(+)-Catechin to Cyanidin and a Novel Procyanidin

FEBS Letters 580 (2006) 1642–1648

Anthocyanidin synthase from Gerbera hybrida catalyzes the conversion of (+)-catechin to cyanidin and a novel procyanidin

Frank Wellmanna,1, Markus Griesserb, Wilfried Schwabb, Stefan Martensa, Wolfgang Eisenreichc, Ulrich Materna,*, Richard Lukacˇina,2 a Institut fu¨r Pharmazeutische Biologie, Philipps-Universita¨t Marburg, Deutschhausstrasse 17 A, D-35037 Marburg, Germany b Fachgebiet Biomolekulare Lebensmitteltechnologie, Technische Universita¨tMu¨nchen, Lise-Meitner-Str. 34, D-85354 Freising, Germany c Lehrstuhl fu¨r Organische Chemie und Biochemie, Technische Universita¨tMu¨nchen, Lichtenberg-Str. 4, D-85747 Garching, Germany

Received 12 January 2006; revised 7 February 2006; accepted 8 February 2006

Available online 17 February 2006

Edited by Ulf-Ingo Flu¨gge

1. Introduction Abstract Anthocyanidins were proposed to derive from (+)- naringenin via (2R,3R)-dihydroflavonol(s) and (2R,3S,4S)-leucocyanidin(s) which are eventually oxidized by anthocyanidin Flavonoids comprise a class of abundant secondary metab- synthase (ANS). Recently, the role of ANS has been put into olites which contribute in many ways to the growth and question, because the recombinant enzyme from Arabidopsis subsistence of plants [1,2]. Prominent examples are the antho- exhibited primarily flavonol synthase (FLS) activity with negligi- cyanin pigments and the related oligomeric proanthocyani- ble ANS activity. This and other studies led to the proposal that dins which are under investigation for their medicinal ANS as well as FLS may select for dihydroflavonoid substrates potential [3]. The biosynthesis of flavonoids has been studied carrying a ‘‘b-face’’ C-3 hydroxyl group and initially form the 3- extensively over the last decades [4]. However, while the early geminal diol by ‘‘a-face’’ hydroxylation. Assays with recombi- steps have been unequivocally unraveled the reactions leading nant ANS from Gerbera hybrida fully supported the proposal to anthocyanidins and, in particular, to proanthocyanidins and were extended to catechin and epicatechin isomers as potential substrates to delineate the enzyme specificity. Gerbera ANS have remained under debate. All flavonoids derive from the converted (+)-catechin to two major and one minor product, flavanone (2S)-naringenin, which may be oxidized to the cor- whereas ent()-catechin (2S,3R-trans-catechin), ()-epicate- responding flavone by the action of flavone synthase (FNS) or chin, ent(+)-epicatechin (2S,3S-cis-epicatechin) and ()-gallo- hydroxylated in 3b-configuration to the (2R,3R)-dihydrofl- catechin were not accepted. The Km value for (+)-catechin was avonol [5,6]. Substitution reactions may proceed at any stage. determined at 175 lM, and the products were identified by Reduction of the dihydroflavonol by dihydroflavonol 4-reduc- n LC–MS and NMR as the 4,4-dimer of oxidized (+)-catechin tase (DFR, Fig. 1) [7] leads to (2R,3S,4S)-leucoanthocyanidin (93%), cyanidin (7%) and quercetin (trace). When these incuba- which was considered as the immediate precursor of anthocy- tions were repeated in the presence of UDP-glucose:flavonoid 3- anidin [8] initially based on supplemention experiments with O-glucosyltransferase from Fragaria · ananassa (FaGT1), the acyanic flowers of genetically defined lines of Matthiola incana product ratio shifted to cyanidin 3-O-glucoside (60%), cyanidin (14%) and dimeric oxidized (+)-catechin (26%) at an overall [9]. A branch pathway was proposed to convert cis-leucocy- equivalent rate of conversion. The data appear to identify (+)- anidin to (2R,3S)-trans-flavan-3-ol ((+)-catechin) [10] as the catechin as another substrate of ANS in vivo and shed new light likely start unit to oligomeric proanthocyanidins (Fig. 1). on the mechanism of its catalysis. Moreover, the enzymatic The condensation to proanthocyanidins (PA) or condensed dimerization of catechin monomers is reported for the first time tannins (CT) was assumed to require flavan-3,4-diols (leuco- suggesting a role for ANS beyond the oxidation of leucocyani- cyanidins) as extension units [10,11], but more recently dins. (2R,3R)-cis-flavan-3-ols (()-epicatechins) (Fig. 1) derived 2006 Federation of European Biochemical Societies. Published from anthocyanidins by the action of anthocyanidin reductase by Elsevier B.V. All rights reserved. (ANR) [12,13] have been suggested as the more likely precursors [11–14]. Nevertheless, the oligomerization reaction has Keywords: Flavonoid biosynthesis; Oligomeric proanthocyanidins; (+)-Catechin; 2-Oxoglutarate-dependent not yet been accomplished in vitro and the precise mechanism dioxygenase; Anthocyanidin synthase remains to be established [15]. Although the stereoconfiguration of flavonoids at C-2 appears to be set at the level of (2S)-naringenin and resulting necessarily in (+)-catechin and ()-epicatechin (Fig. 1), the diastereomers ent()-catechin ((2S,3R)-trans-catechin) and ent(+)-epicatechin ((2S,3S)-cis- * Corresponding author. Fax: +49 6421 282 6678. epicatechin) have also been reported as natural plant products E-mail address: matern@staff.uni-marburg.de (U. Matern). (cf. [15]). Furthermore, in some plants a bypass may exist 1 Present address: Werthenstein Chemie AG, CH-6105 Schachen, avoiding the 3b-hydroxylation of (2S)-flavanones and leading Switzerland. to (2R,4R)-flavan-4-ols (phlobaphenes) [16] and 3-deoxyanth- 2 Present address: Chromsystems Instruments & Chemicals GmbH, ocyanidins. Heimburg-Str. 3, D-81243 Munchen, Germany. ¨ ANS belongs to the 2-oxoglutarate iron-dependent oxygen- Abbreviations: FaGT1, Fragaria · ananassa cv. Elsanta glucosyltrans- ases and was cloned first from Perilla frutescens [17]. Four ferase recombinant ANSs were used subsequently with commercial

