 7.5), 2 mM UDP-glucose, 51,000 mM 7-deoxyloganetic acid or 2 mM of other tested compounds and 501,000 ng of puried enzyme. Reactions were incubated at 32 C for 15 min and stopped by adding 1 volume of methanol, mixed by vortexing and kept on ice for 10 min. The tubes were centrifuged at 4,000 g for 10 min, and the supernatants were passed through 0.22-mm nylon lters.

Oxidoreductase activity was detected in 1-ml reaction buffer containing 50 mM bis-tris propane (pH 9 for oxidation and pH 7.5 for reduction), 21,000 mM 8-OH-geraniol, 8-oxogeraniol, 8-OH-geranial, 8-oxogeranial or other tested compounds, 22,000 mM NAD or NADH and 501,000 ng of puried enzyme.

Reactions were incubated for 15 min at 32 C, stopped by adding 0.2 volumes of 1 M sodium citrate (pH 3) and centrifuged and ltered as above. Quantitative assays were carried out by measuring NADH production at 340 nm in a Nanodrop2000c (Thermoscientic)

Chromatographic analysis of 8-HGO and 7-DLGT products. Identication of 8-HGO enzyme products was performed using capillary gas chromatography mass spectrometry. The ethyl acetate extract of the reaction mixture was separated on a Agilent GC 7890A series equipped with a 5975C MSD and DB-5 capillary column (30 m 0.25 mm, lm thickness of 0.25 mm; JpW Scientic). Helium gas was used

as a carrier at a ow rate of 1.2 ml min 1. The separation conditions were as follows: split mode 1:5, injection volume 5 ml, injector temperature 230 C, initial oven temperature 60 C, and then linear gradient to 100 C at a rate of20 C min 1 followed by a linear gradient to 160 C at a rate of 2 C min 1 (run time 32 min). Analysis of 7-DLGT products was carried out using an Agilent series 1200 HPLC with a diode array detector and a Polymer Laboratories PL-ELS 2100 ICE evaporative light scattering detector and a Phenomenex Luna 5 micron 150 4.6-mm C18 column. The injection volume was 10 or 100 ml. The binary

solvent system consisted of acetonitrile and 0.1% triuoroacetic acid in water. The elution program was as follows: 5 min isocratic 10% acetonitrile and then 25 min gradient until 95% acetonitrile. Peak areas were calculated using Agilent ChemStation.

NMR. Structures of enzyme products were analysed by NMR spectroscopy in 750 ml of acetone-d6 or methanol-d4 in a 5-mm NMR glass tube45.

Subcellular localization studies. C. roseus MP183L cell suspension cultures were maintained by weekly 10-fold dilution in 50 ml Linsmaier & Skoog (LS) medium containing 88 mM sucrose, 2.7 mM 1-NAA and 0.23 mM kinetin (LS-13). The cells were grown at 25 C with an 18/6-h lightdark cycle. The cells were bombarded28 using the plasmids pTH2-IO, pTH2-7-DLH, pTH2-7-DLGT, pTH2BN-7-DLGT, pTH2-8-HGO and pTH2BN-8-HGO. The rst two were combined with equal amounts of the ER marker ER-rk, and the others with the nucleocytosolic marker mCherry44. Bombarded cells were placed on Petri dishes with LS-13 medium and viewed after 24 h using a Zeiss Observer laser scanning microscope.

NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2014 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606

In situ hybridization. pBluescript plasmid derivatives containing the cDNAs for 8-HGO, IO, 7-DLGT and 7-DLH were used for the synthesis of antisense and sense digoxigenin-labelled riboprobes as previously described19. G8O antisense probes19 and SGD antisense probes46 were used as internal phloem-associated parenchyma and epidermis markers, respectively. Parafn-embedded serial longitudinal sections of young developing leaves were hybridized with digoxigenin-labelled riboprobes and localized with antidigoxigenin-alkaline phosphatase-conjugated antibodies47.

