Molecular cloning and functional expression of codeinone reductase: the penultimate enzyme in morphine biosynthesis in the opium poppy Papaver somniferum


  • Bernhard Unterlinner,

    1. Laboratorium für Molekulare Biologie, Universität München, Karlstrasse 29, 80333 München, Germany
    Search for more papers by this author
  • Rainer Lenz,

    Search for more papers by this author
      †Present address: Abteilung Pharma-QA-EZ2, Bayer AG, Gebäude E 39, 51368 Leverkusen, Germany.
  • Toni M. Kutchan

    Corresponding author
      *For correspondence (fax +11 49 89 5902611;
      †Present address: Abteilung Pharma-QA-EZ2, Bayer AG, Gebäude E 39, 51368 Leverkusen, Germany.
    Search for more papers by this author

*For correspondence (fax +11 49 89 5902611;
†Present address: Abteilung Pharma-QA-EZ2, Bayer AG, Gebäude E 39, 51368 Leverkusen, Germany.


The narcotic analgesic morphine is the major alkaloid of the opium poppy Papaver somniferum. Its biosynthetic precursor codeine is currently the most widely used and effective antitussive agent. Along the morphine biosynthetic pathway in opium poppy, codeinone reductase catalyzes the NADPH-dependent reduction of codeinone to codeine. In this study, we have isolated and characterized four cDNAs encoding codeinone reductase isoforms and have functionally expressed them in Escherichia coli. Heterologously expressed codeinone reductase-calmodulin-binding peptide fusion protein was purified from E. coli using calmodulin affinity column chromatography in a yield of 10 mg enzyme l-1. These four isoforms demonstrated very similar physical properties and substrate specificity. As least six alleles appear to be present in the poppy genome. A comparison of the translations of the nucleotide sequences indicate that the codeinone reductase isoforms are 53% identical to 6′-deoxychalcone synthase from soybean suggesting an evolutionary although not a functional link between enzymes of phenylpropanoid and alkaloid biosynthesis. By sequence comparison, both codeinone reductase and 6′-deoxy- chalcone synthase belong to the aldo/keto reductase family, a group of structurally and functionally related NADPH-dependent oxidoreductases, and thereby possibly arise from primary metabolism.


The search for useful drugs of defined structure from plants began with the isolation of morphine from dried latex, or opium, of the opium poppy Papaver somniferum in 1806 ( Sertürner 1806). The narcotic analgesic morphine and the antitussive and narcotic analgesic codeine, the antitussive and apoptosis inducer noscapine ( Ye et al. 1998 ), and the vasodilator papaverine are currently the most important physiologically active alkaloids from opium poppy. Of these four alkaloids, only papaverine is prepared by total chemical synthesis for commercial purposes. Opium poppy therefore serves as one of the most important renewable resources for pharmaceutical alkaloids. Per annum, 90–95% of the approximately 160 tons of morphine that are purified are chemically methylated to codeine, which is then used either directly or is further converted to a variety of derivatives such as dihydrocodeinone and 14-hydroxydihydrocodeinone that find use as antitussives and analgesics ( Kutchan 1998). The elicit production of morphine for acetylation to heroin is unfortunately almost 10 times that amount, more than 1200 tons per year ( Zenk 1994).

The enzymatic synthesis of morphine in opium poppy has been almost completely elucidated by M.H. Zenk and co-workers and is summarized by Kutchan (1998). Opium poppy produces more than 100 different alkaloids that are derived from the amino acid l-tyrosine and have the tetrahydrobenzylisoquinoline alkaloid (S)-reticuline as a common intermediate. There are three NADPH-dependent reductases involved in the conversion of (S)-reticuline to morphine. (S)-Reticuline must first be converted to (R)-reticuline before the phenanthrene ring with the correct stereochemistry at C-13 can be formed. The inversion of stereochemistry at C-1 of (S)-reticuline occurs by oxidation to the 1,2-dehydroreticulinium ion followed by stereospecific reduction to the R-epimer by 1,2-dehydroreticulinium ion reductase ( EC1.5.1.27) ( De-Eknamkul & Zenk 1992). The second reduction occurs after formation of the phenanthrene nucleus with stereospecific reduction of salutaridine to salutaridinol by salutaridine reductase ( EC1.1.1.248) ( Gerardy & Zenk 1993). The third reduction is the penultimate step in the biosynthetic pathway to morphine, the reduction of codeinone to codeine by codeinone reductase ( EC1.1.1.247) ( Fig. 1) ( Lenz & Zenk 1995a; Lenz & Zenk 1995b). The substrate for codeinone reductase, codeinone, exists in an equilibrium with its positional isomer neopinone. In vitro, as codeinone is reduced, this equilibrium is continually driven from neopinone towards codeinone until the substrates are depleted ( Gollwitzer et al. 1993 ). Each of the known enzymes of morphine biosynthesis has been detected in both P. somniferum plants and cell suspension culture, yet plant cell cultures have never been shown to accumulate morphine ( Kutchan 1998).

