Enzymatic oxidations in the biosynthesis of complex alkaloids


*For correspondence (fax +49 89 5902611;
e-mail kutch@lrz.uni-muenchen.de).


The biosynthesis of complex alkaloids in plants involves enzymes that, due to high substrate specificity, appear to have evolved solely for a role in secondary metabolism. At least one class of these enzymes, the oxidoreductases, catalyze transformations that are in some cases difficult to chemically mimick with an equivalent stereo- or regiospecificity and yield. Oxidoreductases are frequently catalyzing reactions that result in the formation of parent ring systems, thereby determining the class of alkaloid that a plant will produce. The oxidoreductases of alkaloid formation are a potential target for the biotechnological exploitation of medicinal plants in that they could be used for biomimetic syntheses of alkaloids. Analyzing the molecular genetics of alkaloid biosynthetic oxidations is requisite to eventual commercial application of these enzymes. To this end, a wealth of knowledge has been gained on the biochemistry of select monoterpenoid indole and isoquinoline biosynthetic pathways, and in recent years this has been complemented by molecular genetic analyses. As the nucleotide sequences of the oxidases of alkaloid synthesis become known, consensus sequences specific to select classes of enzymes can be identified. These consensus sequences will potentially facilitate the direct cloning of alkaloid biosynthetic genes without the need to purify the native enzyme for partial amino acid sequence determination or for antibody production prior to cDNA isolation. The current state of our knowledge of the biochemistry and molecular genetics of oxidases involved in alkaloid biosynthesis is reviewed herein.


Early research on the enzymes of secondary metabolite biosynthesis using crude protein extracts prepared from differentiated plant material encountered many difficulties such as high content of phenols and very low levels of the biosynthetic enzyme activities. This was particularly true for the alkaloid field where the early precursors of alkaloid biosynthesis are often labile phenolic compounds and where alkaloids accumulate only very slowly over a period of several months to several years. Identification and purification of enzymes of alkaloid biosynthesis were greatly facilitated by the use of plant cell cultures as a source tissue ( Zenk 1991). Plant cell cultures have a much compressed growth and alkaloid production period compared to differentiated plants, and in numerous cultures alkaloid accumulation can be additionally increased by treatment of a cell culture with elicitor preparations ( Eilert et al. 1985 ;Gundlach et al. 1992 ;Schumacher et al. 1987 ). Overall, the development of medicinal plant cell culture techniques has led to the identification of more than 80 enzymes of alkaloid biosynthesis (reviewed in Kutchan 1998).

Enzymatic oxidations in alkaloid biosynthesis were previously assigned to enzymes such as copper-containing laccases and phenolases or the hemoprotein peroxidases. These enzymes are relatively abundant in the plant cell and have broad substrate- as well as limited regio- and stereospecificity. Many low molecular weight substances that contain a phenol or catechol moiety are indeed non-specifically transformed by one or more of these enzyme types. Formation of the complex parent ring systems of many alkaloids from simple aromatic amino acid precursors such as l-tyrosine and l-tryptophan, however, involves multiple oxidations catalyzed in a stereo- and regiospecific manner, suggesting that only specific oxidases are involved in biosynthesis in vivo.

In more recent years, we have come to learn that oxidations in alkaloid biosynthesis are catalyzed by microsomal cytochrome P-450-dependent enzymes, 2-oxoglutarate-dependent dioxygenases and flavoproteins. These have been described as highly substrate specific enzymes that appear to be dedicated to alkaloid biosynthesis. In phenylpropanoid anabolism, a fundamentally new type of protein, a dirigent protein, is responsible for the stereoselectivity of radical coupling in lignan formation. Whereas the actual oxidation can be carried out by a number of enzymic and non-enzymic single electron carriers, such as laccases and riboflavin derivatives, the dirigent protein functions as a scaffold to hold the substrate in the correct configuration to achieve stereospecific coupling ( Davin et al. 1997 ).

Our current understanding of the biochemistry and gene structure of oxidoreductases involved in the biosynthesis of select plant alkaloids is the topic of this review.

Tetrahydrobenzylisoquinoline-derived alkaloids

Cytochrome P-450 monooxygenases of berberine, macarpine and morphine biosynthesis

Cytochromes P-450 catalyze many of the oxidations in the biosynthesis of tetrahydrobenzylisoquinoline-derived alkaloids. Unlike the hepatic cytochromes P-450 that oxidize a broad series of xenobiotics, the plant alkaloid biosynthetic cytochromes P-450 appear to be quite substrate specific catalyzing monooxygenations that lead to either hydroxylated or demethylated substrate or to formation of a methylenedioxy bridge.

