Further Studies on the Mechanism of Interaction of Polyphenols, Oxygen, and Sulfite in Wine

Transcription

1 Further Studies on the Mechanism of Interaction of Polyphenols, Oxygen, and Sulfite in Wine John C. Danilewicz 1 * and Peter J. Wallbridge 2 Abstract: Further evidence is presented that Fe in association with Cu is an essential catalyst in oxidative processes of wine. The model polyphenol, 4-methylcatechol (4-MeC), is shown not to react with oxygen at a significant rate without addition of these metals in model wine. Similarly, sulfite autoxidation, which is a radical chain reaction, is demonstrated to dependent on these metals. However, the free radical scavenging activity of polyphenols prevents free radical chain propagation, so that sulfite cannot react with oxygen directly but with the hydrogen peroxide that is produced when polyphenols are oxidized. The Fe-catalyzed autoxidation of (+)-catechin is markedly accelerated by Cu, which it is proposed facilitates the redox cycling of Fe. The addition of these metals to wine accelerates the reaction with oxygen, while their removal with potassium ferrocyanide slows oxidation and in white wine can prevent it entirely. In the autoxidation of (+)-catechin in model wine, the rate of reaction of depends on (+)-catechin concentration and is faster when (-)-epicatechin is oxidized. The dependence of the rate of reaction of on the catechol is further evidence that is reacting with the hydrogen peroxide it produces. However, by measuring the changes in concentration of (+)-catechin and (-)-epicatechin when oxidized, it is evident that sulfite largely reduces the quinones back to the original catechol. With (+)-catechin, sulfite and benzenesulfinic acid markedly accelerate oxygen consumption, indicating that the rate of oxidation of polyphenols depends on nucleophiles, which by reacting with quinones displace the reversible catechol-iron redox interaction to drive the overall process forward. Key words: polyphenols, oxygen, sulfur dioxide, iron, copper Winemakers are concerned about iron (Fe) and copper (Cu) concentrations in wine as they wish to avoid instability due to the development of iron and copper casse. However, interest in these metals does not seem to extend beyond that to include their potential effects on oxidative processes. No doubt this limited interest is due to the conflicting evidence and opinions in the literature as to the extent of their involvement in these reactions. In studies of the coupled oxidation of polyphenols and ethanol to produce acetaldehyde, no metals were added to the model systems (Wildenradt and Singleton 1974). However, the oxidation of ethanol by hydrogen peroxide is dependent on metals, such as iron and copper, to generate hydroxyl radicals by way of the Fenton reaction (Danilewicz 2003, Waterhouse and Laurie 2006, du Toit et al. 2006). It is therefore possible that the initial oxidation of the polyphenols, which was conducted under forcing conditions (pure oxygen at 55 C over 35 days) to achieve an adequate reaction rate at ph 3.5, would also have been mediated by 1 44 Sandwich Road, Ash, Canterbury, Kent CT3 2AF, UK; and 2 Department of Wine, Plumpton College (University of Brighton), Ditchling Road, Plumpton, Nr Lewes, East Sussex BN7 3AE, UK. *Corresponding author ( jdanilewicz@btconnect.com; tel: ) Acknowledgment: The authors thank Matthew Ingram of the School of Pharmacy and Biomolecular Sciences, University of Brighton, for use of HPLC equipment and for NMR and mass spectra. Manuscript submitted Sept 2009, revised Nov 2009, accepted Nov 2009 Copyright 2010 by the American Society for Enology and Viticulture. All rights reserved. traces of these metals. In a later study with caffeic acid, the rate of oxidation increased with ph (range 4 to 8) such that a good correlation was found with phenolate ion concentration (Cilliers and Singleton 1989, 1991). Consequently, it was proposed that the reaction was initiated by single electron transfer from phenolate ions directly to oxygen, producing the semiquinone and superoxide, which then reacted further to yield a quinone and hydrogen peroxide. However, this proposal has to be set against the conclusion that the autoxidation of 6-hydroxydopamine, which contains the highly reducing benzene-1,2,4-triol system, depends on the presence of trace amounts of transition metals at ph 7 to 8 (Bandy and Davison 1987, Gee and Davison 1984). This is also the finding with ascorbic acid, which is oxidized by a similar mechanism to that of catechols. It no longer reacts with oxygen after careful removal of these metals with Chelex 100 at ph 7 (Buettner 1988, Miller and Aust 1989, Miller et al. 1990). The rate of oxidation of (+)-catechin in model wine has been shown to increase with iron concentration (Oszmianski et al. 1996). However, a decrease in (+)-catechin concentration was also observed without added iron, and since some products differed from those observed with added iron, an iron-independent degradation pathway was proposed. One of these products had spectral properties similar to those of the (+)-catechin oxidation product, dehydrodicatechin A, which indicated that the iron-free reaction involved oxidation with formation of the quinone. However, as noted above, trace metal impurities may still mediate this proposed iron-free process. Some studies have demonstrated the existence of a metal-independent pathway, 166

