Sanguinarine reductase, a key enzyme of benzophenanthridine detoxification

Authors


Werner Roos. Fax: +345 5527006; e-mail: werner.roos@pharmazie.uni-halle.de

ABSTRACT

Cultured cells of Eschscholzia californica respond to a yeast glycoprotein elicitor by producing benzophenanthridine alkaloids, which are excreted into the cell wall and the outer medium. These compounds, preferentially sanguinarine, are efficient phytoalexins because of their ability to intercalate double-stranded DNA (dsDNA), penetrate membranes and inhibit various enzymes containing SH-groups. Externally added sanguinarine is rapidly taken up by intact cells and converted to dihydrosanguinarine, which is substituted intracellularly according to the biosynthetic route. A 29.5 kDa soluble enzyme that catalyses the reduction of sanguinarine and chelerythrine by either NADPH or NADH has been isolated and purified to homogeneity. Benzophenanthridines that accumulate in the outer medium, mainly 10-OH-chelerythrine, chelirubine and macarpine, are converted by the isolated enzyme and by intact cells at much slower rates than sanguinarine. The cellular capacity of uptake and conversion of sanguinarine largely surpasses the rate of alkaloid production. We conclude that the sanguinarine produced by intact cells, after excretion and binding to cell wall elements, is rapidly reabsorbed and reduced to the less toxic dihydrosanguinarine, which then undergoes further biosynthetic reactions. This recycling process would allow the presence of the toxic phytoalexin at the cellular surface without taking the risk of injuring the producing cell.

INTRODUCTION

Many plant secondary compounds are potent toxins with essential functions in the defence against microbes and herbivores. This is especially true for alkaloids, one of the largest and best-investigated groups of plant metabolites. Accumulating evidence shows that the molecular targets of such phytoalexins are usually the basic elements of cellular metabolism, signalling and gene expression, which also makes them potentially dangerous to the producing cell (e.g. Matile 1984; Van Tellingen 1992; Schmeller, Latz-Brüning & Wink 1997; Wink 1999; Wilson, Sauer & Hooser 2001). Thus, in order to benefit from their ability to produce defence compounds of high pharmacological activity, plants need to protect vulnerable sites of the living cytoplasm from their own toxins (Matile 1984).

The best-known strategy for controlling toxic alkaloids and their biosynthetic intermediates rests on vesicular and vacuolar compartmentation (Deus-Neumann & Zenk 1984, 1986; Roos & Luckner 1986; Wink & Roberts 1998; Martinoia, Massonneau & Frangne 2000). Much less is known about the handling of plant toxins that do not accumulate in vacuoles but are instead directed to the cellular surface and the outer medium. Such defence compounds do not require the destruction of intracellular compartmentation (Matile 1984) in order to interact with microbial invaders or herbivores. Among them are benzophenanthridines, a subgroup of isoquinoline alkaloids that are built via S-reticuline and S-scoulerine (Fig. 1, see Kutchan & Zenk 1993 for a review). Their toxicity is mainly based on two structural elements: (1) as planar heteroaromatic cations they intercalate dsDNA, thus inhibiting both transcription and DNA replication (Kakiuchi et al. 1987; Bajaj et al. 1990; Sen, Ray & Maiti 1996; Schmeller et al. 1997; Das et al. 2003), and bind to negatively charged membrane surfaces (Babich et al. 1996; Schmeller et al. 1997) and (2) because of their reactivity with SH-compounds, they inhibit a number of cytosolic and membrane enzymes, including Na+/K+ ATPases (Scheiner-Bobis 2001) and lipoxygenases (Vavreckova, Gawlik & Muller 1996), and interact with cytoskeletal proteins, mainly tubulin (Wolff & Knipling 1993; Faddeeva & Beliaeva 1997; Slaninova et al. 2001).

Figure 1.

A short overview of the biosynthesis of benzophenanthridine alkaloids (according to Kutchan & Zenk 1993; Haider et al. 2000). BBE, berberine bridge enzyme.

Benzophenanthridines are inducible defence compounds, i.e. their production is triggered by contact with elicitors of biotic (e.g. fungal glycoproteins, bacterial lipoproteins) or abiotic (e.g. vanadate) origin, as well as by endogenous signals like jasmonates (Schumacher et al. 1987; Gundlach et al. 1992; Roos et al. 1998; Haider et al. 2000; Villegas, Sommarin & Brodelius 2000). In cultured cells of Eschscholzia californica, the object of the present study, we have shown that different signal paths can induce alkaloid biosynthesis: low concentrations of the yeast glycoprotein elicitor signal via the activation of phospholipase A2 and cytosolic pH shifts (Roos et al. 1999; Viehweger, Dordschbal & Roos 2002), whereas high elicitor concentrations trigger a peak of jasmonate and induce alkaloid biosynthesis together with hypersensitive cell death (Färber et al. 2003).