Figure 1.

Proposed biosynthetic pathway leading from thebaine to morphine in the opium poppy.

The reduction of codeinone to codeine by codeinone reductase drives the non-enzymatic equilibrium between neopinone and codeinone in a physiologically forward direction. The demethylation of thebaine and codeine are each thought to be catalyzed by cytochrome P450-dependent enzymes.

To date, none of the genes specific to morphine biosynthesis in opium poppy have been isolated. Tyrosine/dopa decarboxylase and a cytochrome P-450 reductase have been investigated at the molecular genetic level, but are involved in multiple biochemical processes in this plant (Facchini & De Facchini & Luca 1994 ; Rosco et al. 1997 ). Morphine, along with the chemotherapeutic agents vincristine, vinblastine and camptothecin, is one of the most important alkaloids commercially isolated from medicinal plants. Isolation of the genes of morphine biosynthesis would facilitate metabolic engineering of opium poppy to produce plants with specific patterns of alkaloids and could ultimately lead to an understanding of the inability of plant cell cultures to accumulate morphine. In this paper, we take the first step towards both of these goals with the isolation and characterization of cDNAs and genes that encode codeinone reductase isoforms in opium poppy. To our knowledge, this is the first report of the cloning of genes specific to morphine biosynthesis.


Purification and amino acid sequence analysis of opium poppy codeinone reductase

Codeinone reductase was purified to apparent electrophoretic homogeneity from opium poppy cell suspension cultures and the amino acid sequence of seven endoproteinase Lys-C-generated peptides was determined ( Fig. 2). A comparison of these amino acid sequences with those available in the GenBank/EMBL sequence database allowed a relative positioning of peptides 7, 14 and 16 due to sequence homology with an NADPH-dependent reductase from members of the Fabaceae – alfalfa, glycyrrhiza and soybean (6′-deoxychalcone synthase) that synthesizes 4,2′,4′-trihydroxychalcone in coaction with chalcone synthase ( Fig. 3) ( Welle et al. 1991 ). PCR primers were then designed based on the codeinone reductase peptide sequences. The sequences of the primers used in the first round of PCR were:

Figure 2.

Partial amino acid sequences of native codeinone reductase.

Codeinone reductase was purified to apparent electrophoretic homogeneity from cell suspension cultures of opium poppy and hydrolyzed with endoproteinase Lys-C. The resultant peptide mixture was resolved by HPLC and the amino acid sequences of seven peptides were obtained.

Figure 3.

Amino acid sequence homology of codeinone reductase internal peptides.

Codeinone reductase peptides 3, 7, 14, 16 and 17 aligned with the reductase subunit of the 6′-deoxychalcone synthase complex from alfalfa, glycyrrhiza and soybean allowing the relative positioning of these internal peptides from opium poppy. The position of the sense and antisense oligodeoxynucleotide primers used in the first PCR experiments to generate partial codeinone reductase cDNAs is shown by arrows.


(derived from Peptide 14) and


(derived from Peptide 7).

Resolution of an aliquot of the first PCR experiment by agarose gel electrophoresis revealed a mixture of DNA products, none of which was the expected band of approximately 480 bp. This was presumably due to the relatively low specificity of the degenerate primers coupled to a low abundance of codeinone reductase transcript. Another aliquot of the first PCR reaction mixture was therefore used as template for nested PCR with the following primers:


(same as Peptide 14 primer above) and


(nested primer derived from Peptide 16)

to yield an approximately 360 bp DNA fragment and the following primers to yield an approximately 180 bp DNA product:


(nested primer derived from Peptide 16) and


(same as Peptide 7 primer above).

The results from the nested PCR were bands of the expected size. The translation of the nucleotide sequences of these PCR products indicated that they encode codeinone reductase.

Isolation of cDNAs encoding codeinone reductase

Screening of approximately 200 000 clones of a primary cDNA library prepared from opium poppy RNA isolated from capsule and cell suspension culture did not result in the identification of codeinone reductase clones. Likewise, difficulty was also confronted when detecting a band on RNA gel blots that corresponds to the size expected for codeinone reductase. In order to overcome the apparent problem of low steady state levels of codeinone reductase transcript, RACE-PCR was used to generate both the 5′- and 3′-portions of the cDNA ( Frohman 1993). A series of non-degenerate primers based on the nucleotide sequence information determined for the PCR product generated as described in the previous section were used for 5′- and 3′-RACE. The nucleotide sequence of the resultant 5′- and 3′-partial clones were thus determined in three major fragments and suggested the presence of isoforms. The full length cDNA clones were then generated by RT-PCR using the following primers and RNA isolated from opium poppy cell suspension culture as template:


(located at the 5′-terminus) and


(located in the 3′-flanking region)

followed by nested PCR with the following primer pair:


(located at the 5′-terminus) and


(located at the 3′-terminus).