2 Interaction of Polyphenols, Oxygen, and Sulfite in Wine 167 but it remains to be established that it is sufficiently fast to contribute significantly to polyphenol oxidation in wine conditions (Mandal et al. 2005). In a study of the autoxidation of (+)-catechin, under forcing conditions (45 C for one month), Cu(II) at concentrations greater than 0.3 mg/l increased the rate of browning (Clark and Scollary 2002). Little (+)-catechin oxidation was observed, but yellow xanthylium cations were identified. These pigments are not derived from (+)-catechin quinone but by the condensation of glyoxylic acid, presumably formed by the degradation of tartaric acid, with (+)-catechin. Similar results had previously been obtained by autoxidizing (+)-catechin in the presence of high concentrations of Fe (Fulcrand et al. 1997). Paradoxically, therefore, it appears that in the presence of copper alone tartaric acid is degraded, presumably by autoxidation, while (+)-catechin, which is a much stronger potential reductant, is relatively stable. Emile Peynaud took the clear view that oxygen cannot combine directly with the reducing substances of wine and can only do so with the assistance of iron salts, traces of copper enhancing considerably its catalytic action (Peynaud 1984). It has been reported that removal of metals with potassium ferrocyanide markedly reduces the rate of reaction of wines with oxygen, although precise experimental details of this work, which was carried out in China, could not be obtained (Li et al. 2008). Studies where the autoxidation of 4-MeC was examined in model wine have shown that no significant reaction occurs without addition of these metals. Furthermore, the metals alone only exerted a relatively weak catalytic effect but marked synergism was observed when they were combined as proposed by Peynaud (Danilewicz 2007). It has long been held that sulfite protects wine by removing oxygen (Peynaud 1984, Ribéreau-Gayon et al. 2000). However, this reaction should be suppressed in wine, as sulfite oxidation is a free radical chain reaction, which would be prevented by radical scavengers such as polyphenols (Danilewicz 2007). The rate of sulfite oxidation in model wine is dependent on the concentration of a catechol (Danilewicz 2007). This finding is consistent with the proposal that sulfite reacts not with oxygen but with the hydrogen peroxide that is produced when polyphenols are oxidized (Boulton et al. 1996). Also to be explained is the observation that the rate of reaction of catechols with oxygen is markedly accelerated by sulfite, which reacts not only with hydrogen peroxide but also with quinones (Danilewicz et al. 2008). The interaction of oxygen, sulfite, and polyphenols in wine is clearly complex and the aim of this study is to provide a further insight into the reaction mechanisms that are involved and, in particular, the involvement of Fe and Cu. The European Commission is currently financing the Winesulfree project, which aims to find alternatives to sulfite in wine production (www. eurekanetwork.org; project number E!4506). In order to do so, a full understanding of how sulfite interacts with other wine constituents is essential. Materials and Methods Materials. Water (ISO 3696:1987 grade 3), copper(ii) sulfate pentahydrate, ferric chloride hexahydrate, sodium hydroxide, (+)-tartaric acid (BDH AnalaR grade), and ethanol (96% GRP grade) were obtained from VWR International (Lutherworth, UK); benzenesulfinic acid sodium salt, (+)-catechin, (-)-epicatechin, ferrous sulfate heptahydrate, 4-MeC, and potassium metabisulfite were from Sigma-Aldrich (Poole, Dorset, UK). The experiment, in which the rates of oxidation of (+)-catechin and (-)-epicatechin were compared, used potassium metabisulfite (Kadifit, Erbslöh Geisenheim AG, Geisenheim, Germany) and ethanol >99.5 % (Sigma-Aldrich). Kieselgel 60 F 254 plates (Merck, Darmstadt, Germany) were used for thin layer chromatography and compounds were visualized by UV light and by spraying with aqueous ferric chloride. 1 H NMR spectra were recorded using a Bruker 360 MHz spectrometer and mass measurement was determined in electrospray ionization (ESI) mode using a Daltonics microtof spectrometer (Bruker, Billerica, MA). Fe and Cu analyses were conducted by Corkwise Ltd. (Nutfield, UK) using a PerkinElmer 3110 atomic absorption spectrometer (PerkinElmer, Norwalk, CT). Mean of triplicate values are reported. Wines. Three wines were used: (1) Reichensteiner (Riesling x Sylvaner x Madeleine Angevine cross) Canterbury Choice, Dry Reserve 2008 (Barnsole Vineyard, Canterbury, UK), 11.5% (v/v) alcohol, ph 3.53, 6.0 g/l acid as tartaric acid, 44.5 mg/l free, 112 mg/l bound, 1.2 mg/l Fe, 0.05 mg/l Cu; (2) South African Colombard- Chardonnay 2008 (BrandPhoenix Ltd., Dorking, UK), 13.0% (v/v) alcohol, ph 3.47, 41.0 mg/l free, 0.8 mg/l Fe, 0.32 ± 0.02 mg/l Cu; and (3) First Cape, South African Merlot 2007 (BrandPhoenix), 13.5% (v/v) alcohol, ph 3.4, 28.8 mg/l free, 64.8 mg/l bound, 2.9 ± 0.1 mg/l Fe, and 0.26 ± 0.01 mg/l Cu. Preparation of model wine solutions. (+)-Tartaric acid (8.0 g) was dissolved in water (~800 ml) in a 1 L volumetric flask. Ethanol was added to give a 12% (v/v) final concentration. 10 ml of a freshly made up solution of Fe(II) sulfate heptahydrate (248.9 mg) in water (100 ml) and 1 ml of a solution of Cu(II) sulfate pentahydrate (117.8 mg) in water (100 ml) were added to give a final concentration of 5 mg/l Fe and 0.3 mg/l Cu. This was followed by solid potassium metabisulfite to give the required free concentration. The ph was increased to 3.60 with 2.5 N sodium hydroxide, also adding water progressively to the mark as the required ph was approached. The concentration of tartaric acid was selected so that it would simulate the action of all the wine acids and produce a final titratable acidity of ~4 g/l. When the effect of different concentrations of Fe and Cu were compared, it was found to be more convenient to add the metals to the final model wine, using from 36 to 250 μl/l of solutions of the above sulfates at an appropriate concentration. Reaction of model wines with O 2 in a closed system. A small amount of the above model wine solution (~10 ml) was withdrawn to wash in the required amount of 4-MeC,

