Opium poppy and Madagascar periwinkle: model non-model systems to investigate alkaloid biosynthesis in plants


*(fax +1 403 289 9311; e-mail pfacchin@ucalgary.ca).


Alkaloids represent a large and diverse group of compounds that are related by the occurrence of a nitrogen atom within a heterocyclic backbone. Unlike other types of secondary metabolites, the various structural categories of alkaloids are unrelated in terms of biosynthesis and evolution. Although the biology of each group is unique, common patterns have become apparent. Opium poppy (Papaver somniferum), which produces several benzylisoquinoline alkaloids, and Madagascar periwinkle (Catharanthus roseus), which accumulates an array of monoterpenoid indole alkaloids, have emerged as the premier organisms used to study plant alkaloid metabolism. The status of these species as model systems results from decades of research on the chemistry, enzymology and molecular biology responsible for the biosynthesis of valuable pharmaceutical alkaloids. Opium poppy remains the only commercial source for morphine, codeine and semi-synthetic analgesics, such as oxycodone, derived from thebaine. Catharanthus roseus is the only source for the anti-cancer drugs vinblastine and vincristine. Impressive collections of cDNAs encoding biosynthetic enzymes and regulatory proteins involved in the formation of benzylisoquinoline and monoterpenoid indole alkaloids are now available, and the rate of gene discovery has accelerated with the application of genomics. Such tools have allowed the establishment of models that describe the complex cell biology of alkaloid metabolism in these important medicinal plants. A suite of biotechnological resources, including genetic transformation protocols, has allowed the application of metabolic engineering to modify the alkaloid content of these and related species. An overview of recent progress on benzylisoquinoline and monoterpenoid indole alkaloid biosynthesis in opium poppy and C. roseus is presented.


Alkaloids are a group of nitrogen-containing natural products that are found in approximately 20% of plant species. Many of the approximately 12 000 plant alkaloids for which structures have been elucidated exhibit potent biological activity. Unlike other types of secondary metabolites, the numerous and diverse alkaloid structural types have independent biosynthetic origins. A discussion of general alkaloid metabolism in plants reveals several interesting patterns, but specific details regarding the biochemistry, molecular and cell biology of alkaloid biosynthesis are unique to each category. Over the past several decades, two alkaloid-producing plants have emerged as model systems based on the number of characterized cDNAs encoding biosynthetic enzymes and regulatory proteins. With the establishment of cellular localization models and the application of genomics technologies, Papaver somniferum (opium poppy) and Catharanthus roseus (Madagascar periwinkle) have emerged as key species in the study of alkaloid metabolism in plants.

Opium poppy accumulates a number of benzylisoquinoline alkaloids (BIA), a term that collectively describes approximately 2500 elucidated natural product structures found mainly in the Papaveraceae, Ranunculaceae, Berberidaceae and Menispermaceae. The most prominent compounds are the narcotic analgesic morphine, the cough suppressant codeine, the muscle relaxant papaverine, and the anti-microbial agents sanguinarine and berberine. Molecular clones encoding at least 20 BIA biosynthetic enzymes have been isolated, and the pace of gene discovery will undoubtedly increase with the extensive application of new technologies. Catharanthus roseus is a valuable and important medicinal plant that produces complex anti-tumor drugs that are effective against a variety of cancers (van der Heijden et al., 2004). The monoterpenoid indole alkaloids (MIAs) are one of the largest and most diverse groups of plant secondary metabolites (approximately 3000 structures), occurring mainly in the Apocynaceae, Loganiaceae and Rubiaceae. The complex chemistry of MIAs is related to powerful biological activities, and several drugs have been developed, including reserpine for the treatment of neurological disorders, vinblastine and vincristine as anti-neoplastic agents, and yohimbine as a vasodilator.

The biosynthesis (Facchini, 2006) and metabolic engineering (Larkin and Harrigan, 2007; Sato et al., 2007) of BIAs have recently been reviewed. Several reviews have also been published on the analysis (Hisiger and Jolicoeur, 2007), chemistry (O’Connor and Maresh, 2006), biosynthesis (El-Sayed and Verpoorte, 2007; Hedhili et al., 2007; Loyola-Vargas et al., 2007; Oudin et al., 2007a), regulation (Memelink and Gantet, 2007), intra-cellular, inter-cellular and organ-specific compartmentation (Mahroug et al., 2007), plant cell culture-based production (Zhao and Verpoorte, 2007), biotechnological production (Zárate and Verpoorte, 2007), functional genomics (Goossens and Rischer, 2007) and transport (Roytrakul and Verpoorte, 2007) of MIAs in C. roseus. Together, these reviews reflect the profusion of scientific achievements that highlight opium poppy and C. roseus as two of the most thoroughly investigated medicinal plants. This review outlines recent developments involving the application of a wide array of biochemical, molecular, cellular and genomic techniques have allowed opium poppy and C. roseus to emerge as model non-model systems for the investigation of alkaloid biosynthesis in plants.

Benzylisoquinoline alkaloid metabolism

Genes for the biosynthesis of simple BIAs

Fundamentally, BIA biosynthesis involves the condensation of two tyrosine derivatives, and begins with the decarboxylation of tyrosine or dihydroxyphenylalanine (DOPA) to yield tyramine and dopamine, respectively. Tyrosine decarboxylase (TYDC) is encoded by a large gene family with about 15 members in opium poppy (Facchini and De Luca, 1994). Dopamine is the precursor for the isoquinoline moiety, and 4-hydroxyphenylacetaldehyde (4-HPAA), the deamination product of tyramine, is incorporated into the benzyl component (Figure 1). Dopamine and 4-HPAA are coupled by a Pictet–Spengler condensation catalyzed by norcoclaurine synthase (NCS), the first committed step in the pathway. The enzyme was purified from Thalictrum flavum, and corresponding cDNAs have been isolated and functionally characterized from T. flavum and opium poppy (Liscombe et al., 2005; Samanani and Facchini, 2002; Samanani et al., 2004). NCS is related to the pathogenesis-related protein pathogenesis-related 10 (PR10) and Bet v 1 allergen protein families. However, homologous PR10 proteins from opium poppy and other plants are not catalytically active (Liscombe et al., 2005). Biochemical characterization of recombinant T. flavum NCS using a continuous enzyme assay based on circular dichroism spectroscopy following the generation of one enantiomer – (S)-norcoclaurine – revealed a reaction mechanism involving a two-step cyclization with a direct electrophilic aromatic substitution (Luk et al., 2007). Recently, a second enzyme that also produces only (S)-norcoclaurine, but shows sequence similarity to 2-oxoglutarate-dependent dioxygenases was isolated from Coptis japonica (Minami et al., 2007). However, this enzyme does not possess a 2-oxoglutarate-binding domain and requires ferrous ions, rather than 2-oxoglutarate or oxygen, for activity. It is remarkable that two proteins belonging to different protein families can catalyze the same reaction in vitro, and the relative participation of each enzyme in BIA biosynthesis in vivo remains to be determined.

Figure 1.

 Biosynthesis of BIAs including morphine, sanguinarine and berberine.
Enzymes for which corresponding cDNAs have been isolated are shown in red. The key branch-point intermediate (S)-reticuline is highlighted in yellow. Abbreviations: 4′OMT, 3′-hydroxy-N-methylcoclaurine 4′-O-methyltransferase; 6OMT, norcoclaurine 6-O-methyltransferase; 7OMT, reticuline 7-O-methyltransferase; BBE, berberine bridge enzyme; CFS, cheilanthifoline synthase; CNMT, coclaurine N-methyltransferase; COR, codeinone reductase; Cyp719A1, canadine synthase; Cyp719A2, stylopine synthase; Cyp719B1, salutaridine synthase; Cyp80A1, berbamunine synthase; Cyp80B1, N-methylcoclaurine 3′-hydroxylase; DBOX, dihydrobenzophenanthridine oxidase; DRR, 1,2-dehydroreticuline reductase; DRS, 1,2-dehydroreticuline synthase; MSH, N-methylstylopine 14-hydroxylase; NCS, norcoclaurine synthase; P6H, protopine 6-hydroxylase; SalAT, salutaridinol 7-O-acetyltransferase; SOMT, scoulerine 9-O-methyltransferase; SalR, salutaridine:NADPH 7-oxidoreductase; STOX, (S)-tetrahydroxyprotoberberine oxidase; TNMT, tetrahydroprotoberberine cis-N-methyltransferase; TYDC, tyrosine decarboxylase.