Of the biosynthetic enzymes, the cytochrome P450-dependent monooxogenases are localized in vesicles derived from the endoplasmic reticulum. This has been shown in most detail in Berberis with the scoulerine-forming berberine bridge enzyme (Amann, Wanner & Zenk 1986; Facchini & Bird 1998) and in Eschscholzia with the protopin-6-hydroxylase, which catalyses the formation of the benzophenanthridine skeleton (Tanahashi & Zenk 1990a). Microsomal vesicles are most probably the sites of dihydrobenzophenanthridine biosynthesis, as they contain the full range of these alkaloids following cell fractionation (Tanahashi & Zenk 1990b), and electron microscopic observations revealed the fusion of numerous vesicles with the vacuole where they liberated osmiophilic material (Bock, Wanner & Zenk 2002). As other biosynthetic enzymes are located in the cytosol, transport of intermediates across vesicular membranes must also be assumed, most probably via highly specific transport systems as characterized in the isolated vacuoles of Fumaria (Deus-Neumann & Zenk 1984, 1986).

Almost nothing is known about the mechanism of exporting benzophenanthridines or the metabolic fate of the extracellular alkaloids. In intact cells, accumulation of the dihydroalkaloids, but not of benzophenanthridines, is detectable through their characteristic blue and orange-red fluorescence, respectively. Hence, the formation of benzophenanthridines catalysed by dihydrobenzophenanthridine oxidase is likely coordinated with their export across the plasma membrane.

The present paper demonstrates that the most cytotoxic species among benzophenanthridines can be rapidly reabsorbed from the outer medium and detoxified by a novel enzyme.

MATERIALS AND METHODS

Plant cell cultivation

Suspension cultures of E. californica were grown in a sterile medium according to Linsmaier & Skoog (1965) with the hormones 2,4-dichlorophenoxyacetic acid and α-naphthaleneacetic acid (1 µm each). Cultivation was performed on a gyrotary shaker (100 r.p.m.) at 24 °C in continuous light (c. 7 µmol m−2 s−1) in a 10 d growth cycle. Suspensions grown for 5 or 6 d were used for the following experiments. Cell densities were examined with a high-frequency cell counter (Casy1, Schärfe Systems, Reutlingen, Germany).

Elicitor treatment

Cells were harvested by filtering without pressure through a 200-mesh nylon filter, washed by resuspension in the fivefold volume of sterile phosphate-free culture liquid and finally resuspended [60 mg fresh weight (FW) mL−1] in 100 mL 75% phosphate-free culture liquid, which either contained a yeast elicitor (1 µg mL−1, if not indicated otherwise) or not (control suspensions). Yeast elicitor was prepared from bakers’ yeast by ethanol precipitation according to Schumacher et al. (1987) and the obtained glycoprotein mixture was further purified by ultrafiltration (30 kDa), fast protein liquid chromatography (FPLC) (anion exchange and size exclusion) and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). The active fraction is a mixture of glycoproteins of 30–42 kDa with a mannose content of ≈ 40% (Steighard & Roos, unpublished results).

Incubation of cells with alkaloids

Twelve millilitre portions of non-elicited cell suspensions, prepared as described above, received alkaloids from aqueous stock solutions and were shaken in 25 mL culture flasks. At the times indicated, 2 mL aliquots were withdrawn, immediately cooled on ice and either extracted with MeOH/HCl (in order to determine total alkaloids, cf. below) or centrifuged at 4 °C, 16 000 g for 15 min to order to quantify the alkaloids of the supernatant by their fluorescence (sanguinarine and alkaloid mixture: Ex 485 nm, Em 580 nm; chelerythrine or berberine: Ex 450 nm, Em 540 nm).

Fluorescence assay of benzophenanthridines and dihydrobenzophenanthridines

One millilitre cell suspension or culture filtrate was mixed with 1 mL MeOH 96% (v/v) containing 1% HCl, extracted for 30 min in a thermoshaker at 50 r.p.m. and 40 °C and then centrifuged at 16 000 g for 10 min. In 500 µL of the supernatant, benzophenanthridines were quantified by reading their fluorescence at Ex 490 nm and Em 570 nm, and dihydrobenzophenanthridines at Ex 365 nm and Em 450 nm.

Extraction of alkaloids for high-performance liquid chromatography (HPLC) separations

Two or three millilitres of cell suspension were filtered through a glass filter pad (GF/F, Whatman Brentford, UK) in a 5 mL syringe and the cell pellet washed three times with demineralized water. The cell pellet (≈ 200 mg FW) was suspended in 1 mL MeOH and incubated in a thermoshaker at 40 °C for 20 min, followed by 15 min centrifugation at 16 000 g. The supernatant was made up to 2 mL with MeOH and used for HPLC separation.

The culture medium was adjusted to pH 9.0 with NaOH and extracted three times by shaking with 3 mL EtOAc at room temperature. The glass filter was washed with 500 µL MeOH. The solvents were combined, dried with sodium sulfate by using a rotary evaporator, and the residue was dissolved in 1 mL MeOH.