3 168 Danilewicz and Wallbridge (+)-catechin, (-)-epicatechin, and benzenesulfinic acid sodium salt as required. The resulting solution was shaken in air until saturated with oxygen (i.e., when concentrations no longer increased). It was then sealed in 12 ~60 ml brown glass bottles with screwcaps fitted with a plastic cone liner, ensuring the exclusion of any air bubbles, and then stored in the dark. Maximum and minimum temperatures were noted during the course of each experiment. Initial free concentration was determined in duplicate and O 2 concentration in triplicate as described below. Three bottles were taken at each time point to measure O 2 concentrations in triplicate. Free was also measured when the O 2 : molar reaction ratio was determined. Mean values (±SD) were calculated and figures drawn using Excel software (Microsoft, Redmond, WA). (Where error bars denoting ± SD are not shown, bars were smaller than the data point symbol dimensions.) Reaction of wine with oxygen. concentrations were checked to ensure free levels were initially above 25 mg/l. Three to four liters of wine were poured into a 5 L container, which was shaken periodically and oxygen concentration measured until saturation was attained. One liter was used to fill 12 bottles as described above. The required amount of potassium ferrocyanide as an aqueous solution (100 to 250 μl) was added to a second liter, which was used to fill a second set of 12 bottles. Aqueous Fe(II) sulfate and copper(ii) sulfate (40 to 400 μl) were added to achieve the required concentration of Fe and Cu to additional liters of wine, which were used to fill additional sets of 12 bottles. These sets of bottles were placed side by side in the dark and oxygen concentration measured over the same time as described above. Oxygen measurement. An HI-9146 dissolved oxygen meter fitted with a HI Clarke type electrode was used for oxygen measurement (Hanna Instruments, Leighton Buzzard, UK). The manufacturer specifies a resolution and limit of detection (LOD) of the meter as 0.01 mg/l O 2. For measurements the bottle caps were quickly removed and a small stirrer bar inserted followed by the electrode, which displaced ~13 ml liquid. The electrode tip was lowered to ~5 mm of the briskly stirred magnetic bar. Readings stabilized within 30 sec and then remained stable for more than 5 min, indicating that, although the system was not sealed during measurement, no measurable amount of external oxygen reached the measurement area during that time. Measurement of. Free and bound concentration was measured by the aeration-oxidation method (Iland et al. 2004). When only free concentration was required in white wine and model wine, the modified Ripper procedure, using potassium iodate and starch-ki, was used (Ough and Amerine 1988). Reaction of in model wine at O 2 saturation. (+)-Catechin or (-)-epicatechin were added to model wine (500 ml; several hours were required for dissolution). Free concentration was adjusted to the required level and 150 ml aliquots of these solutions were placed in three 750 ml glass bottles. This afforded ~600 ml air headspace, which provided adequate oxygen exposure, as stirring was unnecessary even at the most rapid rates of oxidation. The bottles were closed with a plastic cap and maintained in the dark at ambient temperature, with maximum and minimum temperatures recorded during each series of experiments. The bottles were shaken and only brief ly exposed to light when analyzed. Free and bound concentrations, measured by the aeration-oxidation method, were followed over time in triplicate. The experiments, in which the effect of (+)-catechin concentration on rate of oxidation was examined, were conducted using 500 ml volumetric flasks instead of 750 ml bottles and free concentration was measured by the modified Ripper method. Mean time and free concentrations (± SD) were then calculated and figures drawn as described above. HPLC analysis. Analyses were carried out using a UV975 UV-Vis detector and PU980 pump (Jasco, Great Dunmow, UK). A C18 reversed-phase column was used (Fortis Technology, Neston, UK) and injection of samples was performed using a Rheodyne injection loop (IDEX Corp., Northbrook, IL). Isocratic elution was performed using a 1:1 mixture of HPLC grade water methanol at a flow rate of 1.0 ml/min. Solutions of the flavanol to be analyzed were injected in triplicate at concentrations of 60, 50, 41.7, 30, and 15 mg/l in 0.8% aqueous tartaric acid to obtain a calibration curve. Concentrations of 30 and 15 mg/l were used for the (+)-catechin quinone benzenesulfinic acid adduct. The detection wavelength was set at 280 nm and model wine