The PCR product was digested with the restriction endonucleases NheI/BamHI, ligated into NheI/BamHI digested pCAL-c and transformed into Escherichia coli BL21(DE)pLysS. Each cDNA was hence constructed in frame in front of DNA encoding a 25 amino acid long calmodulin-binding peptide to facilitate eventual heterologous protein purification. Single colonies were grown in 3 ml medium and were assayed for the ability to reduce codeinone. Of 40 colonies tested, 10 were found to contain functional enzymes. Nucleotide sequence determination of these 10 cDNAs resulted in the identification of four alleles encoding codeinone reductase. The analogous PCR products had also been prepared with the cDNAs placed behind the calmodulin-binding peptide gene in pCAL-n-EK, but only the C-terminal fusion proteins bound the calmodulin affinity resin, indicating that the amino terminus of the fusion protein lies within the folded polypeptide.

By sequence comparison, codeinone reductase clearly belongs to the aldo/keto reductase family, a group of structurally and functionally related NADPH-dependent oxidoreductases. Members of this family possess three consensus sequences that are also positionally conserved ( Fig. 4): aldo/keto reductase consensus 1 (amino terminus) – G (F,Y) R (H,A,L) (L,I,V,M,F) D (S,T,A,G,C) (A,S) X X X X X E X X (L,I,V,M) G [cor1.1–G Y R H F D T A A A Y Q T E E C L G]; aldo/keto reductase consensus 2 (central) – (L,I,V,M,F,Y) X X X X X X X X X (K,R,E,Q) X (L,I,V,M) G (L,I,V,M) (S,C) N (F,Y) [cor1.1– M E E C Q T L G F T R A I G V C N F]; and aldo/keto reductase consensus 3 (carboxy terminus) – (L,I,V,M) (P,A,I,V) (K,R) (S,T) X X X X R X X (G,S,T,A,E,Q,K) (N,S,L) X X (L,I,V,M,F,A) [cor1.1–V V K S F N E A R M K E N L K I]. This third consensus sequence is centered on a lysine residue, the modification of which has been shown to effect the catalytic efficiency of aldose and aldehyde reductases ( Morjana et al. 1989 ).

Figure 4.

Amino acid sequence comparison of codeinone reductase isoforms.

The amino acid sequences derived from translation of the nucleotides sequences of cor1.1–1.4 as compared to the reductase subunit of the 6-´deoychalcone synthase complex from soybean (6 ′dcs) indicate the very high sequence identity between isoforms (95–96%) and to this reductase of phenylpropanoid metabolism (53%). The complete amino acid sequence of cor1.1 is shown, but only those non-identical residues of the four subsequent sequences. Shaded areas indicate the positions of the consensus sequences indicative of the members of the aldo/keto reductase family.

The four functional full-length cDNAs (cor1.1cor1.4) that encoded codeinone reductase shared approximately 95–96% sequence identity ( Fig. 4). In addition, a similar cDNA generated by PCR (cor2) was 70% identical to the codeinone reductase cDNAs, but was not functional. These opium poppy cDNAs were 53% identical to soybean NADPH-dependent reductase 6′-deoxychalcone synthase ( Welle et al. 1991 ) ( Fig. 4); 33% identical to rat 3α-hydroxysteroid dehydrogenase [EC1.1.1.50]; 38% identical to bovine prostaglandin F synthase [EC1.1.1.188]; 37% identical to apple d-sorbitol-6-phosphate dehydrogenase [EC1.1.1.200]; 38% identical to bacterial (Pseudomonas putida) morphine 6-dehydrogenase [EC1.1.1.218]; and 35% identical to yeast (Pichia stipitis) xylose reductase ( Amore et al. 1991 ).

Genomic DNA analysis and gene expression pattern

Genomic DNA was used as template for a PCR analysis of cor1.1–cor1.4. Each gene was found to contain one intron that was conserved in size (443 bp) and location (beginning after nucleotide +561) within the open reading frame, but not in nucleotide sequence. In comparison, cor2 contained two introns beginning after nucleotide +321 and +514. Genomic DNA gel blot analysis using cor1.1 as a hybridization probe resulted in a complex hybridization pattern which suggests the presence of at least six genes that could encode codeinone reductase in opium poppy ( Fig. 5). From the isolation and nucleotide sequence analysis of cDNA clones, it is certain that at least six genes are expressed in the plant and plant cell suspension culture. (Two additional partial cDNAs (cor1.5 and cor1.6) were generated by RT-PCR using plant RNA as template.) When the peptide sequences presented in Fig. 2 are compared with the translations of the cDNA sequences in Fig. 4, it is clear that a mixture of isoforms was purified for amino acid sequence analysis. From the initial biochemical analysis of codeinone reductase, evidence for only two isoforms in the poppy plant and one isoform in poppy cell suspension culture was observed ( Lenz & Zenk 1995b).