0014-5793/$32.00 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2006.02.004 F. Wellmann et al. / FEBS Letters 580 (2006) 1642–1648 1643

FHT (2S)-flavanones (2R/3R)-dihydroflavonols

DFR

R2 R2 OH OH

HO O HO O R1 LAR R1 OH OH OH OH OH (2R,3S,4S)-leucocyanidins (2R,3S)-flavan-3-ols (+)-catechin

ANS proanthocyanidin

R2 R2 ? OH OH

+ HO O HO O R1 R1 ANR OH OH OH OH anthocyanidins (2R,3R)-flavan-3-ols (-)-epicatechin

FGT

anthocyanins

Fig. 1. Role of ANS in the biosynthesis of anthocyanins and proanthocyanidins. Broken arrows designate reactions which have not been confirmed experimentally (R1, R2 = H or OH). ANS, anthocyanidin synthase; ANR, anthocyanidin reductase; DFR, dihydroflavonol 4-reductase; FHT, flavanone 3b-hydroxylase; FGT, flavonoid 3-O-glucosyltransferase; LAR, leucoanthocyanidin reductase. leucoanthocyanidin substrates and in combination with UDP- ation of the substrate and surprisingly revealed two molecules glucose:flavonoid 3-O-glucosyltransferase to investigate the of the substrate analogue in the active site with (2R,3R)-trans- reaction mechanism [8]. A mechanism was postulated that pro- DHQ closest to the iron atom, whereas either enantiomer was ceeds from leucoanthocyanidin via 2-flaven-3,4-diol (pseudo- bound at the other location [21]. Clearly, additional data are base) followed by isomerization of the 2,3-double bond to the required to define the substrate specificity of ANS. 3,4-position concomitant with a shift of the hydroxyl group Recombinant ANS from Gerbera hybrida was used to deter- from C-4 to C-2 and removal of the C-2 hydroxyl anion under mine the activity with various flavan-3-ol substrates. The selec- acidic conditions to yield anthocyanidin (flavylium ion) [8]. tive conversion of (+)-catechin to anthocyanidin and a dimeric However, the synthesis of pure leucoanthocyanidin enantiomers flavan-3-one sheds new light on the mode of action of ANS, is rather difficult, and the instability of flavan-3,4-diols in aque- because neither the mechanism proposed via isomerisation of ous solution as well as the substrate specificity of ANS observed 3-flaven-2,3-diol concomitant with the C-4/C-2 shift of the hy- in vitro cast some doubt on the role of leucoanthocyanidin as a droxyl group [8] nor the lack of configurational requirement at natural precursor of anthocyanidin [18]. In a series of studies on C-2 apply in this instance. Furthermore, the dimerization reac- recombinant ANS from Arabidopsis thaliana evidence was tion might be considered as a precedent for proanthocyanidin presented for the initial oxidation of the substrate at C-3 [19].Fur- formation. thermore, the ANS formed predominantly quercetin and cis- and trans-dihydroquercetin (DHQ) with cyanidin being a minor 2. Materials and methods product only [19,20], and the product pattern from (2R,3S,4S)- cis-leucocyanidin vs. that from (2R,3S,4R)-trans-leucocyanidin 2.1. Chemicals implied that cis-DHQ, trans-DHQ and cyanidin resulted mostly Biochemicals of analytic grade were purchased from Roth from the unnatural (2R,3S,4R)-trans-leucocyanidin [20]. More- (Karlsruhe, Germany). Reference samples of (+)-catechin, ()-catechin, over, co-crystallization of the ANS with Fe2+, 2-oxoglutarate (+)-epicatechin, ()-epicatechin, ()-gallocatechin, cyanidin chloride and quercetin were from Roth (Karlsruhe, Germany) or Sigma and racemic trans-DHQ or enantiomerically pure (2R,3R)- (Deisenhofen, Germany). LiChroprep RP18 (40–63 lm) was obtained DHQ as a substrate analogue in the absence or presence of from Merck (Darmstadt, Germany). The solutions of flavonoids were molecular oxygen supported the stereoselective C-3 hydroxyl- freshly prepared in methanol when used for enzyme incubations. 1644 F. Wellmann et al. / FEBS Letters 580 (2006) 1642–1648