P450 expression in yeast and enzyme assays. P450-coding sequences were amplied from pRT101 source vectors using specic primers to introduce USER sites at the ends of each sequence. The genes were subsequently transferred to the plasmid pYeDP60 using the USER cloning technique (New England Biolabs, Ipswich, UK)48. The resulting recombinant plasmids were introduced intoS. cerevisiae strain WAT11, cultivated at 28 C and P450 expression was induced as described21. Cells were harvested by centrifugation and manually broken with0.45-mm glass beads in 50 mM TrisHCl buffer (pH 7.5) containing 1 mM EDTA and 600 mM sorbitol. The homogenate was centrifuged for 10 min at 10,000 g and the supernatant was centrifuged for 1 h at 30,000 g. The pellet, comprising microsomal membranes, was resuspended in 50 mM TrisHCl (pH 7.4), 1 mM EDTA and 30% (v/v) glycerol with a Potter-Elvehjem homogenizer and storedat 20 C. All procedures for microsomal preparation were carried out at 04 C. P450 expression was evaluated as described49 and enzymatic activities were determined in a standard 0.1 ml assay comprising for IO 2.3 nmol cytochrome P450, 0.6 mM NADPH and substrate in 20 mM sodium phosphate (pH 7.4). The reaction was initiated by the addition of NADPH and was stopped after 5 min by the addition of 10 ml 1 M HCl. Iridoids were extracted in 1 ml ethyl acetate, and the organic phase was concentrated to 200 ml before GC-FID analysis on a Varian 3900 gas chromatograph (Agilent Technologies) equipped with a ame ionization detector and a DB-5 column (30 m, 0.25 mm, 0.25 mm; Agilent Technologies)

with splitless injection, at 250 C injector temperature, and with a temperature programme of 0.5 min at 50 C, 10 C min 1 to 320 C, and 5 min at 320 C.

For 7-DLH, 10 ml of yeast microsomes expressing 7-DLH (130 mg of microsomal protein) were incubated for 20 min at 27 C, in 0.1 ml of 20 mM Na-phosphate (pH 7.4) containing 0.6 mM NADPH and substrate. The reaction was initiated by the addition of NADPH and was stopped after 20 min on ice. After addition of 50 ml of 50% acetic acid, tubes were vortexed and centrifuged. The supernatant was run on reverse-phase HPLC (Alliance 2695 Waters system, NOVA-PAK C184.6 250 mm column) with photo-diode array detection at 236.5 nm. Peak areas of

the product(s) were used to calculate the catalytic parameters of each enzyme.

Leaf disc assays. Five-week-old N. benthamiana leaves were inltrated withA. tumefaciens transformed with vector pCAMBIA1300 containing the relevant candidate genes, plus the helper plasmid p19. Five days post inltration, leaves were used in a leaf disc assay as previously described21.

Pathway reconstruction in N. benthamiana. The pathway genes were transiently expressed in the leaves of ve-week-old N. benthamiana plants by agroinltration as previously described25. Briey, bacteria carrying the relevant expression constructs (PaGPPS, VoGES, G8O, 8-HGO, IS, IO, 7-DLGT, 7-DLH, LAMT, SLS, TDC, STR, empty vector or TBSV p19 (ref. 50)) were grown individually at 28 C for 24 h. Cells were harvested by centrifugation and then resuspended in inltration buffer containing 10 mM MES (Duchefa Biochemie), 10 mM MgCl2 and 100 mM acetosyringone (40-hydroxy-30,50-dimethoxyacetophenone, Sigma) to a nal OD600 of B0.5. For all gene combinations that compared subsequent steps in the pathway, the amounts of cell suspension for each expression construct were kept constant by adding the corresponding amount of A. tumefaciens carrying an empty vector. All inltrations were performed in three replicates. In several experiments, pathway intermediates were injected into the same leaves 3 days after agroinltration. Compounds used for inltration were diluted to a nal concentration of 400 mM in methanol/water (1:19), with the same ratio of methanol/water alone as a negative control and 400 mM 7-deoxyloganic acid, 8-carboxygeranic acid, secologanin or strictosidine as positive controls. Leaves were harvested for metabolite analysis 5 days after agroinltration. Frozen, powderedN. benthamiana leaves (200 mg aliquots) were extracted in 0.6 ml 99.867% methanol, 0.133% formic acid and 5 ml of the extract was analysed using a Waters

Alliance 2795 HPLC connected to a QTOF Ultima V4.00.00 mass spectrometer (Waters, MS Technologies, UK). Measurements were taken in negative ionization mode. LC-MS data were acquired using MassLynx 4.0 (Waters) and processed using MetAlign version 1.0. The normalized and log-transformed data matrix was used for principal component analysis implemented in GeneMath XT v 2.1. ANOVA was used to evaluate the statistical signicance of differences in metabolite levels between all treatments.

References

1. Dinda, B., Debnath, S. & Harigaya, Y. Naturally occurring iridoids. A review, part 1. Chem. Pharm. Bull. (Tokyo) 55, 159222 (2007).

