The benzylisoquinoline alkaloids of opium poppy, including the narcotic analgesics morphine and codeine, accumulate in the multinucleate cytoplasm of specialized laticifers that accompany vascular tissues throughout the plant. In mature opium poppy plants, immunofluorescence labeling using specific antibodies showed that four alkaloid biosynthetic enzymes, (S)-norcoclaurine 6-O-methyltransferase (6OMT), (S)-coclaurine N-methyltransferase (CNMT), (S)-3′-hydroxy-N-methylcoclaurine-4′-O-methyltransferase (4′OMT) and salutaridinol-7-O-acetyltransferase (SAT) were restricted to sieve elements of the phloem adjacent or proximal to laticifers. The identity of sieve elements was confirmed by (i) the specific immunogold labeling of the characteristic cytoplasm of this cell type, (ii) the co-localization of a sieve element-specific H+-ATPase with all biosynthetic enzymes and (iii) the strict association of sieve plates with immunofluorescent cells. The localization of laticifers was demonstrated antibodies specific to major latex protein (MLP), which is characteristic of this cell type. In situ hybridization using antisense RNA probes for 6OMT, CNMT, 4′OMT and SAT showed that the corresponding gene transcripts were found in the companion cell paired with each sieve element. Seven benzylisoquinoline alkaloid biosynthetic enzymes, (S)-N-methylcoclaurine 3′-hydroxylase (CYP80B1), berberine bridge enzyme, codeinone reductase, 6OMT, CNMT, 4′OMT and SAT were localized by immunofluorescence labeling to the sieve elements in the root and hypocotyl of opium poppy seedlings. The abundance of these enzymes increased rapidly between 1 and 3 days after seed germination. The localization of seven biosynthetic enzymes to the sieve elements provides strong support for the unique, cell type-specific biosynthesis of benzylisoquinoline alkaloids in the opium poppy.
Alkaloids are a large and diverse group of nitrogenous secondary metabolites found in about 20% of plant species. Although alkaloids are typically derived from amino acid precursors, metabolic pathways leading to different alkaloid groups are generally unrelated in terms of both biosynthesis and phylogeny (Facchini, 2001). A complex and diversified relationship couples cellular differentiation and the metabolism of specific alkaloid types. Putrescine N-methyltransferase (PNMT) and hyoscyamine 6β-hydroxylase catalyze the first and last steps in the biosynthesis of the tropane alkaloid scopolamine and are localized to the pericycle in the roots of Atropa belladonna and Hyoscyamus muticus (Hashimoto et al., 1991; Suzuki et al., 1999). Putrescine N-methyltransferase also catalyzes the first step in nicotine biosynthesis and has been localized to the endodermis, outer cortex, and xylem in Nicotiana sylvestris (Shoji et al., 2000, 2002). In contrast, tropinone reductase I, an intermediate enzyme in the tropane alkaloid pathway, resides in the endodermis and nearby cortical cells (Nakajima and Hashimoto, 1999); thus, intermediates of tropane alkaloid metabolism must be transported between cell types. Intercellular translocation of monoterpenoid indole alkaloid pathway intermediates was also postulated between internal phloem, epidermis, laticifers and idioblasts in the leaves of Catharanthus roseus (Burlat et al., 2004; Irmler et al., 2000; Murata and De Luca, 2005; St-Pierre et al., 1999). Pyrrolizidine alkaloids are synthesized in endodermal and cortical cells immediately adjacent to the phloem in roots of Senecio vernalis (Moll et al., 2002), or throughout the cortex in Eupatorium cannabinum roots (Anke et al., 2004). The biosynthesis and storage of acridone alkaloids are also associated with endodermis in Ruta graveolens (Junghanns et al., 1998). In Papaver somniferum, benzylisoquinoline alkaloids accumulate in specialized laticifers, but biosynthetic gene transcripts and enzymes are restricted to companion cells and sieve elements, respectively, of the phloem (Bird et al., 2003; Facchini and De Luca, 1995). In contrast, gene transcripts involved in the biosynthesis of the benzylisoquinoline alkaloid berberine are restricted to the immature endodermis and pericycle in roots, and the protoderm of leaf primordia in leaves of Thalictrum flavum (Samanani et al., 2005). Overall, a remarkable variety of cell types and subcellular compartments (Facchini and St-Pierre, 2005) have been implicated in the biosynthesis and/or accumulation of alkaloids in plants.
The benzylisoquinoline alkaloids comprise >2500 compounds with potent pharmacological activity, including the analgesic morphine, the muscle relaxant papaverine and the antimicrobial agent sanguinarine (Facchini, 2001). Benzylisoquinoline alkaloid biosynthesis begins with the conversion of l-tyrosine to dopamine and 4-hydroxyphenylacetaldehyde (4-HPAA) via a lattice of ortho-hydroxylations, deaminations and decarboxylations. Dopamine is derived from l-dopa by an aromatic amino acid decarboxylase (tyrosine/dopa decarboxylase; TYDC) condensed with 4-HPAA then condensed by norcoclaurine synthase (NCS; Samanani et al., 2004) to yield (S)-norcoclaurine, the central precursor to all benzylisoquinoline alkaloids (Figure 1). (S)-Norcoclaurine is converted to (S)-reticuline by a 6-O-methyltransferase (6OMT), a N-methyltransferase (CNMT), a P450-dependent hydroxylase (CYP80B), and a 4′-O-methyltransferase (4′OMT; Figure 1). (S)-Reticuline is a key branch-point intermediate in the biosynthesis of most benzylisoquinoline alkaloids, including those with a morphinan (e.g. morphine), benzophenanthridine (e.g. sanguinarine) or protoberberine (e.g. berberine) nucleus. (S)-Reticuline can be converted to laudanine by (R,S)-reticuline 7-O-methyltransferase (7OMT; Ounaroon et al., 2003), to (S)-scoulerine by the berberine bridge enzyme (BBE) or to 1,2-dehydroreticuline (Hirata et al., 2004). The reaction catalyzed by BBE represents the first committed step in the branch pathway leading to the benzophenanthridine alkaloid sanguinarine, which is most abundant in roots. The opium poppy genes encoding BBE and 7OMT have been isolated and characterized (Facchini, 2001; Facchini et al., 1996; Ounaroon et al., 2003).