The conversion of (S)-norcoclaurine to (S)-reticuline involves an O-methylation at position 6, an N-methylation, a 3′-hydroxylation and a second 4′-O-methylation (Figure 1). The two O-methyltransferases – norcoclaurine 6-O-methyltransferase (6OMT) and 3′-hydroxy-N-methylcoclaurine 4′-O-methytransferase (4′OMT) – display strict regiospecificity, and cDNAs for each have been obtained from opium poppy and Coptis japonica (Morishige et al., 2000; Ounaroon et al., 2003; Ziegler et al., 2005). Cognate cDNAs encoding coclaurine N-methyltransferase (CNMT) have been isolated from opium poppy and Coptis japonica and are more similar to S-adenosyl-L-methionine (SAM)-dependent cyclopropane fatty acid synthases than to other N-methyltransferases (Choi et al., 2002; Facchini and Park, 2003). The hydroxylation of N-methylcoclaurine is catalyzed by a P450 mono-oxygenase of the Cyp80B sub-family (Huang and Kutchan, 2000; Ikezawa et al., 2003; Pauli and Kutchan, 1998; Samanani et al., 2005).

Most BIA structural types are derived from the central pathway intermediate (S)-reticuline (Figure 1). Dimeric bisbenzylisoquinoline alkaloids such as berbamunine, which result from the regio- and stereo-selective oxidative carbon–oxygen phenol coupling of N-methylcoclaurine by Cyp80A1, are an exception (Kraus and Kutchan, 1995). The recent isolation of a cDNA from opium poppy encoding a 7-O-methyltransferase specific for norreticuline (N7OMT), but not N-methylated analogs, suggests that precursors for (S)-reticuline might also enter the pathway to form the N-methylated BIA papaverine (J. Ziegler, University of Calgary, personal communication). A cDNA for an O-methyltransferase acting on (S)-reticuline and catalyzing the formation of laudanine was previously isolated from opium poppy (Ounaroon et al., 2003). Interestingly, (R,S)-reticuline 7-O-methyltransferase (7OMT) does not accept N-demethylated BIAs, but is active towards phenolic compounds.

Genes for the biosynthesis of protoberberine alkaloids

Several branch pathways leading to many BIA structural types begin with the formation of (S)-scoulerine by the berberine bridge enzyme (BBE) (Figure 1). BBE cDNAs have been isolated from several sources (Dittrich and Kutchan, 1991; Facchini et al., 1996; Samanani et al., 2005), and the enzyme has been thoroughly characterized (Winkler et al., 2006, 2007). BBE belongs to a novel family of flavoproteins with two covalent binding sites – one histidine and one cysteine – for flavin adenine dinucleotide (FAD). The mid-point redox potential of BBE is higher than that of other flavoproteins due to cysteinylation of the co-factor, and results in hydride abstraction of (S)-reticuline, the first half-reaction toward the conversion to (S)-scoulerine. The biosynthesis of benzophenanthridine alkaloids, such as sanguinarine, is initiated by the formation of two methylenedioxy bridges, resulting in (S)-stylopine (Figure 1). Both reactions are catalyzed by P450-dependent mono-oxygenases, and cDNAs encoding stylopine synthase (Cyp719A2 and Cyp719A3) have been isolated from Eschscholzia californica (Ikezawa et al., 2007). Both recombinant proteins showed the same region-specificity for methylenedioxy bridge formation, but Cyp719A2 only converts (S)-cheilanthifoline to (S)-stylopine, whereas Cyp719A3 also accepts compounds without an existing methylenedioxy bridge. (S)-Stylopine is then converted to (S)-cis-N-methylstylopine by tetrahydroprotoberberine N-methyltransferase (TNMT). A cDNA encoding TNMT was isolated and functionally characterized from opium poppy based on homology to CNMT (Liscombe and Facchini, 2007). TNMT is one of only a few plant enzymes that are able to catalyze the formation of quaternary ammonium compounds, and converts only protoberberine alkaloids with dimethoxy or methylenedioxy functional groups at positions 2/3 and 9/10. Two subsequent P450-dependent hydroxylations yield dihydrosanguinarine, which is oxidized to sanguinarine. However, cDNAs for these final three enzymes have not yet been isolated.

Alternatively, (S)-scoulerine can be methylated to (S)-tetrahydrocolumbamine by scoulerine 9-O-methyltransferase (SOMT) (Figure 1) (Takeshita et al., 1995). Tetrahydrocolumbamine is converted to columbamine, which is methylated by the protoberberine-specific enzyme columbamine O-methyltransferase (CoOMT) to yield palmatine (Morishige et al., 2002). Canadine synthase (Cyp719A1) catalyzes the formation of a methylenedioxy bridge to form canadine from tetrahydrocolumbamine. The inability of Cyp719A1 to accept columbamine as a substrate precludes the involvement of this compound in berberine biosynthesis. (S)-Canadine is ultimately oxidized to berberine.

Genes for the biosynthesis of morphinan alkaloids

Conversion of (S)-reticuline to its (R) epimer is required to initiate the morphinan alkaloid biosynthetic pathway (Figure 1). The epimerization of (S)-reticuline is a two-step process, involving the oxidation of (S)-reticuline by 1,2-dehydroreticuline synthase, and the reduction of 1,2-dehydroreticuline to (R)-reticuline (De-Eknamkul and Zenk, 1992; Hirata et al., 2004). However, cDNAs for these enzymes have not yet been identified. Intramolecular carbon–carbon phenol coupling between the C2 of the benzyl moiety and the C4 of the isoquinoline moiety leads to salutaridine (Gerardy and Zenk, 1993a). A cDNA encoding salutaridine synthase (Cyp719B1) was recently isolated from opium poppy (J. Ziegler, University of Calgary, personal communication). The protein shows high homology to the methylenedioxybridge-forming P450-dependent enzymes. Salutaridine reductase (SalR) converts salutaridine to salutaridinol, and study of a cDNA for SalR isolated from opium poppy has shown that the enzyme belongs to the family of short-chain dehydrogenases/reductases (SDR). The next step in the pathway is catalyzed by salutaridinol 7-O-acetyltransferase (SalAT), which specifically acetylates the 7(S) epimer of salutaridinol to salutaridinol-7-O-acetate. The SalAT cDNA shows considerable homology to the BAHD family of acetyltransferases (Grothe et al., 2001). Depending on pH, the introduced acetyl group can be spontaneously eliminated, leading to the formation of an oxide bridge between C4 and C5 to yield thebaine, the first pentacyclic morphinan alkaloid. The final steps in morphine biosynthesis consist of two demethylations and one reduction. The demethylations are not yet understood and no enzymes capable of catalyzing either reaction have been detected. However, cDNAs for codeinone reductase (COR), which belongs to the aldo-keto reductase family, have been identified in opium poppy (Unterlinner et al., 1999).

Monoterpenoid indole alkaloid metabolism

Genes for the MEP and mevalonic acid pathways

Monoterpenoid indole alkaloids are condensation products of a nitrogen-containing indole moiety derived from tryptamine and a monoterpenoid component derived from the iridoid glucoside secologanin (Figure 2). Incorporation experiments with isotopically labelled glucose in C. roseus cell cultures showed that the initial steps for biosynthesis of the monoterpenoid backbone proceed via the 2-C-methyl-d-erythritol-4-phosphate (MEP) pathway (Contin et al., 1998). Four separate cDNAs involved in the MEP pathway – encoding deoxyxylulose 5-phosphate synthase (DXS) (Chahed et al., 2000), deoxyxylulose 5-phosphate reductoisomerase (DXR) (Veau et al., 2000), methylerythritol 2,4-diphosphate synthase (MECS) (Veau et al., 2000) and hydroxymethylbutenyl diphosphate synthase (HDS) (Oudin et al., 2007b) – have been isolated. However, a cDNA for only one mevalonic acid (MVA) pathway enzyme, 3-hydroxy-3-methyl-glutaryl (HMG) CoA reductase (Maldonado-Mendoza et al., 1992), has been reported. Recent sequencing efforts in cDNA libraries derived from leaf epidermis-enriched mRNA have identified 4-diphosphocytidyl-2-C-methyl-d-erythritol synthase (CMS), four MVA pathway genes [3-hydroxy-3-methylglutaryl CoA reductase (HMGR), 3-ketoacyl CoA thiolase (AACT1), acetoacetyl CoA thiolase and HMG CoA synthase (HMGS)] and three genes common to the MEP/MVA pathways [isopentenyl pyrophosphate isomerase (IPPI), farnesyl pyrophosphate synthase (GPPS) and geranyl pyrophosphate synthase (GPS)] (Murata et al., 2008).