solutions were diluted 1:10 before injection. Data were collected and analyzed using a Nelson 900 series interface and Totalchrom software (PerkinElmer). Elution times were the following: (+)-catechin, ± min; (-)-epicatechin, ±.007 min; and (+)-catechin quinone benzenesulfinic acid adduct, ±.017 min. (+)-Catechin quinone benzenesulfinic acid adduct. [(2R, 3S)-2-(2-benzenesulfonyl- 4,5-dihydroxyphenyl) -3,4-dihydro-1(2H)-benzopyran-3,5,7-triol]. Solutions from the experiment where (+)-catechin was oxidized in the presence of benzenesulfinic acid were saved after the determination of O 2 concentration and 310 ml placed in a liter bottle. After two months stored in the dark with periodic shaking to maintain O 2 saturation, the resulting orange-yellow solution was extracted with EtOAc, the extract washed with water and dried over MgSO 4. Evaporation gave an orange foam, which was chromatographed on silica, elution with a 17:3 mixture of CH 2 Cl 2 :MeOH giving a white solid (138 mg). This material was dissolved in the minimum amount of MeOH, and the required product crystallized out as a white solid on adding water (110 mg). TLC: R f 0.4 (CH 2 Cl 2 :MeOH, 17:3). 1 H NMR (CD 3 OD): similar to that previously described (Dangles et al. 2000). Mass: m/z = 453 (M + Na+). Results and Discussion The dependence of catechol oxidation on metal catalysis is demonstrated in two separate experiments (Figure 1). When a solution of 4-MeC in model wine containing sulfite

4 Interaction of Polyphenols, Oxygen, and Sulfite in Wine 169 was saturated with aerial oxygen and oxygen uptake followed over time in sealed bottles, no significant oxidation occured over five days without iron or copper addition. The mean of the average iron concentration in wines from different regions of the world is 5.5 mg/l (range 2.7 to 8.8 mg/l) and 0.2 mg/l (range 0.1 to 0.36 mg/l) for copper, taking out some extreme values (Ough and Amerine 1988). Addition of these metals, even at concentrations below these overall average values, resulted in prompt oxygen consumption (Figure 1). Taking 4-MeC as a suitable model compound that would be oxidized in a similar manner to wine polyphenols such as (+)-catechin or caffeic acid, it can be concluded that a metal-free oxidation mechanism cannot play a significant role in wine oxidation at reaction rates that are normally observed. The requirement for the catalytic intervention of metals in the oxidation of polyphenols is explained by the fact that oxygen is in effect a diradical, the two highest energy electrons occupying separate π * 2p antibonding orbitals with their spins aligned. It cannot accept paired electrons in a concerted manner as in reactions with polyphenols and with sulfite. Electrons can only be accepted singly and these are conveniently provided by metals or free radicals (Danilewicz 2003). For instance, it is proposed that the oxidation of sulfite is initiated by coordination to a transition metal, such as Fe(III), to form a sulfito-complex (Scheme 1). Fe(III) is hexa-coordinated, L representing ligands, which in wine would include tartaric acid. Electron transfer within the complex results in formation of a sulfite radical with release of Fe(II). The sulfite radical is now able to add oxygen to produce the highly oxidizing peroxomonosulfate radical (SO 5 ), which is capable of oxidizing sulfite to regenerate sulfite radicals and so continue the radical chain reaction (Danilewicz 2007). Similarly, catechols are able to coordinate to Fe(III), and again electron transfer within the unstable complex results in generation of intermediate semiquinones (Scheme 2) (Powell and Taylor 1982, Hynes and Coinceanainn 2001). In both cases the Fe(II) that is generated requires to be reoxidized to the ferric state to continue the catalytic cycle, which it is able to do by reacting directly with oxygen to produce hydroperoxyl radicals. These radicals may in turn be intercepted by polyphenols to produce hydrogen peroxide. The mechanism of oxidation may differ at neutral ph. Studies with 6-hydroxydopamine mentioned above indicate that oxygen reacts with the catechol-fe complex (Bandy and Davison 1987, Gee and Davison 1984). A similar mechanism is proposed for the Cu-catalyzed autoxidation of hydroquinone at ph 7 to 8 (Mandal et al. 2005). Reactive intermediate species generated may differ from those produced under acid conditions and so lead to different reaction products. The possible change in mechanism of catalysis may also contribute to the increased rate of oxidation of polyphenols with increasing ph. It has been shown that in a model wine containing 4-MeC the rate of oxidation is dependent on the concentration of the catechol (Danilewicz 2007). The same can be demonstrated for (+)-catechin (Figure 2). Had reacted only with oxygen, reaction rates would not have differed, since oxygen concentration was the same in these three experiments. The increase in reaction rate with increasing (+)-catechin concentration is consistent with Scheme 2 Iron-catalyzed autoxidation of a catechol to produce a quinone and hydrogen peroxide. Figure 1 Effect of Fe and Cu on rate of autoxidation of 4-MeC (1.0 g/l) in model wine containing sulfite. Scheme 1 Metal-catalyzed autoxidation of sulfite. Oxygen reacts with the sulfite radical, not with sulfite directly. Figure 2 Effect of (+)-catechin concentration on the rate of reaction of in model wine kept saturated with air: Fe 5.0 mg/l, Cu 0.3 mg/l.