The oxidation of (S)-reticuline to 1,2-dehydroreticuline, and subsequent reduction to (R)-reticuline, are the first committed steps in morphinan alkaloid biosynthesis (De-Eknamkul and Zenk, 1992; Hirata et al., 2004. (R)-reticuline is converted in two enzymatic steps to salutaridinol by a P450-dependent enzyme and an NADPH-dependent oxidoreductase. Acetyl coenzyme A:salutaridinol-7-O-acetyltransferase (SAT) catalyzes the conversion of salutaridinol to salutaridinol-7-O-acetate (Figure 1), which spontaneously produces thebaine. The subsequent methylation of thebaine results in the formation of either neopinone or oripavine. Neopinone spontaneously forms codeinone, which is reduced by the NADPH-dependent enzyme codeinone reductase (COR) to form codeine (Figure 1). In turn, codeine is demethylated to yield morphine. An alternate route for morphine biosynthesis involves the production of morphinone from oripavine, followed by reduction catalyzed by COR. Molecular clones encoding SAT and COR have been isolated from opium poppy, but other morphinan-specific biosynthetic enzymes are not well characterized.
Recently, an interesting debate has emerged concerning the identity of cell types involved in benzylisoquinoline alkaloid biosynthesis in opium poppy. Immunofluorescence localization using antibodies raised against 4′OMT, 7OMT, BBE and SAT was purported to show that the early stages of morphine biosynthesis occur in parenchyma cells surrounding laticifers, and that downstream enzymes (e.g. COR) are found in the laticifers themselves (Weid et al., 2004). The role of sieve elements as reported by Bird et al. (2003) was disputed. Partially in response to the suggestions of Weid et al. (2004), we report the cell type-specific localization of four enzymes (i.e. 6OMT, CNMT, 4′OMT, and SAT) and corresponding gene transcripts involved in benzylisoquinoline alkaloid biosynthesis in a commercial opium poppy cultivar used for the production of morphine. We demonstrate the presence of alkaloid biosynthetic gene transcripts in companion cells and corresponding biosynthetic enzymes in sieve elements, as shown previously for CYP80B1, BBE and COR (Bird et al., 2003). The co-localization of a sieve element-specific H+-ATPase with all biosynthetic enzymes, the strict association of sieve plates with immunofluorescent cells and the immunogold labeling of the characteristic cytoplasm of sieve elements are used to confirm the identity of the cell type. Antibodies specific for major latex protein are used to localize laticifers, which are generally adjacent or proximal to sieve elements (Thureson-Klein, 1970). We also show that the sieve element-specific biosynthesis of benzylisoquinoline alkaloids is established in opium poppy seedlings in association with the differentiation of primary vascular tissues. Our results support the co-localization of benzylisoquinoline alkaloid biosynthetic enzymes to sieve elements in all opium poppy organs throughout plant development.
Alkaloid biosynthetic enzyme and gene transcript levels in two opium poppy cultivars
Levels of benzylisoquinoline alkaloid biosynthetic enzyme and gene transcript were compared in all organs of two opium poppy cultivars (cv.) – Louisiana, used in this study, and Marianne, which was used previously (Bird et al., 2003). Under growth chamber conditions, cv. Louisiana and cv. Marianne produced similar quantities of morphine (92 ± 30 and 88 ± 20 mg ml−1 latex, respectively). Ribonucleic acid gel-blot hybridization showed mostly conserved, but some differential, aspects of alkaloid biosynthetic gene transcript accumulation (Figure 2a). In both cultivars, gene transcripts for 6OMT, CNMT, 4′OMT and SAT were detected in all organs of the plant, but were generally highest in roots and stems and lowest in leaves. The only notable difference in the accumulation of gene transcripts between the two cultivars was the higher levels of 6OMT and 4′OMT mRNA in the flower bud of cv. Marianne, compared with cv. Louisiana (Figure 2a).
Immunoblot analysis using polyclonal antibodies prepared against recombinant 6OMT, CNMT, 4′OMT and SAT demonstrated the specificity of the antibodies. One band was detected in each lane of an immunoblot containing total protein extracts from different opium poppy organs (Figure 2b). The proteins with each antiserum were consistent with the expected molecular weights of 6OMT (38.5 kDa), CNMT (41.0 kDa), 4′OMT (39.4 kDa) and SAT (52.6 kDa). All four enzymes were present in each organ of both cv. Louisiana and cv. Marianne (Figure 2b). Each protein was typically most abundant in stems, but high levels of some enzymes were also found in other organs. The only notable difference between the two cultivars was the marginally lower levels of 6OMT and CNMT in the roots and flower buds of cv. Louisiana.