2.2. Cloning and expression of ANS from Gerbera hybrida 2.7. Concerted reaction of ANS and FaGT1 Recombinant ANS was expressed in yeast [22]. The growing of yeast The ANS standard tests were supplemented by addition of 50 ll par- transformants and protein isolation were performed as previously de- tially purified FaGT1 (0.2 mg protein/ml) and 10 ll of 125 mM scribed [23]. UDP-glucose to a final volume of 200 ll and incubated at 37 C for 60 min. The flavonoids were analyzed directly by LC–MSn as described 2.3. Cloning and expression of FaGT1 from Fragaria · ananassa cv. above. Elsanta The construction of the cDNA library from strawberry fruit and 2.8. Isolation of the dimer the identification of the UDP-glucose:flavonoid 3-O-glucosyltransfer- For preparative purposes numerous single incubations of a total ase (FaGT1) sequence has been described [24]. The open reading amount of 10 mg (+)-catechin and approximately 25 mg ANS protein frame of FaGT1 (GenBank Accession number AAU09442) which were carried out under the above mentioned conditions. Successively, codes for a protein of 466 amino acids was cloned into the pET- the flavonoids were isolated by repeated extraction with ethyl acetate 29a (+) expression vector (Novagen, Schwalbach, Germany) using (twice 75 ll), the organic fractions pooled and finally dried under vac- the EcoRI and the XhoI restriction sites. Protein expression was car- uum. After resolving the residue in 2 ml of 60% aqueous methanol the ried out with the Escherichia coli strain BL21 (DE3) pLysS (Nova- dimer was separated from the reaction products by preparative RP-18 gen) which was grown at 37 C until an A600 of approximately 0.5 column chromatography (15 cm · 3 cm). Solvent A was water (Merck, was reached. Induction of protein expression was initiated by adding Darmstadt, Germany) acidified with 0.05% formic acid (Roth, isopropyl-1-thio-b-D-galactopyranoside (IPTG) to a final concentra- Karlsruhe, Germany) and solvent B was acetonitrile (Merck, Darms- tion of 1 mM, followed by further growth at 16 C for 8 h. Protein tadt, Germany). Products were eluted using a stepwise gradient from purification was carried out using the Talon metal affinity resin 100 ml of 100% A to 100 ml of 50% B, whereas during each step the (BD Biosciences Clontech, Palo Alto, USA) following the manufac- concentration of B was increased by 5%. A flow rate of 2 ml/min turers instructions with minor modifications. In brief, cells were har- was applied and 20-ml fractions were collected. Fractions were ana- vested by centrifugation and were disrupted with chilled mortar, lyzed by LC–MSn. Fractions 21 and 22 containing the dimer were pestle and glass beads (Sigma, Deisenhofen, Germany). All buffers pooled, concentrated to dryness, dissolved in d6-acetone and analyzed were modified to contain 10% glycerol, 10 mM b-mercaptoethanol by NMR spectroscopy. and 500 mM sodium chloride. The batch/gravity-flow column purification was applied and the recombinant protein was eluted using a 2.9. NMR spectroscopy buffer containing 150 mM imidazole. Protein concentrations were NMR spectra of the catechin derived dimer were recorded at 25 C determined using the Bradford microprotein assay [25]. Negative con- using an AVANCE 500 spectrometer (Bruker Instruments, Karlsruhe, trols were carried out with E. coli BL21 (DE3) pLysS cells harbour- Germany) at transmitter frequencies of 500.1 and 125.6 MHz for 1H ing an empty expression vector. 13 and C, respectively. Samples were dissolved in d6-acetone (1.5 mg in 0.5 mL). Two-dimensional COSY, NOESY, HMQC, and HMBC 2.4. Enzyme assays experiments were performed according to standard Bruker software Active ANS from filtered (PD10 column, Pharmacia, Freiburg, Ger- (XWINNMR). The mixing time was 1 s in the NOESY experiment. many) crude extract from yeast cells harbouring the ANS cDNA was 1H and 13C NMR chemical shifts were predicted by Specinfo. The sig- incubated at 37 C for 15 min. The standard reaction mixture (100 ll) nal assignments are based on proton–proton (COSY, NOESY) and contained 100 lM (+)-catechin, 100 lM 2-oxoglutarate, 50 lM ammo- proton–carbon correlation experiments (HMQC, HMBC). nium iron(II) sulfate, 2.5 mM sodium ascorbate, 2 mg/ml bovine catalase and 50 ll protein (0.8 lg/ll) in 200 mM potassium phosphate buffer pH 6.0. The incubations were carried out in open vials under gentle shaking and the reaction was terminated by the addition of 15 ll saturated aqueous EDTA solution. After centrifugation (5 min, 3. Results and discussion 10000 · g) the supernatant was directly subjected to LC–MSn analysis as described below. 3.1. ANS assays with catechins and epicatechins Activity assays carried out recently with recombinant ANS 2.5. Liquid chromatography – mass spectrometry (LC-MSn) from A. thaliana and employing either (2R,3S,4S)-cis-leucocy- The system used for the analysis of the enzyme assay products was a plus anidin or (2R,3S,4R)-trans-leucocyanidin as a substrate sur- Bruker Daltronics esquire 3000 ion trap mass spectrometer (Bruker prisingly revealed quercetin or (2R,3S)-cis-dihydroquercetin Daltronics, Bremen, Germany) connected with an Agilent 1100 HPLC system (Agilent Technologies, Waldbronn, Germany) equipped with a as the primary product rather than the anticipated product quaternary pump and a variable wavelength detector. Components cyanidin which accounted for less than 5% only [20]. These were separated on a Eurospher 100 C18 column, particle size 5 lm, incubations require particular experimental care, because in 10 cm · 2 mm (Grom, Rottenburg, Germany) which was held at aqueous solution leucocyanidins epimerize easily and cyani- 25 C. Solvent A was water (Merck, Darmstadt, Germany) acidified with 0.05% formic acid (Roth, Karlsruhe, Germany) and solvent B din is degraded rapidly [8]. Analogous incubations were car- was acetonitrile (Merck, Darmstadt, Germany). Products were sepa- ried out subsequently with ANS from G. hybrida expressed in rated using a linear gradient from 100% A to 100% B in 30 min with yeast cells [22] and using dihydrokaempferol as a substrate in a flow rate of 0.2 ml/min. The detection wavelength was either 280 combination with dihydroflavonol reductase from G. hybrida, or 520 nm, as indicated in the text. The electrospray ionization voltage which basically confirmed the low rate of pelargonidin forma- of the capillary was set to 4000 V and the end plate to 500 V. Nitro- gen was used as dry gas at a temperature of 330 C and a flow rate of tion (Martens et al., unpublished). Accordingly, a more de- 9 l/min. The full scan mass spectra were measured in a scan range from tailed examination of the physiological role of ANS appears 50 to 800 m/z with a scan resolution of 13000 m/z/s. Tandem mass to be necessary. In a first approach, we used the partially spectrometry was carried out using helium as collision gas purified recombinant ANS from G. hybrida for activity assays (3.56 · 106 mbar) with the collision voltage set at 1 V. All spectra were acquired in the positive ionization mode. Data analysis was per- with enantiomerically pure (+)-catechin, ()-catechin, and formed using the DataAnalysis 3.1 software (Bruker Daltronics). (+)-epicatechin, respectively, each over a range of concentrations from 16 to 100 lM. The assay conditions were adapted from previous work [22,26], and crude extracts from yeast 2.6. Kinetic properties cells harbouring the empty pYES2 expression vector were Ranges of flavonoid substrate concentrations of between 16 and routinely used as negative controls. The incubations were 100 lM were used for Km determination. The apparent Michaelis con- n stant for ANS was calculated from Lineweaver–Burk plot. The protein subjected directly to LC–MS analysis, thus avoiding the amounts were quantified according to Bradford [25]. acidification of the reaction mixture prior to the separation F. Wellmann et al. / FEBS Letters 580 (2006) 1642–1648 1645 as had been done in case of the incubations employing re- Detailed mass spectral analysis revealed also quercetin combinant ANS from Perilla frutescens [8]. The stability of ([M + H]+, m/z 303) and an additional product co-eluting dissolved leucocyanidins and potential products delicately de- with cyanidin (Fig. 2a). This product exhibited a pseudomo- pends on the pH which requires particular attention. Only lecular ion [M + H]+ of m/z 575 yielding a fragment of (+)-catechin was accepted as substrate, yielding a red col- m/z 287 in the MS2 spectrum (Fig. 2c and d). Thus, we con- oured compound. LC–MSn analysis confirmed the formation cluded a dimeric oxidized catechin as structure for the new of cyanidin from (+)-catechin due to the mass spectrum, product. Semiquantitative analysis of the product pattern product ion spectrum of the pseudomolecular ion [M + H]+ was performed by integration of the MS signals in the respec- of m/z 287 and the retention time of the new compound. tive ion traces. The catechin derived dimer (93%) with a molecular weight of 574 g mol1 and cyanidin (7%) were determined as major products besides traces of quercetin (Fig. 2a). a b The product formation was essentially dependent on the m/z x105 m/z presence of ferrous ions and 2-oxoglutarate, as is typical for x10 6 A 575 4 E 575 1 this class of dioxygenases (27), and maximal activity required the addition of ascorbate (300 lM) in accordance with previous studies [17], whereas catalase had no significant effect on the catalytic efficiency. The substrate affinity of the Gerbera ANS to (+)-catechin with an apparent Km value of 175 lM 0 0 is within range to the substrate affinities reported for related m/z x105 m/z B F plant 2-oxoglutarate-dependent oxygenases. x10 6 287 4 287 1 The specifity of the Gerbera ANS for (+)-catechin is note- worthy, because the ANSs from A. thaliana, Petunia hybrida or G. hybrida are all capable of hydroxylating (2S)- as well as (2R)-naringenin at C-3, beyond their catalytic activities on (2R,3S,4S)-cis-or(2R,3S,4R)-trans-leucocyanidin and 0 0 (2R,3R)-trans-dihydroquercetin as a substrate [22,27]. In case 5 C m/z x10 G m/z of the leucocyanidins, the stereoconfiguration at C-4 affected x10 6 449 4 449 1 the product selectivity leading to different ratios of quercetin vs. dihydroquercetin [20]. This does not apply to catechins or epicatechins. However, the 3-hydroxylation of naringenin