5 170 Danilewicz and Wallbridge reacting with the hydrogen peroxide and quinone generated by oxidation of the flavanol. is slowly oxidized in model wine conditions, but it is proposed that this interaction should be prevented by catechols. Sulfite autoxidation is a metal-dependent radical chain process, which is blocked by radical scavengers, catechols being highly effective in this respect (Danilewicz 2007). The intermediate peroxomonosulfate radical is a powerful oxidant, the formal reduction potential (E 0 ) for the SO 5 /HSO 5 couple being 1.41 V (Das et al. 1999), and so catechol oxidation is highly favored to produce semiquinone and sulfate (Scheme 3). Removal of the central peroxomonosulfate radical would prevent radical chain propagation. The metal dependence of sulfite autoxidation in model wine has been previously demonstrated by measuring its rate of reaction with and without the addition of Fe and Cu (Danilewicz 2007). The same may be observed by following the rate of oxygen consumption (Figure 3). Sulfite does not react with oxygen at a significant rate unless the metals, which are added at concentrations normally present in wine, are also present. The reaction of oxygen with a catechol is greatly suppressed without addition of Fe and Cu (and then probably reliant on trace impurities) and sulfite also fails to react at a significant rate in this earlier experiment (Figure 1). As also previously noted, when sulfite is oxidized alone in model wine, bound is found to increase in keeping with a radical chain mechanism in which ethanol is oxidized to acetaldehyde by highly-oxidizing sulfate radicals (Danilewicz 2007). This was also observed in the above study (Figure 3). Free was 51.6 ± 0.3 mg/l and bound was 1.2 ± 0.4 mg/l by the aeration-aspiration method at the end of the experiment, when no metals had been added and no oxidation was observed (curve a), but at the higher metal concentration (curve c) final free was 33.1 ± 0.9 mg/l and bound was 6.7 ± 1.2 mg/l. is not acting as an antioxidant, but alone it enhances the oxidizing power of oxygen. Similarly, polyphenol oxidation results in the oxidation of ethanol (Wildenradt and Singleton 1974). It is only when and polyphenols are combined that oxygen is removed without producing its damaging effect. Polyphenols remove oxidizing -derived radicals and removes hydrogen peroxide (Scheme 4). The rate of reaction of when 4-MeC is oxidized in a model wine demonstrates that Cu alone is a relatively poor catalyst, but when combined with iron there is a pronounced synergistic effect (Danilewicz 2007). This synergistic interaction can also be observed in the autoxidation of (+)-catechin by measuring oxygen uptake directly (Figure 4). Here the effect of Cu is so weak that it may not have any significant catalytic activity alone and the limited oxygen uptake that is observed may be due to its interaction with trace impurities of Fe. The rate of reduction of Fe(III) by 4-MeC is fast (Danilewicz et al. 2008) and much faster than the oxidation of Fe(II) by oxygen (R. Elias and A. Waterhouse, personal communication, 2007). Therefore, it could be that the overall rate of catechol oxidation may be determined by the rate at which Fe(II) can be reoxidized to the ferric state to continue the catalytic cycle (Scheme 5). This possibility is supported by the finding that Cu markedly accelerates the oxidation of Fe(II) in model wine (Figure 5), Scheme 3 Interception of peroxomonosulfate radicals by polyphenols, so terminating radical chain propagation and preventing sulfite autoxidation. Scheme 4 Interaction of sulfite and polyphenols to remove oxygen while preventing the formation of damaging free radicals. Figure 3 Sulfite autoxidation in model wine: effect of Fe and Cu; free concentration 53.8 mg/l. Figure 4 Autoxidation of (+)-catechin (1.0 g/l): effect of Fe, Cu, and Cu + Fe; free 56.0 mg/l.

6 Interaction of Polyphenols, Oxygen, and Sulfite in Wine 171 where high concentrations of iron were used to accelerate the reaction but importantly the relative concentrations of the two metals were in the normal wine range. Cu(II) does not oxidize Fe(II) directly in model wine conditions (R. Elias and A. Waterhouse, personal communication, 2007) and so the independent redox cycling of copper, where Cu(I) is reoxidized to the cupric state by oxygen as previously proposed (Danilewicz 2007), is unlikely. The involvement of a binuclear Fe/Cu complex capable of reacting with oxygen more rapidly could be a possibility. The pronounced influence of Cu in the presence of iron is also seen in a low-metal white wine (Figure 6). This wine, which reacted with oxygen extremely slowly, had 1.2 mg/l Fe and 0.05 mg/l Cu. A linear extrapolation from the last two data points (Figure 6, curve a) indicates that it would take at least 45 days (t½) for half of the oxygen saturation concentration to be taken up. Adding just 0.1 mg/l Cu shortened this time to ~8.5 days (curve d) and to 4.7 days upon addition of a further 0.3 mg/l (curve e). A similar augmentation in reaction rate was obtained by raising Fe concentration to 5.0 mg/l and Cu concentration to 0.3 mg/l. On the other hand, reducing metal concentration by the addition of potassium ferrocyanide prevented oxidation. A precipitate, which was presumably ferri ferrocyanide (Prussian Blue), was observed after several days. It is understood that it has been the practice in Germany to treat wine with potassium ferrocyanide (blue fining) to slow maturation. A similar effect was observed on adding potassium ferrocyanide to a Colombard-Chardonnay wine (Figure 7). This wine had a high initial Cu concentration (0.32 mg/l) but low Fe concentration (0.8 mg/l). Increasing Fe concentration to 5.0 mg/l shortened t½ from 12 to 7 days. Potassium ferrocyanide appears to be less effective in removing metals from three red wines that were studied, as its inhibitory effect on autoxidation was less marked than in white wines. In the South African Merlot wine, Fe and Cu addition shortened t½ from 5.9 to 3.4 days, while treatment with potassium ferrocyanide extended it to 9.4 days (Figure 8). Clearly, Fe and Cu concentrations have a major Figure 6 Effect of Fe and Cu concentration and potassium ferrocyanide on the rate of autoxidation of a Reichensteiner wine; free 44.5 mg/l. Scheme 5 Autoxidation of a catechol. Redox cycling of Fe and participation of Cu. Figure 7 Effect of added Fe and potassium ferrocyanide on the rate of autoxidation of a Colombard-Chardonnay wine; free 41.0 mg/l. Figure 5 The effect of Cu on the rate of autoxidation of Fe(II) in model wine. Figure 8 Effect of added Fe and Cu and potassium ferrocyanide on the rate of autoxidation of a Merlot wine; free 28.8 mg/l.

