Integrative metabolomics-flavoromics to monitor dynamic changes of ‘Nam Dok Mai’ mango (Mangifera indica Linn) wine during fermentation and storage

Integrative metabolomics-flavoromics to monitor dynamic changes of ‘Nam Dok Mai’ mango (Mangifera indica Linn) wine during fermentation and storage

Journal Pre-proof Integrative metabolomics-flavoromics to monitor dynamic changes of ‘Nam Dok Mai’ mango (Mangifera indica Linn) wine during fermentat...

3MB Sizes 0 Downloads 6 Views

Journal Pre-proof Integrative metabolomics-flavoromics to monitor dynamic changes of ‘Nam Dok Mai’ mango (Mangifera indica Linn) wine during fermentation and storage Nutthapol Wattanakul, Sumallika Morakul, Yaowapa Lorjaroenphon, Kriskamol Na Jom PII:

S2212-4292(19)30381-5

DOI:

https://doi.org/10.1016/j.fbio.2020.100549

Reference:

FBIO 100549

To appear in:

Food Bioscience

Received Date: 24 April 2019 Revised Date:

25 February 2020

Accepted Date: 25 February 2020

Please cite this article as: Wattanakul N., Morakul S., Lorjaroenphon Y. & Jom K.N., Integrative metabolomics-flavoromics to monitor dynamic changes of ‘Nam Dok Mai’ mango (Mangifera indica Linn) wine during fermentation and storage, Food Bioscience (2020), doi: https://doi.org/10.1016/ j.fbio.2020.100549. This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier Ltd.

CRediT author statement Nutthapol Wattanakul: Formal analysis, Methodology, Writing- Original draft preparation. Sumallika Morakul: Resources, Visualization. Yaowapa Lorjaroenphon: Software, Validation. Kriskamol Na Jom: Conceptualization, Writing- Reviewing, Editing and Supervision.

1

Integrative metabolomics-flavoromics to monitor dynamic changes of ‘Nam Dok Mai’

2

mango (Mangifera indica Linn) wine during fermentation and storage

3 4

Nutthapol Wattanakula, Sumallika Morakulb, Yaowapa Lorjaroenphona, and Kriskamol

5

Na Joma*

6 7 8 9 10

a

Department of Food Science and Technology, Faculty of Agro-Industry, Kasetsart University, Bangkok, Thailand

11 12 13

b

Department of Biotechnology, Faculty of Agro-Industry, Kasetsart University, Bangkok, Thailand

14 15 16 17 18 19 20 21

* Correspondance to: Dr. Kriskamol Na Jom, Department of Food Science and Technology,

22

Faculty of Agro-Industry, Kasetsart University, Bangkok, 10900 Thailand, Telephone: +66 (0)

23

2562 5020, Fax: +66 (0) 2562 5021, email: [email protected]

1

24

Abstract

25

The ‘Nam Dok Mai’ variety is an important mango variety with a short shelf-life. Wine

26

production is one of the new opportunities to process such low-quality mangoes. ‘Nam Dok Mai’

27

mango juice with 22 °Brix and pH 4 was prepared. Laboratory scale fermentation was done at

28

20±2°C for 18 d. After fermentation was completed, the mango wine was bottled and stored at

29

10 to 12°C for 3 months. The pH of the mango wine decreased from 0 to 4 d and then remained

30

stable at ~pH 3 to the end of fermentation. The Brix sucrose of the product decreased from 22 to

31

3.6 °Brix at 12 d of fermentation and was stable thereafter. The alcohol percentage of the

32

samples increased to 13% by the last fermentation day and then dropped to 11% during storage.

33

Fifty-six compounds could be identified using metabolomic analysis accounting for ~40% of all

34

metabolites during fermentation and storage. Principal component analysis was used to

35

differentiate all samples based on the metabolites and flavor compounds observed. The finished

36

mango wine was separated from other samples along a PC biplot by acid compounds. Relative

37

targeted quantification using heat plots of the metabolite and flavor components showed

38

decreases in fatty acid methyl esters (FAME), free fatty acids, fatty alcohols, sugars, and

39

terpenes during fermentation. On the other hand, acids, glycerol, ethyl esters, and higher alcohols

40

increased during fermentation. Correlation analysis resulted in a strong positive relationship

41

between FAME with sterols, sugars, higher alcohols, ketones, aldehydes, terpenes, and volatile

42

acids.

43 44 45

2

46

Keywords: mango wine, metabolite profiles, flavor profiles, Nam Dok Mai, fruit wine analysis,

47

Mangifera indica Linn

48

1. Introduction

49

Mango (Mangifera indica Linn) is an important fruit worldwide. Mango production was

50

the seventh-highest in world fruit production behind bananas, grapes, oranges, apples, coconuts,

51

and plantains, with Thailand being third-ranked in 2014 behind India and China (Food and

52

Agriculture Organization of the United Nations, 2014). In Thailand, there are many mango

53

cultivars such as Maha Chanok, Thong Dam, Okrong Thong, Khiaw Sawoei, and Nam Dok Mai,

54

which are eaten as ripened fruit. The variety ‘Nam Dok Mai’ is popular not only with Thai

55

people, but also with international consumers because of its unique aroma and taste. It also

56

contains higher amounts of vitamin A, vitamin C, and α-carotene, which provide health benefits

57

(Reddy and Reddy, 2005). However, mangoes are climacteric fruits with a short shelf-life and

58

rapid deterioration (Tharanathan et al., 2006). The excess amounts are preserved in various

59

manners, including stirred mango sheets, dried mango slices, canned mango, and mango pickles.

60

Wine production is one opportunity to improve the value of low-quality mangoes (Reddy et al.,

61

2014). However, there are many factors affecting mango wine characteristics. The effects of

62

fermentation temperature and SO2 addition have been reported by Reddy and Reddy (2011), with

63

the fermentation temperature affecting not only the duration and rate of yeast fermentation, but

64

also the yeast metabolism, which combine to define the chemical composition and the quality of

65

the wine. Furthermore, the addition of SO2 can affect volatile compound synthesis during

66

fermentation such as acetaldehyde formation (Liu and Pilone, 2000). Not only the conditions for

67

wine-making, but also the yeast strains can influence the character of mango wine. Changes in

3

68

mango wine characteristics following fermentation using three different Saccharomyces

69

cerevisiae yeast stains have been reported (Li et al., 2012).

70

Metabolomics is one of the omic family that focuses on small molecule metabolites from

71

cells, tissues, organs, and organisms. Metabolites can be divided into primary metabolites, that

72

are essential to sustain the life of every organism, while secondary metabolites are not essential

73

but are necessary for survival in any stressed environment (Limwiwattana et al., 2016). Changes

74

in metabolites are relative to changes in the proteome, transcriptome, and genome.