Figure 2.

 Biosynthesis of MIAs including vinblastine, hörhammericine and ajmaline.
Enzymes for which corresponding cDNAs have been isolated are shown in red. The key branch-point intermediate dehydrogeissoschizine is highlighted in yellow. Abbreviations: 16OMT, 16-hydroxytabersonine-16-O-methyltransferase; AAE, acetylajmalan esterase; ANAMT, acetylnorajmalan methyltransferase; Cyp71D12, tabersonine 16-hydroxylase; Cyp72A1, secologanin synthase; Cyp76B6, geraniol 10-hydroxylase; D4H, desacetoxyvindoline 4-hydroxylase; DAT, deacetylvindoline 4-O-acetyltransferase; DHVR, dihydrovomilenine reductase; LAMT, loganic acid methyltransferase; MAT, minovincinine acetyltransferase; NAMT, norajmalan methyltransferase; NMT, N-methyltransferase; PER, peroxidase; PNAE, polyneuridine aldehyde esterase; SBE, sarpagan bridge enzyme; SGD, strictosidine β-d-glucosidase; STR, strictosidine synthase; TDC, tryptophan decarboxylase; VH, vinorine hydroxylase; VR, vomilenine reductase; VS, vinorine synthase. Although not shown in the figure, NADPH cytochrome c reductase (CPR) is involved in the Cyp-catalyzed reactions.

Genes for the biosynthesis of secologanin

Four cDNAs involved in secologanin biosynthesis and encoding P450-dependent geraniol 10-hydroxylase (Cyp76B6) (Collu et al., 2001), acyclic monoterpene primary alcohol dehydrogenase (ADH) (Ikeda et al., 1991), loganic acid methyltransferase (LAMT) (Murata et al., 2008) and P450-dependent secologanin synthase (Cyp72A1) (Irmler et al., 2000) have been isolated. Of particular interest is the recent substrate specificity and kinetic analysis of LAMT (Murata et al., 2007), which identified loganic acid as the preferred iridoid substrate in C. roseus, suggesting that O-methylation occurs after (Madyastha et al., 1973) rather than before (Yamamoto et al., 1998) hydroxylation of 7-deoxyloganic acid in this species (Figure 2). Candidate cDNAs encoding 10-oxogeranial cyclase, and several P450 mono-oxygenases potentially involved in the conversion of iridodial to loganic acid, were also identified (Murata et al., 2007).

Monoterpenoid indole alkaloid biosynthesis in C. roseus

Strictosidine, the common precursor to all MIAs, is formed by a Pictet–Spengler condensation of tryptamine and secologanin (Figure 2). Tryptophan decarboxylase (TDC) (De Luca et al., 1989) converts tryptophan to tryptamine, which is then coupled to secologanin by strictosidine synthase (STR) (Kutchan et al., 1988; McKnight et al., 1990) to form strictosidine, the central intermediate leading to all MIAs (Ma et al., 2006). C. roseus accumulates at least 130 MIAs, which are detected in small quantities in various organs and in specialized cell cultures (van der Heijden et al., 2004). Strictosidine β-glucosidase (SGD) is a remarkable strictosidine-specific enzyme that yields a highly reactive ring-opened dialdehyde intermediate that is involved in the biosynthesis of MIAs with corynantheine, iboga and aspidosperma chemical backbones (Geerlings et al., 2000). However, little is known about the mechanisms or enzymes involved in directing the formation of each type. The branch pathways leading from the branch-point intermediate dehydrogeissoschizine to tabersonine and catharanthine remain uncharacterized.

In contrast, the biochemistry of the six-step enzymatic conversion of tabersonine to vindoline has been studied extensively (Loyola-Vargas et al., 2007). The sequence of reactions involves the ordered conversion of tabersonine to 16-hydroxytabersonine by cytochrome P450-dependent tabersonine-16-hydroxylase (Cyp71D12) (Schröder et al., 1999), followed by 16-hydroxytabersonine-16-O-methyltransferase (16OMT) (Levac et al., 2008). The substrate specificity and kinetic properties of 16OMT suggest it has a high affinity for both 16-hydroxytabersonine and SAM. As 16OMT does not accept 16-hydroxy-2,3-dihydrotabersonine as a substrate, O-methylation might occur prior to substitution of the 2,3 position on the molecule (Levac et al., 2008). The conversion of 16-methoxytabersonine to 16-methoxy-2,3-dihydrotabersonine occurs via an uncharacterized oxidation. The subsequent step involves a chloroplast thylakoid membrane-associated N-methyltransferase to yield desacetoxyvindoline. The two terminal reactions in vindoline biosynthesis involve a light-regulated desacetoxyvindoline 4-hydroxylase (D4H) (Vazquez-Flota et al., 1997) and deacetylvindoline 4-O-acetyltransferase (DAT) (St-Pierre et al., 1998). Detection of 16-demethoxyvindoline in C. roseus suggests that the last four reactions in vindoline biosynthesis can transform tabersonine as well as 16-methoxytabersonine into their respective end products.

In contrast to C. roseus leaves and stems, which produce vindoline, underground tissues accumulate other tabersonine derivatives including lochnericine and hörhammericine (Figure 2). An early step in this pathway involves tabersonine 6,7-epoxidase, which catalyzes the NADPH- and molecular oxygen-dependent conversion of tabersonine to lochnericine. This P450 mono-oxygenase was associated with microsomes of transformed C. roseus hairy root cultures (Rodriguez et al., 2003). A cDNA for the root tip-specific minovincinine-O-acetyltransferase (MAT) involved in the biosynthesis of 6,7-dehydroechitovenine and/or 19-O-acetylhörhammericine has been functionally characterized (Laflamme et al., 2001).

Monoterpenoid indole alkaloid biosynthesis in Rauvolfia serpentina

The biosynthesis of ajmaline in cell cultures of Rauvolfia serpentina involves ten well-characterized enzymatic steps (Figure 2) (Ruppert et al., 2005). In addition, cDNAs have been isolated for six genes involved in general MIA and ajmaline metabolism in R. serpentina– STR, SGD, polyneuridine aldehyde esterase (PNAE), vinorine synthase (VS), cytochrome P450 reductase (CPR) and acetylajmalan acetylesterase (AAE) (Ruppert et al., 2005). The three-dimensional structures for several of these enzymes have been elucidated, which has provided insights into the reaction mechanisms involved in MIA biosynthesis (Stöckigt et al., 2007).

The role of ‘omics’ technologies to study plant alkaloid biosynthesis

The Arabidopsis (Arabidopsis thaliana) genome sequence has been available since 2000, and projects targeting other model species including rice, corn, soybean, poplar and the legume Medicago truncatula are at various stages of completion. However, genome sequence databases are not yet available for plants that produce complex alkaloids. Nevertheless, application of high-throughput technologies such as genome sequencing, expressed sequence tag (EST) and DNA microarray analyses, proteomics and metabolomics has recently accelerated the discovery of new genes and biological processes involved in plant secondary metabolism.

Expressed sequence tags

Compared with model plants such as Arabidopsis, rice and tomato, the number of ESTs for alkaloid-producing plants is relatively low. An exception is tobacco, for which more than one million ESTs are publicly available due to the importance of this plant as a model for investigation of plant–pathogen and plant–virus interactions. As alkaloid metabolism is so diverse and few research groups work on the same plant species, large-scale EST projects are limited. However, extensive EST databases are now available for opium poppy (and several related BIA-producing plants) and for C. roseus. The release of a database consisting of 4500 ESTs derived from the laticifers of morphine-accumulating opium poppy was reported in 2001 (Pilatzke-Wunderlich and Nessler, 2001). Further efforts with this plant have yielded a total of 25 000 ESTs from seedlings, stems, roots and elicitor-treated cell cultures (Ziegler et al., 2005, 2006; Zulak et al., 2007). A recent floral genome project contributed 11 000 ESTs from Eschscholzia californica (Carlson et al., 2006), and the release of a Coptis japonica database has also been reported (Inui et al., 2007; Kato et al., 2007). NAPGEN (Natural Products Genomic Research), a consortium of Canadian investigators, has recently established a database of over 400 000 ESTs from a wide variety of plants that accumulate potentially health-related bioactive metabolites. Within this project, more than 46 000 ESTs were reported from cell cultures for eight BIA-accumulating species of four plant families.