Alkaloid biosynthetic enzymes occur in phloem sieve elements in mature organs
Immunofluorescence labeling using resin-embedded cross-sections of various opium poppy (cv. Louisiana) organs showed the co-localization of 6OMT, CNMT, 4′OMT and SAT to a specific cell type associated with phloem tissue throughout the plant (Figure 3). In roots, bundles composed of sieve element/companion cell pairs and laticifers are interspersed among parenchyma tissue throughout the secondary phloem, which surrounds a core of secondary xylem (Figure 3a). Immunofluorescence experiments using polyclonal antibodies showed that 6OMT (Figure 3b), CNMT (Figure 3c), 4′OMT (Figure 3d) and SAT (Figure 3e) are localized to the same cells in the vascular tissues of the root. In contrast, co-localization using rabbit major latex protein (MLP) antibodies, which specifically labeled latex proteins, and mouse CNMT antibodies showed that laticifers are usually adjacent or proximal to cells containing 6OMT, CNMT, 4′OMT and SAT (Figure 3c). An initial identification of the cell type involved in benzylisoquinoline alkaloid production was achieved using a sieve element-specific H+-ATPase monoclonal antibody (Langhans et al., 2001), which produced a localization result identical to that of each biosynthetic enzyme (Figure 3f).
Identical results were obtained using sections of stem (Figure 3h–l), leaf (Figure 3n–r), and carpel (Figure 3t–x) from opium poppy (cv. Louisiana). Antibodies to 6OMT, CNMT, 4′OMT and SAT were all detected in the same cells adjacent or proximal to laticifers, which were definitively identified in stem (Figure 3i), leaf (Figure 3o) and carpel (Figure 3u) using MLP-specific antibodies. In stems, laticifers are generally larger than those in roots and located closer to the cortex than sieve elements and companion cells (Figure 3i). Similarly, the large laticifers in leaves are generally abaxial to other phloem tissues (Figure 3m). The sieve element-specific H+-ATPase monoclonal antibody (Langhans et al., 2001) was again used to support the identity of the cell type containing each of the benzylisoquinoline alkaloid biosynthetic enzymes. Tissue sections treated with pre-immune serum from individual mice used to produce each antibody did not show labeling (data not shown).
Immunofluorescence labeling using resin-embedded stem longitudinal sections (Figure 4a) was used to corroborate the co-localization of 6OMT (Figure 4b), CNMT (Figure 4c), 4′OMT (Figure 4d), SAT (Figure 4e) and phloem-specific H+-ATPase (Figure 4f) to sieve elements adjacent or proximal to laticifers. Sieve elements were positively identified by staining sections with aniline blue to visualize callose, a β-1,3-linked glucan lining the walls of plasmodesmata and characteristically associated with sieve plates (Smith and McCully, 1978). Sieve plates were associated with all immunofluorescent, elongated sieve elements, which were found adjacent or proximal to laticifers (Figure 4c). Consistent with results obtained using cross-sections, immunofluorescence was restricted to the peripheral cytoplasm of sieve elements. Labeling was not detected in phloem parenchyma (Figure 4).
The identity of the immunolabeled cells as sieve elements was confirmed by the characteristic ultrastructure of phloem cell types observed using electron microscopy. Companion cells were distinguished by their relatively small size and dense cytoplasm, which contained an abundance of rough endoplasmic reticulum (ER) and a full complement of organelles, including plastids and mitochondria. Sieve elements displayed a thin layer of peripheral cytoplasm consisting mostly of ER, and a few plastids and mitochondria, but lacking vacuoles and a nucleus (Figure 5a). The cytoplasm of phloem parenchyma was more uniformly distributed throughout the cell. Laticifers were clearly identified by a dense cytoplasm containing many large, membranous vesicles, an abundance of ER and numerous plastids and mitochondria (Figure 5a). Immunogold labeling studies showed the strict association of 6OMT (Figure 5b), CNMT (Figure 5c), 4′OMT (Figure 5d) and SAT (Figure 5e) with electron dense regions of the peripheral cytoplasm of sieve elements. Antibody-specific immunogold labeling was not detected in any other cell type above the background levels resulting from the use of pre-immune serum (Figure 5f).
Alkaloid biosynthetic gene transcripts occur in the companion cells of phloem sieve elements
In situ hybridization using digoxigenin (DIG)-labeled antisense RNA probes complementary to 6OMT (Figure 6a), CNMT (Figure 6b), 4′OMT (Figure 6c) and SAT (Figure 6d) mRNAs were used to localize alkaloid biosynthetic gene transcripts to the companion cells adjacent to larger, angular sieve elements in the phloem of opium poppy stems. Gene transcripts for all four enzymes were also detected in companion cells of the phloem in roots, leaves and carpels (data not shown). No other cell type was labeled. Laticifers in each section could be identified by a characteristically thick cell wall, non-angular shape and a diameter generally >30 μm. Hybridization of gene transcripts was not detected in stem sections exposed to sense RNA probes for 6OMT (Figure 6e), CNMT, 4′OMT or SAT (data not shown) at a concentration up to fivefold higher than that used for antisense RNA probes. Similarly, no hybridization signal was produced in root, leaf or carpel sections exposed to sense RNA probes (data not shown). The lack of a signal in sections treated with sense RNA probes demonstrates the specificity of hybridization between the antisense RNA probes and transcripts localized in companion cells throughout the plant.
