Leucoanthocyanidin reductase and anthocyanidin reductase gene expression and activity in flowers, young berries and skins of Vitis vinifera L. cv. Cabernet-Sauvignon during development

Plant Physiology and Biochemistry 47 (2009) 282–290

Contents lists available at ScienceDirect

Plant Physiology and Biochemistry journal homepage: www.elsevier.com/locate/plaphy

Research Article

Leucoanthocyanidin reductase and anthocyanidin reductase gene expression and activity in flowers, young berries and skins of Vitis vinifera L. cv. Cabernet-Sauvignon during development Se´verine Gagne´, Soizic Lacampagne, Olivier Claisse, Laurence Ge´ny* Universite´ de Bordeaux, UMR 1219 Oenologie, INRA, ISVV, 210 Chemin de Leysotte, CS 58008, 33 140 Villenave D’Ornon, France

a r t i c l e i n f o

a b s t r a c t

Article history: Received 25 March 2008 Accepted 8 December 2008 Available online 16 December 2008

Proanthocyanidins, or condensed tannins, are crucial polyphenolic compounds for grape and wine quality. Recently, significant advances were achieved in understanding the biosynthesis of their main subunits: (þ)-catechin and ()-epicatechin, produced by catalysis of leucoanthocyanidin reductase (LAR) and anthocyanidin reductase (ANR), respectively. Expression studies had been published but no data were available on enzyme activity. In our work, we devised assays to measure LAR and ANR activity and determine their development throughout the growth of flowers, young berries, and skins of Vitis vinifera L. cv. Cabernet-Sauvignon. We also investigated the accumulation of compounds in these tissues and focused on the expression of both the structural genes and the transcription factors involved in regulating them: VvMYB5a and VvMYBPA1. Biosynthetic genes were expressed early and LAR and ANR were already active during flowering and at the beginning of berry growth, as well as during colour-change in skins. The profiles we determined correlated with total tannin, catechin, and epicatechin concentrations. The involvement of VvMYB5a and VvMYBPA1 was confirmed and specific expression patterns were also established for VvLAR transcripts. Ó 2008 Elsevier Masson SAS. All rights reserved.

Keywords: Vitis vinifera L. Proanthocyanidin Leucoanthocyanidin reductase Anthocyanidin reductase Skin Flowers

1. Introduction Proanthocyanidins, also known as condensed tannins, are polyphenolic secondary metabolites synthesized via the flavonoid pathway. In plants, they act as protective factors for radiation [37], disease [12], and predators [3,18]. Their beneficial dietary effects on animal [28] and human [5] health are widely documented. The organoleptic properties they confer on fruits, i.e. bitterness and astringency [10], and their ability to interact with proteins and other polyphenols [4,6] make them crucial components in grape and wine quality. In grapes, they are mainly located in seeds and in stems but also in skins, where they are reactive and easily extractible. They accumulate early then decrease continuously in grape skins [13,16,31,35]. They contribute to the intensity and stability of wine colour and take part in wine structure. The biosynthesis mechanism of proanthocyanidins, especially in the later stages, has not yet been completely elucidated, but the

Abbreviations: ANR, anthocyanidin reductase; DAA, days after anthesis; FW, fresh weight; LAR, leucoanthocyanidin reductase; NADPH, reduced nicotinamide adenine dinucleotide phosphate. * Tel.: þ33 5 57 57 58 54; fax: þ33 5 57 57 58 13. E-mail address: [email protected] (L. Ge´ny). 0981-9428/$ – see front matter Ó 2008 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.plaphy.2008.12.004

varied, positive effects of these compounds have recently attracted growing interest. Even if flavanol polymerization mechanisms are not yet known, significant progress has been made concerning monomer production. Xie et al. [40] and Tanner et al. [38] clearly established the roles of anthocyanidin reductase (ANR) and leucocyanidin reductase (LAR) in ()-epicatechin and (þ)-catechin synthesis, respectively. ANR (E.C. 1.3.1.77) converts anthocyanidins to 2,3-cis-flavanols via NAPDH-mediated reduction, thus inversing the stereochemistry of the pyran ring at C3 and producing mainly ()-epicatechin. LAR (E.C. 1.17.1.3) catalyzes the NADPH-dependent reduction of leucoanthocyanidins to 2,3-trans-flavanols, such as (þ)-catechin. Several recent studies have focused on the expression and regulation of VvANR and VvLAR in different grape-berry organs. They were highly expressed in leaves and flowers, as well as in skins and seeds up to the colour-change period, when the metabolic pathway shifted to anthocyanin production. These patterns were correlated to flavanol accumulation [19]. Important data were reported concerning the control of their expression. Until now, only two transcription factors related to tannin metabolism were identified in grape. VvMYB5a expression [15] early in berry development was apparently closely correlated to proanthocyanidin accumulation. According to Bogs et al. [8], VvMYBPA1 specifically

S. Gagne´ et al. / Plant Physiology and Biochemistry 47 (2009) 282–290

controlled flavanol synthesis by inducing VvANR and VvLAR expression. Further research is still required to connect the elements involved in proanthocyanidin biosynthesis. No data are currently available on enzyme activities, which would complement the genomic approach. In this work, we developed enzyme assays for LAR and ANR activity throughout grape skin development. Since we hypothesized that the continuous decrease in tannin content we measured in skins was linked to tannin metabolism in flowers (i.e. flower organs differentiate into the different grape-berry tissues), we also analyzed flowers and immature berries. At the same time, we correlated enzyme data with the expression of LAR, ANR, their transcription factors, VvMYB5a and VvMYBPA1, and catechin and epicatechin concentrations both as monomers and polymerized subunits. 2. Materials and methods 2.1. Plant material and sample collection Vitis vinifera L. cv. Cabernet-Sauvignon flowers and berries were sampled from a commercial vineyard in the Pessac-Le´ognan appellation near Bordeaux (France) and collected at different development stages in 2006, according to the phenological stages defined by Eichhorn and Lorenz [17]. The north/south-oriented vineyard had been planted in 1990 and grafted onto 101-14 rootstock. Planting density was 6500 vines per ha, and the pruning method was Guyot double. Flowers were collected at three stages: floral bud (stage 17), flowering onset corresponding to 50% of flowers at inflorescence (stage 21), and full flowering (stage 23). Random samples of five grape clusters on ten vines were selected at ten phenological stages. Four green stages were collected: berry set (stage 27), pea-sized berries (stage 31), ten days after stage 31, and berry touch (stage 33), corresponding to 14, 24, 34, and 45 days after anthesis (DAA), respectively. Four samples were collected during colour-change: 10% red ripe (RR) berries (stage 35), 50% RR (stage 36), 80% RR, and 100% RR (stage 37), corresponding to 59, 61, 63, and 67 DAA, respectively. Two samples were taken during fruit ripening: 2 weeks after the end of the colour-change (83 DAA) and at maturity (harvest, 110 DAA, stage 38). Table 1 summarizes the samplings made for this study. Floral and berry samples were immediately frozen in liquid nitrogen, and stored at 80  C until analysis. Grape skins were carefully removed for analysis using razor blades, starting with pea-

