The Pennsylvania State University

The Graduate School

Department of Food Science

INVESTIGATING THE FORMATION AND FATE OF ETHYL RADICALS IN WINE

A Thesis in

Food Science

by

Gal Y. Kreitman

 2013 Gal Y. Kreitman

Submitted in Partial Fulfillment of the Requirements for the Degree of

Master of Science

May 2013

The thesis of Gal Y. Kreitman was reviewed and approved* by the following:

Ryan J. Elias Assistant Professor of Food Science Thesis Advisor

Joshua D. Lambert Assistant Professor of Food Science

John E. Hayes Assistant Professor of Food Science

Robert F. Roberts Professor of Food Science Interim Head of the Department of Food Science

*Signatures are on file in the Graduate School

ii

ABSTRACT

The shelf life of wine is greatly affected by its oxidative stability. The non- enzymatic oxidation of wine is thought to proceed by metal-catalyzed reactions capable of oxidizing phenolics and reducing oxygen to quinones and hydroxyl radicals, respectively. Previous studies have focused upon the important reactions that occur between quinones and other wine components, but the fate of oxygen radicals within the context of wine oxidation is less understood.

Ethanol is the prime target for hydroxyl radicals in wine due to its abundance (~2

M). I investigated the oxidative stability of model wine solutions with respect to major wine constituents after ethanol (i.e. phenolics, glycerol, and tartaric acid), knowing ethanol forms 1-hydroxyethyl radicals (1-HER) using spin trapping with phenyl-N-tert- butyl nitrone (PBN) paired with electron paramagnetic spectroscopy (EPR). I then focused on the potential reaction between hydroxyl radicals and tartaric acid, and measured other important oxidative markers including hydrogen peroxide consumption and acetaldehyde formation, a known oxidation product of ethanol. As expected, 4- methylcatechol (4-MeC), a model phenolic, accelerated the oxidation of ethanol by hydroxyl radicals, and had resulted in higher 1-HER/PBN adduct formation. Glycerol, despite its presence at relatively high concentrations, did not significantly affect spin adduct formation, suggesting that it was not a major target of hydroxyl radical-mediated oxidation. Interestingly, tartaric acid, seemed to shift from slight antioxidant activity to pro-oxidant activity when oxygen was purged from the model wine system.

I investigated the reaction between 1-HER and various wine-related phenolics and thiols, including gallic acid, caffeic acid, ferulic acid, 3-mercaptohexan-1-ol (3MH), cysteine (Cys), and glutathione (GSH) by using competitive spin trapping with detection by EPR and mass spectrometry. In addition, I measured the ability of these compounds

iii

to prevent the formation of acetaldehyde. The hydroxycinnamic acids, which include caffeic acid and ferulic acid, significantly inhibited 1-HER formation compared to treatment-free model wine by 69% and 55%, respectively. The thiol treatments, Cys,

GSH, and 3MH also significantly inhibited 1-HER formation by 88, 87, and 96%, respectively. As expected, I observed similar trends when I measured spin adduct formation with mass spectrometry and when I measured final acetaldehyde yields. This confirms that hydroxycinnamic acids and thiols react with 1-HER under model wine conditions. I also measured the oxidative loss of 3MH, an important varietal thiol, in model wine resulting from either 2-electron quinone trapping or direct reaction with 1-

HER. I observed, for the first time, evidence that 1-HER reacts with 3MH; however, the well-known quinone trapping mechanism is likely the dominant mechanism for 3MH loss in wine.

Finally, with the knowledge that 1-HER formation in wine is catalyzed by transition metals, I investigated the use of chelators as a means to control 1-HER formation. I focused on the effect of exogenous Fe(II) (bipyridine, Ferrozine) and Fe(III) chelators (EDTA, phytic acid) on 1-HER and acetaldehyde formation using spin trapping with EPR and HPLC-PDA, respectively. I then investigated the ability for these chelator treatments to prevent the oxidative loss of the important aroma-active thiol, 3MH. The

Fe(II)-specific chelators were more effective antioxidants than the Fe(III)-specific chelators during the early stages of oxidation, and completely inhibited 1-HER formation compared to the chelator-free control. However, while the addition of Fe(III) chelators was less effective or even pro-oxidative initially, the Fe(III) chelators proved to be more effective antioxidants compared to Fe(II) chelators over the course of the study. In addition, I show for the first time that Fe(II) and Fe(III) chelators can significantly inhibit the oxidative loss of 3MH in model wine.

iv

TABLE OF CONTENTS LIST OF FIGURES ...... vii LIST OF SCHEMATICS ...... ix LIST OF TABLES ...... x ACKNOWLEDGEMENTS ...... vii Chapter 1: Literature review ...... 1 1.1 Introduction ...... 1 1.2 Wine Oxidation ...... 3 1.2.1 Role of Oxygen in Wine Oxidation ...... 4 1.2.2 Role of Transition Metals in Wine Oxidation Catalysis ...... 5 1.2.3 Role of Phenolics in Wine Oxidation ...... 6 1.2.4 Role of Sulfur Dioxide in Wine Oxidation ...... 9 1.2.5 Oxidation of Wine Constituents ...... 10 1.2.5.1 Reaction with Quinones ...... 10 1.2.5.2 Hydroxyl radical reaction ...... 12 1.2.5.3 1-Hydroxyethyl Radical Reaction ...... 14 1.3 Importance of Varietal Thiols in Wine ...... 15 1.4 Preventing oxidation...... 17 1.4.1 Bottle closure and environment ...... 17 1.4.2 Antioxidants ...... 18 1.4.3 Metal Chelation ...... 19 1.5 Purpose and Significance ...... 22 1.6 Hypothesis and Objectives ...... 23 Chapter 2: Hydroxyl radical scavenging capability of major wine components ...... 24 2.1 ABSTRACT ...... 24 2.2 INTRODUCTION ...... 25 2.3 MATERIALS AND METHODS ...... 27 2.4 RESULTS AND DISCUSSION ...... 31 Chapter 3: Investigation of Ethyl Radical Quenching by Phenolics and Thiols in Model Wine ...... 43 3.1 ABSTRACT ...... 43 3.2 INTRODUCTION ...... 44 3.3 MATERIALS AND METHODS ...... 47

v

3.4 RESULTS AND DISCUSSION ...... 53 Chapter 4: The effect of metal chelators on the oxidative stability of model wine...... 67 4.1 ABSTRACT ...... 67 4.2 INTRODUCTION ...... 68 4.3 MATERIALS AND METHODS ...... 71 4.4 RESULTS AND DISCUSSION ...... 76 CONCLUDING REMARKS ...... 86 REFERENCES ...... 88

vi

LIST OF FIGURES

Figure 1: 1-HER/PBN adduct formation under air. All samples contained 12% EtOH, 90μM Fe(II), and 30mM PBN (specific treatments in Table 5)...... 32

Figure 2: 1-HER/PBN spin adduct formation with varying concentrations of tartaric acid, glycerol, and 4-methylcatechol under Fenton conditions (90μM Fe(II) and 300μM H2O2) in aerated model wine (12% EtOH, pH 3.6) and 30mM PBN ...... 35

Figure 3: PBN/1-HER EPR spin intensity after induced Fenton reaction in deoxygenated (black bars) and aerated (grey bars) model wine solution at varying tartaric acid concentration. Note: deoxygenated samples were measured at 3 scans whereas aerated samples were measured at 10 scans...... 37

Figure 4: Residual H2O2 after induced Fenton reaction in deoxygenated (black bars) and aerated (grey bars) model wine solution at varying tartaric acid concentration...... 39

Figure 5: Acetaldehyde formation after induced Fenton reaction in deoxygenated (black bars) and aerated (white bars) model wine solution at varying tartaric acid concentration...... 41

Figure 6: Possible PBN/1-HER adducts formed: (1) oxidized (222 m/z), (2) radical (223 m/z), (3) reduced (224 m/z), and (4 and 5) bi-adducts (268 m/z)...... 50

Figure 7: Chemical structures of caffeic acid (CA), gallic acid (GA), ferulic acid (FA), 3-mercaptohexan-1-ol (3MH), cysteine (Cys), and glutathione (GSH)...... 52

Figure 8: Representative experimental EPR spectrum corresponding to the PBN/1-HER spin adduct in model wine conditions...... 54

Figure 9: EPR spin adduct intensities of PBN/1-HER in the presence of 4-MeC, PBN, and selected treatments in model wine in the absence of oxygen. Spectra were obtained at room temperature after 1 min following the addition of Fe(II) and H2O2. *Statistically significant difference (P < 0.05) from control by one-way ANOVA with Dunnett‘s post-test...... 55

Figure 10: 3MH disulfide and 4-MeC-3MH adduct formation from the following treatments: 3MH only (measured immediately); 3MH only; 3MH + 4-MeC; Fenton reaction (Fe(II) + H2O2); Fenton reaction (Fe(II) + H2O2) + 4-MeC. Reactions were run in model wine (pH 3.6) in the absence of oxygen and measured after 30 min reaction at ambient temperature ...... 57

Figure 11: Total MS spin adduct intensities of all possible PBN/1-HER adducts (oxidized, 222 m/z; radical, 223 m/z; reduced, 224 m/z; bi-adducts, 268 m/z) in the presence of 4-MeC, PBN, and selected treatments in model wine in the absence of oxygen. Spectra were obtained at room temperature after 1 min following the addition of Fe(II) and H2O2. *Statistically significant difference (P < 0.05) from control by one-way ANOVA with Dunnett‘s post-test...... 59

vii

Figure 12: Acetaldehyde formation in the presence of 4-MeC and phenolic and thiol treatments in model wine in the absence of oxygen. *Statistically significant difference (P < 0.05) from control by one-way ANOVA with Dunnett‘s post-test...... 61

Figure 13: 3MH loss resulting from the following treatments: H2O2 only; Fenton reaction (Fe(II) + H2O2); Fe(II) + 4MeC; Fenton reaction (Fe(II) + H2O2) + 4MeC. Reactions was run in either model wine (pH 3.6) or mannitol solution (pH 3.6) in the absence of oxygen. *Statistically significant difference (P < 0.05) from model wine by Student‘s t-test...... 64

Figure 14: Intensity of PBN/1-HER spin adducts (arbitrary units) in the presence of various iron chelator treatments in model wine (12% EtOH, 8 g/L tartaric acid, pH 3.6) and 4-MeC under air at ambient temperature ...... 77

Figure 15: Intensity of PBN/1-HER spin adducts (arbitrary units) in the presence of various iron chelators in pinot gris adjusted to pH 3.6 under air at ambient temperature...... 79

Figure 16: Acetaldehyde formation over time in the presence of various iron chelator treatments in model wine (12% EtOH, 8 g/L tartaric acid, pH 3.6) and 4- MeC under air at ambient temperature...... 80

Figure 17: Acetaldehyde formation over time in the presence of various iron chelator treatments in model wine (12% EtOH, 8 g/L tartaric acid, pH 3.6) and 4- MeC under air at 50°C...... 81

Figure 18: 3MH loss over time in the presence of various iron chelator treatments in model wine (12% EtOH, 8 g/L tartaric acid, pH 3.6) and 4-MeC under air at 50°C...... 84

viii

LIST OF SCHEMATICS

Scheme 1: Proposed metal catalyzed reduction of oxygen to hydrogen peroxide and hydroxyl radicals, and subsequent oxidation of ethanol to 1-hydroxyethyl radical...... 3

Scheme 2: (A) formation of quinone upon the oxidation of catechol by loss of 2 electrons and 2 protons. (B) metal catalyzed oxidation of catechol and regeneration of Fe(II) ...... 8

Scheme 3: The competitive reaction of sulfite and iron for hydrogen peroxide...... 9

Scheme 4: The radical chain reaction of sulfites to yield potent sulfite radicals...... 10

Scheme 5: The reaction between electrophilic quinone and various nucleophiles found in wine...... 11

Scheme 6: The reaction of hydroxyl radical with major wine constituents...... 12

Scheme 7: Proposed reaction of 1-hydroxyethyl radical in wine conditions ...... 14

Scheme 8: 3-mercaptohexan-1-ol production during fermentation...... 16

Scheme 9: Proposed reaction mechanisms of 1-HER with: (1) ferric ions or oxygen, (2) thiols, (3), and hydroxycinnamic acids...... 45

Scheme 10: Proposed mechanisms for the loss of 3MH by (1) reaction with 1- HER and subsequent disulfide formation, (2) Michael-type addition reaction to o- quinone, (3) reaction with semiquinone radicals...... 62

ix

LIST OF TABLES

Table 1: Sources for introduction of atmospheric oxygen in wine...... 1

Table 2: Calculated values of relevant reactive oxygen species to wine at pH 3.6 .. 4

Table 3: Values obtained by voltammetry in model wine Ph 3.6 and 0.5Mm concentration, with 207 Mv added to give values versus standard hydrogen electrode...... 7

Table 4: Chemical makeup of wine...... 13

Table 5: Full factorial design used for investigating the effect of major wine constituents on 1-HER formation...... 28

x

ACKNOWLEDGEMENTS

My utmost sincere thanks go to my advisor, Dr. Ryan Elias, for giving me the opportunity to join his lab. I thank him for introducing me to the frustrating yet rewarding field of wine research. Without his patience, support, and dedication this would not have been possible.

I would also like to thank my committee members, Dr. Joshua Lambert and Dr.

John Hayes, as well as Ms. Denise Gardner for their willingness to talk with me whenever I stopped by. Their insight and guidance have been incredibly useful throughout my time here.

I would like to thank the Department of Food Science for providing salary and tuition support. I would also like to thank the Pennsylvania Wine Research and

Marketing Board for providing some funding support for this project.

I thank all my lab mates and classmates for being supportive of me. They have taught me many valuable skills and helped me develop as a scientist. They have made my experiences here much more enjoyable by being great friends socially and academically. My family and friends at home have always provided love and support, and for that, I am eternally grateful.

xi

Chapter 1: Literature review

1.1 Introduction

Oxygen is ubiquitous in the winemaking process, and can be introduced in virtually every step (Table 1). The impact of oxidation in finished wine can lead to both desirable and deleterious effects depending on the type of wine. For example, oxidation can stabilize the color of red wine by anthocyanin polymerization, resulting in a persistent brick red color which is highly desirable (1). In addition, tannin precipitation and polymerization can result in softening of astringent phenolics in red wine during barrel and bottle aging, yielding a wine with improved mouthfeel (2). Oxidation of white wines, however, can result in browning due to catechin polymerization (3,4) as well as yellowing due to phenolic reactions with oxidation products such as ketones and aldehydes (5,6). Since white wines tend to have a low phenolic content, there are no known improvements in mouthfeel that arise from wine oxidation and, in fact, oxidation may accent the bitterness of some white wines (7,8).

Table 1: Sources for introduction of atmospheric oxygen in wine (9,10)

Operation Dissolved Oxygen (mg/L) Crushing, Pressing Saturation Racking 2.2 to saturation Pumping 0.1 – 0.2 Centrifugation 1.0 Filtration 0.7 Bottling 1.6

1

Aroma is perhaps the most important attribute affecting the quality of wine but is one of the major attributes that change during oxidation. There are some beneficial effects on aroma due to oxidation, such as the removal of unwanted reductive thiols which contribute to malodorous aromas; however, by the same mechanism, there is a loss of important thiols that are vital in varietal wines as they provide pleasant fruity aromas (11). Oxidation contributes to the generation of various aldehydes from ethanol, organic acids, and sugars in wine (12). These aldehydes are usually associated with wine defects in table and dessert wines, but there are special cases where the presence of these aldehydes is highly desirable, such as in maderized wines (e.g., sherry,

Madeira). For the most part, the oxidation of wine post-bottling is rarely beneficial, particularly in white wines, and winemakers are in need of better tools to control these processes.

There are relatively few ways to reliably and reproducibly control oxidation reactions in wine. The most commonly employed (and most important) strategy is the careful monitoring of oxygen throughout the winemaking process. Bottle closures, such as metal roll-on tamper evident (ROTE), screwcaps, glass bottle ampoules, among other physical deterrents of oxygen ingress have been proven successful at minimizing the oxygen transfer rate and have been utilized in the industry (13,14). The use of antioxidants has also been employed; the most commonly used antioxidant in wine is sulfur dioxide, which is produced naturally during alcoholic fermentation to a minor extent and is almost always added to the wine at bottling. Glutathione (GSH) and ascorbic acid have also been explored as antioxidants and as potential sulfur dioxide replacements, yet with disappointing results (15–17).

The purpose of this review is to provide a concise but comprehensive overview of the oxidative stability of wine in regards to both major and minor wine constituents. In

2

addition, beneficial thiols in wine will be reviewed due to their importance in wine. Ways of preventing oxidation, particularly via chelators and metal redox inactivation, will also be reviewed.

1.2 Wine Oxidation

The non-enzymatic oxidation of wine yields two important reactive intermediates: quinones and hydroxyl radicals (18). These species are now thought to be produced by the metal-catalyzed oxidation of phenolics during the sequential one-electron reduction of oxygen to hydroperoxyl radical, hydrogen peroxide, hydroxyl radical, and eventually water (19,20) (Scheme 1). Oxygen, phenolics, transition metals, and sulfites all play a major role during this initial oxidation process.

OH O

R OH R OH H+ OH-

O2 HOO HOOH HO

Fe2+ Fe3+ Fe2+ Fe3+

CH3CH2OH

CH3CHOH O OH

R OH R OH H2O

Scheme 1: Proposed metal catalyzed reduction of oxygen to hydrogen peroxide and hydroxyl radicals, and subsequent oxidation of ethanol to 1-hydroxyethyl radical.

3

1.2.1 Role of Oxygen in Wine Oxidation

It is obvious that oxygen is the fuel in the wine oxidation process. Because of this, there is a high interest in controlling for oxygen, and many publications discuss accurate measurement and control of oxygen (21–23). Concentrations of dissolved oxygen are highly variable with temperature and partial pressure, but at room temperature it reaches saturation at approximately 8.4 mg/L (24). As the temperature is decreased, the solubility of oxygen increases, but the rate of oxygen consumption decreases (24). Dissolved oxygen in must and wine reacts with phenolics and sulfites in the presence of transition metals (18,24–26). Oxygen is quite stable in its ground state and is spin forbidden from reacting directly with organic compounds, so it must be activated to its singlet state first. This could potentially occur by photolysis if exposed to light for extended periods of time (18,27), but since this is not often the case, the most common way for oxygen to be activated is by metal catalysis. As shown in Scheme 1, oxygen is eventually reduced to water, but in the process it forms very potent reactive oxygen species, such as the hydroperoxyl radical (HOO•) and hydroxyl radical (HO•), the reduction potential of these reactive oxygen species is reported in Table 2. The metal- catalyzed breakdown of H2O2 to form HO• radicals is well known as the Fenton reaction

(19). The HO•/H2O couple has a reduction potential of 2.53 V at pH 3.6, which is high enough to oxidize any given organic molecule in wine at a diffusion-limited rate (18).

Table 2: Calculated values of relevant reactive oxygen species to wine at pH 3.6 (28).

Reactive oxygen species Redox potential (V)

Ground state oxygen (O2) -0.09 Hydroperoxyl radical (HOO•) 1.22

Hydrogen peroxide (H2O2) 0.59 Hydroxyl radical (HO•) 2.53

4

1.2.2 Role of Transition Metals in Wine Oxidation Catalysis

Transition metals are thought to be essential catalysts in the oxidation of wine as they facilitate one electron transfers to oxygen to yield reactive oxygen species. These trace levels of metals can come from a variety of sources including the grapes themselves, dust, fungicide residues, and the winemaking equipment (7). The concentration of iron, copper, and manganese in wine have been reported as 0.11-

3.6mg/L, 0.9-16mg/L, and 0.2-1.2mg/L respectively (7,29). It has been shown recently that these transition metals, particularly iron and copper, are essential as catalysts for the reduction of oxygen and oxidation of wine components including polyphenols, sulfites, and ethanol (30–33). Ribéreau-Gayon showed that the rate of oxidation could be slowed and eventually stopped in wine by the removal of iron and copper with potassium ferrocyanide (34). This was recently confirmed in another study by Danilewicz and Wallbridge (35).