Over 25 000 C. roseus ESTs have been produced by partial sequencing of cDNA libraries prepared from whole organs and tissues (Murata et al., 2006; Rischer et al., 2006; Shukla et al., 2006). Previously, only 372 C. roseus sequences were available in public databases. Despite the high MIA biosynthetic activity of the selected C. roseus tissues (Laflamme et al., 2001; St-Pierre et al., 1999), no known transcripts involved in MIA biosynthesis, except for those encoding Cyp76B6 and SGD, were represented in the clones obtained from young leaf, root tip and other cDNA libraries (Murata et al., 2006; Rischer et al., 2006; Shukla et al., 2006). The absence of MIA biosynthesis genes is consistent with results obtained in cDNA/amplified fragment length polymorphism studies conducted with induced C. roseus cell cultures (Rischer et al., 2006). Together, these results suggest that gene transcripts relevant to MIA metabolism are rare in induced cell cultures or in young leaves/root tips compared with transcripts encoding cellular components that are not related to MIA biosynthesis. The isolation of leaf epidermis mRNA by carborundum abrasion facilitated the production of a representative cDNA library, and more than 10 000 unique ESTs were identified (Murata et al., 2008 and unpublished results). Remarkably the leaf epidermis-enriched cDNA library contained virtually all known MIA biosynthesis and regulatory genes (Murata et al., 2008).

Transcriptome analysis

The earliest attempt to identify preferentially expressed genes in alkaloid-producing plants involved differential screening by subtractive hybridization of low and normal alkaloid-producing lines of tobacco to isolate and characterize putrescine N-methyltransferase, an early step in nicotine biosynthesis, (Hibi et al., 1994). Several other developments first pioneered in tobacco (Goossens and Rischer, 2007) are now being applied to opium poppy, C. roseus and other medicinal plants. Recently, a cDNA/amplified fragment length polymorphism method was used to profile cell cultures in terms of metabolites and transcripts at various times after jasmonate (JA) treatment (Rischer et al., 2006). The analysis for MIA pathway genes included the identification of a number of MEP pathway genes (DXS, HDS, MECS), as well as Cyp76B6, which commit geraniol to the secoiridoid pathway, and SGD, which converts strictosidine to its aglycone. Other putative MIA pathway genes were identified, but require functional characterization. The metabolite and transcript profiles were integrated, which yielded some intriguing correlations, but the results are equivocal without more accurate tools to select candidate MIA biosynthetic or regulatory genes for further study. More recent efforts included small-scale transcriptome analyses using cDNA libraries produced from C. roseus mRNA extracted from leaves and roots, respectively (Shukla et al., 2006). Random small-scale sequencing of leaf, root and subtracted cDNAs did not reveal new information about MIA biosynthesis.

DNA macroarray analysis of differential gene expression among morphine-containing opium poppy plants and eight other Papaver species that do not accumulate morphine produced a set of cDNAs with higher expression in opium poppy, one of which encoded 4′OMT (Ziegler et al., 2005) and another encoded SalR (Ziegler et al., 2006). DNA microarray analysis of the top1 mutant variety of opium poppy, which accumulates thebaine and oripavine rather than codeine and morphine, revealed ten down-regulated transcripts, although none appeared to encode proteins responsible for the high-thebaine, low-morphine phenotype (Millgate et al., 2004). Large-scale transcriptome and metabolite profiling has been performed in opium poppy cell cultures treated with a fungal elicitor (Zulak et al., 2007), and in C. roseus and tobacco cell cultures treated with methyl JA (Goossens et al., 2003; Rischer et al., 2006). In all cases, a coordinated increase was observed in the expression of genes implicated in alkaloid metabolism. Moreover, profound changes in the level of gene transcripts encoding primary metabolic enzymes also occurred. Interestingly, transcripts involved in SAM recycling increased in all three systems, indicating a high demand for this co-factor in the modification of alkaloid backbone structures.

Proteomics and metabolomics

Proteomics has been used to identify proteins found in the exuded latex of opium poppy (Decker et al., 2000), and subsequently to characterize 6OMT and 7OMT within the protein population (Ounaroon et al., 2003). A proteomics approach was also undertaken to identify novel proteins involved in alkaloid biosynthesis during the growth cycle of C. roseus cell cultures (Jacobs et al., 2005). Eighty-eight protein spots were selected for identification by MALDI-MS/MS, of which 58 were identified, including two STR isoforms and tryptophan synthase. These studies demonstrate that proteomics has the potential to become a valuable approach once protein databases are established.

Several pioneering metabolomic analyses have recently been performed on MIA-producing plant species, including C. roseus (Choi et al., 2004) and Strychnos spp. (Frederich et al., 2004). Comprehensive metabolic profiling of various phytoplasma-treated C. roseus leaves by NMR spectroscopy followed by principle component analysis identified compounds present at different levels in infected versus healthy tissues (Choi et al., 2004). Phytoplasma infection caused an increase in phenylpropanoid and MIA pathway intermediates and products, including chlorogenic acid, loganic acid, secologanin and vindoline. Another metabolomic study was conducted using liquid chromatography coupled with photodiode array detection and ESI-MS on methanol extracts from Ophiorrhiza pumila, Nothapodytes foetida and Camptotheca acuminata to compare the profiles of camptothecin-related alkaloids being produced (Yamazaki et al., 2003). Considerable differences were detected in the constituents of the various plants, hairy roots and cell cultures.

Compartmentalization and trafficking

Although the biosynthetic pathways leading to BIAs, MIAs and other alkaloid types are of polyphyletic origin, some intriguing patterns are apparent, including the involvement of multiple cell types for alkaloid biosynthesis and/or accumulation, the targeting of pathway enzymes to multiple subcellular compartments, and the possible ubiquity of multi-enzyme complexes. Such complex organization implies extensive intra- and inter-cellular transport of pathway intermediates, products and even biosynthetic enzymes. Alkaloids generally accumulate in specific cell types due to their cytotoxicity and probable role in plant defense responses. For example, alkaloids are sequestered to isolated idioblasts and laticifers in C. roseus (St-Pierre et al., 1999), root endodermis and stem cortex/pith in Thalictrum flavum (Samanani et al., 2005), and laticifers in opium poppy (Bird et al., 2003). The cellular localization of alkaloid biosynthesis is remarkably diverse and complex. Work on the cellular and developmental complexities and organization of alkaloid biosynthesis in opium poppy and C. roseus, in particular, has established new paradigms in the cell biology of secondary metabolism.

A tale of three cell types in opium poppy

In all organs of opium poppy, BIA accumulation occurs in the articulated laticifers found adjacent or proximal to sieve elements of the phloem (Figure 3) (Facchini and De Luca, 1995). The cytoplasm of laticifers (latex) contains a full complement of cellular organelles and many large vesicles to which alkaloids are sequestered. Although laticifers were long considered the site of BIA biosynthesis and accumulation, the cellular localization of BIA biosynthetic enzymes and gene transcripts has shown that alkaloid synthesis involves other cell types (Bird et al., 2003; Facchini and De Luca, 1995; Weid et al., 2004). Initial in situ hybridization experiments demonstrated localization of TYDC gene transcripts to the phloem, but not to laticifers (Facchini and De Luca, 1995). The morphinan pathway enzymes Cyp719B1 and SalR were also not detected in isolated latex (Gerardy and Zenk, 1993a,b). Seven other biosynthetic enzymes (6OMT, CNMT, CYP80B1, 4′OMT, BBE, SalAT and COR) have been localized to sieve elements in opium poppy and their corresponding gene transcripts to associated companion cells (Figure 3) (Bird et al., 2003; Samanani et al., 2006). The localization of BIA metabolism to phloem cells distinct from laticifers indicates inter-cellular transport of pathway intermediates or products, and is contrary to a long-standing view in plant biology (Figure 4). Previously, sieve elements were not known to support complex metabolism, and were assumed to possess only a limited number of proteins required for cell maintenance and solute transport. The expanding physiological roles for sieve elements include the transport of information macromolecules such as RNA (Jorgensen et al., 1998), and the biosynthesis of jasmonic acid (Hause et al., 2003), ascorbic acid (Hancock et al., 2003) and defense-related compounds (Walz et al., 2004). Sieve elements clearly possess a previously unrealized biochemical potential.

