SMOKE TAINT: Impacts on the Chemical and Microbiological Profile of Grapes and Wine. Kerry Anita Pinchbeck. B.Sc., Flinders University

SMOKE TAINT: Impacts on the Chemical and Microbiological Profile of Grapes and Wine by Kerry Anita Pinchbeck B.Sc., Flinders University B.Sc. (Hons) The University of Adelaide A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy The University of Adelaide School of Agriculture, Food and Wine April, 2011

TABLE OF CONTENTS ABSTRACT... DECLARATION... STATEMENT OF THE CONTRIBUTIONS OF JOINTLY AUTHORED PAPERS... ACKNOWLEDGEMENTS... i iii iv viii CHAPTER 1: INTRODUCTION... 1 1.1 Introduction to the Australian wine industry... 2 1.2 History of vineyard exposure to smoke... 3 1.3 The effects of smoke on nature... 8 1.4 Composition of smoke... 10 1.5 Smoke taint in grapes and wine... 13 1.6 Research aims... 19 CHAPTER 2: SYNTHESIS OF GUAIACOL GLUCOSIDES... 21 2.1 Introduction to glycosides in grapes and wine... 22 2.2 Introduction to guaiacol β-d-glucopyranoside... 24 2.3 Introduction to glycosylation... 25 2.4 Glycosylation of guaiacol... 26 2.5 Results and discussion... 27 2.5.1 Preparation of guaiacol β-d-glucopyranoside... 27 2.2.5.1 Preparation of guaiacol glucoside using 2,3,4,6-tetra-O-acetyl-α-D-glucopyranosyl bromide (Method 1)... 27 2.5.1.2 Preparation of guaiacol glucoside using 2,3,4,6-tetra-O-pivaloyl-α-D-glucopyranosyl bromide. (Method 2)... 27 2.5.2 Preparation of deuterated guaiacol β-d-glucopyranoside... 28

2.6 Materials and methods... 30 2.6.1 Solvents and reagents... 30 2.6.2 Chromatography... 30 2.6.3 Nuclear magnetic resonance (NMR) spectroscopy. 30 2.6.4 Ultra violet/visible spectroscopy and fluorescence spectroscopy... 31 2.6.5 Microwave synthesis... 31 2.6.6 Gas chromatography-mass spectrometry (GC-MS) 31 2.6.7 High performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS)... 32 2.6.8 Synthesis... 33 2.7 Conclusion... 42 CHAPTER 3: PROVENENCE OF GUAIACOL GLUCOSIDE IN SMOKE AFFECTED FRUIT..... 43 Paper 1: Identification of a β-d-glucopyranoside precursor to guaiacol in grape juice following grapevine exposure to smoke. 45 CHAPTER 4: QUANTIFICATION OF GUAIACOL GLYCOSIDES IN SMOKE AFFECTED FRUIT. 51 Paper 2: Quantitative analysis of glycoconjugate precursors of guaiacol in smoke-affected grapes using liquid chromatography-tandem mass spectrometry based stable isotope dilution analysis. 53 CHAPTER 5: QUANTIFICATION OF GUAIACOL GLYCOCONJUGATES IN GRAPES AND WINE... 59 5.1. Introduction... 60 5.2. Results and discussion... 61 5.2.1 Method development... 61 5.2.1.1Calibration function for guaiacol β-d-glucopyranoside... 61 5.2.1.2 Mass transitions used for HPLC-SRM analysis... 62

5.2.2 Method validation... 63 5.2.2.1Instrument repeatability... 63 5.2.2.2 Reproducibility... 63 5.2.2.3 Recovery... 63 5.2.3 Application of wine based SIDA method to winemaking trials... 66 5.2.3.1Hydrolysis of guaiacol glycoconjugates during fermentation... 66 5.2.3.2 Influence of winemaking techniques on the glycoconjugate content of wine... 69 5.2.3.3 Glycoconjugate content of wine and potential for smoke taint to intensify with bottle age... 72 5.2.3.4 Potential for carryover of glycoconjugates between growing seasons... 74 5.3 Materials and methods... 76 5.3.1 Method development... 76 5.3.1.1 Preparation of wine samples for HPLC-MS/MS analysis... 76 5.3.1.2 Calibration function for guaiacol β-d-glucopyranoside... 77 5.3.1.3 Instrumental analysis... 77 5.3.1.4 High performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS)... 77 5.3.2 Application of the quantitative guaiacol glycoconjugate method to winemaking trials... 79 5.3.2.1Smoke affected grapes... 79 5.3.2.2 Winemaking... 80 5.3.3 Statistical analysis... 81 5.4. Conclusion... 82 CHAPTER 6: THE EFFECT OF WINEMAKING TECHNIQUES ON THE INTENSITY OF SMOKE TAINT IN WINE... 83

Paper 3: The effect of winemaking techniques on the intensity of smoke taint in wine... 85 CHAPTER 7: IMPACT OF SMOKE ON GRAPE BERRY MICROFLORA AND YEAST FERMENTATION... 97 Paper 4: Impact of smoke on grape berry microflora and yeast fermentation... 99 CHAPTER 8: SUMMARY... 104 APPENDIX... 109 REFERENCES... 117

ABSTRACT Guaiacol β-d-glucopyranoside was prepared via a modified Koenigs-Knorr glycosylation method as a reference compound to confirm its presence in smoke affected grapes, using high performance liquid chromatography-tandem mass spectrometry (HPLC-MS/MS) analysis. The β-d-glucopyranoside of guaiacol was identified in extracts of Sangiovese grapes exposed to bushfire smoke and Chardonnay grapes exposed to smoke under experimental conditions. Only trace levels of the glucoside were identified in the corresponding control (i.e. unsmoked) Chardonnay grapes, indicating glycosylation of smoke-derived guaiacol occurred in response to smoke exposure. The reference compound and the glucoside present in smoked juice samples remained largely unaffected following strong acid hydrolysis but were highly susceptible to β-glucosidase enzyme hydrolysis, providing a plausible explanation for the release of guaiacol during fermentation of smoke affected grapes. Following the identification of additional guaiacol glycoconjugate precursors, the d 4 -labelled analogue of guaiacol β-d-glucopyranoside was synthesised for use as an internal standard in the development of a quantitative stable isotope dilution analysis (SIDA) method, using HPLC-MS/MS. This method was subsequently applied to the analysis of several grape varieties exposed to either experimental or bushfire smoke, to investigate the accumulation of guaiacol glycoconjugates following grapevine smoke exposure. Experimentally smoked grapes contained glycoconjugate concentrations up to 300 µg/kg; whereas grapes affected by bushfire smoke contained up to 2,000 µg/kg glycoconjugates, attributed to the different durations of smoke exposure. Analysis of separated berry components indicated that the majority of guaiacol glycoconjugates were present in skin and pulp fractions, although approximately 6.7 times higher concentrations were found in the skins by mass. As such, i

glycoconjugate extraction from berry homogenate was considered to be more efficient than from juice. To investigate the potential for smoke taint carry over to subsequent growing seasons, grapes were collected from control and smoked Merlot and Viognier grapevines in the season following smoke exposure. Subsequent analysis showed no evidence to suggest grapevine sequestration of glycoconjugates. The HPLC-MS/MS based SIDA method was adapted for the quantification of guaiacol glycoconjugates in wine and the method applied to several winemaking trials to investigate glycoconjugate metabolism during fermentation. Reduced skin contact achieved using a cold soak winemaking technique, yielded wines with significantly lower concentrations of guaiacol glycoconjugates, compared with traditional red winemaking practices involving extended skin contact at ambient temperature. This suggests winemaking processes which limit precursor extraction might offer opportunities for winemakers to ameliorate the impact of smoke taint in wine. A yeast selection trial demonstrated only partial metabolism of glycoconjugates during fermentation, with glycoconjugate concentrations of finished wines not significantly influenced by choice of yeast strain. The fact that a considerable portion of the glycoconjugate pool remained after fermentation has important implications for industry; i.e. hydrolysis of glycoconjugates after bottling could result in enhanced smoke characters with ageing. The effect of grapevine smoke exposure on grape berry microflora and the performance of several winemaking yeast in the presence of smoke-derived volatiles were also investigated. The growth of indigenous and winemaking yeast on yeast media agar plates spiked with guaiacol, 4-methylguaiacol or a liquid smoke preparation was investigated to further determine the impact of smoke-derived volatile compounds on yeast performance. ii

ACKNOWLEDGEMENTS Throughout the last three years I have been fortunate enough to have a great team of people behind me supporting and helping me the whole way through my study. Firstly many thanks to my principal supervisor Dr. Kerry Wilkinson who convinced me to do a PhD, your guidance and support was crucial to the success of this project and I am grateful to have had such a supporting supervisor. Thankyou also to my supervisors Prof. Dennis Taylor for his help and advice throughout the last three years and Dr. Yoji Hayasaka from the Australian Wine Research Institute, who has enabled all the HPLC-MS analysis for this work, he is truly the master of the HPLC- MS and without his expertise this project would have been a lot more difficult to achieve. Without valuable funding from the GWRDC this work would not have been conducted so I thank them for their financial support. Also to the Department of Agriculture, Fisheries and Forestry for awarding me the 2009 Science and Innovation award for young people in Agriculture, Fisheries and Forestry and the associated funding that supported the microbiological research to be conducted. The University of Adelaide provided laboratories and facilities for this research for which I am grateful. I have been fortunate enough to build some important collaborations during my research. Firstly I would like to thank Dr. Alan Pollnitz for his helpful discussions and guidance. Also to Mrs Gayle Baldock from the Australian Wine Research Institute who helped with GC-MS and HPLC-MS analysis. There are so many people who have helped me with this research, but I have to start by thanking all the members of the Wilkinson group especially Anthea and Renata for their assistance with the challenging field work and great friendship as well. Crista viii

Burbidge provided me with microbiological technical assistance, without which I would have been lost. Various other people have been essential in helping me with my research and being there as friends, including all the guys and girls in the Taylor group. Finally and most importantly I need to thank my friends and family for always supporting me with everything I have done. To my parents who helped support me throughout my study, I am so grateful for having those opportunities. My husband Chris you have been there for me when I was stressed and unsure about where I was going, you have pushed me to succeed and I will love you always. ix