75

Not only metabolic changes, but also the flavor changes could be observed during the

76

mango wine process. Flavoromics, which is also called flavor omics or flavor metabolomics, is

77

used to study targeted and non-targeted flavor compounds in food. Consequently, the

78

flavoromics approach provides an overview of the chemical components involved in flavor

79

formation (Charve, 2011). Therefore, metabolomics and flavoromics might be applied to wines

80

to study changes in chemical components during fermentation and storage.

81

The purposes of this study were to investigate changes in metabolites together with flavor

82

components during ‘Nam Dok Mai’ mango wine fermentation using a combination of

83

metabolomics and flavoromics techniques.

84 85

2. Materials and Methods

86

2.1. Chemicals

87

Chemicals for the extraction and derivatization, including hydroxy ammonium chloride,

88

methyl tertiary-butyl ether, pyridine, and sodium-methylate were HPLC grade purchased from

89

Sigma Aldrich Co. (St. Louis, MO, USA). N-Methyl-N-(trimethylsilyl) trifluoroacetamide

90

(MSTFA) and N-trimethylsilylimidazole (TMSIM) were GC derivatization grade purchased

4

91

from Sigma Aldrich. Acetonitrile, dichloromethane, hexane, and methanol were analytical grade

92

purchased from RCI Labscan Ltd. (Pathumwan, Bangkok, Thailand). All internal standards for

93

metabolomics, including tetracosane, 5α-cholestane-3β-ol, phenyl-β-D-glucopyranoside, and ρ-

94

chloro-L-phenylalanine were standard grade purchased from Sigma Aldrich. Reference standards

95

for metabolomics, including lauric acid methyl ester (12:0 FAME), myristic acid methyl ester

96

(14:0 FAME), pentadecanoic acid methyl ester (15:0 FAME), pentadecenoic acid methyl ester

97

15:1 FAME, palmitic acid methyl ester (16:0 FAME), palmitoleic acid methyl ester (16:1

98

FAME), steric acid methyl ester (18:0 FAME), oleic acid methyl ester (18:1 FAME), linoleic

99

acid methyl ester (18:2 FAME), linolenic acid methyl ester (18:3 FAME), erucic acid methyl

100

ester (22:1 FAME), tricosanoic acid methyl ester (23:0 FAME), nonanoic acid (9:0 FFA), lauric

101

acid (12:0 FFA), methyl-p-hydroxy cinnamate, methyl ferulate, hexadecanol (16:0-OH), steric

102

acid (18:0 FFA), oleic acid (18:1 FFA), linoleic acid (18:2 FFA), linolenic acid (18:3 FFA),

103

arachidyl alcohol (20:0-OH), linoleyl alcohol (9,12-OH 18:0), behenyl alcohol (22:0-OH), 1-

104

octacosanol (28:0-OH), campesterol, β-sitosterol, sitostanol, stigmasterol, gramisterol, ∆7-

105

avenasterol, citrostradienol, glycerol, fructose, glucose, mannitol, sorbitol, myo-inositol, sucrose,

106

trehalose, glycine, 4-hydroxybutyric acid, leucine, isoleucine, proline, succinic acid, fumaric

107

acid, threonine, β-aminoisobutyric acid, malic acid, pyroglutamic acid, γ-aminobutyric acid,

108

threonic acid, glutamine, citric acid, histidine, and tyrosine were purchased from Sigma Aldrich.

109

A C6-C30 n-alkane mixture used for determination of linear retention indices was purchased

110

from Sigma-Aldrich. Reference standards for flavoromics, including isobutyl acetate, butanoic

111

acid ethyl ester, 1-propanol, 2-methyl-1-propanol (isobutanol), 1-butanol-3-methyl acetate

112

(isoamyl acetate), 3-methyl-1-butanol (isoamyl alcohol), hexanoic acid ethyl ester, acetic acid

113

hexyl ester, acetoin, 3-hexen-1-ol acetate, 6-methyl-5-hepten-2-one, 3-hexen-1-ol, octanoic acid

5

114

methyl ester, isopentyl hexanoate, 2-ethyl-1-hexanol, benzaldehyde, nonanoic acid ethyl ester,

115

linalool, 1-octanol, 2-methyl propanoic acid (isobutyric acid), caryophyllene, butanoic acid,

116

acetophenone, decanoic acid ethyl ester, octanoic acid 3-methylbutyl ester, humulene, ethyl 9-

117

decenoate, acetic acid 2-phenylethyl ester, dodecanoic acid ethyl ester, phenylethyl alcohol,

118

trans-β-ionone, octanoic acid, and hexadecanoic acid ethyl ester were standard grade purchased

119

from Sigma Aldrich. Citronellol and geraniol analytical standards were purchased from

120

Bedoukian (Danbury, CT, USA). Benzyl alcohol was purchased from J.T. Baker (Phillipsburg,

121

NJ, USA).

122

2.2. Selection of ‘Nam Dok Mai’ mango

123

Ten kg of ‘Nam Dok Mai’ mangoes (Mangifera indica L.) were sourced from a

124

commercial mango orchard (certified by the Department of Agriculture, Ministry of Agriculture

125

and Cooperatives of Thailand) in Kaeng Khoi district, Saraburi province in central Thailand.

126

They were all harvested in May 2017 after 90 to 100 d of the fruit-set stage and immediately

127

transported to Kasetsart University, Bangkok, Thailand with controlled temperature (4°C) in a

128

refrigerated truck. The transportation time was ~1 h. Mangoes were selected with weights of 400

129

to 420 g, lengths of 14 to 16 cm and widths of 7 to 9 cm. Their densities were determined using

130

floatation in a 3% NaCl solution and sinking in 1% NaCl for their homogeneity maturity index in

131

term of a specific gravity value (Sriwimon and Boonsupthip, 2011). The selected samples were

132

kept at 30±2°C and a relative humidity of 75±2% for ripening. Samples were taken after 8 d of

133

ripening. Mangoes were washed with tap water, peeled by hand with a knife, and cut into small

134

pieces (~2 cm3) before being stored at -20°C for a maximum of 3 wk.

135

2.3. Mango wine-making

6

136

The mango wine-making process was modified from Li et al. (2012a). Frozen mango was

137

thawed at room temperature (25±2°C) for 1 h. After puréeing the mango using a spatula without

138

filtration, tap water was added to make 30% w/w mango juice called must. A sample of 5 L was

139

adjusted to 22 °Brix by adding sucrose and adjusted to pH 4 using 50% w/w citric acid solution.

140

Then, 200 mg/mL potassium metabisulfite (KMS) and 500 mg/mL diammonium phosphate

141

(DAP) were added and the mixture was left in a 7 L fermentation tank for 1 d at room

142

temperature. Five g of dry commercial wine yeast, Saccharomyces bayanus Lalvin EC 1118

143

(Lallemand Inc., Montreal, Quebec, Canada) were activated in 20 mL of tap water at 35°C for 15

144

min and then 4 mL of yeast solution was inoculated into the must. Fermentation was done in

145

quadruplicates at 20±2°C for 18 d with 100 mL sampling every 2 d. After fermentation, the

146

mango wine tanks were stabilized in a chilled controlled temperature room (0-4°C) for 2 wk.