detector response detector response was reported recently to be important in biasing the substrate selectivity of ANS towards the natural C-2-stereo- 0 0 chemistry (27), and the specificity of the Gerbera ANS for D 520 H 520 nm nm (2R,3S)-trans-catechin ((+)-catechin) is thus fully compatible with this proposal. 1.5 2.0 3.2. Incubation of (+)-catechin with ANS and FaGT1 1.0 Anthocyanidins are unlikely to accumulate in vivo, and 1.0 ANS was suggested to act in concert with UDP-glucose:flavonoid 3-O-glucosyltransferase, at least, as part of a membrane- 12 14 16 [min] 12 14 16 [min] associated enzyme complex which forms the water-soluble and stable anthocyanidin-3-O-glycosides [4,28]. The advantage of c 575 such an interaction has also been exploited in in vitro incuba- 100 tions [17]. Accordingly, mixed incubations were conducted % with Gerbera ANS and flavonoid 3-O-glucosyltransferase from 557 Fragaria · ananassa cv. Elsanta (FaGT1) [24]. Under these 200 250 300 350 400 450 500 550 m/z conditions the ratio of products shifted drastically towards anthocyanin formation (14% cyanidin, 60% cyanidin 3-O-glu- 100 d 557 coside) at the expense of the catechin derived dimer (26%) % 287 (Fig. 2b). Control incubations carried out with (+)-catechin and the 3-O-glucosyltransferase only did not produce a glycos- 200 250 300 350 400 450 500 550 m/z ylated product, which was fully compatible with the fact that the enzyme requires C-2/C-3-oxidized flavonoid substrates n Fig. 2. LC–MS analyses of enzyme assays with (+)-catechin as [29,30]. The results suggested that most of the cyanidin formed substrate and ANS (a) as well as of concerted enzyme assays with ANS during ANS incubation was conjugated immediately by the ac- and FaGT1 (b) and mass spectrum (c) and product ion spectrum (MS2) of m/z 575 (d) of the catechin derived dimer. Catechin derived dimer, tion of FaGT1 and thus became stabilized by the glycosylation cyanidin and cyanidin-3-glucoside were detected at ion traces m/z 575 reaction. (A,E), 287 (B,F) and 449 (C,G), respectively. UV analyses at 520 nm Nevertheless it could not be excluded that in vivo cyanidin confirmed the formation of cyanidin and cyanidin-3-glucoside (D,H). formation occurs via 4S-flav-2en-3,4-diol intermediate chan- All substances except for the catechin derived dimer were identified using authentic reference material comparing retention times and neled directly to FGT and subsequently transformed into molecular masses. anthocyanin as proposed recently [20]. 1646 F. Wellmann et al. / FEBS Letters 580 (2006) 1642–1648