Preparative isolation of benzophenanthridines from elicited cultures

The 6 d cultures were challenged by adding 1 µg mL−1 elicitor, and after 48 h the spent medium was obtained by filtration, extracted with EtOAc, and the alkaloids were precipitated as chlorides according to Boulware & Schlowsky (1990). The resulting alkaloid mixture contained 25% 10-OH-chelerythrine, 40% chelirubine and 35% macarpine, as quantified by HPLC.

HPLC separation of Eschscholzia alkaloids

The Gynkotek HPLC system (Gynkotek, Germering, Germany) was used to operate a reversed-phase C18 column (EC 250/4, Nucleosil 120-10 C18, Macherey-Nagel, Düren, Germany) at 1.5 mL min−1 by a stepped solvent gradient made of demineralized water/TFA at pH 1.9 and HPLC-grade acetonitrile. The content of the HPLC-grade acetonitrile was 40% between 0 and 2 min, 50% between 7 and 17 min and 80% between 25 and 30 min. Peaks were analysed with a diode array detector (UVD 340, Gynkotek) that produced absorption spectra from 200 to 600 nm. Benzophenanthridines and dihydrobenzophenanthridines were quantified by their optical density (OD) peaks at 280 nm [area under the curve (AUC)] with authentic sanguinarine as a standard compound for calibration. The AUC of sanguinarine was linear, with its concentration within the range of 2.5 to 20 µg mL−1. Chelerythrine was quantified by using the authentic compound as a standard.

Identification of alkaloids by mass spectrometry (MS)

The peak fractions eluted by the above HPLC procedure were collected and analysed by electrospray ionization (ESI) MS with an ESQUIRE-LC ionic trap mass spectrometer (Bruker Daltonik, Bremen, Germany).

FPLC-based isolation of sanguinarine reductase

Cell fractionation

After 8 d growth, 1000 g of cells (FW) were harvested by filtration, frozen in liquid nitrogen, grounded in a mortar with three parts dry ice and then homogenized in a cell mill at 15 000 r.p.m. (Ultra-Zentrifugalmühle ZM 1000, Retsch, Haan, Germany). After evaporation of the dry ice, the cell material was suspended in 1700 mL buffer A and stirred at 4 °C for 30 min. A fraction of soluble proteins, that contained the main sanguinarine-reducing activity of the homogenate, resulted from the centrifugation steps (performed with Optima LE-80K, Beckman coulter, Fullerton, USA) and precipitations described as follows: (1) removal of cell debris at 1000 g for 10 min; (2) 15 000 g for 30 min, supernatant stirred for 30 min with 1% protamine sulfate; and (3) 15 000 g for 30 min. The supernatant was subjected to two subsequent ammonium sulfate precipitations (each for 1 h at 4 °C, followed by centrifugation at 15 000 g for 30 min). The proteins precipitating between 40 and 80% saturation were dissolved in 100 mL buffer B and used for FPLC analysis.

Buffer A is made up of 50 mm K-MOPS at pH 7.5, 330 mm sucrose, 5 mm each of Na2EDTA, DTE and ascorbate and 1 mm phenylmethylsulfonyl fluoride (PMSF), while buffer B contains 10 mm Tris–HCl at pH 5.7, 10% (v/v) glycerol, 0.5 mm Na2EDTA, 1.7 mmβ-mercaptoethanol and 1 m ammonium sulfate.

FPLC

The following separations were performed with the ÄKTA Explorer FPLC System (Amersham Pharmacia, Freiburg, Germany) at columns obtained from the same manufacturer:

  • 1Phenylsepharose 6 FF 26/16 (hydrophobic interaction).
  • The sample (50 mL in buffer B) was loaded at 2 mL min−1 and elution occurred at 3 mL min−1 by a linear gradient of buffer C (0–100% over 300 mL), followed by an isocratic elution (150 mL buffer C). The main activity eluted between 230 and 280 mL and these fractions were combined. Proteins were precipitated with 80% ammonium sulfate (1 h at 4 °C, centrifugation at 16 000 g for 30 min) and dissolved in buffer C. For complete desalting, the solution was passed through a HiPrep 26/10 desalting column (Pharmacia Amersham) at 6 mL min−1. Buffer C is made up of the same compounds as buffer B, only without ammonium sulfate.

  • 2Q Sepharose FF 16/10 (anion exchange).
  • The protein-containing samples were loaded at 2 mL min−1 and elution occurred at 3 mL min−1 by a linear gradient of buffer D (0–100% over 600 mL), followed by an isocratic elution (150 mL of buffer D). The fractions of the highest enzyme activity (240–330 mL) were precipitated with 80% ammonium sulfate (cf. above) and the pellet resuspended in buffer D. The composition of buffer D is the same as buffer B, only that 0.5 m NaCl was used instead of ammonium sulfate.