CHAPTER 1 INTRODUCTION. Chapter 1 1

CHAPTER 1: INTRODUCTION. 1.1 Introduction to the Australian wine industry. The Australian wine industry contributes significantly to the Australian economy, with in excess of $2.0 billion in domestic sales and $2.5 billion in export sales reported in 2007/08. 1 In 2006, Australia became the 4 th largest exporter of wine in the world, behind France, Italy and Spain; with 776.6 million litres of wine exported, primarily to the UK, USA, Canada, China, Germany and New Zealand. 2 In any given year, the success of the Australian industry is determined by both the yield and the quality of grapes produced. In 2009, 1.73 million tonnes of grapes were crushed to produce 1.16 billion litres of wine, but in 2010 the winegrape crush decreased by almost 12%, to 1.53 million tonnes. 3 Over the past 10 years, annual grape yields have fluctuated from approximately 1.4 M tonne (in 2003 and 2007) to approximately 1.9 M tonne (in 2004 and 2006). 4 Reduced yields have been attributed to atypical environmental conditions, in particular drought and frost, as well as several major bushfires. 2,5 The main issue arising from bushfires is not fire damage to vineyards and wineries, although in some cases this has occurred, but rather grapevine exposure to smoke (Figure 1), which can result in objectionable smoke characters being observed in subsequent wines. 6 Given the incidence of vineyard exposure to smoke is likely to increase due to the prolonged warm, dry conditions associated with climate change, 7 together with the potential for significant financial losses, smoke taint has become an issue of increasing concern for grapegrowers and winemakers. To ensure the continued demand for Australian wine in both domestic and export markets, industry needs to gain a better understanding of the impacts of grapevine smoke exposure. Chapter 1 2

NOTE: This figure is included on page 3 of the print copy of the thesis held in the University of Adelaide Library. Figure 1: A bushfire occurring in close proximity to a vineyard, resulting in grapevine smoke exposure. 8 1.2 History of vineyard exposure to smoke. In recent years, vineyard exposure to smoke has been reported in wine regions throughout the world, including Canada (Okanagan Valley), USA (California), South Africa and Australia. 9 The first incidence of vineyard smoke exposure in Australia was reported in mid-january 2003, following bushfires in Victoria and Canberra. Numerous fires broke out in the Kosciuszko and Namadgi national parks, as a result of extreme weather conditions, such as strong wind, lightning and high temperatures. 10,11 Escalation of the fires occurred rapidly, burning many outer suburbs of Canberra and casting thick smoke over the King and Alpine Valley wine Chapter 1 3

regions. As a consequence, smoke affected juice and wine submitted to the Australian Wine Research Institute (AWRI) from these regions, were described as exhibiting objectionable smoky, burnt, ash, ashtray and smoked salmon aromas, with an excessively drying back-palate and a retro-nasal ash character. 7 Smoke affected fruit became a significant concern for winemakers, in particular determination of the extent of smoke exposure and consequences for wine quality. Financial losses to grapegrowers and winemakers were subsequently estimated at $4 million. 6 During the 2003 vintage, the AWRI, reported an inundation of samples for smoke taint analysis and enquiries from concerned grapegrowers and winemakers. In response to the 2003 bushfires, AWRI conducted a series of preliminary investigations which were published in their annual report. 7 The major outcomes resulting from this work included: Identification of the volatile phenols, guaiacol and 4-methylguaiacol, as the major contributors to smoke taint; the concentration of these phenols was found to be strongly correlated with the intensity of perceived taint, but AWRI acknowledged other smoke-derived compounds were likely to be present also. Detection of guaiacol and 4-methylguaiacol in skin rather than pulp fractions of smoke affected grapes. The discovery that increased maceration times or maceration with leaf material gave increased guaiacol concentrations in resultant wines. Subsequent to the above findings, AWRI conducted several vineyard and winery trials in an attempt to identify potential amelioration methods for reducing smoke taint in grapes and wine. 7 A vineyard washing trial was conducted, which involved the Chapter 1 4

application of cold water, cold water plus a wetting agent, warm water, cold water plus 5% ethanol and milk treatments to smoke affected grapevines. However, none of these treatments gave significantly reduced juice guaiacol concentrations. The washing liquids were found to contain some particulate matter, but guaiacol and 4- methylguaiacol were not detected in any of these samples, indicating vineyard washing did not effectively reduce volatile phenol concentrations. 7 However, the vineyard water wash was still considered to be beneficial, as it removed up to 90% of smoke-derived ash and particulate matter which could potentially have contained other smoke components capable of contributing undesirable sensory attributes. 7 A fining trial was also conducted and the capacity of various fining agents to remove guaiacol from smoke affected wine was investigated. Of the fining agents trialled, only activated carbon was found to remove guaiacol; however the 5% reduction achieved was minor, and therefore not especially beneficial to winemakers. 7 Based on these findings, the AWRI made a number of recommendations to the Australian wine industry to assist grapegrowers and winemakers to minimise the effects of smoke on grapes and wine. 7 They recommended leaf plucking followed by a cold water vineyard wash, hand picking and whole bunch pressing fruit would most likely minimise the intensity of smoke taint in resultant wines. AWRI also suggested that reduced maceration times would limit the extraction of guaiacol during fermentation. 7 The recommendations provided by AWRI were based on the outcomes of their trials, but these trials lacked detailed experimental design, in particular replication. As such these findings cannot be considered conclusive and further research in this area is warranted. Following their preliminary trials, AWRI reported there is a possibility that smoke taint might become a sporadic but more common occurrence in the future. This prediction Chapter 1 5

proved accurate and subsequent bushfires indeed occurred, with a series of major fires occurring in north eastern Victoria between the 1 st December 2006 and the 7 th February 2007. 6 As a consequence, smoke taint was identified by grapegrowers and winemakers in the King and Ovens Valleys, Milawa, Beechworth and Glenrowan regions; i.e. not only wine regions in close proximity to fires but also more distant regions, due to wind patterns which caused smoke to drift. 12 Direct financial losses of up to $20 million were estimated in 2007, being the cost associated with discarding smoke affected fruit. However, total losses were estimated to be closer to $90 million, being the value of expected profits from wine sales, together with subsequent loss of shelf space and impact on reputation of brands. 9 Severe bushfires occurred again in February 2009, in areas surrounding the Yarra Valley wine region. 13 While many vineyards reported financial losses associated with fire damage to vineyards, drifting smoke plumes also resulted in smoke exposure of fruit in a more widespread area of the Yarra Valley and Victoria (Figure 2). The Yarra Valley Winegrowers Association reported fire damage or destruction of 29 vineyards, with some individual growers losing up to 40 hectares of vineyards. 13 Chapter 1 6

NOTE: This figure is included on page 7 of the print copy of the thesis held in the University of Adelaide Library. Figure 2: Satellite image of smoke from the Black Saturday bushfire, taken on February 7 th 2009; the Yarra Valley wine region is circled in red. 14 Although bushfires have been the major cause of smoke taint in wines, prescribed burns conducted in the vicinity of wine regions, have also resulted in smoke affected grapes and wine. 15 Winemakers in Western Australia sought millions of dollars in compensation from the WA Department of Environment and Conservation following vineyard exposure to smoke as a result of prescribed burns conducted during the 2004 growing season. 15 Forestry Tasmania also fielded complaints from grapegrowers regarding the detrimental impact of smoke from prescribed burns on their vineyards. 16 In response, some government agencies responsible for prescribed burning have introduced new guidelines detailing the schedule of prescribed burns in efforts to work more closely with growers and to identify more compatible burn times which minimise the likelihood of smoke damage to vineyards. 16 While improved communication between government agencies and the wine industry will reduce the Chapter 1 7

impact of smoke from prescribed burns, the occurrence of bushfires is expected to increase as a consequence of the hot and dry conditions caused by climate change. As such, further research concerning the impact of smoke on grape and wine production is warranted. 1.3 The effects of smoke on nature. Prior to 2003, the effects of smoke on grape and wine composition and quality had not been considered. However, considerable research has been undertaken to investigate the role of smoke in seed germination. The application of smoke to seeds from a variety of plants has been shown to stimulate seed germination, in some cases prompting dormancy to be broken in seeds of threatened species. 17 Although smoke application doesn t positively influence the seed germination rate of all species, it has enabled the regeneration and conservation of many plant species; Brown and van Staden 17 reported improved germination of seeds from 45 of 94 native Western Australian plants following smoke exposure. Flematti and coworkers 18 attempted to identify the smoke constituents responsible for seed germination by isolating an active fraction of an aqueous smoke extract using a combination of solvent partitioning, acid-base separation, column chromatography and semi-preparative high performance liquid chromatography (HPLC). The active smoke fraction obtained was then applied to seeds of three plant species: Lactuca sativa L. Grand Rapids, Conostylis aculeate R. Br. and Stylidium affine Sonder, which improved seed germination rates by more than 100%. 18 Subsequent HPLC-MS analysis of this active fraction revealed the presence of 10 different compounds, each of which was separated by collecting 1 minute elution fractions from the HPLC-MS. 18 The seed germination potential of each individual compound was investigated using Chapter 1 8

Grand Rapids seeds, and the active compound, with an elution time of 25 to 27 mins and a quasi-molecular ion of m/z 151 was tentatively identified. 18 The active constituent was later confirmed to be a butenolide, following comparison with a synthetic reference. 19 While the role of smoke in seed germination has been extensively studied and documented in the literature, there is little research concerning the effect of smoke on plant physiology. Gilbert and Ripley investigated the photosynthetic response of Chrysanthemoides monilifera (more commonly known as Boneseed in Australasia) following smoke exposure, being the only physiological study conducted to date. 20 Greenhouse grown plants, of no less than seven months in age, were exposed to grass derived smoke using a commercial bee smoker, for a duration of one minute. 20 Smoke exposure resulted in a significant decrease in carbon dioxide assimilation rates, stomatal conductance and internal carbon dioxide concentrations. 20 Complete recovery of photosynthetic gas exchange rates was reported 24 hours post-smoke exposure. Plants subjected to five consecutive days of one minute smoke treatments showed no physiological response to smoke application on the fifth day, which suggested the plants had developed resistance to smoke exposure. 20 Longer periods of smoke exposure, i.e. 5 minutes, had a more pronounced effect on plant health and resulted in leaf necrosis and shoot death. 20 This indicated that extended periods of smoke exposure might damage plant tissue. The anti-microbial effects of smoke have also been investigated. Although smoke can have a detrimental effect on the health of living plants, it has long been used for the preservation of food products, such as fish, cheese and meat. 21 Specifically smoke inhibits the growth of micro-organisms in food and kills bacteria known to cause disease, 22 for example, Listeria monocytogenes, is a micro-organism present Chapter 1 9

