Analysis of selected carbonyl oxidation products in wine by liquid chromatography with diode array detection

Transcription

1 analytica chimica acta 626 (2008) available at journal homepage: Analysis of selected carbonyl oxidation products in wine by liquid chromatography with diode array detection Ryan J. Elias a,1, V. Felipe Laurie b,, Susan E. Ebeler a, Jon W. Wong c,2, Andrew L. Waterhouse a a Department of Viticulture and Enology, University of California at Davis, CA 95616, USA b Centro Tecnológico de la Vid y el Vino, Facultad de Ciencias Agrarias, Universidad de Talca, Casilla 747-Talca, Chile c Alcohol and Tobacco Tax and Trade Bureau (formerly Bureau of Alcohol, Tobacco, and Firearms), Beverage Alcohol Laboratory, 6000 Ammendale Road, Beltsville, MD 20705, USA article info abstract Article history: Received 8 April 2008 Received in revised form 28 July 2008 Accepted 29 July 2008 Published on line 7 August 2008 Keywords: Carbonyl compounds Aldehydes 2,4-Dinitrophenylhydrazine Wine Oxidation Sulfite A high performance liquid chromatography (HPLC) method for the detection and quantitation of acetaldehyde, glyceraldehyde, pyruvic acid, 2-ketoglutaric acid, and formaldehyde in wine, based on the formation of the 2,4-dinitrophenylhydrazones, is presented. These carbonyl compounds often result from the chemical oxidation of major wine components, and are known to affect flavor and color stability. Their analysis in wine is complicated due to their instability and their tendency to react reversibly with bisulfite to form hydroxysulfonates. Published methods that break down the sulfonates for the quantitation of total carbonyls in wine involve alkaline hydrolysis of sulfite-bound carbonyls, but we show, for the first time, that this alkaline treatment step significantly increases the concentration of carbonyls during analysis. A solution based on oxygen exclusion is described. The technique offers good specificity, reproducibility (%RSD ), and limits of detection ( gl 1 ). The method was successfully used to monitor concentration changes of these compounds in both white and red wines Elsevier B.V. All rights reserved. 1. Introduction Many carbonyls are formed in wine as normal byproducts of microbial fermentation and chemical oxidation, or from oak-barrels during winemaking and aging [1 3]. Their concentration can vary significantly from wine to wine, mainly as a result of variations in winemaking and storage conditions [4 7]. In sweet wines, for instance, the content of carbonyl compounds surpasses that of dry table wines, probably due to sugar oxidation [8]. So far, well-characterized concentrations of important carbonyl substances in wine include acetaldehyde (3 494 mg L 1 ), pyruvic acid ( mg L 1 ), 3- hydroxybutanone or acetoin ( mg L 1 ), and diacetyl ( mg L 1 ) [2,9]. The contribution of these carbonyls to the chemistry of wine is a complex subject in which their effects on flavor Corresponding author. Tel.: ; fax: address: flaurie@utalca.cl (V.F. Laurie). 1 Present address: Department of Food Science, The Pennsylvania State University, University Park, PA 16802, USA. 2 Present address: Food and Drug Administration, Center for Food Safety and Applied Nutrition, Office of Regulatory Science, College Park, MD 20740, USA /$ see front matter 2008 Elsevier B.V. All rights reserved. doi: /j.aca