3.3. Structure of the catechin-derived dimer The configuration of the molecule cannot be derived unequiv- About 1.5 mg of the pure catechin-derived dimer was col- ocally on the basis of the present data. However, the apparent lected from numerous incubations of (+)-catechin with recom- absence of a NOESY crosspeak between H-2 and H-4 could binant ANS followed by separation on a preparative RP18 indicate trans configuration of the respective atoms. column. The formation of the dimer required (+)-catechin as the substrate as well as an active ANS and was always accom- 3.4. Mechanistic implications panied by the formation of cyanidin. In contrast to cyanidin Several studies have been conducted recently on flavonoid- the dimeric product was stable at room temperature. Its struc- related 2-oxoglutarate dependent dioxygenases aiming primar- ture was analysed by NMR spectroscopy (Table 1). ily at the mechanism of C-ring oxidation [6,19,22,27,31]. These In light of the molecular weight of the dimer, it was surpris- enzymes were commonly proposed to catalyze the hydroxyl- ing that the 1H NMR spectrum displayed only a limited num- ation of carbon-3 configurated either to the b-face (FHT/ ber of signals. In the downfield region of the 1H NMR FNS I) or to the a-face (ANS/FLS) of the flavonoid-ring plane spectrum, a set of well-resolved signals were observed resem- [27]. The initial hydroxylation suggested for the ANS reaction bling the full signal pattern of (+)-catechin. This indicates that: is likely to occur at carbon-3, although oxidations at either C-2 (i) the structure of the dimer is closely related to the structure or C-4 have not been fully excluded. In case of (+)-catechin, of the catechin substrate; (ii) the molecule is characterized by only the a-face hydroxylation at C-3 or a concerted oxidation an inherent symmetry giving rise to isochronous signals of cor- at C-3/C-2 are plausible, because a hydroxylation at C-4 would responding atoms in the symmetrical subunits of the molecule; generate leucocyanidins as a source of different products [20] (iii) the two monomers are not connected via carbon atoms of (Fig. 3, I). It is thus conceivable to assume at the first stage the aromatic rings bearing hydrogen atoms in the original cat- of the oxidation flav-3-en-3-ol and flav-2-en-3-ol with flavan- echin substrate (e.g., C-8 which is the typical nucleophilic cen- 3-one in tautomeric equilibrium, which spontaneously oxidize ter for condensations with catechin). further to anthocyanidin (cyanidin from (+)-catechin) in the In the upfield region of the spectrum, a spin system compris- presence of oxygen [32] (Fig. 3, II). Alternatively, a second cy- ing hydrogen atoms at C-2–C-4 of the catechin substrate was cle of ANS-catalyzed hydroxylation at carbon-3 might be pos- not observed for the dimer. On the other hand, two singlets tulated (Fig. 3, III), in analogy to the mechanistic proposal at 5.21 and 3.16 ppm were observed with intensities displaying suggested for the ANS-catalyzed formation of quercetin from one hydrogen atom for each site per symmetrical subunit of leucocyanidins [27]. the molecule. A two-dimensional COSY experiment afforded The symmetric cyanidin derived dimer formed by ANS three pairs of correlated hydrogen atoms displaying the spin in vitro has, to our knowledge, no precedent in the literature. systems of the aromatic rings in the catechin motif (i.e., H- It was obviously produced by the ANS activity, most likely 6–H-8, H-14–H-13, and H-14–H-10). Additionally, the signal from the flav-2-en-3-ol intermediate by abstraction of hydro- at 5.21 was correlated with the H-10 and H-14 signals and gen from C-4 leaving a radical that stabilizes by spontaneous was therefore clearly recognized as H-2. The singlet signal at dimerization (Fig. 3, IV). Natural procyanidins are highly di- 3.16 ppm gave weak correlation peaks with the H-6 and H-8 verse in size and composition, and the monomers are usually signals. Tentatively, this signal was therefore assigned as H-4. lined up by C-4/C-8 or C-4/C-6 linkages [33]. So far only the Information about 13C NMR chemical shifts could be NADPH-dependent anthocyanidin (ANR) and leucoanthocy- gleaned from two-dimensional HMQC and HMBC experi- anidin (LAR) reductases have been characterized, which pro- ments revealing 24 pairs of connectivities between hydrogen duce leucoanthocyanidin and catechin, respectively, the and carbon atoms via one bond (HMQC experiment) or via precursors of procyanidins [12,33] (Fig. 1). Biomimetic models multiple bonds (HMBC). This connectivity pattern established have demonstrated a non-enzymatic condensation reaction, the structure of the molecule as the C-4–C-4 dimer of catechin. suggesting the extension of a (+)-catechin start unit by