  • 3Superdex 200 prep grade (size exclusion).
  • Four millilitre samples were loaded and eluted with buffer C at 2 mL min−1. Enzyme activity left the column between 200 and 225 mL.

  • 4Mono Q HR 5/5 (anion exchange).
  • Thirty millilitre fractions obtained from the third separation were loaded at 0.7 mL min−1 and the column was eluted at 0.8 mL min−1 with the following stepped gradient of buffer D: 0–15 mL: 0–20%; 15–25 mL: 20%; 25–45 mL: 20–50%; and 45–65 mL: 50–100%. The main activity was eluted between 6 and 8 mL.

  • 5Superose 12 HR 10/30 (size exclusion).
  • Two hundred microlitre samples were loaded and eluted with buffer E at 0.4 mL min−1. Enzyme activity and protein content peaked coincidently at 14–15 mL. All other fractions were almost free of enzyme activity. In parallel runs, the following marker protein peaks were separated: aldolase (160 kDa), bovine serum albumine (65 kDa), chymotrypsinogen A (25 kDa), myoglobin (17.8 kDa) and ribonuclease A (13.7 kDa). According to their elution volumes, the molecular mass of sanguinarine reductase was interpolated as 31.5 kDa. The composition of buffer E is the same as buffer B, only that 2 m NaCl was used instead of ammonium sulfate.

Sanguinarine reductase activity assay

  • 1The conversion of benzophenanthridines, e.g. sanguinarine, was measured in 500 µL mixtures containing 10 mm Tris–HCl buffer, pH 7.5, 1 mm PMSF, 40 µm NADH or NADPH, 7 µm alkaloid substrate and 2–5 µg of test protein. After 10 min incubation the reaction was stopped by adding 500 µL MeOH containing 1% HCl. 4 × 100 µL of the reaction mixture were transferred to a 96 well microtiter plate and the fluorescence was read with a Fluorolite reader (Dynatech Laboratories, Chantilly, VA, USA) at Ex 490 nm and Em 570 nm. The decrease of fluorescence was taken as a measure of substrate conversion, which was calibrated with sanguinarine or chelerythrine solutions of known concentration. Blanks were used to correct for some fluorescence decrease in the absence of the test protein.
  • 2The consumption of NADPH was measured in 1 mL mixtures containing 50 mm Tris–HCl, pH 7.0, 20 µm NADPH, 7 µm of the substrate alkaloid and 2.6 µg enzyme protein. After 15 min incubation at 22 °C, each 20 µL of diaphorase-FAD (3.3 nkat) and INT/Triton solution were added to give final concentrations of 100 µm iodonitrotetrazolium chloride (INT) and 0.1% Triton X-100. After a further 30 min of incubation in the dark, the OD 500 of the formed formazan was read (Spectrophotometer Ultrospec 3000, Amersham Pharmacia) and converted to NADPH concentrations by using calibration data obtained with authentic NADPH concentrations in the absence of the test enzyme protein.

Dihydrobenzophenanthridine oxidase activity assay

In a 500 µL mixture, 7.5 µm dihydrosanguinarine (prepared by chemical reduction of sanguinarine with NaBH4, cf. Tanahashi & Zenk 1990a) were incubated with the test proteins under similar conditions as described for sanguinarine reductase, but without NAD(P)H.

RESULTS AND DISCUSSION

Benzophenanthridines are excreted from the cytoplasm of elicitor-treated cells

Cell suspension cultures of E. californica produce a variety of dihydrobenzophenanthridine alkaloids [≈ 220 µg g−1 dry weight (DW) in 24 h] with dihydromacarpine as the dominant species (Fig. 2). After contact with a yeast glycoprotein elicitor (1 µg mL−1) the total alkaloid content increases nearly fourfold within 24 h. Among the newly formed alkaloids are 25–30% benzophenanthridines, mainly macarpine and chelirubine (Fig. 2, Table 1). The benzophenanthridines are mostly excreted into the outer medium, whereas the corresponding dihydroalkaloids are found exclusively in the cells (Fig. 2). Small amounts of benzophenanthridines that are detectable in the cell pellet, mainly 12-OH-chelirubine, are most probably contained in export vesicles but above all are bound to the cell wall, as suggested from microscopic examinations: as shown in Fig. 3, the typical fluorescence of benzophenanthridines is almost exclusively confined to the cell wall region. The binding affinity of benzophenanthridines to cell wall material as implied by Figs 3 and 4 is supported by their tight adherence to either fibres or microcrystals of cellulose (Roos, Weiss & Viehweger, unpublished results).

Figure 2.

High-performance liquid chromatography (HPLC) profiles of cellular and excreted alkaloids in elicited and control cell cultures. Twenty-four hours after the addition of a 1 µg mL−1 yeast elicitor to a 6 d culture, elicited and elicitor-free cell suspensions of the same batch were filtered, and cells and culture media were separately extracted and analysed by HPLC. Peak compounds were identified by their absorption spectra and the molecular masses were obtained by mass spectroscopy (MS) (cf. Methods).