The first cytochrome P-450 dependent hydroxylation of tetrahydrobenzylisoquinoline alkaloid biosynthesis is the conversion of (S)-N-methylcoclaurine to (S)-3′-hydroxy-N-methylcoclaurine (6-O-methyllaudanosoline) along the biosynthetic pathway that leads from two molecules of l-tyrosine to the branchpoint intermediate of isoquinoline alkaloid biosynthesis (S)-reticuline ( Pauli & Kutchan 1998) ( Fig. 1). (S)-Reticuline can then undergo a variety of oxidative transformations that result in the formation of the different parent ring structures of isoquinoline alkaloids such as berberine, the benzo[c]phenanthridine macarpine and the phenanthrine morphine. (S)-N-Methylcoclaurine 3′-hydroxylase (CYP80B1) was identified by molecular cloning from the California poppy Eschscholzia californica. This is a first example of an enzyme of alkaloid biosynthesis being discovered by a direct molecular genetic as opposed to a reverse genetic approach. (S)-N-Methylcoclaurine 3′-hydroxylase demonstrates high specificity for (S)-N-methylcoclaurine. Of 16 potential substrates tested, only (S)-N-methylcoclaurine was oxidized. Two alleles were isolated and the two isozymes, differing in only five amino acids, were functionally expressed in insect cells (Spodoptera frugiperda Sf9) and showed similar physical characteristics with a Km of 15 μm for the substrate (S)-N-methylcoclaurine. Co-infection of the insect cells with two baculovirus constructions that contained a cDNA encoding either (S)-N-methylcoclaurine 3′-hydroxylase or the E. californica cytochrome P-450 reductase ( Rosco et al. 1997 ) resulted in a 10-fold increase in this specific hydroxylase activity in isolated microsomes (130 versus 13 pkatal 300 μg–1 microsomal protein (approximately 0.5 pmol P-450)). This indicates that either reductase is limiting in the insect cell microsomes or that the transfer of electrons proceeds more efficiently with a plant as opposed to insect reductase.

Figure 1.

Proposed scheme of the central biosynthetic pathway leading from l-tyrosine to a variety of species-specific tetrahydrobenzylisoquinoline-derived alkaloids.

At least two branches can occur at the position of (S)-N-methylcoclaurine and (S)-reticuline, depending on the plant species. Species-specific oxidative enzymes are largely responsible for the transformations that form the different parent ring systems of the isoquinoline alkaloids. CYP80 A1, berbamunine synthase; CYP80B1, (S)-N-methylcoclaurine 3′-hydroxylase; BBE, berberine bridge enzyme; dashed arrow indicates more than one enzymatic transformation.

Cell suspension cultures of E. californica produce the antimicrobial benzo[c]phenanthridine alkaloids sanguinarine, chelirubine and macarpine when challenged with elicitor. The nature of this elicitor can vary widely from preparations of fungal, bacterial or yeast origin to barbiturates or to methyl jasmonate and synthetic derivatives thereof ( Blechert et al. 1997 ;Gundlach et al. 1992 ;Haider et al. 1997 ;Schumacher et al. 1987 ). The specific activity of the final six reactions mediated by cytochromes P-450 of macarpine biosynthesis ( Fig. 2) [(S)-cheilanthifoline synthase, EC1.1.3.33; (S)-stylopine synthase,; (S)-cis-N-methylstylopine 14-hydroxylase,; protopine 6-hydroxylase,; dihydrosanguinarine 10-hydroxylase,; and dihydrochelirubine 12-hydroxylase,] is increased in elicited cell cultures with respect to control cultures ( Blechert et al. 1995 ;De-Eknamkul et al. 1992; Kammerer et al. 1994;Tanahashi & Zenk 1990). The level of (S)-N-methylcoclaurine 3′-hydroxylase transcript is also increased in response to methyl jasmonate and barbiturates, reaching a maximum 6–9 h after the addition of elicitor ( Pauli & Kutchan 1998). This indicates that the enzyme activities of all seven cytochromes P-450 of macarpine biosynthesis are elicitor inducible. Sanguinarine biosynthesis is likewise induced in response to elicitor addition in cell suspension cultures of the opium poppy Papaver somniferum ( Eilert et al. 1985 ). (S)-N-Methylcoclaurine 3′-hydroxylase transcript level also increases as expected in P. somniferum in response to methyl jasmonate treatment (F.-C. Huang and T.M. Kutchan, unpublished results).

Figure 2.

Proposed scheme of the biosynthetic pathway that leads from (S)-reticuline to the antimicrobial benzo[c]phenanthridine alkaloids in various species of the Papaveraceae.

Six cytochromes P-450, one covalently flavinylated oxidase and one copper-containing enzyme are necessary for the biosynthesis of the most highly oxidized benzophenanthridine alkaloid, macarpine. Dashed arrow indicates more than one enzymatic transformation.