Table 1 Plant material collected for this study. Stage of development

Number of days after anthesis

Phenological stage according to Ref. [17]

Floral bud Flowering onset (50% flowers per inflorescence) Full flowering Berry set Pea-sized berry Pea-sized berry þ 10 days Berry touch 10% red ripe berries per bunch 50% RR berries per bunch 80% RR berries per bunch 100% RR berries per bunch 100% RR þ 2 weeks Maturity or harvest

– –

17 21

– 14 24 34 45 59 61 63 67 83 110

23 27 31 – 33 35 36 – 37 – 38

283

sized berries. Before this stage, the tissue is not sufficiently differentiated for the skins to be removed. 2.2. Enzyme extraction All procedures were carried out at 4  C. LAR and ANR were simultaneously extracted using a method adapted from Ref. [14] to isolate dihydroflavonol 4-reductase (DFR). Previous reports [33,32] presented this method for simultaneous crude extraction of ANR and LAR and validated it for both enzymes. Moreover, this procedure, similar to the crude enzyme extract preparation method used in our laboratory [21], was also rapid, easy and removed anthocyanins from the enzyme extract. Plant material (2 g) was ground to powder in liquid nitrogen, homogenized in 3 mL lysis buffer (0.1 M HEPES pH 7.3, 1% sucrose (w/v), 1% PEG (w/v), 25 mM CaCl2) and mixed with 200 mg PVPP. The homogenate was centrifuged at 20 000 g for 10 min and the supernatant incubated with Dowex 1  2 mesh 200 (Sigma, Saint Quentin Fallavier – France), equilibrated with the lysis buffer. After centrifugation at 20 000 g for 5 min, the supernatant was percolated through a Sephadex G-25 column (GE Healthcare – Amersham Biosciences, Orsay – France). The recovered suspension was used as crude extract to determine enzyme activity. Extractions were performed in triplicates. 2.3. Determining assay parameters Protocols for LAR and ANR assays were adapted from Ref. [32] and from Ref. [40,33,32], respectively. The volume of reaction mixture was initially set (200 mL for LAR and 500 mL for ANR) and different concentrations of dihydroquercetin (Sigma, Saint Quentin Fallavier – France) or cyanidine (Extrasynthe`se, Genay – France) were tested together with different volumes of crude extract, different pHs, and different temperatures with crude extracts from skins at two distinct phenological stages. Kinetic study was done to determine the optimum incubation time. For each assay, five independent assays were performed per phenological stage to confirm the efficiency of both methods. Negative controls, consisting of boiled crude extracts, showed neither production of catechin or epicatechin, nor substrate degradation. The separation method described by Ref. [32] for HPLC analysis was adapted to quantify catechin and epicatechin. 2.4. LAR assay LAR activity (expressed in pkatals g1 FW) was determined by monitoring the conversion of dihydroquercetin to (þ)-catechin. The assay mixture contained 10 mL dihydroquercetin (1 g L1 in methanol), 10 mL NADPH 20 mM and 110 mL Tris–HCl buffer 0.1 M pH 7.5. The reaction was initiated by adding 70 mL crude extract, incubated at 25  C for 30 min, and stopped by adding 200 mL ethyl acetate with vigorous vortexing. Extraction was repeated and the ethyl acetate phases were pooled and dried under nitrogen gas. Residues were dissolved in 100 mL HPLC-grade methanol for HPLC analysis. LAR products were separated on a Beckman Ultrasphere ODS (250  4.6 mm, 5 mm) reversed-phase column and eluted with acetic acid (5% v/v) in water (solvent A) and methanol (solvent B), according to the following program: 5% B from 0 min to 5 min, 5%– 10% B from 5 min to 10 min, 10% from 10 min to 16 min, 10%–90% B from 16 min to 21 min, 90% B from 21 min to 31 min, 90%–5% B from 31 min to 36 min, and 5% B up to 45 min. The flow rate was set at 1 mL/min, the detection wavelength was 280 nm, and the injection volume was 50 mL. Identification and quantification were performed using an external (þ)-catechin standard (Sigma, Saint Quentin Fallavier – France). Data represent the mean of three assays per extract  standard deviation (SD).

S. Gagne´ et al. / Plant Physiology and Biochemistry 47 (2009) 282–290

284

2.5. ANR assay The ANR enzyme assay consisted of quantifying ()-epicatechin formed from cyanidin. The 200 mL reaction mixture contained 40 mL cyanidin chloride (5 g L1in methanol), 50 mL NADPH 20 mM, and 210 mL 0.1 M Tris–HCl buffer at pH 6.0. 200 mL crude extract was added and the mixture was incubated at 35  C for 45 min. Reaction products were extracted twice with 500 mL ethyl acetate, dried under nitrogen gas and dissolved in 100 mL HPLC-grade methanol for HPLC analysis. The HPLC conditions were the same as those in the LAR assay. The elution program was: 5% B from 0 min to 5 min, 5%–10% B from 5 min to 10 min, 10% from 10 min to 15 min, 10%– 15% B from 15 min to 25 min, 15% B from 25 min to 40 min, 15%–90% B from 40 min to 65 min, 90% B from 65 min to 75 min, 90%–5% from 75 min to 80 min, and 5% up to 80 min. The external standard was ()-epicatechin (Sigma, Saint Quentin Fallavier – France). Data (expressed in pkatals g1 FW) represent the mean of three assays per extract  standard deviation (SD). 2.6. RNA extraction and cDNA synthesis Total RNAs were isolated from grape skins as described by Ref. [1]. No DNA contamination was detected by PCR amplification (40 cycles) with VveEF1g (AF176496) and VvUbiquitin1 (BN000705) primers (Table 2). DNase-treated total RNA (1 mg) was reverse transcribed with oligo(dT)15, using MMLV reverse transcriptase (Promega, Charbonnie`res – France) following the supplier’s instructions. cDNA synthesis was controlled by PCR using 1 mL cDNA in a 50 mL reaction with VveEF1g and VvUbiquitin1 primers. 2.7. Expression analysis by real-time quantitative-PCR Transcript levels in grape flowers and skins were measured by real-time Q-PCR, using the IQ-SYBR green supermix on a MyIQÔ Single Colour Real-Time PCR Detection System (Bio-Rad, Marnes La Coquette – France) monitored via the Bio-Rad iQ5 2.0 Standard edition Optical System Software (Bio-Rad, Marnes La Coquette – France). The reaction mixture (20 mL) contained 5 mL cDNA template (0.4 g mL1) and 0.25 mM each of the forward and reverse primers specific to each gene (Table 2). Calibration curves were prepared for each primer set as described by Ref. [25] for a library of grape-berry cDNA. Thermal cycling conditions were as follows: 95  C for 3 min, followed by 60 cycles of 95  C (10 s), 60  C (20 s), and 72  C (10 s). Annealing temperature (60  C) was determined hypothetically when designing primers. Amplification specificity (specific melt temperature) was checked for each gene product at the end of each