Table 1.  The main benzophenanthridines and dihydrobenzophenanthridines of the used cell culture
 Control cultureElicited 24 h
t = 024 h
  1. In an experiment like Fig. 2, alkaloids were extracted from elicited and control cultures, separated by HPLC, identified by MS and quantified as described in the Methods section, except that 100 mL samples were analysed after suction filtration through a nylon mesh.

  2. Figures are in µg alkaloid g−1 DW, means of eight independent experiments, with SD between 26 and 67%.

  3. +, alkaloid was identified but content is below 5 µg g−1 DW; HPLC, high-performance liquid chromatography; MS, mass spectroscopy; DW, dry weight.

Dihydrobenzophenanthridines184230786
 Dihydrochelirubine 50 66634
 Dihydromacarpine 95122133
 10-OH dihydrochelerythrine 39 42+
 Dihydrosanguinarine++ 19
Benzophenanthridines++245
 Chelirubine++ 74
 12-OH chelirubine +100
 Macarpine++ 34
 10-OH chelerythrine + 37
Figure 3.

Benzophenanthridine alkaloids are bound to the cell wall region. Confocal images of a cell string in a 5 µL cell suspension that was spread on a microscopic slide and covered with a gel disc made of culture liquid with 2% agarose. (a) Cells before addition of a 1 µg mL−1 elicitor. (b) Cells 18 h after addition of a 1 µg mL−1 elicitor. (c) Elicited cells, plasmolysed for 1 min by 3 m NaCl. The Leica TCS-SP confocal microscope (Leica, Mannheim, Germany) was used to scan fluorescence images in two emission channels that showed no substantial overlap: (1) emission of benzophenanthridines at Ex 488 nm, Em 580–630 nm, shown in red. The typical maxima of benzophenanthridines are at Em 580 nm, 590 nm and 610 nm (2) emission of the vacuolar marker DM-NERF at Ex 488 nm, Em 535–555 nm, shown in green. This pH indicator, used to stain vacuoles, was allowed to accumulate during 2 h prior to the experiment. Each image contains the superimposed signals of both channels. Note that benzophenanthridine fluorescence adheres to the cell wall even during plasmolysis.

Figure 4.

Uptake and conversion of sanguinarine by intact cells, monitored by fluorescence microscopy. A 5 mL cell suspension received 50 µm sanguinarine from an aqueous stock solution. At 3, 10 and 20 min (top to bottom), fluorescence images were obtained with the Nikon Optiphot fluorescence microscope (Nikon, Tokyo, Japan) at Ex 330–380 nm, Em > 420 nm. Note that the red extracellular- and cell wall-localized fluorescence of sanguinarine turns into the blue intracellular fluorescence of dihydrobenzophenanthridines, which first accumulates in cytoplasmic areas (cyt) and later enters the vacuole (vac).

External benzophenanthridines are taken up and reduced by intact cells

Sanguinarine, the first benzophenanthridine of the biosynthetic route (cf. Fig. 1), was usually undetectable among the alkaloids of elicitor-treated intact cells, although its immediate precursor dihydrosanguinarine was present (cf. Fig. 2, Table 1). The production of sanguinarine could be demonstrated only in cultures that had been treated with high elicitor concentrations (50 µg mL−1), where it was present at substantial amounts, i.e. ≈ 1 µg g−1 DW, and could be easily identified by HPLC–MS (MW 332) and UV spectra (typical peaks at 275 and 329.5 nm) that coincided with the authentic compound. High elicitor concentrations are known to cause alkaloid production in concert with cellular injuries (e.g. loss of K+), leading to hypersensitive cell death (Fig. 5, Roos et al. 1998; Färber et al. 2003). The exclusive appearance of sanguinarine under such conditions argued for a specific sanguinarine-converting mechanism that might be active only in intact cells.

Figure 5.

Intact and non-viable cells detected by a fluorescence viability test. Typical cell strings are shown of a culture that was incubated with a 50 µg elicitor mL−1 for over 24 h. A phase contrast image is superimposed by a fluorescence image (Ex 460 nm, Em > 535 nm) of the same specimen. Yellow fluorescence represents the viability probe 5-(and 6)-carboxySNARF-1 (Molecular Probes, Eugene, OR, USA) liberated from its non-fluorescent acetoxymethyl ester and accumulated in intact cells, preferentially vacuoles, during a 1.5 h pre-incubation. Red fluorescence indicates benzophenanthridines accumulating in the nuclei of injured cells. Lacking accumulation of the yellow viability probe indicates a compromised plasma membrane and as a consequence thereof, the cellular interior becomes accessible to the benzophenanthridines of the culture medium (Roos et al. 1998). Five to ten per cent of non-viable cells are typically seen in the used cell suspensions. This percentage does not significantly change after contact with a 1 µg mL−1 yeast elicitor, but increases up to 60% after treatment with elicitor concentrations of 10–100 µg mL−1.