147

Stabilized wine was filtered using a cellulose membrane (Whatman filter paper No. 4, pore size

148

20-25 µm, Whatman International Ltd., Maidstone, UK) and filled into 200 mL green bottles

149

then closed with sterilized screw caps. All wine bottles were stored in a controlled temperature

150

room at 10-12°C (the recommended storage temperature for commercial dry white wine) for 3

151

months. Samples were taken every month.

152

2.4. Physico-chemical analysis of mango juice and wine

153

Total soluble solids as °Brix were measured using a refractometer (DR6000, Krüss,

154

Hamburg, Germany) (Sadineni et al., 2012). The pH was measured using a pH meter (Model

155

3510, Jenway, West Thurrock, UK) (Garde-Cerdán and Ancín-Azpilicueta, 2008). The

156

%titratable acidity was measured using 0.1 M NaOH solution and phenolphthalein was used as

157

an indicator. The results were shown as g citric acid/100 mL mango wine according to AOAC

158

(1990) method number 942.15. The total sugar content was measured using the phenol-sulfuric

7

159

method (Dubois et al., 1956). Sample or glucose standard solution (2 mL) was mixed with 1 mL

160

5% phenol solution and 5 mL 96% sulphuric acid. The mixture was mixed using a Vortex mixer

161

(VX100, Labnet International Inc., Edison, NJ, USA) and allowed to stand at room temperature

162

for 30 min. The absorbance was measured at 490 nm using a spectrophotometer (Genesys 10S

163

UV-Vis, Thermo Fisher Scientific Inc., Waltham, MA, USA). The reducing sugar content,

164

including glucose and fructose, was measured using the dinitro salicylic acid (DNS) method

165

modified from Miller (1959) and Zeng et al. (2018). Sample or glucose standard solution (1 mL)

166

was mixed with 1 mL DNS reagent. The mixture was incubated in a boiling water bath for 5 min.

167

After cooling in an ice bath for 5 min, the absorbance was measured at 540 nm. The alcohol

168

percentage was measured using the dichromate method modified from Zoecklein et al. (1995).

169

The wine (1 mL) was distilled into 25 mL acidified dichromate at 60±2°C for 25 min. The

170

solution was titrated with ferrous ammonium sulphate until the color changes from emerald

171

green to brown.

172

2.5. Metabolomics

173

2.5.1. Sample extraction and fractionation

174

The extraction was modified from Limwiwattana et al. (2016). Both non-polar (lipid) and

175

polar low molecular weight (MW) compounds could be determined. A sample of 10 mL of the

176

wine was obtained using liquid-liquid extraction by adding 10 mL of methanol:water solution

177

(80:20 v/v) and 4 mL of dichloromethane. Then, 100 µL 1 mg/mL tetracosane and 100 µL 1

178

mg/mL 5α-cholestane-3β-ol were added as the internal standard for the lipid fraction. For the

179

polar fraction, 250 µL 1 mg/mL phenyl-β-D-glucopyranoside and 250 µL 1 mg/mL ρ-chloro-L-

180

phenylalanine were added. The solution was mixed and sonicated in an ultrasonic bath (Model

8

181

575T, 400 W, Crest Ultrasonics Corp., Ewing Township, NJ, USA) for 15 min at room

182

temperature (25±2°C) before separation of the two phases.

183

2.5.2. Lipid fractionation

184

Before fractionation, 1.5 mL of lipid extract was taken from the dichloromethane phase

185

and evaporated until dry using a parallel evaporator (Syncore Polyvap, Buchi, Flawil,

186

Switzerland). Then re-dissolved with 500 µL methyl tertiary-butyl ether, 300 µL methanol, and

187

50 µL 5.4 M sodium-methylate in methanol for transesterification. After reaction for 90 min in

188

the dark at room temperature, selective hydrolysis was done by adding 1 mL dichloromethane

189

and 2 mL 0.35 M HCl solution. The HCl phase was removed and the lower phase containing the

190

transmethylated lipid was evaporated until dry using the parallel evaporator at 50°C and then re-

191

dissolved with dichloromethane. Transmethylated lipids, i.e., FAME and polar lipids were

192

fractionated using elution with different concentrations of hexane:methyl tertiary-butyl ether

193

(hexane:MTBE) solution in a solid-phase microextraction C18-LP cartridge (Vertical

194

Chromatography Co., Ltd., Nonthaburi, Thailand). The FAME fraction (fraction 1) was eluted

195

with 6 mL 100:2 v/v hexane:MTBE solution and then evaporated until dry. The residue was re-

196

dissolved using 300 µL hexane and 1 µL was used for GC. The polar lipids fraction (fraction 2),

197

which still remained in the cartridge, was eluted with 6 mL 70:30 v/v hexane:MTBE solution

198

and then evaporated until dry. The residue of fraction 2 was re-dissolved using 250 µL pyridine

199

and 50 µL MSTFA. After argon flushing, silylation was done in a water bath at 70°C for 30 min

200

and 1 µL was measured using GC. All lipid fractions were immediately stored at -20°C until GC

201

analysis. The GC analysis was done within 2 wk.

202

2.5.3. Polar fractionation

9

203

The polar extract was fractionated into two fractions of sugars and acids. A sample of 200

204

µL polar extract was taken from the 80:20 v/v methanol:water solution and silylated using 100

205

µL TMSIM as a silylating agent and 300 µL pyridine. Silylation was done in a water bath at

206

70°C for 15 min. Then, the solution was diluted with 300 µL hexane and selective hydrolysis of

207

amino acids and fatty acids was done adding 300 µL deionized water (generated from Purelab

208

Option, ELGA LabWater Co., High Wycombe, UK). The upper phase of hexane was collected,

209

which contained silylated sugars and sugar alcohols (fraction 3) and made ready for GC analysis.

210

For the acid fractionation (fraction 4), 200 µL polar extract was oximated by adding 300 µL 2

211

mg/mL hydroxy ammonium chloride in pyridine. After oximation at 70°C for 30 min, 100 µL N-

212

methyl-N-(trimethylsilyl) fluoroacetamide was added for silylation. The sample was silylated in

213

a water bath at 70°C for 15 min. Selective hydrolysis was done by adding 500 µL hexane and

214

300 µL deionized water. The upper phase of hexane containing silylated sugars was removed and

215

the lower phase containing amino and organic acids was re-silylated adding 50 µL MSTFA in

216

200 µL acetonitrile. Silylation was done in a water bath at 70°C for 60 min. The sample was

217

ready for GC analysis.