Alkaloid biosynthetic enzymes are rapidly induced in seedlings after seed germination
Immunoblot analysis using polyclonal antibodies prepared against recombinant 6OMT, CNMT, CYP80B1, 4′OMT, BBE, SAT and COR showed that several benzylisoquinoline alkaloid biosynthetic enzymes are induced in opium poppy (cv. Louisiana) seedlings after seed germination (Figure 7). One day after imbibition 6OMT, CYP80B1 and BBE were present in seeds, whereas CNMT, 4′OMT, SAT and COR were not detected until 4 days post-imbibition. The level of each enzyme was maximal by day 7 and remained constant until at least 16 days after imbibition. Major latex protein antibodies did not cross-react with seedling proteins through the first 10 days post-imbibition, but MLP was detected at relatively low levels on days 13 and 16 (Figure 7).
Alkaloid biosynthetic enzymes are restricted to phloem sieve elements in seedlings
Immunofluorescence labeling using resin-embedded cross-sections of roots and hypocotyls from opium poppy (cv. Louisiana) seedlings 13 days after imbibition showed the co-localization of seven benzylisoquinoline alkaloid biosynthetic enzymes (i.e. 6OMT, CNMT, CYP80B1, 4′OMT, BBE, SAT and COR) to a specific cell type associated with the phloem (Figure 8). Young roots display a diarch stele composed of a core of primary xylem, surrounded by sieve element and companion cell pairs interspersed among phloem parenchyma tissue (Figure 8a). The stele in hypocotyls consists of two adjacent bundles of primary xylem surrounded by opposing arcs of primary phloem (Figure 8j). Immunofluorescence localization using polyclonal antibodies showed that 6OMT (Figure 8b,k), CNMT (Figure 8c,l), CYP80B1 (Figure 8d,m), 4′OMT (Figure 8e,n), BBE (Figure 8f,o), SAT (Figure 8g,p) and COR (Figure 8h,q) were localized to the same cells in the phloem of roots and hypocotyls. Polyclonal antibodies against MLP failed to produce a fluorescent signal in either organ above the level of background autofluorescence (data not shown). A conclusive identification of the cell type involved in benzylisoquinoline alkaloid metabolism in seedlings was achieved using a sieve element-specific H+-ATPase monoclonal antibody (Langhans et al., 2001), which produced a localization result identical to that of each biosynthetic enzyme in roots (Figure 8i) and hypocotyls (Figure 8r). Tissue sections treated with pre-immune serum from individual mice used to produce each polyclonal antibody did not show labeling (data not shown).
In this study, and in our previous work (Bird et al., 2003), we have shown that seven benzylisoquinoline alkaloid biosynthetic enzymes are localized to sieve elements of the phloem in opium poppy plants and seedlings, and that corresponding gene transcripts are found in adjacent companion cells. This study was conducted on the opium poppy cv. Louisiana, which is grown commercially for the production of morphine and codeine. Our previous work was performed on the opium poppy cv. Marianne, which also produces morphine but is not cultivated for the manufacture of pharmaceutical alkaloids. In addition to similar alkaloid profiles when cultivated under growth chamber conditions, no substantial differences in gene transcript or protein levels were detected for 6OMT, CNMT, 4′OMT or SAT in any organ of cv. Louisiana compared with those of cv. Marianne (Figure 2). Both cultivars displayed similar levels of the alkaloid biosynthetic gene transcripts and enzymes used in this study and in our previous work (Bird et al., 2003; data not shown); thus, the suggestion of Weid et al. (2004) that our prior localization experiments were hampered by the use of cv. Marianne is unsubstantiated.
In addition to the seven enzymes we have localized to sieve elements, two other benzylisoquinoline alkaloid biosynthetic enzymes, TYDC and 7OMT, have also been associated with vascular tissues in opium poppy (El-Ahmady and Nessler, 2001; Facchini and De Luca, 1995; Weid et al., 2004). Tyrosine/dopa decarboxylase is involved in several biochemical processes in opium poppy including the formation of l-dopa (Figure 1), and thus is not an exclusive marker for benzylisoquinoline alkaloid biosynthesis. Nevertheless, it is notable that TYDC gene transcripts were not detected in laticifers (El-Ahmady and Nessler, 2001; Facchini and De Luca, 1995). 7-O-Methyltransferase converts (R,S)-reticuline to laudanine (Figure 1) and was originally identified by proteomic analysis of opium poppy latex (Ounaroon et al., 2003). Soluble latex proteins were fractionated by two-dimensional gel electrophoresis and 7OMT was isolated by internal amino acid sequencing of an abundant polypeptide spot, followed by amplification of the corresponding cDNA using degenerated oligodeoxynucleotide PCR primers. In contrast to the isolation of 7OMT as a purported latex protein, immunofluorescence localization did not associate the enzyme with laticifers (Weid et al., 2004). Instead, 7OMT was localized to a cell type adjacent or proximal to laticifers and identified as phloem parenchyma (Weid et al., 2004). We propose that TYDC and 7OMT are co-localized to sieve elements of the phloem along with other benzylisoquinoline alkaloid biosynthetic enzymes. We also suggest that previous biochemical (Fairbairn and Wassel, 1964; Fairbairn et al., 1968; Roberts et al., 1983) and genomic (Decker et al., 2000; Ounaroon et al., 2003) work involving the isolation of latex from lanced seed capsules was confounded by unavoidable contamination from sieve element sap, since both laticifers and sieve tubes exhibit positive turgor pressure. The crude lancing of seed capsules is certain to result in the exudation of unanchored proteins from both cell types. Consistent with this suggestion, the soluble enzymes 6OMT, 7OMT (Ounaroon et al., 2003) and COR (Decker et al., 2000) have been identified in crude seed capsule exudates, but membrane-bound enzymes of alkaloid biosynthesis are consistently absent (Gerardy and Zenk, 1993a,b).