2 analytica chimica acta 626 (2008) [1] and color [10,11] are the most noticeable. For example, acetaldehyde (ethanal) is an important flavor compound associated with the oxidation of wine, and contributes aromas described as overripe apples, nutty, or sherry-like [1,12]. Carbonyls are also known to take part in wine aging reactions, with potential benefits to the color of red wines. Studies on the formation of novel, potentially stable, colored structures arising from the linkages between flavonoids and acetaldehyde [10], glyoxylic acid [13], and pyruvic acid [14] have been published. Likewise, possible effects on color due to the presence of furfural derivatives [15] and glycerol oxidation products (glyceraldehyde and dihydroxyacetone) have also been suggested [16,17]. Some of the available techniques for analyzing carbonyl compounds include non-specific methods such as distillation or reaction with bisulfite, low sensitivity methods based on colorimetric procedures, and non-quantitative thin layer and paper chromatography methods [1,2,18,19]. Enzymatic redox reaction methods are good alternatives for the analysis of single compounds such as acetaldehyde, pyruvate or ketoglutarate; however, these assays can be time-consuming [8,20]. One of the more reliable alternatives for the analysis of these compounds is gas chromatography (GC), although some problems associated with volatilization and even decomposition of the analytes in the GC-injector port have been reported [2,21,22]. In wine, carbonyl compounds have been analyzed with GC MS after o-(2,3,4,5,6-pentafluorobenzyl)- hydroxylamine [5,8] and cysteamine derivatization [23]. Recently, a method to analyze acetaldehyde concentration by headspace solid-phase microextraction (SPME) with on-fiber derivatization and GC FID has been developed [24]. Alternatively, liquid chromatography (LC) methods, more suitable for non-volatile compounds, with equivalent accuracy, sensitivity and specificity have been developed [22]. The use of LC allows for the detection of more polar and higher molecular weight compounds than those detected by GC [22]. Most of the LC methods available for other food matrices are based on the detection of stable carbonyl derivatives. Hydrazines, such as 2,4-dinitrophenylhydrazine (DNPH), selectively react with aldehydes and ketones to form stable hydrazones [21,25,26]. The aromatic hydrazine reacts with carbonyls under acidic conditions, forming insoluble hydrazone derivatives [27]. The rate of this reaction must be sufficiently fast to allow for the quantitative determination of the generated products [21]. When working with liquid chromatography, it is more convenient to keep the hydrazone derivatives in solution by using an acidified organic phase such as acetonitrile and perchloric acid [20]. Separation is typically performed on a C18 chromatography column with a variety of gradients using water and acetonitrile [27]. During the preparation of DNPH standards, contamination with trace acids can cause anomalies in their analysis. A bicarbonate wash of the crystallized derivatives has been proposed to solve this problem. Hydrazones are good chromophores for UV vis or fluorescence spectroscopy, and good levels of quantification are often observed. Typically, detection of this type of derivative is carried out at a wavelength of 360 nm, but due to the spectral variations between different analytes, a diode array detector or a mass spectrometer may provide more comprehensive information [20,26 28]. Wine carbonyls can react with alcohols to produce acetals and with nitrogen and sulfur containing compounds, including DNA, to produce various metabolites, some of which are toxic [2,23,29,30]. An added complication in wine arises from the fact that naturally produced and added sulfur dioxide (SO 2 ), specifically the bisulfite ion (HSO 3 ), reacts rapidly but reversibly with carbonyl compounds to form hydroxysulfonates (sulfite-bound carbonyls), thus decreasing the amount of free sulfite and available carbonyls [18,31]. Some of the sulfite-binding compounds in wines are acetaldehyde, pyruvic acid, 2-ketoglutaric acid, galacturonic acid, 2,5-diketogluconic acid, l-xylosone, d-threo-2,5-hexodiulose, 2-ketogluconic