2. Dinda, B., Debnath, S. & Harigaya, Y. Naturally occurring secoiridoids and bioactivity of naturally occurring iridoids and secoiridoids. A review, part 2. Chem. Pharm. Bull. (Tokyo) 55, 689728 (2007).

3. Tundis, R., Loizzo, M. R., Menichini, F., Statti, G. A. & Menichini, F. Biological and pharmacological activities of iridoids: Recent developments. Mini Rev. Med. Chem. 8, 399420 (2008).

4. Viljoen, A., Mncwangi, N. & Vermaak, I. Anti-inammatory iridoids of botanical origin. Curr. Med. Chem. 19, 21042127 (2012).

5. Birkett, M. A. & Pickett, J. A. Aphid sex pheromones: from discovery to commercial production. Phytochemistry 62, 651656 (2003).

6. Birkett, M. A., Hassanali, A., Hoglund, S., Pettersson, J. & Pickett, J. A. Repellent activity of catmint, Nepeta cataria, and iridoid nepetalactone isomers against Afro-tropical mosquitoes, ixodid ticks and red poultry mites. Phytochemistry 72, 109114 (2011).

7. Oudin, A. et al. Spatial distribution and hormonal regulation of gene products from methyl erythritol phosphate and monoterpene-secoiridoid pathways in Catharanthus roseus. Plant Mol. Biol. 65, 1330 (2007).

8. Loyola-Vargas, V. M., Galaz-valos, R. M. & K-Cauich, R. Catharanthus biosynthetic enzymes: the road ahead. Phytochem. Rev. 6, 307339 (2007).

9. Collu, G. et al. Geraniol 10-hydroxylase, a cytochrome P450 enzyme involved in terpenoid indole alkaloid biosynthesis. FEBS Lett. 508, 215220 (2001).

10. Irmler, S. et al. Indole alkaloid biosynthesis in Catharanthus roseus: new enzyme activities and identication of cytochrome P450CYP72A1 as secologanin synthase. Plant J. 24, 797804 (2000).

11. Murata, J., Roepke, J., Gordon, H. & De Luca, V. The leaf epidermome of Catharanthus roseus reveals its biochemical specialization. Plant Cell 20, 524542 (2008).

12. Geu-Flores, F. et al. An alternative route to cyclic terpenes by reductive cyclization in iridoid biosynthesis. Nature 492, 138142 (2012).

13. Simkin, A. J. et al. Characterization of the plastidial geraniol synthase from Madagascar periwinkle which initiates the monoterpenoid branch of the alkaloid pathway in internal phloem associated parenchyma. Phytochemistry 85, 3643 (2013).

14. Mahroug, S., Burlat, V. & Saint-Pierre, B. Cellular and sub-cellular organisation of the monoterpenoid indole alkaloid pathway in Catharanthus roseus. Phytochem. Rev. 6, 363381 (2007).

15. Verma, P., Mathur, A. K., Srivastava, A. & Mathur, A. Emerging trends in research on spatial and temporal organization of terpenoid indole alkaloid pathway in Catharanthus roseus: a literature update. Protoplasma 249, 255268 (2012).

16. Van Moerkercke, A. et al. CathaCyc, a metabolic pathway database built from Catharanthus roseus RNA-seq data. Plant Cell Physiol. 54, 673685 (2013).

17. van der Fits, L. & Memelink, J. ORCA3, a jasmonate-responsive transcriptional regulator of plant primary and secondary metabolism. Science 289, 295297 (2000).

18. Guirimand, G. et al. The subcellular organization of strictosidine biosynthesis in Catharanthus roseus epidermis highlights several trans-tonoplast translocations of intermediate metabolites. FEBS J. 278, 749763 (2011).

19. Burlat, V., Oudin, A., Courtois, M., Rideau, M. & St-Pierre, B. Co-expression of three MEP pathway genes and geraniol 10-hydroxylase in internal phloem parenchyma of Catharanthus roseus implicates multicellular translocation of intermediates during the biosynthesis of monoterpene indole alkaloids and isoprenoid-derived primary metabolites. Plant J. 38, 131141 (2004).

20. Pompon, D., Louerat, B., Bronine, A. & Urban, P. Yeast expression of animal and plant P450s in optimized redox environments. Cytochrome P450, Pt B 272, 5164 (1996).

21. Hfer, R. et al. Geraniol hydroxylase and hydroxygeraniol oxidase activities of the CYP76 family of cytochrome P450 enzymes and potential for engineering the early (seco)iridoid pathway. Metab. Eng. 08, 001 (2013).