When added to cell suspensions, sanguinarine rapidly disappeared from the outer medium. In parallel, the cellular content of dihydrosanguinarine and related dihydroalkaloids increased, as shown accordantly by fluorescence microscopy (Fig. 4) and HPLC analysis (Fig. 6). Clearly, external sanguinarine was absorbed by the cells and immediately reduced to dihydrosanguinarine; the latter was derivatized according to the biosynthetic sequence (hydroxylation, O-methylation, cf. De Eknamkul, Tanahashi & Zenk 1992) leading to dihydrochelirubine and dihydromacarpine (Fig. 6, lower panel). Other benzophenanthridines were taken up much slower than sanguinarine, especially the mixture of benzophenanthridines that were dominantly produced by the used Eschscholzia culture, essentially 10-OH-chelerythrine, chelirubine and macarpine (Fig. 7). In each case, uptake was closely linked to reduction: no increase of cellular benzophenanthridines was detectable and the newly formed dihydroalkaloids appeared exclusively in the cells. Further derivatization occurred only with dihydrochelerythrine (10-OH-dihydrochelerythrine was identified, data not shown). Berberine, an alkaloid of the structurally related protopine type, which is not produced in the used Eschscholzia culture, was neither taken up nor converted (Fig. 7). The rate of uptake and reduction of sanguinarine did not change in the presence of energy poisons as azide or carbonylcyanide-4-trifluoro methoxyphenylhydrazone (FCCP) (Fig. 7), indicating that no active transport step or other processes that require ATP was involved.

Figure 6.

Alkaloid profiles of the cells and the outer medium during incubation with external sanguinarine. After addition of 13 µm sanguinarine, cultures were filtered by rapid suction and extracts were prepared separately from cells and culture media and analysed by high-performance liquid chromatography (HPLC). The peak compounds were identified by their absorption spectra and molecular masses (cf. Methods).

Figure 7.

Selectivity of uptake of benzophenanthridines from the outer medium. Cell suspensions were incubated with each 13 µm sanguinarine (○), chelerythrine (▵), berberine (▪), or an alkaloid mix (▿) isolated from elicited cultures that contained 10-OH chelerythrine, chelirubine and macarpine (cf. Methods). In addition, cells that had been pre-treated for 10 min with 50 µm sodium azide (*) or 5 µm carbonylcyanide-4-trifluoro-methoxyphenylhydrazone FCCP (×) were incubated with sanguinarine. At the times indicated, aliquots were centrifuged and the alkaloids of the supernatant quantified by their fluorescence (cf. Methods). The alkaloid mixture was treated as a single compound of average MW. Data are external alkaloid concentrations of a typical experiment which was repeated twice with a similar outcome.

The cellular capacity for the uptake and reduction of sanguinarine is surprisingly high: as shown in Fig. 8, up to 5 nmoles min−1 per Mio cells (≈ 300 nmol min−1 g−1 DW) can be converted, an activity that by far surpasses the maximum rate of alkaloid production (≈ 6 nmoles min−1 per g DW). As a result, a well-grown culture that contains ≈ 0.8 Mio cells mL−1 (≈ 5 mg DW mL−1) was able to completely absorb and reduce the sanguinarine of a 500 µm solution within 3–4 h, whereas the maximum concentration of alkaloids in the outer medium of an elicited culture never exceeded 15 µm.

Figure 8.

Capacity of intact cells to convert sanguinarine from a broad range of concentrations. In an experiment similar to Fig. 6, cell suspensions [5 mg dry weight (DW) mL−1] were incubated with sanguinarine of the indicated concentrations. Aliquots were withdrawn at 10 min intervals and extracted with MeOH/HCl, and sanguinarine was quantified by its fluorescence at Ex 490 nm, Em 570 nm (cf. Methods). Data are initial rates of conversion (loss of sanguinarine), means ± SD of three parallel samples. At the end of the experiment, cell viability was tested by monitoring the uptake and vacuolar accumulation of the viability probe cSNARF1, as explained in Fig. 5. No significant differences between sanguinarine-treated and control cells were observed.

Sanguinarine reductase, a soluble enzyme of Eschscholzia cells

The preceding observations allow us to assume the presence in Eschscholzia cells of an hitherto unknown enzyme that catalyses the reduction of benzophenanthridines, preferentially sanguinarine, to the corresponding dihydroalkaloids. After cell fractionation, sanguinarine-reducing enzyme activity was detectable in the 100 000 g supernatant. Through an FPLC-based isolation procedure (Fig. 9, cf. Methods) a protein of 29 432 Da (measured by HPLC–MS) was enriched and purified to homogeneity (Fig. 10) that catalysed the reduction of sanguinarine to dihydrosanguinarine with either NADPH or NADH as a hydrogen donor (Fig. 11). It was termed sanguinarine reductase [sanguinarine: NAD(P)H oxidoreductase]. With 40 µm NADH, sanguinarine reductase displays a specific activity of 30.3 nkat mg−1 protein and a KM for sanguinarine of 9.5 µm.