The rate of oxidation is highly dependent on the reduction potential of the

Fe(III)/Fe(II) couple (18,28,36). The lower the reduction potential, the greater the reducing power. Therefore, if the reduction potential of the Fe(III)/Fe(II) couple is low, dioxygen will be reduced to its active peroxyl radical form more readily. It also contributes to the reduction of H2O2 to HO• radicals via the Fenton reaction (Scheme

1). The reduction potential of the Fe(III)/Fe(II) is dependent on the ligands present in the wine solution. Tartaric acid and phenolics in wine bind to iron and affect its reduction potential and it has been suggested that the reduction potential of Fe(III)/Fe(II) couple in wine pH falls between 0.18-0.36 V (18). Once Fe(II) is oxidized into Fe(III), it quickly reduces back to Fe(II) in the presence of phenolics (36). The iron speciation in wine has been examined, and it has been suggested that the majority of free iron is present as

Fe(II) (36,37), though a large fraction of iron is bound to the organic fraction of wine such

5

as tartrate (38) and tannins (39). Fe(II) is the major species of iron in wine due to its acidic pH and abundance of phenolics.

1.2.3 Role of Phenolics in Wine Oxidation

Phenolic compounds in wine are mainly extracted from grape skins and seeds, but a minor fraction is also extracted from oak during fermentation (1). The concentration of phenolics depends on the grape variety, the form of oak used, and processing methods. The final concentration of phenolics vary widely across wine varietals and styles; however, the catechin equivalent of white wines are reported to be in the range of

50 mg/L – 1500 mg/L (7,40). In general, the concentration of catechin equivalent in white wines is 10-fold less than in red wines. Phenolic compounds in wine can quickly consume oxygen (in the presence of iron) and it has been shown that wines with higher phenolic contents, such as reds, are able to consume more oxygen relative to whites before showing any obvious signs of oxidation (24,41). Wine phenolics are often considered to be antioxidants due to their ability to form a resonance stabilized structure by the delocalization of the free radical (42). However, in wine conditions, the ability to be quickly oxidized propagates the oxidation of wine due to the formation of H2O2 (7)

(Scheme 1). There are varying reduction potentials for the different phenolics in wine

(Table 3), which in turn have different reaction rates with reactive oxygen species

(18,28). In general, the lower the reduction potential, the more readily the phenolic compound oxidizes. It has been shown that polyphenols containing catechol (1,2- diphenol) or gallate (1,2,3-triphenol) moieties are the most easily oxidized due to their low reduction potential (18,43). Wine phenolics that contain a catechol group include catechin, epicatechin, caftaric acid, and condensed tannins. The gallate containing group include gallic acid, gallocatechin, and epigallocatechins (7,12,18). Quercetin has lower reduction potential then most wine phenolics, as it has an extended conjugation

6

with the heterocyclic ring providing further stabilization then catechin. This contributes to its quick loss despite its presence in low concentrations (18,44).

Table 3: Values obtained by voltammetry in model wine pH 3.6 and 0.5mM concentration (43), with 207 mV added to give values versus standard hydrogen electrode (28).

Compound Structure E3.6 (mV) (+)-Catechin 577

(-)-Epicatechin 568

Caffeic acid 604

Gallic acid 582

Quercetin 567

4-Methylcatechol 541

1,2-Benzenediol 582

Pyrogallol 468

7

Catechols are oxidized to quinones upon the successive loss of two electrons and two protons (18) (Scheme 2A). The two major paths for oxidation of catechols in wine are via metal-catalyzed oxidation and radical-mediated oxidation. The most likely radical to oxidize the catechol in wine is the HOO• radical which can react to yield a semiquinone while simultaneously reducing the HOO• radical into H2O2 (Scheme 1); however, it is also possible for the phenolics to react with HO• radicals and sulfite radicals (30). The semiquinone radical would then react again to be oxidized into a quinone (7,12,18). Phenolic compounds have a pKa>9, and at wine pH (range of 3.0 -

3.9), almost all groups are completely protonated to give the neutral form (7,12). In order for metal-catalyzed oxidation reactions to proceed, the catechol needs to be ionized for the metal to bind, which is why oxidation rates are much faster at higher pH as the equilibrium shifts towards the ionic form (7,18). Catechol and gallate groups can reduce an equivalent of two Fe(III) ions to Fe(II) ions (Scheme 2B).

Scheme 2 – (A) formation of quinone upon the oxidation of catechol by loss of 2 electrons and 2 protons. (B) metal catalyzed oxidation of catechol and regeneration of Fe(II)

Quinones are highly electrophilic species and therefore can participate in a variety of reactions leading to detrimental sensory characteristics, by the loss of aromas and browning due to polymerization. In the presence of sulfur dioxide, quinones can be cycled back to their hydroquinone (reduced) form. Due to this ability to regenerate, more

8

oxygen can be consumed prior to showing any signs of oxidative characteristics (24).

However, this regeneration also allows for the cycling of Fe(III)/Fe(II) and/or further generation of H2O2.

1.2.4 Role of Sulfur Dioxide in Wine Oxidation

- At wine pH, sulfur dioxide is most commonly found in its bisulfite form (HSO3 ),

- which is thought to be the form that is reactive with H2O2. HSO3 acts as an important preventive antioxidant in wine as it can react with unstable quinones and remove H2O2

- (12,18,31). In addition, HSO3 can react with aldehyde and ketone compounds via addition reactions (45) which are formed after the oxidation of major components such as ethanol and tartaric acid. This can reduce the negative aromas associated with these compounds (46) and is considered to be bound sulfur dioxide. However, these reactions are reversible and as oxidation progresses, loss of sulfite will cause a shift in equilibrium

- and HSO3 will become unbound from the aldehydes and ketones. This will remove the masking effect of sulfur dioxide on these oxidized aroma compounds.

- HSO3 reacts with H2O2 by an acid-catalyzed mechanism that does not involve radical intermediates (47) (Scheme 3). By removing H2O2 at a faster rate than the

- Fenton reaction, HSO3 effectively prevents the formation of the HO• radical.

- Scheme 3: The competitive reaction of HSO3 and iron for H2O2

- HSO3 can reduce quinones back to their original hydroquinone form, but it has

- also been shown that the quinones can react with HSO3 to form addition products, both

9

- of which effectively regenerate the catechol (30,31) (Scheme 5). While HSO3 seems to behave mostly as an antioxidant, it has been suggested that its presence is necessary for oxidation to proceed at a fast rate. The reasoning for this is that the presence of

- HSO3 and other nucleophiles can effectively remove the quinone thus shifting the equilibrium and allowing further generation of H2O2 and Fe(II) by oxidation of catechols

(32).

The balance of both phenolics and sulfites in wine is important for the rate of oxidation in wine. In the presence of oxygen, sulfite autoxidation is catalyzed by metals to promote radical chain propagation. This is prevented by phenolics in wine as they are effective radical scavengers (30). In the absence of phenolics, sulfite autoxidation can yield sulfate radicals, which have comparable reduction potentials to HO• radicals (E3.6 =

2.43 V) (Scheme 4) (30,48).

Scheme 4: The radical chain reaction of sulfites to yield potent sulfite radicals

1.2.5 Oxidation of Wine Constituents

1.2.5.1 Reaction with Quinones

Quinones are oxidation products of catechol and gallate moieties from the primary wine oxidation process (Schemes 1, 2); they are highly unstable and can participate in a variety of reactions. As quinones are highly electrophilic they react with nucleophilic molecules in wine that include amino acids, sulfites, thiols, and flavanols

(Scheme 5).

10

Scheme 5: The reaction between electrophilic quinone and various nucleophiles found in wine.

Quinones can polymerize by reaction with flavanols including catechin and epicatechin (5,49). These polymerization reactions contribute to the browning of white

- wines, which is undesirable. In the presence of antioxidants including HSO3 and ascorbic acid, quinones quickly react and are reduced back to hydroquinones (31,50).

The reaction between quinones and amino acids can also occur in wine, which is proceeded by Strecker degradation reactions to yield important aldehydes including methional and phenylacetaldehyde (51) (Scheme 5).

3MH, an important varietal thiol, is significantly depleted by binding of its thiol group with ortho-quinones (11). Upon addition of the thiol to quinone, a catechol adduct is formed (52). The formation of these adducts has been directly shown using electrospray ionization paired with mass spectrometry (53) and nuclear magnetic resonance spectroscopy by both enzymatic and nonenzymatic chemical oxidation (54).

11

GSH has been shown to effectively reduce the loss of 3MH in wine (55) due to its reaction with quinones at a more competitive rate compared to 3MH.

A recent study by Nikolantonaki and Waterhouse have shown that wine

- antioxidants including HSO3 , GSH, and ascorbic acid, as well as hydrogen sulfide, have very high reaction rates with quinones, followed by volatile and varietal thiols, and lastly by nucleophilic amino acids (including methionine, phenylalanine) (56).

1.2.5.2 Hydroxyl radical reaction

The HO• radical is a very potent oxidant and can abstract a hydrogen atom from practically all organic compounds in wine in a concentration based manner (18,57).

Since ethanol is present at the highest concentration in wine at approximately 2 M, it is most likely to react with the HO• radical due to its abundance (Scheme 6). The HO• radical abstracts the α-hydrogen 85% of the time to form the 1-hydroxyethyl radical (1-

HER) and has been confirmed by electron paramagnetic resonance coupled with spin trapping techniques (48). 2-hydroxyethyl radicals are also formed as a minor species.

Scheme 6: The reaction of HO• radical with major wine constituents.

12

However, other minor wine constituents must be taken into consideration when it comes to the HO• radical (Table 4). Glycerol and tartaric acid are the most abundant compounds in wine after water and ethanol, present at 54-217 mM and 13-53 mM, respectively (46,58). It is expected that approximately 5-10% of the hydroxyl radicals will react with these compounds in a similar fashion to ethanol to produce aldehydes

(Scheme 6). It has been reported that tartaric acid and glycerol oxidize to yield glyoxylic acid and glyceraldehyde, respectively (58,59). However, it is worth noting that tartaric acid may not readily react with HO• radicals. Tartaric acid binds to Fe(III) and it has been suggested that Fe(III) tartrate complexes are not targets for HO• radicals formed from the Fenton reaction (60). In addition, more recent reports suggest that no glyoxylic acid is formed from tartaric acid in the absence of light (61,62). More research needs to be done to fully understand the role of tartaric acid in wine oxidation.

Table 4: Chemical makeup of wine (7).

Group Typical Range (mM) Group Typical Range (mM) Alcohol 1300-3500 Cations 0-64 Polyols 11-160 Phenolics 0-2 Sugars 0-560 Esters 0.6-1.7 Acids 7-33 Nitrogenous compounds 0.4-6 Anions 1-30 Vitamins 0-0.8

Phenolics can be oxidized by the HO• radical to form semiquinone radicals and eventually further oxidized to quinones in the presence of other radicals or Fe(III)

(Scheme 2). Although phenolics are found at relatively low concentrations in wine relative to ethanol and tartaric acid, it is possible that while phenolics chelate iron during its reduction, the Fenton reaction occurs in close proximity and the HO• radical can quickly react with the phenolic (63).

13

1.2.5.3 1-Hydroxyethyl Radical Reaction

1-HER is highly reducing and can quickly oxidize in the presence of Fe(III) or oxygen to produce acetaldehyde (18,36) (Scheme 7). Upon the reaction of 1-HER with

Fe(III), 1-HER will quickly be oxidized and Fe(III) will be simultaneously reduced to

Fe(II), which will be able to participate in the Fenton reaction again. In the presence of oxygen, 1-HER forms a 1-hydroxyethyl peroxyl radical intermediate which can then break down to yield acetaldehyde and HOO• radical (48). Acetaldehyde can then further react with sulfites and polyphenols. It is also possible for 1-HER to be reduced back to ethanol in the presence of highly reducing compounds, such as phenolics and thiols.

Scheme 7: reaction of 1-hydroxyethyl radical in wine conditions

It has recently been shown that 1-HER is reactive towards the α,β-unsaturated side chains of hydroxycinammic acids including caffeic and ferulic acids (64). 1-HER attacks the side chain double bond, and then attaches to the α-carbon. This forms a relatively stable benzyl radical which can then be oxidized by Fe(III) to form a

14

carbocation which will readily decarboxylate and yield an allylic alcohol (64) (Scheme 7).

It has also been shown that in beer isohumulones which contain highly resonant structures can react competitively for 1-HER with POBN spin trap in low levels (65). It has been suggested mechanistically that these isohumulones can either reduce 1-HER to ethanol and/or attach to 1-HER (66).

In biological systems, GSH and ascorbic acid can competitively reduce 1-HER radical back to ethanol (67), and there has been evidence for the oxidation of GSH to yield glutathione disulfide (67). This mechanism may also be applicable to sulfites and thiols in wine, as well as phenolics with a low enough reduction potential. We have recently demonstrated that important wine thiols are target of ethyl radicals in model wine conditions (discussed in Chapter 3).

1.3 Importance of Varietal Thiols in Wine

Varietal thiols are extremely important and associated with a range of young wines including white, rosé, and red (68). Varietal thiols are usually present in ng/L levels in wine (69) and contribute to pleasant aromas such as grapefruit, passionfruit, and blackcurrant (68). These compounds are produced during fermentation by the cleavage of odorless precursors in grapes by yeast (69–71). The non-varietal thiols are small sulfurous compounds such as hydrogen sulfide and carbon sulfide. They can be produced at high concentrations during fermentation and are associated with wine defects due to their reductive odors such as sweaty and egg-like odors (72), and as such, they are generally harmful to the quality of wine.

Some of the most studied varietal thiols in wine include 3-mercaptohexan-1-ol

(3MH), 3-mercaptohexyl acetate (3MHA), and 4-mercapto-4-methylpentan-2-one

15

(4MMP) (68,73) as they have very low sensory thresholds. They contribute to the beneficial aromas associated with the varietal wines. For the purposes of this review, we will focus on 3MH.

3MH can be generated by 3 known pathways during fermentation (Scheme 8).

The major precursor to 3MH is S-3-(hexan-1-ol)-cysteine (Cys3MH). This precursor has been identified in Sauvignon Blanc (74), Merlot, and Cabernet Sauvignon grapes (70), and it has been shown that Cys3MH is cleaved by yeast through its beta lyase activity(75). Another precursor, S-3-(hexan-1-ol)-glutathione (G3MH) has been identified in the grapes of Sauvignon Blanc (76). The two mechanisms by which G3MH can be used to form 3MH could be by acting as a precursor for Cys3MH (77) and as a precursor for 3MH directly (78). 3MH can also be formed from hexanal; E-2-hexanal undergoes a sulfur addition during alcoholic fermentation forming 3MH. (71)

Scheme 8: 3-mercaptohexan-1-ol production during fermentation

3MH provides important fruity notes in varietal wines, primarily as passion fruit and grapefruit aromas (69,79,80). 3MH exists as its R- and S- enantiomeric forms, which have sensory thresholds of 50ng/L and 60ng/L, respectively (81). The S- form

16

contributes to passion fruit notes (82), whereas the R- form is responsible for grapefruit notes (83). The enatiomeric distribution of 3MH in dry Sauvignon Blanc is approximately

50:50 (84). Despite the presence of 3MH at low levels, it is almost always above the human detection threshold in young white wines such as sauvignon blanc. 3MH is usually present at higher levels compared to the other varietal thiols, ranging from 26-

18,000 ng/L in sauvignon blanc, whereas 4MMP is present at levels of 4-40 ng/L (85). It is important to note that 3MH and its acetate (3MHA) both contribute to similar aromas and could potentially have synergistic effects, and both may need to be present to give the key aromas associated with sauvignon blanc and rosé wines (15,43)

Since these key aroma compounds are expected to be present in young white wines, their oxidative stability is a major issue. Varietal thiols are labile to oxidation and can be lost within several months depending on storage conditions. This is because thiols are easily oxidized to odorless disulfides under mild oxidative conditions (86,87).

They can also be lost quickly to other oxidation products such as quinones (Scheme 5) and radical species (Scheme 7). The use of cork as a bottle closure can also cause partitioning of these volatile compounds into the cork, known as ‗scalping‘, and could be responsible for loss of fruit aroma in sauvignon blanc (88). It is essential to protect these compounds for a long duration to meet consumer expectations for these fruity wines.

1.4 Preventing oxidation

1.4.1 Bottle closure and environment

Storage conditions play an important role with respect to the oxidative stability of wine. It is important to store wines in the dark as HO• radicals can form via photolysis.

UV-photolysis can also generate superoxide and oxidize ethanol to acetaldehyde

17

(18,27). In addition, photoactivation of the Fe(III) tartrate complex in wine can cause yellowing in wines by the reaction of glyoxylic acid and phenolics (61). Storage temperature and bottle closure are important as temperature may change the rate of oxygen consumption, and varying bottle closures allow varying oxygen ingress levels

(13,14,89). In general, ROTE caps have the lowest oxygen ingress; however, they are usually associated with reductive notes (89,90). Corks that have a higher oxygen ingress rate result in faster oxidation and loss of the fruity aroma, but can also cause problems with loss of varietal thiols as they can partition into the cork (88).

1.4.2 Antioxidants

Sulfur dioxide is the most commonly used antioxidant in wine, and has been discussed previously. However, in a very small part of the population, sulfites can induce allergic reactions. In addition, some consumers are concerned when they see the

―contains sulfites‖ warning on wine labels. Because of this, other alternatives have been considered for use. Other antioxidants have been added into wine include ascorbic acid,

GSH, and caffeic acid (91).

Ascorbic acid has a lower reduction potential than phenolics in wine so it quickly reacts with HOO• radicals (18). However, it produces H2O2 and the reactive oxidation product dehydroascorbic acid which can further degrade and react. It also quickly

- reduces Fe(III) to Fe(II) that can participate in the Fenton reaction (18). HSO3 is therefore essential with the addition of ascorbic acid as it is needed to react with H2O2 and dehydroascorbic acid. There is some evidence that ascorbic acid has a beneficial effect on wine color over time by reducing quniones in the presence of sufficient levels of

- HSO3 (15) (Scheme 5).

18

- GSH is similar to HSO3 as it can react with H2O2 and quinones. It can potentially react with 1-HER as seen in physiological conditions, and recent studies suggest the same in model wine (Chapter 3). There have been studies showing its beneficial effects on preventing oxidation and protecting other varietal thiols (55). GSH has a protective effect preventing the loss of fruity notes in wine, and it has been considered to preserve the aroma of young white wine during storage (68).

Antioxidants work well by either reacting with the precursors or end products of oxidation, but they are useless when it comes to the HO• radicals. Some of these antioxidants actually promote its formation, such as ascorbic acid. Most antioxidants or nucleophiles can in fact promote oxidation by shifting the equilibrium by the removal of quinones.

1.4.3 Metal Chelation

The use of chelators in wine research has been limited. However, in light of recent studies highlighting the essential role that transition metals play in wine oxidation, chelators are of interest due to their ability to disrupt the redox cycling of these metals.

As previously discussed, it has been shown that removing trace metals from wine can effectively stop oxidation processes (35). Iron chelators contain ligands that bind to iron, which can effectively sequester iron and disrupt redox cycling, making it unavailable to interact with other ions or molecules in solution. In wine, this could effectively stop the formation of oxygen radicals (Scheme 1).