22. Morrone, D., Chen, X., Coates, R. M. & Peters, R. J. Characterization of the kaurene oxidase CYP701A3, a multifunctional cytochrome P450 from gibberellin biosynthesis. Biochem. J. 431, 337344 (2010).

23. Guirimand, G. et al. Spatial organization of the vindoline biosynthetic pathway in Catharanthus roseus. J. Plant Physiol. 168, 549557 (2011).

24. Schmidt, A. et al. A bifunctional geranyl and geranylgeranyl diphosphate synthase is involved in terpene oleoresin formation in Picea abies. Plant Physiol. 152, 639655 (2010).

25. Dong, L. et al. Characterization of two geraniol synthases from Valeriana ofcinalis and Lippia dulcis: similar activity but difference in subcellular localization. Metab. Eng. 20, 198211 (2013).

26. Uesato, S., Kanomi, S., Iida, A., Inouye, H. & Zenk, M. H. Mechanism for iridane skeleton formation in the biosynthesis of secologanin and indole alkaloids in Lonicera tatarica, Catharanthus roseus and suspension cultures of Rauwola serpentina. Phytochemistry 25, 839842 (1986).

27. Ikeda, H. et al. Acyclic monoterpene primary alcohol:NADP oxidoreductase of Rauwola serpentina cells: the key enzyme in biosynthesis of monoterpene alcohols. J. Biochem. 109, 341347 (1991).

10 NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2014 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms4606 ARTICLE

28. Hallahan, D. L. et al. Purication and characterization of an acyclic monoterpene primary alcohol:NADP oxidoreductase from catmint (Nepeta racemosa). Arch. Biochem. Biophys. 318, 105112 (1995).

29. Jensen, S. R., Kirk, O. & Nielsen, B. J. Application of the vilsmeier formylation in the synthesis of 11-C-13 labeled iridoids. Tetrahedron 43, 19491954 (1987).

30. Jensen, S. R., Kirk, O. & Nielsen, B. J. Biosynthesis of the iridoid glucoside cornin in Verbena ofcinalis. Phytochemistry 28, 97105 (1989).

31. van der Fits, L. & Memelink, J. Comparison of the activities of CaMV 35S and FMV 34S promoter derivatives in Catharanthus roseus cells transiently and stably transformed by particle bombardment. Plant Mol. Biol. 33, 943946 (1997).

32. Zuo, J., Niu, Q. W. & Chua, N. H. Technical advance: an estrogen receptor-based transactivator XVE mediates highly inducible gene expression in transgenic plants. Plant J. 24, 265273 (2000).

33. Eisen, M. B., Spellman, P. T., Brown, P. O. & Botstein, D. Cluster analysis and display of genome-wide expression patterns. Proc. Natl Acad. Sci. USA 95, 1486314868 (1998).

34. Tohge, T. et al. Toward the storage metabolome: proling the barley vacuole. Plant Physiol. 157, 14691482 (2011).

35. Laemmli, U. K. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680685 (1970).

36. Shevchenko, A., Wilm, M., Vorm, O. & Mann, M. Mass spectrometric sequencing of proteins silver-stained polyacrylamide gels. Anal. Chem. 8, 850858 (1996).

37. Schneider, T. et al. A proteomics approach to investigate the process of Zn hyperaccumulation in Noccaea caerulescens (J & C. Presl) F.K. Meyer. Plant J 73, 131142 (2013).

38. Keller, A., Nesvizhskii, A. I., Kolker, E. & Aebersold, R. Empirical statistical model to estimate the accuracy of peptide identications made by MS/MS and database search. Anal. Chem. 74, 53835392 (2002).

39. Nesvizhskii, A. I., Keller, A., Kolker, E. & Aebersold, R. A statistical model for identifying proteins by tandem mass spectrometry. Anal. Chem. 75, 46464658 (2003).

40. Menke, F. L. H., Champion, A., Kijne, J. W. & Memelink, J. A novel jasmonate- and elicitor-responsive element in the periwinkle secondary metabolite biosynthetic gene Str interacts with a jasmonate- and elicitor-inducible AP2-domain transcription factor, ORCA2. EMBO J. 18, 44554463 (1999).

41. Tpfer, R., Matzeit, V., Gronenborn, B., Schell, J. & Steinbiss, H. H. A set of plant expression vectors for transcriptional and translational fusions. Nucleic Acids Res. 15, 5890 (1987).