acid, glucuronic and glyoxylic acids [32,33]. A quantitative method for the analysis of total (i.e., free and sulfite-bound) carbonyl compounds in wine, as for the analysis of wine sulfites, must include a sulfonate hydrolysis step. The dissociation of sulfite-bound carbonyls is most often carried out under strong alkaline conditions [2,20]; however, such conditions may promote the oxidation of phenolic compounds, thus resulting in the formation of further carbonyl compounds [34,35]. Recent developments on the general mechanism of wine oxidation have established the possibility that after phenolic oxidation, hydrogen peroxide is formed and via a Fenton reaction, this can generate a broad pool of carbonyl-oxidation products by the hydroxyl radical ( OH) oxidation of hydroxy-substituted substances [3,16]. The rate of phenolic oxidation in wine increases as a function of increasing ph [3,36], however, the mechanism by which this occurs in wine is not clear. One of the more traditionally accepted mechanisms in the wine chemistry literature involves the direct reaction between dioxygen and the anionic (phenolate) form of phenol, which dominate as the ph surpasses the phenols pk a s, typically 9 10 [37]. The rate of oxygen consumption in wine is markedly faster at high ph values [38], and it has been shown that caffeic acid in model wine solution is almost completely oxidized within 30 min at ph 8 [36,39]. However, the two-electron reaction between oxygen (triplet ground state) and phenolic compounds (singlet ground state) is forbidden by the Pauli Exclusion Principle without spin-inversion and one-electron processes [3,40,41]. Transition metals, such as iron and copper, are typically encountered at catalytic concentrations in wine [2,3], and may be important in the oxidation of wine phenolics at high ph. For example, it has been demonstrated that the autoxidation of the catechol 6-hydroxydopamine is prevented at ph 8 in the presence of the metal chelator diethylenetriaminepentaacetate (DTPA), catalase, and superoxide dismutase, thus supporting the role of metal catalysis and reactive oxygen species intermediates in this reaction [41]. Further studies are needed to elucidate this very complex mechanism in wine, and such questions lie outside the scope of this study. As suggested, the analysis of carbonyls in wine is complicated due to their low concentration, volatility, and their ability to form complexes with other wine components. Consequently, the aim of this study was to develop an accurate and sensitive method for the determination of selected carbonyl compounds in wine by direct DNPH derivatization and HPLC- DAD. In particular, in this method, special measures were taken to eliminate the production of interfering carbonyls during the alkaline hydrolysis of sulfite-bound analytes.

3 106 analytica chimica acta 626 (2008) Experimental 2.1. Reagents, model solutions, and wine samples DNPH (30% water) was obtained from Alfa Aesar (Ward Hill, MA, USA) and was purified by recrystallization from acetonitrile. DL-glyceraldehyde (2,3-dihydroxypropanal), 2- ketoglutaric acid (2-oxopentanedioic acid), formaldehyde, pyruvic acid, and acetaldehyde were purchased from Sigma Aldrich (St. Louis, MO, USA). The corresponding DNPH derivatives of the standards were prepared as described previously and recrystallized from acetonitrile [26]. All solutions were prepared with Milli-Q water from Millipore (Bedford, MA, USA), and all chemical and solvents used were of reagent or HPLC grade and were obtained from Fisher (Fairlawn, NJ, USA) or Sigma Aldrich. Single and mixed carbonyl standard solutions were prepared using 12% ethanol and adjusted to ph 3.6 with hydrochloric acid (1N). Model wine solutions consisted of 12% ethanol in (+)-tartaric acid solution (5 g L 