Fe(III) ions have small atomic radii and high charge density compared to Fe(II) and as such, behave as hard Lewis acids. Fe(III) ions have a greater affinity for hard bases such as hydroxide, phenolate, and carboxylate ions (92). Fe(II), on the other hand, is considered a borderline Lewis acid. Fe(II) ions preferentially bind to borderline

19

bases containing sulfur, nitrogen, and pyridine ligands (92). Depending on the presence of these specific ligands, the equilibrium can be shifted and result in a different reduction potential of the Fe(III)/Fe(II) couple. Iron is always bound to a ligand in solution and under acidic solutions, Fe(II) hexaaquo-complex is present and reacts slowly with oxygen (18). However, under wine conditions and the presence of the hydroxy acid, tartaric acid, the reduction potential is decreased as Fe(II) is chelated at a 3:1 ratio, and

Fe(II) can be quickly oxidized in wine (18).

Phenolics also bind to Fe(III), however they must be deprotonated in order for this binding to occur. Catechol and gallate groups are effective metal chelators upon deprotonation. At physiological pH, polyphenols easily deprotonate in the presence of iron and form stable complexes (93). This contributes to the antioxidant activity associated with phenolics at physiological pH but is not applicable for wine. The Fe(III) and catechol or gallate ligand binding is less stable at low pH and the polyphenol quickly reduces the Fe(III) to Fe(II) while simultaneously oxidizing the phenol into a semiquinone radical (Scheme 2B) (94,95). The semiquinone radical is then capable of reducing another Fe(III) by being oxidized into a quinone (93). The regenerated Fe(II) can reduce oxygen and H2O2 which results in ROS generation. This is the main reason phenolics containing catechol and gallate groups in wine behave as pro-oxidants.

Synthetic chelators have been used in the food industry as additives to inhibit metal-catalyzed oxidation reactions. One such example is EDTA which has been used to prevent lipid oxidation (96,97). EDTA is a hexadentate ligand and has a strong affinity for

Fe(III), and as such, it quickly shifts the equilibrium of iron in wine to form Fe(III). In the presence of oxygen, Fe(II) is quickly oxidized as it reduces oxygen or other ROS such as H2O2. This quickly promotes the production of HOO• radicals from oxygen and HO• radicals from H2O2. In the presence of EDTA at pH 4, the reduction potential of

20

Fe(III)/Fe(II) couple is reduced to 0.12V which allows this quick reaction (98). EDTA has been shown to have pro-oxidant activity in the presence of phenolics and iron (99,100).

The reaction is highly dependent on the pH of the system. In a lipid system containing trace metals and EGCG it has been shown that, at neutral pH, EDTA rapidly generated

HO• radicals, but at pH 3, it inhibited HO• radical generation (97). As such, the initial addition of EDTA to wine would result in a prooxidant effect due to the high presence of

Fe(II) followed by a quick shift in equilibrium to be oxidized into Fe(III). As redox cycling of iron is not completely inactivated upon binding to EDTA, catechols could potentially reduce Fe(III) while complexed with EDTA by outersphere electron transfer (63,100).

Fe(II) chelators such as 2,2‘-bipyridine stabilize Fe(II) ions, and can therefore prevent oxygen from binding and being reduced to ROS. In a lipid system, it has been shown that 2,2‘-bipyridine increases the concentration of Fe(II) which can generate hydroxyl radicals (97). This is also pH dependent and has been shown to have an antioxidant effect at pH 7 and a prooxidant effect at pH 3. However, the same has not been observed under wine conditions (Chapter 4).

Siderophores are small compounds secreted by microorganisms and grasses.

They contain sidechains of hydroxamate, catecholate, or a mixture of both, and as such they behave as hard base ligands and have a high affinity for Fe(III) (101). Since they have such a high affinity for iron they can successfully inactivate iron redox cycling.

Deferroxamine, for example, is a hexadantate with no free coordination sites that can prevent phenolic oxidation and ROS formation (102).

21

1.5 Purpose and Significance

Domestic wine consumption has increased steadily over the past few years and is now estimated to be 2.54 gallons per capita as of 2010 (103). The United States is the

4th largest wine producing country in the world with a reported production of 2,211,300 tons in 2010 and total sales estimated at 32.5 billion dollars (103). Total worldwide production of wine has been reported to be 26,216,967 tonnes in 2010 (104), and as such it is an extremely important commodity. Due to its economic importance, the loss of wine quality due to non-enzymatic chemical oxidation can lead to major financial losses.

There are currently few ways to control oxidation in wines. It is particularly important to be able to control oxidation in white wines, as there are no benefits of oxidation after bottling. This is especially important for wines containing varietal thiols such as 3MH, as these compounds contribute to the fruity aromas which are quickly lost during oxidation. Different fermentation processes, bottling, and addition of antioxidants have been considered to reduce oxidation, but they do not prevent it over long periods of time. Currently the only completely effective method for inhibiting oxidation reactions is oxygen exclusion, but this can be deleterious as it results in reductive aromas.

Further elucidation of the mechanisms of wine oxidation that lead to both the losses of important aroma compounds as well as the generation of oxidized qualities should be further investigated. Understanding the underlying mechanisms can help further develop tools to prevent wine oxidation in a reproducible manner; tools which winemakers do not currently have.

22

1.6 Hypothesis and Objectives

Wine quality is important and can be attributed to a variety of factors including color, mouthfeel, flavor, and aroma. There has been an increasing interest in understanding the chemistry behind wine in order to preserve its quality, yet few tools are available to control oxidation. This study will investigate key factors of 1-HER formation via metal catalyzed and radical mediated reactions. In addition, the effect of 1-

HER on 3MH and its subsequent loss will be examined. Lastly, the use of metal chelators as a treatment for the prevention of 1-HER formation as well as other oxidation markers will be investigated.

I hypothesize that major wine components will affect the formation of 1-HER. In addition, I hypothesize 1-HER can further react with phenolics and thiols in wine and the use of exogenous metal chelators will inhibit the formation of 1-HER and subsequent loss of 3MH in wine.

I test this hypothesis by investigating:

1. The mechanism for 1-HER formation as affected by major wine components.

2. The mechanisms of phenolic and thiol loss by reaction with 1-HER.

3. The ability of chelators to inhibit oxidation markers and preventing the oxidative

loss of 3MH.

23

Chapter 2: Hydroxyl radical scavenging capability of major wine components

2.1 ABSTRACT

HO• radicals are formed from the metal catalyzed reduction of oxygen, and they react non-selectively. Ethanol is the prime target for HO• radicals in wine due to its abundance; however, other major wine constituents may also act as substrates. In this study, the effect of major wine constituents including glycerol, tartaric acid, and phenolics on 1-HER formation was monitored using spin trapping with electron paramagnetic resonance (EPR). Glycerol was not observed to affect 1-HER formation, which is in contrast with previous reports, and 4-methylcatechol (4-MeC) was seen to promote radical formation. Tartaric acid suggested antioxidant effect by inhibition of 1-

HER formation. However, upon further investigation, H2O2 consumption and acetaldehyde formation suggest either no activity or prooxidant activity depending on the presence or absence of oxygen.

24

2.2 INTRODUCTION

The quality of wine is directly linked to its oxidative stability. Wine that is under high oxidative stress is labile to a variety of oxidative reactions that can be deleterious to its quality, particularly in the case of white wine. The most notable quality changes of white wine include color, such as browning or yellowing, and aroma, such as loss of characteristic aroma or the production of aldehydes and ketones. Both changes are clear signs of oxidation.

In recent years there has been active research concerning the nonenzymatic chemical oxidation of wine, especially with respect to elucidating the role of transition metal catalysts and identifying free radical intermediates. It appears that HO• radicals and quinones are key intermediates to most oxidation related reactions during wine aging (12,18,48). Phenolics, transition metals, and sulfites are all ubiquitous in wine and their concentrations greatly affect the formation of these oxidation intermediates

(30,32,35,36,57). The oxidation product of catechols include electrophilic quinones which can react with a variety of nucleophiles by 2-electron reactions in wine, including sulfites, thiols, selected amino acids, and some phenolics (31,49,51,56). HO• radicals are formed by the reaction of H2O2 and Fe(II), classically referred to as the Fenton reaction (19,105), and can initiate a variety of radical chain reactions. Due to their reactivity, HO• radicals can react with any organic wine component in a diffusion controlled manner (18). It is now known that 1-HER is the major radical species in wine as ethanol (1.3-3.5 M) quickly scavenges hydroxyl radicals due to its abundance (48,57)

(Scheme 1).

Other major wine constituents – and potential substrates for HO• radicals - include glycerol (54-217 mM) and tartaric acid (13-53 mM) (7,58). The proposed major

25

oxidation products of ethanol, tartaric acid, and glycerol are acetaldehyde (24), glyoxylic acid (59), and glyceraldehyde (58), respectively. Scalzo et al. showed that the glycerol can efficiently scavenge HO• radicals from wine extracts (106). However, there have been conflicting reports as to whether tartaric acid is a substrate for HO• radicals under wine conditions (60). While the oxidation products of these constituents have been identified, the radical scavenging properties of these major wine constituents have not been elucidated in wine solutions. The interaction of flavonoids and the resulting aldehydes has been shown to yield a variety of novel phenolic condensation products which affect the color of wine (6,58,107–109). While the presence of ethanol, tartaric acid, and glycerol are too high to be significantly depleted by radical reactions, their intermediate oxidation products and final oxidation products may have varying effects on the final wine composition. Since the concentrations of these components vary widely between different wine styles, it is important to know how each of these compounds can affect oxidation.

In this study, the fate of HO• radicals was measured by following 1-HER formation using a spin trapping EPR spectroscopic technique. The concentration of glycerol, tartaric acid, and phenolics may be sufficiently high to inhibit 1-HER formation.

However, the presence of these compounds may also affect the initial formation of the hydroxyl radicals or final acetaldehyde yield, resulting in either prooxidant or antioxidant effects. The aim of this study was to investigate how the presence of tartaric acid, glycerol, and 4-MeC affect the formation of 1-HER, and to confirm their status as Fenton chemistry targets. In addition, the presence of tartaric acid on the consumption of H2O2 and final acetaldehyde yield was investigated as they are important markers of the oxidative progress.

26

2.3 MATERIALS AND METHODS

Materials. Iron(II) sulfate heptahydrate, 4-methylcatechol, glycerol, xylenol orange tetrasodium salt, and acetaldehyde-DNPH analytical standard were obtained from

Sigma-Aldrich (St. Louis, MO). 2,4-dinitrophenyl-hydrazine (DNPH) was purchased from

MCB laboratory chemicals (Norwood, OH). L-Tartaric acid and D-Sorbitol were obtained from Alfa Aesar (Ward Hill, MA). Hydrogen peroxide (30% w/w; H2O2) was obtained from

EMD Chemicals (Gibbstown, NJ). The spin trap phenyl-N-tert-butyl nitrone (PBN) was purchased from GeroNova Research Inc. (Garson City, NV). Water was purified through a Millipore Q-Plus system (Milipore Corp., Bedford, MA). All other chemicals and solvents were of analytical or HPLC grade.

Model wine preparation. A full factorial experiment (Table 5) was conducted based on the presence or absence of major wine constituents including tartaric acid, glycerol, and

4-MeC. All model solutions contained 12% ethanol and 30mM PBN. Stock solutions of each were made fresh daily in ultrapure water: 20% EtOH containing 50 mM PBN (15%

EtOH with or without 37.5 mM PBN for second set of reactions), 10 mM 4-MeC, 266.5 mM tartaric acid, and 1.3 M glycerol solution. All stock solutions, including pure water were adjusted to pH 3.6 with NaOH and combined immediately prior to the experiment to achieve the final concentration of each treatment as shown in Table 5.

For the second set of reactions, tartaric acid was the focus and so the stock solutions were combined to achieve 0 g/L, 4 g/L, and 8 g/L of tartaric acid in 12% EtOH with or without 30 mM PBN.

27

Table 5: full factorial design used for investigating the effect of major wine constituents on 1-HER formation. Treatment 4-MeC (mM) Tartaric acid (mM) Glycerol (mM) A 0 0 0 B 1 0 0 C 0 53.3 0 D 1 53.3 0 E 0 0 130 F 1 0 130 G 0 53.3 130 H 1 53.3 130

Accelerated model wine oxidation reaction. For accelerated oxidation reactions, 1 mL of each of the given treatments (Table 5) was added to 1.8 mL capacity microcentrifuge tubes and vortexed for 30 sec to achieve air saturation. Fe(II) was added from a 9 mM stock solution (10 μL; 90 μM final concentration) and was allowed to react over 13 hours in the dark. 50 μL aliquots were taken at set time intervals and measured on the EPR.

Fenton reaction conditions. Reactions were carried out under either nitrogen gas or air saturation. For the air-saturated treatments, 1 mL of sample was added to 1.8 mL capacity microcentrifuge tube and vortexed for 30 sec to achieve air saturation. For the deoxygenated treatments, 1 mL of samples was added into 5 mL test tube and purged with nitrogen gas for 3 min. following the aeration or deoxygenation, Fenton reagents

([H2O2] = 300 μM, [Fe(II)] = 90 μM) added to initiate the oxidation reaction. Stock solutions of Fe(II) (9 mM) and H2O2 (30 mM) were prepared daily in acidified water.

Fe(II) (10 μL; 90 μM final concentration) was added added either under nitrogen or air and H2O2 (10 μL, 300 μM final concentration) was then added to initiate the reaction, which was allowed to proceed at ambient temperature for 1 min under nitrogen or air.

Sample aliquots were withdrawn and analyzed immediately by EPR or were further treated for analysis for H2O2 by FOX-1 assay and acetaldehyde by DNPH derivatization.

28

EPR spin-adduct analysis. 1-HER was measured using spin trapping with PBN and the

EPR spectrum observed for the 1-HER/PBN adducts produced a triplet of doublets

(hyperfine coupling constants: aN = 15.7 G, aH = 3.3 G) based on previously reported methods (36,48). Sample aliquots (50 μL) were loaded in 50 μL borosilicate micropipets

(VWR, Radnor, PA, USA) and the EPR spectra were immediately recorded at room temperature on a Bruker eScan R spectrometer (Bruker BioSpin, Rheinstetten,

Germany) operating in X-band. The sweep width was set to 50 G and the microwave power was set to 37.86 mW. The modulation frequency and modulation amplitude were set to 86.00 kHz and 2.45 G, respectively. The receiver gain was set to 4.48 103. The conversion time and sweep time were set to 20.48 msec and 10.49 sec respectively.

The total number of scans per sample was 10. For the deoxygenated samples the total number of scans per sample was 3.

Hydrogen Peroxide Analysis. H2O2 was analyzed according to a modified version of the concentrated ferrous oxidation-xylenol orange (FOX) assay (110). FOX reagent contained xylenol orange (1 mM), ferrous sulfate (2.5 mM), and sorbitol (1.0 M) in sulfuric acid solution (0.5 N), and was prepared weekly from stock solutions. The H2O2 analysis was performed by adding the assay solution (20 μL) to diluted model wine (140

μL). The model wine was diluted at a 1:20 ratio with water to remain in the linear portion of the standard curve. Samples were mixed by vortex and allowed to incubate for 30 min at ambient temperature, and were then measured at 560 nm using Thermo Scientific

Multiskan spectrophotometer (Thermo Scientific, Waltham, MA) blanked against diluted model wine without any added H2O2. Samples were quantified based on an external

29

standard curve prepared with an authentic H2O2 and validated using hydrogen peroxide

–1 –1 extinction coefficient ε240 = 40 M cm . No differences between standard curves were observed in the presence or absence of tartaric acid.

Acetaldehyde analysis. Acetaldehyde was measured in model wine solutions as its

2,4-dinitrophenylhydrazone (DNPH) derivative by HPLC according to previous methods

(33,64). The DNPH solution was prepared by dissolving DNPH (200 mg) in acetonitrile

(100 mL), followed by the acidification with 70 wt% perchloric acid (4 mL). Following the

Fenton reaction, as described above (in the absence of PBN), 40 μL of sulfuric acid (25 wt%), 240 μL of DNPH, and 100 μL aliquot of model wine in a 1.8 mL capacity microcentrifuge tube. The derivatization reaction was carried out at ambient temperature for 3 h, and then diluted with 480 μL of 60:40 acetonitrile:water. The sample was then filtered over a PTFE syringe tip filter (0.45 μm; 13 mm). Chromatographic separation was achieved isocratically using a ZORBAX Eclipse Plus C18 column (4.6 x 150mm, 5

μm; Agilent Technologies) with a mobile phase consisting of 70:30 methanol:water. The acetaldehyde-DNPH derivative was detected using a diode array detector at 365 nm and quantified based on an external standard curve prepared with an authentic acetaldehyde-DNPH analytical standard.

Statistical analysis. Data were analyzed using either two-way ANOVA and Bonferroni correction or one-way ANOVA and Tukey‘s HSD to determine differences from control

(Minitab 16 Statistical Software, State College, PA, USA). Treatments were significant when p < 0.05. All experiments were performed in triplicate.

30

2.4 RESULTS AND DISCUSSION

Effects of major wine constituents on 1-HER formation. Ethanol is the major reactant for oxidation by HO• radicals in wine, and is thought to react in a concentration based manner. However, the presence of other major wine constituents can contribute to the reaction with HO• radicals. Some may behave as antioxidants due to high concentration and quench the HO• radical, whereas others may accelerate HO• generation. Some of the major wine constituents after ethanol include glycerol, sugars, tartaric acid, and phenolics.

For the factorial experiment, the initial compounds were run under accelerated oxidation conditions under air (Table 5) to attempt to take into accounts most of the major wine components that make up white wines. No oxidation progressed in treatment

A which contained only 12% EtOH and 30 mM PBN (Figure 1). Fe(II) can reduce oxygen, however, the oxidation did not progress in the absence of exogenous compounds (i.e. phenolics) that may shift the equilibrium to the product side. In addition, the Fe(II) hexaaquo complex reacts slowly in acidic conditions (111), which further explains why no radical formation was observed over the time frame of the reaction.

31

Figure 1: 1-HER/PBN adduct formation under air. All samples contained 12% EtOH, 90μM Fe(II), and 30mM PBN (specific treatments in Table 1).

As soon as 4-MeC was added in treatment B, there was a clear spin adduct formation in as a little as 1.2 hours, with a linear increase after 2.3 hours (Figure 1). This is expected as Fe(II) can be oxidized to Fe(III) and in the process reduce oxygen to yield

HOO• which can then quickly be scavenged by the catechol to form H2O2 (12,18,48)

(Scheme 1). HO• that is formed by the Fenton reaction quickly reacts with EtOH to yield

1-HER which is trapped by PBN. The removal of HOO• by 4-MeC can allow for quick oxidation of Fe(II) due to the equilibrium shift to the product side which will then be oxidized to Fe(III). Due to the reducing nature of the catechol, Fe(III) can then be quickly reduced to Fe(II). It has been suggested that Fe(III) can be reduced to Fe(II) in a matter of minutes (36).

32

In treatment C, which contained both 4-MeC and tartaric acid, spin adduct formation was observed within 1.2 hours and there was significantly lower spin adduct formation compared to treatment B after 2.3 hours (Figure 1). This suggests that the presence of tartaric acid inhibits the formation of 1-HER, but does not prevent it.

Interestingly, treatment D, which contained only tartaric acid had a similar rate of oxidation and 1-HER formation to treatment C, and by the last time point at 13 hours practically all treatments containing tartaric acid had similar readings (Figure 1). Despite the fact that no phenolics were present to reduce HOO• to H2O2 and subsequent HO•, the HOO• could be reduced to the peroxide in the presence of Fe(II) under these acidic conditions to HO•. The reason this was observed in treatments containing tartaric acid and not in treatment A could be due to the fact that the iron tartrate complex is more favorable than the iron hexaaquo complex (18), and the hydroxy group that binds to iron preferentially binds to Fe(III), allowing for the quick oxidation of Fe(II) once it is bound. In addition, it is possible for the HOO• (E3.6 = 1.22 V) to abstract hydrogen from ethanol

(E3.6 = -1.2 V), as it is thermodynamically favorable, although unlikely in the presence of oxygen (28).

Treatment E, which contained only 12% EtOH, 30 mM PBN and 130 mM glycerol, also showed no signs of 1-HER formation, similar to treatment A (Figure 1).