in soft cheese, milk, meat and fish which is known to cause illness and food poisoning when ingested. 22 Niedziela et al. reported the inhibition of Listeria monocytogenes in salmon following treatment with smoke. 22 Similarly, Faith et al. demonstrated the anti-listerial properties of liquid smoke preparations. 23 In addition to the anti-microbial nature of smoke, a range of smoke components isolated from liquid smoke preparations, for example lignin dimers, have been shown to exhibit antioxidant and organoleptic properties. 24 Smoke preservation techniques therefore offer potential benefits in addition to the inhibition of harmful micro-organisms. Aside from preservation properties, the unique flavour and aroma imparted by smoke has become an important characteristic of foods prepared using smoke preservation techniques. 21 The number and nature of chemicals which contribute to the aroma and flavour of smoke has been the subject of considerable research. Of the compounds identified to date, the volatile phenols are considered by many researchers to be the major contributors of wood smoke aroma. 21,25-28 The volatile phenolic fraction of smoke is also thought to be the major contributor of the smoky aroma and flavour of smoked food products. 21 As such, the composition of smoked food has been well investigated and guaiacol and 4-methylguaiacol have been identified as two of the major volatile organic compounds to which the smoky aroma in foods has been attributed. 29 1.4 Composition of smoke. Smoke is a highly complex matrix and the precise composition of smoke depends on the nature and moisture content of the fuel source, the temperature of combustion and the availability of oxygen. 28 More than 400 volatile organic compounds have been identified in smoke and liquid smoke preparations. 27,28,30 These compounds Chapter 1 10

include: acids; alcohols; aldehydes; ketones; esters; furan and pyran derivatives; lactones; phenols; ethers; hydrocarbons; and nitrogenated derivatives. 26,28 Although the combination of these compounds provide smoke with its unique flavour profile, the volatile phenols have been identified as the major contributors. 21,25,26,28,30-33 Smoke is produced by the thermal combustion of a fuel source such as wood or plant material. The volatile phenol fraction of smoke is principally derived from the pyrolytic degradation of lignin to give ferulic acid, which has been shown to undergo further thermal decomposition to produce a series of volatile phenols. 31 Volatile phenols comprise an aromatic ring with one or more hydroxyl groups, as well as other functional groups such as aldehydes, ketones, acids and esters. 28 Guaiacol, 4-methylguaiacol, 4-ethylguaiacol, 4-ethylphenol and eugenol are the more abundant volatile phenols identified in smoke; their chemical structure and sensory descriptors are shown below (Figure 3). 21,28 OCH 3 OCH 3 OCH 3 OCH 3 OH 1 guaiacol 'sweet', 'smoky', 'pungent' OH 2 4-methylguaiacol 'sweet', 'smoky', 'toasted', 'ash' OH 3 4-ethylguaiacol 'sweet', 'smoky', 'spicy' OH 4 eugenol 'clove', 'woody' OH 5 4-ethylphenol 'pungent', 'horsey', 'barnyard' Figure 3: Volatile phenols identified in wood smoke and liquid smoke preparations, and their sensory descriptors. 27,28,34 Chapter 1 11

Smoke-derived volatile phenols are important to the flavour and aroma qualities of smoke and not surprisingly they have been identified in commercial liquid smoke preparations used to artificially flavour foods. 25,26,35 Liquid smoke preparations are commonly applied to foods such as meat, fish and cheese to impart desirable smoke attributes without necessitating the use of specialised smoke equipment. 26 Liquid smoke flavourings are prepared by introducing smoke into a liquid matrix, often distilled water. 25,26 Smoke flavourings can differ in viscosity, colour and odour, depending on the matrix used to retain smoke-derived volatiles, and the concentration and ratio of individual components within the matrix; which are influenced by fuel source and parameters used for combustion. 24 Commercial smoke preparations have generally been found to contain different ratios of carbonyl and phenolic compounds, with those containing a higher proportion of carbonyl compounds (for example 2-propanone, 2,3-dimethyl-2-cyclopenten-1-one and 2- ethyl-2,5-dimethylcyclopenten-2-one), considered to best reflect the sensory characteristics of smoke. 24 Not surprisingly, many of the volatile organic compounds present in smoke and liquid smoke flavourings are also identified in smoked food products, and some of these volatiles could potentially be responsible for smoke taint in grapes and wine. The volatile phenols guaiacol and 4-methylguaiacol have not only been identified in wine as a result of smoke, but are typically attributed to oak maturation. 36-38 Wine is traditionally aged in oak barrels to enhance aroma, flavour and complexity. During barrel maturation, oak-derived volatile compounds including guaiacol and 4- methylguaiacol can be extracted from the oak wood into the wine. 36,37,39 Oak aged wines typically contain between 10 and 100 µg/l of guaiacol and between 1 and 20 µg/l of 4-methylguaiacol 39, and at concentrations exceeding their detection thresholds, (Table 1) are considered to contribute desirable sensory characters. 39 Chapter 1 12

However, in smoke tainted wines these phenols may contribute to the objectionable smoky, burnt, ashtray, smoked salmon characters 7, which anecdotally are thought to be more apparent in white wine varieties. Table 1: Aroma detection thresholds and wine concentrations for guaiacol and 4- methylguaiacol. Compound Aroma detection threshold (µg/l) wine concentration water white juice red wine guaiacol 0.48 40 <6 41 9.5 42 0-100 39 4-methylguaiacol 10 34 65 34 65 34 0-20 39 1.5 Smoke taint in grapes and wine. Until recently there was no scientific literature concerning smoke taint in grapes and wine. However, the recurrence of bushfires in close proximity to wine regions prompted several research groups to investigate the effects of smoke on grape and wine production. The first scientific literature concerning smoke taint comprised a series of papers by Kennison and collegues. 43-45 Their first paper aimed to demonstrate the link between smoke exposure of grapes and smoke taint in wine, by comparing the composition and sensory attributes of smoke affected and control wines. 45 Verdelho bunches were exposed to straw-derived smoke post-harvest for 1 hour. Control (i.e. no smoke exposure) and smoked grapes were then fermented according to two different winemaking protocols: one involving juice clarification and primary fermentation, i.e. reflecting commercial white winemaking; and one involving oxidative primary Chapter 1 13

fermentation with skin contact, followed by malolactic fermentation, i.e. reflecting commercial red winemaking. Wines were then subjected to chemical and sensory analyses to determine the effect of smoke exposure and winemaking treatments. 45 The concentrations of a range of volatile phenols including guaiacol, 4-methylguaiacol, 4-ethylguaiacol, 4-ethylphenol and eugenol were determined by gas chromatography-mass spectroscopy (GC-MS). Irrespective of the winemaking treatment employed, the volatile phenols were not detected in wines made from control (unsmoked) grapes, but were detected in wines made from smoked grapes (Table 2). Guaiacol and 4-methylguaiacol in particular were reported at elevated levels; while the concentrations of 4-ethylphenol, 4- ethylguaiacol and eugenol were within ranges previously reported in wine (Table 1). 45 Table 2: Concentrations of guaiacol, 4-methylguaiacol, 4-ethylguaiacol, 4- ethylphenol and eugenol present in smoked and unsmoked wines. 45 Concentration a (µg/l) Smoked free run Unsmoked free run Smoked free run on skins Unsmoked free run on skins guaiacol 1470 a n.d. 969 b n.d. 4-methylguaiacol 326 a n.d. 250 b n.d. 4-ethylguaiacol 128 a n.d. 111 b n.d. 4-ethylphenol 59 a n.d. 67 b n.d. eugenol 20 a n.d. 26 b n.d. a Values followed by a different letter within rows are significantly different. n.d.= not detected. Mean values from three replicates. Values were in agreement to ca. 5%. Sensory analysis confirmed a perceivable difference between smoked and control wines (at the 99.9% confidence level). 45 The sensory panel were able to differentiate smoked wine blended with control wine, until a 98% dilution factor was achieved. On Chapter 1 14

this basis, the authors concluded industry probably couldn t rely on blending to significantly diminish smoke related sensory attributes of smoke tainted wine. Kennison and colleagues subsequently investigated the application of smoke to grapevines in the field and the evolution of volatile phenols during fermentation. 43 Merlot grapevines were exposed to repeated smoke treatments (8 x 30 mins each) at different timepoints between veraison and harvest, using purpose built smoke tents erected around the vines and straw-derived smoke. 43 Control grapevines were also enclosed in tents, but without smoke exposure, to eliminate any effects of the tent. At maturity, grapes from control and smoked grapevines were fermented; primary fermentation was conducted with skin contact followed by malolactic fermentation, with samples collected at various stages of winemaking for analysis by GC-MS to determine volatile phenol concentrations. 43 Consistent with previous findings, Kennison et al. 43 showed that volatile phenols were either not detected or detected at only trace levels in control wines (Table 3). The volatile phenol concentrations of smoked ferment samples increased progressively throughout the winemaking process, with guaiacol and 4-methylguaiacol again being the most abundant phenols (Table 3). This finding supported anecdotal evidence from winemakers that the intensity of smoke taint increased with fermentation. The authors considered these results could indicate the progressive release of phenols from grape skins, except that phenol concentrations continued to increase after the wines were pressed off the skins (Table 3). Further increases were also observed for some phenols after 12 months bottle ageing. The authors instead suggested the evolution of phenols after pressing and bottling might be due to the hydrolysis of precursor forms of the phenols. Chapter 1 15

Table 3: Concentrations of guaiacol, 4-methylguaiacol, 4-ethylguaiacol, 4-ethylphenol and eugenol during fermentation of fruit derived from smoked and unsmoked grapevines. 43 Concentration a (µg/l) Sample guaiacol 4-methyl 4-ethyl 4-ethyl guaiacol guaiacol phenol eugenol unsmoked free run juice n.d. n.d. n.d. n.d. n.d. after 1 day maceration tr. tr. n.d. n.d. tr. after 3 days maceration tr. tr. n.d. n.d. tr. after 5 days maceration tr. tr. n.d. n.d. tr. after 7 days maceration tr. tr. n.d. n.d. tr. after 10 days maceration 1 tr. n.d. n.d. tr. after alcoholic fermentation 1 tr. n.d. n.d. tr. finished wine 4 n.d. tr. tr. tr. 12 months post-bottling 3 tr. tr. tr. n.d. smoked free run juice 1 a tr. n.d. n.d. n.d. after 1 day maceration 68 b 11 a 10 a 5 a 2 ab after 3 days maceration 168 c 26 b 8 a 5 a 1 a after 5 days maceration 203 cd 32 bc 9 a 15 b 2 a after 7 days maceration 249 d 42 c 9 a 17 b 2 a after alcoholic fermentation 249 d 43 c 8 b 23 c 1 a finished wine 388 e 93 d 16 c 58 d 3 b 12 months post-bottling 371 e 124 e 29 c 94 e 4 c a Values are the means from three replicates and were in agreement with ca. 10%. Values followed by a different letter within columns are significantly different (P < 0.05). n.d.= not detected; tr.= trace (i.e. positive identification but < 1µg/L) To investigate their hypothesis a series of hydrolysis studies were performed. 43 Free run juice from control and smoke affected Merlot grapes was hydrolysed under either mildly acidic (i.e. ph=3.5), strongly acidic (i.e. ph=1) or enzymatic (β-glucosidase) conditions. Only trace levels of phenols were detected in control hydrolysates; whereas the concentration of phenols in smoked juice increased significantly following either strong acid or enzyme hydrolysis (Table 4). 43 This data provided Chapter 1 16