Figure 3.

 Three-dimensional representation of vascular tissues in opium poppy showing the location of sieve elements (yellow), companion cells and laticifers involved in the biosynthesis and accumulation of BIAs.

Figure 4.

 BIA biosynthesis involves three cell types in the phloem of opium poppy.
(a) Laticifers are adjacent to sieve elements and companion cells. Stem tissue was fixed using potassium permanganate and sectioned for viewing by transmission electron microscopy. Abbreviations: cc, companion cell; cw, cell wall; er, endoplasmic reticulum; la, laticifer; m, mitochondrion; p, plastid; se, sieve element; ser, sieve element reticulum; v, vacuole. Bar = 10 μm.
(b) Plasmodesmata create symplastic connections between sieve elements and laticifers. Stem cross-sections were exposed to a mouse β-1,3-glucan monoclonal antibody (green, asterisks) to show plasmodesmata, and a rabbit major latex protein (MLP) polyclonal antibody (red) to identify laticifers. An obliquely oriented sieve plate (green, white arrowhead) identifies the sieve element. Bar = 10 μm.

Thalictrum flavum accumulates the anti-microbial agent berberine and other BIAs. In situ RNA hybridization analysis revealed cell type-specific expression of BIA biosynthetic genes in the roots and rhizomes of T. flavum (Samanani et al., 2005). In roots, gene transcripts for all nine enzymes were localized to the pericycle (the innermost layer of the cortex) and adjacent cortical cells. In rhizomes, all biosynthetic gene transcripts were restricted to the protoderm of leaf primordia. As in opium poppy, biosynthetic gene transcripts in T. flavum were localized to cell types that are different from those in which alkaloids accumulate. In roots, protoberberine alkaloids were restricted to endodermal cells upon initiation of secondary growth, and distributed throughout the pith and cortex in rhizomes. Thus, the cell type-specific localization of BIA biosynthesis and accumulation are temporally and spatially separated in T. flavum roots and rhizomes, respectively. Despite the close phylogenetic relationships between corresponding biosynthetic enzymes (Liscombe et al., 2005), different cell types are involved in BIA synthesis and storage in opium poppy, T. flavum and perhaps Coptis japonica.

The cellular source of isopentenyl diphosphate for secologanin biosynthesis

Two separate routes for IPP biosynthesis have been established in plants – the cytosolic MVA pathway and the plastid-localized MEP pathway (Rohmer, 1999). Strong evidence based on feeding studies with [1-13C]-glucose supports the role of the MEP pathway in supply of IPP precursors for secologanin biosynthesis in C. roseus cell cultures (Contin et al., 1998) and Ophiorrhiza pumila hairy roots (Yamazaki et al., 2004). Other studies conducted with R. serpentina cell cultures suggest that the MVA pathway might contribute up to 5% of the precursors for loganin biosynthesis, which implies that some cross-talk occurs between the two pathways. In situ RNA hybridization studies in leaf showed that gene transcripts encoding four MEP pathway enzymes and geraniol 10-hydroxylase (Cyp76B6) were primarily localized in the internal phloem-associated parenchyma (IPAP) (Burlat et al., 2004; Mahroug et al., 2006; Oudin et al., 2007b). This discovery led to the hypothesis that IPAP cells participate in MIA biosynthesis by supplying an isoprenoid pathway intermediate that is translocated to the leaf epidermis for conversion to secologanin and ultimately MIAs (Figure 5). Identification of this intermediate is essential to substantiate this hypothesis, but this will be difficult without whole-plant genetic engineering capabilities.

Figure 5.

 Three-dimensional representation of leaf tissues in Catharanthus roseus showing the location of epidermis, idioblasts, laticifers and internal phloem parenchyma (yellow) involved in the biosynthesis and accumulation of MIAs.

MIA biosynthesis involves multiple cell types in above-ground parts of C. roseus

Initial studies based on in situ RNA hybridization and immunocytochemistry established that MIA biosynthesis is developmentally regulated and compartmentalized in more than one leaf or stem cell type in C. roseus (St-Pierre et al., 1999). TDC and STR transcripts were localized to the epidermis of leaves, stems and flower buds, whereas D4H and DAT transcripts were associated with specialized laticifer and idioblast cells in these organs (Figure 5). In underground tissues, transcripts encoding TDC and STR1 (St-Pierre et al., 1999) and MAT (Laflamme et al., 2001) were localized to the protoderm and cortical cells around the root apical meristem. Neither D4H nor DAT transcripts nor gene products were detected in roots (St-Pierre et al., 1999), which is consistent with the accumulation of vindoline in aerial organs only. The involvement of leaf/stem epidermis and idioblasts/laticifers suggests inter-cellular translocation of at least one pathway intermediate in MIA metabolism (Figures 5 and 6).

Figure 6.

 Putative localization and inter-cellular trafficking of MIA and other metabolic pathways.
Catharanthus roseus leaf epidermal cells express alkaloid, MVA, triterpene, very-long-chain fatty acids (VLFA) and flavonoid biosynthetic pathways. Triterpenes and VFLAs are secreted to the waxy cuticle, whereas flavonoids accumulate in epidermal cell vacuoles. The model proposes that most of the vindoline pathway up to and including 16-methoxytabersonine is expressed in the leaf epidermis, and an undetermined intermediate is then transported to adjacent mesophyll cells, specialized idioblasts and/or laticifers, for final elaboration into vindoline. The internal phloem-associated parenchyma (IPAP) preferentially expresses the MEP pathway and Cyp76B6, which raises questions about the roles played by the epidermis and IPAP in supplying IPP for secologanin biosynthesis.

Recently, laser-capture microdissection (LCM) was used to isolate individual cell types from C. roseus leaves and examine the role of each in relation to MIA biosynthesis (Murata and De Luca, 2005). LCM produced sufficient and relatively homogeneous epidermis, idioblasts, laticifers, mesophyll and vascular tissues to allow the isolation of RNA and subsequent synthesis of cell-specific cDNAs. TDC, STR, D4H and DAT transcript levels in these cDNA populations were consistent with those determined by in situ RNA hybridization (St-Pierre et al., 1999). Preferential expression of Cyp71D12 and SGD and the detection of Cyp76B6 in leaf epidermis suggested that the entire pathway from geraniol to 16-hydroxytabersonine might be completed in this cell type (Murata and De Luca, 2005). However, the low levels of expression of Cyp76B6 in leaf epidermis compared with IPAP, together with preferential expression of the MEP pathway in these cells (Burlat et al., 2004; Mahroug et al., 2006; Oudin et al., 2007b), has raised questions about the supply of iridoid precursors for secologanin biosynthesis (Figure 6).

Carborundum abrasion has been used as a novel and complementary approach to obtain epidermis-enriched leaf extracts to measure alkaloid metabolite, enzyme activity and gene expression levels (Murata and De Luca, 2005). Use of this technique showed that levels of tabersonine and 16-methoxytabersonine were higher in leaf epidermis-enriched extracts compared with vindoline and catharanthine. Similarly, 16OMT was found predominantly in C. roseus leaf epidermis, in contrast to DAT and the N-methyltransferase involved in the vindoline pathway, which were only found in whole-leaf extracts (Murata and De Luca, 2005).

MIA pathways in leaf epidermis and cortical cells of the root apical meristem

Catharanthus roseus hairy roots accumulate tabersonine, lochnericine, hörhammericine and catharanthine, demonstrating the de novo biosynthetic potential of roots to produce these alkaloids (Magnotta et al., 2007; Morgan and Shanks, 1999). However, it has not been determined whether leaf epidermis is also involved in the biosynthesis of catharanthine, or whether this alkaloid is synthesized in roots and is then transported to leaves. However, the biosynthetic competence of C. roseus shoot cultures to produce both catharanthine and vindoline (Hirata et al., 1990) has been reported, suggesting that the leaf epidermis might support the formation of both alkaloids.