This suggests that glycerol does not have a significant impact on the rate of oxidation, and this can be further shown by the fact that treatments F, G, and H were practically the same as treatments B, C, and D respectively in terms of spin adduct formation (Figure

1). The presence of glycerol did not seem to change the rate of reaction compared to the comparable treatment without glycerol. This suggests that glycerol, despite its presence at high concentrations in wine, is neither an antioxidant or pro-oxidant.

33

In addition to the accelerated oxidation reaction, the same set of treatments was run under a forced Fenton mechanism (Figure 2). The same trends were observed as expected based on Figure 1 and the forced oxidation model. Treatments A, B, C, and D were not significantly different from treatments E, F, G, and H, respectively. These data suggest once again that glycerol, despite its relatively high concentration, does not outcompete ethanol for HO• to any meaningful degree. In general, the presence of catechol increased 1-HER spin adduct formation compared to the respective treatment in the absence of catechol. This suggests that 4-MeC is pro-oxidative, which is expected based on previous literature (18,31,57). Despite the short reaction time, it is possible for

4-MeC to very quickly reduce Fe(III), which will drive HO• radical formation via the

Fenton reaction from H2O2, as Fe(II) is the limiting reactant in this system.

34

25×105

20

y

t

i

s

n

e

t

n i

15

t

c

u

d

d

a

n i

p 10 S

5

0 A B C D E F G H

Figure 2: 1-HER/PBN spin adduct formation with varying concentrations of tartaric acid, glycerol, and 4-methylcatechol under Fenton conditions (90μM Fe(II) and 300μM H2O2) in aerated model wine (12% EtOH, pH 3.6) and 30mM PBN.

Once tartaric acid was added, 1-HER spin adduct formation was significantly inhibited in all treatments (treatment C, D, G, and H) (Figure 2), suggesting either strong inhibition of HO• formation or scavenging of HO• directly. This was unexpected given that tartrate is capable of complexing Fe(III) to form a relatively stable Fe(III) tartrate complexes, thereby lowering the reduction potential of the Fe(III)/Fe(II) couple; this would be expected to promote the reduction of H2O2 to the reactive HO•. In addition, glycerol did not show any inhibition of the spin adduct formation, despite its presence at twice the concentration of tartaric acid.

35

Tartaric acid was measured directly in an attempt to quantify any significant loss of the acid over the reaction, which would indicate that tartaric acid is oxidized by scavenging HO• directly; however, no differences were observed (data not shown). In fact, it has been stated previously that Fe(III) tartrate complexes are not likely substrates for HO• oxidation (60). Glyoxylic acid is suggested to be a major oxidation product of tartaric acid (112,113), yet previous studies have not been able to detect this product under wine conditions (36). In recent studies, it has only been detected in the presence of light, suggesting that glyoxylic acid is generated in wine by photo-oxidative processes

(61,62).

Effects of tartaric acid on hydrogen peroxide consumption, 1-HER formation, and acetaldehyde formation. In order to get a better understanding of the factors leading to

1-HER/PBN signal inhibition, the effect of tartaric acid concentration was examined more closely on oxidation markers of model wine. There is strong evidence to suggest that tartaric acid behaves as a pro-oxidant in wine by lowering the redox potential of iron and promoting HO• formation. In order to understand whether tartaric acid is a prooxidant or antioxidant, varying concentrations of tartaric acid were examined by a series of induced

Fenton reactions under both aerated and deoxygenated systems.

In the presence of oxygen, 88% and 89% inhibition of 1-HER by 4 g/L and 8 g/L tartaric acid, respectively, compared to the tartrate-free control (Figure 3), which was consistent with the results described above. However, there was no significant difference between 4 g/L and 8 g/L, indicating that this inhibition is not concentration based in the range tested. This effect is likely observed when tartaric acid is in excess to iron, which would always be the case for table wines. When the system was deoxygenated, there

36

was no difference between treatments with respect to spin adduct formation, suggesting that tartaric acid is neither inhibiting nor promoting 1-HER formation. The presence of oxygen clearly seems to affect 1-HER radical formation in the presence of Fe-tartrate complexes.

6×106 Deoxygenated Aerated

5

y t

i 4

s

n

e

t

n

i

t c

u 3

d

d

a

n

i p S 2

1

0 0 g/L 4 g/L 8 g/L

Figure 3: PBN/1-HER EPR spin intensity after induced Fenton reaction in deoxygenated (black bars) and aerated (grey bars) model wine solution at varying tartaric acid concentration. Note: deoxygenated samples were measured at 3 scans whereas aerated samples were measured at 10 scans.

37

Due to the discrepancy within the data and contradicting results compared to the literature, both H2O2 consumption and acetaldehyde formation were measured. It is expected that if tartaric acid is, in fact, inhibiting HO• formation, less H2O2 would be consumed and vice versa. Acetaldehyde, as the major oxidation product of ethanol, with

1-HER being the intermediate, was also measured to determine whether or not tartaric acid competes for HO• under the conditions described.

H2O2 consumption was measured after the initial addition of Fe(II) (90 μM), which is expected to swiftly reduce an equimolar concentration H2O2 (i.e., 90 μM) via Fenton reaction. Interestingly, for the aerated solutions, the presence of both 4 g/L and 8 g/L of tartaric acid did not affect the consumption of H2O2 significantly compared to the tartrate- free control (Figure 4). This suggests that tartaric acid does not inhibit nor affect H2O2 consumption via Fenton reaction in the presence of oxygen.

38

Deoxygenated Aerated

150

)

M

µ

(

e

d

i

x o

r 100

e

P

n

e

g

o

r

d

y H 50

0 0 g/L 4 g/L 8 g/L

Figure 4: Residual H2O2 induced Fenton reaction in deoxygenated (black bars) and aerated (grey bars) model wine solution at varying tartaric acid concentration.

When the solution was deoxygenated, there was a significant increase of H2O2 consumption in the absence of tartaric acid in the deoxygenated solution compared to the aerated solution (Figure 4), this was expected as oxygen can react with 1-HER to form hydroperoxyl ethyl radicals which will subsequently degrade to acetaldehyde (48).

In the absence of oxygen, 1-HER can directly reduce Fe(III) to Fe(II), thereby regenerating Fe(II) to undergo Fenton reaction (Scheme 7). Once tartaric acid was added, there was a significant increase of H2O2 consumption in both 4 g/L and 8 g/L treatments. This is supported by previous literature as the Fe(III) tartrate solution lowers the reduction potential of Fe(II) thereby making it more reducing (18). 1-HER can then

39

quickly regenerate Fe(II) in the absence of oxygen. This suggests that tartaric acid is clearly a pro-oxidant in the absence of oxygen, which is in contrast to the oxygen- saturated system.

The final ethanol oxidation product, acetaldehyde, was measured to establish if tartaric acid directly competes with ethanol for HO•. For the aerated solution, no significant differences in acetaldehyde formation were observed (Figure 5). This supports the observation from the H2O2 consumption, suggesting that the addition of tartaric acid does not affect the rate of radical formation, yet directly contradicts the results obtained from our EPR analysis. In deoxygenated systems, the tartrate-free control had significantly higher acetaldehyde formation compared to the aerated solution

(Figure 5), correlating with the H2O2 consumption (Figure 4). Both 4 g/L and 8 g/L tartaric acid significantly increased acetaldehyde formation, however, the difference between these two treatments was not significant. This, once again, directly contradicts the EPR data.

40

12 Deoxygenated Aerated

10 )

M 8

P

P

(

e

d y

h 6

e

d

l

a

t

e

c A 4

2

0 0 g/L 4 g/L 8 g/L

Figure 5: Acetaldehyde formation after induced Fenton reaction in deoxygenated (black bars) and aerated (white bars) model wine solution at varying tartaric acid concentration.

The data suggests that in the absence of oxygen, tartaric acid behaves as a pro- oxidant by facilitating the Fenton reaction in model wine conditions. In addition, there is no observed difference between the tartaric acid containing (4 g/L and 8 g/L) treatments, thus suggesting that once tartaric is in excess to iron (binds in 3:1 ratio) it shows a similar effect at varying concentrations past saturation (with respect to Fe). In the presence of oxygen, however, slight antioxidant effect by tartaric acid is observable

(although not significant), perhaps due to the Fe(III)-tartrate-oxygen complex, which may prevent the reaction between Fe and H2O2 which is present at relatively low levels.

However, this needs to be investigated further, particularly at intermediate and low

41

concentrations of oxygen that are found in wine conditions. From a practical perspective, care should be taken in conducting EPR studies in wine-like solutions, as tartaric acid affects observed intensities, as discussed above. It is interesting that there was no effect in the absence of oxygen, yet in the presence of oxygen all signals reach similar EPR intensity (Figure 1) further suggesting that tartaric acid may stabilize the signal at a specific level. The effect of tartaric acid needs to be studied further by varying oxygen levels in model wine solutions to further understand the underlying mechanism for oxidation in the presence of both iron and oxygen.

42

Chapter 3: Investigation of Ethyl Radical Quenching by Phenolics and Thiols in

Model Wine

Published as:

Kreitman G.Y., Laurie V.F., Elias R.J. Investigation of Ethyl Radical Quenching by Phenolics and Thiols in Model Wine. Journal of Agricultural and Food Chemistry. 2013 (in press).

3.1 ABSTRACT

In the present study, the reaction between 1-HER and various wine-related phenolics and thiols, including gallic acid, caffeic acid, ferulic acid, 3-mercaptohexan-1-ol

(3MH), cysteine (Cys), and glutathione (GSH) was studied by using competitive spin trapping with electron paramagnetic resonance (EPR) and mass spectrometry. Previous studies have reported several important reactions occurring between quinones and other wine components, but the fate of 1-HER within the context of wine oxidation is less understood. Furthermore, the ability of these compounds to prevent the formation acetaldehyde, a known non-enzymatic oxidation product of ethanol, was measured. The hydroxycinnamic acids and thiol compounds tested at 5 mM concentrations significantly inhibited spin adduct formation, indicating their reactivity towards 1-HER. In addition, we confirm that the loss of 3MH under model wine conditions is due to quinone trapping as well as 1-HER induced oxidation.

43

3.2 INTRODUCTION

Nonenzymatic oxidation greatly affects the stability and, thus, economic value of wines. This is particularly true in the case of white wines, in which oxidation results in browning and a loss of important aroma-active compounds that contribute to desirable sensory attributes (114,115). Varietal thiols are produced during fermentation from the enzymatic cleavage of glutathione and cysteine conjugates by yeast activity (69,75,76).

These thiols are present at exceedingly low concentrations, but are critically important to the sensory attributes of wine by contributing pleasant aromas (e.g., grapefruit, passionfruit, and blackcurrant) (68). However, these same compounds are labile to oxidation and can be rapidly lost in wines (116), especially if bottled under closures with high oxygen transmission rates or if stored under improper conditions (55,88). In recent studies, it has been shown that these thiols, which are strong nucleophiles, are particularly susceptible to loss by reacting with electrophilic quinones by Michael-type addition reactions resulting from enzymatic (54) and nonenzymatic (31,52,53,116) oxidation reactions.

Non-enzymatic wine oxidation is thought to be initiated by the metal-catalyzed reduction of dioxygen by transition metals, particularly iron and copper (12,18,30).

Oxygen is reduced to HOO• radicals in the presence of reduced transition metals

(Scheme 1), and these HOO• radicals are thought to react quickly with phenolics bearing a catechol or gallate group to yield H2O2 and semiquionone radicals (18,49). In the presence of Fe(II) or perhaps Cu(I), H2O2 is then further reduced to yield highly reactive HO• radicals (E3.6 = 2.5 V for HO•/H2O couple) via the Fenton reaction. These

HO• radicals react at diffusion-limited rates and are thought to react non-selectively with wine components (12,18). As the major non-water component in wine, ethanol is known to be a major target for these radicals. The reaction between HO• radicals and ethanol yield ethyl radicals (1-hydroxyethyl radicals and, to a lesser extent, 2-hydroxyethyl

44

radicals), which have been shown to be the major radical species in oxidizing wine

(48,57). While the fate of 1-HER is not fully understood in wine, it is clear that significant amounts of this radical are subsequently oxidized to acetaldehyde (Scheme 9, Reaction

1).

Scheme 9: Proposed reaction mechanisms of 1-HER with: (1) ferric ions or oxygen, (2) thiols, (3), and hydroxycinnamic acids.

In recent years, there has been great interest in the stability and fate of thiols in wines, especially 3MH (11,31,54). While thiol loss resulting from Michael-type addition reactions with benzoquinones has been the focus of nearly all of these studies, the present study considers the possibility that thiol loss in wine is also a result of ethyl radical (e.g., 1-HER) oxidation (Scheme 9, Reaction 2). While the HO• radical is certainly capable of directly oxidizing thiols (117), this reaction is predicted to be of lesser importance in wine given the non-selectivity of the radical and the relative abundance of ethanol (i.e., a more likely substrate for hydroxyl radical oxidation). 1-HER is reported to have a lower reduction potential (E3.6 = 1.2 V for the

45

CH3CH•OH/CH3CH2OH couple (28)) relative to the HO• radical (E3.6 = 2.5 V for HO•/H2O couple) and, as such, should react more selectively with wine components compared to hydroxyl radicals. Gislason et al recently showed that the α,β-unsaturated side-chains of hydroxycinnamic acids can react directly with 1-HER to form allylic alcohols (Scheme 9,

Reaction 3) (64). Similarly, de Almeida et al reported that 1-HER is reactive towards certain isohumulones in beer, resulting in decreased bitterness (65,66). In biological systems, it has been shown that 1-HER is reactive towards GSH, which is oxidized to produce glutathione disulfide (67). In light of recent studies demonstrating the prevalence and relative stability of 1-HER in wine, our objective was to directly assess the reactivity of 1-HER towards several wine-relevant hydroxycinnamic acids and thiols, including Cys, GSH, and 3MH.

46

3.3 MATERIALS AND METHODS

Materials. Iron(II) sulfate heptahydrate, 4-methylcatechol (4-MeC), ferulic acid (FA), D- mannitol, L-cysteine (Cys), acetaldehyde-DNPH analytical standard and catalase from bovine liver were obtained from Sigma-Aldrich (St. Louis, MO). 2,4-dinitrophenyl- hydrazine (DNPH) was purchased from MCB laboratory chemicals (Norwood, OH). L-

Tartaric acid, 3MH, reduced L-glutathione (GSH), 3,4,5-trihydroxybenzoic acid (gallic acid; GA), 3,4-dihydroxycinnamic acid (caffeic acid; CA), and 5,5‘-dithiobis(2- nitrobenzoic acid) (99%+) (DTNB) were obtained from Alfa Aesar (Ward Hill, MA).

Hydrogen peroxide (30% w/w; H2O2) was obtained from EMD Chemicals (Gibbstown,

NJ). The spin trap phenyl-N-tert-butyl nitrone (PBN) was purchased from GeroNova

Research Inc. (Carson City, NV). Water was purified through a Millipore Q-Plus system

(Milipore Corp., Bedford, MA). All other chemicals and solvents were of analytical or

HPLC grade.

Preparation of model wine. Model wine was prepared according to previous methods

(20) by dissolving 8.0 g of tartaric acid in approximately 700 mL of purified water in a 1 L volumetric flask. Absolute ethanol (120 mL) was added and the pH was adjusted to 3.6 using 5 N sodium hydroxide. Sufficient water was added to bring the solution to a final volume of 1 L.

Fenton reaction conditions in model wine. The PBN spin trap was dissolved directly into model wine solutions to achieve a final concentration of 5 mM. GA, CA, FA, Cys,

GSH, or 3MH were also added directly to the model wine solution to achieve final concentrations of 5 mM; these concentrations were chosen in order to establish competitive kinetic conditions between the spin trap and test compounds. A control consisting of model wine containing only 5 mM PBN was used. All reactions were carried

47

out under nitrogen gas. Model wine solutions (1 mL) containing 5 mM of either of the phenolic- or thiol-containing compounds and PBN were transferred to 5 mL test tubes and gently sparged with nitrogen gas via glass dispersion tube at low flow rate for 2 min to achieve deoxygenated conditions. Following deoxygenation, the Fenton reagents

([H2O2] = 1 mM, [Fe(II)] = 100 μM) were added to initiate oxidation reactions according to the method described below. Stock solutions of Fe(II) (10mM), 4-methylcatechol (4-

MeC) (100 mM), and H2O2 (100 mM) were prepared daily in water acidified with HCl (pH

2) and mixed by vortex. 4-MeC (10 μL; 1mM final concentration), was added to approximate the phenolic fraction of a white wine, and Fe(II) (10 μL; 100 μM final concentration) was added to the model wine under nitrogen. H2O2 (10μL, 1 mM final concentration) was then added to initiate the Fenton reaction, which was allowed to proceed at ambient temperature for 1 min under nitrogen. Sample aliquots were withdrawn and analyzed without delay by EPR and LC/MS (methods described below).

For treatments requiring acetaldehyde analysis, experiments were performed as described above except in the absence of PBN. Samples used for acetaldehyde analysis were incubated at ambient temperature for 5 min prior to derivatization

(described below).

EPR analysis of PBN/1-HER spin adducts. EPR was used to detect PBN/1-HER spin adducts in model wine solution, as described previously (36). Sample aliquots (50 μL) were loaded into borosilicate micropipets (VWR, Radnor, PA, USA) and the EPR spectra were immediately recorded at room temperature on a Bruker eScan R spectrometer

(Bruker BioSpin, Rheinstetten, Germany) operating in X-band. The sweep width was set to 50 G and the microwave power was set to 37.86 mW. The modulation frequency and modulation amplitude were set to 86.00 kHz and 2.45 G, respectively. The receiver gain was set to 4.48 103. The conversion time and sweep time were set to 20.48 msec and

48

10.49 sec respectively. The total number of scans per sample was 10. The reaction reached a maximum EPR absorbance within one minute and the signal remained constant for 30 minutes before starting to decay (data not shown). The intensity was quantified by adding the maximum and minimum values of the central doublet.

HPLC-MS analysis of PBN/1-HER spin adducts. Samples were diluted 1:20 in

Millipore water and were filtered over polytetrafluoroethylene (PTFE) syringe tip filters

(0.45 μm, 13 mm; AcrodiscTM, Ann Arbor, MI, USA). The HPLC system consisted of a binary pumping system (Shimadzu LC-10ADvp) with high pressure mixing and sample introduction by means of a Shimadzu SIL 10ADvp autosampler. PBN/1-HER spin adducts were separated on a ZORBAX Eclipse Plus C18 column (4.6 x 150 mm, 5 μm;

Agilent Technologies, Santa Clara, CA, USA). The mobile phase consisted of 0.1% v/v formic acid (A) and 0.1% v/v formic acid in methanol (B). The PBN/1-HER adducts were eluted by gradient according to the following program: 0 min, 0% B; 0-5 min, 80% B; 5-

18 min, 90% B; 18-22 min, 90% B; 22-23 min, 0% B; 23-26 min, 0% B.

Detection and quantification of spin adducts were achieved using a Waters Quattro micro triple quadrupole mass spectrometer (Waters Laboratory Informatics, Milford, MA,

USA) coupled to the HPLC. Mass spectra were collected in negative ion mode using electrospray ionization (ESI). The ESI capillary spray was held at 0.50 kV. The cone source voltage was set to 60 V and the source temperature was set to 120°C. The desolvation gas flow was 250 L/h. Selected ion monitoring mode was set to monitor ions with m/z of 178, 222, 223, 224, and 268, which correspond to unreacted PBN, oxidized

PBN-1-HER adduct, radical PBN-1-HER adduct, reduced PBN-1-HER adduct, and the

PBN-HER bi-adduct, respectively (Figure 6). Due to the complexity of having pure standards for all of these adducts, their quantification was performed based on raw areas.

49

Figure 6: Possible PBN/1-HER adducts formed: (1) oxidized (222 m/z), (2) radical (223 m/z), (3) reduced (224 m/z), and (4 and 5) bi-adducts (268 m/z).