further evidence to support the authors hypothesis that guaiacol might be bound within the grape in precursor form. Furthermore, evolution of guaiacol following treatment with β-glucosidase enzymes leads the authors to suggest these precursors might be glycoconjugate in nature. Table 4: Volatile phenol concentrations before and after mild acid (ph=3.5), strong acid (ph=1) and β-glucosidase enzyme hydrolysis. 43 Sample guaiacol Concentration a (μg/l) 4-ethyl guaiacol 4-methyl guaiacol 4-ethyl phenol eugenol unsmoked free run juice n.d. n.d. n.d. n.d. n.d. mild acid hydrolysate tr. tr. tr. tr. n.d. strong acid hydrolysate tr. tr. tr. tr. 2 enzyme hydrolysate tr. tr. tr. tr. n.d. smoked free run juice 1 tr. n.d. n.d. n.d. mild acid hydrolysate tr. tr. tr. tr. n.d. strong acid hydrolysate 431 162 31 48 5 enzyme hydrolysate 325 82 13 27 n.d. a Values are the means from three replicates for juice samples and two replicates for hydrolysate samples. Values were in agreement to ca. 10%. n.d.= not detected; tr.= trace (i.e. positive identification but < 1µg/L). Kennison and co-workers then investigated the effect of timing and duration of grapevine exposure to smoke. 44 Merlot grapevines were exposed to a single smoke treatment (for 30 min) at one of eight different time points between veraison and harvest. A second treatment involved Merlot grapevines being exposed to eight repeated smoke applications. In each case, treatments were taken to a wine outcome for chemical and sensory analysis. Once again, only trace levels of phenols were detected in control wines. For single smoke treatments, all smoked wines were found to contain smoke-derived volatile phenols and to exhibit some degree of Chapter 1 17

smoke related sensory characters. However, the highest phenol levels and most apparent smoke taint was reported for wines made from grapes exposed to smoke 7 days post-veraison, suggesting at this phenological timepoint grapes are most vulnerable to smoke. 44 Considerably higher phenol levels were observed for wines derived from repeated smoke treatments, demonstrating that prolonged or repeated smoke exposure will result in a more intense smoke taint. A similar study investigating the effects of timing of smoke exposure was conducted by Sheppard et al. 46 Fruit from Chardonnay, Pinot Gris and Merlot grapevines were exposed to smoke produced from pine, burned in a modified barbeque and pumped into a box surrounding the vines. Smoke was applied to vines at three different stages of growth, preveraison, postveraison and maturity, and the grapes were harvested and analysed by GC-MS to determine guaiacol and 4-methylguaiacol concentrations. The authors of this study also concluded that the timing of grapevine smoke exposure influenced guaiacol and 4-methylguaiacol concentration, and that grape variety might also affect the uptake of smoke. 46 Kennison s research strongly suggests smoke-derived volatile phenols accumulate in grapes in glycoconjugate forms (i.e. as glucose derivatives), following grapevine exposure to smoke. Industry currently relies on quantification of guaiacol and 4- methylguaiacol using existing GC-MS based analytical methods to assess the extent of taint in smoke affected grapes. 43 However these methods do not account for bound or precursor forms of guaiacol and 4-methylguaiacol, so there is significant potential for smoke taint to be under-estimated. Grapes with low or undetectable volatile phenol levels might release significant levels of these phenols, and therefore smoke taint, during fermentation. Therefore, the provenance of glycoside derivatives of volatile phenols in smoke affected fruit needs to be established and analytical methods specific to these precursors subsequently developed. Chapter 1 18

1.6 Research aims. Given the close proximity of many grape growing regions to bushland and forests, and the warm, dry conditions experienced during summer, the incidence of bushfires, and therefore smoke taint is likely to continue. A number of volatile phenols have been identified in smoke tainted grapes and wine, but research findings to date suggest these compounds might accumulate in smoke affected grapes in precursor forms, i.e. as glycosides. This project therefore aimed to investigate the provenance of glycoside precursors of guaiacol, as the most abundant of the smoke-derived volatile phenols. The project aimed to: 1. Synthesise the β-d-glucoside of guaiacol as a reference compound to confirm its presence in smoke affected grapes. 2. Synthesise the deuterated β-d-glucoside of guaiacol for use as an isotopically labelled internal standard for the development of a quantitative high performance liquid chromatography tandem mass spectrometry (HPLC- MS/MS) based Stable Isotope Dilution Analysis (SIDA) method. 3. Apply the HPLC-MS/MS method to the analysis of smoke affected grapes, to investigate the accumulation and distribution of glycoconjugates. 4. Apply the HPLC-MS/MS method to the analysis of smoke affected wines to investigate the behaviour of the glycoconjugates during fermentation and bottle storage (ageing). To date, the primary focus of smoke taint research conducted has concerned the chemical composition and sensory characteristics of smoke affected grapes and wine. However, the anti-microbial, preservative properties exhibited by smoke could Chapter 1 19

potentially influence the growth of indigenous microflora on grapes or the performance of winemaking yeast during fermentation. As such, this project also aimed to: 5. Investigate the impact of smoke on grape berry microflora and fermentation rates. 6. Investigate the growth of winemaking yeast in the presence of smoke-derived volatile compounds. Chapter 1 20

CHAPTER 2 SYNTHESIS OF GUAIACOL GLUCOSIDES. Chapter 2 21

CHAPTER 2: SYNTHESIS OF GUAIACOL GLUCOSIDES. 2.1 Introduction to glycosides in grapes and wine. Glycosides comprise an aglycone, with one or more sugar units attached. A range of different glycosides have been identified in grapes including rutinosides, disaccharides and glucosides. 47 Glycosides are thought to play a role in the storage and transport of hydrophobic compounds in the plant, facilitated by the increased solubility afforded by sugar units, as well as reduced reactivity and potential toxicity of aglycones. 47 Glycosides are ubiquitous, occurring frequently in nature. For example, many fruits including the cupuacu, anise, green vanilla beans, cape gooseberry and tomatoes have been found to contain glycosides of volatile phenols. 48-53 The glycosides within these fruits are responsible for the containment of a portion of volatile aroma compounds which may provide significant aroma potential for these fruits. For example, 24 out of the 47 aglycones identified in the Amazonian fruit capuacu, were found only in the enzyme hydrolysates of the glycoside fraction and not in the free volatile fraction; identifying the important role glycosides play in the flavour profile of this fruit. 48 It is well known that many grape derived aroma volatiles occur in glycosidic precursor forms. 54-62 Although glycosides possess no odour or flavour properties, they can be metabolised by yeast and enzymes during primary and malolactic fermentation to release odour active aglycones. 58,60 As such, acid and enzyme hydrolysis have been employed in many wine flavour related studies to release volatile compounds from glycosidic precursors, for example to enable the identification of novel molecules. 42,55,57,58,60,63,64 Many compounds have been isolated and identified in grapes by this method including monoterpenes, alcohols, aliphatic Chapter 2 22

alcohols and shikimates. 62,65-67 The norisoprenoid, β-damascenone is an important aroma compound present in grapes and wine, contributing to stewed apple, exotic fruit and honey characters. 68 β-damascenone occurs in grapes in either free or glycosidically-bound forms; the glycoside typically being quantified by release of β- damascenone following hydrolysis. 64 Similarly, benzenoid compounds such as vanillin and phenol have been identified in the acid and enzyme hydrolysates of Merlot and Cabernet Sauvignon grapes. 60 Glycosides are not only associated with grape-derived volatiles but also oak-derived volatiles. Cis- and trans-oak lactone, possibly one of the most important oak-derived volatiles, responsible for coconut, citrus and vanilla characters, 69 can also exist in glycosidic forms. 70,71 The galloyl-β-d-glucoside of cis-oak lactone has been isolated from oak wood and shown to liberate oak lactone under strongly acidic, enzymatic and pyrolytic conditions. 71 The toasting process of cooperage and enzyme activity during fermentation can also release oak lactone from its glycosidic precursors. Glycosides therefore play an important role in the liberation of aroma volatiles during winemaking. For this reason, the accumulation of glycosides of smoke-derived volatile phenols and their subsequent metabolism during fermentation, could explain the results reported by Kennison et al. 43 i.e. the release of volatile phenols from freerun juice of smoke affected Merlot grapes following treatment with β-glucosidase enzyme strongly supports Kennison s hypothesis that precursors are glycosidic in nature. However, further work is required to validate the provenance of the guaiacol β-d-glucoside in smoke affected grapes and wine. Chapter 2 23

2.2 Introduction to guaiacol β-d-glucopyranoside. Guaiacol has been isolated in the enzyme and acid hydrolysates of a number of plants and fruit, such as tomatoes, cape gooseberries and grapes, suggesting that it is present as a glycosidic precursor. 48,52,53,65,67 For example, hydroponically grown tomato cultivars, Jorge, Durinta and p73, chosen for their economical importance and usefulness for genetic transformation, were hand harvested at commercial maturity, homogenised and hydrolysed to enable analysis of their volatile flavour components. 53 Tomato juice from each variety was analysed for free and bound equivalents of volatile flavour compounds, by gas chromatography (GC). Glycosidic fractions were isolated by solid phase extraction, and volatile components released by pectinase hydrolysis. 53 The concentration of free guaiacol in the tomato varieties ranged from 503-945 µg/l, while bound concentrations ranged from 70-113 µg/l. 53 Similarly fresh cape gooseberries were harvested, homogenised, selectively concentrated by solid phase extraction, hydrolysed with a non-selective glucosidase enzyme and analysed by capillary gas chromatography-mass spectrometry (GC- MS). 52 Guaiacol was measured at concentrations ranging from 300-700 µg/kg of fruit in the hydrolysates, indicating the presence of guaiacol glycoside precursors in the cape gooseberry. 52 Interestingly, guaiacol has also been identified in grape juice hydrolysates derived from a range of varieties, including Shiraz and Merlot. 60,67 Shiraz berries subjected to enzymatic hydrolysis contained 17 µg/kg of guaiacol, well above the detection threshold of 9.5 µg/l in red wine. 67 Enzyme and acid hydrolysates of Merlot juice were found to contain up to 50 µg/l of guaiacol and the concentrations of guaiacol were seen to increase when a greater quantity of enzyme was used during Chapter 2 24

hydrolysis. 60 The natural occurrence of guaiacol and its glycosidic precursors, supports the presence of glycosidic precursors within smoke affected fruit. 2.3 Introduction to glycosylation. Glucosides are comprised of a glucose unit attached to the hydroxyl group of an aglycone in either an α or β-position (Figure 4). Glycosylation reactions usually involve two synthetic steps: (i) linkage of a protected glucose moiety to the aglycone unit, and (ii) deprotection, whereby the protecting groups on the glucose moiety are removed. Glycosidic extracts isolated from plants show that β-glucosides occur more commonly in nature than α-glucosides, due to the effectiveness of the β-glucosidase enzyme in releasing volatile compounds. 59,60,72 Various synthetic strategies have been developed to direct β-glycosylation, but the most effective is considered to be the modified Koenigs-Knorr method, which involves the use of a protected glucose unit as the sugar donor, in the presence of silver triflate as a catalyst. 73 OCH 3 O H O H OH O O OH OCH 3 HO i) HO ii) OH OH OH OH Figure 4: i) β-guaiacol glucoside and ii) α-guaiacol glucoside Chapter 2 25