1 ), adjusted to ph 3.6 with sodium hydroxide (1N), and including either 4- methylcatechol (1.5 g L 1 ) or catechin (3.65 g L 1 ), two phenolic compounds widely used in modeling the polyphenol fraction of wines. White and red wine samples used in the study were either donated to, or produced by the Department of Viticulture and Enology at the University of California, Davis. Variable concentrations of sulfur dioxide, added as potassium metabisulfite, were used for method development and validation Procedure for DNPH solution preparation DNPH solutions were freshly prepared by dissolving 200 mg recrystallized DNPH reagent in 100 ml acetonitrile, then acidifying with 4 ml perchloric acid (70%). Contamination from airborne aldehydes (e.g., formaldehyde) was observed when DNPH solutions were kept at room temperature. When cold storage (4 C) and a gaseous nitrogen blanket were used, the integrity of these preparations was extended to at least one week (checked by LC DAD comparison with freshly prepared DNPH solutions). Fig. 1 Formation of acetaldehyde in model wine solution containing 4-methylcatechol (4-MeC) or catechin following the addition of NaOH or water (control). free carbonyls and bisulfite. DNPH derivatization took place at room temperature for a total of 3 h Chromatography, carbonyl compounds detection, and sample handling An Agilent Technologies (Palo Alto, CA, USA) 1100 series reverse phase HPLC with a photodiode array UV visible detector was used for separation and detection of the analytes. The identification of the observed carbonyls was based on their retention time compared with those of the carbonyl standards tested at 365 nm as well as their spectral characteristics. Data analysis and peak integration was carried out using the Agilent Chemstation (A 09.03) software package Sample preparation and procedure for carbonyl derivatization Hydrolysis of the -hydroxysulfonates in model solutions and wines was carried out under alkaline conditions using NaOH (1N) that had been deoxygenated with nitrogen gas (99% N 2(g) ; Airgas, Woodland, CA, USA) for at least 30 min. Sample aliquots (100 L) were dispensed to glass culture tubes (15 mm 85 mm, Fisher) under a N 2 blanket and allowed to react with 100 L of the N 2 -sparged NaOH solution for 10 min (final ph 13). Great care was taken not to strip volatile carbonyl analytes (e.g., acetaldehyde) during hydrolysis, so the N 2(g) flow rate into the hydrolysis tube was kept low, and the gas was not passed through the solution. Following hydrolysis, the samples were acidified with 40 L of 25% sulfuric acid solution (final ph 1), and 240 L of the DNPH reagent was added without delay in order to prevent the reaction of Fig. 2 Effect of the presence of 4-methylcatechol (4-MeC), SO 2 and air on the formation of acetaldehyde in model wine. Model wine samples excluding and including 4-MeC were treated as follows: (A) NaOH treatment under air; (B) NaOH treatment under air and in the presence of 120 mg L 1 SO 2 ; (C) NaOH treatment under anoxic conditions in the presence of 120 mg L 1 SO 2.

4 analytica chimica acta 626 (2008) Table 1 Analysis of carbonyl compounds (n = 3, technical replicates) in Cabernet Sauvignon (Talus Collection, 2002; Lodi, CA) under aerial and anoxic alkaline hydrolysis conditions Compound Hydrolysis under air Hydrolysis under nitrogen No added SO mg L 1 SO 2 No added SO mg L 1 SO 2 Glyceraldehyde 72 ± ± ± ± Ketoglutaric acid 172 ± ± ± ± 1.9 Pyruvic acid 41 ± ± ± ± 0.4 Formaldehyde 3.1 ± ± ± ± 0.1 Acetaldehyde 38 ± ± ± ± 0.1 Carbonyl-DNPH derivatives were separated using a C18 LiChrospher column (4 mm 250 mm, 5 m particle size; Merck, Whitehouse Station, NJ, USA) protected with a guard column of the same material. The mobile phase consisted of a binary gradient of (A) 