Table 2 PCR primers used for real-time PCR. Gene name

GenBank accession no.

Orientation

Sequence 50 / 30

VvUbiquitin1

BN000705

VveEF1g

AF176496

VvLAR1

DQ129685

VvLAR2

DQ129686

VvANR

DQ129684

VvMYB5a

AY555190

VvMYBPA1

AM259485

F R F R F R F R F R F R F R

CTT GAT GGG ACA GGT CTG TGT CTT GGA GGC AGG ATA GAA GGT TGA CCT CTC GGA CAG AAG AGC CTC TCC CTC CTC CAA CGG ATT TCT TCC CGT CCA CTG TTT TCA TCG T TAA ACG AGC TGG CAT CAC GCA GCG GCT AGT AGG TCA CTT GAT GGG ACA GGT CTG GT TGT CTT GGA GGC AGG ATA GC CCG CCT AAC CTG GAT CAG TA CCA TGG GTG CTT TGT AGT CC AGA TCA ACT GGT TAT GCT TGC T AAC ACA AAT GTA CAT CGC ACA C

F: forward; R: reverse.

run by performing a melting cycle from 50  C to 96  C and by gel electrophoresis and sequence analysis. PCR efficiency and standard curve linearity were also evaluated in preliminary experiments to ensure % efficiency between 85 and 115 and an r2 value  0.95. Transcript levels (mmoles cDNA) for each flavonoid gene were normalized to VvUbiquitin1 and VveF1g constitutively expressed transcripts (182-bp and 87-bp products, respectively). Three replicates of all real-time Q-PCR reactions were carried out per sample. 2.8. Tannin extraction and total tannin assays Tannins were extracted according to Ref. [21] from approximately 1 g (fresh weight) plant material, ground to powder in liquid nitrogen. Samples were macerated twice in 40 mL MeOH/12 N HCl (99.9:0.1; v/v), with stirring, at room temperature, for 3 h. After incubation, the extract was filtered on a 100 mm FlaconÒ filter (VWR, Fontenay-sous-Bois – France) and stored at 80  C until analysis. Each sample was extracted three times. Total tannin content was determined by spectrophotometry, according to the method of Ribe´reau-Gayon and Stonestreet [34]. After an appropriate dilution in distilled water, the sample was mixed (v/v) with 12 N HCl and separated into two aliquots. One was incubated for 30 min at 100  C to hydrolyse proanthocyanidins into anthocyanins, then rapidly cooled and completed with EtOH 95% (v/v), while the other aliquot was not submitted to hydrolysis. Finally, optic density was measured at 550 nm. The difference between the two measurements specifically corresponds to tannin optic density. The coefficient determined by Ref. [34] was used to calculate tannin contents. Data are expressed in milligrams of total tannins per gram of fresh weight (mg g1 FW), or in milligrams of total tannins per berry (mg/berry), and represent the mean  standard deviation (SD) of three replicates per extract. 2.9. Quantification of catechin and epicatechin monomers and polymers Flavan-3-ols were quantified by analysis of 50 mL methanol extracts of phenolic compounds according to Ref. [36] using the same HPLC conditions as above and the following elution program: 30–60% B from 0–15 min, 100% B from 16–25 min, 30% B from 26– 30 min. Data represent the mean  standard deviation (SD) of three replicates per extract. The catechin and epicatechin composition of tannin polymers was determined after thioacidolysis of 100 mL phenolic extract (prepared as described above) in 100 mL thiolysis reagent (benzylmercaptan/2 N HCl/MeOH; 0.5/2/7.5; v/v/v) at 60  C for 5 h as previously described by Ref. [2]. 150 mL aliquot of 100 mg L1 4methylcatechol solution was added to stop the reaction. HLPC analysis gradients were modified for the thiolysis products as follows: 30–40% B from 0–15 min, 40–70% B from 15–30 min, 70– 100% B from 30–40 min, 100% B from 40–45 min, 100–30% B from 45–46 min, 30% B from 46–50 min. Flavan-3-ols and thioethers were identified by retention time and absorption spectrum in comparison with authentic commercial molecules and thiolysis products previously purified and characterized in the laboratory. Data represent the mean  standard deviation of three replicates per extract. 2.10. Statistical analysis Results were compared by one-way ANOVA and Newman–Keuls multiple range test, taking p < 0.05 to assess statistical significance between stages of development, using Statistica VI Software (StatSoft Inc., Tulsa, OK – USA).

S. Gagne´ et al. / Plant Physiology and Biochemistry 47 (2009) 282–290

3. Results and discussion 3.1. LAR and ANR enzyme activity during floral and berry development The first stage of our study consisted of developing assays to investigate LAR and ANR activity in grape tissue. We adapted the extraction procedure described by Dellus et al. [14] to isolate both enzymes. For the LAR assay, we applied the protocol described by Pfeiffer et al. [32], with slight modifications, and adapted procedures from and Xie et al. [40], Punyasiri et al. [33] and Pfeiffer et al. [32] for the ANR assay. LAR and ANR activity was determined at three stages in the flower development and two in young berries (berry set and peasized berries), as well as throughout grape skin growth. Both enzymes were characterized by different patterns and levels of activity but high enzyme activity was detected at the onset of flowering and in the early green stage for both LAR and ANR. Since no data were available on enzyme activity, we compared our results to previous data on enzyme expression and flavanol accumulation. LAR presented two activation peaks in flowers (Fig. 1A): in floral buds (phenological stage 17) and at berry set (phenological stage 27). In grape skins (Fig. 1B), the highest LAR activity level was detected in pea-sized berries (24 DAA, phenological stage 31) and the activity dropped rapidly. A second activation point occurred at 67 DAA (stage 37), corresponding to the end of colour-change. However, maximum LAR activity in skin was 7.5-fold lower than in flowers. These data are consistent with expression studies showing strong LAR expression in flowers and lower transcript levels in skins [7,19], with VvLAR1 as the most abundant transcript in Shiraz and VvLAR2 in Cabernet-Sauvignon.