Figure 5.

Genomic DNA gel blot analysis of the codeinone reductase gene family in opium poppy.

Genomic DNA isolated from opium poppy cell suspension cultures was hybridized to cor1.1 full-length cDNA and was visualized by phosphorimagery. The numbers following the restriction enzyme names indicate the number of recognition sites that occur within the cor1.1 reading frame. This high stringency Southern analysis indicates the presence of at least six alleles in the opium poppy genome.

RNA gel blot analysis indicated the presence of a very weakly hybridizing RNA of approximately 1.4 kb in the poppy leaf, root and stem of a mature plant 2 days after petal fall ( Fig. 6). Since cor1 transcript was apparently present at very low levels, further analysis was undertaken by nested RT-PCR. Morphinan alkaloids begin to accumulate rapidly in poppy seedlings 4–7 days after germination ( Rush et al. 1985 ; Wieczorek et al. 1986 ). An analysis of codeinone reductase enzyme activity and transcript accumulation showed that enzyme activity is at 310 pkat g–1dry tissue weight (dwt) already at day 7 after germination ( Table 1). This activity remains at that level throughout a 3 week growth period then decreases to 150 pkat g–1dwt by the eighth week. In comparison, opium poppy cell suspension culture also contains 330 pkat mg–1dwt enzyme activity. Transcript was detected by RT-PCR for cor1.1–cor1.4 at all developmental stages. Since two PCR amplifications were necessary in order to detect cor1 transcript, a comparative quantitation was not undertaken.

Figure 6.

RNA gel blot analysis of distribution of codeinone reductase transcript in a mature opium poppy.

The gel blot was prepared from RNA isolated from leaf mid rib, lateral root and 12 cm of stem tissue directly beneath the receptacle of an opium poppy plant 2 days after petal fall. 50 μg of total RNA were loaded per gel lane. The RNA was hybridized to cor1.1 full-length cDNA and was visualized by phosphorimagery.

Table 1.  Analysis of codeinone reductase enzyme activity and transcript in developing opium poppy and in plant suspension culture Thumbnail image of

The distribution of codeinone reductase enzyme activity and transcript was also investigated in mature opium poppy plants 2 days after petal fall. On a dry tissue weight basis, most activity was present in the capsule (730 pkat g–1dwt), then the lateral root (560 pkat g–1dwt) followed by stem and leaf lamina ( Table 2). Again, no differences could be found in the distribution pattern of the four isoforms by RT-PCR.

Table 2.  Analysis of codeinone reductase enzyme activity and transcript in opium poppy 2 days after petal fall Thumbnail image of

Functional characterization of the codeinone reductase alleles

The four codeinone reductase–calmodulin-binding peptide fusion proteins were purified from E. coli lysates in one step with a calmodulin affinity column. Beginning with 250 mg total protein in the bacterial extract, 10.5 mg codeinone reductase with a specific activity of 5.2 nkat mg–1protein could be obtained in 73% yield. Aliquots from a typical purification analyzed by SDS-PAGE are shown in Fig. 7. Codeinone reductase purified by this method is nearly homogeneous and demonstrated properties that compared favorably to those of the native enzyme ( Lenz & Zenk 1995b).

Figure 7.

SDS-PAGE analysis of fractions from the purification of codeinone reductase fusion protein from E. coli.

Codeinone reductase was expressed as a C-terminal fusion with a 25 amino acid calmodulin-binding peptide in E. coli BL21(DE3)pLysS. Protein bands were visualized with Coomassie brilliant blue R-250. Lane 1, 15 μg crude protein from an extract of E. coli BL21(DE3)pLysS containing the codeinone reductase cDNA before IPTG induction; lane 2, 10 μg crude protein from an extract of E. coli BL21(DE3)pLysS containing the codeinone reductase cDNA 3 h after IPTG induction; lane 3, 5 μg protein from the calmodulin affinity chromatography eluate after concentration using a Centriprep 30 column (Amicon); lane 4, Rainbow Marker protein standards (Amersham). The arrow indicates the position of codeinone reductase fusion protein.

The temperature optimum, pH optimum and Km values for codeinone, codeine, NADPH and NADP were determined for each of the isoforms ( Table 3). Significant differences in these values were not found. For all isoforms, the optimum temperature for reduction (physiologically forward reaction) was 28°C and for oxidation (physiologically reverse reaction) was 30°C, the pH optimum for reduction was 6.8 and for oxidation was 9.0. The isoforms were also tested for their ability to transform morphinan alkaloids structurally related to codeinone and codeine. The reductive reaction with NADPH as co-factor functions with morphinone, hydrocodone and hydromorphone as substrate. The oxidative reaction with NADP as co-factor functions with morphine and dihydrocodeine as substrate. The Km values for, and structures of, these additional substrates with COR1.3 are shown in Fig. 8. In all cases, the physiologically forward reaction yielded lower Km values than the physiologically reverse reaction, with codeinone having the lowest Km value at 48 μm. No differences in temperature or pH optimum were observed whether codeinone or morphinone were used as substrate in the assay. NADH could not substitute for NADPH with any of the isoforms. Tritium was enzymatically transferred to codeinone from [4R-3H]NADPH, but not from [4S-3H]NADPH, indicating that codeinone reductase stereospecifically abstracts the pro-R hydrogen from the co-factor.