1% acetonitrile in acidic phosphate buffer (25 mm, ph 2.2) and (B) 100% acetonitrile as follows: 0 min, 25% B; min, 50%; min, 75% B; min, 25%; and min, 5% B. Sodium azide (0.2 g L 1 ) was added to the aqueous mobile phase in order to extend its shelf-life and did not affect chromatographic separation or analysis. The sample injection volume was 25 L and the flow rate was 0.4 ml min 1. UV visible spectra were recorded from 200 to 600 nm with automatic detection traces at 365 nm. To avoid hydrazone crystallization during chromatography, samples were diluted 1:1 in mobile phase A following DNPH derivatization, and filtered through 0.45 m polytetrafluoroethylene (PTFE), 13 mm, syringe tip filters (ArcodiscTM, Ann Arbor, MI, USA) into 2 ml HPLC vials and sealed with PTFE crimp caps. Completely derivatized wine samples were stored at 4 C before injection into the HPLC Method validation To check the linearity of the method, an external calibration curve for the hydrazone standards in model wine solution (12% ethanol, ph 3.6), in the range of mg L 1,was run and recorded in triplicate. This concentration range was based on the carbonyl content of wine samples, allowing the determination of all the analytes in a single chromatographic run. The regression curve was calculated as y = ax + b, where y corresponded to the peak areas of the analyte and x was the standard concentration injected. The calibration curve was obtained using the linear least squares regression procedure built into the Microsoft Excel 2000 software package. The precision of the method was obtained by calculating the relative standard deviation (%RSD) of the analytes concentration for repeated samples, while recovery studies were done by spiking pure carbonyl standards (8 mg L 1 ) into the model wine system. Accuracy was assessed over model wine samples with spiked pure standards, as no other method for carbonyl Fig. 3 HPLC/DAD chromatograms of (A) unsulfited red wine (Petite Syrah, Frey Vineyards, 2004; Redwood Valley, CA), (B) aged white wine (Chardonnay, University of California, 1999; Davis, CA), and (C) young red wine (Cabernet Sauvignon, Woodbridge, 2007; Lodi, CA). The observed peaks were glyceraldehyde (1), 2-ketoglutaric acid (2), pyruvic acid (3), formaldehyde (4), and acetaldehyde (5). Peak I is the excess, unreacted DNPH reagent.

5 108 analytica chimica acta 626 (2008) Table 2 Linearity and limits of detection of carbonyl analytes Compound Linear equation R 2 Limit of detection ( gl 1 ) Linear range (mg L 1 ) Glyceraldehyde y = 276.5x Ketoglutaric acid y = 171.7x Pyruvic acid y = 382.5x Formaldehyde y = 977.3x Acetaldehyde y = 619.1x compounds in wine addresses the interferences caused by the hydrolysis of sulfonates. Statistical significance was determined using Student s t-test. 3. Results and discussion 3.1. Hydrolysis of sulfite-bound carbonyls As previously indicated, the alkaline hydrolysis of sulfitebound carbonyl compounds can result in the oxidation of wine phenolics, inadvertently producing additional carbonyls in the process. Based on the findings of others [26,40,41], it is possible that the production of carbonyls at neutral or high ph is due to radical oxygen species. It has been previously demonstrated that hydrogen peroxide is formed as a result of catechol oxidation in wine, and that the rate of oxidation and thus hydrogen peroxide formation rates increased with rising ph [34,42]. Given the prevalence of trace quantities of transition metals in wine, the reduction of hydrogen peroxide to the highly reactive hydroxyl radical is predicted (i.e., the Fenton reaction), which is capable of oxidizing wine constituents, often to aldehydes and ketones, in proportion to their concentration [3,35]. The effect of high ph on carbonyl production during the assay was