218

2.5.4. GC-FID analysis

219

All samples were measured using GC coupled with a flame ionization detector (Hewlett

220

Packard, Palo Alto, CA, USA). The separation was done using a DB-1 capillary column (60 m ×

221

0.32 mm × 0.25 µm film thickness) with a 100% dimethylpolysiloxane stationary phase (Agilent

222

Technologies, Santa Clara, CA, USA). Each sample of 1 µL was injected into the GC-FID in

223

splitless mode and helium was used as the carrier gas at a constant flow rate of 1.8 mL/min. The

224

inlet temperature was 280°C. The oven temperature program started at 100°C, then ramped to

225

320°C at 4°C/min and was held at 320°C for 25 min. The detector temperature was 320°C.

10

226

2.6. Flavoromics

227

2.6.1. Sample extraction

228

The extraction was modified from Torrens et al. (2004). The extraction was done using

229

the autosampler headspace solid-phase microextraction (HS-SPME) method. A sample of 5 mL

230

was spiked with 10 µL 0.1 mg/mL methyl nonanoate in the methanol as an internal standard.

231

Then, 1 g NaCl was added to increase volatility in the headspace. The sample was automatically

232

equilibrated at 50°C for 20 min. A 50/30 µm divinylbenzene/carboxen/polydimethylbenzene

233

(DVB/CAR/PDMS) fiber (Supelco, Bellefonte, PA, USA) was inserted and exposed to the

234

headspace with a penetration length of 10 mm at 50°C for 40 min. After the absorption step, the

235

fiber was desorbed in the GC injection port at 250°C for 5 min.

236

2.6.2. GC-ToF-MS analysis

237

All volatile compounds were measured and identified using a GC 7890A coupled with a

238

time of flight mass spectrometer (Agilent Technologies). The separation was done using a

239

Stabilwax fused silica column (30 m × 0.25 mm × 0.25 µm film thickness) with a cross-bond

240

polyethylene glycol stationary phase (Restek, Bellefonte, PA, USA). The temperature program

241

started at 40°C for 5 min and was ramped at 4°C/min to 225°C and then held for 15 min. For

242

mass spectrometry, the ion source temperature was 200°C and the transfer line temperature was

243

225°C. Mass spectra were scanned at -70 eV in the m/z range of 35 to 500 atomic mass units

244

(amu). Identification and quantification of flavor compounds were processed using the

245

ChromaTOF-GC Software v4.50.8.0 (Leco, St. Joseph, MI, USA).

246

2.7. Statistical data analysis

247

Data of all physicochemical measurements were obtained from 4 replications and

248

reported as means±standard deviation (SD). One-way analysis of variance (ANOVA) was done

11

249

using the International Business Machines’ Statistical Package for the Social Sciences (SPSS)

250

software version 21 (SPSS Inc., Chicago, IL, USA). Significant differences between treatments

251

were calculated at the 95% confidence level (p≤0.05) using Duncan’s multiple range test.

252

The peak areas of metabolites were integrated using the HP-ChemStation A.06.03

253

program (Hewlett Packard). Metabolites were identified using the reference standards

254

comparison technique. The concentration of each identified metabolites as mg/mL related to the

255

internal standard of each fractions was reported.

256

The tentative identification of targeted compounds related to the mango wine was done

257

using their mass spectra, linear retention indices (RI), and authentic standards. Mass spectra were

258

compared with the NIST mass spectral database version 2.0 (National Institute of Standards and

259

Technology, Gaithersburg, MD, USA). RI were computed using the retention time of the n-

260

alkanes series (C6-C30) and compared with RI data found in literatures that used the same GC

261

column polarity (Li et al., 2012b; Li et al., 2013b; Pino et al., 2005). Some tentatively identified

262

flavor compounds were confirmed using the authentic standards. Relative concentration

263

(mg/mL) of all flavor compounds was quantified using the internal standard technique. The

264

calibration curve of each standard compounds and internal standard showed good linearity (R2

265

>0.98).

266

Relative concentration of all metabolites and flavor compounds were subjected to

267

principal component analysis (PCA) and agglomerative hierarchical clustering analysis (AHC)

268

using the XLSTAT-base version 2018.3 software (Addinsoft, New York, NY, USA). Correlation

269

network analysis using non-parametric Spearman’s rank correlation of all metabolites and flavor

270

compounds was done at a significance level of p≤0.05 with the correlation/association tests mode

271

in the XLSTAT-base software.

12

272 273 274

3. Results and Discussion 3.1. Physicochemical attributes of mango juice and wine

275

The physicochemical results are shown in Table 1. During the ‘Nam Dok Mai’ mango

276

wine fermentation, the Brix value decreased significantly. The Brix changes were confirmed by

277

measurement of sugar concentrations. The total sugar content of the wine decreased

278

continuously, whereas the reducing sugar content increased until 25% fermentation and then

279

decreased to 0.86 mg/mL at the end of fermentation. A similar result was reported in other

280

varieties of Indian mango wine, made from Banginapalli, Alphonso, Totapuri, and Raspuri

281

varieties (Reddy and Reddy, 2009). This may have been due to yeast metabolism, with the yeast

282

hydrolyzing sucrose to become glucose and fructose in the initial stage. In the following stage,

283

the yeast used glucose and fructose as substrates to produce alcohol (Ribéreau-Gayon et al.,

284

2006a).

285

There was a rapid decrease in the pH during 0-25% fermentation of the wine (from

286

21.7 down to 16.6 °Brix) and then a drop to ~pH 3.50 at the end of fermentation. On the other

287

hand, there was an increase in the %titratable acidity of mango wine. This increase in the

288

%titratable acidity might have resulted from the products of the yeast’s metabolism (succinic and

289

α-ketoglutaric acid) during alcoholic fermentation (Berenguer et al., 2016). The increasing

290

%titratable acidity was continued through the initiation step of fermentation (0-25%) then

291

slightly increased until the end of fermentation. This might have been due to the buffer effect

292

when the organic acids interacted with amino acids. Therefore, after 25% fermentation, the

293

change in the %titratable acidity and the pH would be stable (Ribéreau-Gayon et al., 2006a).

13

294

An acceptable mango vinification process was indicated using the alcohol percentage

295

during fermentation. The alcohol percentage rose until the end of fermentation to a value of 13%.

296

This final alcohol content was similar to pomegranate wine (Berenguer et al., 2016) and

297

Banginapalli mango wine, which was produced by co-fermentation with S. cerevisiae and M.

298

pulcherrima (Sadineni et al., 2012). The final alcohol content of 13% v/v in the wine from the

299

initial total sugar content of 216 mg/mL could indicate the end of fermentation. A theoretical

300

conversion of 106 mg CO2 and 110 mg ethanol from 216 mg sugar would be expected during

301

wine fermentation (Boulton et al., 1999). Therefore, 216 mg/mL total sugar content should be

302

converted to ~14% v/v alcohol in the mango wine. However, 13% v/v alcohol in the wine from

303

these results were equal to 93% of expected yield because some sugars were used to produce cell

304

mass, glycerol, and other products. Similar results have been reported with Pasteur’s original

305

experiment, and Bioletti’s studies, in which, 92-95% yield of alcohol were obtained from the

306

fermentation (Boulton et al., 1999).