Figure 9.

Snapshot of the fast protein liquid chromatography (FPLC)-based isolation procedure of sanguinarine reductase. An elution profile of anionic exchange chromatography on Q Sepharose FF is shown to exemplify the purification protocol given in the Methods section. Sanguinarine reductase (▪, SR) is separated from dihydrobenzophenanthridine oxidase (▴, DHBO). Protein content is symbolized by ○.

Figure 10.

Enrichment and purification of sanguinarine reductase documented by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Protein samples of enzymatically active fractions obtained during the purification process are compiled. Lane 1: MW standards; Lane 2: crude extract, first 15 000 g supernatant; Lane 5: eluate from Q Sepharose FF; Lane 8: eluate from Superose 12.

Figure 11.

Sanguinarine reductase catalyses 1:1 conversion of sanguinarine and nicotinamide adenosine dinucleotide phosphate (NADPH). Purified sanguinarine reductase (2.64 µg protein mL−1) was incubated with 20 µm NADPH plus different concentrations of sanguinarine. After the reaction was completed (c. 30 min), benzophenanthridines, dihydroalkaloids and NADPH were quantified in the reaction mixture per fluorescence assay and INT/Diaphorase assay, respectively. Data show the yield of dihydrosanguinarine (□) and the loss of NADPH (•), each correlated to the loss of sanguinarine, means ± SD of three independent experiments. Similar data were obtained with chelerythrine or the mixture of 10-OH chelerythrine, chelirubine and macarpine, the latter incubated with 13 µg enzyme protein mL−1 (not shown). Insert: Electrospray ionization (ESI) mass spectra obtained before (left) and 10 min after incubation of purified sanguinarine reductase (2.64 µg protein mL−1) with 7 µm sanguinarine (right). The newly appearing peak at m/z 334 represents protonated dihydrosanguinarine and was not observed in the absence of either NADPH or sanguinarine reductase. Similar data exist for chelerythrine (not shown).

At its pH optimum of 6.5–7.5, the catalytic properties of the enzyme change with the hydrogen donor and the concentration range of the alkaloid substrate (Fig. 12): below 10 µm, the reaction velocity is about threefold higher with NADPH than with NADH and increases with the alkaloid concentration. Higher alkaloid concentrations cause the NADPH-dependent reduction to slow down, but not the NADH-driven reduction. The same phenomenon is seen with the substrate chelerythrine.

Figure 12.

Kinetic properties of sanguinarine reductase. The purified enzyme (2.64 µg protein mL−1) was incubated with increasing concentrations of sanguinarine (○) or chelerythrine (▿) and 40 µm of either nicotinamide adenosine dinucleotide phosphate (NADPH) (dark symbols, solid lines) or NADH (light symbols, broken lines) as hydrogen donor. Samples were withdrawn at t = 0 and 5 min and the alkaloids quantified by the fluorescence assay. Data represent initial rates of decrease of benzophenanthridine concentrations, mean ± SD of three independent experiments.

Selective influences of the hydrogen donors on the enzyme/substrate interaction are also implicated from the finding that sanguinarine shows maximum conversion rates with NADPH, but chelerythrine with NADH (Fig. 12).

Under conditions that are likely to prevail in the cytoplasm, i.e. NADPH concentration far higher than NADH (Møller 2001), sanguinarine concentration far below 5 µm, sanguinarine reductase shows a similar substrate preference as intact cells: sanguinarine is converted 1.3-times faster than chelerythrine and > 10-times faster than the genuine alkaloid mix (10-OH-chelerythrine, chelirubine and macarpine).

Likewise, neither sanguinarine reductase nor cell suspensions convert berberine or phenanthridine cations as ethidium or propidium (data not shown).

The reduction catalysed by sanguinarine reductase could not be reversed by increasing the product concentrations, i.e. even a hundredfold excess of NAD(P)+ did not cause a detectable oxidation of added dihydrosanguinarine. Thus, at the pH optimum of 7.0, hydrogen transfer from NAD(P)H to sanguinarine is thermodynamically highly preferred to the back reaction. Oxidation of dihydrobenzophenanthridines in Eschscholzia occurs via a reaction with molecular oxygen that is converted to H2O2. This reaction, which finalizes benzophenanthridine biosynthesis, is catalysed by dihydrobenzophenanthridine oxidase, an enzyme that shows no preference for sanguinarine (Schumacher & Zenk 1988; Arakawa et al. 1992). In our experiments, this enzyme was detected as a soluble protein and could easily be separated from sanguinarine reductase by the used FPLC procedure (Fig. 9).