examined in model wine solution containing either 4-methylcatechol or catechin (Fig. 1). When the ph of model solutions was raised to produce the hydrolysis of sulfite-bound carbonyls ( 13.5 with NaOH), a rapid increase in the concentration of acetaldehyde (from ca. 2 to more than 9 mg L 1 in less than 1 h) was observed compared to a control solution in which no base was added. To prevent such interference, an oxygen exclusion step was employed in which N 2 -sparged NaOH and headspace blanketing with N 2 during the sample hydrolysis was used. In order to validate this protocol, excess acetaldehyde production during alkaline hydrolysis was measured in spiked-acetaldehyde (20 mg L 1 final) model wine samples containing or excluding 4-methylcatechol (1.5 g L 1 ), SO 2 (120 mg L 1 ), and air (Fig. 2). A blank was initially analyzed (data not shown), and a baseline acetaldehyde concentration of 2.2 mg L 1, attributed to impurities in the ethanol used to prepare the model wine solution, was established. In samples in which 4-methylcatechol was excluded, no increase in acetaldehyde was detected during hydrolysis. In addition, this experiment served to demonstrate that a satisfactory recovery of acetaldehyde is achieved in the presence of excess bisulfite (100.61% recovery), and that no acetaldehyde is volatilized by the N 2 blanket (100.56% recovery). In samples containing 4-methylcatechol, a large increase in acetaldehyde concentration was observed when the alkaline hydrolysis step was carried out under air. Clearly, a good hydrogen donor such as 4-methylcatechol is needed to catalyze the oxidation of ethanol to acetaldehyde in the presence of oxygen. The presence of SO 2 (120 mg L 1 ) in this system inhibited the production of acetaldehyde by some degree, perhaps by reacting with hydrogen peroxide resulting from catechol oxidation [43]. Alternatively, the reduction in acetaldehyde concentration could be due to the formation of its bisulfate adduct. When model wine samples containing 4-methylcatechol were treated with N 2 -sparged NaOH and hydrolyzed under a light N 2(g) blanket, no significant (p 0.05) increase in acetaldehyde was observed. Thus, the combination of using both deoxygenated NaOH and a N 2(g) blanket during hydrolysis proved sufficient to inhibit phenol oxidation, therefore deoxygenating the sample itself was unnecessary. The interfering effect of base-catalyzed oxidation on carbonyl analysis was tested and confirmed in a real system, in which red wine was treated with NaOH under both aerial and anoxic condition (Table 1). Alkaline hydrolysis carried out under normal, aerial conditions resulted in significantly higher concentrations of all analyzed carbonyls (p 0.05), despite the presence of added SO 2 (120 mg L 1 added) Chromatography A satisfactory chromatographic separation was achieved, with all compounds eluting in less than 32 min, and only minor coelution problems for 2-ketoglutaric acid in some of the wine samples analyzed (Fig. 3). Peaks 1 5 correspond to glyceraldehyde, 2-ketoglutaric acid, pyruvic acid, formaldehyde, Table 3 Recovery and precision of pure standards spiked into model wine matrix Compound Spike concentration (mg L 1 ) Measured concentration (mg L 1 ) Recovery (%) S.D. %RSD Glyceraldehyde Ketoglutaric acid Pyruvic acid Formaldehyde Acetaldehyde