Previously, and in this study, we have used the occurrence of sieve plates at the apical and basal junction of adjoining, elongated cells as a distinct characteristic for identifying the immunolabeled cell type as sieve elements of the phloem (Figure 4; Bird et al., 2003). In a similar study by Weid et al. (2004), the apparent absence of sieve plates in similarly elongated cells labeled with 4′OMT, SAT and 7OMT antibodies was used to identify the cell type as phloem parenchyma rather than sieve elements. Cells with sieve plates were reportedly not labeled (Weid et al., 2004). In our present work, we also used a definitive marker for sieve elements – a monoclonal antibody specific for an H+-ATPase isoform found only in this cell type (Figures 3, 4 and 8; Langhans et al., 2001). In addition, we relied on the characteristic ultrastructure of sieve elements, observed using electron microscopy, to confirm the identity of the immunolabeled cells. We conclude that alkaloid biosynthetic enzymes are, indeed, associated with phloem sieve elements. Assuming that neither our localization results nor those of Weid et al. (2004) are plagued by non-specific labeling problems, we offer the following suggestions to explain the apparent disparity in the data. Since it is not known whether sieve elements involved in benzylisoquinoline alkaloid biosynthesis are also functional in the transport of solutes and information macromolecules, perhaps only immature sieve elements (i.e. lacking sieve plates) were labeled in the study by Weid et al. (2004). Indeed, differences in the specificity and concentration of antibodies used in the two studies might have produced differential labeling of cells with either high or low levels of alkaloid biosynthetic enzymes. It is also notable that plants used in our work were grown in climate-controlled growth chambers, whereas those used by Weid et al. (2004) were field grown. Different developmental programs, perhaps mediated by variations in environmental conditions, might affect the spatial limits of expression of alkaloid biosynthetic genes. Interestingly, the in situ hybridization results presented by Weid et al. (2004) appear consistent with our data supporting the companion cell-specific localization of alkaloid biosynthetic gene transcripts (Figure 6; Bird et al., 2003). Clearly, more work is necessary to clarify these issues.
Alkaloid biosynthetic enzymes currently localized to sieve elements include two O-methyltransferases (6OMT and 4′OMT), an N-methyltransferase (CNMT), a P450-dependent monooxygenase (CYP80B1), an FAD-dependent oxidoreductase (BBE), an acetyl-CoA-dependent acyltransferase (SAT) and an aldo-keto reductase (COR). It has been proposed that CYP80B1 and its associated cytochrome P450 reductase, which are membrane-bound enzymes, would be inactive in the cytosol of sieve elements due to an alluded absence of ER (Weid et al., 2004). However, the ER is well preserved in sieve elements (Sjolund and Shih, 1983; Thorsch and Esau, 1981a,b) in the form of a sieve element reticulum (SER), which is proposed to function in the trafficking of proteins through plasmodesmata that connect sieve elements and companion cells (Facchini and St-Pierre, 2005; Lucas et al., 2001; Oparka and Turgeon, 1999; van Bel and Knoblauch, 2000). The empirical movement of phloem proteins from companion cells to sieve elements has been demonstrated (Balachandran et al., 1997; Stadler et al., 2005). Many sieve element-specific polypeptides, such as P-proteins, are not translocated along the solute stream perhaps because they are anchored to the SER (Knoblauch and van Bel, 1998). Alkaloid biosynthetic enzymes should also be associated with the peripheral region of sieve elements to prevent dislodging and translocation. Oligopeptide anchors have been suggested to immobilize the parietal SER and other organelles to the plasma membrane, forming a parietal cytoplasmic channel where the majority of phloem proteins are thought to reside (Ehlers et al., 2000). It is notable that each enzyme was localized to a peripheral cellular region corresponding to the parietal layer of sieve elements (Figures 3, 4 and 8). Immunogold labeling, in particular, implicates the parietal layer in the localization of alkaloid biosynthetic enzymes (Figure 5).
The SER also provides an appropriate cellular compartment for the unique functional requirements of BBE, which possesses a signal peptide (Bird and Facchini, 2001) and was localized to the ER in cultured opium poppy cells (Alcantara et al., 2005). Sanguinarine was also present in the ER, suggesting that the entire biosynthetic pathway occurs in association with the ER (Alcantara et al., 2005). The localization of seven enzymes to the peripheral region of sieve elements (Figures 3 and 8; Bird et al., 2003) is consistent with the possible assembly of metabolic channels for benzylisoquinoline alkaloid biosynthesis associated with the SER in opium poppy. The RNAi-mediated silencing of COR genes in transgenic opium poppy plants resulted in the accumulation of (S)-reticuline rather than codeinone (Figure 1; Allen et al., 2004); thus, the removal of a single enzyme prevented general benzylisoquinoline alkaloid intermediates from entering the morphine-specific branch pathway. Multi-enzyme complexes have been demonstrated for flavonoid (Achnine et al., 2004; Burbulis and Winkel-Shirley, 1999; He and Dixon, 2000) and polyamine (Panicot et al., 2002) metabolism in Arabidopsis thaliana. Such complexes involve weak physical interactions between biosynthetic enzymes, which allow the efficient and isolated channeling of intermediates along a metabolic pathway (Winkel, 2004).