The biosynthesis of the antibiotic alkaloids berberine and macarpine involves the cytochromes P-450 canadine synthase [EC1.1.3.36] ( Fig. 3a), (S)-cheilanthifoline synthase and (S)-stylopine synthase ( Fig. 2) that each convert an aromatic methoxy moiety into a methylenedioxy bridge ( Bauer & Zenk 1991;Rueffer & Zenk 1994). Methylenedioxy bridge formation is thought to proceed by hydroxylation of an ortho-methoxyphenol resulting in formation of a hemiformal that, via either a radical or cationic mechanism, eliminates water during or prior to ring closure ( Fig. 3b) ( Barton et al. 1963 ;Bauer & Zenk 1991). A mechanism of this type is in contrast to most aliphatic methoxyl cytochrome P-450 mediated hydroxylations that in hepatic xenobiotic metabolism, for example, result in demethylation ( Kronbach 1991).

Figure 3.

(a) Biosynthesis of the antibacterial alkaloid berberine from (S)-reticuline in various members of the Berberidaceae and Ranunculaceae. Three of the four enzymes in this short pathway are oxidases, two covalently flavinylated and one cytochrome P-450.

(b) Proposed mechanism of methylenedioxy bridge formation cataylzed by select cytochromes P-450.

A third type of cytochrome P-450 of isoquinoline alkaloid biosynthesis, in addition to the hydroxylases and methylenedioxy bridge forming enzymes, are the demethylases in the biosynthesis of morphine, the opiate analgesic produced by P. somniferum ( Fig. 4). Although the activities of the enzymes that catalyze enolether cleavage of thebaine and demethylation of codeine have not yet been detected in vitro, these two enzymes may belong to the cytochrome P-450 family, analogous to hepatic demethylation of codeine to morphine by the cytochrome P-450 hydroxylase CYP2D6 ( Wilhelm & Zenk 1997).

Figure 4.

Proposed scheme of the biosynthetic pathway that leads from (R)-reticuline to analgesic opiate alkaloid morphine in P. somniferum.

The cytochrome P-450 salutaridine synthase catalyzes the C-12 to C-13 phenol coupling of (R)-reticuline that leads to formation of the phenanthrine ring system found in salutaridine. Two additional cytochromes P-450 are thought to demethylate thebaine and codeine. Dashed arrow indicates more than one enzymatic transformation.

Cytochrome P-450 oxidases in phenol coupling reactions

It has been postulated that the mechanism of phenol coupling in alkaloid biosynthesis proceeds by formation of phenolate radicals generated through oxidative attack followed by radical pairing to form either an inter- or intramolecular C-O or C-C bond ( Barton & Cohen 1957). In the past, enzymatic phenol-coupling reactions were frequently attributed to the action of laccases, phenolases and peroxidases ( Kametani et al. 1970, 1973 ). These oxidases have broad substrate specifities and are relatively predominant in plants. The abundance of these unspecific oxidoreductases apparently masked reactions that today are known to be catalyzed by highly substrate specific cytochromes P-450 ( Zenk et al. 1989 ). There are now several examples of phenol coupling cytochromes P-450 from alkaloid biosynthesis. The first that was identified was salutaridine synthase [EC1.1.3.35] from P. somniferum cell suspension culture and plants. This enzyme catalyzes the stereo- and regiospecific intramolecular carbon-carbon phenol coupling that transforms the benzyltetrahydroisoquinoline (R)-reticuline into the phenanthrine alkaloid salutaridine along the biosynthetic pathway that leads to morphine ( Fig. 5a) ( Gerardy 1993). This enzyme differs from most cytochromes P-450 in that it transforms substrate without a concomitant incorporation of oxygen into the product. It functions therefore as an oxidase rather than as a monooxygenase. In the P. somniferum plant, salutaridine synthase was found to occur in roots, shoots, capsule and leaf vein. Enzyme activity was noticeably missing from latex. This is of interest in the opium poppy because the major site of morphine accumulation is in the latex, the oxidized, dried form of which is the brown substance referred to as opium.

Figure 5.

Comparison of the (a) C-O phenol coupling reaction catalyzed by the cytochrome P-450 berbamunine synthase along the biosynthetic pathway leading to the bisbenzylisoquinoline alkaloids and the (b) C-C phenol coupling reaction catalyzed by the cytchrome P-450 salutaridine synthase along the biosynthetic pathway leading to the morphinan alkaloids.