Table 1 NMR data of the C-4–C-4 dimer derived from (+)-catechin Position Chemical shifts (ppm) Coupling constant (Hz) Correlation pattern 1 13 d H d C JHH COSY NOESY HMBC Observed Predicted Observed Predicted 10 6.94(d, 1H) 6.49 114.7 117.9 2.2 14 2 2 14 6.81(dd, 1H) 6.58 119.4 123.3 8.5, 2.2 13, 10 2, 13 10, 2 13 6.75(d, 1H) 6.49 114.8 117.2 8.5 14 14 14 6 5.93(d, 1H) 5.71 95.2 95.3 2.2 8 8 5.81(d, 1H) 5.75 95.7 94.0 2.2 6 2(w), 6(w) 2 5.21(s, 1H) 5.84 82.3 95.3 10, 14 14, 10 10, 14 4 3.16(s, 1H) 4.24 24.1 31.1 6(w), 8(w) 2 4a 94.6 99.7 4(s), 2, 8(w) 9 130.5 128.3 13, 2 11 144.6 144.6 13 12 145.2 142.9 14, 10 8a 154.7 165.6 2, 6(w) 7 153.6 158.1 8(w) 5 160.5 3 207.1 F. Wellmann et al. / FEBS Letters 580 (2006) 1642–1648 1647

OH OH OH OH OH OH H I HO O HO O HO O OH C3-oxidation OH O OH OH OH OH (+)-catechin H2O- flavan-3-one elimination

OH OH OH OH OH OH

HO O HO O HO O III IV

OH C3-oxidation OH C4-hydrogen- OH O abstraction (IV) OH H H OH H H OH H H Fe (IV) Fe O flav-2-en-3-ol

OH OH OH OH

+ HO O HO O (II) O Fe no enzyme FGT OH OH · (III) OH H H II VI OH H Fe O H

radical abreaction OH 11 10 OH OH OH 8 12 HO 7 8a O 9 OH OH 13 2 14 + + 4 HO O HO O 4a O V 6 5 3 VII OH OH OH OSugar O FGT FGT OH OH HO cyanidin cyanidin 3-glucoside O OH HO C-4–C-4-catechin derived dimer

Fig. 3. Oxidation of (+)-catechin catalyzed by ANS. Flav-2-en-3-ol intermediates might be further oxidized to cyanidin and catechin derived dimer, respectively, or may serve as substrate for other reactions in ﬂavonoid or proanthocyanidin biosynthetic pathways.