Carborundum abrasion to study the C. roseus leaf epidermone

Several approaches have been used to clone 16OMT, which catalyzes the second step in the conversion of tabersonine to vindoline. Homology-based methods (Cacacea et al., 2003; Schröder et al., 2004) and attempts to purify the protein from cultured cells were unsuccessful (Schröder et al., 2004). Use of caborundum abrasion for large-scale isolation of leaf epidermal proteins facilitated the purification of 16OMT and isolation of the cognate cDNA (Levac et al., 2008). Recombinant 16OMT shows a narrow substrate range and high affinity for 16-hydroxytabersonine. The caborundum abrasion approach also clearly demonstrates that 16OMT is predominantly expressed in C. roseus leaf epidermis. The biochemical specialization of young leaf epidermis for MIA biosynthesis (Levac et al., 2008) suggests that the remaining uncharacterized biosynthetic and regulatory genes in the MIA pathway will be found in the epidermis-enriched EST database. A homology-based cloning strategy was used to identify LAMT, the second-to-last step in secologanin biosynthesis, in the EST database (Murata et al., 2008). As expected, LAMT activity was enriched in protein extracts from leaf epidermis compared with whole-leaf extracts.

In addition to MIA biosynthesis, several MVA pathway genes and genes common to downstream pathways were identified in the EST database (Murata et al., 2008). Enrichment of the cytosolic isoprenoid and related pathways appears important for leaf epidermis-localized biosynthesis of the oleanane-type triterpenes, which are secreted into to the cuticular wax layer of C. roseus leaves (Figure 6). These triterpene alcohols are composed primarily of ursolic acid and small amounts of oleanolic acid (Usia et al., 2005), and constitute approximately 2.5% (dry weight) of the young C. roseus leaves (Murata et al., 2008). In support of this hypothesis, ESTs representing a putative squalene mono-oxygenase and β-amyrin synthase, which are required for oleanane-type triterpene biosynthesis, were identified in the leaf epidermis-enriched cDNA library (Murata et al., 2008). The specialization of leaf epidermis for triterpenoid biosynthesis and accumulation raises important issues and questions regarding the ability of the epidermis to accommodate the MEP pathway for secologanin biosynthesis. Extensive sequencing of leaf epidermis enriched C. roseus cDNA libraries showed that several steps in the MVA pathway were well-represented, whereas only a single MEP pathway step and a single G10H EST could be identified (Murata et al., 2008). These results add support to in situ hybridization studies that show preferential expression of the MEP pathway and G10H in IPAP cells (Burlat et al., 2004). However, this remains to be proven by more rigorous experimentation to establish whether the pathways for triterpenoid and the entire pathway for secoiridoid biosynthesis co-exist in the same cell type (Murata et al., 2006, 2008) or whether part of the secoiridoid pathway occurs in IPAP cells (Mahroug et al., 2007).

The biochemical capacity of C. roseus leaf epidermis for flavonoid biosynthesis was first suggested by localization of transcripts encoding phenylalanine ammonia lyase (PAL), cinnamate-4-hydroxylase (C4H) and chalcone synthase (CHS) in epidermal cells (Figure 6) (Mahroug et al., 2006). Nine putative genes encoding PAL, C4H, CHS, chalcone reductase, flavanone 3-hydroxylase, 2-hydroxyisoflavanone dehydratase, 2′-hydroxyisoflavone/dihydroflavonol reductase, quercetin 3-O-glucoside-6′-O-malonyltransferase and anthocyanidin rhamnosyl transferase were also identified in the leaf epidermis-enriched EST database (Murata et al., 2008). ESTs encoding more than 20 enzymes involved in the pathway for very-long-chain fatty acid biosynthesis were also detected in the leaf epidermis-enriched cDNA library (Figure 6) (Murata et al., 2008). These results clearly illustrate the biochemical potential of C. roseus leaf epidermis for the production of diverse secondary metabolites.

Subcellular compartmentalization of alkaloid biosynthetic enzymes

Several enzymes involved in BIA biosynthesis are associated with a subcellular compartment other than the cytosol. For the sanguinarine pathway, density gradient fractionation suggested the localization of BBE and P450 mono-oxygenases to microsomes with a density slightly higher than that of typical endoplasmic reticulum (ER) (Amann et al., 1986; Bauer and Zenk, 1991; Rueffer and Zenk, 1987; Tanahashi and Zenk, 1990). Although BBE is not an integral membrane protein, it is initially targeted to the ER and subsequently sorted to a vacuolar compartment (Bird and Facchini, 2001). NCS has also been predicted to possess a signal peptide targeting the ER (Samanani et al., 2004), and the final oxidation of alkaloids, such as dihydrosanguinarine, probably occurs in an ER-derived compartment (Amann et al., 1986). The association of BIA biosynthetic enzymes with endomembranes led to speculation that specialized alkaloid-synthesizing vesicles are present in alkaloid-producing cells (Amann et al., 1986). Recently, however, Cyp80B3, BBE and sanguinarine were co-localized with the ER in opium poppy cell cultures (Alcantara et al., 2005). As the pH optimum for BBE is approximately 8.8 (Steffens et al., 1985), sanguinarine biosynthesis is unlikely to involve the vacuole; thus, BIA metabolism could be entirely associated with the ER. Indeed, the vacuole might not even be the site of BIA accumulation in cultured cells, as previously thought. Instead, sanguinarine and related alkaloids could be secreted and bound to cell-wall components, re-absorbed, reduced to less toxic dihydrosanguinarine, and further metabolized (Weiss et al., 2006).

In addition to various cell types, enzymes involved in MIA biosynthesis in C. roseus have also been localized to at least five subcellular compartments: TDC, D4H and DAT are found in the cytosol (De Luca and Cutler, 1987), STR and the peroxidase involved in the coupling of MIA monomers were localized to the vacuole (McKnight et al., 1991; Sottomayor et al., 1998), SGD is a soluble enzyme that is purported to associate with the cytoplasmic face of the endoplasmic reticulum (Stevens et al., 1993), the P450 mono-oxygenases Cyp76B6, Cyp72A2 and Cyp71D12 are integral endomembrane proteins (Collu et al., 2001; St-Pierre and De Luca, 1995; Yamamoto et al., 2000), and the N-methyltransferase involved in vindoline biosynthesis is localized to chloroplast thylakoid membranes (Dethier and De Luca, 1993).

Enzyme complexes and metabolic channels

It is generally assumed that many metabolic enzymes interact physically or are in close proximity with other enzymes that participate in common pathways (Winkel, 2004). Theoretically, the existence of multi-enzyme complexes – also known as metabolic complexes/channels or metabolons – promote more efficient cellular metabolism. RNAi-mediated silencing of COR genes suggested the possible existence of a multi-enzyme complex in BIA metabolism (Allen et al., 2004). Although seven enzymatic steps occur between (S)-reticuline and codeinone (the substrate for COR), only (S)-reticuline accumulated at elevated levels. No intermediates between (S)-reticuline and morphine were detected. Removal of COR was suggested to disrupt a metabolic channel composed of morphinan branch pathway enzymes, resulting in the accumulation of an alkaloid intermediate produced by enzymes that are not part of the same complex. Interestingly, thebaine and oripavine, which are both intermediates upstream of substrates acted upon by COR, accumulate to high levels in some opium poppy cultivars (Millgate et al., 2004); thus, an accumulation of morphinan branch pathway intermediates was expected. An alternative interpretation not considered by Allen et al. (2004) is the possible suppression of 1,2-dehydroreticuline reductase as a side-effect of silencing COR. 1,2-Dehydroreticuline reductase is one of two enzymes involved in the epimerization of (S)-reticuline to (R)-reticuline. The potential homology between the two reductases could lead to co-silencing.

Gene regulation and signal transduction

The biosynthesis and accumulation of MIAs and BIAs are coordinately regulated by development-, tissue- and environment-specific factors (Memelink and Gantet, 2007). Early studies with opium poppy and C. roseus cell cultures showed that alkaloid biosynthesis and accumulation could be triggered by alterations to the composition of the culture medium or by treatment with specific elicitors such as JA or fungal cell-wall components (El-Sayed and Verpoorte, 2007; Facchini, 2006). Similarly, studies with hairy root cultures, organ cultures, seedlings and plant organs at various stages of development have shown that BIA and MIA metabolism are temporally and developmentally regulated.