Table 3.  Comparison of properties of codeinone reductase isoforms
Amino acid identity (%)100959696
Km codeinone (μm) 58624850
Km NADPH (μm) 180220205197
Km codeine (μm) 220200187140
Km NADP (μm) 53584555
Calculated Mr35 80835 70435 79735,705
Calculated pI6.255.716.326.33
Figure 8.

Chemical structures of alkaloids serving as substrates for codeinone reductase.

Of the 26 potential substrates tested, only seven were transformed by codeinone reductase. The names of the untransformed compounds are given in the Results section. Codeinone is the physiological substrate for this enzyme in most, if not all, varieties of opium poppy. Morphinone also serves as a physiological substrate in Tasmanian varieties. The Km values provided for those seven substrates were determined for COR1.3.

The reduction of codeinone to codeine is the last of three NADPH-dependent reductions that occur along the biosynthetic pathway leading from (S)-reticuline to morphine in opium poppy. The two other potential substrates for reduction, the 1,2-dehydroreticulinium ion and salutaridine ( Fig. 1), or for the physiologically reverse reaction, salutaridinol and (R)-reticuline, were tested as substrates with the codeinone reductase isoforms. None of these alkaloids served as substrate indicating that codeinone reductase can catalyze only one reductive step in morphine biosynthesis. In addition, the following analogs were also inactive: (S)- and (R)-norreticuline (S)-reticuline and norcodeine.

Since codeinone reductase showed sequence similarity to several members of the aldo/keto reductase family, a series of substrates were tested to reflect members from carbohydrate and steroid metabolism. d-Sorbitol-6-phosphate, d-xylose, prostaglandin D1, 5-androstene-3β,17β-diol, 5α-androstan-17β-ol-3-one, 5α-cholestane-3β-ol, β-estradiol, cyclohexanone and 2-cyclohexene-1-one were not transformed by codeinone reductase. The highest amino acid sequence identity (53%) was, however, to the reductase subunit of the 6′-deoxychalcone synthase complex from soybean ( Welle et al. 1991 ). In order to test for a functional evolutionary relationship between isoflavonoid and alkaloid anabolism, codeinone reductase was analyzed for the ability to substitute for the reductase in the formation of 6′-deoxychalcone in co-action with either native chalcone synthase or native stilbene synthase from Pinus sylvestris. In the presence of 4-coumaryl-CoA, malonyl-CoA, NADPH, chalcone synthase and codeinone reductase or cinnamoyl-CoA, malonyl-CoA, NADPH, stilbene synthase and codeinone reductase, formation of product was not observed. Likewise, the reductase of the 6′-deoxychalcone synthase complex could neither reduce codeinone in the presence of NADPH nor oxidize codeine in the presence of NADP.


The opium poppy has been bred by mankind for therapeutic use since antiquity ( Zenk 1994). Only two species of Papaver, P. somniferum and P. setigerum, produce the opiate alkaloid morphine. It is possible that the opium poppy with which we are familiar today is the result of centuries of breeding P. setigerum for optimal physiological effects (i.e. morphine production) much in the same way that our present day cereal crops were derived from wild varieties that today may only vaguely resemble one another. Within the realm of this view, the ability to produce morphine has evolved only once in the plant kingdom. Although it can be converted through acetylation to the substance of abuse heroin, morphine is still one of the most important natural product pharmaceuticals derived from plants. It remains one of the most powerful analgesic substances known and also serves as the commercial source of the antitussive and analgesic codeine.

Morphine biosynthesis in the opium poppy has been intensely investigated for decades. It is known that morphinan alkaloids accumulate in latex vesicles within poppy laticifers. Laticifer ultrastructure and differentiation has been analyzed and related to opium alkaloid accumulation ( Kutchan et al. 1985 ; Kutchan et al. 1986 ; Nessler & Mahlberg 1977; Nessler & Mahlberg 1978; Roberts et al. 1983 ; Rush et al. 1985 ), but the precise site of morphine alkaloid biosynthesis is not yet elucidated. To this end, we have isolated cDNAs that encode an enzyme specific to morphine biosynthesis, codeinone reductase. Four full-length reading frames and two partial clones were isolated that represent six alleles from a gene family that may have at least 10 members. An analysis of RNA and enzyme activity from various stages of developing opium poppy seedlings and roots, stem, leaf and capsule of mature poppy plants indicated that transcript from these alleles is present throughout the plant at all developmental stages, and that the highest total enzyme activity is localized to the capsule after petal fall. These results would suggest that morphine biosynthesis occurs in all major plant organs starting within the first 7 days after seed germination. This biosynthesis continues throughout the lifecycle of this annual with the highest biosynthetic activity taking place in the capsule after petal fall, consistent with the amount of biosynthetic enyzme present. The amount of extractable RNA remained high in the capsule until 3 days after petal fall, after which time the quantity of extractable RNA decreased rapidly.

A biochemical analysis of four functionally expressed alleles, cor1.1–cor1.4, revealed no significant differences in the temperature or pH optima, Km values or substrate specificity of the isoforms. All isoforms were able to reduce morphinone to morphine. In 1983, the occurrence of a new opium alkaloid called oripavine was reported from a Tasmanian variety of opium poppy ( Nielsen et al. 1983 ). This discovery led to a proposal for an alternative biosynthetic pathway leading from thebaine to morphine that would not proceed through codeinone and codeine, but rather through oripavine and morphinone ( Fig. 9) ( Brochmann-Hanssen 1984). Although the variety of poppy that was used in this study contains no oripavine, suggesting differences in the demethylases between the Tasmanian and Munich varieties, the codeinone reductase isoforms that are expressed retain the capacity to catalyze this alternative reduction.

Figure 9.

Proposed alternative biosynthetic pathway leading from thebaine to morphine in opium poppies from Tasmania.

This alternative biosynthetic pathway was proposed after oripavine was discovered in Tasmanian varieties of opium poppy ( Brochmann-Hanssen 1984). Codeinone reductase from non-Tasmanian varieties can also catalyze the reduction of morphinone to morphine ( Lenz & Zenk 1995b). COR1.1–COR1.4 each catalyzed this reduction with equivalent specific activity. The demethylation of thebaine and codeine is thought to be catalyzed by cytochrome P450-dependent enzymes.

An NADH-dependent reductase, morphinone reductase, from Pseudomonas putida M10 has been identified that can reduce codeinone to hydrocodone and morphinone to hydromorphone ( French & Bruce 1994; French et al. 1995 ). The reverse (oxidative) reaction was not observed. In comparison, none of the P. somniferum codeinone reductase isoforms, which catalyze both the oxidative and reductive reactions, could utilize NADH as co-factor. Likewise, the flavoprotein morphinone reductase was not active in the presence of NADPH. A second enzyme from P. putida M10, the NADP-dependent morphine 6-dehydrogenase, can oxidize morphine to morphinone and codeine to codeinone ( Bruce et al. 1991 ). In the presence of NADPH, morphine 6-dehydrogenase can reduce codeinone to codeine. P. somniferum codeinone reductase is 38% similar to morphine 6-dehydrogenase, which by sequence comparison is also related to aldo/keto reductases. In addition, a 3β-hydroxysteroid dehydrogenase from Pseudomonas testosteroni, a member of the short-chain alcohol dehydrogenase family that utilizes NADH as co-factor, can accept a broad range of substrates including steroids and the alkaloids morphine and codeine as substrate ( Liras et al. 1975 ; Yin et al. 1991 ). Codeinone reductase is 21% similar to 3β-hydroxysteroid dehydrogenase from P. testosteroni, but only after the introduction of 10 gaps into the amino acid sequence. Although codeinone reductase and these two bacterial reductases share overlapping substrate specificity, the amino acid sequence comparison suggests that codeinone reductase potentially evolved from plant aldo/keto reductases. The very similar phenylpropanoid gene encoding 6′-deoxychalcone synthase could also have arisen from the aldo/keto reductase family. A cor1/6′-deoxychalcone synthase- related gene, cor2, was also isolated from opium poppy and heterologously expressed, but demonstrated neither codeinone reductase nor 6′-deoxychalcone synthase enzyme activity. 5-Deoxyflavones and 5-deoxyisoflavones, the formation of which is dependent upon the presence of 6′-deoxychalcone synthase, are present in select species such as soybean, Phaseolus vulgaris and peanut ( Welle et al. 1991 ). There are, however, no reports in the literature of these classes of deoxyflavonoids occurring in opium poppy. The 53% sequence identity between codeinone reductase and 6′-deoxychalcone synthase suggests that they were derived from a common aldo/keto reductase gene, possibly from primary metabolism. The two enzymes do not show, however, measurable overlapping substrate specificity. The ability to synthesize 5-deoxyflavonoids could have evolved in opium poppy, but one has to speculate that a unique selection pressure led to the formation of morphine alkaloids instead.

At six alleles, the number of alleles present in the opium poppy genome was suprisingly large. This is the first time that we observe more than two or three genes encoding an enzyme specific to alkaloid biosynthesis. No obvious biochemical or expression pattern differences could be observed in this study. More refined methods of expression analysis will await the isolation of the promoters driving the expression of the various alleles. It could be, however, that this six-member gene family is simply the result of gene amplification due to tens of centuries of breeding opium poppy for optimal physiological effects in humans that are associated with morphine accumulation.