6 analytica chimica acta 626 (2008) and acetaldehyde, respectively, while peak I is the unreacted DNPH. As is evident from Fig. 3, wine type and age, as well as winemaking style affect the concentration of carbonyl compounds, with increasing carbonyl concentration mostly notably the presence of acetaldehyde for older wines (Fig. 3B) and those produced without added SO 2 (Fig. 3A). The unlabeled peaks in the chromatograms represent other carbonyl substances that were not identified as part of this study Validation of the method Table 2 demonstrates that good linearity of the detector response was obtained over the concentration range used. For all of the compounds analyzed, the determination coefficient (R 2 ) was at least The method showed good detection limits, ranging from 1.29 to 7.53 gl 1 for all analytes, well below the reported concentrations of most aldehydes and ketones in wine. Minimum detectable concentrations of analytes were calculated based on instrument noise, as described previously [44]. To evaluate accuracy, model wine samples were spiked with 8 mg L 1 of pure standards, and recovery results were within 97.9 and 101.1% (Table 3). Precision, measured as RSD% (n = 7), was between 0.45 and 4.57% for formaldehyde, acetaldehyde, 2-ketoglutaric acid and pyruvic acid, and 10.6% for glyceraldehyde (Table 3) Application of the method Several California wines were analyzed for glyceraldehyde, 2-ketoglutaric acid, pyruvic acid, formaldehyde, and acetaldehyde using the above-described method. The results obtained for three of the most representative wines of this set are given in Fig. 4: an unsulfited ( no added sulfites ) Petite Syrah Fig. 5 Kinetics of formation of DNPH-carbonyl derivatives and effect of prolonged storage of derivatized wine samples at ambient temperature. (Frey Vineyards, 2004; Redwood Valley, CA), an aged Chardonnay (University of California, 1999; Davis, CA), and a young Cabernet Sauvignon (Woodbridge, 2007; Lodi, CA). Wine samples were injected onto the HPLC after 3 h of derivatization with the DNPH reagent. While some carbonyl analytes were observed to react quickly with DNPH (e.g., acetaldehyde), other compounds (e.g., pyruvic acid, 2-ketoglutaric acid) proved less reactive in the wine matrix (Fig. 5). It was determined that 3 h was sufficient for the complete derivatization of all target carbonyls; however, a marked increase in all analytes, with the exception of formaldehyde, was observed if the DNPHderivatized wine samples were allowed to stand for 24 h or more at room temperature (Fig. 5). The additional carbonyls observed during this period may result from oxidation reactions of wine components occurring in the assay matrix, and demonstrate the need for timely sample analysis. As such, all wine samples were stored at 4 C following derivatization until analysis by HPLC, which satisfactorily minimized this effect. Further work with this methodology should address some of the many volatile aldehydes that have been identified over threshold limits in wine [6]. For minor carbonyl species, a solid-phase extraction (SPE) step could be included as a way to concentrate the samples and allow for a lower limit of detection of these derivatives [45]. 4. Conclusions Fig. 4 Survey of targeted carbonyls in wine samples: (A) Petite Syrah, Frey Vineyards, 2004; Redwood Valley, CA; (B) Chardonnay, University of California, 1999; Davis, CA, and (C) Cabernet Sauvignon, Woodbridge, 2007; Lodi, CA. A robust HPLC method for the determination of several carbonyl oxidation products in wine has been developed. For the first time, the production of interfering carbonyl species derived from the commonly used alkaline hydrolysis of hydroxysulfornates have been identified. A simple method for preventing such artifact formation has been described, which serves as an improved alternative to the current technique available for the evaluation of carbonyl compounds in wines, and it could be readily adapted to other procedures using this sulfonate hydrolysis step.