307

During the storage period, the Brix increased after 2 wk. There was no significant

308

difference until 3 months of storage. The pH of the wine remained stable during storage. The

309

%titratable acidity decreased from 0.41 to 0.37%, which might have been due to the decrease in

310

tartaric acid, which was one of the main acids in mango during the cold stabilization and

311

filtrating steps. Tartrate salt was precipitated at chilling temperature and separated at the

312

filtration step before wine bottling (Gao and Fleet, 1995).

313

The alcohol content dropped during the cold stabilization and filtration. The final alcohol

314

content of the wine during storage was ~11% v/v. These results were different from those

315

reported for Indian mango wine made from Banginapalli, Alphonso and Neelum varieties, which

316

were 8.5, 7.5, and 6.5% v/v alcohol, respectively (Reddy and Reddy, 2005), and 6.2-7.7% v/v

14

317

alcohol for durian wine (Lu et al., 2015). The lower alcohol percentage in the fruit wines might

318

have resulted from a lower initial sugar content.

319

3.2. Metabolomics and flavoromics of mango wine

320

For the metabolomics results, there were more than 150 chromatography peaks, including

321

unidentified and identified peaks. There were 56 peaks that could be identified; these accounted

322

for ~40% of all metabolites in the mango wine. All 56 metabolite compounds could be divided

323

into 4 main groups: 1) 13 FAME compounds; 2) 20 polar lipid compounds (FFA, fatty alcohols,

324

and phytosterols); 3) 8 sugar compounds (organic sugars and sugar alcohols); and 4) 15 acid

325

compounds (organic acids and amino acids). The flavoromics study gave more than 300 peaks

326

from the total chromatograms. Among these peaks, 73 compounds could be identified,

327

accounting for ~25% of all flavor components in the mango wine. All 73 flavor compounds

328

could be arranged into 8 groups: 31 esters, 20 alcohols, 7 ketones and aldehydes, 7 terpenes, 4

329

volatile acids, 2 furans, 1 volatile phenol, and 1 lactone. There were ~230 unknown volatile

330

compounds. All metabolic and flavor compounds were subjected to PCA. A PCA biplot (Figure

331

1) showed the metabolites and flavor compounds responsible for sample differentiation using the

332

mango wine processing period shifted along the PC1 and PC2 axes representing ~60% of total

333

variables. The PCA biplot was used to determine the grouping of samples based on all

334

compounds observed in the ‘Nam Dok Mai’ mango wine. The first group was pre-fermentation,

335

consisting of 0% fermentation samples, which was separated from the other groups and

336

contained 15 metabolites and 8 flavor compounds. The second was the in-process group,

337

representing 25-75% fermentation samples. There were 16 compounds in this group, consisting

338

of 4 metabolites and 12 flavor compounds. The third group was post-fermentation, which

15

339

consisted of 100% fermentation final products and 3 months storage products. There were 22

340

compounds in this group, consisting of 9 metabolites and 12 flavor compounds.

341

Agglomerative hierarchical clustering analysis (AHC) in similarity mode was done as

342

shown in Figure 2 and resulted in the wine being divided into three groups by processing period

343

as previously shown in the PC biplot. The dendrogram differentiated three groups as: the group

344

of pre-fermentation samples, the group of in-process and the group of post-fermentation or final

345

products.

346

Heat plots based on relative quantification were used to investigate more specific trends

347

of compounds during mango wine processing. Heat plots of all metabolites (Figure 3) showed a

348

decrease in the FAME content during fermentation and storage. However, the FFA concentration

349

increased due to FA synthesis from acetyl-CoA during the yeast fermentation. The main FA

350

observed in the ‘Nam Dok Mai’ mango pulp were palmitic acid (16:0), palmitoleic acid (16:1),

351

steric acid (18:0), oleic acid (18:1), linoleic acid (18:2), and linolenic acid (18:3), which were

352

also reported in Alphonso and Pairi mango pulp (Deshpande et al., 2016). Unsaturated long-

353

chain FA are anaerobic fermentation activators. The most important of these are oleic and

354

linoleic acid (Ribéreau-Gayon et al., 2006a). Moreover, they react with alcohol by esterification,

355

resulting in the formation of ester compounds. The formation of esters continues throughout the

356

aging process due to the presence of various acids in wine, along with large quantities of ethanol

357

(Ribéreau-Gayon et al., 2006a).

358

Most sugars, including sucrose, glucose, and fructose were depleted by wine yeast during

359

their alcoholic metabolism. Sugars can be a potential substrate for the formation of phenols and

360

aromatic amino acids. Moreover, mannitol could be used as a wine quality indicator. Mannitol

361

results from the metabolism of fructose by lactic acid bacteria (LAB) (Pilone et al., 1991).

16

362

Therefore, high amounts of mannitol could indicate LAB contamination. However, the mannitol

363

concentration in the current study decreased during the wine-making period, indicating no LAB

364

contamination. On the other hand, organic acids, such as citric acid, which was one of the main

365

acids observed in mangoes and mango wine, increased during the fermentation process due to

366

the aerobic glycolysis channel. The malic acid concentration remained unchanged and lactic acid

367

was not observed during the wine processing. This could indicate that malolactic fermentation

368

did not occur (Li et al., 2011). An increase in glycerol was observed during fermentation, which

369

was perhaps due to it being the end product of glycerol pyruvic fermentation (Ribéreau-Gayon et

370

al., 2006b).

371

Volatile esters are one of the major aroma groups in the wine, including acetate and ethyl

372

esters. The most abundant acetate esters observed were 1-butanol-3-methyl acetate (isoamyl

373

acetate), isobutyl acetate, acetic acid 2-phenylethyl ester (2-phenylethyl acetate), and octyl

374

acetate. Acetate esters resulted from the reaction of alcohols and acetyl-CoA. Alcohols are

375

produced with amino acid metabolism and acetyl-CoA is derived from pyruvate by yeast

376

metabolism (Lee et al., 2004; Sofie et al., 2010). However, ethyl esters are produced from

377

medium- and long-chain FA, which originated from the degradation of acyl-CoA interacting

378

with the hydroxyl group of alcohols (Alves et al., 2010; Saerens et al., 2008). Various ethyl

379

esters were observed. The dominant ethyl esters in the product were ethyl butanoate, ethyl

380

hexanoate, ethyl octanoate, ethyl nonanoate, ethyl decanoate, ethyl dodecanoate, and ethyl

381

hexadecanoate. The increasing amounts of ethyl esters are shown in Figure 3. This might have

382

been due to the decrease in medium- and long-chain FA, which are precursors of ethyl ester

383

formation. The formation of esters is also dependent on yeast strains, fermentation conditions,

384

and the variety of mango (Quilter et al., 2003).