A

a

200

In flowers (Fig. 2A), ANR exhibited a similar pattern to LAR with two activation peaks, but the ANR activity at berry set (phenological stage 27) was higher than in floral buds (stage 17). In berry skins (Fig. 2B), ANR activity profiles were distinct from LAR, remaining at maximum level for at least 10 days in the early stages of berry development, then decreasing more gradually to a minimum and then rising at harvest. ANR activity was at least fivefold higher in skin than in flowers and young berries, which could reflect different mRNA levels between these tissues as previously shown in Shiraz [7]. However, this differential patterning had not previously been observed between Cabernet-Sauvignon flowers and skin [19]. Considering enzyme activity level, such dramatic variations have already been shown between enzymes from the upper flavonoid pathway. For instance, Hrazdina et al. [24] measured a chalcone isomerase (CHI) activity and a 4-coumarate: coenzyme A ligase (4CL) activity in grape berry 100-fold and 10-fold respectively higher than phenylalanine ammonia-lyase (PAL) activity. Similarly, Chen et al. [11] determined that cinnamate-4-hydroxylase (C4H) enzyme activity was 10-fold the PAL activity though C4H is located immediately downstream PAL in the biosynthetic pathway. ANR was more active than LAR in skins, possibly due to a previous higher activity. ANR was active in the whole immature berries whereas LAR activity became negligible after berry set: ANR enzyme activity was a 3.5-fold higher enzyme activity than LAR at berry set (phenological stage 27), which may explain the differences in activity level measured in the isolated skins of pea-sized berries (24 days after anthesis, stage 31). It can be noticed that this higher ANR activity in young green skins also correlated with the difference in transcript levels previously determined: Fujita et al. [19] detected that VvANR expression in Cabernet-Sauvignon skin was two to three times higher than that of VvLAR in young green

150

ANR activity pkat.g-1 FW

LAR activity pkat.g-1 FW

A

full flowering

100

c b

50

b d

0 17

21

23

27

285

c

250 full flowering

200 a

150

d

100

0 17

31

21

23

a colour-change 20 b b

b

b b

b

b

b

0 20

40 31

33

60 35 36 37

80

ANR activity pkat,g-1 FW

LAR activity pkat. g-1 FW

B

30

10

27

31

Phenological stage

Phenological stage

B

b

b

50

a

1000

a

colour-change

750 b

500

b cc

250

c d

d

0

100 38

DAA Phenological stage Fig. 1. LAR activity, expressed in picokatals per gram fresh weight, in: (A) grape flowers and young berries and (B) grape skins throughout growth. Data represent the mean of three determinations  standard deviation (error bars). Values denoted by the same letter are not statistically different according to the Newman–Keuls test (p < 0.05).

20

31

40

60 80 33 35 36 37 DAA Phenological stage

100

38

Fig. 2. ANR activity, expressed in picokatals per gram of fresh weight, in: (A) grape flowers and young berries and (B) grape skins throughout growth. Data represent the mean of three determinations  standard deviation (error bars). Values denoted by the same letter are not statistically different according to the Newman–Keuls test (p < 0.05).

S. Gagne´ et al. / Plant Physiology and Biochemistry 47 (2009) 282–290

berries and Bogs et al. [7] observed that VvANR expression was ten times that in Shiraz skin after flowering than VvLAR1 and VvLAR2. Moreover, the tannin-specific flavanol enzyme activities established in our study showed different patterns compared to upperpathway enzymes measured elsewhere [24,11]. These latter enzymes are characterized by two activation periods and stronger activity in coloured than green berries. Tannin biosynthesis is, therefore, nearly complete at an early stage in grape development, and upstream enzyme activity is likely related to anthocyanin production.

A 1,E+04

b colour-change

1,E+02

VvLAR1

286

a ac ac

1,E+00 1,E-02

d dd

1,E-04 10

30 31

3.2. Changes in LAR and ANR expression during grape ripening in skins

50 33

70 35

90

37

110 38

DAA Phenological stage

B

colour-change 8 b

b

VvLAR2

6 4 a c

c

2

e

dd

d

0 10

30 31

50 33

70 35 36 37

90

110 38

DAA Phenological stage a

C 1,E+01

colour-change

b

1,E+00

VvANR

The mRNA levels of VvLAR1, VvLAR2, VvANR, VvMYB5a, and VvMYBPA1 were determined by real-time quantitative-PCR during skin development. Results are shown in Fig. 3 for biosynthetic genes and Fig. 4 for transcription factor genes. Considering LAR, two LAR gene families have been reported in Vitis [7,19] and we analyzed the expression of the two corresponding transcripts, VvLAR1 and VvLAR2, considering that our LAR enzyme assay can be a mixture of two LAR isoforms. It has to be pointed out that the functional activity has already been shown only for VvLAR1 and not for VvLAR2 [7,30]. VvLAR1, VvLAR2, and VvANR had different transcription patterns but levels for all three genes were always highest in the skins of young berries. VvLAR1 mRNA peaked at a very high level 34 days after anthesis and then declined rapidly (Fig. 3A). VvLAR1 expression increased again from the middle of the colour-change period (63 DAA) to the middle of ripening (Fig. 3A). Transcript levels were still high at maturity (110 DAA, stage 38). VvLAR2 transcription exhibited two accumulation periods (Fig. 3B). The first peak also occurred in small-berry skins 34 days after anthesis, while a second peak was detected at the 100% ve´raison stage (at 67 days after anthesis, phenological stage 37). Levels dramatically increased as soon as the middle of the colour-change period (61 days after anthesis, phonological stage 36). The absolute values of these maxima were considerably lower than those of VvLAR1 mRNA, i.e. around 392 at 31 days after anthesis, compared to six for VvLAR2. Our results are only partially consistent with the literature [7,19]. VvLAR2 was previously found to be expressed at a higher level than VvLAR1, but the dramatic increase in VvLAR1 mRNA in green-berry skins had not previously been reported. However, our results confirmed the major role of VvLAR2 during colour-change, while VvLAR1 expression was negligible. At this time, LAR enzyme activity would be due to VvLAR2 expression. In green berries, LAR activity and transcript abundance could also be correlated. At the beginning of skin development (24 days after anthesis, phenological stage 31), LAR activity may be related to both VvLAR1 and VvLAR2 expression since these two transcripts were synthesized at that time. Afterwards, the constant low enzyme activity measured during the growth may be supported by the dramatic increase in VvLAR1 and VvLAR2 expression, although this was not followed by a similar increase in enzyme activity. We can suppose that other regulation mechanisms might be involved to explain this discrepancy. Our results show an accumulation of VvLAR2 mRNA at the end of colour-change, when tannins are also synthesized [9,13,16]. We can then hypothesize that VvLAR2 product may be efficient in competing with anthocyanin synthase during colour-change when functional. This differential pattern between VvLAR1 and VvLAR2 could also suggest the involvement of two LAR isoforms, specific to the developmental stage or purpose, as demonstrated for phenylalanine ammonia-lyase (PAL) [23,11] for instance. These works described two PAL isoforms, with different isoelectric points and molecular masses. In both studies, only one