7 110 analytica chimica acta 626 (2008) Acknowledgements Partial support of this work was provided by the American Vineyard Foundation and the U.S. Tobacco, Tax, and Trade Bureau (formerly the Bureau of Alcohol, Tobacco, and Firearms). references [1] S.E. Ebeler, R.S. Spaulding, In: A.L. Waterhouse, S.E. Ebeler (Eds.), Chemistry of Wine Flavor, ACS Symposium Series 714, San Francisco, 1997, p [2] C.S. Ough, M.A. Amerine, Methods for Analysis of Musts and Wines, J. Wiley, New York, [3] J.C. Danilewicz, Am. J. Enol. Vitic. 54 (2003) 73. [4] G. de Revel, N. Martin, L. Pripis-Nicolau, A. Lonvaud-Funel, A. Bertrand, J. Agric. Food Chem. 47 (1999) [5] R. Flamini, G. De Luca, R. Di Stefano, Vitis 41 (2002) 107. [6] L. Cullere, J. Cacho, V. Ferreira, J. Agric. Food Chem. 55 (2007) 876. [7] A. Escudero, E. Asensio, J. Cacho, V. Ferreira, Food Chem. 77 (2002) 325. [8] J.C. Barbe, G. de Revel, A. Joyeux, A. Lonvaud-Funel, A. Bertrand, J. Agric. Food Chem. 48 (2000) [9] C. Bayonove, R. Baumes, J. Crouzet, Z. Gunata, In: C. Flanzy (Ed.), Enología: Fundamentos científicos y tecnológicos, AMV Mundiprensa, Madrid, [10] C. Timberlake, P. Birdie, J. Sci. Food Agric. 28 (1977) 539. [11] J. Bakker, C.F. Timberlake, J. Agric. Food Chem. 45 (1997) 35. [12] S.Q. Liu, G.J. Pilone, Int. J. Food Sci. Technol. 35 (2000) 49. [13] H. Fulcrand, V. Cheynier, J. Oszmianski, M. Moutounet, Phytochemistry 46 (1997) 223. [14] H. Fulcrand, C. Benabdeljalil, J. Rigaud, V. Cheynier, M. Moutounet, Phytochemistry 47 (1998) [15] N.E. Es-Safi, V. Cheynier, M. Moutounet, J. Agric. Food Chem. 50 (2002) [16] V.F. Laurie, A.L. Waterhouse, J. Agric. Food Chem. 54 (2006) [17] V.F. Laurie, A.L. Waterhouse, J. Agric. Food Chem. 54 (2006) [18] L.F. Burroughs, A.H. Sparks, J. Sci. Food Agric. 15 (1964) [19] J.F. Guymon, E.A. Crowell, J. A. O. A. C. 46 (1963) 276. [20] A.G.H. Lea, G.D. Ford, S. Fowler, Int. J. Food Sci. Technol. 35 (2000) 105. [21] M. Vogel, A. Buldt, U. Karst, Fresenius J. Anal. Chem. 366 (2000) 781. [22] S.D. Richardson, T.V. Caughran, T. Poiger, Y.B. Guo, F.G. Crumley, Ozone-Sci. Eng. 22 (2000) 653. [23] M.N. Lau, J.D. Ebeler, S.E. Ebeler, Am. J. Enol. Vitic. 50 (1999) 324. [24] W.K. Carlton, B. Gump, K. Fugelsang, A.S. Hasson, J. Agric. Food Chem. 55 (2007) [25] S.K. Shannon, G. Barany, J. Comb. Chem. 6 (2004) 165. [26] M. Behforouz, J.L. Bolan, M.S. Flynt, J. Org. Chem. 50 (1985) [27] S.M. van Leeuwen, L. Hendriksen, U. Karst, J. Chromatogr. A 1058 (2004) 107. [28] C. Zwiener, T. Glauner, F.H. Frimmel, Anal. Bioanal. Chem. 372 (2002) 615. [29] P. Ribéreau-Gayon, Y. Glories, A. Maugean, D. Dubourdieu, Handbook of Enology. The Chemistry of Wine Stabilization and Treatments, John Wiley & Sons Ltd., Chichester, [30] L. Pripis-Nicolau, G. de Revel, A. Bertrand, A. Maujean, J. Agric. Food Chem. 48 (2000) [31] L.F. Burroughs, G.C. Whitting, The sulfur dioxide combining power of cider, Ann. Report Agr. Hort. Res. Station, Long Ashton, Bristol, England, 1960, p [32] L.F. Burroughs, A.H. Sparks, J. Sci. Food Agric. 24 (1973) 187. [33] L.F. Burroughs, A.H. Sparks, J. Sci. Food Agric. 24 (1973) 207. [34] H.L. Wildenradt, V.L. Singleton, Am. J. Enol. Vitic. 25 (1974) 119. [35] A.L. Waterhouse, V.F. Laurie, Am. J. Enol. Vitic. 57 (2006) 306. [36] J.A. Rossi Jr., V.L. Singleton, Am. J. Enol. Vitic. 17 (1966) 231. [37] V.L. Singleton, J.A. Rossi Jr., Am. J. Enol. Vitic. 16 (1965) 144. [38] J. Ribéreau-Gayon, Contributions a l etude des oxidations et réductions dans le vins., Delmas, Bordeux, [39] V.L. Singleton, Am. J. Enol. Vitic. 38 (1987) 69. [40] B. Bandy, A.J. Davison, Arch. Biochem. Biophys. 259 (1987) 305. [41] P. Gee, A.J. Davison, Arch. Biochem. Biophys. 231 (1984) 164. [42] J.J.L. Cilliers, V.L. Singleton, J. Agric. Food Chem. 37 (1989) 890. [43] R.B. Boulton, V.L. Singleton, L.F. Bisson, R.E. Kunkee, Principles and Practices of Winemaking, Chapman, New York, [44] D.C. Harris, Quantitative Chemical Analysis, W.H. Freeman and Co., New York, [45] R.J. Kieber, K. Mopper, Environ. Sci. Technol. 24 (1990) 1477.