17

385

The second largest group of volatile compounds observed was alcohols. The results

386

showed that alcohols in the must included 3-hexen-1-ol and 1-penten-3-ol, which decreased

387

during fermentation. On the other hand, levels of ethanol and other higher alcohols (1-propanol,

388

isoamyl alcohol, and phenylethyl alcohol) increased due to their formation using the Ehrlich

389

pathway with the amino acids during fermentation. In the Ehrlich pathway, amino acids, such as

390

leucine and valine, were trans-aminated to α-keto acids, then decarboxylated to aldehydes and

391

finally passed through the reduction reaction to become higher alcohols through NADH-

392

dependent chemical reactions (Vuralhan et al., 2003). However, isoamyl acetate, 2-phenylethyl

393

acetate, and other branched-chain esters in the wine could be produced from higher alcohols

394

Most terpene levels decreased during fermentation. Monoterpenes (trans-β-ocimene)

395

occurred in a simple hydrocarbon form, which decreased a few days after the fermentation had

396

commenced. Sesquiterpenes, such as caryophyllene and humulene, decreased more slowly than

397

the monoterpenes. A similar result has been reported by Li et al. (2012a). However, isoprenoid

398

monoterpene (citronellol), as well as C13-norisoprenoid (i.e., trans-β-ionone) increased during

399

the fermentation process. These increases indicated that the compounds had formed from a

400

mevalonate precursor, which was derived from acetyl-CoA in the wine yeast (Iriti and Faoro,

401

2006). Geraniol is also a dominant volatile compound in the mango juice. However, it decreased

402

during fermentation of the wine. This might have been due to the geraniol being converted to

403

citronellol and other terpenols by S. cerevisiae (Li et al., 2013a).

404

The volatile acids observed were acetic, butanoic, and octanoic acid. Butanoic and

405

octanoic acid decreased throughout the mango wine processing period as substrates for

406

esterification. Li et al. (2012a) also reported that the ‘Nam Dok Mai’ mango wine contained

407

higher octanoic acid content than mango wines made from the R2E2 variety from Australia and

18

408

the Harum Manis variety from Indonesia. On the other hand, the level of acetic acid increased

409

during wine processing. This might have been due to the wine being exposed to air during

410

fermentation (Joyeux, 1984).

411

Of the other miscellaneous volatile compounds, dihydro-2-methyl-3(2H)-thiophenone

412

(sulfur ketone) increased in the fermentation and tended to drop during storage. However, the

413

compound was not observed in the mango juice. The appearance of sulfur-containing compounds

414

might be due to KMS, which was added into the wine must before fermentation as an antioxidant

415

and antimicrobial agent. Moreover, it could be formed by metabolism of sulfur-containing amino

416

acids by wine yeast during fermentation. However, the level of dihydro-2-methyl-3(2H)-

417

thiophenone decreased during aging (Masuda and Nishimura, 1982).

418

Pairwise Spearman’s rank correlation analysis was done to investigate the relationship

419

between metabolites and flavor compounds during the wine fermentation and storage (Figure 4).

420

Rank order correlation was selected for the analysis, because the data were not bivariate and

421

normally distributed. These results showed that there were 8,385 correlative pairs, with 2,328

422

pairs having significant (p≤0.05) correlation, of which 1,242 pairs had a strong positive

423

correlation (r >0.7) and 1,086 pairs had a strong negative correlation (r <-0.7), while 138 pairs

424

had no correlation (-0.3< r <0.3). Four groups of metabolic constituents in relation to flavor

425

compounds have been observed. FAME (e.g., 16:0, 16:1, 18:0, 18:1, 18:2, 18:3, and 22:1) were

426

significantly and strongly positively correlated with sterols, sugars, and flavor groups (higher

427

alcohols, ketones, aldehydes, terpenes, and volatile acids). However, they were significantly and

428

strongly negatively correlated with acids, esters and some flavor compounds, such as ethanol,

429

phenylethyl alcohol, and acetic acid. On the other hand, there was no significant correlation

430

between the FAME and polar lipids. The polar lipids and phytosterols were significantly and

19

431

strongly negatively correlated with esters and other minor flavor groups (ketones, aldehydes, and

432

terpenes). On the other hand, glycerol had a significantly and strongly positive correlation with

433

acids and esters. This might be due to those compounds being products from yeast fermentation.

434

Sugars, including sucrose, glucose, and fructose, were significantly and strongly negatively

435

correlated with esters and ethanol. This could be the reason that ethanol formation from sugars

436

related to ester fabrication. On the other hand, they were significantly and strongly positively

437

correlated with higher alcohols, terpenes, and volatile acids. Acid compounds, including amino

438

and organic acids, were correlated positively with FFA, esters, and ethanol, but negatively

439

correlated with higher alcohols, volatile acids and other groups of flavor compounds. This might

440

be due to amino acids being substrates for the production of higher alcohols and volatile acids

441

using the Ehrlich pathway. These results might reflect the relationship between metabolites and

442

flavor compounds during the fermentation of the wine. In addition, important metabolites would

443

suggest that they were responsible for the formation of fragrant flavor compounds during the

444

production of the wine. The results might suggest information for the control of these factors and

445

the conditions of mango wine-making, to produce ‘Nam Dok Mai’ mango wine with a special

446

aroma and high levels of bioactive compounds.

447 448

4. Conclusions

449

Mango wine production from the ‘Nam Dok Mai’ variety of mango was successfully

450

prepared at the laboratory scale with the same characteristics as a dry white wine. Changes in the

451

physicochemical properties, metabolomics, and flavoromics were investigated to determine the

452

overall characteristics of the ‘Nam Dok Mai’ mango wine during processing and storage.

453

Multivariate and correlation analysis contributed to a better understanding of the interactions

20

454

between the metabolites and flavor constituents in the wine. The technique could be used to mark

455

important bioactive compounds, flavors, and characters of the wine, which might be used to

456

improve the quality of Asian mango wine processing in the future. Oleic (18:1) and linoleic

457

(18:2) acid could be used as biomarkers for monitoring anaerobic fermentation, whereas, malic

458

and lactic acid as biomarkers for malolactic fermentation. Mannitol could be used as a biomarker

459

for LAB contamination. Acetic acid and ethyl acetate could be used as biomarkers for improper

460

fermentation. The integration of metabolomics and flavoromics techniques provided an effective

461

alternative tool for comprehensive monitoring of the dynamic changes during mango wine

462

fermentation and storage.

463 464 465 466

Acknowledgements

467

Financial support for this study was provided by the Graduate School of Kasetsart

468

University, Bangkok, Thailand and by the Center for Advanced Studies for Agriculture and Food

469

(CASAF). English proof-reading by a native expert was supported by Kasetsart University

470

Research and Development Institute (KURDI).

471 472 473

Conflict of interest The authors declare that there are no conflicts of interest with respect to the publication.

474 475

References

476

Alves, J.A., Lima, L.C.O., Dias, D.R., Nunes, C.A., & Schwan, R.F. (2010). Effects of

21

477

spontaneous and inoculated fermentation on the volatile profile of lychee (Litchi chinensis

478

Sonn) fermented beverages. International Journal of Food Science and Technology, 45(11),

479

2358-2365.