Acetaldehyde analysis. Acetaldehyde was measured in model wine solutions as its

DNPH derivative by HPLC according to previous methods (33,64). Following the Fenton reaction, as described above (in the absence of PBN), 40 μL of sulfuric acid (25%) and

240 μL of DNPH reagent were added to a 100 μL aliquot of model wine in a 1.8 mL capacity microcentrifuge tube. The derivatization reaction was carried out at ambient temperature for 3 h, after which 480 μL of 60:40 acetonitrile:water was added to the sample. The sample was then filtered over a PTFE syringe tip filter (0.45 μm; 13 mm).

Chromatographic separation was achieved isocratically using a ZORBAX Eclipse Plus

C18 column (4.6 x 150mm, 5 μm; Agilent Technologies) with a mobile phase consisting of 70:30 methanol:water. The acetaldehyde-DNPH derivative was detected using a diode array detector at 365 nm and quantified based on an external standard curve prepared with an authentic acetaldehyde-DNPH analytical standard.

Quantification of 3MH loss. In order to study the loss of 3MH by 1-HER, an alternative model wine solution, in which the ethanol was replaced by mannitol, was also prepared

50

alongside a standard model wine (i.e., with ethanol). Mannitol, unlike ethanol, does not produce stable radicals that could react with 3MH. To prepare this solution, mannitol (22 g) was added to a 100 mL solution containing 8 g/L tartaric acid at pH 3.6 with a final mannitol concentration of 1.2 M, despite the fact that this is approximately half the concentration of ethanol in model wine, it still achieves an overwhelming molar excess to

3MH. To either deoxygenated mannitol or deoxygenated ethanol model wine solutions, a final concentration of 100 μM 3MH was established. Fenton reagents and 4-MeC solutions were prepared as described above. The reagents were added to achieve

Fenton conditions with or without 4-MeC; in addition, controls which included H2O2 only or 4-MeC and Fe(II) were added to verify that the reaction is attributed to radical and quinone addition type reactions. Reaction vessels were capped and held at ambient temperature for ten minutes. 3MH was measured in model wine solutions using the

Ellman‘s assay (118). Following the Fenton reaction, 25 μL of catalase (100 IU) was added directly to model wine solutions followed by 100 μL of phosphate/tris buffer (1 M; pH 8.1) to achieve a final pH of 7.0. Catalase was added to prevent further reaction by

H2O2 formed at this pH, as previously documented (33). DTNB (375 μL; 2 mM) in phosphate buffer (100 mM; pH 7.0) was then added and the reaction was allowed to proceed for 10 min. The DTNB derivative was then quantified using a Genesys 10S UV-

Vis spectrophotometer (ThermoScientific, Waltham, MA, USA) at 412 nm. Quantification of 3MH was based on an external standard curve.

Statistical analysis. Data were analyzed using one-way ANOVA and Dunnett‘s posttest or Student‘s t-test to determine differences from control (Minitab 16 Statistical Software,

State College, PA, USA). Treatments were significant when p < 0.05. All experiments were performed in triplicate.

51

Figure 7: Chemical structures of caffeic acid (CA), gallic acid (GA), ferulic acid (FA), 3- mercaptohexan-1-ol (3MH), cysteine (Cys), and glutathione (GSH).

52

3.4 RESULTS AND DISCUSSION

Analysis of PBN/1-HER adducts by EPR. GA, CA, FA, Cys, GSH, and 3MH (Figure 7) were investigated for their ability to quench 1-HER in model wine solutions in which oxidation was initiated by exogenous Fenton reagents (i.e., Fe(II) and H2O2). The HO• radicals formed under these conditions are extremely reactive and are thought to react with the organic fraction of wine in a concentration-dependent manner. Ethanol (~2 M in wine) is present in molar excess compared to other wine components, and is thus predicted to be the major target of HO• radicals, resulting in the generation of hydroxyethyl radicals (36,57). It has been previously shown that the accelerated oxidation conditions (i.e., Fenton reaction) employed in the present study yield the same radicals that result from the unforced oxidation of wine (36,57). Treatment (GA, CA, FA,

Cys, GSH, and 3MH) concentrations (5 mM) used in this study are higher than what is typically present in wine, but were chosen to establish equimolar concentrations with

PBN in order to create competitive kinetic conditions.

53

Figure 8: Representative experimental EPR spectrum corresponding to the PBN/1-HER spin adduct in model wine conditions.

A triplet of doublets was observed in the EPR spectrum of oxidized model wine samples

(Figure 8), with hyperfine coupling constants (aN = 15.5 G, aH = 3.3 G) indicative of

PBN/1-HER adducts (36). As expected, the intensity of PBN/1-HER spin adduct formation for GA was not significantly different from the control (Figure 9). GA was, in essence, used as a negative control as it plays a similar role to 4-MeC which was present in all other treatments. GA was used at 5 mM to verify that the gallate group does not react with either 1-HER or HO• radicals to a significant extent at the concentrations used. As these reactions were run in the absence of oxygen, Fe(II) was not considered to be limiting in the Fenton reaction, as 1-HER quickly reduce Fe(III) under low oxygen conditions (Scheme 9, Reaction 1) (36). If the experiment was

54

conducted in the presence of oxygen, 1-HER would be more likely to react with oxygen to form peroxyl radicals rather than reduce Fe(III) to Fe(II) ions. In the absence of a reducing agent capable of cycling Fe(III) back to its Fe(II) state, it would be expected that the PBN/1-HER spin adduct intensity observed in the GA treatment would be higher than the control, as gallate groups facilitate Fe(III)/Fe(II) redox cycling (18,31).

9×106

8

7

)

y

t

i

s

n e

t 6

n

i

R P

E 5

(

s

t

c u

d 4

d

A

n * i

p 3

S

N * B P 2

* 1 * *

0 Control GA CA FA Cys GSH 3SH

Figure 9: EPR spin adduct intensities of PBN/1-HER in the presence of 4-MeC, PBN, and selected treatments in model wine in the absence of oxygen. Spectra were obtained at room temperature after 1 min following the addition of Fe(II) and H2O2. *Statistically significant difference (P < 0.05) from control by one-way ANOVA with Dunnett‘s post- test.

Significantly lower PBN/1-HER spin adducts were observed in solutions containing hydroxycinnamic acids (FA and CA) and thiol-containing compounds (Cys,

GSH, 3MH) compared to the control model wine (Figure 9). Based on the results seen

55

from the GA treatment, it is unlikely that the concentrations used were sufficiently high to compete with ethanol for HO• radicals, as ethanol is present at ca. 400 molar excess.

Therefore, these results suggest evidence of direct competition between the treatments and PBN for 1-HER.

CA inhibited the production of PBN/1-HER spin adducts by ca. 69% compared to the control, despite the presence of a catechol group and a similar reduction potential to that of GA (18). A decrease in observed spin adducts can be attributed to the high reactivity of CA‘s α,β-unsaturated side-chain group, which has recently been shown by

Gislason et al to efficiently scavenge 1-HER at the α-position (64). The resulting complex is a stable benzyl radical and in the presence of Fe(III), rearranges to its carbocation form and, eventually, into an allylic alcohol (64). The ability of FA to compete with PBN for 1-HER was also investigated, as it contains an α,β-unsaturated side-chain yet, unlike

CA, does not contain a catechol group and therefore does not form an o-benzoquinone upon oxidation. As was the case with CA, FA was observed to scavenge 1-HER radicals, thereby suppressing PBN/1-HER adduct formation by ca. 55% compared to the control.

The thiol-containing treatments inhibited spin adduct formation to the greatest extent, indicating their reactivity towards 1-HER under wine conditions. Cys, GSH, and

3MH supressed the formation of PBN/1-HER spin adducts by 88, 87, and 96%, respectively (Figure 9). It is possible that these compounds showed high reduction in spin adduct formation due to the fact that thiol groups have a relatively low reduction potential and can effectively reduce 1-HER. Reduction potentials do not necessarily predict the reaction rates. For the 2-electron reduction potential, RSH/RSSR couple is a stronger reducing agent than polypenols (119), however, a sulfhydryl group has a reduction potential of approximately -1.16 V for the RSH/RS• couple at wine pH (31), this is lower than the catechol system (E3.6 = -1.0 V for catechol/semiquinone couple). This

56

indicates that upon the formation of a thiyl radical, it is likely quickly scavenged by a catechol which is present in excess to the thiol in wine. We observed that the dimerized form of 3MH was formed when undergoing a Fenton reaction in the absence of 4-MeC, but we did not observe significance in the presence of 4-MeC (Figure 10). Although it is possible for the disulfide to form, as it has been observed in botrytized and aged wines

(120) by reducing 1-HER back to ethanol while simultaneously forming a thiyl radical

(Scheme 2, Reaction 2).

3MH disulfide 4-MeC-3MH adduct 2×106

106

)

U

A

(

a e

r 500000 A

200000

t = 0 t = 30 4-MeC Fenton Fenton+4-MeC

Figure 10: 3MH disulfide and 4-MeC-3MH adduct formation from the following treatments: 3MH only (measured immediately); 3MH only; 3MH + 4-MeC; Fenton reaction (Fe(II) + H2O2); Fenton reaction (Fe(II) + H2O2) + 4-MeC. Reactions were run in model wine (pH 3.6) in the absence of oxygen and measured after 30 min reaction at ambient temperature.

57

Analysis of PBN/1-HER adducts by LC-MS. In order to confirm the EPR spin trapping results reported above, PBN/1-HER adducts were also measured using LC-MS. This was important due to the potential instability of the EPR-active form of the adduct under the conditions employed in the study. Both hydroxylamine and nitrone forms of the spin adducts (i.e., reduced and oxidized radical adducts, respectively) are EPR-silent species due to the loss of their unpaired electrons. Previous studies have demonstrated that under some conditions (e.g., biological systems), reducing agents such as Cys and GSH can result in the reduction of paramagnetic radical spin adducts to their EPR-silent hydroxylamine forms (121). It has also been shown that under beer conditions, α-(4- pyridyl-1-oxide)-N-tert-butylnitrone (POBN)/1-HER adducts, which are structurally analogous to PBN/1-HER, can be oxidized to their nitrone forms or react with a second

1-HER to yield a bi-adduct, both of which are undetectable by EPR (65). Therefore, the nitrone, nitroxide radical, hydroxylamine, and bi-adduct forms of the PBN/1-HER adducts were measured by LC-MS (222 m/z, 223 m/z, 224 m/z, and 268 m/z, respectively) to account for any losses of the EPR-active nitroxide adducts.

58

Figure 11: Total MS spin adduct intensities of all possible PBN/1-HER adducts (oxidized, 222 m/z; radical, 223 m/z; reduced, 224 m/z; bi-adducts, 268 m/z) in the presence of 4-MeC, PBN, and selected treatments in model wine in the absence of oxygen. Spectra were obtained at room temperature after 1 min following the addition of Fe(II) and H2O2. *Statistically significant difference (P < 0.05) from control by one-way ANOVA with Dunnett‘s post-test.

No significant differences between the control and treatments with respect to formation of the nitrone (oxidized form of the adduct, 222 m/z) were observed (Figure

11). This was expected as samples were analyzed immediately following the induction of the Fenton reaction and there was limited opportunity for oxidation of the nitroxide radical adduct. Samples that were allowed to stand for extended periods prior to LC-MS analysis were found to contain higher peak area of bi-adducts and nitrone forms, and lower peak area of hydroxylamine and nitroxide radical forms (data not shown). The peak area for the EPR-active PBN/1-HER adduct (nitroxide radical; 223 m/z) detected in

59

model wine solutions containing the GA treatment was not significantly different from the control; however, the total nitroxide radical peak area was significantly lower for the hydroxycinnamic acid and thiol treatments compared to the control, thus corroborating our EPR spin trapping results. Interestingly, the total peak area value for the reduced

(hydroxylamine) PBN/1-HER adduct (224 m/z) for 3MH was significantly lower than those observed in the control. Previous studies have suggested that the addition of a thiol (e.g., GSH) to a pre-formed PBN/1-HER adduct does not, in fact, affect EPR signal intensity, but would effectively inhibit spin adduct formation if the thiol were present prior to the reaction (67), thereby suggesting competition with PBN for 1-HER. With respect to bi-adduct (268 m/z) formation, the peak area for the GA treatment was significantly higher than the control, yet for the hydroxycinnamic acids (CA, FA), there was no difference from control. Bi-adduct formation was significantly lower in the thiol-containing treatment compared to control.

Acetaldehyde analysis. With respect to acetaldehyde formation (i.e. an alternative way to asses ethanol oxidation), the results were consistent with the above reported EPR and LC-MS analysis of 1-HER. GA did not significantly prevent acetaldehyde formation compared to the control (Figure 12). Significantly less acetaldehyde was observed in samples containing the hydroxycinnamic acid and thiol treatments compared to the control. The overall trend for acetaldehyde formation was similar to 1-HER formation, as measured both by EPR and LC-MS, thus supporting our proposed mechanism.

60

25

20

) M

μ 15

(

e

d

y

h

e

d

l

a t

e 10

c A

* 5 * * * *

0 Control GA CA FA Cys GSH 3SH

Figure 12: Acetaldehyde formation in the presence of 4-MeC and phenolic and thiol treatments in model wine in the absence of oxygen. *Statistically significant difference (P < 0.05) from control by one-way ANOVA with Dunnett‘s post-test.

Contribution of 1-HER to 3MH loss. In order to directly assess the extent to which a test thiol (3MH in this case) is oxidized by 1-HER, wine samples were oxizided in either the presence or absence of 4-MeC. This was done to determine the relative contribution of quinone reactions (via 1,4-Michael addition reactions) versus 1-HER induced oxidation reactions to the overall loss of the model thiol under wine conditions. Mannitol, a sugar alcohol that is a known substrate for hydroxyl radicals (122), was substituted for ethanol in some experiments. Unlike ethanol, mannitol has been shown to be rapidly degraded following its reaction with hydroxyl radicals and, thus, would not be expected to promote thiol (e.g., 3MH) oxidation as 1-HER would.

61

(1) SH S OH OH S OH S CH3CHOH OH S

OH SH (2) OH O S OH HO R O R OH

HO

S HO HO (3) HO S S S H O OH O H O OH

R OH R O R O R OH R OH

Scheme 10: Proposed mechanisms for the loss of 3MH by (1) reaction with 1-HER and subsequent disulfide formation, (2) Michael-type addition reaction to o-quinone, (3) reaction with semiquinone radicals.

When model wine samples were oxidized via the Fenton reaction (same conditions as described above) in the absence of 4-MeC, a 26.6 ± 2.2% loss of 3MH was observed (Figure 13). We propose that, under these conditions, hydroxyl radicals oxidize ethanol to 1-HER, which then proceed to oxidize 3MH. The likely product of this reaction is a 3MH thiyl radical, which is expected to react with a second thiyl radical to yield a disulfide (Scheme 10, Reaction 1). In order to support this hypothesis and to verify that there is no direct loss of 3MH to H2O2, as has been suggested previously (52), a treatment containing H2O2 (1 mM) only (i.e., without added iron) was included. No significant loss of 3MH was observed under these conditions, indicating that in the time frame of the reaction, H2O2 was not capable of directly oxidizing the thiol. In the presence of 4-MeC, which is readily oxidized to quinones under wine conditions (18,28), the total loss of 3MH increased to 75.8 ± 1.8%, indicating that quinones play a major role in the loss of this thiol due to Michael type addition reactions, as has been shown

62

previously (52–55). This loss is likely attributed to the scavenging of 1-HER by 3MH to yield thiyl radicals which can eventually dimerize. In addition, these thiyl radicals may be scavenged by catechols which upon their oxidation may react with 3MH by Michael-type addition reactions. While quinones can in theory be formed by the direct reaction of catechols with hydroxyl radicals, it is more likely that these quinones are formed by metal-catalyzed reactions. The subsequent formation of Fe(III) from the Fenton reaction is likely to quickly yield quinones, Elias and Waterhouse showed that in the presence of

0.68 mM 4-MeC, 50 μM of Fe(III) is completely reduced to Fe(II) within 200 s (36). It is also possible, yet probably less likely, for the thiyl radical to react with a 4-MeC semiquinone radical to form a catechol-thiol adduct (123) (Scheme 10). It seems that

3MH can scavenge 1-HER, yet it is difficult to determine the exact reason for its eventual loss. Further studies should investigate the formation of the disulfide.

63

80 Ethanol Mannitol 70

60

50

s

s

o L

40 *

H

M

3

% 30

20

* 10

0

H2O2 Fenton Fe(II)+4MeC Fenton+4MeC

Figure 13: 3MH loss resulting from the following treatments: H2O2 only; Fenton reaction (Fe(II) + H2O2); Fe(II) + 4MeC; Fenton reaction (Fe(II) + H2O2) + 4MeC. Reactions was run in either model wine (pH 3.6) or mannitol solution (pH 3.6) in the absence of oxygen. *Statistically significant difference (P < 0.05) from model wine by Student‘s t-test.

In model solutions where mannitol was substituted for ethanol, the Fenton reaction (in the absence of 4-MeC) is expected to yield hydroxyl radicals that quickly react with mannitol, which is present in molar excess. However, unlike ethanol, mannitol is not known to form stable radicals upon its reaction with hydroxyl radicals (124).

Consequently, when ethanol was replaced with mannitol, only a 7.3 ± 0.9% loss of 3MH was observed, compared to a 26.6 ± 2.2% loss of 3MH in the presence of ethanol, which was likely due to oxidation by1-HER radicals. It is unlikely that a significant portion of

3MH reacted directly with hydroxyl radicals, as mannitol is present in large molar excess

(ca. 240 x) of 3MH; however, it is possible that 3MH is lost to oxidation by Fe(III) in this

64

system slowly, as was suggested previously (31). As was the case with model wine

(i.e., in the presence of ethanol), H2O2 did not directly oxidize 3MH during the timeframe of the experiment. When the Fenton reaction was carried out in the presence of 4-MeC, there was a 37.5 ± 0.7% loss of 3MH in mannitol-containing solutions, compared to a

75.8 ± 1.8% loss of 3MH in the presence of ethanol (and 1-HER radicals). This loss is likely attributed almost entirely due to Michael-type addition reactions. The hydroxyl radicals will be scavenged by mannitol which will then quickly degrade and will not react further. The subsequent formation of Fe(III) will then cause the quick oxidation of catechol to quinone, which can then react with 3MH to yield the catechol-3MH adduct and effectively cause its loss. A treatment containing 4-MeC and Fe(II) (i.e., no H2O2) was also used as a control, as no reaction would be expected to occur in the absence of oxygen. No significant loss of 3MH was observed over the time frame of this experiment. This suggests, in addition to the fact that the system was deoxygenated, that the conditions of the DTNB derivatization did not result in accelerated oxidation of the catechol, and demonstrates that 3MH is stable in the presence of 4-MeC or Fe(II) alone under our conditions.

This study provides further evidence that selected hydroxycinnamic acids (i.e.,

CA and FA) react directly with ethyl radicals in wine, thus supporting the mechanism recently proposed by Gislason et al. (64). We also demonstrate, for the first time to our knowledge, direct evidence of the reactivity between selected thiol compounds with 1-

HER under wine conditions, which argues for a mechanism other than non-radical quinone adduction as a contributor to thiol loss. However, our results indicate that the loss of thiols, such as 3MH, to quinones by two-electron, Michael-type additions reactions is probably the dominant mechanism in wine due to the abundance of phenolics. In addition, there may be an initial radical reaction by the formation of thiyl

65

radicals, followed by catechol scavenging and Michael-type addition reaction. Future work should be conducted in real wine to confirm these results.

66

Chapter 4: The effect of metal chelators on the oxidative stability of model wine.