The oak lactone glucosides have been synthesised using a modified Koenigs-Knorr method. 70,71,74 A similar method was also used by Fudge et al. to synthesise deuterium labelled cis-oak lactone; however, the glucose unit contained acetyl protecting groups and the deprotection method used KOH and MeOH. 70 Both methods gave the desired reaction product, although Wilkinson et al. 71 achieved yields of 66% and 98% in the glucosylation and deprotection steps, respectively, whereas Fudge et al. 70 achieved significantly lower yields of 14% and 73%. These methods are similar to those previously used to synthesise guaiacol β-d-glucoside. 2.4 Glycosylation of guaiacol. The synthesis of the β-d-glucopyranoside of guaiacol has been reported previously by Dignum et al. who reported an 18% yield using α-d-acetobromoglucose (Scheme 1). 49 Zhou et al. reported the synthesis of an acetyl-protected guaiacol glucoside, also using the α-d-acetobromoglucose as a reagent. 75 These investigations provided the basis for the preparation of guaiacol β-d-glucopyranoside in the current study. α-(oac) 4 Glu-Br, KOH, EtOH, CHCl 3 (CH 3 ) 2 CO/NaOH OCH 3 OCH 3 OCH 3 OH guaiacol OGlu(OAc) 4 protected guaiacol glucoside OGlu guaiacol glucoside Scheme 1: Synthetic scheme for preparation of guaiacol β-d-glucopyranoside as performed by Dignum et al. 49 Chapter 2 26

2.5 Results and discussion. 2.5.1 Preparation of guaiacol β-d-glucopyranoside. 2.5.1.1 Preparation of guaiacol glucoside using 2,3,4,6-tetra-O-acetyl-α-Dglucopyranosyl bromide (Method 1). The glycosylation of guaiacol was attempted using the methods reported by Dignum et al. 49 Low yields were obtained for both the glycosylation and deprotection steps, and thin layer chromatography (TLC) indicated the presence of several by-products. The major drawback of this method was the formation of significant quantities of the α-isomer, confirmed by the characteristic anomeric proton signal, i.e. with a coupling constant between 2-5 Hz. This glycosylation method involves non-selective isomeric attack of the acetyl glucose giving a mixture of α- and β-isomers, which are difficult to separate. Therefore, this glycosylation method was considered to be unsuitable for preparation of guaiacol β-d-glucopyranoside. 2.5.1.2 Preparation of guaiacol glucoside using 2,3,4,6-tetra-O-pivaloyl-α- D-glucopyranosyl bromide (Method 2). A modified Koenigs-Knorr method 71, using 2,3,4,6-tetra-O-pivaloyl-α-Dglucopyranosyl bromide in the presence of lutidine and silver triflate was instead employed for the glycosylation of guaiacol. The steric bulk of the pivaloyl protecting groups inhibit nucleophilic attack from the α-face, thereby improving selectivity for β- glycosylation. Chapter 2 27

The guaiacol β-d-glucopyranoside was successfully synthesised using this method with improved yields, i.e. 25% and 85% for the glycosylation and deprotection steps respectively, compared with 21% and 16% obtained using method 1. NMR analysis of the purified product showed no indication of the presence of the α-isomer. The glycoside was characterised by 2D 1 H and 13 C NMR spectroscopy and HPLC-MS. The anomeric proton produced a coupling constant of 6.6 Hz, characteristic of β-dglucopyranosides. HPLC-MS analysis of the guaiacol glucopyranoside further confirmed purity; the glucoside eluted at 5.7 min with a dominant ion of m/z 345.5, i.e. as an acetic acid adduct ion [M-H + CH 3 COOH], with a minor ion of m/z 285.0 as the molecular ion [M-H] in the mass spectrum. 2.5.2 Preparation of deuterated guaiacol β-d-glucopyranoside. The current analytical quantification of guaiacol and 4-methylguaiacol in smoke affected grapes and wine utilises stable isotope dilution analysis (SIDA) in conjunction with GC-MS. SIDA is a technique commonly used for quantification, generally using either GC-MS or HPLC-MS. 39,69,70,76,77 GC-MS based SIDA is typically used for volatile compounds such as guaiacol, 39 whereas HPLC-MS based SIDA is better suited to non-volatile compounds such as glycosides. SIDA employs an internal standard in the form of an isotopically labelled analogue of the analyte to be quantified, added at a known concentration prior to sample preparation and analysis. The HPLC or GC peak area of the analyte and the internal standard are compared to determine the concentration of the analyte. Any loss of the analyte that might occur during the extraction process is accounted for, by an equal loss of the isotopically labelled internal standard. Chapter 2 28

GC-MS based SIDA methods have been developed for the analysis of a wide range of compounds in wine, for example β-ionone, β-damascenone and cis- and trans- oak lactone. 39,64,78,79 SIDA methods developed for wine and oak analysis are currently used to quantify smoke-derived volatile phenols in grapes and wine. 39,76 SIDA methods have also been developed for the quantification of aroma precursors, for example the oak lactone glucosides. 80 The use of HPLC-MS based SIDA for the quantification of oak lactone glucosides provides the basis for the development of a method for the quantification of the guaiacol β-d-glucopyranoside. Isotopically labelled guaiacol β-d-glucopyranoside was synthesised in the same fashion as the unlabelled glucoside, but from deuterated guaiacol. Pollnitz et al. reported the synthesis of d 3 -guaiacol from catechol (i.e. via methylation of a hydroxyl group), albeit with only a 30% yield. 39 Deuterium exchange of the aromatic ring would enable incorporation of four deuterium atoms, improving the molecular mass difference between the analyte and the deuterated internal standard for HPLC-MS analysis. Pollnitz et al. reported deuterium exchange of 4-ethylphenol using deuterium oxide and thionyl chloride, in a reaction performed over 5 days. 76 In the current study, the reaction was instead performed with guaiacol using a microwave reactor to significantly reduce the duration of the reaction. The microwave assisted synthesis gave d 4 -guaiacol in 80% yield after just 30 hours reaction time. Conversion of guaiacol to its d 4 -equivalent was monitored by 1 H NMR and deuterium exchange was considered complete when aromatic 1 H peaks could no longer be detected. Deuterium exchange was confirmed by GC-MS analysis, with a molecular ion of m/z 128.2 [M + ] observed. Synthesis of the isotopically labelled guaiacol β-d-glucopyranoside was subsequently performed, according to glycosylation method 2. Confirmation of deuterium retention Chapter 2 29

on the aromatic ring was carried out by HPLC-MS; the d 4 -glucoside eluted at 5.97 min, and gave an acetic acid adduct ion of m/z 349.2 [M-H + CH 3 COOH], and molecular ion of m/z 289.1; i.e. 4 atomic mass units heavier than the unlabelled glucoside, as expected. 2.6 Materials and methods. 2.6.1 Solvents and reagents. Hexane was distilled at atmospheric pressure under nitrogen. Dichloromethane was dried over 4 Å molecular sieves (2.5 5 mm). All other solvents and reagents were used as purchased from Sigma Aldrich or Crown Scientific. 2.6.2 Chromatography. Analytical thin layer chromatography was performed with aluminium backed silica gel 60 F 254 sheets from Merck. Column chromatography was performed with silica gel 60 F 254 obtained from Scharlau (230-400 mesh). 2.6.3 Nuclear magnetic resonance spectroscopy (NMR). 1 H and 13 C NMR spectra were recorded with a Varian Gemini spectrophotometer operating at either 200 MHz, 300 MHz or 600 MHz. Spectra were recorded in either deuterated chloroform (CDCl 3 ) or deuterated pyridine (C 5 D 5 N). Chemical shifts are reported in parts per million (ppm) downfield. The following abbreviations are used in the assignment of 1 H spectra: s=singlet, d=doublet, m=multiplet, dd=doublet of doublets, ddd=doublet of doublets of doublets, t=triplet. Chapter 2 30

2.6.4 Ultra violet/visible spectroscopy and fluorescence spectroscopy. UV/Vis spectra were recorded with a Varian Cary 5000 UV-Vis-NIR spectrophotometer. Methanol was used as the solvent and as the blank. Fluorescence spectra were recorded with a Varian Cary Eclipse fluorescence spectrophotometer with methanol as the solvent. 2.6.5 Microwave synthesis. Microwave assisted synthesis was performed using a CEM Discover microwave reactor. 2.6.6 Gas chromatography-mass spectrometry (GC-MS). GC-MS analysis was performed with an Agilent 5973N mass spectrometer (MS) coupled to an Agilent 6890 gas chromatograph (GC) equipped with a GERSTEL MPS2 Multi Purpose Sampler (Agilent Technologies, Forest Hill, N.S.W., Australia). A 1 µl sample was injected and chromatographed on a ZB-WAX fused silica capillary column (Phenomenex, 7H6 6007 11, 30 m x 0.25 mm, 0.25 µm film thickness). The carrier gas used was helium with a flow rate of 1.9 ml/min. During analysis, oven temperature was started at 40 C, held at this temperature for 4 mins, increased to 130 C at a rate of 5 C/min and then increased to 220 C at a rate of 10 C/min and held at this temperature for a further 5 mins. The injector was set to split mode (ratio 50:1) and set at a temperature of 250 C. The transfer line was also set at a temperature of 250 C. Positive Ion electron impact mass spectra were recorded in selected ion monitoring (SIM) mode over a scan range of m/z 20-210. Chapter 2 31

2.6.7 High performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS). Mass spectrometric analysis was performed with a 4000 Q TRAP hybrid tandem mass spectrometer equipped with a Turbo ion source (Applied Biosystems/MDS Sciex) and coupled to an Agilent 1200 HPLC system equipped with binary pump, degasser, autosampler and column oven (Agilent Technologies, Santa Clara, CA, U.S.A.). Data acquisition and processing were performed using Analyst software (version 1.5.1, Applied Biosystems/MDS Sciex). Chapter 2 32