(2R,3S)/(2R,3R)-quinone methide derived via flav-3-en-ol from the corresponding o-quinone [36]. This proposal followed the leucoanthocyanidin. Flav-3-en-3-ols have more generally been assumption of polyphenol oxidases converting flavan-3-ols to considered as precursors in flavonoid- and proanthocyanidin procyanidins [37]. biosynthesis [15,34] and were postulated to arise also during the reverse NADPH/NADH-dependent reduction of anthocy- References anidins to 2,3-cis-flavan-3-ols catalysed by ANR [12]. Irrespec- tive of chemical model reactions there is strong evidence for an [1] Harborne, J.B. and Williams, C.A. (2000) Advances in flavonoid enzyme condensing catechins and leucocyanidins [35], how- research since 1992. Phytochemistry 55, 481–504. ever, the mechanism of condensation is poorly understood. [2] Cooper-Driver, G.A. (2001) Contributions of Jeffrey Harborne The dimerization of (+)-catechin by ANS might be considered and co-workers to the study of anthocyanins. Phytochemistry 56, 229–236. as a lead to proanthocyanidin condensation, in particular since [3] Kong, J.-M., Chia, L.-S., Goh, A.-K., Chia, T.-F. and Brouillard, the ANS crystallized from A. thaliana has been shown recently R. (2003) Analysis and biological activities of anthocyanins. to accommodate two flavonoids in the active site pocket [21]. Phytochemistry 64, 923–933. The catalytic activity documented in this report for ANS [4] Forkmann, G. and Heller, W. (1999) Biosynthesis of flavonoids in: Comprehensive Natural Products Chemistry (Sankawa, U., clearly differs from the mode of action proposed recently for Ed.), Polyketides and Other Secondary Metabolites Including the polymerisation of flavonoids in Arabidopsis seed coat testa, Fatty Acids and their Derivatives, Vol. 1, pp. 713–748, Pergamon which relies on a laccase-type enzyme oxidizing epicatechin to Press, Oxford. 1648 F. Wellmann et al. / FEBS Letters 580 (2006) 1642–1648