4.1 ABSTRACT

Wine oxidation is a major problem with respect to quality, and winemakers have few tools at their disposal to control it. In this study, the effect of exogenous Fe(II)

(bipyridine, Ferrozine) and Fe(III) chelators (EDTA, phytic acid) on non-enzymatic wine oxidation was examined. The ability of these chelators to affect the formation of 1-HER and acetaldehyde was measured using a spin trapping technique with EPR and by

HPLC-PDA, respectively. The chelators were then investigated for their ability to prevent the oxidative loss of an important aroma-active thiol, 3-mercapthexan-1-ol (3MH). The

Fe(II)-specific chelators were more effective antioxidants than the Fe(III) chelators during the early stages of oxidation, and significantly reduced oxidation markers compared to control during the study. However, while the addition of Fe(III) chelators was less effective or even showed an initial prooxidant activity, the Fe(III) chelators proved to be more effective antioxidants compared to Fe(II) chelators over the course of the study. In addition, we show for the first time that Fe(II) and Fe(III) chelators can significantly inhibit the oxidative loss of 3MH in model wine.

67

4.2 INTRODUCTION

The oxidation of wine constituents often results in deleterious quality defects, including browning reactions (49), loss of characteristic aroma compounds (11), and the production of carbonyls associated with undesirable aromas (51,58). The non-enzymatic wine oxidation is thought to be catalyzed by trace quantities of transition metals, in particular iron and copper (18,30). While the direct reaction between molecular dioxygen and the organic fraction of wine is, in fact, thermodynamically favorable, it is kinetically restricted. The reaction between triplet (e.g., dioxygen) and singlet species

(e.g., phenolics, alcohols, acids, etc.) is spin forbidden by Pauli‘s exclusion principle.

Therefore, ground state oxygen must be excited to the singlet state before it can react with other organic molecules in wine. This is possible via a phytochemical excitation

(27), but the most likely mechanism under wine conditions is by the one-electron reduction of oxygen by transition metal catalysts (18).

In the presence of Fe(II), oxygen is reduced by a sequential one electron reduction to yield a superoxide anion radical, which is quickly converted to a HOO• radical under acidic wine conditions (18) followed by the reaction with phenolics and

- formation of H2O2 (Scheme 1). H2O2 is thought to either react quickly with HSO3 (125)

(when present) or react by a relatively slower reaction with reduced transition metals

(36) (e.g., Fe(II) or Cu(I) ions). This latter reaction, known classically as the Fenton reaction yields highly oxidizing HO• radicals (19) (Scheme 1). As extremely reactive species, HO• radicals are capable of reacting at a diffusion-limited rate with organic wine components. Due to the fact that ethanol is the major organic component in wine, it is thus the principal target for these radicals (18,48), which has been shown previously to yield ethyl radicals, with the major radical species being 1-HER (48). 1-HER can be further oxidized to acetaldehyde (36,57) or, perhaps, react with other wine components

68

(64). For example, we have recently demonstrated that wine thiols, such as 3MH are a target of ethyl radicals in wine (see Chapter 3).

In addition to the reactive radical species formed during non-enzymatic wine oxidation, phenolics bearing catechol or gallate groups can be oxidized to quinones (24).

These catechol and gallate moieties can be quickly oxidized by reacting with HOO• or

Fe(III) to form semiquinone radicals; upon the loss of another electron by reaction with a radical or Fe(III), a quinone is formed (18) (Scheme 2B). These quinones are responsible for many defects in wine, such as browning due to polymerization (3,49), or

Michael-type addition reactions, which can lead to the loss of important aroma-active thiols (11,53,54).

Fe(II) and Fe(III) ions represent the two oxidation states of iron in wine. Iron speciation in wine has been measured in various studies, and it has been reported that the majority of free iron is present as Fe(II) (36,37), though a large fraction of iron is bound to tartrate (38) and tannins (39). The reason for the dominance of the ferrous species is likely due to the acidic, reducing environment of wine. In addition, wine phenolics are thought to be able to maintain iron in its reduced state (36). Danilewicz argued that the equilibrium of Fe(II):Fe(III) is important with respect to oxygen consumption and phenolic oxidation (32). In a simple model wine system, it has been shown that oxidation reactions effectively cease once an equilibrium between

Fe(II):Fe(III) is established. However, once other compounds are introduced that are capable of disrupting this equilibrium, such as sulfites and other nucleophiles, the equilibrium is disrupted and allows for oxidation reactions to progress (32). The importance of the Fe(II):Fe(III) ratio has also been demonstrated in lipid peroxidation, yet remains controversial (20,126).

69

Clearly, transition metals are responsible for catalyzing a number of reactions in wine, often leading to many undesirable effects. In order to completely prevent metal- catalyzed oxidation processes, the most effective way would be to completely remove all trace iron and copper from the juice, must, or wine. In previous studies, the progressive removal of transition metals from wine slowed and eventually shut down oxidation reactions (35). However, the complete removal of transition metals is not practical and many of these methods lead to the unintentional stripping of important polyphenolic species (127). An alternative approach would be to bind metals by chelation in a way that inactivates its redox cycling ability.

A chelator‘s ability to disrupt iron‘s redox cycling is key to its efficacy as an inhibitor of oxidation. Wine already contains many small molecules that are able to complex iron and effectively alter its reduction potential (E0). The more positive the E0, the greater the reducing power of iron; the lower the E0, the greater the oxidizing power of iron (28). Tartaric acid, a hydroxy acid which is present at high concentrations in wine, can bind to Fe(II) at an acid-to-metal ratio of 3:1 (128). This effectively lowers the E0 of

Fe(II) (18), making it more reducing and increasing the rate of the reduction of O2 and

H2O2 to HOO• and HO•, respectively. Phenolics containing either a catechol or gallate groups can effectively bind to Fe(III) ions upon their deprotonation. At physiological pH, this forms a relatively stable complex (93) and can retard metal-catalyzed oxidation reactions. However, under the acidic conditions of wine, ligand binding is less stable and appears to result in the polyphenol quickly reducing Fe(III) ions to Fe(II) ions (94,95).

This can effectively restart the metal catalyzed reduction of oxygen.

The potential application of selective metal chelators capable of strongly binding to iron, and inactivating all its binding sites, could effectively mitigate its reactivity. In this work, various chelators with high affinities for Fe(II) and Fe(III) were used to assess

70

their effect on the rate of oxidation in model wine solutions. The rate of oxidation was tested by measuring for 1-HER formation along with acetaldehyde, which are two major markers for wine oxidation. In addition, the ability of chelators to prevent the oxidative loss of 3MH was investigated.

4.3 MATERIALS AND METHODS

Materials. Iron(II) sulfate heptahydrate, 4-methylcatechol (4-MeC), 3-(2-Pyridyl)-5,6- diphenyl-1,2,4-triazine-p,p′-disulfonic acid monosodium salt hydrate (Ferrozine), 2,2′- bipyridyl (BiPy), 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU), and acetaldehyde-DNPH analytical standard were obtained from Sigma-Aldrich (St. Louis, MO, USA). Phytic acid was obtained from TCI America (Portland, OR, USA). (Ethylenedinitrilo)tetracetic acid

(EDTA) was purchased from Mallinckrodt Chemicals (St. Louis, MO, USA). 2,4- dinitrophenyl-hydrazine (DNPH) was purchased from MCB laboratory chemicals

(Norwood, OH, USA). L-Tartaric acid, 3-mercaptohexan-1-ol (3MH), 2,3,4,5,6- pentafluorobenzyl bromide (PFBBr), 3-mercapto-1,2-propanediol (mercaptoglycerol,

90%) were obtained from Alfa Aesar (Ward Hill, MA, USA). Hydrogen peroxide (30% w/w; H2O2) was obtained from EMD Chemicals (Gibbstown, NJ, USA). The spin trap phenyl-N-tert-butyl nitrone (PBN) was purchased from GeroNova Research Inc. (Reno,

NV, USA). Water was purified via a Millipore Q-Plus system (Milipore Corp., Bedford,

MA). All other chemicals and solvents were of analytical or HPLC grade. The wine used in this study was vinified from mechanically-harvested Pinot gris and was generously donated by Mazza Vineyards (2010, North East, PA, USA). The endogenous concentrations of Fe and Cu in the wine were 17.9 μM Fe and 1.73 μM Cu as determined by inductively coupled plasma mass spectrometry. The pH of the wine was

71

3.22. and the total SO2 concentration was 98 mg/L at the start of all experiments, as measured by the aeration/oxidation method (129).

EPR Spin Trapping. PBN (30 mM) was dissolved directly into the model wine solution which contained 12% EtOH and 8.0 g/L tartaric acid with pH adjusted to 3.6 with NaOH

(as described in previous method) and shaken to achieve air saturation. Stock solutions

(20 mM) of each chelator (Ferrozine, BiPy, EDTA, DFO, and Phytic acid) were made in ultrapure water in advance and stored at -80°C until needed. Stock solutions of Fe(II) (9 mM) and 4-methylcatechol (4-MeC) (100 mM) were freshly prepared in water acidified with HCl (pH 2). The model wine (0.955 mL) was added to 1.8 mL capacity microcentrifuge tubes. To that, each given chelator treatment (25 μL; 500 μM final concentration), 4-MeC (10 μL; 1 mM final concentration), and Fe(II) (10 μL; 90 μM final concentration) were added to the model wine. Samples were mixed by vortex, and a 50

μL aliquot was immediately withdrawn for measurement by EPR (as described below).

Samples were stored in the dark at ambient temperature and subsequent aliquots were taken at various time intervals.

The wine‘s pH was adjusted to 3.6 to match that of the model wine system. The sulfur dioxide level was reduced by three consecutive additions of H2O2 (3% v/v) spaced in 20 min intervals under constant agitation and headspace blanketing with nitrogen gas, as described previously (48). The wine was allowed to sit overnight at room temperature and the final total SO2 measured was 30 mg/L. PBN (30 mM) was dissolved directly into the wine and was subsequently saturated with air. The wine (0.967 mL) was then added to 1.8 mL capacity microcentrifuge tubes. To the wine, the given chelator treatment was added (25 μL; 500 μM final concentration) in addition to Fe(II) (8 μL; 90μM final concentration) to achieve the same concentrations used in model wine experiments.

Samples were stored in the dark at ambient temperature for the duration of the study.

72

All samples were saturated with air throughout the experiment, and as such, oxygen was not limiting over the time frame of the reaction. Wine or model wine samples

(50 μL) containing PBN were loaded into 50 μL borosilicate micropipets. The 1-

HER/PBN adduct was quantified and the EPR spectra were recorded on a Bruker eScan

R (Bruker BioSpin, Rheinstetten, Germany) spectrometer operating in X-band at room temperature. The sweep width was set to 50 G and the microwave power was set to

37.86 mW. Modulation frequency and modulation amplitude were set to 86.00 kHz and

2.45 G, respectively. The receiver gain was set to 4.48 103. The conversion time and sweep time were set to 20.48 msec and 10.49 sec respectively. The total number of scans per sample was 10.1-HER adducts produced a triplet of doublets (hyperfine coupling constants: aN = 15.7 G, aH = 3.3 G) as observed in previous studies (36,48).

The intensity was quantified by adding the maximum and minimum values of the central doublet.

Acetaldehyde measurement. Model wine (9.55 mL) was added into a 20 mL capacity headspace vial (23 x 46 mm, 20 mm clear crimp). To the model wine, each chelator treatment (250 μL; 500 μM final concentration), 4-MeC (100 μL; 1mM final concentration), and Fe(II) (100 μL; 90 μM final concentration) were added. The sample was then capped with a red crimp cap with a blue silicone/PTFE septum. The sample was mixed by vortex and 100 μL sample aliquots were withdrawn using a Hamilton 100

μL capacity syringe and dispensed directly into DNPH reagent (procedure described below). Samples were stored in the dark at either ambient temperature or 50°C.

Acetaldehyde was measured in model wine solutions as its 2,4- dinitrophenylhydrazone (DNPH) derivative by HPLC according to previous methods

(33,64). DNPH reagent solution was prepared by dissolving DNPH (200 mg) in acetonitrile (100 mL), followed by the acidification with 70 wt% perchloric acid (4 mL).

73

25% wt% sulfuric acid (40 μL) and DNPH reagent (240 μL) were added to a 100 μL aliquot of model wine in a 1.8 mL capacity microcentrifuge tube. The derivatization reaction was carried out at ambient temperature for 3 h, at which point 60:40 acetonitrile:water (480 μL) was added to the sample. The sample was then filtered over a PTFE syringe tip filter (0.45 μm; 13 mm). Chromatographic separation was achieved isocratically using a ZORBAX Eclipse Plus C18 column (4.6 x 150 mm, 5 μm; Agilent

Technologies, Santa Clara, CA. USA) with a mobile phase consisting of 70:30 methanol:water. The acetaldehyde-DNPH derivative was detected using a diode array detector at 365 nm and quantified based on an external standard curve prepared with an authentic acetaldehyde standard which was reacted with DNPH as described in the procedure.

3MH measurement: The sample preparation was performed in a similar manner to the acetaldehyde measurement (described above) with the following modification: 3MH was added from a freshly made stock solution (100 μg/L) to achieve a final concentration of approximately 1 μg/L. Samples were stored in the dark at 50°C. The whole sample (10 mL) was used per time point, as described below. Extraction and derivatization was performed as described by previous methods (130,131) with the following modifications:

Solid-phase extraction was performed using an Agilent SampliQ 12-position SPE vacuum manifold. The wine sample (10 mL) was passed through a 50 mg Bond Elut-

ENV cartridge conditioned with 1 mL dichloromethane, methanol and water. Phosphate buffer (6 mL; 0.2 M 40% methanol) was then passed followed by water (1 mL), DBU (1 mL; 6.7% in water), and PFBBr (100 μL; 2 g/L in hexane) and allowed to react at ambient temperature for 20 min. Mercaptoglycerol (100 μL; 2 g/L in 6.7% DBU) was then passed through and allowed to react for an additional 20 min. The cartridge was rinsed with phosphate buffer (6 mL; 0.2 M 40% methanol) and water (1 mL). The samples were

74

eluted into 13 x 100mm glass culture tube containing sodium sulfate with hexane/ether solution (1 mL; 1:3 ratio). Samples were mixed by vortex and 600 μL was carefully withdrawn using a 1 mL Hamilton syringe and placed in a 10 mL screw cap vial, they were then evaporated to dryness under nitrogen and sealed with a silver screw cap with silicone/PTFE septa. All the samples and solutions were poured into a reservoir on the top of the cartridge and left to pass through the cartridge with a pressure of 5 in Hg.

The sample was extracted for 30 min at 110 °C with a SPME fiber

(DVB/CAR/PDMS) using GERSTEL MPS2 (Linthicum, MD). The thiol adduct was desorbed from the fiber directly in the GC injector for 10 min using splitless mode.

Helium was used as carrier gas at constant flow 1.2 mL min−1 and injector temperature was set at 250 °C. Chromatographic separation and quantification was achieved using

HP 6890/5972 GC/MSD. The capillary column was a DB-FFAP (30 m x 0.25 mm x 0.25

μm) from Agilent Technologies (Santa Clara, CA). The column oven temperature was as follows: 80 °C for 10 min, then heated to 220 °C at 5 °C min−1. Detection was perfomed by negative electrochemical ionization in selective ion monitoring (SIM) mode. 133,181, and 314 m/z were monitored and 314 m/z was used as the quantification ion as it had the least noise in the spectra.

After each run, the fiber was baked out for 25 min at split mode 1:100. Helium was used as carrier gas at constant flow 1.2 mL min−1 and injector temperature was set at 270 °C. The column oven temperature ramp was as follows: 220 °C for 5 min, then heated to 245 °C at 10 °C min−1 for 17.5 min.

Statistical analysis. Data were analyzed using two-way ANOVA and Bonferroni posttest to determine differences between treatments (Minitab 16 Statistical Software,

75

State College, PA, USA). Treatments were considered statistically significant when p <

0.05. All experiments were performed in triplicates.

4.4 RESULTS AND DISCUSSION

Effect of chelators on 1-HER formation. 1-HER has been shown to be a key radical intermediate of wine oxidation and, as such, was used to monitor the progress of the oxidation in model wine. As discussed previously, the reaction of hydroxyl radicals with ethanol has been demonstrated to yield ethyl radicals in wine. The 1-HER has been shown to be the major radical species in wine (48,57). This radical is sufficiently stable to be trapped using a nitrone spin trap (e.g., PBN) and quantified by measuring the intensity of the EPR spectrum corresponding to the spin adduct (Figure 8). In model wine system, PBN/1-HER adduct formation was observed within the first 4 hours in the chelator-free control (Figure 14).

76

Figure 14: Intensity of PBN/1-HER spin adducts (arbitrary units) in the presence of various iron chelator treatments in model wine (12% EtOH, 8 g/L tartaric acid, pH 3.6) and 4-MeC under air at ambient temperature.

The addition of each chelator was observed to significantly influence the rate of

PBN/1-HER formation. The addition of EDTA resulted in an immediate increase in spin adducts, demonstrating pro-oxidant activity (Figure 14). The presence of the EDTA‘s ligands are known to lower the complexed metal‘s E0 at acidic pH, thereby increasing the reducing power of Fe(II) (98). In addition, while EDTA is capable of binding both Fe(II) and Fe(III) ions, it forms a more stable complex with Fe(III), and results in the rapid oxidation of Fe(II) while, in the process, reducing O2 and H2O2 to HOO• and HO•, respectively (105). Interestingly, phytic acid, which like EDTA preferentially binds to

Fe(III) and binds all coordination site of iron, did not seem to cause a pro-oxidant effect

77

over the time frame of the reaction, and in fact inhibited 1-HER formation compared to control (Figure 14). This observation warrants further study.

The Fe(II) chelators, BiPy and Ferrozine, were able to completely inhibit the formation of 1-HER over the time frame of the reaction (Figure 14). The nitrogen ligands on both Ferrozine and BiPy preferentially bind to Fe(II) and significantly increase the E0 by stabilizing the low oxidation states of metals (132). This, in effect, is thought to stabilize Fe(II) and thus limit its reactivity with O2 and H2O2.

The same reaction was also carried in Pinot gris with reduced SO2 levels. The same general trend that was observed for model wine was observed in this system, although an initial lag phase was seen, likely due to the quenching of H2O2 by residual

SO2 (Figure 15). Interestingly, while the EDTA containing treatment produced the highest spin adduct intensity, indicating a higher rate of radical formation, it produced a slightly longer lag phase as compared to the control. However, unlike the initial results seen in model wine for phytic acid, phytic acid was not significantly different than control

(except for 47.5 hours). As was seen in the above-described experiments in model wine, no evidence of PBN/1-HER radical adduct formation was observed for both BiPy or

Ferrozine treatments over the course of the study (Figure 15).

78

Figure 15: Intensity of PBN/1-HER spin adducts (arbitrary units) in the presence of various iron chelators in pinot gris adjusted to pH 3.6 under air at ambient temperature.

Effect of chelators on acetaldehyde generation. 1-HER is thought to be oxidized to acetaldehyde (Scheme 1), although there is evidence that it can react with other wine components. Acetaldehyde generation rates were monitored to corroborate EPR spin trapping data. This was deemed important as the PBN/1-HER spin adducts are known to degrade over time, and can be reduced or oxidized to an EPR silent form, thus artificially depressing signal intensities. During the first 24 hours of the study, acetaldehyde analysis showed the trend as was observed by EPR spin trapping experiments (Figure

16). EDTA containing treatments showed an immediate prooxidant effect phytic acid did have an elevated formation of acetaldehyde; however, this was not significant. At the

79

last time point at 24 hours, EDTA and phytic acid treatments had 108% and 10.4% more acetaldehyde compared to control, respectively. Although the Fe(II) chelator treatments did not completely inhibit oxidation, as observed by the EPR experiments, they did significantly reduce total acetaldehyde formation compared to control, with BiPy and

Ferrozine treatments inhibiting acetaldehyde by 54.6% and 93.5% at 24 hours, respectively. However, the readings for acetaldehyde were lower than expected, especially compared to the EPR experiments, indicating that the amount of oxidation was minimal.