2.6.8 Synthesis. Guaiacol 2,3,4,6 -tetra-o-acetyl β-d-glucopyranoside (1) (Method 1). OCH 3 Oglu(Ac) 4 To a solution of 2,3,4,6-tetra-O-acetyl-α-D-glucopyranosyl bromide (2.5 g, 6 mmol), in chloroform (5 ml), was added a solution of guaiacol (620 mg, 5 mmol) and potassium hydroxide (280 mg, 5 mmol) in ethanol (5 ml) and heated at reflux for 2 hours. The reaction mixture was cooled, filtered and ice water added. The organic phase was then concentrated in vacuo to give a pale orange solid, which was purified by column chromatography (66% ethyl acetate in hexane). The crude product was then recrystallised from ethanol to give 1 as colourless needles (476 mg, 21%) (m.p.: 150-151 C). 1 H NMR (CDCl 3 ): 7.12 (1H, dd, J = 7.8, 1.8 Hz, ArH), 7.07 (1H, ddd, J = 7.8, 1.8 Hz, ArH), 6.90 (1H, ddd, J = 7.8, 1.8 Hz, ArH), 6.87 (1H, dd, J = 7.8, 1.8 Hz, ArH), 5.29 (2H, m, H 1, 3 ), 5.17 (1H, m, H 4 ), 4.96 (1H, dd, J = 5.4, 2.4 Hz, H 2 ), 4.28 (1H, dd, J = 12, 5.4 Hz, H 6a ), 4.16 (1H, dd, J = 12, 2.4 Hz, H 6b ), 3.82 (3H, s, OCH 3 ), 3.76 (1H, m, H 5 ), 2.08, 2.07, 2.04, 2.04 (12H, 4 x s, Ac). 13 C NMR (CDCl 3 ): 170.6, 170.3, 169.4, 169.4, 150.7, 146.1, 124.7, 120.9, 120.3, 112.8, 100.9, 72.7, 72.0, 71.3, 68.5, 62.0, 56.0, 20.7, 20.6, 20.6, 20.5. Chapter 2 33

Guaiacol β-d-glucopyranoside (2) (Method 1). OCH 3 Oglu 1 (200 mg, 0.4 mmol) was added to a mixture of 1M sodium hydroxide solution (5 ml) and acetone (5 ml). The mixture was stirred at room temperature for 1 hour and monitored by TLC (60% ethyl acetate in hexane). The mixture was then stirred for a further 30 min in the presence of acidified Dowex (H + ) ion exchange resin. The reaction mixture was filtered and concentrated in vacuo, and recrystallised with ethanol to give 2 as a white solid (20 mg, 16%). 1 H NMR (C 5 D 5 N): 7.6 (1H, ArH), 6.92-7.02 (3H, m, ArH), 5.67 (1H, d, J = 6.6 Hz, H 1 ), 4.52 (1H, dd, J = 12, 2.4 Hz, H 6a ), 4.33-4.41 (4H, m, H 2, 4, 5, 6b ), 4.10-4.12 (1H, m, H 3 ), 3.70 (3H, s, OCH 3 ). 13 C NMR (C 5 D 5 N): 150.7, 148.6, 123.0, 121.9, 116.9, 113.7, 102.7, 79.3, 79.0, 75.4, 71.7, 62.8, 56.4. Chapter 2 34

1,2,3,4,6-Penta-O-pivaloyl-β-D-glucopyranoside 81 (3). OPiv PivO PivO O OPiv OPiv D-Glucose (4.0 g, 22.2 mmol) was added portionwise to a solution of pivaloyl chloride (27 ml, 222.0 mmol) and pyridine (18.0 ml, 222.0 mmol) in chloroform (100 ml) and heated at reflux for 72 hours. The solvent was evaporated, the residue dissolved in water (100 ml) and extracted with ethyl acetate (5 x 70 ml). The organic extracts were combined and washed with water (100 ml), hydrochloric acid (1M, 100 ml), saturated sodium bicarbonate (100 ml) and saturated sodium chloride (100 ml). The solution was then dried over magnesium sulphate and the solvent removed in vacuo, to afford the crude product which was then recrystallised from ethanol to give 3 as a white crystalline solid (11.9 g, 89%) (m.p.: 148-150 C). 1 H NMR (CDCl 3 ): 5.70 (1H, d, J = 9.6, H 1 ), 5.37 (1H, t, J = 9.3, H 3 ), 5.22 (1H, dd, J = 9.3 and 8.4, H 2 ), 5.16 (1H, t, J = 9.6, H 4 ), 4.16 (1H, dd, J = 12.3, 2.7, H 6a ), 4.10 (1H, dd, J = 12.3, 4.8, H 6b ), 3.86 (1H, ddd, J = 10.2, 4.8, 2.7, H 5 ), 1.24, 1.20, 1.18, 1.15, 1.12 (45H, 5 x s, CH 3 ). Chapter 2 35

2,3,4,6-tetra-O-pivaloyl-α-D-glucopyranosyl bromide 81 (4). OPiv PivO PivO O PivO Br Hydrobromic acid solution (33%) in acetic acid (5 ml) was added dropwise to a solution of 3 (5.4 g, 9.0 mmol) in dichloromethane at 0 C, and the mixture was stirred at room temperature for 16 hours. The reaction mixture was then concentrated, coevaporating with benzene (2 x 20 ml) and diethyl ether (2 x 20 ml). The residue obtained was then dissolved in diethyl ether (40 ml), washed with saturated sodium bicarbonate solution (3 x 30 ml), then water (40 ml) and finally dried over magnesium sulphate and concentrated. The crude product was recrystallised from ethanol to give 4 as a white crystalline solid (2.2 g, 42%). 1 H NMR (CDCl 3 ): 6.62 (1H, d, J = 3.9, H 1 ), 5.63 (1H, t, J = 9.6, H 3 ), 5.21 (1H, dd, J = 10.3 and 9.4, H 4 ), 4.81 (1H, dd, J = 9.9 and 4.2, H 2 ), 4.34-4.28 (1H, ddd, J = 10.5, 3.9 and 3.3, H 5 ), 4.18-4.16 (2H, m, H 6a,6b ), 1.22, 1.19, 1.17, 1.13 (36H, 4 x s, CMe 3 ). Chapter 2 36

Guaiacol β-d-2, 3, 4, 6 -tetrapivaloyl glucopyranoside (5) (Method 2). OCH 3 Oglu(Piv) 4 4 (469 mg, 0.81 mmol) was added to a solution of guaiacol (100 mg, 0.81 mmol) in anhydrous dichloromethane (6 ml) containing silver triflate (210 mg, 0.81 mmol) and 2,6-lutidine (100 µl, 0.81 mmol). The reaction mixture was then stirred in darkness for 16 hours at ambient temperature. The reaction was quenched with saturated sodium bicarbonate solution (20 ml) and extracted with dichloromethane (2 x 15 ml). The organic extracts were then combined, washed with saturated sodium chloride solution, dried and concentrated in vacuo. The crude product was then purified by column chromatography (20-60% ethyl acetate in hexane) to afford 5 as a white crystalline solid (142 mg, 25%) (m.p.: 131-132 C). 1 H NMR (CDCl 3 ): 7.10 (1H, d, J = 7.8 Hz, ArH), 7.02 (1H, t, J = 7.2 Hz, ArH), 6.82-6.90 (2H, m, ArH), 5.41 (1H, t, J = 9 Hz, H 3 ), 5.33 (1H, t, J = 8.4, H 2 ), 5.17 (1H, t, J = 9.6, H 4 ), 5.07 (1H, d, J = 7.8 Hz, H 1 ), 4.23 (1H, d, J = 12 Hz, H 6a ), 4.05 (1H, dd, J = 12, 6.6 Hz, H 6b ), 3.80 (1H, m, H 5 ), 3.79 (3H, s, OCH 3 ), 1.18, 1.17, 1.16, 1.14 (36H, 4 x s, C(CH 3 ) 3 ). 13 C NMR (CDCl 3 ): 178.0, 177.2, 176.5, 176.2, 150.2, 146.0, 124.0, 120.7, 118.9, 112.5, 100.0, 72.5, 72.2, 71.1, 68.1, 62.1, 55.7, 38.8, 38.8, 38.8, 27.2, 27.1, 27.0, 27.0. Chapter 2 37

Guaiacol β-d-glucopyranoside (2) (Method 2). OCH 3 Oglu Sodium metal (36 mg, 1.56 mmol) was dissolved in methanol (5 ml) and the resulting sodium methoxide solution added to a solution of 5 (100 mg, 0.14 mmol) in methanol (5 ml). The reaction mixture was stirred for 16 hours at room temperature. Acidified Dowex (H + ) ion exchange resin was added to the reaction mixture and stirred for a further 30 mins. The reaction mixture was then filtered and concentrated in vacuo to produce 2 as a white solid (52 mg, 85%) (m.p.: 148-150 C). 1 H NMR (C 5 D 5 N): 7.6 (1H, ArH), 6.92-7.02 (3H, m, ArH), 5.67 (1H, d, J = 6.6 Hz, H 1 ), 4.52 (1H, dd, J = 12, 2.4 Hz, H 6a ), 4.33-4.41 (4H, m, H 2, 4, 5, 6b ), 4.10-4.12 (1H, m, H 3 ), 3.70 (3H, s, OCH 3 ). 13 C NMR (C 5 D 5 N): 150.7, 148.6, 123.0, 121.9, 116.9, 113.7, 102.7, 79.3, 79.0, 75.4, 71.7, 62.8, 56.4. MS: [M-H] = m/z 285.2 and [M+CH 3 COO] = m/z 345.2 (APCI in negative mode). Chapter 2 38

d 4 -Guaiacol (6). D D D D OCH 3 OH Guaiacol (2 g, 16 mmol) and a solution of thionyl chloride (2 ml, 27 mmol) in deuterium oxide (23 ml) were added to the reactor tube of a Discover SP-D microwave apparatus (CEM, Matthews NC, USA). The tube was capped and irradiated for 30 hours at 100 C. The reaction mixture was then neutralised with potassium carbonate and extracted with pentane (3 x 40 ml). The combined organic extracts were dried and concentrated to give 6 as a pale yellow liquid (1.65 g, 80%). 1 H NMR (CDCl 3 ): 3.89 (3H, s, OCH 3 ) GC-MS: retention time of 24.35 mins (100% pure). Mass spectrum = m/z 128.2 [M + ], 113.2, 85.2 Chapter 2 39