A second phenol coupling cytochrome P-450 that has been identified is berbamunine synthase [EC1.1.3.34]. This enzyme catalyzes the intermolecular carbon-oxygen coupling of one molecule of (R)-N-methylcoclaurine and one molecule of (S)-N-methylcoclaurine to form the bisbenzylisoquinoline alkaloid (R,S)-berbamunine or two molecules of (R)-N-methylcoclaurine to form (R,R)-guattegaumerine in cell suspension cultures of Berberis stolonifera ( Fig. 5b) ( Stadler & Zenk 1993). The best known member of the bisbenzylisoquinolines is probably the muscle relaxant (+)-tubocurarine from Chondodendron tomentosum, which in the form of tube-curare is used as an arrow poison by South American Indian tribes. A cDNA encoding berbamunine synthase has been isolated and functionally expressed in insect cell culture and yeast ( Kraus & Kutchan 1995;Rosco et al. 1997 ). Co-expression of the E. californica cytochrome P-450 reductase with berbamunine synthase in insect cell culture indicated that the amount of plant reductase present exerted an influence on the ratio of the products, although not the total amount of products, that were enzymatically formed ( Rosco et al. 1997 ). Native berbamunine synthase produces the two dimeric products in a ratio of R,S:R,R of 90:10. Reconstitution of purified heterologously expressed berbamunine synthase with Berberis reductase or with porcine reductase resulted in a shift of this product ratio in favor of the R,R-dimer ( Kraus & Kutchan 1995). Increasing the ratio of plant reductase with respect to cytochrome P-450 oxidase also resulted in a shift in the ratio of products formed by berbamunine synthase from R,S:R,R in a ratio of 15:85 in the absence of E. californica reductase to R,S:R,R (37:63) when a two- to fivefold excess of baculovirus containing the E. californica reductase was used for the co-infection. These results indicate that the cytochrome P-450 reductase may influence the binding of substrate to berbamunine synthase. As with salutaridine synthase, there is no concomitant incorporation of oxygen into the bisbenzylisoquinoline products during the C-O phenol coupling reaction catalyzed by berbamunine synthase.

Intramolecular C-C phenol coupling of autumnaline leads to formation of the tropolone ring of the colchicine biosynthetic precursor isoandrocymbine in seeds of the autumn crocus or meadow saffron Colchicum autumnale ( Nasreen et al. 1996 ). Colchicine is used in the treatment of acute gout attacks due to its anti-inflammatory properties. It is also an established inhibitor of microtubule assembly. Although it is too toxic to be used as an anti-neoplastic, it does find use as a mitotic poison in plant breeding. Under colchicine treatment, plant chromosomes can be multiplied without cell division. The resultant polyploidal cells resume cell division after the colchicine is removed and can potentially lead to new plant varieties ( Dewick 1997).

Flavin oxidases in berberine and macarpine biosynthesis

Oxidases containing a covalently bound flavin as cofactor are involved in the biosynthesis of (S)-scoulerine-derived alkaloids ( Fig. 1). The berberine bridge enzyme [EC1.5.3.9] catalyzes ring closure of (S)-reticuline to form the tetrahydroprotoberberine ring system by oxidatively cyclizing the N-methyl group of (S)-reticuline to the berberine bridge carbon (C-8) of (S)-scoulerine thereby consuming one equivalent of O2 and producing one equivalent of H2O2 ( Fig. 2) ( Rink & Böhm 1975;Steffens et al. 1985 ). A berberine bridge enzyme encoding cDNA bbe1 from E. californica has been functionally expressed in yeast and in insect cell culture ( Dittrich & Kutchan 1991;Kutchan et al. 1994 ). Insect cell culture production of the heterologous enzyme was high enough at 15–20 mg l–1 to allow a biochemical characterization. An initial comparison of the translated nucleotide sequence of bbe1 had demonstrated 25% identity to 6-hydroxy- d-nicotine oxidase from Arthrobacter oxidans ( Brandsch et al. 1987 ;Brandsch & Bichler 1985). 6-Hydroxy- d-nicotine oxidase was known to be covalently flavinylated ( Brandsch & Bichler 1991). Most importantly, bbe1 retained the peptide sequence known to contain the histidine residue to which an FAD in 6-hydroxy- d-nicotine oxidase is attached. In addition to this initial indirect evidence, a concentrated solution (6 mg ml–1) of the berberine bridge enzyme was deeply yellow in color. The fluorescence emission spectrum that was obtained was also typical of that for a flavoenzyme. Laser desorption time-of-flight mass spectral measurements confirmed the identification of the flavin cofactor as FAD. The molar ratio of FAD to protein was spectrophotometrically determined to be 1:1.03. The berberine bridge enzyme was found to oxidize seven substrates in addition to (S)-reticuline, including (S)-protosinomenine, (R,S)-crassifoline, (R,S)-3′-hydroxy-N-methylcoclaurine, (R,S)-laudanosoline, (S)-N-methylcoclaurine, (S)-coreximine and (S)-isocoreximine. Mechanistically, the enzyme first oxidizes the N-methyl group of (S)-reticuline to the methylene iminium ion, probably via two single electron abstractions. In the second step, stereospecific ionic ring closure forms the berberine bridge of (S)-scoulerine ( Bjorkland et al. 1995 ;Kutchan & Dittrich 1995).