ac

c

bc bc

1,E-01 c

c

c

1,E-02

c

1,E-03 e

1,E-04 10 31

30 33

50

70 35 36 37

90

110 38

DAA Phenological stage Fig. 3. Gene expression of: (A) VvLAR1, (B) VvLAR2, and (C) VvANR in grape skins throughout development. Expression was determined by real-time quantitative-PCR and indicates the molar ratio of the mRNA level of each gene relative to that of the mean level of VvUbiquitin1 and VveEF1g for each sample. Data represent the mean of three determinations  standard deviation (error bars). Values denoted by the same letter are not statistically different according to the Newman–Keuls test (p < 0.05).

isoform was present during colour-change and was assumed to be dedicated to anthocyanin biosynthesis. Indeed, PAL activation is considered as product-specific, i.e. for lignin or tannin or anthocyanin production [29,27,26]. Thus, each VvLAR transcript may be regulated in accordance with its tissue-specificity [7] or with the developmental stage. Like for LAR activity, tannin biosynthesis may be supported by VvLAR2 expression during colour-change. VvANR mRNA was also characterized by a pattern similar to those described for VvLAR expression but levels were quite inferior, especially from the beginning of the colour-change period (Fig. 3C). The highest mRNA level was detected in the skins of pea-sized berries (24 days after anthesis, phenological stage 31), then decreased dramatically within 10 days, remaining very low and nearly constant until the end of skin development. This profile is similar to those previously established [7,19] and suggests a precocious expression of VvANR, as observed in Shiraz and

S. Gagne´ et al. / Plant Physiology and Biochemistry 47 (2009) 282–290

A 15

genes to draw definitive conclusions about the role of each transcription factor. Further studies are needed to specify if VvANR and VvLAR genes can be regulated at different developmental stages either by VvMYBPA1 or VvMYB5a or both of them.

10

colour-change

3.3. Tannin and flavan-3-ol composition during grape growth

b

b 5

c

c c d

d

0 10

30 31

50 33

70 35

90

37

110 38

DAA Phenological stage b

colour-change 4

A

a

a dd

c

d

d

c

0 10

30 31

50 33

35

70 37

90

110 38

DAA Phenological stage Fig. 4. Gene expression of (A) VvMYB5a and (B) VvMYBPA1 in grape skins throughout development. Expression was determined by real-time quantitative-PCR and indicates the molar ratio of the mRNA level of each gene relative to that of the mean level of VvUbiquitin1 and VveEF1g for each sample. Data represent the mean of three determinations  standard deviation (error bars). Values denoted by the same letter are not statistically different according to the Newman–Keuls test (p < 0.05).

Cabernet-Sauvignon flowers [7,19]. Moreover, VvANR expression and ANR enzyme activity profiles were similar and included the expected delay in enzyme activity compared to expression. Despite a low level, the expression detected during maturation (from colour-change to maturity) for each transcript described could explain that enzyme activity of both enzymes was detected at that time of skin development. The VvMYB5a and VvMYBPA1 transcription factors exhibited similar expression patterns to VvLAR2 with two accumulation peaks: the first in green-berry skins 24 days after anthesis (stage 31) and the second towards colour-change. These profiles reflect flavanol accumulation (see below), as well as gene expression and enzyme activity, supporting their essential involvement in controlling the grape proanthocyanidin metabolism. VvMYB5a (Fig. 4A) peaked during the first growth period and then declined up to the beginning of the colour-change. Expression increased again slightly during the colour-change stage, which is a little later than previously reported [15]. The VvMYBPA1 expression profile (Fig. 4B) was characterized by transient, more dramatic variations. The level was relatively low at all developmental stages except at 34 and 63 days after anthesis, when it reached 11 and two, respectively. Until now, only the second peak of VvMYBPA1 expression in grape skin, in the middle of ripening, had been reported [8]. The profiles of these transcription factor genes parallel the peaks of the enzyme transcripts supporting the involvement of VvMYBPA1 and VvMYB5a in the control of VvLAR and VvANR expression. The transcript levels are very similar between VvMYBPA1 and VvMYB5a (Fig. 4) and big differences in the expression values (Fig. 3) can be noticed among the three structural

Tannin contents mg,g-1 FW

8 full flowering 60 a

a

a

40

a b

20

0 17

21

23

27

31

Phenological stage

B

Tannin contents mg,g-1 FW

VvMYBPA

B 12

Total tannin, catechin and epicatechin (both as monomers and polymerized forms) contents were determined in grape flowers and skins during growth (Figs. 5–7, respectively). Although thiolysis makes it possible to quantify other flavan-3-ols (epigallocatechin and epicatechin-3-O-gallate, for instance), the results only focus on catechin and epicatechin, since the capability of ANR to synthesize epigallocatechin has never been clearly demonstrated. Total tannin content was measured in both flowers and skins. A typical profile was observed for skins (Fig. 5B), contents decreasing from berry set to harvest [16,31,22]. A slight increase, from 59 to 67 DAA, was noticed during colour-change, when anthocyanins are

75

a

60

b

bb

b

b

b

b

b

45 30

colour-change 15 0 10

30 31

50 33

70 35

90

37

110 38

DAA Phenological stage

C

Tannin contents mg/berry

VvMYB5a

a a

287

10 a 8

a

ab

ab

b b

6

b c

4

b

2

colour-change

0 10

30 31

50 33

70 35

37

90

110 38

DAA Phenological stage Fig. 5. Total tannin content in: (A) grape flowers and young berries and (B, C) grape skins throughout growth. Data represent the mean of three independent replicates  standard deviation (error bars). Letters indicate statistical differences, according to the Newman–Keuls test (p < 0.05).