7 172 Danilewicz and Wallbridge effect on the rate at which a wine will react with oxygen. However, other factors would be anticipated to contribute, such as ph, polyphenol concentration, and extent of previous exposure to oxygen, when more readily oxidized polyphenols would be consumed. Studies in model wine with 4-MeC have shown that the O 2 : molar reaction ratio is close to 1:2, indicating that reacts with the quinone and the hydrogen peroxide. Some of the quinone is reduced back to the catechol and the remainder undergoes 1,4-Michael addition to the sulfonic acid (Danilewicz et al. 2008). A similar ratio is evident with (+)-catechin (Figure 9). It was proposed that in wine other nucleophiles may compete with sulfite for quinones. Figure 9 Autoxidation of (+)-catechin in model wine: O 2 : molar reaction ratio; Fe 5.0 mg/l; Cu 0.15 mg/l. The rates of oxidation of (+)-catechin and (-)-epicatechin were compared in model wine maintained at oxygen saturation by measuring uptake, and (-)-epicatechin reacted faster than (+)-catechin (Figure 10). Here again it is evident that is not reacting solely with oxygen, as the experiments were conducted side by side and the rates of reaction would therefore have been the same, as O 2 and concentrations were identical. The amount of consumed after 27 days in the two experiments is shown (Table 1), from which is calculated the amount of each flavanol that was oxidized, assuming : O 2 : flavanol molar reaction ratios of 2:1:1. The amount of quinone that was reduced was calculated from the final (+)-catechin and (-)-epicatechin concentrations. The results show that largely protects these compounds from oxidation by regenerating them from the quinone, although it appears to be less effective in doing so with (-)-epicatechin, presumably because of the greater instability of the quinone (Richard-Forget et al. 1992). In contrast to 4-MeC, no product corresponding to the sulfonic acid, produced by addition of sulfite to the quinone, was observed by HPLC. This difference may be due to increased steric crowding at the B-ring 5 position (Scheme 6). It has previously been proposed that quinones, generated by polyphenoloxidase, are instantly reduced back to catechols on quenching the reaction with sulfite (Cheynier et al. 1989a). It has also been shown that sulfite does not function as a reducing agent in the Folin-Ciocalteu and metmyoglobin assays used to determine antioxidant activity. However, it potentiates values obtained for (+)-catechin, which led to the proposal that it recycled the quinone back to the catechol (Saucier and Waterhouse 1999). The reaction of O 2 and with little apparent change in polyphenol concentration would lead to the mistaken conclusion that the two simply reacted together. It has been proposed that the mechanism of reduction involved initial 1,2-addition to quinones, as occurs with other carbonyl compounds, followed by elimination of sulfate (Danilewicz et al. 2008). In the Figure 10 reaction on autoxidation of (+)-catechin and (-)-epicatechin, each at 500 mg/l in model wine kept saturated with air: Fe 5.0 mg/l; Cu 0.3 mg/l. Scheme 6 1,4-Michael addition of benzenesulfinic acid to (+)-catechin quinone. Table 1 Free consumed and (+)-catechin (cat) and (-)-epicatechin (epi) remaining after autoxidation for 27 days. From these values is calculated the amount of each quinone that was reduced to regenerate the catechol (Figure 10). values on day 27 are the mean of values obtained on day 26 and 28. (+)-Catechin is taken as the monohydrate, and (-)-epicatechin as anhydrous. Initial concn consumed in 27 days (mg/l) a Theoretical cat/epi oxidized (mg/l) Final cat/epi concn after 27 days (mg/l) cat/epi lost (mg/l) Quinone reduced (%) Catechin (500 mg/l) 55.5 ± ± ± Epicatechin (500 mg/l) 97.5 ± ± ± a By aeration-oxidation.

8 Interaction of Polyphenols, Oxygen, and Sulfite in Wine 173 above experiments, where (+)-catechin and (-)-epicatechin were oxidized in the presence of sulfite, bound did not increase from a low baseline value, indicating that the removed all the hydrogen peroxide that was produced. None escaped to oxidize ethanol to produce acetaldehyde. It could be argued that the quinones were not reduced in the above experiments and that the catechols were simply not oxidized in the first place, the having reacted with O 2. The quinone derived from (+)-catechin may be trapped with benzenesulfinic acid, which forms an adduct (Scheme 6) (Cheynier et al. 1989b). Consequently, when the above experiment with 500 mg/l (+)-catechin was repeated without sulfite but in the presence of 6.2 equivalents of benzenesulfinic acid, 93.5 ± 10.5 mg/l of (+)-catechin was lost after 27 days, which should yield ± 14.6 mg/l of the adduct. The amount actually found by HPLC was ± 1.9 mg/l, and so the quinone must be formed in these model wine conditions. The adduct was isolated and characterized as described under Materials and Methods. When (+)-catechin and (-)-epicatechin at 500 mg/l were oxidized at ambient temperature in model wine at oxygen saturation without protection, the solutions became deep orange-yellow. However, only 39.7 ± 7.1 mg/l (+)-catechin and 96.3 ±3.1 mg/l (-)-epicatechin were lost after 4 months. Again, (-)-epicatechin is less stable, but it is apparent that the rates of oxidation are much slower than in the presence of (Table 1). markedly accelerates the oxidation of 4-MeC (Danilewicz et al. 2008), which is also found with (+)-catechin (Figure 11). In the absence of little oxidation took place after two weeks, as all the oxygen uptake can be accounted for by the oxidation of