The participation of multiple cell types in the biosynthesis and accumulation of alkaloids in opium poppy raises intriguing questions about the transport of products from sieve elements to laticifers. Recently, a multidrug-resistance-type ATP-binding cassette (ABC) protein from Coptis japonica capable of transporting the benzylisoquinoline alkaloid berberine was reported (Shitan et al., 2003). An ABC-transporter transcript identified in the phloem of rice (Asano et al., 2002) suggests that a similar protein might reside in plasma membranes at the interface between sieve elements and laticifers in opium poppy. However, the symplastic transport of alkaloids between cell types must also be considered. The proximal relationship between sieve elements and laticifers in opium poppy (Nessler and Mahlberg, 1977; Thureson-Klein, 1970) supports the potential for the translocation of small molecules between these cell types.
The occurrence of benzylisoquinoline alkaloids in near-basal angiosperm families, such as the Papaveraceae and Ranunculaceae, suggests an ancient evolutionary origin for this group of secondary metabolites (Facchini et al., 2004; Liscombe et al., 2005). A monophyletic origin is supported by the extensive sequence homology among biosynthetic enzymes from different plant families operating at corresponding points in the benzylisoquinoline alkaloid pathway (Liscombe et al., 2005; Samanani et al., 2005). Nevertheless, benzylisoquinoline alkaloid metabolism involves endodermis, pericycle, protoderm, cortex or pith tissues in T. flavum (Ranunculaceae) and does not involve vascular cell types as in opium poppy (Papaveraceae). The emergence of a complex benzylisoquinoline alkaloid pathway in an ancestral, basal angiosperm is in contrast to the proposed independent recruitment of pyrrolizidine alkaloid biosynthetic enzymes in at least four different angiosperm lineages (Reimann et al., 2004). In the case of pyrrolizidine alkaloids, the differential localization of a key pathway enzyme has been interpreted as evidence for the independent origin of the biosynthetic pathway in various lineages (Anke et al., 2004). In contrast, the differential cell type-specific localization of benzylisoquinoline alkaloid gene transcripts in T. flavum and opium poppy implicates the migration of established pathways between cell types as a key feature of phytochemical evolution. Clearly, benzylisoquinoline alkaloid pathways have become established in cell types other than those that participate in the biosynthesis of other alkaloid groups.
Opium poppy (P. somniferum L. cv. Marianne and cv. Louisianne) plants were maintained in a growth chamber (Conviron, Winnipeg, MB, Canada) at 20°C/18°C (light/dark) with a photoperiod of 14 h. Plant organs were harvested 2–3 days after anthesis except for carpels, which were collected 3–5 days after anthesis. A combination of fluorescent (Cool White, Sylvania, Mississauga, ON, Canada) and incandescent lighting was used. Seedlings were grown at 23°C in sterile Petri plates (100 × 15 mm) on Phytagar (Gibco, Burlington, ON, Canada) containing B5 salts and vitamins (Gamborg et al., 1968), under a photoperiod of 16 h using wide-spectrum fluorescent tubes (Gros-Lux Wide Spectrum, Sylvania) with a fluency rate of 35 μmol m−2 sec−1. Before germination, the seeds were surface sterilized with 20% (v.v) sodium hypochlorite for 15 min, and thoroughly rinsed with sterile water.
Heterologous expression and purification of proteins
Opium poppy 6OMT (Facchini and Park, 2003; Ounaroon et al., 2003), 4′OMT (Facchini and Park, 2003), CNMT (Facchini and Park, 2003) and SAT (Grothe et al., 2001) open reading frames (ORFs) were inserted in-frame into pRSETB (Invitrogen, Carlsbad, CA, USA), and the constructs introduced into Escherichia coli strain ER2566 pLys S (New England Biolabs, Beverly, MA, USA). The assembly of opium poppy CYP80B1 (Huang and Kutchan, 2000), BBE (Facchini et al., 1996) and COR (Unterlinner et al., 1999) expression constructs has been described previously (Bird et al., 2003). Heterologous expression was performed by inoculating 1 l of Lauria–Bertani broth (Sambrook et al., 1989) containing 100 mg l−1 ampicillin with 5 ml of overnight bacterial culture and incubating at 37°C. At a density OD600 = 0.5, the cultures were induced for 4 h with 100 μm isopropyl-β-D-thiogalactopyranoside. Cells were pelleted, resuspended in homogenization buffer [50 mm Tris-HCl, pH 7.5, 10 mm EDTA, 10 μm phenylmethylsufonyl fluoride (PMSF), and 5 mm 2-mercaptoethanol (2-ME)] and ruptured using a French press (Spectronic Instruments, Rochester, NY, USA). Cell debris and protein inclusion bodies were recovered by centrifugation. The rinsed pellet was solubilized in homogenization buffer containing 6 m urea and the solution was passed through a 0.20 μm filter. Recombinant proteins were affinity purified using a Ni2+-charged HiTrap column according to the manufacturer's instructions (GE Healthcare, Piscataway, NJ, USA).
Preparation of antibodies
Purified recombinant antigen proteins were dialyzed against PBS (100 mm sodium phosphate buffer, pH 7.2, 140 mm NaCl), resuspended at a concentration of 100 μg ml−1 and emulsified (1:1) with Freund's complete adjuvant. Repeated subcutaneous injections of 100 μl were performed in mice, with booster injections every 3 weeks until a sufficient titer was achieved.