d 4 -guaiacol 2,3,4,6 -tetrapivaloyl β-d-glucopyranoside (7). D D D D OCH 3 Oglu(Piv) 4 4 (4.7 g, mmol) was added to a solution of 6 (1.0 g, 7.81 mmol) in anhydrous dichloromethane (10 ml) containing silver triflate (2.10 g, 7.81 mmol) and 2,6-lutidine (1 ml, 7.81 mmol). The reaction mixture was then stirred in the darkness for 16 hours at room temperature. The reaction was quenched with saturated sodium bicarbonate solution (30 ml) and extracted with dichloromethane (2 x 30 ml). The organic extracts were combined, washed with saturated sodium chloride solution, dried and concentrated in vacuo. The crude product was then purified by column chromatography (20% ethyl acetate in hexane) to give 7 as colourless needles (142 mg, 25%) (m.p.:130-131 C). 1 H NMR (CDCl 3 ): 5.41 (1H, t, J = 9Hz, H 3 ), 5.33 (1H, dd, J = 9.6, 8.4Hz, H 2 ), 5.17 (1H, t, J = 9.6 Hz, H 4 ), 5.07 (1H, d, J = 7.2 Hz, H 1 ), 4.23 (1H, dd, J = 12.6, 1.8 Hz, H 6a ), 4.05 (1H, dd, J = 12, 6.6 Hz, H 6b ), 3.83 (1H, ddd, J = 9.6, 6, 1.8, H 5 ), 3.79 (3H, s, OCH 3 ), 1.18, 1.17, 1.16, 1.14 (36H, 4 x s, C(CH 3 ) 3 ). 13 C NMR (CDCl 3 ): 178.0, 177.2, 176.5, 176.5, 150.1, 146.0, 100.1, 72.5, 72.2, 71.1, 68.1, 62.1, 55.7, 38.8, 38.8, 38.7, 27.1, 27.0, 27.0. Chapter 2 40

d 4 -guaiacol β-d-glucopyranoside (8). D D D D OCH 3 Oglu Sodium metal (115 mg, 5 mmol) was dissolved in methanol (5 ml), and the resulting sodium methoxide solution added a solution of 7 (300 mg, 0.42 mmol) in methanol (10 ml). The reaction mixture was stirred for 16 hours at room temperature. Acidified Dowex (H + ) ion exchange resin was added to the reaction mixture and stirred for a further 30 mins. The mixture was then filtered and the solvent removed in vacuo to produce 8 as a white solid (133 mg, 84%) (m.p.: 150-151 C). 1 H NMR (C 5 D 5 N): 5.67 (1H, d, J = 6.6 Hz, H 1 ), 4.54 (1H, dd, J = 12.6, 2.4 Hz, H 6a ), 4.34-4.42 (4H, m, H 2, 4, 5, 6b ), 4.12-4.14 (1H, m, H 3 ), 3.71 (3H, s, OCH 3 ). 13 C NMR (C 5 D 5 N): 150.7, 148.5, 102.7, 79.3, 79.0, 75.4, 71.7, 62.8, 56.4. MS: [M-H] = m/z 289.3 and [M+CH 3 COO] = m/z 349.5 (APCI in negative mode) Chapter 2 41

2.7 Conclusion. The most efficient method for the synthesis of β-d-guaiacol glucopyranoside (2) utilised 2,3,4,6-tetra-O-pivaloyl-α-D-glucopyranosyl bromide (4) as a reagent in the presence of silver triflate. This method gave improved yields compared to those methods previously reported, but most importantly the reaction selectivity favoured β- glycosylation over α-glycosylation. The β-d-glucopyranoside was used as an authentic reference sample to confirm the provenance of 2 in smoke affected grapes, and therefore the glycosylation of guaiacol following grapevine exposure to smoke (as described in chapter 3). Deuterated guaiacol was prepared using microwaveassisted synthesis, significantly reducing deuterium exchange reaction times. d 4 - Guaiacol was then glycosylated to give d 4 -guaiacol β-d-glucopyranoside (8), which was subsequently used as an internal standard for the development of a quantitative SIDA method using HPLC-MS/MS (as described in Chapter 4). Chapter 2 42

CHAPTER 3 PROVENANCE OF GUAIACOL GLUCOSIDE IN SMOKE AFFECTED FRUIT. Chapter 3 43

CHAPTER 3: PROVENANCE OF GUAIACOL GLUCOSIDE IN SMOKE AFFECTED FRUIT. Guaiacol is one of several volatile phenols considered to contribute significantly to the unique aroma of smoke 28 and has been identified as a component of both wood smoke 28,32,33 and smoke tainted wines. 45 The evolution of guaiacol during the fermentation of smoke affected Merlot grapes was attributed to the degradation of one or more precursor compounds by Kennison et al. 43 and the precursors were thought to be glycosidic in nature, given significant levels of guaiacol were also released following the addition of β-glucosidase enzymes to Merlot juice from the same grapes. 43 However the presence of a guaiacol β-d-glucopyranoside in smoke affected grapes had yet to be confirmed. This paper concerns an investigation into the provenance of guaiacol β-dglucopyranoside in smoke affected grapes, using HPLC-MS/MS analysis. The guaiacol β-d-glucopyranoside previously synthesised in Chapter 2 was used as an authentic reference compound to develop an HPLC-MS/MS method for its detection in juice. The release of guaiacol from its β-d-glucopyranoside precursor following treatment with acid and enzyme hydrolysis are also described. Chapter 3 44

A Hayasaka, Y., Dungey, K.A., Baldock, G.A., Kennison, K.R. & Wilkinson, K.L. (2010) Identification of a β-d-glucopyranoside precursor to guaiacol in grape juice following grapevine exposure to smoke Analytica Chimica Acta, v. 660 (1-2), pp. 143-148 A NOTE: This publication is included on pages 45-50 in the print copy of the thesis held in the University of Adelaide Library. A It is also available online to authorised users at: A http://dx.doi.org/10.1016/j.aca.2009.10.039 A

CHAPTER 4 QUANTIFICATION OF GUAIACOL GLYCOSIDES IN SMOKE AFFECTED FRUIT. Chapter 4 51

CHAPTER 4: QUANTIFICATION OF GUAIACOL GLYCOSIDES IN SMOKE AFFECTED FRUIT. Following identification of guaiacol β-d-glucopyranoside as a component of smoke affected fruit, 82 Hayasaka et al. 83 identified several guaiacol disaccharides using stable isotope tracer experiments involving the application of a 50:50 mixture of guaiacol and d 3 -labelled guaiacol to grapevine leaves. Subsequent HPLC-MS/MS screenings enabled the tentative identification of seven different guaiacol conjugates; a glucose-glucose disaccharide (glucosylglucoside), the glucoside, four glucosepentose disaccharides and a rutinoside. To investigate the glycosylation of guaiacol in smoke affected grapes, a quantitative analytical method was required for glycoconjugate determination. This paper concerns the development and validation of an HPLC-MS/MS based SIDA method using the d 4 -labelled guaiacol β-d-glucopyranoside synthesised in Chapter 2 as an internal standard. The method was subsequently applied to the analysis of grapes sourced from grapevines exposed to experimental smoke and from commercial vineyards exposed to bushfire smoke. The accumulation of guaiacol glycoconjugates within berry components was also investigated. Chapter 4 52

A Dungey, K.A., Hayasaka, Y. & Wilkinson, K.L. (2011) Quantitative analysis of glycoconjugate precursors of guaiacol in smoke-affected grapes using liquid chromatography -tandem mass spectrometry based stable isotope dilution analysis Food Chemistry, v. 126(2), pp. 801-806 A NOTE: This publication is included on pages 53-58 in the print copy of the thesis held in the University of Adelaide Library. A It is also available online to authorised users at: A http://dx.doi.org/10.1016/j.foodchem.2010.11.094 A

CHAPTER 5 QUANTIFICATION OF GUAIACOL GLYCOCONJUGATES IN GRAPES AND WINE. Chapter 5 59

CHAPTER 5: QUANTIFICATION OF GUAIACOL GLYCOCONJUGATES IN GRAPES AND WINE. 5.1 Introduction. In Chapter 4, the development and validation of an HPLC-based SIDA method for the quantification of guaiacol glycoconjugates in grapes was described. The application of this method to various experimental field trials subsequently enabled the glycosylation of smoke-derived guaiacol following grapevine exposure to smoke to be determined, as well as the distribution of glycoconjugates within different berry components. The preferential accumulation of glycoconjugates in grape skins suggested winemaking techniques which involve reduced skin contact time might offer potential methods of amelioration. Therefore, to investigate the behaviour of guaiacol glycoconjugates during fermentation, the HPLC-MS/MS method was adapted for wine analysis. This chapter describes the development and validation of a SIDA based HPLC- MS/MS method for quantification of guaiacol glycoconjugates in wine and its application to various winemaking trials. Chapter 5 60

5.2 Results and discussion. 5.2.1 Method development. 5.2.1.1 Calibration function for guaiacol β-d-glucopyranoside in wine. A calibration function for guaiacol β-d-glucopyranoside was constructed by plotting the peak area ratio of the target mass transition of guaiacol β-d-glucopyranoside (2) to that of its deuterated equivalent (8), against known concentrations of the glucoside, ranging from 10 to 100,000 µg/l, in a control rosé Grenache wine. This wine was made with grapes known to contain negligible levels of guaiacol glycoconjugates. A correlation coefficient of 0.998 was obtained, indicating a high degree of linearity for the working range (0-5000 µg/l) (Figure 5). As with the SIDA method developed for grape analysis (Chapter 4), the absence of labelled analogues for the glucosylglucoside, glucose-pentose disaccharides and rutinoside required their relative concentrations to be determined using the deuterated guaiacol β-dglucopyranoside (i.e. 8) as internal standard. Again a high degree of reproducibility in glycoconjugate measurements was observed (Table 5) which leads to the assumption that relative changes in glucosylglucoside, glucose-pentose disaccharides and rutinoside concentrations can be accurately determined, but direct comparison with glucoside levels are, at best, approximations. Chapter 5 61

Analyte peak area/is peak area 6.0 5.0 R² = 0.998 4.0 3.0 2.0 1.0 0.0 0 1000 2000 3000 4000 5000 Concentration (μg/l) Figure 5: Calibration function for guaiacol β-d-glucopyranoside in control rosé Grenache wine. 5.2.1.2 Mass transitions used for HPLC-SRM analysis. Using the deuterated guaiacol glucoside (8) as an internal standard, an HPLC-SRM based SIDA method for the direct quantification of guaiacol glucoside and relative quantification of the glucosylglucoside, glucose-pentose disaccharides and rutinoside in smoke affected wines was developed, i.e. as an adaptation of the method previously developed for use in extracts of smoke affected grapes (Chapter 4). The glycoconjugates again predominately gave the respective acetic acid adduct ([M-H + CH 3 COOH] ) ions under the APCI conditions employed, therefore quantification was carried out by HPLC-SRM, monitoring the mass transition from the respective [M-H + CH 3 COOH] ions to m/z 161 for the glucoside, m/z 293 for the glucose-pentose disaccharides, m/z 307 for the rutinoside or m/z 323 for the glucosylglucoside. Chapter 5 62