Fe(II) to the ferric state. As with 4-MeC, the rate of oxidation of (+)-catechin is also accelerated by benzenesulfinic acid. It should be noted that benzenesulfinic acid is not oxidized under these model wine conditions and forms adducts cleanly with quinones. It was therefore proposed that the metal-catalyzed oxidation of catechols is reversible and very slow unless a nucleophile is present to trap the quinone and drive the reaction forward (Scheme 7). The same mechanism has been proposed for the marked acceleration produced by glutathione on the rate of trans-caftaric acid oxidation by polyphenoloxidase (Cheynier and Van Hulst 1988). The semiquinone in Scheme 7 may disproportionate to generate the quinone or possibly add oxygen to give a peroxyl radical that would eliminate a hydroperoxyl radical to produce the quinone (Mandal et al. 2005). For the reverse reaction 1,4-benzoquinones are known to be reduced back to hydroquinones by Fe(II) (Baxendale et al. 1951). To summarize, in the absence of sulfite the oxidative process may be represented as in Scheme 8 (upper section). Oxygen does not react with the catechol directly but with Fe(II) (with the assistance of Cu), returning it to the ferric state to continue its catalytic function, while oxygen is reduced to hydrogen peroxide. It is proposed that the overall rate of oxygen consumption is controlled by the rate at which iron redox cycles and the quinone is removed by reaction with water or other wine constituents. (-)-Epicatechin is oxidized faster than (+)-catechin, and it is further proposed that this is due to the greater reactivity of the quinone. Oxygen does not react with sulfite directly (Scheme 8, lower section), but the net result is the same. Two sulfites react as one oxygen molecule is consumed, with the intermediate redox cycling of the metals and the catechol. In the case of (+)-catechin the process is very efficient, as ~96% of the quinone was reduced in the model system studied. However, the process is less efficient for (-)-epicatechin, as under the same conditions ~20% was lost; it was also previously found that ~38% of the quinone derived from 4-MeC added sulfite to produce the sulfonic acid (Danilewicz et al. 2008). Consequently, sulfite may not prevent the loss of different polyphenols equally in wine and this is currently being investigated. Scheme 7 Displacement of the catechol-iron reversible redox interaction by nucleophiles. Figure 11 Effect of free and benzenesulfinic acid, added as the sodium salt (BSAS), on rate of oxidation of (+)-catechin (Cat): Fe 5.0 mg/l, Cu 0.3 mg/l. Scheme 8 Proposed interaction of oxygen with a polyphenol in the absence and presence of sulfite in wine.

9 174 Danilewicz and Wallbridge Importantly, the differences in the apparent rate of oxidation of polyphenols may not be due to differences in their reducing ability, that is, in their reduction potentials, but rather due to differences in the rate at which the quinones react with sulfite or other nucleophile to displace the reversible catechol-iron redox interaction and so allow the oxidative process to advance. Quinones become less stable with increasing ph because of their increased vulnerability to nucleophilic attack (Proudfoot and Ritchie 1983). It follows from the above conclusion that the increase in rate of polyphenol oxidation with ph may be in large measure due to the increased reactivity of the quinone. The reasons why very little Cu catalyzed (+)-catechin oxidation was observed over a month at 45 C now become apparent (Clark and Scollary 2002). Copper may be involved in the degradation of tartaric acid to produce glyoxylic acid, which condenses with (+)-catechin to produce yellow pigments, but alone it is a very weak polyphenol oxidation catalyst; also no nucleophile was present to allow the oxidation to proceed. The degradation of tartaric acid was therefore possible in the presence of a potentially more powerful reductant, which could only react with oxygen very slowly under these conditions. Sulfite is very effective in preventing the oxidative yellowing of (+)-catechin in model wine, as it reacts both with hydrogen peroxide and the quinone as well as forming an adduct with any glyoxylic acid that might form. Benzenesulfinic acid did not prevent this yellowing reaction, as it presumably only reacts with the quinone, which is not involved in forming xanthylium ions. Conclusions This work has provided further evidence both in model wine and wine itself that iron is an essential catalyst in oxidative processes and that its action is greatly influenced by copper. The concentration of these metals will therefore be a major determinant as to how rapidly a wine will react with oxygen. It is further argued that sulfite does not react with oxygen directly but with hydrogen peroxide. However, its mechanism of action is far more complex. The superficial view that it functions as an antioxidant and that it may be replaced by another is a gross oversimplification. It remains to be determined how sulfite interacts with other polyphenols, but in terms of (+)-catechin and (-)-epicatechin, sulfite allows these substances to react with oxygen much more rapidly and so maintain an anaerobic wine environment, while largely regenerating them. It removes hydrogen peroxide, the primary oxygen reduction product, very efficiently. Hydrogen peroxide is highly damaging by way of the Fenton reaction and any antioxidant that reacts with oxygen to produce it cannot replace sulfite fully. Importantly, sulfite reacts with aldehydes and ketones to give nonvolatile addition compounds, preventing their reaction and also masking oxidation aromas. In addition, sulfite has an antimicrobial action. Finding a simple, safe alternative to sulfite that possesses all these properties is therefore a demanding objective and its removal from winemaking will probably require major changes in winemaking techniques. Literature Cited Bandy, B., and A. J. Davison Interactions between metals, ligands, and oxygen in the autoxidation of 6-hydroxydopamine: Mechanisms by which metal chelation enhances inhibition by superoxide dismutase. Arch. Biochem. Biophys. 