Experimental procedures

Plant material

Cell suspension cultures of the opium poppy Papaver somniferum were provided by the cell culture laboratory of this department. Cultures were routinely grown in 1 litre conical flasks containing 400 ml of Linsmaier-Skoog medium ( Linsmaier & Skoog 1965) over 7 days at 23°C on a gyratory shaker (100 r.p.m.) in diffuse light (750 lux). Differentiated opium poppy plants were grown outdoors in Upper Bavaria. Seedlings were grown on substrate from 7 to 56 days in a greenhouse at 20°C, 65% relative humidity and 12 h cycles of light and dark.

Purification of native enzyme and amino acid sequence analysis

A mixture of codeinone reductase isoforms was purified from opium poppy cell suspension cultures exactly according to Lenz & Zenk (1995b). The purified enzyme preparation was subjected to SDS/PAGE to remove traces of impurities and the Coomassie brilliant blue R-250-visualized band representing codeinone reductase was digested in situ with endoproteinase Lys-C as reported previously ( Dittrich & Kutchan 1991; Eckerskorn & Lottspeich 1989). The peptide mixture thereby obtained was resolved by reversed phase HPLC (column, Merck Lichrospher RP18; 5 μm (4 × 125 mm); solvent system (A) 0.1% trifluoroacetic acid (B) 0.1% trifluoroacetic acid/60% acetonitrile; gradient of 1% per min; flow rate of 1 ml min–1) with detection at 206 nm. Microsequencing of seven of the peptides thus purified was accomplished with an Applied Biosystems model 470 gas-phase sequencer.

Generation of partial cDNAs from opium poppy

Partial cDNAs encoding codeinone reductases from opium poppy were produced by PCR using cDNA produced by reverse transcription of total RNA isolated from 3- to 5-day-old suspension cultured cells. DNA amplification using either Taq or Pfu polymerase was performed under the following conditions: 4 min at 94°C, 35 cycles of 94°C, 30 sec; 45°C, 30 sec; 72°C, 1 min. At the end of 35 cycles, the reaction mixtures were incubated for an additional 5 min at 72°C prior to cooling to 4°C. Re-amplification of DNA using nested primers was performed as above, but the primer annealing temperature was raised from 45 to 55°C. The amplified DNA was then resolved by agarose gel electrophoresis, the bands of approximately the correct size were isolated and subcloned into pGEM-T (Promega) prior to nucleotide sequence determination. The specific sequences of the oligodeoxynucleotide primers used are given in the Results section.

Generation of full-length cDNAs

The sequence information requisite to the generation of full-length cDNAs was derived from the nucleotide sequences of the partial cDNAs generated as described in the Results section. The complete nucleotide sequence of one reading frame was determined using codeinone reductase specific oligodeoxynucleotide primers in 5′- and 3′-RACE-PCR experiments with a MarathonTM cDNA amplification kit (Clontech). RACE-PCR was performed using the PCR cycles described above. The amplified DNA was then resolved by agarose gel electrophoresis and the band of the approximate expected size was isolated, subcloned into pGEM-T and sequenced.

Nested primer pairs were then used to generate full-length clones for heterologous expression by RT-PCR using opium poppy cell suspension culture RNA as template. The final primers used in clone amplification contained the restriction endonuclease recognition sites NheI and BamHI that were appropriate for subcloning directly into the pCAL-c (Stratagene) expression vector. The specific sequences of these primers are given in the Results section. RT-PCR was carried out using the PCR cycles given above. The amplified DNA was then resolved by agarose gel electrophoresis and the band of the correct size (972 bp) was excised and isolated for further subcloning into the expression vector.

Heterologous expression and enzyme purification

Full-length cDNAs generated by RT-PCR were ligated into p-CAL-c and transformed into the E. coli strain BL21(DE3)pLysS. For enzyme assays, single colonies were picked and grown in 3 ml Luria-Bertani medium containing 100 μg ml–1ampicillin at 37°C to an OD590 of 0.8. For protein purification, single colonies were picked and grown in 1 l Luria-Bertani medium containing 100 μg ml–1ampicillin at 37°C to an OD590 of 1.8. Cells were collected by centrifugation for 5 min at 4000 g and 4°C. The bacterial pellet was resuspended in either 0.1 m potassium phosphate buffer pH 6.8 for the reduction of codeinone or 0.1 m glycine buffer pH 9 for the oxidation of codeine. The bacterial pellet from a 3 ml culture was resuspended in 0.5 ml buffer and from a 1 l culture in 100 ml buffer. The cells were ruptured by sonication. Cellular debris was removed by centrifugation for 5 min at 4000 g and 4°C and the supernatant used directly for either affinity chromatography purification using the AffinityTM Protein Expression and Purification System according to the manufacturer’s instructions (Stratagene) or for enzyme activity measurements according to Lenz & Zenk (1995b).