[5] Lukacˇin, R. and Britsch, L. (1997) Identification of strictly [22] Martens, S., Forkmann, G., Britsch, L., Wellmann, F., Matern, conserved histidine and arginine residues as part of the active site U. and Lukacˇin, R. (2003) Divergent evolution of flavonoid 2- in Petunia hybrida flavanone 3b-hydroxylase. Eur. J. Biochem. oxoglutarate-dependent dioxygenases in parsley. FEBS Lett. 544, 249, 748–757. 93–98. [6] Wellmann, F., Matern, U. and Lukacˇin, R. (2004) Significance of [23] Urban, P., Mignotte, C., Kazmaier, M., Delorme, F. and C-terminal sequence elements for Petunia flavanone 3b-hydrox- Pompon, D. (1997) Cloning, yeast expression, and characteriza- ylase activity. FEBS Lett. 561, 149–154. tion of the coupling of two distantly related Arabidopsis thaliana [7] Martens, S., Teeri, T. and Forkmann, G. (2002) Heterologous NADPH-cytochrome P450 reductases with P450 CYP73A5. J. expression of dihydroflavonol 4-reductases from various plants. Biol. Chem. 272, 19176–19186. FEBS Lett. 531, 453–458. [24] Aharoni, A. and O’Conell, A.P. (2002) Gene expression analysis [8] Nakajima, J., Tanaka, Y., Yamazaki, M. and Saito, K. (2001) of strawberry achene and receptacle maturation using DNA Reaction mechanism from leucoanthocyanidin to anthocyanidin microarrays. J. Exp. Bot. 53, 2073–2087. 3-glucoside, a key reaction for coloring in anthocyanin biosyn- [25] Bradford, M.M. (1976) A rapid and sensitive method for the thesis. J. Biol. Chem. 276, 25797–25803. quantification of microgram quantities of protein utilizing the [9] Heller, W., Britsch, L., Forkmann, G. and Grisebach, H. (1985) principle of protein dye binding. Anal. Biochem. 72, 248–254. Leucoanthocyanidins as intermediates in anthocyanidin biosyn- [26] Schro¨der, G., Wehinger, E., Lukacˇin, R., Wellmann, F., Seefel- thesis in flowers of Matthiola incana R. BR.. Planta 163, 191–196. der, W., Schwab, W. and Schro¨der, J. (2004) Flavonoid methyl- [10] Stafford, H.A. (1990) Flavonoid Metabolism, CRC Press, Inc., ation: a novel 40-O-methyltransferase from Catharanthus roseus, New York. and evidence that partially methylated flavanones are substrates [11] Kitamura, S., Shikazono, N. and Tanaka, A. (2004) TRANS- of four different flavonoid dioxygenases. Phytochemistry 65, PARENT TESTA 19 is involved in the accumulation of both 1085–1094. anthocyanins and proanthocyanidins in Arabidopsis. Plant J. 37, [27] Turnbull, J.J., Nakajima, J., Welford, R.W.D., Yamazaki, M., 104–114. Saito, K. and Schofield, C.J. (2004) Mechanistic studies on three [12] Xie, D.-Y., Sharma, S.B., Paiva, N.L., Ferreira, D. and Dixon, 2-oxoglutarate-dependent oxygenases of flavonoid biosynthesis: R.A. (2003) Role of anthocyanidin reductase, encoded by / anthocyanidin synthase, flavonol synthase, and flavanone 3b- BANYULS/ in plant flavonoid biosynthesis. Science 299, 396– hydroxylase. J. Biol. Chem. 279, 1206–1216. 399. [28] Winkel-Shirley, B. (2001) Flavonoid biosynthesis. A colorful [13] Xie, D.-Y., Sharma, S.B. and Dixon, R.A. (2004) Anthocyanidin model for genetics, biochemistry, cell biology, and biotechnology. reductases from Medicago truncatula and Arabidopsis thaliana. Plant Physiol. 126, 485–493. Arch. Biochem. Biophys. 422, 91–102. [29] Yamazaki, M., Yamagishi, E., Gong, Z., Fukuchi-Mizutani, M., [14] Xie, D.-Y., Jackson, L.A., Cooper, J.D., Ferreira, D. and Paiva, Fukui, Y., Tanaka, Y., Kusumi, T., Yamaguchi, M. and Saito, K. N.L. (2004) Molecular and biochemical analysis of two cDNA (2002) Two flavonoid glucosyltransferases from Petunia hybrida: clones encoding dihydroflavonol-4-reductase from Medicago molecular cloning, biochemical properties and developmentally truncatula. Plant Physiol. 134, 979–994. regulated expression. Plant Mol. Biol. 48, 401–411. [15] Marles, M.A.S., Ray, H. and Gruber, M.Y. (2003) New perspec- [30] Fukuchi-Mizutani, M., Okuhara, H., Fukui, Y., Nakao, M., tives on proanthocyanidin biochemistry and molecular regulation. Katsumoto, Y., Keiko, Y.-S., Kasumi, T., Hase, T. and Tanaka, Phytochemistry 64, 367–383. Y. (2003) Biochemical and molecular characterization of a novel [16] Chopra, S., Gevens, A., Svabek, C., Wood, K.V., Peterson, T. UDP-glucose:anthocyanin 30-O-glucosyltransferase, a key enzyme and Nicholson, R.L. (2002) Excision of the Candystripe1 trans- for blue anthocyanin biosynthesis, from gentian. Plant Physiol. poson from a hyper-mutable Y1-cs allele shows that the 132, 1652–1663. sorghumY1 gene controls the biosynthesis of both 3-deoxyanth- [31] Lukacˇin, R., Wellmann, F., Britsch, L., Martens, S. and Matern, ocyanidin phytoalexins and phlobaphene pigments. Physiol. Mol. U. (2003) Flavonol synthase from Citrus unshiu is a bifunctional Plant Pathol. 60, 321–330. dioxygenase. Phytochemistry 62, 287–292. [17] Saito, K., Kobayashi, M., Gong, Z., Tanaka, Y. and Yamazaki, [32] Coetzee, J., Malan, E. and Ferreira, D. (2000) Synthesis and M. (1999) Direct evidence for anthocyanidin synthase as a 2- reactions of flav-3-en-3-ols. Tetrahedron 56, 1819–1824. oxoglutarate-dependent oxygenase: molecular cloning and func- [33] Tanner, G.J., Francki, K.T., Abrahams, S., Watson, J.M., tional expression of cDNA from a red forma of Perilla frutescens. Larkin, P.J. and Ashton, A.R. (2003) Proanthocyanidin biosyn- Plant J. 17, 181–189. thesis in plants: purification of legume leucoanthocyanidin [18] Turnbull, J.J., Sobey, W.J., Aplin, R.T., Hassan, A., Schofield, reductase and molecular cloning of its cDNA. J. Biol. Chem. C.J., Firmin, J.L. and Prescott, A.G. (2000) Are anthocyanidins 278, 31647–31656. the immediate products of anthocyanidin synthase? Chem. [34] Stafford, H.A. (1983) Enzymic regulation of procyanidin biosyn- Commun., 2473–2474. thesis; lack of a flav-3-en-3-ol intermediate. Phytochemistry 22, [19] Welford, R.W.D., Turnbull, J.J., Claridge, T.D.W., Schofield, 2643–2646. C.J. and Prescott, A.G. (2001) Evidence for oxidation at C-3 of [35] Abrahams, S., Tanner, G.J., Larkin, P.J. and Ashton, A.R. (2002) the flavonoid C-ring during anthocyanin biosynthesis. Chem. Identification and biochemical characterization of mutants in the Commun., 1828–1829. proanthocyanidin pathway in Arabidopsis. Plant Physiol. 130, [20] Turnbull, J.J., Nagle, M.J., Seibel, J.F., Welford, R.W.D., Grant, 561–576. G.H. and Schofield, C.J. (2003) The C-4 stereochemistry of [36] Pourcel, L., Routaboul, J.-M., Kerhoas, L., Caboche, M., leucocyanidin substrates for anthocyanidin synthase affects prod- Lepiniec, L. and Debeaujon, I. (2005) TRANSPARENT TES- uct selectivity. Bioorg. Med. Chem. Lett. 13, 3853–3857. TA10 encodes a laccase-like enzyme involved in oxidative [21] Wilmouth, R.C., Turnbull, J.J., Welford, R.W.D., Clifton, I.J., polymerisation of flavonoids in Arabidopsis seed coat. Plant Cell Prescott, A.G. and Schofield, C.J. (2002) Structure and mecha- 17, 2966–2980. nism of anthocyanidin synthase from Arabidopsis thaliana. [37] Dixon, R.A., Xie, D.Y. and Sharma, S.B. (2005) Procyanidins – a Structure 10, 93–103. final frontier in flavonoid research? New Phytol. 165, 9–28.