480 481

AOAC. (1990). Official Methods of Analysis of AOAC International (15th edition, Vol. 2). Arlington, VA, USA: Association of Official Analytical Chemists, Inc.

482

Berenguer, M., Vegara, S., Barrajón, E., Saura, D., Valero, M., & Martí, N. (2016).

483

Physicochemical characterization of pomegranate wines fermented with three different

484

Saccharomyces cerevisiae yeast strains. Food Chemistry, 190, 848-855.

485 486 487 488

Boulton, R.B., Singleton, V.L., Bisson, L.F., & Kunkee, R.E. (1999). Principles and Practices of Winemaking (pp. 137). New York, NY, USA: Springer. Charve, J.I.M. (2011). Prediction of Mandarin Juice Flavor: A Flavoromic Approach (pp. 1-13). Ph.D. Thesis, Saint Paul, MN, USA: University of Minnesota.

489

Deshpande, A.B., Chidley, H.G., Oak, P.S., Pujari, K.H., Giri, A.P., & Gupta, V.S. (2016). Data

490

on changes in the fatty acid composition during fruit development and ripening of three

491

mango cultivars (Alphonso, Pairi and Kent) varying in lactone content. Data in Brief, 9,

492

480-491.

493 494

Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., & Smith, F. (1956). Colorimetric method for determination of sugars and related substances. Analytical Chemistry, 28(3), 350-356.

495

Food and Agriculture Organization of the United Nations. (2014). Statistical Yearbook 2014:

496

Europe and Central Asia Food and Agriculture (pp. 65-99). Bangkok, Thailand: FAO

497

Regional Office for Asia and the Pacific.

498 499

Gao, C., & Fleet, G.H.H. (1995). Degradation of malic and tartaric-acids by high-density cellsuspensions of wine yeasts. Food Microbiology, 12(1), 65-71.

22

500

Garde-Cerdán, T., & Ancín-Azpilicueta, C. (2008). Effect of the addition of different quantities

501

of amino acids to nitrogen-deficient must on the formation of esters, alcohols, and acids

502

during wine alcoholic fermentation. LWT - Food Science and Technology, 41(3), 501-510.

503 504

Iriti, M., & Faoro, F. (2006). Grape phytochemicals: A bouquet of old and new nutraceuticals for human health. Medical Hypotheses, 67(4), 833-838.

505

Joyeux, A. (1984). Evolution of acetic acid bacteria during fermentation and storage of wine

506

evolution of acetic acid bacteria during fermentation and storage of wine. Applied and

507

Environmental Microbiology, 48(1), 1034-1038.

508

Kumar, V., Jnawali, P., Goud, P.V., Bhasin, J.K., & Yildiz, F. (2016). Effect of maturation on

509

physico-chemical and sensory quality characteristics of custard apple wine. Cogent Food &

510

Agriculture, 2(1), 1180660.

511

Lee, S.J., Rathbone, D., Asimont, S., Adden, R., & Ebeler, S.E. (2004). Dynamic changes in

512

ester formation during Chardonnay juice fermentations with different yeast inoculation and

513

initial brix conditions. American Journal of Enology and Viticulture, 55(4), 346-354.

514

Li, X., Chan, L.J., Yu, B., Curran, P., & Liu, S.Q. (2012a). Fermentation of three varieties of

515

mango juices with a mixture of Saccharomyces cerevisiae and Williopsis saturnus var.

516

mrakii. International Journal of Food Microbiology, 158(1), 28-35.

517 518

Li, X., Lim, S.L., Yu, B., Curran, P., & Liu, S.Q. (2013a). Impact of pulp on the chemical profile of mango wine. South African Journal of Enology and Viticulture, 34(1), 119-128.

519

Li, X., Lim, S.L., Yu, B., Curran, P., & Liu, S.Q. (2013b). Mango wine aroma enhancement by

520

pulp contact and β-glucosidase. International Journal of Food Science & Technology, 48,

521

2258-2266

522

Li, X., Yu, B., Curran, P., & Liu, S.Q. (2011). Chemical and volatile composition of mango

23

523

wines fermented with different Saccharomyces cerevisiae yeast strains. South African

524

Journal of Enology and Viticulture, 32(1), 117-128.

525

Li, X., Yu, B., Curran, P., & Liu, S.Q. (2012b). Impact of two Williopsis yeast strains on the

526

volatile composition of mango wine. International Journal of Food Science & Technology,

527

47(4), 808-815.

528

Limwiwattana, D., Tongkhao, K., & Na Jom, K. (2016). Effect of sprouting temperature and air

529

relative humidity on metabolic profiles of sprouting black gram (Vigna mungo L.). Journal

530

of Food Processing and Preservation, 40(2), 306-315.

531

Liu, S.Q., & Pilone, G.J. (2000). An overview of formation and roles of acetaldehyde in

532

winemaking with emphasis on microbiological implications. International Journal of Food

533

Science & Technology, 35(1), 49-61.

534

Lu, Y., Huang, D., Lee, P.R., & Liu, S.Q. (2015). Effects of cofermentation and sequential

535

inoculation of Saccharomyces bayanus and Torulaspora delbruckii on durian wine

536

composition. International Journal of Food Science & Technology, 50(12), 2653-2663.

537

Masuda, M., & Nishimura, K.I.C. (1982). Changes in volatile sulfur compounds of whisky

538 539 540

during aging. Journal of Food Science, 47(1), 101-105. Miller, G.L. (1959). Use of dinitrosalicylic acid reagent for determination of reducing sugar. Analytical Chemistry, 31(3), 426-428.

541

Pilone, G. J., Clayton, M. G., & Duivenboden, R. J. (1991). Characterization of wine lactic acid

542

bacteria: Single broth culture for tests of heterofermentation, mannitol from fructose, and

543

ammonia from arginine. American Journal of Enology and Viticulture, 42(2), 153-157.

544

Pino, J.A., Mesa, J., Muñoz, Y., Martí, M.P., & Marbot, R. (2005). Volatile components from

545

mango (Mangifera indica L.) cultivars. Journal of Agricultural and Food Chemistry, 53(6),

24

546

2213-2223.

547

Quilter, M.G., Hurley, J.C., Lynch, F.J., & Murphy, M.G. (2003). The production of isoamyl

548

acetate from amyl alcohol by Saccharomyces cerevisiae. Journal of the Institute of Brewing,

549

109(1), 34-40.

550

Reddy, L.V.A., & Reddy, O.V.S. (2005). Production and characterization of wine from mango

551

fruit (Mangifera indica L). World Journal of Microbiology and Biotechnology, 21(8-9),

552

1345-1350.

553 554

Reddy, L.V.A., & Reddy, O.V.S. (2009). Production, optimization and characterization of wine from mango (Mangifera indica Linn.). Natural Product Radiance, 8(4), 426-435.