Enzyme assay and product identification

The oxidative and reductive reactions catalyzed by codeinone reductase were assayed according to Lenz & Zenk (1995b). The oxidation of codeine to codeinone by heterologously expressed enzyme in a crude bacterial extract was used for large-scale production of enzymic product for structure elucidation by 1H NMR, 13C NMR and mass spectrometry. A typical assay consisted of 15 μmol codeine, 25 μmol NADP, 1.8 mmol glycine buffer pH 9.0 and 100 pkat codeinone reductase in a total volume of 10 ml. Incubation was allowed to proceed for 2 h under mild agitation at 30°C before extraction twice with two volumes (20 ml) of CHCl3. The volume of the combined organic phase was reduced in vacuo and resolved by semi-preparative HPLC using the following gradient: (column, Knauer Lichrospher 100 RP18 endcapped; 5 μm (16 × 250 mm); solvent system (A) 97.99% (v/v) H2O, 2% CH3CN, 0.01% (v/v) H3PO4 (B) 1.99% (v/v) H2O, 98% CH3CN, 0.01% H3PO4; gradient: 0–9 min 0–8% B, 9–24 min 8% B, 24–45 min 8–25% B, 45–75 min 25% B, 75–75.3 min 25–0%B, 75.3–90 min 0%B; flow 4.5 ml min–1) with detection at 204 nm using authentic codeine (retention time, 38 min) and codeinone (retention time, 49 min) as reference materials. In this manner, 32 mg of codeine were converted and purified to 10 mg of codeinone.

Codeinone –δH (360 MHz, CDCl3) 1.87 (1H, dd, J15a/15e 12.2, J15e/16a 3.1, H-15e), 2.08 (1H, ddd, J15a/16a 12.3, J15a/16e 4.5, J15a/15e 12.2, H-15a) 2.29 (1H, ddd, J15a/16a 12.3, J15e/16a 3.1, J16a/16e 11.8, H-16a), 2.35 (1H, dd, J10a/10e 18.5, J9/10a 5.9, H-10a), 2.47 (3H, s, CH3N-), 2.63 (1H, dd, J16a/16e 11.8, J15a/16e 4.5, H-16e), 3.12 (1H, d, J10a/10e 18.5, H-10e), 3.21 (1H, m, H-14), 3.43 (1H, m, H-9), 3.85 (3H, s, CH3O-), 4.71 (1H, s, H-5), 6.09 (1H, dd, J7/8 10.1, J7/14 2.8, H-7), 6.62 (1H, d, J1/2 8.3, H-1), 6.66 (1H, dd, J7/8 10.1, J8/14 1.5, H-8), 6.68 (1H, d, J1/2 8.3, H-2); δC (90.6 MHz, CDCl3) 20.5 (C-10), 33.8 (C-15), 41.3 (C-14), 42.8 (NMe), 43.0 (C-13), 46.8 (C-16), 56.8 (OMe), 59.1 (C-9), 88.0 (C-5), 114.8 (C-2), 119.9 (C-1), 125.7 (C-11), 128.9 (C-12), 132.6 (C-7), 142.6 (C-3), 144.9 (C-4), 148.7 (C-8), 194.4 (C-6); EI-MS (70 eV), m/z 297 (M+, 100%), 282 (8), 268 (9), 254 (8), 238 (9), 229 (23), 214 (17), 188 (15), 165 (11), 152 (13), 139 (16), 128 (22), 115 (41).

General methods

Total RNA was isolated and RNA gels were run and blotted as described previously ( Pauli & Kutchan 1998). Genomic DNA was isolated and DNA gels were run and blotted according to Bracher & Kutchan (1992). cDNA clones were labeled by random-primed labeling with [α-32P]dCTP and oligodeoxynucleotides were end-labeled with [γ-32P]ATP. Hybridized RNA on Northern blots and DNA on Southern blots were visualized with a Raytest BAS-1500 phosphorimager. The entire nucleotide sequence on both DNA strands of full-length cDNA clones in either pGEM-T or pCAL-c was determined by dideoxy cycle sequencing using internal DNA sequences for the design of deoxyoligonucleotides as sequencing primers.


We thank, as always, Dr Friedrich Lottspeich (Max-Planck-Institut für Biochemie, Martinsried) for the amino acid sequence determination and Professor Joachim Schröder, Universität Freiburg, for help with the 6′-deoxychalcone synthase enzyme assays. This work was supported by SFB 369 of the Deutsche Forschungsgemeinschaft, Bonn and Fonds der Chemischen Industrie, Frankfurt.


  1. GenBank accession numbers AF108432 (cor1.1), AF108433 (cor1.2), AF108434 (cor1.3), AF108435 (cor1.4), AF108436 (cor1.5), AF108437 (cor1.6) and AF108438 (cor2).