The induction of the berberine bridge enzyme activity as well as bbe1 transcription in cell suspension culture is well established. An analysis of the time course of the elicitation process using the cDNA clone as a hybridization probe in RNA gel blot experiments revealed that the berberine bridge enzyme transcript reaches maximum levels within 6 hours after addition of a yeast cell wall preparation to E. californica cell suspension cultures. Enzyme activity increases up to 22 h after elicitation and total benzophenanthridine alkaloids continue to accumulate for several days ( Dittrich & Kutchan 1991). This scenario is indicative of de novo transciption as observed for stress-induced phenylpropanoid genes. The bbe1 gene has also been isolated from E. californica and an initial promoter analysis attempted ( Hauschild et al. 1997 ). There are at least two alleles, each containing no intron. The bbe1 gene is induced by methyl jasmonate, which has been shown to stimulate accumulation of low molecular weight compounds in a large number of plant species in culture, including benzophenanthridines in Eschscholzia ( Gundlach et al. 1992 ;Kutchan 1993). Numerous synthetic derivatives of jasmonic acid and various barbiturates can likewise induce the bbe1 gene ( Blechert et al. 1997 ;Haider et al. 1997 ). An elicitor inducible bbe1 gene has also been isolated from P. somniferum ( Facchini et al. 1996 ). Eventual identification of the cis elements and trans factors necessary for elicitor-induced transcription of bbe1 should serve to elucidate the complex benzophenanthridine alkaloid defense response in these plants.

The next flavinylated enzyme along the biosynthetic pathway that leads to protoberberine-derived alkaloids is (S)-tetrahydroprotoberberine oxidase [EC1.3.3.8] ( Fig. 3a) ( Amann et al. 1988 ). In Berberis and Thalictrum species, (S)-tetrahydroprotoberberine oxidase catalyzes the stereospecific oxidation of the C-1 to nitrogen atom bond of (S)-canadine. The intermediate molecule is in turn immediately non-enzymatically oxidized to berberine. Analogous to the berberine bridge enzyme, one equivalent of O2 is consumed and one equivalent of H2O2 is formed during catalysis. (S)-Tetrahydroprotoberberine oxidase has a relatively broad substrate specificity within the tetrahydroisoquinoline class and can act on at least 20 different alkaloids with a Km in the micromolar range, but has an absolute requirement for the (S)-enantiomeric form. The enzyme is very stable in immobilized form. The purified oxidase from Berberis wilsoniae (from the barberry family) demonstrated a half life of 200 days at 25°C when immobilized on activated glass compared to only 4 days for the soluble enzyme ( Amann & Zenk 1987). This broad substrate specificity and high thermal stability when immobilized make (S)-tetrahydroprotoberberine oxidase an enzyme useful for biomimetic syntheses.

Both the berberine bridge enzyme and (S)-tetrahydroprotoberberine oxidase are compartmentalized within a vesicle of a specific gravity of ρ = 1.14 g cm–3 ( Amann et al. 1986 ). The vesicles occur in a broad range of berberine bridge enzyme and (S)-tetrahydroprotoberberine oxidase-containing species both in culture and in intact plants. E. californica, in which only the berberine bridge enzyme occurs, also has this same type of vesicle. Both oxidases have an activity optimum at pH >9, which indicates that the internal pH of the vesicles must be very basic. Such small vesicles are believed to fuse into vacuoles that in turn perhaps fuse to form the central vacuole of the cell where alkaloids such as berberine in Berberis species accumulate. Ultimately the pH drops to the acidic values found within the vacuole, which would lead to the inactivation of both enzymes. This model has, however, not been experimentally proven.

A small family of covalently flavinylated oxidases that span in origin from bacteria to plants to mammals and vary widely in the nature of the substrate that is transformed now begins to emerge. A comparison of the derived amino acid sequences of the enzymes reveals that there are highly conserved amino acid residues ( Fig. 6). One of these regions is the site of covalent attachment of FAD and contains an essential histidine residue ( Brandsch & Bichler 1991;Kutchan & Dittrich 1995). Although the natural substrates of these flavin oxidases appear to be structurally unrelated, it should be possible in the future to isolate additional new members of this enzyme family from plants using a PCR-based homology cloning approach.

Figure 6.

Comparison of the derived amino acid sequences of a series of covalently flavinylated proteins.