Figure 16: Acetaldehyde formation over time in the presence of various iron chelator treatments in model wine (12% EtOH, 8 g/L tartaric acid, pH 3.6) and 4-MeC under air at ambient temperature.

80

In order to be representative of a longer wine oxidation process, acetaldehyde formation was measured over a longer time frame at 50°C. While the very initial time points show the same trend as observed in the previous experiments, the roles of Fe(III) and Fe(II) chelators appeared to be reversed. While all chelators significantly inhibited acetaldehyde formation rates compared to the control (Figure 17), treatments containing

BiPy or Ferrozine resulted in significantly higher acetaldehyde formation rates compared to the EDTA and DFO treatments.

Figure 17: Acetaldehyde formation over time in the presence of various iron chelator treatments in model wine (12% EtOH, 8 g/L tartaric acid, pH 3.6) and 4-MeC under air at 50°C.

81

Both BiPy and Ferrozine containing treatments had practically the same effect throughout the experiment. While initially they inhibited acetaldehyde formation the most compared to EDTA and control treatments, by day 5 and 8 they had significantly higher acetaldehyde formation compared to the Fe(III) chelators, but still significantly lower than the control treatment. By day 8 BiPy and Ferrozine containing treatments inhibited acetaldehyde formation by 24.3% and 23.3%, respectively, relative to the control. EDTA treatment showed an initial prooxidant activity, but interestingly, the reaction drastically slowed at the subsequent days and after just one day it was at the same acetaldehyde levels as the rest of the treatments. The phytic acid treatment did not show this same prooxidant activity as EDTA, but after day 1 they had practically the same acetaldehyde readings through the rest of the experiment. By day 8, the Fe(III) chelator proved to be the most effective oxidation prevention treatments, with EDTA and phytic acid treatments inhibiting acetaldehyde formation by 60.8% and 59.1%, respectively, compared to the control treatment (Figure 17).

The data suggest that EDTA, an Fe(III) chelator, quickly oxidizes Fe(II) to Fe(III), thus causing a short-term prooxidant effect, as evidenced by an initial burst of 1-HER.

However, once all iron was bound by EDTA, the complexed metal appears to be relatively stable and inhibits further oxidation reactions. As the redox cycling ability of iron is not completely prevented upon binding to EDTA, catechols could potentially reduce ferric ions while complexed with EDTA by outersphere electron transfer (63,100), albeit at a much slower rate. Phytic acid, on the other hand, did not show a significant prooxidant effect as would be expected as it preferentially binds to Fe(III) and warrants further studies.

The Fe(II) chelator treatments do not completely inactivate iron either, as O2 and H2O2 can still occasionally bind to iron leading to the eventual oxidation of ethanol to

82

acetaldehyde (Scheme 1). The high concentration of tartaric acid and catechol may shift the equilibrium of iron speciation to Fe(III) as they preferably bind to Fe(III) allowing the oxidation of Fe(II) and reduction of O2 and H2O2.

Effect of chelators on oxidative stability of 3MH. For simplicity, two representative chelator treatments were investigated compared to a chelator-free control: BiPy as a

Fe(II) chelator, and EDTA as a Fe(III) chelator (Figure 18). Interestingly, the initial prooxidant activity of EDTA as observed by 1-HER formation and acetaldehyde yield did not translate to 3MH loss. In fact, at no point did the EDTA containing treatment have a lower concentration of 3MH compared to control and BiPy. At day 1, the EDTA containing treatment had significantly higher concentration of 3MH than control and BiPy treatment was not significantly different from control. Up until day 6, EDTA and BiPy treatments were not significantly different from each other, but had significantly higher concentration of 3MH compared to control. By day 10 and 17, the EDTA containing treatment had significantly higher 3MH concentration than the BiPy treatment. By day 17 there was 75% loss of 3MH for control, 46% loss for BiPy, and 34% loss for EDTA

(Figure 18).

83

Figure 18: 3MH loss over time in the presence of various iron chelator treatments in model wine (12% EtOH, 8 g/L tartaric acid, pH 3.6) and 4-MeC under air at 50°C

There are two proposed mechanisms by which these chelators prevent the loss of 3MH. The first mechanism involves preventing the formation of 1-HER, which is capable of directly oxidizing the thiol, followed by subsequent dimerization or potentially scavenging of the thiyl radical by a catechol and forming subsequent semi-quinones and quinones in close proximity to the thiol (Chapter 3). The second mechanism involves preventing the formation of quinones that can react with 3MH via a Michael type addition reaction to form thiol-catechol adducts, effectively causing a loss of 3MH (53,54)

This work focuses primarily on a model wine system containing only the major components of wine (i.e. ethanol, tartaric acid, and phenolics); however there are other

84

components capable of complexing transition metals in wines. While 4-MeC was used at a concentration representative of the phenolic content of white wines, in reality, there are a variety of polyphenolic compounds capable of forming a diverse set of ferric complexes. In addition, the presence of various nucelophiles including sulfites, thiols, and amino acids which were not added in this system can interact with iron and greatly affect the rate of oxidation.

While Ferrozine and BiPy both show inhibition of wine oxidation, the obvious downside is that they are toxic and affect the color of wine. Future work should focus on food safe Fe(II) chelators and various peptides that can be used safely without adversely affecting the quality of wine. While EDTA is food safe, it cannot be used in wine.

However, phytic acid is allowable in some regions and proved to be quite successful without showing the prooxidant activity as observed by Fe(III). Phytic acid warrants further studies in real wine and its effects on oxidative stability of color and aroma.

In the present study, we have shown a clear inhibition of ethyl radical formation in model wine through the use of exogenous metals chelators. This inhibition of radical formation should, theoretically, inhibit the formation quinones as well, however this requires further study. In addition, we show for the first time that chelators can be used as an effective treatment for preventing the loss of 3MH in wine, which may translate well to other varietal thiols. Further work should focus on real wine and, perhaps most importantly, the effect of chelators‘ oxidation inhibition on the sensory attributes of wine.

85

CONCLUDING REMARKS

The influences of several major wine constituents (i.e. glycerol, phenolics, and tartaric acid) were investigated and their effect on the oxidative stability of model wine solutions has been confirmed. Glycerol, despite its high concentration, did not affect the oxidative stability of wine. Phenolics, as expected, can propagate radical reactions under wine conditions. Tartaric acid has shown slight antioxidant effects at aerobic conditions, but when the system was shifted to anaerobic conditions it was clearly a pro-oxidant.

These results need to be further confirmed and investigated at varying levels of oxygen.

In addition, the effect of tartaric acid and its stabilization of the spin adducts by EPR measurement need to be further investigated.

The reactivity between 1-HER and some important wine components, including hydroxycinnamic acids and thiols, has been confirmed. However, further studies should focus on the reaction of thiols with 1-HER to elucidate the exact reaction mechanism.

We speculate possible reaction mechanisms for 3MH with strong proof of its reaction, but the exact reaction that leads to the eventual loss of 3MH still remains in question. If possible, these reactions should be investigated by more sensitive methods (such as

GC-MS) at concentrations more relevant to wine.

Lastly, the use of both Fe(II) and Fe(III) chelators was examined as a strategy for inhibiting oxidation, both in the context of initial oxidation markers and preventing the oxidative loss of important aroma-active compounds (e.g. 3MH). Both chelator treatments used in this study proved to be effective at preventing wine oxidation; although EDTA, an Fe(III) chelator, showed initial prooxidant activity, it proved to be effective over the course of the study. In addition, we were able to show for the first time that these chelators can inhibit the oxidative loss of 3MH. While many factors were

86

investigated, further studies are needed to confirm these results. The oxidative stability of the phenolic fraction should be further investigated to verify that it is not being oxidized. In addition, sensory tests should be conducted to corroborate with the analytical results to verify that wines containing chelators are more acceptable than the chelator-free wines. Lastly, these studies mainly used chelators which are not allowed as food additives, particularly the Fe(II) chelators. In future work, phytic acid, which is an allowable additive should be pursued further and food safe Fe(II) chelators should be investigated.

The results of our work are preliminary and understanding wine oxidation as whole remains a challenging task. This work will hopefully contribute to the wine industry and help yield a tool for winemakers in controlling wine oxidation.

87

REFERENCES

(1) Boulton, R.; Singleton, V.; Bisson, L.; Kunkee, R. Principles and practices of winemaking; 1st ed.; Springer, 1996.

(2) Castellari, M.; Arfelli, G.; Riponi, C.; Amati, A. Evolution of phenolic compounds in red winemaking as affected by must oxygenation. American Journal of Enology and Viticulture 1998, 49, 91–94.

(3) Sioumis, N.; Kallithraka, S.; Makris, D.; Panagiotis, K. Kinetics of browning onset in white wines: influence of principal redox-active polyphenols and impact on the reducing capacity. Food chemistry 2006, 94, 98–104.

(4) Singleton, V.; Kramling, T. Browning of white wines and an accelerated test for browning capacity. American Journal of Enology and Viticulture 1976, 27, 157– 160.

(5) Guyot, S.; Vercauteren, J. Structural determination of colourless and yellow dimers resulting from (+)-catechin coupling catalysed by grape polyphenoloxidase. Phytochemistry 1996, 42, 1279–1288.

(6) Clark, A. C. The production of yellow pigments from (+)-catechin and dihydroxyfumaric acid in a model wine system. European Food Research and Technology 2007, 226, 925–931.

(7) Laurie, V. F.; Talca, U. De; Clark, A. C. Wine oxidation. In Oxidation in foods and beverages and antioxidant applications; 2010; pp. 445–475.

(8) Margalit, Y. Concepts in wine technology; Wine Appreciation Guild: San Francisco, 1997.

(9) Vidal, J.; Dufourcq, T.; Boulet, J.; Moutounet, M. Les apports d‘oxygène au cours des traitements des vins. Bilan des observations sur site, 1ère partie. Revue Francaise d’Oenologie 2001, 190, 24–30.

(10) Castellari, M.; Simonato, B.; Tornielli, G.; Spinelli, P.; Ferrarini, R. Effects of different enological treatments on dissolved oxygen in wines. Italian Journal of Food Science 2004, 16, 387–397.

(11) Blanchard, L.; Darriet, P. Reactivity of 3-mercaptohexanol in red wine: Impact of oxygen, phenolic fractions, and sulfur dioxide. American Journal of Enology and Viticulture 2004, 33, 115–120.

(12) Waterhouse, A. L.; Laurie, V. F. Oxidation of Wine Phenolics : A Critical Evaluation and Hypotheses. American Journal of Enology and Viticulture 2006, 3, 306–313.

(13) Lopes, P.; Saucier, C.; Teissedre, P. Main routes of oxygen ingress through different closures into wine bottles. Journal of agricultural 2007, 55, 5167–5170.

88

(14) Lopes, P.; Saucier, C.; Teissedre, P.-L.; Glories, Y. Impact of storage position on oxygen ingress through different closures into wine bottles. Journal of agricultural and food chemistry 2006, 54, 6741–6.

(15) Skouroumounis, G. K.; Kwiatkowski, M. J.; Francis, I. L.; Oakey, H.; Capone, D. L.; Peng, Z.; Duncan, B.; Sefton, M. A. The influence of ascorbic acid on the composition , colour and flavour properties of a Riesling and a wooded Chardonnay wine during five years storage. Australian Journal of Grape and Wine Research 2005, 355–368.

(16) Wildenradt, H.; Singleton, V. The production of aldehydes as a result of oxidation of polyphenolic compounds and its relation to wine aging. American Journal of Enology and Viticulture 1974, 25.

(17) Bradshaw, M. P.; Prenzler, P. D.; Scollary, G. R. Ascorbic acid-induced browning of (+)-catechin in a model wine system. Journal of agricultural and food chemistry 2001, 49, 934–9.

(18) Danilewicz, J. C. Review of reaction mechanisms of oxygen and proposed intermediate reduction products in wine: Central role of iron and copper. American journal of enology and viticulture 2003, 54, 73–85.

(19) Miller, D. M.; Buettner, G. R.; Aust, S. D. Transition Metals as Catalysts of ―Autoxidation‖ Reactions. Free Radical Biology and Medicine 1990, 8, 95–108.

(20) Welch, K. D.; Davis, T. Z.; Van Eden, M. E.; Aust, S. D. Deleterious iron-mediated oxidation of biomolecules. Free Radical Biology and Medicine 2002, 32, 577–583.

(21) Nevares, I.; Del Alamo, M. Measurement of dissolved oxygen during red wines tank aging with chips and micro-oxygenation. Analytica chimica acta 2008, 621, 68–78.

(22) Nevares, I.; Del Alamo, M.; Cárcel, L. M.; Crespo, R.; Martin, C.; Gallego, L. Measure the Dissolved Oxygen Consumed by Red Wines in Aging Tanks. Food and Bioprocess Technology 2008, 2, 328–336.

(23) Laurie, V. F.; Law, R.; Joslin, W. S.; Waterhouse, A. L. In situ measurements of dissolved oxygen during low-level oxygenation in red wines. American Journal of Enology and Viticulture 2008, 59, 215–219.

(24) Singleton, V. Oxygen with phenols and related reactions in musts, wines, and model systems: Observations and practical implications. American Journal of Enology and Viticulture 1987, 38, 69–77.

(25) Mazauric, J.-P.; Salmon, J.-M. Interactions between yeast lees and wine polyphenols during simulation of wine aging: I. Analysis of remnant polyphenolic compounds in the resulting wines. Journal of agricultural and food chemistry 2005, 53, 5647–53.

89

(26) Silva, a; Lambri, M. Oxygen measures and consumption in must and wine. Analytica Chimica Acta 2006, 563, 391–395.

(27) Bielski, B. H. J.; Gebicki, J. M. Generation of superoxide radicals by photolysis of oxygenated ethanol solutions. Journal of the American Chemical Society 1982, 104, 796–798.

(28) Danilewicz, J. C. Review of Oxidative Processes in Wine and Value of Reduction Potentials in Enology. American Journal of Enology and Viticulture 2011, 63, 1– 10.

(29) Pyrzyńska, K. Analytical Methods for the Determination of Trace Metals in Wine. Critical Reviews in Analytical Chemistry 2004, 34, 69–83.

(30) Danilewicz, J. C. Interaction of Sulfur Dioxide , Polyphenols , and Oxygen in a Wine-Model System : Central Role of Iron and Copper. American Journal of Enology and Viticulture 2007, 1, 53–60.

(31) Danilewicz, J. C. Mechanism of interaction of polyphenols, oxygen, and sulfur dioxide in model wine and wine. American Journal of Enology and Viticulture 2008, 59, 128–136.

(32) Danilewicz, J. C. Mechanism of Autoxidation of Polyphenols and Participation of Sulfite in Wine: Key Role of Iron. American Journal of Enology and Viticulture 2011, 62, 319–328.

(33) Elias, R.; Laurie, V.; Ebeler, S.; Wong, J.; Waterhouse, A. Analysis of selected carbonyl oxidation products in wine by liquid chromatography with diode array detection. Analytica chimica acta 2008, 626, 104–110.

(34) Ribéreau-Gayon Contribution à l‘étude des oxydations et réductions dans les vins, Thèse Doctorat ès Sciences Physiques, Bordeaux University Institute of Enology, 1931.

(35) Danilewicz, J. C.; Wallbridge, P. J. Further studies on the mechanism of interaction of polyphenols , oxygen and sulfite in wine . American Journal of Enology and Viticulture 2010, 2, 166–175.

(36) Elias, R. J.; Waterhouse, A. L. Controlling the fenton reaction in wine. Journal of agricultural and food chemistry 2010, 58, 1699–707.

(37) Tašev, K.; Karadjova, I.; Arpadjan, S.; Cvetković, J.; Stafilov, T. Liquid/liquid extraction and column solid phase extraction procedures for iron species determination in wines. Food Control 2006, 17, 484–488.

(38) Weber, G. Speciation of iron using HPLC with electrochemical and flame-AAS detection. Fresenius’ Journal of Analytical Chemistry 1991, 340, 161–165.

90

(39) Paleologos, E. K.; Giokas, D. L.; Tzouwara-Karayanni, S. M.; Karayannis, M. I. Micelle mediated methodology for the determination of free and bound iron in wines by flame atomic absorption spectrometry. Analytica Chimica Acta 2002, 458, 241–248.

(40) De Beer, D.; Harbertson, J. F.; Kilmartin, P. A.; Roginsky, V.; Barsukova, T.; Adams, D. O.; Waterhouse, A. L. Phenolics: A comparison of diverse analytical methods. American Journal of Enology and Viticulture 2004, 4, 389–400.

(41) Rossi, J. A.; Singleton, V. L. Contributions of grape phenols to oxygen absorption and browning of wines. American Journal of Enology 1966, 231–239.

(42) Vanderhaegen, B.; Neven, H.; Verachtert, H.; Derdelinckx, G. The chemistry of beer aging – a critical review. Food Chemistry 2006, 95, 357–381.

(43) Kilmartin, P. a; Zou, H.; Waterhouse, a L. A cyclic voltammetry method suitable for characterizing antioxidant properties of wine and wine phenolics. Journal of agricultural and food chemistry 2001, 49, 1957–65.

(44) Castellari, M.; Matricardi, L.; Arfelli, G.; Galassi, S. Level of single bioactive phenolics in red wine as a function of the oxygen supplied during storage. Food Chemistry 2000, 69, 61–67.

(45) Barbe, J. C.; De Revel, G.; Joyeux, A.; Lonvaud-Funel, A.; Bertrand, A. Role of carbonyl compounds in SO2 binding phenomena in musts and wines from botrytized grapes. Journal of agricultural and food chemistry 2000, 48, 3413–9.

(46) Ribéreau-Gayon, P.; Glories, Y.; Maujean, A.; Dubourdie, D. Handbook of Enology, The Chemistry of Wine: Stabilization and Treatments; John Wiley & Sons Ltd.: Brisbane, 2006.

(47) Hoffmann, M.; Edwards, J. Kinetics of the oxidation of sulfite by hydrogen peroxide in acidic solution. The Journal of Physical Chemistry 1975, 79, 2096– 2098.

(48) Elias, R. J.; Andersen, M. L.; Skibsted, L. H.; Waterhouse, A. L. Identification of free radical intermediates in oxidized wine using electron paramagnetic resonance spin trapping. Journal of agricultural and food chemistry 2009, 57, 4359–65.

(49) Li, H.; Guo, A. Mechanisms of oxidative browning of wine. Food Chemistry 2008, 108, 1–13.

(50) Barril, C.; Clark, A. C.; Scollary, G. R. Chemistry of ascorbic acid and sulfur dioxide as an antioxidant system relevant to white wine. Analytica Chimica Acta 2011, 1–8.

(51) Culleré, L.; Cacho, J.; Ferreira, V. An assessment of the role played by some oxidation-related aldehydes in wine aroma. Journal of Agricultural and Food Chemistry 2007, 55, 876–881.

91

(52) Nikolantonaki, M.; Chichuc, I.; Teissedre, P.-L.; Darriet, P. Reactivity of volatile thiols with polyphenols in a wine-model medium: impact of oxygen, iron, and sulfur dioxide. Analytica chimica acta 2010, 660, 102–9.

(53) Laurie, V. F.; Zúñiga, M. C.; Carrasco-Sánchez, V.; Santos, L. S.; Cañete, Á.; Olea-Azar, C.; Ugliano, M.; Agosin, E. Reactivity of 3-sulfanyl-1-hexanol and catechol-containing phenolics in vitro. Food Chemistry 2012, 131, 1510–1516.

(54) Nikolantonaki, M.; Jourdes, M.; Shinoda, K.; Teissedre, P.-L.; Quideau, S.; Darriet, P. Identification of adducts between an odoriferous volatile thiol and oxidized grape phenolic compounds: kinetic study of adduct formation under chemical and enzymatic oxidation conditions. Journal of agricultural and food chemistry 2012, 60, 2647–56.