42. Chiu, W. et al. Engineered GFP as a vital reporter in plants. Curr. Biol. 6, 325330 (1996).

43. Niwa, Y., Hirano, T., Yoshimoto, K., Shimizu, M. & Kobayashi, H. Non-invasive quantitative detection and applications of non-toxic, S65T-type green uorescent protein in living plants. Plant J. 18, 455463 (1999).

44. Nelson, B. K., Cai, X. & Nebenfuhr, A. A multicolored set of in vivo organelle markers for co-localization studies in Arabidopsis and other plants. Plant J. 51, 11261136 (2007).

45. Kim, H. K., Choi, Y. H. & Verpoorte, R. NMR-based metabolomic analysis of plants. Nat. Protoc. 5, 536549 (2010).

46. Guirimand, G. et al. Strictosidine activation in Apocynaceae: towards a nuclear time bomb?. BMC. Plant Biol. 10, 182 (2010).

47. Mahroug, S., Courdavault, V., Thiersault, M., St-Pierre, B. & Burlat, V. Epidermis is a pivotal site of at least four secondary metabolic pathways in Catharanthus roseus aerial organs. Planta 223, 11911200 (2006).

48. Nour-Eldin, H. H., Hansen, B. G., Norholm, M. H., Jensen, J. K. & Halkier, B. A. Advancing uracil-excision based cloning towards

an ideal technique for cloning PCR fragments. Nucleic Acids Res. 34, e122 (2006).49. Omura, T. & Sato, R. The carbon monoxide-binding pigment of liver microsomes. i. evidence for its hemoprotein nature. J. Biol. Chem. 239, 23702378 (1964).

50. Voinnet, O., Rivas, S., Mestre, P. & Baulcombe, D. An enhanced transient expression system in plants based on suppression of gene silencing by the p19 protein of tomato bushy stunt virus. Plant J. 33, 949956 (2003).

Acknowledgements

We are grateful to Richard Twyman for critical reading of the manuscript and to Sren Rosendal Jensen for advice on iridoid substrate synthesis. David Nelson is acknowledged for naming P450 enzymes. The research leading to these results has received funding from the European Union Seventh Framework Programme FP7/2007-2013 under grant agreement number 222716SMARTCELL. T.I. received funding from the People Programme (Marie Curie Actions) of the European Unions 7th Framework Programme (FP7/20072013) under REA Grant Agreement 289217.

Author contributions

J.M., A.G., K.-M.O.-C., R.V., E.M., H.B., N.N. and D.W. conceived the project and general strategy; A.G. and J.P. performed the expression analysis and gene selection. K.M. isolated the candidate genes and prepared most vectors, and expressed and functionally characterized 8-HGO and 7DLGT, performed subcellular localization experiments and parts of the synthesis of iridoid substrates; L.D. did and analysedthe pathway reconstruction in N. benthamiana; N.N. assisted by T.I. expressed and functionally characterized the candidate P450 genes; T.S. assisted by L.W. did the proteomics analysis; V.B. carried out the in situ hybridizations; N.N. and R.L. contributed to metabolite analysis; D.W.-R., J.M., A.G., S.vd K. and H.B. wrote the manuscript; D.W.-R. acted as coordinator.

Additional information

Accession Codes: Transcriptomic data have been deposited in the GenBank Sequence Read Archive (SRA) under the accession code SRP026417. Gene sequences have been deposited in GenBank nucleotide core database under the accession codes KF302066 (IO), KF302067 (7DLH), KF302068 (7DLGT), KF302069 (8-HGO), KF302070 (CYP81Q32), KF302071 (CYP81C13), KF302072 (CYP71AY2), KF302073 (CYP71BT1), KF302074 (CYP72A225), KF302075 (CYP76T24), KF302076 (CYP71D2), KF302077 (ADH9), KF302078 (ADH13), KF302079 (ADH14), KF302080 (UGT9) and KF309243 (CYP71AY1).

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Miettinen, K. et al. The seco-iridoid pathway from Catharanthus roseus. Nat. Commun. 5:3606 doi: 10.1038/ncomms4606 (2014).

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/3.0/

Web End =http://creativecommon http://creativecommons.org/licenses/by-nc-nd/3.0/

Web End =s.org/licenses/by-nc-nd/3.0/

NATURE COMMUNICATIONS | 5:3606 | DOI: 10.1038/ncomms4606 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

& 2014 Macmillan Publishers Limited. All rights reserved.