The amino acid sequences that are highly conserved are shaded grey. The histidine residue that is the site of covalent attachment of FAD is denoted by an asterisk. This position corresponds to H-104 in EcaBBE1. EcaBBE1, E. californica berberine bridge enzyme (GenBank/EMBL database accession number S65550); PsoBBE1, P. somniferum berberine bridge enyzme ( U59232); BstBBE1, B. stolonifera berberine bridge enzyme ( AF049347); HOX, Chondrus crispus hexose oxidase ( U89770) (Hansen and Stougaard, 1997); MCRA, Streptomyces lavendulae mitomycin C resistance gene mcrA ( L29247) (August et al. 1994); 6-HDNO, A. oxidans 6-hydroxy- d-nicotine oxidase ( M28330); L-GULO, rat liver l-gulonolactone oxidase ( E01923) (Koshizaka et al. 1988).

Copper-containing oxidase of sanguinarine, chelirubine and macarpine formation

Dihydrobenzophenanthridine oxidase [EC1.5.3.12] that aromatizes ring C of dihydrobenzophenanthridine alkaloids was first characterized from elicited cell suspension cultures of E. californica ( Fig. 2) ( Schumacher & Zenk 1988). Dihydrobenzophenanthridine alkaloids are non-planar molecules that lack antimicrobial activity. In fully oxidized form, benzophenanthridines are planar, fluorescent molecules capable of intercalating with DNA and RNA ( Bock et al. 1998 ), thereby possessing antibiotic activity ( Cline & Coscia 1988;Dzink & Socransky 1985). In E. californica, the dihydro-forms are prevalent, where upon elicitation, they are converted to the biologically active forms. Dihydrobenzophenanthridine oxidase has been identified as a copper-containing enzyme from elicited cell suspension cultures of Sanguinaria canadensis ( Arakawa et al. 1992 ). Whether this enzyme is a multicopper ion-containing laccase-type or a single copper ion-containing phenolase-type oxidase is not yet clear. It does, however, demonstrate some degree of substrate specificity within the dihydrobenzophenanthridine family and its occurrence was specific to cell cultures that produce this type of alkaloid. The oxidase was very weakly, if at all, induced in elicited E. californica cell cultures, but was more strongly induced in S. canadensis cell cultures that have been treated with methyl jasmonate ( Ignatov et al. 1996 ).

Monoterpenoid indole and tropane alkaloids

Cytochromes P-450 of ajmaline and vindoline biosynthesis

Cytochromes P-450 play an important role not only in the biosynthesis of isoquinoline alkaloids, but also in the biosynthesis of monoterpenoid indole alkaloids. One of the first cytochromes P-450 that was identified from plants was geraniol/nerol 10-hydroxylase from the Madagascar periwinkle Catharanthus roseus (Madyastha et al. 1976) . Geraniol 10-hydroxylase lies on the pathway that leads to the secoiridoid secologanin that provides the terpenoid moiety of 3α(S)-strictosidine, the precursor to the monoterpenoid indole alkaloids ( Fig. 7) ( Smith 1968). Along the biosynthetic pathway that leads specifically to sapargine- and ajmaline-type alkaloids in Rauwolfia serpentina, a cytochrome P-450 is thought to be involved in the formation of the sarpagan bridge of polyneuridine aldehyde from geissoschizine ( Schmidt & Stöckigt 1995). The mechanism of this aliphatic C-C bond formation as catalyzed by a cytochrome P-450 is not yet understood. Later along the pathway that specifically leads to ajmaline, a cytochrome P-450 stereospecifically hydroxylates vinorine at position 21 to form vomilenine ( Falkenhagen & Stöckigt 1995). This monooxygenase demonstrated a very high substrate specificity so that only slight structural modifications to vinorine, such as imine reduction, resulted in little or no enzymatic hydroxylation ( Falkenhagen et al. 1995 ).

Figure 7.

Proposed scheme of the biosynthetic pathway leading to the monoterpenoid indole alkaloid ajmaline in R. serpentina.

Three cytochromes P-450 are believed to be involved in this pathway. Geraniol 10-hydroxylase and vinorine 21-hydroxylase have been clearly identified as cytochromes P-450. Dashed arrow indicates more than one enzymatic transformation.

A fourth cytochrome P-450 of monoterpenoid indole alkaloid biosynthesis is tabersonine 16-hydroxylase (St-Pierre & De Luca 1995). Tabersonine is a precursor to the Aspidosperma alkaloid vindoline ( Fig. 8), which is a structural component, along with catharanthine, of the chemotherapeutic dimeric indole alkaloid vinblastine produced by C. roseus. The monooxygenase is found both in cell culture and in young leaves, and to a lesser extent in root and flower buds. Three of the six transformations that convert tabersonine into vindoline are hydroxylations. These three introductions of oxygen into the Aspidosperma skeleton are catalyzed by three very different types of enzymes. The first monooxygenation is carried out by a microsomal cytochrome P-450, the second hydroxylation is likely to be hydration of a double bond by a yet unidentified enzyme and the third is catalyzed by a 2-oxoglutarate-dependent dioxygenase. The evolution of these three different modes of oxygenation within such a short biosynthetic pathway is representative of the elegant biochemistry to be found in plant alkaloid metabolism.