555

Reddy, L.V.A., & Reddy, O.V.S. (2011). Effect of fermentation conditions on yeast growth and

556

volatile composition of wine produced from mango (Mangifera indica L.) fruit juice. Food

557

and Bioproducts Processing, 89(4), 487-491.

558 559

Reddy, L.V.A., Reddy, O.V.S., & Joshi, V.K. (2014). Production of wine from mango fruit: A review. International Journal of Food and Fermentation Technology, 4(1), 13-25.

560

Ribéreau-Gayon, P., Glories, Y., Maujean, A., & Dubourdieu, D. (2006a). Handbook of Enology,

561

The Chemistry of Wine: Stabilization and Treatments (2nd edition) (pp. 3-60). Hoboken,

562

NJ, USA: Willey.

563

Ribéreau-Gayon, P., Glories, Y., Maujean, A., & Dubourdieu, D. (2006b). Handbook of

564

Enology: The Microbiology of Wine and Vinifications (2nd edition) (pp. 58). Hoboken, NJ,

565

USA: Willey.

566

Sadineni, V., Kondapalli, N., & Reddy, O.V.S. (2012). Effect of co-fermentation with

567

Saccharomyces cerevisiae and Torulaspora delbrueckii or Metschnikowia pulcherrima on

568

the aroma and sensory properties of mango wine. Annals of Microbiology, 62(4), 1353-

25

569

1360.

570

Saerens, S.M.G., Delvaux, F., Verstrepen, K.J., Van Dijck, P., Thevelein, J.M., & Delvaux, F.R.

571

(2008). Parameters affecting ethyl ester production by Saccharomyces cerevisiae during

572

fermentation. Applied and Environmental Microbiology, 74(2), 454-461.

573

Saerens, S.M.G., Delvaux, F.R., Verstrepen, K.J., & Thevelein, J.M. (2010). Production and

574

biological

function

of

volatile

575

Biotechnology, 3(2), 165-177.

esters

in

Saccharomyces

cerevisiae.

Microbial

576

Sriwimon, W., & Boonsupthip, W. (2011). Utilization of partially ripe mangoes for freezing

577

preservation by impregnation of mango juice and sugars. LWT - Food Science and

578

Technology, 44(2), 375-383.

579 580

Tharanathan, R.N., Yashoda, H.M., & Prabha, T.N. (2006). Mango (Mangifera indica L.), “the king of fruits” - An overview. Food Reviews International. 22, 95-123.

581

Torrens, J., Riu-Aumatell, M., López-Tamames, E., & Buxaderas, S. (2004). Volatile

582

compounds of red and white wines by headspace-solid-phase microextraction using

583

different fibers. Journal of Chromatographic Science, 42(6), 310-316.

584

Vuralhan, Z., Morais, M., Tai, S., Piper, M.D.W., & Pronk, J.T. (2003). Identification and

585

characterization of phenylpyruvate decarboxylase genes in Saccharomyces cerevisiae

586

identification

587

Saccharomyces cerevisiae. Applied and Environmental Microbiology, 69(8), 4534-4541.

588

Zeng, Z., Luo, S., Liu, C., Hu, X., Gong, E., & Miao, J. (2018). Phenolic retention of brown rice

589 590 591

and

characterization

of

phenylpyruvate

decarboxylase

genes

in

after extrusion with mesophilic α-amylase. Food Bioscience, 21, 8-13. Zoecklein, B.W., Fugelsang, K.C., Gump, B.H., & Nury, F.S. (1995). Wine Analysis and Production (pp. 411). New York, NY, USA: Chapman & Hall.

26

592 593 594 595 596

27

597 598

Table 1. Physicochemical properties, total sugar contents, reducing sugar contents, %titratable acidity and alcohol contents of ‘Nam Dok Mai’ mango wine during fermentation and storage.

599

Treatment

°Brix

pH

Total sugar

Reducing sugar

%Titratable

Alcohol content

content (mg/mL)

content (mg/mL)

acidity

(% v/v)

0% ferment

21.7±0.2a

4.29±0.03a

220±10a

10±1d

0.31±0.03d

1.2±0.4f

25% ferment

16.6±0.5b

3.39±0.04d

131±10b

85±5a

0.35±0.01c

4.3±0.3e

50% ferment

11±1c

3.34±0.02e

79±15c

61±4b

0.38±0.03abc

7.7±0.6d

75% ferment

7.4±0.5d

3.36±0.01de

37±9d

30±6c

0.39±0.02ab

8.7±1.5c

100% ferment

3.6±0.3f

3.54±0.02c

2.6±0.2e

0.86±0.04e

0.41±0.03a

13±1a

Day 0 storage

5.3±0.3e

3.59±0.01b

2.6±0.2e

0.85±0.01e

0.38±0.01abc

10.8±0.2b

Month 1 storage

5.4±0.4e

3.62±0.01b

2.5±0.1e

0.83±0.01e

0.38±0.01abc

10.8±0.2b

Month 2 storage

5.4±0.5e

3.61±0.01b

2.6±0.3e

0.87±0.01e

0.38±0.01abc

10.9±0.1b

Month 3 storage

5.3±0.3e

3.62±0.01b

2.6±0.2e

0.84±0.01e

0.37±0.01bc

11.1±0.1b

600

* Value are means±standard deviation (SD). Means in the same column indicated by different letters are significantly different at 95%

601

confidence level (p≤0.05).

28

602

Figure captions

603

Figure 1. Biplot of principal component analysis from all identified metabolites (●) and flavor

604

compounds ( ) in ‘Nam Dok Mai’ mango wine during wine processing periods: pre-

605

fermentation group (♦), in-process group (■), and post-fermentation or final product group (▲).

606

Figure 2. Dendrogram (similarity mode of agglomerative hierarchical clustering analysis) of

607

mango wine; red line: pre-fermentation group; green line: in-process group; blue line: post-

608

fermentation or final product group.

609

Figure 3. Heat plots of metabolite and flavor compounds (fatty acid methyl esters, polar lipids,

610

sugars, acids, esters, alcohols, ketone, aldehydes, terpenes, volatile acids, furans, phenol, and

611

lactone) during ‘Nam Dok Mai’ mango wine processing and storage; FAME=fatty acid methyl

612

ester, FFA=free fatty acid.

613

Figure 4. Lower triangular heat map represents pairwise correlation analysis between metabolite

614

and flavor compounds during mango wine processing and storage. Each square represents the

615

Spearman’s rank correlation coefficient at a significance level of p≤0.05. Positive correlations (r

616

>0.3) are shown with an orange-red scale (r >0.7 indicates a strong positive correlation).

617

Negative correlations (r <-0.3) are shown with a green scale (r <-0.7 indicates a strong negative

618

correlation).

619 620 621 622 623 624

29

625 626

Figure 1.

627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648

30

649 650

Figure 2.

651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674

31

675 676

Figure 3.

677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698

32

699 700

Figure 4.

701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718

33