5.2.2 Method validation. 5.2.2.1 Instrument repeatability. Instrument repeatability was tested by repeating the analysis of a heavily smoke tainted Shiraz wine (5 replicates). Glycoconjugate concentrations were highly consistent, with coefficients of variation of less than 4% obtained for each (Table 5). 5.2.2.2 Reproducibility. Reproducibility of the method was evaluated by measuring the guaiacol β-dglucopyranoside concentration of five replicates of addition samples spiked with 50 or 1,000 μg/l of the glucoside. The method demonstrated a high level of consistency, with coefficients of variation of 2.8 and 1.6% respectively (Table 5). Reproducibility of glycoconjugate analysis was also evaluated by repeating the analysis of smoke affected Shiraz (5 replicates) and Merlot (4 replicates) wines. Analysis of the glycoconjugates demonstrated a high level of consistency with coefficients of variation between 0.7 and 5.0% (Table 5), for the glucoside, glucose-pentose disaccharides and rutinoside. The unusually high coefficients of variation obtained for the glucosylglucoside (i.e. 10.0 and 28.3% for Shiraz and Merlot respectively) are attributed to the extremely low levels present in these wines (i.e. 6 µg/l), compared with the other glycoconjugates. 5.2.2.3 Recovery. Glycoconjugate recovery was evaluated by comparing the glucoside content of a control Grenache red wine, and the same wine spiked with 1,000 µg/l of the guaiacol Chapter 5 63

β-d-glucopyranoside (i.e. the 1,000µg/L standard used for construction of the calibration function). Recovery was calculated to be 97%, which demonstrates the method can be applied to accurately quantify guaiacol β-d-glucopyranoside in wine samples (Table 5). Chapter 5 64

Table 5: Method validation for the quantification of guaiacol glycoconjugates in wine. Sample (a) Instrument repeatability Smoke affected Shiraz wine Mean a (μg/l) CV b (%) Glucosylglucoside 2 3.3 10 Glucoside 21 0.9 10 Glucose-pentose disaccharides 195 0.9 10 Rutinoside 45 1.4 10 n c (b) Reproducibility 50 μg/l addition d 64 2.8 5 1,000 μg/l addition d 982 1.6 5 Smoke affected Shiraz wine Glucosylglucoside 6 10.0 5 Glucoside 81 1.9 5 Glucose-pentose disaccharides 581 0.8 5 Rutinoside 113 0.7 5 Smoke affected Merlot wine Glucosylglucoside 6 28.3 4 Glucoside 45 2.5 4 Glucose-pentose disaccharides 657 4.5 4 Rutinoside 121 5.0 4 (c) Recovery Control Grenache red wine 11 4.4 5 Control Grenache red wine with 1,011 1,000 µg/l addition (expected) Control Grenache red wine with 982 1.6 5 1,000 µg/l addition (observed) Recovery (%) e 97 a In wine sample coefficient of variation number of replicates d Control Grenache rosé wine spiked with a known amount of guaiacol glucoside. e (observed/expected) x 100 Chapter 5 65

5.2.3 Application of wine based SIDA method to winemaking trials. 5.2.3.1 Hydrolysis of guaiacol glycoconjugates during fermentation. The concentration of guaiacol glycoconjugates was monitored throughout the fermentation of smoke affected grapes to investigate their hydrolysis during winemaking. Three seperate experiments were conducted using smoke affected Grenache, Shiraz and Viognier grapes. Grenache and Viognier grapes were obtained from grapevines exposed to experimentally produced smoke (for 20 or 30 min, respectively, i.e. a relatively short duration of smoke exposure), whereas Shiraz grapes were sourced from a vineyard exposed to bushfire smoke over a 5 week period (i.e. prolonged smoke exposure). Must from crushed Grenache and Shiraz grapes were inoculated with a commercial yeast strain (i.e. PDM), whereas the Viognier must was fermented using indigenous (or wild ) yeast. The smoke affected Grenache grapes contained 294 µg/kg total guaiacol glycoconjugates. Assuming a 70% juice extraction rate, 84 complete extraction of the glycoconjugate pool would result in juice glycoconjugate concentrations of approximately 420 µg/l. Instead, free run juice contained only 123 µg/l total glycoconjugates. Glycoconjugate levels increased to 197 µg/l after 1 day maceration and to 272 µg/l after 4 days maceration, but there was no significant change in precursor concentrations from then on (Table 6). Smoke-derived volatile phenols, including guaiacol, have been shown to evolve during fermentation, purportedly due to the hydrolysis of glycoconjugate precursors extracted from smoke affected fruit. 43 However, it is clear from the current study that a significant proportion of the glycoconjugate pool remains in the wine, after fermentation has been completed (Table 6). Chapter 5 66

Table 6: Concentration of guaiacol glycoconjugates throughout fermentation of smoke affected Grenache grapes, according to red style winemaking protocols. Total guaiacol Sample glycoconjugate concentration (µg/l) grapes a 294.2 ± 35.7 free run juice 123 ± 36.8 red winemaking after 1 day maceration 197 b ± 32.0 after 4 days maceration 272 c ± 39.8 end of alcoholic fermentation (i.e. post-pressing) 265 c ± 34.2 finished wine 290 c ± 37.0 a expressed as µg/kg Each value represents the mean of three replicates ± standard error. Means in columns followed by different letters are significantly different. Similar results were obtained during the fermentation of smoke affected Shiraz grapes. As expected, the increased duration of smoke exposure gave considerably higher grape glycoconjugate concentrations, being 875 µg/kg. A greater proportion of the glycoconjugate pool was extracted into the Shiraz must than occured for Grenache; i.e. approximately 1,000 µg/l of an estimated 1,250 µg/l maximum (Table 7). Again, there was no significant difference in glycoconjugate concentration during the first 7 days of maceration. However, a significant reduction in glycoconjugate levels had occured by the time fermentations underwent pressing, i.e. approximately 20%, presumably due to hydrolysis by yeast and/or enzymes. That said, as with the Grenache wines, the finished Shiraz wines still contained a large proportion of the initial glycoconjugate pool. Chapter 5 67

Table 7: Concentrations of guaiacol glycoconjugates throughout fermentation of smoke affected Shiraz grapes. Treatment Total glycoconjugates (µg/l) grapes a 875 ± 111.5 after 3 days maceration 1027 b ± 50.4 after 4 days maceration 1112 b ± 90.8 after 7 days maceration 1025 b ± 64.0 after alcoholic fermentation (post-pressing) 832 c ± 28.4 finished wine 825 c ± 29.2 a expressed as µg/kg Each value represents the mean of three replicates ± standard error. Means in columns followed by different letters are significantly different. Control and smoke affected Viognier grapes were fermented with indigenous yeast, to determine the effects of wild fermentation on total guaiacol glycoconjugate concentrations. Control Viognier grapes were found to contain a reasonable quantity of glycoconjugates, being 116 µg/kg, but after fermentation, only 26 µg/l remained in the resulting wine (Table 8). In contrast, smoke affected Viognier grapes contained 536 µg/kg glycoconjugates, but 197 µg/l remained after fermentation. Again, these results are consistent with those obtained in the trials involving Grenache and Shiraz, although much more variation was observed between the wild fermentation replicates, than the inoculated fermentations, as indicated by the significantly higher standard errors (Table 8). This is perhaps not unexpected, since populations of indigenous yeast will differ in species and cell number, causing the observed variations in fermentative ability between replicates. 87 These fermentations were also conducted on micro-scale (i.e. 250 ml) which likely gave much less controlled winemaking conditions; in particular, temperature. Irrespective, the results clearly demonstrate that guaiacol glycoconjugates can be hydrolysed during fermentation with indigenous yeast, but that again only partial metabolism occurs, so that glycoconjugates remain in the finished wine. Chapter 5 68

Table 8: Concentration of guaiacol glycoconjugates in control and smoke affected Viognier grapes and wine (produced by wild fermentation). Sample Total glycoconjugates (μg/l) Control Viognier grapes a 116 ± 9.3 Control Viognier wine 26 b ± 3.8 Smoke affected Viognier grapes a 536 ± 105.2 Smoke affected Viognier wine 197 c ± 74.4 a expressed as µg/kg Each grape value represents the mean of three replicates, while each wine value represents the mean of twelve replicates, ± standard error. Means in columns followed by different letters are significantly different. The presence of glycoconjugates in finished wines has important implications for the wine industry, since their hydrolysis in the bottle over time could result in liberation of additional quantities of guaiacol, and therefore the intensification of smoke related sensory attributes with ageing. This is considered in more detail, i.e. with a broader sample set, below (i.e. in 5.2.3.3). 5.2.3.2 Influence of winemaking techniques on the glycoconjugate content of wine. To investigate the effect of skin contact on guaiacol glycoconjugate concentration, a winemaking trial was conducted, in which traditional red and rosé winemaking techniques were compared. Red and rosé style wines were made from smoke affected and control grapes; with samples collected at various stages of fermentation, including pre-inoculation, during alcoholic fermentation, post pressing and bottling, and glycoconjugate concentrations determined using the wine based SIDA method. Glycoconjugate concentrations were significantly higher in smoke affected red style wines, compared to rosé style wines, although both contained elevated Chapter 5 69

Concentration (μg/l) glycoconjugate levels compared to their corresponding control wines (Figure 6). The lower glycoconjugate concentrations of rosé style wines is attributed to the reduced skin contact time, given glycoconjugates were shown to preferentially accumulate in the skins of grapes (Chapter 4). Wine style can therefore have a significant influence on the extraction of glycoconjugates and thus the extent of smoke taint. As such, winemaking practices need to be a consideration for winemakers when processing smoke affected grapes. 350 300 250 200 150 100 50 0 smoked red style wine control red style wine smoked rosé style wine control rosé style wine Figure 6: Guaiacol glycoconjugate concentrations of control and smoke affected Grenache wines made according to different winemaking techniques. Although determination of total guaiacol glycoconjugate concentrations provides an indication of the bound guaiacol content of wine, the relative concentrations of individual glycoconjugates in Grenache grapes and wine was also investigated. For both grapes and wines, the glycoconjugate pool largely comprised the glucosepentose disaccharides (55-65%); the rutinoside and glucoside were less abundant, at 20-30% and 6-10% respectively (Figure 7). The glucosylglucoside concentration decreased from 13% in grapes, to less than 1% in wine, suggesting that of the Chapter 5 70

various glycoconjugates, the glucosylglucoside is probably the most susceptible to hydrolysis during fermentation. It is possible that a proportion of the disaccharide glycoconjugates might be hydrolysed to the β-d-glucopyranoside. *The glycoconjugate concentration of grapes was converted from µg/kg to µg/l assuming a 70% juice extraction rate. Figure 7: Relative concentrations of guaiacol glycoconjugates of smoke affected Grenache grapes and resulting red and rosé style wines. The concentration of guaiacol glycoconjugates in smoked and control Grenache wines, fermented with eight different yeast strains, was measured using the wine based SIDA method. Control wines contained negligible concentrations of all guaiacol glycoconjugates. Smoke affected wines, fermented using AWRI 1176, showed the highest concentration of glycoconjugates (being 374 µg/l), followed by ICV GRE (being 356 µg/l); with the lowest concentration of glycoconjugates observed in wines fermented with AWRI 1503 (264 µg/l) and BDX (271 µg/l). β- Glucosidase enzyme, present in yeast and responsible for glycoconjugate Chapter 5 71