Figure 8.

Biosynthesis of vindoline from tabersonine in C. roseus

Three hydroxylations occur that are catalyzed by three different types of enzymes, a cytochrome P-450-dependent monooxygenase, a hydratase and a 2-oxo-glutarate-dependent dioxygenase. Dashed arrow indicates more than one enzymatic transformation.

An oxidoreductase of monoterpenoid indole alkaloid biosynthesis that has been of longstanding interest is that which catalyzes the coupling of vindoline and catharanthine to form vinblastine ( Fig. 8). It has been shown that horseradish peroxidase can couple these two monomers ( Goodbody et al. 1988 ) as well as a number of peroxidase-like enzymes from C. roseus cell suspension culture ( Endo et al. 1988 ). What remains to be demonstrated is whether a highly substrate specific enzyme, analogous in its degree of specificity to other oxidoreductases of alkaloid biosynthesis, catalyzes this coupling reaction in vivo.

2-Oxoglutarate-dependent dioxygenases of scopolamine and vindoline biosynthesis

In addition to the heme-containing cytochromes, the non-heme iron-containing dioxygenases are also prevalent in alkaloid biosynthesis. A 2-oxoglutarate-dependent dioxygenase catalyzes epoxidation of the double bond of hyoscyamine to form the hallucinogenic tropane alkaloid scopolamine of henbane, Hyoscyamus niger ( Fig. 9) ( Hashimoto et al. 1993a ). Hyoscyamine 6β-hydoxylase [EC1.14.11.11,] requires 2-oxoglutarate, ferrous ions, ascorbate and molecular oxygen for activity and is bifunctional, catalyzing both the monooxygenation of hyoscyamine to 6β-hydroxyhyoscyamine and epoxide formation to yield scopolamine. Western blot analysis of extracts from various tissues of H. niger showed that hyoscyamine 6β-hydroxylase is abundant in cultured roots and plant roots, but is absent in leaf, stem, calyx, cell cultures and cultured shoots. Immunohistochemical studies using monoclonal antibody and immunogold-silver enhancement detected the enzyme only in the pericycle cells of young root ( Hashimoto et al. 1991 ). Similarly, hyoscyamine 6β-hydroxylase transcript is localized to roots and cultured roots of H. niger, but is not found in stem, leaves, or cell cultures of this same species ( Matsuda et al. 1991 ).

Figure 9.

Proposed scheme of the biosynthetic pathway leading from l-arginine and l-phenylalanine to the hallucinogenic alkaloid scopolamine in a wide range of tropane alkaloid producing species of Atropa, Datura, Duboisia and Hyoscyamus.

The best investigated oxidase of the pathway is the 2-oxo-glutarate-dependent dioxygenase hyoscyamine 6β-hydroxylase that coverts hyoscyamine into scopolamine. Dashed arrow indicates more than one enzymatic transformation.

Another tropane alkaloid-producing species, the deadly nightshade Atropa belladonna, accumulates hyoscyamine as the major alkaloid. As a first attempt at the metabolic engineering of a medicinal plant, transformation experiments were carried out that would alter the alkaloid pattern so that the pharmaceutically desired alkaloid scopolamine, instead of hyoscyamine, would be produced as the major alkaloid. The cDNA encoding hyoscyamine 6β-hydroxylase from H. niger was introduced into A. belladonna using either Agrobacterium tumefaciens- or Agrobacterium rhizogenes-mediated transformation. Elevated levels of scopolamine were detected in the resultant transgenic plants (1.2% dry weight in transgenic leaves compared to trace levels in control leaves) and hairy roots (0.3% dry weight compared to approximately 0.03% dry weight in the controls), thereby successfully demonstrating the viability of tailored alkaloid profiles in plants and plant organ cultures ( Hashimoto et al. 1993b ;Yun et al. 1992 ).