CONCLUSIONS

Sanguinarine reductase and the detoxification of sanguinarine

On the cellular level, our data are best explained by the model shown in Fig. 13. Sanguinarine, the preferred substrate of sanguinarine reductase, is produced in Eschscholzia cells and subjected to a recycling process: after excretion into the extracellular space, the alkaloid most probably is reabsorbed and reduced intracellularly to dihydrosanguinarine, which in turn is converted into higher substituted benzophenanthridines along the biosynthetic sequence. In this way the steady-state concentration of sanguinarine is adjusted below that of other benzophenanthridines and thus might escape detection by the used analytical procedure (detection limit of about 2 µg g−1 DW). A dynamic equilibrium between production, excretion, uptake and reduction would allow to combine two seemingly contradictory demands on the sanguinarine-producing cell: (1) to synthesize the potent phytoalexin sanguinarine and present it at the cellular surface and (2) to protect DNA and vulnerable cytoplasmic sites by avoiding an increase of this toxin in the cytoplasm.

Figure 13.

Conceptual view of the recycling of sanguinarine, embedded in the metabolism of benzophenanthridines in elicited cells of Eschscholzia californica. SR, sanguinarine reductase; DHBO, dihydrobenzophenanthridine oxidase.

This concept is compatible with earlier observations in another cell strain of Eschscholzia: Collinge & Brodelius (1989) reported that sanguinarine and chelerythrine accumulated only transiently followed by their rapid conversion, most probably into macarpine and chelilutine, respectively. As the required hydroxylases and O-methylases only accept dihydroalkaloids, a reduction step must precede either of these derivatizations. Sanguinarine is obviously the most cytotoxic compound among the benzophenanthridines: it exerts the strongest inhibition of growth of Candida albicans and Staphylococcus aureus (inhibition constants: sanguinarine > chelerythrine > chelirubine > berberine, Tolkachev & Vichkanova 1978), and also of HeLa cells, in which it causes a stronger decrease in microtubule numbers than other benzophenanthridines (Slaninova et al. 2001; Wolf & Knipling 1993). The outstanding toxicity of this compound indicates (1) the faster penetration of cellular membranes, as its lower pKA accounts for a higher equilibrium concentration of the membrane permeable pseudobase (hydroxide adduct) (Slaninova et al. 2001); (2) the ability of the pseudobase to form non-covalent complexes with SH-groups (Bartak et al. 2003; Debiton et al. 2003); and (3) a higher degree of planarity that facilitates the intercalation of the cationic form into dsDNA (Maiti et al. 2002; Schmeller et al. 1997).

Cytotoxic effects of sanguinarine against plant cells have not yet been established seen apart from the injuring effects of sanguinarine and chelerythrine, but not berberine, on isolated vacuoles of Chelidonium (Jans 1973). In our experiments, addition of sanguinarine at 5 or 15 µm strongly inhibited the growth of cultured cells of Nicotiana or Arabidopsis. These cultures did not show a significant conversion of the added alkaloid within 24 h. In contrast, the growth of Eschscholzia cells remained unchanged under such conditions (Roos et al. unpublished). These findings, together with the absence of sanguinarine from intact cells and its accumulation in compromised cells (cf. Fig. 5), support the idea that this enzyme is essential for the safe handling of toxic benzophenanthridines in Eschscholzia cells.

The transport mechanism(s) essential for the recycling of benzophenanthridines in intact cells are yet to be characterized. The exclusive accumulation of benzophenanthridines in the extracellular medium and of the corresponding dihydroalkaloids in the cellular interior implies that either transport step across the plasma membrane is coupled to a redox reaction, i.e. the export of benzophenanthridines to their formation via dihydrobenzophenanthridine oxidase, and the uptake to their reduction via sanguinarine reductase. The conversion by this enzyme most probably generates the driving force of sanguinarine uptake by maintaining an inside-directed concentration gradient. This is concluded from the lacking intracellular accumulation of sanguinarine and the insensitivity of its uptake to energy poisons (cf. Fig. 7).

The transfer across the plasma membrane is the limiting step in the conversion of sanguinarine: intact cells absorb/reduce the alkaloid with an apparent KM of around 140 µm, whereas the enzymic reduction requires 10 µm for half-saturation (cf. Figs 8 & 12). A comparison of the kinetic data of sanguinarine and chelerythrine likewise points to the importance of membrane permeability. Intact cells convert chelerythrine only at the slow rates (10–15%) typical for the higher substituted benzophenanthridines, while the enzyme-catalysed reduction by NADPH reaches > 70% of the rate of sanguinarine (cf. Figs 7 & 12). This argues for a slower substrate supply of chelerythrine compared to sanguinarine. Concordantly, because of its lower pKa (8.05 compared to 9.0 for chelerythrine), sanguinarine might penetrate the plasma membrane more rapidly (cf. Slaninova et al. 2001).

ACKNOWLEDGMENTS

The expert help of Dr. Angelika Schierhorn, Max Planck group, Enzymology of protein folding, Halle, Germany with the ESI MS and of Dr Katrin Viehweger, this lab, with the confocal microscopy of Eschscholzia cells are gratefully acknowledged.

Ancillary