Protein extraction, fractionation and Western blot analysis
Tissues were collected by vacuum filtration and ground under liquid nitrogen to a fine powder with a mortar and pestle in the presence of polyvinyl polypyrrolidone (100 mg g−1 of tissue). Ground samples were suspended in extraction buffer (50 mm Tris-HCl, pH 7.5, 5 mm EDTA, 5 mm PMSF and 5 mm 2-ME), incubated on ice for 30 min, and the supernatant collected by centrifugation at 10 000 g for 10 min at 4°C. Protein concentrations were determined according to Bradford (1976) using bovine serum albumin as a standard. Protein samples (25 μg) were fractionated by SDS-PAGE using 10.5% (w/v) polyacrylamide gels (Laemmli, 1970) and transferred to nitrocellulose membranes. Protein blots were incubated with 10 μg ml−1 CYP80B1 antiserum, 5 μg ml−1 BBE antiserum or 25 μg ml−1 COR antiserum for 3 h, washed in Tris-buffered saline with Tween [TBST; 10 mm Tris-HCl pH 7.5, 150 mm NaCl, 0.1% (v.v) Tween-20] and incubated for 2 h with horseradish peroxidase-conjugated anti-mouse secondary antibodies (Bio-Rad, Hercules, CA, USA). The membranes were washed in TBST and incubated for 5 min in SuperSignal West Pico chemiluminescent substrate (Pierce Biotechnology, Rockford, IL, USA). The membranes were then subjected to autoradiography.
RNA isolation and analysis
Total RNA was isolated using the method of Logemann et al. (1987) and poly(A)+ RNA was selected by oligo(dT) cellulose chromatography. For gel blot analysis, 15 μg of RNA was fractionated on a 1.0% (w/v) agarose gel, containing 7.5% (w/v) formaldehyde, before transfer to a nylon membrane. Blots were hybridized with a random-primer 32P-labeled, full-length N-chlorosuccinimide probe. Hybridizations were performed at 65°C in 0.25 mm sodium phosphate buffer, pH 8.0, 7% (w/v) SDS, 1% (w/v) BSA and 1 mm EDTA. Blots were washed at 65°C, twice with 2× SSC and 0.1% (w/v) SDS and twice with 0.2× SSC and 0.1% (w/v) SDS (1× SSC = 0.15 m NaCl and 0.015 m sodium citrate, pH 7.0), and autoradiographed with an intensifier at −80°C.
Tissue fixation and embedding for immunocytochemical localization
Tissue fixation and fluorescence immunocytological localization were performed as described previously (Voznesenskaya et al., 1999). Briefly, tissues were immersed in fixation buffer [50 mm piperazine-1-4-bβ(2-ethanesulphonic acid) (PIPES), pH 7.0, 1.25% (v.v) glutaraldehyde, 2% (v.v) paraformaldehyde and 5 μm PMSF], cut with a razor blade into 1.5–2 mm sections, fixed for 2 h, and rinsed in 50 mm PIPES, pH 7.0, containing 5 μm PMSF. The tissues were dehydrated using a 30–100% (v.v) ethanol series with 2 h incubation in each solution. After dehydration, LR White resin (London Resin Company, London, UK) was introduced into the ethanol series at an initial ratio of 1:4 (v.v) and gradually increased to 1:3, 1:2, 1:1, 2:1 and 3:1 (v.v). Finally, tissues were immersed in pure resin, cast into 1-ml gelatin capsules and incubated at 60°C for 16 h. Sections were cut 1.0 μm thick using a Reichert-Jung Ultracut E microtome (Reichert-Jung, Vienna, Austria).
Tissue fixation and embedding for in situ hybridization
Organs were immersed in FAA [50% (v.v) ethanol, 5% (v.v) acetic acid, 3.7% (v.v) formaldehyde], cut with a razor blade into 2–5 mm segments, and fixed overnight at 4°C. Tissues were dehydrated using an ethanol/tertiary butanol (t-butanol) series (4:1:5; 5:2:3; 5:3.5:1.5; 4.5:5.5:0; 2.5:7.5:0; and 0:1:0 ethanol: t-butanol:water) with 2 h incubation in each solution except for the final step, which was overnight. Paraplast Plus (Oxford Labware, St Louis, MO, USA) was added to a paraffin infiltration series (1:1, 6.7:3.3 and 1:0 wax: t-butanol) with overnight incubations for each step. Embedded tissues were cut into 10 μm sections using an American Optical 620 microtome. Sections were placed onto aminopropyltriethoxysilane (AES)-coated slides and incubated overnight at 37°C to promote firm adhesion of sections to the slides.
Mice antibodies against 6OMT, CNMT, CYP80B1, 4′OMT, BBE, SAT and COR were used at a concentration of 20 μg ml−1. The rabbit anti-MLP (Griffing and Nessler, 1989) IgG fraction was purified using the Affi-Gel Protein A MAPSII Kit (Bio-Rad, Hercules, CA, USA) and used at a concentration of 20 μg ml−1. The H+-ATPase monoclonal antibody clone 30D5C4 (Langhans et al., 2001) was used at a concentration of 10 μg ml−1. Tissue sections were incubated with primary antibodies for 2 h, rinsed three times in Tris-buffered saline (TBS; 10 mm Tris-HCl pH 7.5, 150 mm NaCl) containing 1% (w/v) BSA (Fraction V, Roche Diagnostics, Indianapolis, IN, USA), and twice in TBS for 10 min. Sections were incubated for 1 h with either Alexa 488-conjugated goat anti-mouse IgG or Alexa 594-conjugated goat anti-rabbit IgG (Molecular Probes Inc, Eugene, OR, USA), then rinsed in TBS and water. Slides were sealed with Aquaperm (ThermoShandon, Pittsburg, PA, USA).