The third hydroxylation in the pathway leading from tabersonine to vindoline in C. roseus is also catalyzed by a cytosolic 2-oxoglutarate-dependent dioxygenase that acts upon C-4 of deacetoxyvindoline to form deacetylvindoline ( Fig. 8). The enzyme requires, in addition to the cosubstrate 2-oxoglutarate, ascorbate, ferrous ions and molecular oxygen for activity. The mechanism of desacetoxyvindoline 4-hydroxylase has been investigated and the order of substrate binding found to be 2-oxoglutarate first, followed by O2 and desacetoxyvindoline. The order of products released is deacetylvindoline followed by CO2 and finally succinate ( De Carolis & De Luca 1993). Desacetoxyvindoline 4-hydroxylase is absent from C. roseus cell cultures but could be purified to homogeneity from leaves and, thereby, a cDNA was isolated and functionally expressed in Escherichia coli (Vazquez-Flota et al. 1997). 4-Hydroxylase transcript was found to be predominant in leaf and, to a lesser degree, in fruit. Transcript was absent from the root and petals, and was barely detectable in the stem. This distribution mirrored that found for 4-hydroxylase enzyme activity. Gene transcription appears to be light-regulated during seedling development. The translation of the nucleotide sequence shows high homology to other 2-oxoglutarate-dependent dioxygenases of plant origin ( Fig. 10). This growing list (hyoscyamine 6β-hydroxylase from H. niger, anthocyanidine synthase from Malus domestica, ethylene forming enzyme from Lycopersicon esculente, flavonol synthase from Petunia hybrida and gibberellin 20-oxidase from Arabidopsis thaliana) suggests that PCR-based homology cloning of new 2-oxoglutarate-dependent dioxygenases should be now possible ( Prescott & John 1996).

Figure 10.

Comparison of the derived amino acid sequences of a series of 2-oxo-glutarate-dependent dioxygenases.

The amino acid sequences that are highly homologous and might be useful for oligonucleotide design for PCR-based homology cloning of new 2-oxo-glutarate-dependent dioxygenase genes are shaded grey. D4H, C. roseus desacetoxyvindoline 4-hydroxylase (GenBank/EMBL Database accession number AF008597); G20OX, A. thaliana gibberellin 20-oxidase ( X83379); AOH, apple tree anthocyanidin synthase ( X71360); FLS, petunia flavonol synthase ( Z22543); H6H, H. niger hyoscyamine 6β-hydroxylase ( M62719); EFE, tomato ethylene forming enzyme ( X58885); IPNS, A. nidulans isopenicillin N-synthase ( M18111).

Conclusions and future prospects

The oxidoreductases of alkaloid biosynthesis in plants are a highly substrate specific class of enzymes that catalyze regio-and stereoselective oxidations and oxygenations along the pathways that lead to diverse structural variants. Many years of enzymology have shown us that plant oxidoreductases facilitate chemical reactions that are not readily reproduced in comparable yield and enantiomeric excess by standard chemical means. The potential to use oxidoreductases in biomimetic syntheses of alkaloids makes this class of enzymes of potential biotechnological interest. Many plant alkaloids belonging to the isoquinoline, tropane and monoterpenoid indole classes of alkaloids are of contemporary pharmaceutical importance. Examples such as ajmaline (antihypertensive), berberine (antibiotic), morphine (analgesic), scopolamine (anticholinergic) and vinblastine (antineoplastic) have been provided herein. Many more examples such as camptothecin (antineoplastic), emetine (emetic), quinine (antimalarial), taxol (antineoplastic) and numerous others have not been addressed. In each of these additional biosynthetic pathways, yet unidentified oxidative enzymes should be catalyzing reactions that are of potential regulatory importance and could be exploited for use in biotechnological processes.

As indicated herein, as we learn more about the alkaloid biosynthetic oxidoreductases we see that they can be arranged into distinct classes. This organization also becomes apparent at the gene level when a comparison is made of amino acid sequences derived from nucleotide sequence translations. Regions of high amino acid sequence homology are evident within the cytochromes P-450 ( Holton & Lester 1996), the cytochrome P-450 reductases ( Rosco et al. 1997 ), the covalently flavinylated oxidases (W.-M. Chou and T.M. Kutchan, unpublished results) and the 2-oxoglutarate-dependent dioxygenases ( Vazquez-Flota et al. 1997 ). It should now be possible to use a direct, genetic approach to isolate cDNAs that encode new members of any of these three families of oxidoreductases. Final identification can be made by functional heterologous expression. To this end, bacteria, yeast and insect cell cultures have all been amply demonstrated as appropriate expression systems for enzymes of plant secondary metabolism. An approach of this type has now been used to clone cDNAs encoding an O-methyltransferase of isoquinoline alkaloid biosynthesis ( Frick & Kutchan 1998). New success within the monoterpenoid indole alkaloid field suggests that homology cloning of novel acetyltransferases may eventually also be possible ( St-Pierre et al. 1998 ). Given a greater number of alkaloid biosynthetic genes, the more likely we are to see new examples of metabolically engineered medicinal plants and possibly even combinatorial biochemistry with alkaloidal systems analogous to that accomplished in the polyketide field ( Jacobsen et al. 1997 ;Kramer et al. 1997 ).


Our work reported herein was supported by SFB 369 of the Deutsche Forschungsgemeinschaft, Bonn and by Fonds der Chemischen Industrie.