S. Gagne´ et al. / Plant Physiology and Biochemistry 47 (2009) 282–290

288

4

A

a a

2

a

a

1

b

0 17

21

23

27

31

Phenological stage

Catechin mg.g-1 FW

a

0,1

c

0,0 21

23

27

31

Phenological stage ab

ab

2 a

Epicatechin mg.g-1 FW

B

1

21

23

27

Phenological stage 0,06

a

18

b ab a

12

c 6

0 17

21

23

27

Phenological stage

0,04 b

b

b

b bc

c

0,02

bc

b

C

0 24

31

31

45

59

61

33

35

36

63

67

83

37

110 38

DAA Phenological stage

Catechin mg. g-1 FW

ab a

17

17

D

b

b

0

C

0,2

3

20

Epicatechin mg.g-1 FW

Catechin mg.g-1 FW

B

Epicatechin µg.g-1 FW

3

0,015 a

a

a

0,010

a

b c 0,005

c

b

c

0,000

a

24

31

31

d 10

45

59

61

33

35

36

63

67

83

37

110 38

DAA Phenological stage b

b c

c

c

c

c

67 37

83

110 38

0 24 31

31

45 33

59 35

61 36

63

DAA Phenological stage Fig. 6. Catechin content: (A, C) as free monomers and (B, D) as subunits in polymerized tannins in (A, B) flowers and young berries and (C, D) grape skins. Data represent the mean of three independent replicates  standard deviation (error bars). Values denoted by the same letter are not statistically different according to the Newman– Keuls test (p < 0.05).

D

Epicatechin mg.g-1 FW

Catechin mg.g-1 FW

A

30

a

20 b 10

b

bc

bc

bc

67

83

110

c 0 24 31

known to accumulate rapidly in large amounts, requiring an active biosynthetic flow. As tannins and anthocyanins share the same upstream pathway, the activation of anthocyanin production may deliver precursors, which are partially redirected to tannin formation. This increase is, however, transient and the tannin content then remains stable during fruit maturation. Tannin contents were also expressed per berry (Fig. 5C), which slightly modified the profile shown in Fig. 5B. Considering the

b b

31

45

59

61

33

35

36

63

37

38

DAA Phenological stage Fig. 7. Epicatechin content: (A, C) as free monomers and (B, D) as subunits in polymerized tannins in (A, B) flowers and young berries and (C, D) grape skins. Data represent the mean of three independent replicates  standard deviation (error bars). Values denoted by the same letter are not statistically different according to the Newman–Keuls test (p < 0.05).

S. Gagne´ et al. / Plant Physiology and Biochemistry 47 (2009) 282–290

entire development of the berry, tannin contents decreased. Such a profile differed from patterns published by Downey et al. [16] or Fujita et al. [20] and it can be hypothesized that these differences might reflect cultivar, terroir or vintage features. It can also be noticed that the accumulation occurring at colour-change was not as obvious as in Fig. 5B. This could be mainly due to the rapid increase in berry size from this developmental stage while enzyme activity only slightly increased, as shown above (Figs. 1B and 2B). Similarly, the decrease in contents that followed colour-change was also a consequence of both the slowing down of tannin synthesis and the significant enlargement of the berry. However, the main difference was observed at the beginning of the first growth period (between 24 and 31 DAA) when contents tended to increase. Despite a restricted statistical significance, this trend might be a consequence of the high activity measured for ANR, and in a lesser extent for LAR, at that time. Because profiles were not dramatically different between the two modes of expression (figures B and C), contents are only expressed per gram of fresh weight in the following paragraphs. Flowers exhibited high stable tannin concentrations, which decreased after berry set (Fig. 5A). Higher contents were measured in skins than in flowers, which was due to the accumulation of tannins mainly in grape skins and seeds. Decreasing contents in seed tannins have also been reported [16]. We may thus hypothesize that mechanisms leading to the decrease in tannin contents (polymerization, oxidization, interactions.) may be early initiated in fruit development. For both flowers and skins, the LAR and ANR enzyme profiles tended to parallel the changes in tannin contents (Fig. 5B): a significant activity was measured at the beginning of flowering, in the early stages of grape skin and during colour-change, when total tannin contents were the highest. In developing flowers and berries, catechin was present in equivalent amounts in the free monomer (Fig. 6A) and polymerized subunit forms (Fig. 6B). The monomer content was stable up to berry set, while the proportion of catechin in the polymers increased constantly during flowering. The polymerized catechin content rose during flower development and berry set, but remained constant in immature berries, whereas the monomer concentration decreased considerably in pea-sized berries in a way similar to the drop in total tannin content. This pattern correlated with LAR activity, especially in green berries when the decrease in activity after berry set was accompanied by a sharp drop in catechin monomers. The catechin content of the polymers was not noticeably affected by this reduction in LAR activity, suggesting that the pool of catechin monomers was sufficient to maintain condensed tannin levels. In grape skins, catechin monomers (Fig. 6C) were present in very small amounts, 50- to 250-fold lower than the polymerized forms (Fig. 6D). A significant drop occurred early in growth (31 DAA) while LAR activity also declined. Then concentrations remained nearly constant throughout development. The catechin content of condensed tannins decreased constantly, but rose transiently at 61 DAA, i.e. the onset of colour-change. As discussed above, this may be due to activation of the upstream biosynthetic pathway for anthocyanin production. Epicatechin showed a similar pattern to catechin in developing flowers and berries, but with more marked changes. The 5-fold decrease in epicatechin contents between berry set and pea-sized berries was more noticeable than the decrease in catechin but had a limited impact on total concentrations of flavan-3-ol monomers, due to the low level of epicatechin monomers present. In contrast, polymerized epicatechin (Fig. 7B) was two to eight times more abundant in condensed tannins than catechin. The fact that quantities of epicatechin in free monomer form were negligible, compared to both catechin and polymerized epicatechin, shows