(55) Ugliano, M.; Kwiatkowski, M.; Vidal, S.; Capone, D.; Siebert, T.; Dieval, J.-B.; Aagaard, O.; Waters, E. J. Evolution of 3-mercaptohexanol, hydrogen sulfide, and methyl mercaptan during bottle storage of Sauvignon blanc wines. Effect of glutathione, copper, oxygen exposure, and closure-derived oxygen. Journal of agricultural and food chemistry 2011, 59, 2564–72.

(56) Nikolantonaki, M.; Waterhouse, A. L. A method to quantify quinone reaction rates with wine relevant nucleophiles: a key to understanding oxidative loss of varietal thiols. Journal of agricultural and food chemistry 2012.

(57) Elias, R. J.; Andersen, M. L.; Skibsted, L. H.; Waterhouse, A. L. Key factors affecting radical formation in wine studied by spin trapping and EPR spectroscopy. American Journal of Enology and Viticulture 2009, 60, 471–476.

(58) Laurie, V. F.; Waterhouse, A. L. Oxidation of glycerol in the presence of hydrogen peroxide and iron in model solutions and wine. Potential effects on wine color. Journal of agricultural and food chemistry 2006, 54, 4668–73.

(59) Es-Safi, N.; Guernevé, C. Le; Fulcrand, H.; Cheynier, V.; Moutounet, M. New polyphenolic compounds with xanthylium skeletons formed through reaction between (+)-catechin and glyoxylic acid. Journal of agricultural and food chemistry 1999, 47, 5211–5217.

(60) El-Ghenymy, A.; Garcia-Segura, S.; Rodríguez, R. M.; Brillas, E.; El Begrani, M. S.; Abdelouahid, B. A. Optimization of the electro-Fenton and solar photoelectro- Fenton treatments of sulfanilic acid solutions using a pre-pilot flow plant by response surface methodology. Journal of hazardous materials 2012, 221-222, 288–97.

(61) Clark, A. C.; Dias, D. a; Smith, T. a; Ghiggino, K. P.; Scollary, G. R. Iron(III) tartrate as a potential precursor of light-induced oxidative degradation of white wine: studies in a model wine system. Journal of agricultural and food chemistry 2011, 59, 3575–81.

92

(62) Clark, A. C.; Prenzler, P. D.; Scollary, G. R. Impact of the condition of storage of tartaric acid solutions on the production and stability of glyoxylic acid. Food chemistry 2007, 102, 905–916.

(63) Perron, N. R.; Brumaghim, J. L. A review of the antioxidant mechanisms of polyphenol compounds related to iron binding. Cell biochemistry and biophysics 2009, 53, 75–100.

(64) Gislason, N. E.; Currie, B. L.; Waterhouse, A. L. Novel antioxidant reactions of cinnamates in wine. Journal of agricultural and food chemistry 2011, 59, 6221–6.

(65) De Almeida, N. E. C.; Homem-de-Mello, P.; De Keukeleire, D.; Cardoso, D. R. Reactivity of beer bitter acids toward the 1-hydroxyethyl radical as probed by spin- trapping electron paramagnetic resonance (EPR) and electrospray ionization- tandem mass spectrometry (ESI-MS/MS). Journal of agricultural and food chemistry 2011, 59, 4183–91.

(66) De Almeida, N. E. C.; Nascimento, E. S. P.; Cardoso, D. R. On the Reaction of Lupulones, Hops β-acids, with 1-Hydroxyethyl Radical. Journal of Agricultural and Food Chemistry 2012, 121003070751009.

(67) Stoyanovsky, D.; Wu, D. Interaction of 1-Hydroxyethyl Radical With Glutathione, Ascorbic Acid and α-Tocopherol. Radical Biology and Medicine 1998, 24, 132– 138.

(68) Roland, A.; Schneider, R.; Razungles, A.; Cavelier, F. Varietal thiols in wine: discovery, analysis and applications. Chemical reviews 2011, 111, 7355–76.

(69) Tominaga, T.; Baltenweck-Guyot, R.; Peyrot Des Gachons, C.; Dubourdieu, D. Contribution of Volatile Thiols to the Aromas of White Wines Made From Several Vitis vinifera Grape Varieties. American Journal of Enology and Viticulture 2000, 51, 178–181.

(70) Murat, M. L.; Tominaga, T.; Dubourdieu, D. Assessing the aromatic potential of Cabernet Sauvignon and Merlot musts used to produce rose wine by assaying the cysteinylated precursor of 3-mercaptohexan-1-ol. Journal of agricultural and food chemistry 2001, 49, 5412–7.

(71) Schneider, R.; Charrier, F.; Razungles, A.; Baumes, R. Evidence for an alternative biogenetic pathway leading to 3-mercaptohexanol and 4-mercapto-4- methylpentan-2-one in wines. Analytica Chimica Acta 2006, 563, 58–64.

(72) Mestres, M.; Busto, O.; Guasch, J. Analysis of organic sulfur compounds in wine aroma. Journal of chromatography. A 2000, 881, 569–81.

(73) Cheynier, V.; Schneider, R.; Salmon, J.-M.; Fulcrand, H. 3.26 - Chemistry of Wine. In Comprehensive Natural Products II — Chemistry and Biology; II, H.-W. (Ben) L. B. T.-C. N. P., Ed.; Elsevier: Oxford, 2010; pp. 1119–1172.

93

(74) Peyrot des Gachons, C.; Tominaga, T.; Dubourdieu, D. Measuring the Aromatic Potential of Vitis vinifera L. Cv. Sauvignon Blanc Grapes by Assaying S-Cysteine Conjugates, Precursors of the Volatile Thiols Responsible for Their Varietal Aroma. Journal of Agricultural and Food Chemistry 2000, 48, 3387–3391.

(75) Tominaga, T.; Peyrot des Gachons, C.; Dubourdieu, D. A New Type of Flavor Precursors in Vitis vinifera L. cv. Sauvignon Blanc: S-Cysteine Conjugates. Journal of agricultural and food chemistry 1998, 46, 5215–5219.

(76) Peyrot des Gachons, C.; Tominaga, T.; Dubourdieu, D. Sulfur Aroma Precursor Present in S-glutathione Conjugate Form: Identification of S-3-(Hexan-1-ol)- glutathione in Must from Vitis vinifera L. cv. Sauvignon Blanc. Journal of Agricultural and Food Chemistry 2002, 50, 4076–4079.

(77) Thibon, C.; Cluzet, S.; Mérillon, J. M.; Darriet, P.; Dubourdieu, D. 3- Sulfanylhexanol Precursor Biogenesis in Grapevine Cells: The Stimulating Effect of Botrytis cinerea. Journal of Agricultural and Food Chemistry 2011, 59, 1344– 1351.

(78) Capone, D. L.; Pardon, K. H.; Cordente, A. G.; Jeffery, D. W. Identification and Quantitation of 3-S-Cysteinylglycinehexan-1-ol (Cysgly-3-MH) in Sauvignon blanc Grape Juice by HPLC-MS/MS. Journal of Agricultural and Food Chemistry 2011, 59, 11204–11210.

(79) Tominaga, T.; Furrer, A.; Henry, R.; Dubourdieu, D. Identification of new volatile thiols in the aroma of Vitis vinifera L. var. Sauvignon blanc wines. Flavour and fragrance journal 1998, 13, 159–162.

(80) Ferreira, V.; Ortín, N.; Escudero, A.; López, R.; Cacho, J. Chemical Characterization of the Aroma of Grenache Rosé Wines: Aroma Extract Dilution Analysis, Quantitative Determination, and Sensory Reconstitution Studies. Journal of Agricultural and Food Chemistry 2002, 50, 4048–4054.

(81) Tominaga, T.; Niclass, Y.; Frérot, E.; Dubourdieu, D. Stereoisomeric Distribution of 3-Mercaptohexan-1-ol and 3-Mercaptohexyl Acetate in Dry and Sweet White Wines Made from Vitis vinifera (Var. Sauvignon Blanc and Semillon). Journal of Agricultural and Food Chemistry 2006, 54, 7251–7255.

(82) Weber, B.; Maas, B.; Mosandl, A. Stereoisomeric Flavor Compounds. 72. Stereoisomeric Distribution of Some Chiral Sulfur-Containing Trace Components. Journal of Agricultural and Food Chemistry 1995, 2438–2441.

(83) Tominaga, T.; Niclass, Y.; Frérot, E.; Dubourdieu, D. Stereoisomeric distribution of 3-mercaptohexan-1-ol and 3-mercaptohexyl acetate in dry and sweet white wines made from Vitis vinifera (Var. Sauvignon Blanc and Semillon). Journal of agricultural and food chemistry 2006, 54, 7251–5.

(84) Peña-Gallego, A.; Hernández-Orte, P.; Cacho, J.; Ferreira, V. S-Cysteinylated and S-glutathionylated thiol precursors in grapes. A review. Food Chemistry 2012, 131, 1–13.

94

(85) Coetzee, C.; Du Toit, W. J. A comprehensive review on Sauvignon blanc aroma with a focus on certain positive volatile thiols. Food Research International 2012, 45, 287–298.

(86) Sarrazin, E.; Shinkaruk, S.; Pons, M.; Thibon, C.; Bennetau, B.; Darriet, P. Elucidation of the 1,3-Sulfanylalcohol Oxidation Mechanism: An Unusual Identification of the Disulfide of 3-Sulfanylhexanol in Sauternes Botrytized Wines. Journal of Agricultural and Food Chemistry 2010, 58, 10606–10613.

(87) Hofmann, T.; Schieberle, P.; Grosch, W. Model Studies on the Oxidative Stability of Odor-Active Thiols Occurring in Food Flavors. Journal of Agricultural and Food Chemistry 1996, 44, 251–255.

(88) Brajkovich, M.; Tibbits, N.; Peron, G.; Lund, C. M.; Dykes, S. I.; Kilmartin, P. A.; Nicolau, L. Effect of Screwcap and Cork Closures on SO2 Levels and Aromas in a Sauvignon Blanc Wine. Journal of Agricultural and Food Chemistry 2005, 53, 10006–10011.

(89) Skouroumounis, G.; Kwiatkowski, M. The impact of closure type and storage conditions on the composition, colour and flavour properties of a Riesling and a wooded Chardonnay wine during five years‘. Australian Journal of Grape and Wine Research 2005.

(90) Godden, P.; Francis, L.; Field, J.; Gishen, M.; Coulter, A.; Valente, P.; HØj, P.; Robinson, E. Wine bottle closures: physical characteristics and effect on composition and sensory properties of a Semillon wine 1. Performance up to 20 months post‐bottling. Australian Journal of Grape and Wine Research 2001, 64– 105.

(91) Roussis, I. G.; Lambropoulos, I.; Tzimas, P. Protection of volatiles in a wine with low sulfur dioxide by caffeic acid or glutathione. American Journal of Enology and Viticulture 2007, 2, 274–278.

(92) Pearson, R. G. Hard and soft acids and bases. Journal of the American Chemical Society 1963, 85, 3533–3539.

(93) Hider, R. C.; Liu, Z. D.; Khodr, H. H. Metal chelation of polyphenols. Methods in enzymology 2001, 335, 190–203.

(94) Hynes, M. J.; Coinceanainn, M. O. The kinetics and mechanisms of the reaction of iron(III) with gallic acid, gallic acid methyl ester and catechin. Journal of inorganic biochemistry 2001, 85, 131–42.

(95) Ryan, P.; Hynes, M. J. The kinetics and mechanisms of the reactions of iron(III) with quercetin and morin. Journal of inorganic biochemistry 2008, 102, 127–36.

(96) Hu, M.; McClements, D. J.; Decker, E. A. Impact of chelators on the oxidative stability of whey protein isolate-stabilized oil-in-water emulsions containing ω-3 fatty acids. Food chemistry 2004, 88, 57–62.

95

(97) Zhou, L.; Elias, R. J. Factors influencing the antioxidant and pro-oxidant activity of polyphenols in oil-in-water emulsions. Journal of agricultural and food chemistry 2012.

(98) Ilan, Y. A.; Czapski, G. The reaction of superoxide radical with iron complexes of EDTA studied by pulse radiolysis. Biochimica et biophysica acta 1977, 498, 386– 394.

(99) Laughton, M. J.; Halliwell, B.; Evans, P. J.; Robin, J.; Hoult, S. Antioxidant and pro-oxidant actions of the plant phenolics quercetin, gossypol and myricetin: Effects on lipid peroxidation, hydroxyl radical generation and bleomycin- dependent damage to DNA. Biochemical Pharmacology 1989, 38, 2859–2865.

(100) Puppo, A. Effect of flavonoids on hydroxyl radical formation by Fenton-type reactions; influence of the iron chelator. Phytochemistry 1992, 31, 85–88.

(101) Neilands, J. B. Siderophores: structure and function of microbial iron transport compounds. The Journal of biological chemistry 1995, 270, 26723–6.

(102) Mahoney, J. R.; Graf, E. Role of Alpha-Tocopherol, Ascorbic Acid, Citric Acid and EDTA as Oxidants in Model Systems. Journal of food science 1986, 51, 1293– 1296.

(103) Wine Institute - Statistics. Gomberg, Fredrikson & Associates 2012.

(104) Wine Product - FAOSTAT. Food and Agriculture Organization 2011.

(105) Welch, K. D.; Davis, T. Z.; Aust, S. D. Iron autoxidation and free radical generation: Effects of buffers, ligands, and chelators. Archives of Biochemistry and Biophysics 2002, 397, 360–369.

(106) Scalzo, R. Lo; Morassut, M.; Rapisarda, P. Oxygen radical scavenging capacity of phenolic and non-phenolic compounds in red and white wines. Central European Journal of Biology 2011, 7, 146–158.

(107) Bakker, J.; Timberlake, C. F. Isolation, identification, and characterization of new color-stable anthocyanins occurring in some red wines. J. Agric. Food Chem. 1997, 45, 35–43.

(108) Saucier, C.; Guerra, C.; Pianet, I.; Laguerre, M.; Glories, Y. (+)-Catechin— acetaldehyde condensation products in relation to wine-ageing. Phytochemistry 1997, 46, 229–234.

(109) Es-Safi, N.; Guernevé, C. Le; Cheynier, V.; Moutounet, M. New phenolic compounds formed by evolution of (+)-catechin and glyoxylic acid in hydroalcoholic solution and their implication in color changes of grape-derived foods. Journal of Agricultural and Food Chemistry 2000, 48, 4233–4240.

96

(110) Wolff, S. P. Ferrous ion oxidation in presence of ferric ion indicator xylenol orange for measurement of hydroperoxides. Methods in Enzymology 1994, 233, 182– 189.

(111) Lamb, A. B.; Elder, L. W. The Electromotive Activation of Oxygen. Journal of the American Chemical Society 1931, 53, 137–163.

(112) Es-Safi, N.; Guernevé, C. Le; Fulcrand, H.; Cheynier, V.; Moutounet, M. New polyphenolic compounds with xanthylium skeletons formed through reaction between (+)-catechin and glyoxylic acid. Journal of agricultural and food chemistry 1999, 47, 5211–5217.

(113) Fulcrand, H.; Cheynier, V.; Oszmianski, J.; Moutounet, M. An oxidized tartaric acid residue as a new bridge potentially competing with acetaldehyde in flavan-3- ol condensation. Phytochemistry 1997, 46, 223–227.

(114) Escudero, A.; Asensio, E.; Cacho, J.; Ferreira, V. Sensory and chemical changes of young white wines stored under oxygen. An assessment of the role played by aldehydes and some other important odorants. Food chemistry 2002, 77, 325– 331.

(115) Oliveira, C. M.; Silva Ferreira, A. C.; Guedes De Pinho, P.; Hogg, T. A. Development of a potentiometric method to measure the resistance to oxidation of white wines and the antioxidant power of their constituents. Journal of agricultural and food chemistry 2002, 50, 2121–4.

(116) Nikolantonaki, M.; Waterhouse, A. A method to quantify quinone reaction rates with wine relevant nucleophiles: a key to understanding oxidative loss of varietal thiols. Journal of Agricultural and Food Chemistry 2012, 60.

(117) Blank, I.; Pascual, E. C.; Devaud, S.; Fay, L. B.; Stadler, R. H.; Yeretzian, C.; Goodman, B. A. Degradation of the coffee flavor compound furfuryl mercaptan in model Fenton-type reaction systems. Journal of agricultural and food chemistry 2002, 50, 2356–64.

(118) Robyt, J. F.; White, B. J. Biochemical techniques: theory and practice; Waveland Press, 1990.

(119) Makhotkina, O.; Kilmartin, P. a. Uncovering the influence of antioxidants on polyphenol oxidation in wines using an electrochemical method: Cyclic voltammetry. Journal of Electroanalytical Chemistry 2009, 633, 165–174.

(120) Sarrazin, E.; Shinkaruk, S.; Pons, M.; Thibon, C.; Bennetau, B.; Darriet, P. Elucidation of the 1,3-sulfanylalcohol oxidation mechanism: an unusual identification of the disulfide of 3-sulfanylhexanol in sauternes botrytized wines. Journal of agricultural and food chemistry 2010, 58, 10606–13.

(121) Yue Qian, S.; Kadiiska, M. B.; Guo, Q.; Mason, R. P. A novel protocol to identify and quantify all spin trapped free radicals from in vitro/in vivo interaction of HO•

97

and DMSO: LC/ESR, LC/MS, and dual spin trapping combinations. Free radical biology & medicine 2005, 38, 125–35.

(122) Smirnoff, N.; Cumbes, Q. J. Hydroxyl Radical Scavenging Activity of Compatible Solutes. Phytochemistry 1989, 28, 1057–1060.

(123) Fujimoto, A.; Masuda, T. Chemical Interaction between Polyphenols and a Cysteinyl Thiol under Radical Oxidation Conditions. Journal of Agricultural and Food Chemistry 2012.

(124) Shen, B.; Jensen, R. G.; Bohnert, H. J. Mannitol Protects against Oxidation by Hydroxyl Radicals. Plant physiology 1997, 115, 527–532.

(125) McArdle, J. V; Hoffmann, M. R. Kinetics and mechanism of the oxidation of aquated sulfur dioxide by hydrogen peroxide at low pH. The Journal of Physical Chemistry 1983, 87, 5425–5429.

(126) Minotti, G.; Aust, S. D. The requirement for iron (III) in the initiation of lipid peroxidation by iron (II) and hydrogen peroxide. The Journal of biological chemistry 1987, 262, 1098–104.

(127) Ben t ez, P.; Castro, R.; Barroso, C. G. Removal of iron , copper and manganese from white wines through ion exchange techniques : effects on their organoleptic characteristics and susceptibility to browning. Analytica Chimica Acta 2002, 458, 197–202.

(128) Yokoi, H.; Mitani, T.; Mori, Y.; Kawata, S. Complex formation between iron (III) and tartaric and citric acids in a wide pH range 1 to 13 as studied by magnetic susceptibility measurements. Chemistry Letters 1994.

(129) Iland, P.; Bruer, N.; Edwards, G.; Weeks, S.; Wilkes, E. Chemical analysis of grapes and wine: techniques and concepts; Patrick Iland Wine Promotions, 2004; pp. 56–59.

(130) Mateo-Vivaracho, L.; Cacho, J.; Ferreira, V. Improved solid-phase extraction procedure for the isolation and in-sorbent pentafluorobenzyl alkylation of polyfunctional mercaptans. Optimized procedure and analytical applications. Journal of chromatography. A 2008, 1185, 9–18.

(131) Rodríguez-Bencomo, J. J.; Schneider, R.; Lepoutre, J. P.; Rigou, P. Improved method to quantitatively determine powerful odorant volatile thiols in wine by headspace solid-phase microextraction after derivatization. Journal of chromatography. A 2009, 1216, 5640–6.

(132) Schilt, A. A. Formal Oxidation-Reduction Potentials and Indicator Characteristics of Some Cyanide and 2,2‗-Bipyridine Complexes of Iron, Ruthenium, and Osmium. Analytical Chemistry 1963, 35, 1599–1602.

98