Regulation of gene expression in MIA biosynthesis

Conserved regulatory elements in the STR promoter were used to isolate various regulatory genes that control JA- and elicitor-inducible expression of MIA biosynthesis in C. roseus cell cultures (Memelink and Gantet, 2007). A yeast one-hybrid screen using a JA- and elicitor-responsive regulatory element from the STR1 promoter was used to identify a cDNA encoding the ORCA2 (octadecanoid-responsive C. roseus APETALA-domain protein 2) transcription factor (Menke et al., 1999). In another study, T-DNA activation tagging (van der Fits et al., 2001) led to identification of the cDNA encoding ORCA3 (van der Fits and Memelink, 2000), which regulates several early steps in MIA biosynthesis. Over-expression of ORCA3 in transformed C. roseus cell cultures increased the expression of several MIA pathway genes (van der Fits and Memelink, 2000). Although the possible overlapping functions of ORCA3 and ORCA2 must still be determined, an immediate consequence of treating C. roseus cell cultures with JA is a post-translational modification of ORCA3, possibly via phosphorylation (Memelink and Gantet, 2007).

Several other transcription factors appear to participate in the elicitor-mediated activation of MIA biosynthesis in C. roseus cell cultures (Memelink and Gantet, 2007). A CrBPF1 transcription factor similar to the MYB-like factor BPF1 from parsley was shown to bind to an alternative STR1 promoter site that lies upstream of the JA- and elicitor-responsive regulatory element (van der Fits et al., 2000). CrBPF1 appears to enhance elicitor-mediated STR1 gene expression, but cannot replace ORCA3 to activate MIA biosynthesis in cell cultures and proceeds via a JA-independent signaling pathway (Menke et al., 1999). The G-box, a third regulatory element found between the BPF1 and ORCA binding sites in the STR1 promoter, was used in a yeast one-hybrid screen to isolate CrGBS (G-box binding factor) and CrMYC transcription factors (Ouwerkerk and Memelink, 1999). Transient assays suggested that these transcription factors repress STR1 gene expression, although their exact roles in MIA biosynthesis are not fully characterized (Sibéril et al., 2001). Additional yeast one-hybrid screening using another elicitor-responsive element from the TDC promoter identified three Cys2/His2-type zinc finger proteins, which appear to repress both the TDC and STR1 promoters (Pauw et al., 2004).

Transcriptional regulation of BIA biosynthesis

The recent application of transcript profiling based on DNA macroarrays constructed from ESTs has led to isolation of the first transcription factor putatively involved in the regulation of BIA metabolism. Transcripts encoding a subset of transcriptional regulators showed coordinated expression with respect to BIA biosynthetic genes in berberine-producing Coptis japonica cell cultures. One of these, a WRKY transcription factor, was shown by RNAi and over-expression analysis to specifically regulate the expression of BIA biosynthetic genes (Kato et al., 2007).

Signal transduction

A yeast-derived elicitor was shown to activate the protein phosphorylation- and calcium flux-dependent production of reactive oxygen species as a second messenger in C. roseus cell cultures (Pauw et al., 2004). However, generation of reactive oxygen species was not required for the elicitor-mediated induction of alkaloid biosynthetic genes, suggesting that the oxidative burst is coupled to activation of other defense genes. The necessity for a functional octadecanoid pathway was shown in C. roseus cells cultured in an auxin-starved medium (Gantet et al., 1998). Addition of auxin to cell cultures inhibits accumulation of ajmalicine, but stimulates the accumulation other MIAs such as tabersonine and catharanthine (Rischer et al., 2006). However, exogenous addition of MeJA to auxin-treated cells restores their ability to produce alkaloids. Alkaloid production is also reduced in auxin-starved cells treated with octadecanoid pathway inhibitors, suggesting that JAs are produced in response to auxin depletion and function to coordinate biochemical events that lead to alkaloid biosynthesis. A role for the CaaX-prenyltransferases, protein farnesyltransferase (PFT) and type I protein geranylgeranyltransferase (PGGT-I), in regulating early MIA biosynthetic genes, including those encoding enzymes involved of the MEP pathway, has been demonstrated by RNAi-mediated silencing of the corresponding genes in C. roseus cell cultures (Courdavault et al., 2005). Another component of the JA signal transduction cascade also involves post-translational events that activate an uncharacterized upstream transcription factor that up-regulates ORCA3 gene expression (Vom Endt et al., 2002). Recent studies using one-hybrid transcription factor screening with a bipartite JA-responsive element have identified an AT-hook transcription factor that may be part of the upstream components that activate ORCA3 gene expression (Vom Endt et al., 2007).

Considerable effort has been focused on events associated with the induction of BIA metabolism in E. californica cell cultures, which accumulate benzophenanthridine alkaloids such as sanguinarine in response to treatment with a fungal elicitor. Two signal transduction pathways were identified. One cascade is JA-dependent and responds to high elicitor concentrations. The other is JA-independent and is triggered by low elicitor concentrations (Färber et al., 2003). The JA-independent pathway involves Gα proteins that interact and activate phospholipase A2, which leads to a transient efflux of protons from the vacuole and subsequent activation of other cytoplasmic components (Heinze et al., 2007; Viehweger et al., 2006).

Metabolic engineering of MIA and BIA biosynthesis

Metabolic engineering is defined as the improvement of cellular activities by the manipulation of enzymatic, transport or regulatory functions using recombinant DNA technology. Several biosynthetic and regulatory genes have been used to genetically alter the production of alkaloids in opium poppy and C. roseus. Transgenic plants with altered or novel enzyme activities have also been used as a tool to study the metabolic control architecture of alkaloid pathways (Hughes et al., 2004). Procedures have been developed for the transformation of opium poppy plants (Chitty et al., 2003; Facchini et al., 2008; Park and Facchini, 2000a), root cultures (Le Flem-Bonhomme et al., 2004; Park and Facchini, 2000b) and cell cultures (Belny et al., 1997). Stable transformation protocols for C. roseus are presently restricted to roots and cell cultures (Pasquali et al., 2006; Shanks et al., 1998; Zárate and Verpoorte, 2007; Zhao and Verpoorte, 2007). Genetic transformation of opium poppy and C. roseus has provided the opportunity to alter the activity of individual enzymes of BIA biosynthesis, and to examine the consequences of such modifications on the accumulation of pathway products and intermediates.

Metabolic engineering of BIA biosynthesis

Over-expression of BBE in E. californica root cultures resulted in an increased accumulation of downstream alkaloids and decreased levels of certain amino acids (Park et al., 2003). Conversely, antisense suppression of BBE expression led to non-detectable levels of downstream metabolites and increased cellular amino acid pools, revealing the impact of perturbations in alkaloid metabolism on primary metabolism (Park et al., 2002). Interestingly, antisense suppression of BBE did not result in accumulation of its substrate (S)-reticuline or any other upstream alkaloid, suggesting that metabolic channels might be disrupted in the absence of BBE. In contrast, RNAi lines targeting BBE in E. californica cell cultures resulted in elevated levels of (S)-reticuline (Fujii et al., 2007). Furthermore, the detection of laudanine suggested that the pathway was redirected towards the O-methylation of reticuline by 7OMT. Transgenic opium poppy plants expressing an antisense BBE construct showed increased flux into the morphinan and tetrahydrobenzylisoquinoline branch pathways (Frick et al., 2004). Surprisingly, the BBE reaction product (S)-scoulerine also accumulated in the latex of these transgenic plants, although the roots showed no alterations in metabolite profile. Whether these results are due to an additional role for BBE in BIA metabolism remains unclear.

Opium poppy plants transformed with over-expressed or antisense-suppressed Cyp80B3, which encodes N-methylcoclaurine 3′-hydroxylase, also showed substantial modulations in alkaloid content, but not profile (Frick et al., 2007). Over-expression of COR1, the penultimate step in morphine biosynthesis, also resulted in opium poppy plants with increased levels of morphine, codeine, and (for unexplained reasons) thebaine, which is upstream of COR1 in the pathway (Larkin et al., 2007). Large-scale expression data suggest that over-expression of one alkaloid biosynthetic gene might cause coordinated transcriptional induction of other pathway genes. In RNAi-silenced COR1 plants, expression of other known BIA biosynthetic genes from the pathway was unaffected (Allen et al., 2004). However, for unknown reasons, the plants accumulated the far-upstream intermediate (S)-reticuline rather than codeinone, the substrate of COR. The presence of multi-enzyme complexes in morphine biosynthesis is supported by experiments involving the suppression of SalAT. RNAi-silenced SalAT plants showed accumulation of salutaridine, which is normally not abundant in opium poppy plants (Allen et al., 2007). This is surprising as salutaridinol, the substrate for SalAT, did not accumulate (Figure 1). Salutaridine might be channelled to thebaine via an enzyme complex that includes SalR and SalAT.