289

that this flavanol is mainly used as an extension subunit, in agreement with literature data collected from several plants [39]. Epicatechin in grape skins was also mainly in the polymer form (Fig. 7C and D) and decreased continuously during growth, as observed for ANR activity. Epicatechin monomers decreased during the first active growth period, then increased again in the middle of colour-change, maintaining higher but fluctuating concentrations up to harvest. Although the correlation did not fit for polymerized epicatechin, this slight increase may be explained by the increase in ANR activity. The decrease in total tannin contents observed after berry set both in flowers and berry skins and also after colour-change are supported by decrease in catechin and epicatechin subunits in polymers (Figs. 5–7). The more dramatic variations determined for epicatechin may indicate that condensed epicatechin is one of the main factors controlling overall tannin content. These findings also suggested that other ()-flavan-3-ols, such as epigallocatechin, could be involved in this regulation, even if they are not major components of condensed tannins in grapes. Considering polymerized forms, epicatechin contents were higher than catechin contents in flowers, young berries and throughout skin development. These results are concordant with a constantly higher ANR activity compared to LAR activity, even though variations in contents do not strictly reflect variation in enzyme activity. Focusing on epigallocatechin changes might also provide complementary data to help understanding ANR activation. Correlations between enzyme activity, expression data and flavan-3-ol contents were not easy to establish strictly but some points can be underlined. Firstly, VvANR, VvLAR1 and VvLAR2 expression, as previously described in the literature, did not match the catechin and epicatechin patterns, since variations in flavanol concentrations did not parallel marked changes in mRNA levels [7]. In our experiments, variations in contents and in mRNA levels seem to be correlated, especially when focusing on polymer subunits, but the dramatic changes noticed in expression patterns have a limited impact on catechin and epicatechin contents. In addition, the earlier activity of ANR compared to LAR here supports the higher epicatechin contents compared to catechin in small and pea-sized berries. Secondly, the increase in VvLAR2 transcription during colour-change corresponded to an increase in polymerized catechin and epicatechin in a lesser extent. The enzyme assay we developed does not discriminate between the two LAR isoforms present in Vitis. The functionality of VvLAR1 was proved by Bogs et al. [7] but data have still to be provided for VvLAR2. Recent studies [30] failed to reveal any functional activity for the second LAR gene in Lotus corniculatus but our results suggest that VvLAR2 may be functional and involved during colour-change. 4. Conclusion For the first time, this report provides data on LAR and ANR activity in both grape flowers and young berries, as well as in grape skins throughout growth. Both LAR and ANR are activated early and synthesized during flowering and berry development, facilitating precocious tannin biosynthesis. These patterns are correlated with concentrations of catechin and epicatechin, considered as monomers or polymer subunits, as well as with gene expression. Whereas VvLAR1 and VvLAR2 transcripts were detected in greenberry skins, VvLAR2 was revealed to be specific to ripening skins, suggesting the existence of two active LAR isoforms. Our data also supported the involvement of both VvMYB5a and VvMYBPA1 in the control of VvLAR and VvANR expression. This study thus fills a gap in our knowledge of the tannin metabolism, since no previous studies focused on enzyme activity.

290

S. Gagne´ et al. / Plant Physiology and Biochemistry 47 (2009) 282–290

Little by little, the various steps in tannin synthesis are elucidated and connected to each other. Further work is required to clarify the polymerization mechanisms and identify factors regulating this biosynthesis. We are developing a similar approach to study growth regulators and assess their involvement in controlling the proanthocyanidin metabolism. Acknowledgement The authors acknowledge and thank the CIVB (Bordeaux Wine Council) for funding and Dr. François Barrieu for providing us with the grape cDNA library. References [1] M.H. Asif, P. Dhawan, P. Nath, A simple procedure for the isolation of high quality RNA from ripening banana fruit, Plant Mol. Biol. 18 (2000) 109–115. [2] V. Atanasova, H. Fulcrand, V. Cheynier, M. Moutounet, Effect of oxygenation on polyphenol changes occurring in the course of wine-making, Anal. Chim. Acta 458 (2002) 15–27. [3] N. Aziz, N.L. Paiva, G.D. May, R.A. Dixon, Profiling the transcriptome of alfalfa glandular trichomes, Planta 221 (2005) 28–38. [4] E.C. Bate-Smith, T. Swain, Recent developments in the chemotaxonomy of flavonoid compounds, Lloydia 28 (4) (1965) 313–331. [5] G.R. Beecher, Proanthocyanidins, biological activities associated with human health, Pharm. Biol. 42 (2004) 2–20. [6] S.J. Bloor, R. Falshaw, Covalently linked anthocyanin-flavanol pigments from blue Agapanthus flowers, Phytochemistry 53 (2000) 575–579. [7] J. Bogs, M.O. Downey, J.S. Harvey, A.R. Ashton, G.J. Tanner, S.P. Robinson, Proanthocyanidin synthesis and expression of genes encoding leucoanthocyanidin reductase and anthocyanidin reductase in developing grape berries and grapevine leaves, Plant Physiol. 139 (2) (2005) 652–663. [8] J. Bogs, F.W. Jaffe, A.M. Takos, A.R. Walker, S.P. Robinson, The grapevine transcription factor VVMYBPA1 regulates proanthocyanidin synthesis during fruit development, Plant Physiol. 143 (3) (2007) 1347–1361. [9] P.K. Boss, C. Davies, S.P. Robinson, Analysis of the expression of anthocyanin pathway genes in developing Vitis vinifera L. cv Shiraz grape berries and the implications for pathway regulation, Plant Physiol. 111 (4) (1996) 1059–1066. [10] F. Brossaud, V.A.C.N. Cheynier, Bitterness and astringency of grape and wine polyphenols, Aust. J. Grape Wine Res. 7 (2001) 33–39. [11] J.Y. Chen, P.F. Wen, W.F. Kong, Q.H. Pan, S.B. Wan, W.D. Huang, Changes and subcellular localizations of the enzymes involved in phenylpropanoid metabolism during grape berry development, J. Plant Physiol. 163 (2) (2006) 115–127. [12] G.H. Dai, C. Andary, L. Mondolot-Cosson, D. Boubals, Histochemical studies on the interaction between three species of grapevine, Vitis vinifera, V. rupestris and V. rotundifolia and the downy mildew, Plasmopara viticola, Physiol. Mol. Plant Pathol. 46 (1995) 3–8. [13] V. De Freitas, Y. Glories, Concentration and compositional changes of procyanidins in grape seeds and skins of white Vitis vinifera varieties, J. Sci. Food Agric. 79 (1999) 1601–1606. [14] V. Dellus, W. Heller, H.J. Sandermann, A. Scalbert, Dihydroflavonol 4-reductase activity in lignocellulosic tissues, Phytochemistry 45 (1997) 387–398. [15] L. Deluc, F. Barrieu, C. Marchive, V. Lauvergeat, A. Decendit, T. Richard, J.P. Carde, J.M. Merillon, S. Hamdi, Characterization of a grapevine R2R3-MYB transcription factor that regulates the phenylpropanoid pathway, Plant Physiol. 140 (2) (2006) 499–511. [16] M.O. Downey, J.S. Harvey, S.P. Robinson, Analysis of tannins in seeds and skins of Shiraz grapes throughout berry development, Aust. J. Grape Wine Res. 9 (2003) 15–27. [17] K.W. Eichhorn, D.H. Lorenz, Pha¨nologische Entwicklungsstadien der Rebe, Nachr. Dtsch. Pflanzenschutzd. (Braunschweig) 29 (1977) 119–120. [18] R.E. Forkner, R.J. Marquis, J.T. Lill, Feeny revisited: condensed tannins as antiherbivore defences in leaf-chewing herbivore communities of Quercus, Ecol. Entomol. 29 (2) (2004) 174–187.