In situ hybridization
In situ hybridization was performed as described by St-Pierre et al. (1999) with some modifications. Briefly, 0.5 kb fragments from the 3′ region of 6OMT, CNMT, 4′OMT and SAT ORFs were amplified by PCR and served as templates for the synthesis of sense and antisense DIG-labeled RNA probes using T3 and T7 RNA polymerases. Each PCR product contained T7 and T3 promoter sequences at the 5′ and 3′ ends, respectively.
Sections were deparaffinized and rehydrated using an ethanol series (1:0, 1:0, 9.5:0.5, 7:3 and 1:1 ethanol:water) with 5 min incubation in each solution. Sections were incubated in pre-hybridization buffer (100 mm Tris-HCl, pH 8.0, 50 mm EDTA) containing 5 μg ml−1 proteinase K (Roche Molecular Systems, Alameda, CA, USA) for 30 min, then blocked in TBS containing 2 mg ml−1 glycine. Sections were post-fixed in 3.7% (v.v) formaldehyde in PBS, incubated in 100 mm triethanolamine buffer, pH 8.0, containing 0.25% (v.v) acetic anhydride and, finally, rinsed in TBS. The slides were inverted onto 100 μl of hybridization buffer [10 mm Tris-HCl, pH 6.8, 10 mm sodium phosphate buffer, pH 6.8, 40% (v.v) deionized formamide, 10% (w/v) dextran sulfate, 300 mm NaCl, 5 mm EDTA, 1 mg ml−1 yeast tRNA, 500 ng ml−1 DIG-RNA, 0.8 U ml−1 RNAse inhibitor (Invitrogen)] spread over a coverslip. Slides were sealed in a Petri dish lined with filter paper soaked in 50% (v.v) formamide, and incubated overnight at 50°C.
Slides were immersed in 2× SSC (1× SSC = 300 mm NaCl, 30 mm sodium citrate, pH 7.0) at 37°C until the coverslips fell off. Sections were incubated in 50 mg ml−1 RNase A (Roche Molecular Systems) in 500 mm NaCl, 10 mm Tris-HCl, pH 7.5, 1 mm EDTA for 30 min at 37°C. Slides were washed in 2 l of the following solutions for 1 h: 2× SSC and 1× SSC at room temperature, and 0.1× SSC at 60°C. Slides were rinsed in TBST and blocked for 1 h in TBST containing 2% (w/v) BSA. Slides were inverted onto coverslips carrying 100 μl of goat anti-DIG-AP conjugate (Roche Molecular Systems) diluted 1:200 in TBST containing 1% (w/v) BSA, and incubated for 2 h in sealed Petri dishes lined with filter paper soaked in TBST. After incubation, the slides were rinsed in TBST and AP buffer. Colorimetric development was performed in AP buffer containing 400 mm 5-bromo-4-chloro-3-indolyl phosphate and 428 mm nitro blue tetrazolium for 30–60 min.
Aniline blue and toluidine blue-O staining
For general anatomy, LR White-embedded sections were stained in benzoate buffer (10 mm sodium benzoate, pH 4.4) containing 0.1% (w/v) toluidine blue-O. Callose was detected by staining sections in 67 mm phosphate buffer, pH 8.5, containing 0.05% (w/v) aniline blue.
Fluorescence and light microscopy
Immunofluorescence labeling was viewed using a Leica DM RXA2 microscope (Leica Microsystems, Wetzler Germany) and images acquired with a Retiga EX digital camera (QImaging, Burnaby, BC, Canada). Alexa-488 and Alexa-594 fluorescent labels were detected using Leica L5 and TX2 filters, respectively. False-color imaging was performed using Open Lab version 2.09 (Improvision, Coventry, UK). Light microscopy images were captured using the Leica microscope and Retiga camera mounted with a RGB color liquid crystal filter (QImaging).
Thin sectioning for electron microscopy
Specimen blocks prepared as described for immunocytochemical localization were trimmed to 1 mm2 and sectioned to a thickness of 70–120 nm according to silver–gold refraction using 6.35 mm glass knives, cut at a 45° angle, on a Leica EM UC6 ultramicrotome (Reichert-Jung). Section compression was reduced using chloroform fumes.
Sections collected on nickel grids were blocked in TBS containing 0.05 m glycine for 15 min followed by TBS containing 2% (w/v) BSA and 0.3% Tween 20 for 1 h, incubated with protein A sepharose-purified primary antibodies in a humid chamber for 2 h, and rinsed three times for 15 min each in TBS containing 2% (w/v) BSA and for twice for 10 min each in TBS. Grids were incubated for 1 h with 10 nm colloidal gold goat anti-mouse IgG (Jackson Immunochemicals, West Grove, PA, USA) and rinsed three times for 10 min each in TBS and twice for 10 min each in water. Grids were dried on filter paper before staining.
Grid staining and viewing
Immunogold-labeled grids were stained in a 4:1 2% (w/v) uranyl acetate: 2% (w/v) potassium permanganate solution for 5 min. Grids were viewed at 75 kV using a Hitachi 7000X transmission electron microscope.
We are grateful to Martine Dubedout (Francopia Sanofi-Aventis) for the gift of the Louisiana opium poppy cultivar, Craig Nessler (Virginia Technical University) for the MLP antibodies and Cornelia Ullrich (Technische Universität Darmstadt) for the PM H+-ATPase antibodies. We also thank Jill Hagel for the HPLC analysis and Vincent Franceschi and David Bird for valuable discussions. PJF holds the Canada Research Chair in Plant Biotechnology. This work was funded by a grant from the Natural Sciences and Engineering Research Council of Canada to PJF.