Constitutive expression of two O-methyltransferases from the early BIA pathway in E. californica cells suggested a rate-limiting role for 6OMT (Inui et al., 2007). Constitutive over-expression of 6OMT led to increased alkaloid content. In contrast, over-expression of 4′OMT had little effect. The E. californica cell culture used in this study appeared to lack a functional 6OMT, which might explain the strong effect of over-expressing a 6OMT transgene. Subsequent characterization of 4′OMT from E. californica revealed low 6OMT activity, suggesting a role for 4′OMT as a surrogate for 6OMT to facilitate BIA biosynthesis.

Metabolic engineering of MIA biosynthesis in C. roseus

Plant cell culture technology has been promoted as a platform for the commercial production of useful secondary metabolites. The large-scale production of shikonin (Tabata, 1996), berberine (Zhao and Verpoorte, 2007), ginseng saponins (Moyano et al., 2005) and Taxol (Tabata, 2004) are key examples. Despite these successes, the many failures have been attributed to various biological limitations resulting in low or unstable production of interesting metabolites (Zhao and Verpoorte, 2007). This has been particularly true in the case of C. roseus cell cultures, especially in terms of the inability of cultured cells to produce vindoline. As a result of the present availability of numerous MIA biosynthetic and regulatory genes, several engineering efforts to modify alkaloid metabolism in transgenic C. roseus cell and root cultures have been reported (Pasquali et al., 2006; Zárate and Verpoorte, 2007; Zhao and Verpoorte, 2007).

Transgenic C. roseus cell cultures over-expressing TDC and/or STR1 did not produce higher levels of MIAs compared with controls (Canel et al., 1998; Whitmer et al., 1998). Further experimentation suggested limited precursor availability as supplementation with loganin and tryptamine facilitated an increase in alkaloid accumulation. A number of other studies showed that the simple over-expression of individual genes in the MIA pathway was insufficient to increase stable alkaloid production in the absense of an increase in the production of organized tissues, such as hairy roots (Pasquali et al., 2006). For example, expression of ORCA3 in cell cultures activated the expression of tryptophan biosynthetic genes, as well as of the gene encoding tryptophan decarboxylase. Although tryptamine accumulation increased, the transgenic cell cultures remained limited in their ability to produce monoterpenoid precursors, such as loganin (van der Fits and Memelink, 2000). The potential of hairy root cultures for the stable production of alkaloids and other secondary metabolites has been attributed to a higher degree of genetic and biochemical stability due to increased cellular differentiation compared with cell cultures (Pasquali et al., 2006; Shanks et al., 1998; Zárate and Verpoorte, 2007; Zhao and Verpoorte, 2007).

In recent studies, C. roseus hairy root cultures expressing various forms of anthranilate synthase and/or TDC exhibited higher flux through the tryptophan branch of the MIA pathway (Hughes et al., 2004). Hairy root cultures expressing these genes under the control of constitutive or inducible promoters accumulated higher levels of tryptamine rather than alkaloids. Other attempts to alter the isoprenoid pathway by expressing a hamster 3-hydroxy-3-methylglutaryl CoA reductase gene in hairy roots generated one line that accumulated up to sevenfold higher levels of serpentine compared with controls (Ayora-Talavera et al., 2002). However, no effects on catharanthine levels were detected. Hairy root cultures over-expressing DAT resulted in several lines that accumulated up to fourfold higher levels of hörhammericine compared with controls (Magnotta et al., 2007). To explain these results, evidence was provided to show that an interaction occurring between DAT and the root-specific MAT inhibited conversion of hörhammericine to 19-O-acetylhörhammericine. Studies with inhibitors showed that MIAs turn over during the growth of hairy root cultures (Morgan and Shanks, 1999). Together these results provided a plausible explanation for the accumulation of hörhammericine, assuming that the O-acetylation step is required for turnover of tabersonine, lochnericine and hörhammericine.

The potential for enhancing the MIA content of C. roseus plants has not been analyzed in detail. Although a few reports have addressed whole-plant metabolic engineering (Zárate and Verpoorte, 2007), the use of traditional breeding to improve the yield of MIAs has not been emphasized. Analysis of a variety of C. roseus germplasms suggests that mutation breeding could represent a source of high-MIA genotypes (Dutta et al., 2005; Magnotta et al., 2006).

Metabolic engineering of substrate specificity of MIA pathway enzymes

The molecular architecture of major enzymes in the ajmaline pathway (Figure 2) has recently been reviewed (Stöckigt et al., 2007). Several refined [STR (Ma et al., 2006) and SGD (Barleben et al., 2007)] and partial [raucaffricine glucosidase (Ruppert et al., 2006), perakine reductase (Rosenthal et al., 2006) and vinorine synthase (Ma et al., 2005)] crystal structures for enzymes involved in MIA biosynthesis in R. serpentina have recently been reported (Figure 2). The information obtained from the crystal structure of STR complexed with strictosidine was used to produce the rational site-directed V208A mutation of STR. The mutant STR accepted a larger range of indoleamine and iridoid substrates (Loris et al., 2007). When mutant STR was coupled to SDG from C. roseus, a large alkaloid library for pharmacological screening and novel drug design was generated. Similarly, the STR mutant D177A increased the turnover of certain secologanin analogs (Chen et al., 2006) compared with wild-type STR (McCoy and O’Connor, 2006). The crystal structure of Rserpentina STS complexed with tryptamine (Ma et al., 2006) was also used in the rational generation of two site-directed V214M and F232L mutants able to produce several analogs of strictosidine with various substituents (F, Br or CH3OH) (Bernhardt et al., 2007). When analogs of 5-(R)-hydroxymethyl strictosidine or 10-bromostrictosidine were fed to C. roseus hairy root cultures, a number of putative modified products were isolated.


Opium poppy and C. roseus have emerged as premier model, non-model systems to investigate the biochemistry and biology of alkaloid metabolism in plants. Molecular clones encoding an impressive number of BIA and MIA biosynthetic enzymes have been isolated, and the pace of gene discovery is increasing with the application of ‘omics’ technologies. EST databases and DNA microarray chips are proven tools in these systems, and proteomics and metabolomics platforms are steadily improving. In opium poppy, functional genomic strategies have benefitted from the establishment of genetic transformation protocols, the development of new techniques such as TILLING (Comai and Henikoff, 2006), and the wealth of genetic diversity obtained through intensive classical breeding. Although transformed root cultures are useful for metabolic engineering, the full potential of functional genomics in C. roseus remains restricted by the present inability to transform and regenerate plants. The possibility of reconstituting the complex metabolic pathways leading to morphine, vinblastine and other pharmaceutical alkaloids in heterologous organisms, such as yeast and bacteria, is approaching feasibility as the number of biosynthetic genes rapidly increases.

Future research on the regulation of BIA and MIA biosynthesis should continue to target the isolation of transcription factors involved in the coordinated regulation of pathway enzymes and other metabolic components. The intra- and inter-cellular translocation of pathway intermediates and products, which might involve specific transporters or the symplastic movement of metabolites, is another important topic that requires attention. The complex subcellular organization of the biosynthetic machinery involved in BIA and MIA metabolism is apparent, but remains poorly understood. A thorough understanding of metabolic regulation – at the transcriptional, cellular and biochemical levels – is crucial to realize the biotechnological goal of rationally engineering alkaloid metabolism in opium poppy and C. roseus.


Research in the Facchini and De Luca laboratories has been funded by Discovery and Strategic Project grants from the Natural Sciences and Engineering Research Council of Canada. P.J.F. and V.D.L. also gratefully acknowledge the NAPGEN Genomics Resource Consortium established by the National Research Council–Plant Biotechnology Institute (Saskatoon, Saskatchewan, Canada). P.J.F. holds the Canada Research Chair in Plant Metabolic Processes Biotechnology. V.D.L. holds the Canada Research Chair in Plant Biochemistry and Biotechnology. We thank Jillian Hagel (University of Calgary, Canada) for assistance with the anatomical drawings and Jun Murata (Nara Institute of Technology, Japan) for providing the C. roseus leaf cross-section.