[19] A. Fujita, N. Soma, N. Goto-Yamamoto, A. Mizuno, K. Kiso, K. Hashizume, Effect of shading on proanthocyanidin biosynthesis in the grape berry, J. Jap. Soc. Hort. Sci. 76 (2) (2007) 112–119. [20] A. Fujita, N. Soma, N. Goto-Yamamoto, H. Shindo, T. Kakuta, T. Koizumi, K. Hashizume, Anthocyanidin reductase gene expression and accumulation of flavan-3-ols in grape berry, Am. J. Enol. Vitic. 56 (4) (2005) 336–342. [21] S. Gagne´, K. Este`ve, C. Deytieux, C. Saucier, L. Ge´ny, Influence of abscisic acid in triggering ‘‘ve´raison’’ in grape berry skins of Vitis vinifera L. cv. CabernetSauvignon, J. Int. Sci. Vigne Vin 40 (1) (2006) 7–14. [22] S. Gagne´, C. Saucier, L. Ge´ny, Composition and cellular localization of tannins in Cabernet Sauvignon skins during growth, J. Agric. Food Chem. 54 (25) (2006) 9465–9471. [23] S. Hiratsuka, H. Onodera, Y. Kawai, T. Kubo, H. Itoh, R. Wada, Enzyme activity changes during anthocyanin synthesis in ‘‘Olympia’’ grape berries, Sci. Hort. 90 (2001) 255–264. [24] G. Hrazdina, G.F. Parsons, L.R. Mattick, Physiological and biochemical events during development and maturation of grape berries, Am. J. Enol. Vitic. 35 (4) (1984) 220–227. [25] S.T. Jeong, O.N. Goto-Yamamoto, S. Kobayashi, M. Esaka, Effects of plant hormones and shading on the accumulation of anthocyanins and the expression of anthocyanin biosynthetic genes in grape berry skins, Plant Sci. 167 (2004) 247–252. [26] Y.Y. Kao, S.A. Harding, C.J. Tsai, Differential expression of two distinct phenylalanine ammonia-lyase genes in condensed tannin-accumulating and lignifying cells of Quaking Aspen, Plant Physiol. 130 (2002) 796–807. [27] A. Kumar, B.E. Ellis, The phenylalanine ammonia-lyase gene family in raspberry. Structure, expression, and evolution, Plant Physiol. 127 (2001) 230–239. [28] L.R. McMahon, T.A. McAllister, B.P. Berg, W. Majak, S.N. Acharya, J.D. Popp, B.E. Coulman, Y. Wang, K.J. Cheng, A review of the effects of forage condensed tannins on ruminal fermentation and bloat in grazing cattle, Can. J. Plant Sci. 80 (2000) 469–485. [29] G. Neumann, B. Schwemmle, Flavonoids from Oenothera-seedlings: identification and extracellular control of their biosynthesis, J. Plant Physiol. 142 (2) (1993) 135–143. [30] F. Paolocci, M.P. Robbins, L. Madeo, S. Arcioni, F. Damiani, Ectopic expression of a basic helix-loop-helix gene trans-activates parallel pathways of proanthocyanidin biosynthesis. Structure, expression analysis, and genetic control of leucoanthcyanidin 4-reductase and anthocyanidin reductase genes in Lotus corniculatus, Plant Physiol. 143 (1) (2007) 504–516. [31] A. Pena-Neira, M. Duenas, A. Duarte, T. Hernandez, I. Estrella, E. Loyola, Effects of ripening stages and of plant vegetative vigor on the phenolic composition of grapes (Vitis vinifera L.) cv. Cabernet Sauvignon in the Maipo Valley (Chile), Vitis 43 (2) (2004) 51–57. [32] J. Pfeiffer, C. Ku¨hnel, J. Brandt, D. Duy, N. Punyasiri, G. Forkmann, T.C. Fischer, Biosynthesis of flavan-3-ols by leucoanthocyanidin 4-reductases and anthocyanidin reductases in leaves of grape (Vitis vinifera L.), apple (Malus x domestica Borkh.) and other crops, Plant Physiol. Biochem. (2006). [33] P.A.N. Punyasiri, I.S.B. Abeysinghe, V. Kumar, D. Treutter, D. Duy, C. Gosch, S. Martens, G. Forkmann, T.C. Fischer, Flavonoid biosynthesis in the tea plant Camellia sinensis: properties of enzymes of the prominent epicatechin and catechin pathways, Arch. Biochem. Biophys. 431 (1) (2004) 22–30. [34] P. Ribe´reau-Gayon, E. Stonestreet, Le dosage des tanins dans le vin rouge et de´termination de leur structure, Chim. Anal. 48 (1966) 188–192. [35] G. Roby, J.F. Harbertson, D.A. Adams, M.A. Matthews, Berry size and wine water deficits as factors in winegrape composition: anthocyanins and tannins, Aust. J. Grape Wine Res. 10 (2004) 100–107. [36] C. Saucier, M. Mirabel, F. Daviaud, A. Longieras, Y. Glories, Rapid fractionation of grape seed proanthocyanidins, J. Agric. Food Chem. 49 (2001) 5732–5735. [37] A. Solovchenko, M. Schmitz-Eiberger, Significance of skin flavonoids for UV-Bprotection in apple fruits, J. Exp. Bot. 54 (2003) 1977–1984. [38] G.J. Tanner, K.T. Francki, S. Abrahams, J.M. Watson, P.J. Larkin, A.R. Ashton, Proanthocyanidin biosynthesis in plants. Purification of legume leucoanthocyanidin reductase and molecular cloning of its cDNA, J. Biol. Chem. 278 (34) (2003) 31647–31656. [39] D.Y. Xie, R.A. Dixon, Proanthocyanidin biosynthesis – still more questions than answers? Phytochemistry 66 (2005) 2127–2144. [40] D.Y. Xie, S.B. Sharma, N.L. Paiva, D. Ferreira, R.A. Dixon, Role of anthocyanidin reductase, encoded by BANYULS in plant flavonoid biosynthesis, Science 299 (5605) (2003) 396–399.