259: Baxendale, J.H., H.R. Hardy, and L.H. Sutcliffe Kinetics and equilibria in the system ferrous ion + ferric ion + hydroquinone + quinone. Trans. Faraday Soc. 47: Boulton, R.B., V.L. Singleton, L.F. Bisson, and R.E. Kunkee The role of sulfur dioxide in wine, pp In Principles and Practices of Winemaking. Chapman & Hall, New York. Buettner, G.R In the absence of catalytic metals ascorbate does not autoxidize at ph 7: Ascorbate as a test for catalytic metals. J. Biochem. Biophys. Methods 16: Cheynier, V., and M.W.J. Van Hulst Oxidation of trans-caftaric acid and 2-S-glutathionylcaftaric acid in model solutions. J. Agric. Food Chem. 36: Cheynier, V., N. Basire, and J. Rigaud. 1989a. Mechanism of transcaffeoyltartaric acid and catechin oxidation in model solutions containing grape polyphenoloxidase. J. Agric. Food Chem. 37: Cheynier, V., J. Rigaud, and M. Moutounet. 1989b. High-performance liquid chromatographic determination of the free o-quinones of trans-caffeoyltartaric acid, 2-S-glutathionylcaffeoyl tartaric acid and catechin in grape must. J. Chromatogr. 472: Cilliers, J.J.L., and V.L. Singleton Nonenzymic autoxidative phenolic browning reactions in a caffeic acid model system. J. Agric. Food Chem. 37: Cilliers, J.J.L., and V.L. Singleton Characterization of the products of nonenzymic autoxidative phenolic reactions in a caffeic acid model system. J. Agric. Food Chem. 39: Clark, A.C., and G.R. Scollary Copper(II)-mediated oxidation of (+)-catechin in a model white wine system. Aust. J. Grape Wine Res. 8: Dangles, O., G. Fargeix, and C. Dufour Antioxidant properties of anthocyanins and tannins: A mechanistic investigation with catechin and 3,4,7-trihydroxyflavylium ion. J. Chem. Soc., Perkin Trans. 2: Danilewicz, J.C Review of reaction mechanisms of oxygen and proposed intermediate reduction products in wine: Central role of iron and copper. Am. J. Enol. Vitic. 54: Danilewicz, J.C Interaction of sulfur dioxide, polyphenols, and oxygen in a wine-model system: Central role of iron and copper. Am. J. Enol. Vitic. 58: Danilewicz, J.C., J.T. Seccombe, and J. Whelan Mechanism of interaction of polyphenols, oxygen, and sulfur dioxide in model wine and wine. Am. J. Enol. Vitic. 59: Das, T.N., R.E. Huie, and P. Neta Reduction potentials of SO 3 -, SO 5 - and S 4 O 6 3- radicals in aqueous solution. J. Phys. Chem. 103: du Toit, W.J., J. Marais, I.S. Pretorius, and M. du Toit Oxygen in musts and wine: A review. S. Afr. J. Enol. Vitic. 27: Fulcrand, H., V. Cheynier, J. Oszmianski, and M. Moutounet An oxidized tartaric acid residue as a new bridge potentially competing with acetaldehyde in flavan-3-ol condensation. Phytochemistry 46: Gee, P., and A. J. Davison Hydroxydopamine does not reduce molecular oxygen directly, but requires a coreductant. Archives Biochem. Biophys. 231: Hynes, M. J., and M. Ó. Coinceanainn The kinetics and mechanism of the reaction of iron(iii) with gallic acid, gallic acid methyl ester and catechin. J. Inorg. Biochem. 85:

10 Interaction of Polyphenols, Oxygen, and Sulfite in Wine 175 Iland P., N. Bruer, G. Edwards, S. Weeks, and E. Wilkes Chemical analysis of grapes and wines: Techniques and concepts. Patrick Iland, Campbelltown, Australia. Li, H., A. Guo, and H. Wang Mechanism of oxidative browning of wine. Food Chem. 108:1-13. Mandal, S., N.H. Kazmi, and L.M. Sayre Ligand dependence in the copper-catalyzed oxidation of hydroquinones. Arch. Biochem. Biophys. 435: Miller, D.M., and S.D. Aust Studies of ascorbate-dependent, iron-catalyzed lipid peroxidation. Arch. Biochem. Biophys. 271: Miller, D.M., G.R. Buettner, and S.D. Aust Transition metals as catalysts of autoxidation reactions. Free Radical Biol. Med. 8: Oszmianski, J., V. Cheynier, and M. Moutounet Iron-catalyzed oxidation of (+)-catechin in model systems. J. Agric. Food Chem. 44: Ough, C.S., and M.A. Amerine Methods for Analysis of Musts and Wines. 2d ed. Wiley & Sons, New York. Peynaud, E Knowing and Making Wine. Wiley & Sons, New York. Powell, H.K.J., and M.C. Taylor Interactions of iron(ii) and iron(iii) with gallic acid and its homologues: A potentiometric and spectrophotometric study. Aust. J. Chem. 35: Proudfoot, G.M., and I.M. Ritchie A cyclic voltammetric study of some 4-substituted benzene-1,2-diols. Aust. J Chem. 36: Ribéreau-Gayon, P., Y. Glories, A. Maujean, and D. Dubourdieu Handbook of Enology. Vol. 2. The Chemistry of Wine Stabilization and Treatments. Wiley & Sons, New York. Richard-Forget, F.C., M.A. Rouet-Mayer, P.M. Goupy, J. Philippon, and J.J. Nicolas Oxidation of chlorogenic acid, catechins, and 4-methylcatechol in model solutions by apple polyphenol oxidase. J. Agric. Food Chem. 40: Saucier, C.T., and A.L. Waterhouse Synergetic activity of catechin and other antioxidants. J. Agric. Food Chem. 47: Waterhouse, A.L., and V.F. Laurie Oxidation of wine phenolics: A critical evaluation and hypotheses. Am. J. Enol. Vitic. 57: Wildenradt, H.L., and V.L. Singleton The production of aldehydes as a result of oxidation of polyphenolic compounds and its relation to wine aging. Am. J. Enol. Vitic. 25: