Kinetics of Browning

Kinetics of Browning, Phenolics, and 5‑Hydroxymethylfurfural inCommercial Sparkling WinesA. Serra-Cayuela,† M. Jourdes,‡,§ M. Riu-Aumatell,*,† S. Buxaderas,† P.-L. Teissedre,‡,§

and E. Lopez-Tamames†

†Departament de Nutricio i Bromatologia, Xarxa de Referencia en Tecnologia dels Aliments (XaRTA), Facultat de Farmacia, Institutde Recerca en Nutricio i Seguretat Alimentaria·Universitat de Barcelona (INSA·UB), Campus de l’Alimentacio de Torribera,Avinguda Prat de la Riba 171, 08921 Santa Coloma de Gramenet, Spain‡Universite de Bordeaux, Institut des Sciences de la Vigne et du Vin (ISVV), EA 4577Œnologie, et §Institut National de la RechercheAgronomique (INRA), Institut des Sciences de la Vigne et du Vin (ISVV), USC 1366Œnologie, Faculte d’Œnologie, 210 chemin deLeysotte, CS 50008, 33882 Villenave d’Ornon, France

ABSTRACT: We analyzed the degree of browning (absorbance at 420 nm), the phenolics, and the 5-hydroxymethylfurfural (5-HMF) content in six sparkling wines series kept at three temperatures (4, 16, and 20 °C) for over 2 years. Caffeic acid, trans-coutaric acid, p-coumaric acid, and 5-HMF were the compounds with the greatest correlation with browning and time. 5-HMFwas the only compound that evolved linearly at all temperatures. We propose that 5-HMF is a better time−temperature markerthan the A420 parameter or phenolics, because it shows higher linearity with time at all temperatures, is more sensitive totemperature changes, and has lower variability. The kinetics of 5-HMF was studied showing a zero-order behavior. We proposemathematical models that wineries can use to predict the browning shelf life of their sparkling wines as a function of the storagetime and temperature.

KEYWORDS: kinetics, sparkling wines, browning, phenolics, 5-HMF, storage temperature, shelf life

■ INTRODUCTION

Sparkling wines are produced by the fermentation of a blend ofbase wines through the addition of yeasts and sugar.Fermentation can take place in sealed bottles (Champenoiseor the traditional method) or in bulk tanks (Charmat method).After a period of aging, the lees are removed from the wine.CO2 that has been generated is preserved, and the liqueurd’expedition is added. The most common sparkling wines areChampagne, Cava, Cremant, Mousseux, Prosecco, Spumante,and Sekt. These sparkling wines are distributed worldwide andstored under a range of conditions. Producers are concernedabout the quality of their wines and, especially, about shelf life.Stability is defined as the property of not suffering anyundesirable physical, chemical, or organoleptic changes during agiven period of time. The shelf life depends upon both theenvironmental conditions to which a product is exposed andthe acceptable loss of quality before it should be withdrawnfrom sale. Wineries need a stability parameter that can indicatethe shelf life of sparkling wines and the optimum storageconditions. Browning is a natural process that usually occurs inwhite wines1−3 but also takes place in sparkling wines4−6 overtime. It is a key quality indicator because consumers can noticeit and wineries make every effort to prevent it. Absorbance at420 nm (A420) is a typical, fast, and easy enological parameterthat is used by wineries to determine the degree of browning inwhite wines. However, the value of the A420 parameter is theresult of the oxidation of various compounds, and itsmeasurement has low sensitivity.6 There is general consensusthat browning in the aging and storage period is a non-enzymatic process. Most authors consider that it is due to theoxidation and polymerization of the phenolics present in wines,

although other reactions may occur.7,8 The phenoliccomposition of sparkling wines has been previously re-ported.4,5,9,10 Some studies on white wines address theevolution of phenolics,1,11−14 but few are focused on theirkinetics.2,15 There are hardly any studies on the evolution ofcommercial sparkling wines,4,6 and to the best of ourknowledge, there are no studies on kinetics in sparklingwines during commercial storage. Meanwhile, 5-hydroxyme-thylfurfural (5-HMF), an indicator of the progress of Maillardbrowning, has been used as a quality marker for industrialmanufactured foods.16,17 This compound is present in sparklingwines18 and increases linearly with time.6 The behavior andkinetics of 5-HMF have been studied in overheated fortifiedwines19 and rice wine.20 In both cases, the concentration of 5-HMF was found to be related to time and temperature.Temperature can be monitored and controlled during shelf

life and may affect the content of certain compounds insparkling wine and, consequently, its quality. Thus, the aim ofthis work was to use parameters related to browning (phenolicsand 5-HMF) to construct mathematical models for predictingthe shelf life of commercial sparkling wines. To carry out thesekinetics studies, six series of Spanish sparkling wines, producedat industrial scale, were stored at three temperatures for over 2years.

Received: July 25, 2013Revised: December 21, 2013Accepted: January 21, 2014Published: January 21, 2014

Article

pubs.acs.org/JAFC

© 2014 American Chemical Society 1159 dx.doi.org/10.1021/jf403281y | J. Agric. Food Chem. 2014, 62, 1159−1166

pubs.acs.org/JAFC

■ MATERIALS AND METHODSWines. We studied commercial sparkling wines that had been

produced on an industrial scale at the Freixenet S.A. winery. Theproduction process involved removing the lees and adding theappropriate amount of liqueur d’expedition that corresponded to theBrut category (sugar content < 12 g/L). The wines were producedfrom a blending of the grape varieties Macabeu, Xarel·lo, and Parellada.For over 2 years, six series of sparkling wine (coded A, B, C, D, E, andF) were kept under three temperatures: room (20 °C), cellar (16 °C),and fridge (4 °C). Production involved either the traditional (A, B, E,and F) or Charmat (C and D) methods. Wines E and F were Reservacategory (>15 months of aging). Two bottles from each series of wineat each temperature were taken at 0, 4, 8, 12, and 20 months of storagefor wines A−D and at 0, 4, 12, and 30 months for wines E and F. Intotal, we used 144 bottles of 75 cL from six series of commercialsparkling wines. Table 1 shows some of the enological parameters of

each series at the initial sampling point (time zero). All bottles wereopened in the dark, degassed, and stored at −20 °C until analysis.Chemicals. Deionized water was purified with a Milli-Q water

system (Millipore, Bedford, MA). Acetonitrile [high-performanceliquid chromatography (HPLC) grade] and acetic acid were purchasedfrom Prolabo-VWR (Fontenays/Bois, France). Gallic acid (>98%),tyrosol (>98%), caftaric acid (>97%), caffeic acid (98%), catechin(>98%), p-coumaric acid (>98%), and 5-HMF (>99%) were suppliedby Sigma-Aldrich (Saint Quentin Fallavier, France), while proto-catechuic acid (>90%) and ferulic acid were acquired fromExtrasynthese (Genay, France).Enological Parameters. The total sugar content, alcohol content,

pH, and free and total sulfur dioxide were measured using well-established standard methods21 (Table 1).Absorbance at 420 nm was considered as the browning index

because it is a typical, fast, and easy parameter that is used by theproducers of Spanish sparkling wines. It was measured using a 10 mmpath length quartz cuvette. Values were multiplied 1000-fold andexpressed as milli-absorbance units (mAU). A Shimadzu UV-3600spectrophotometer (Duisburg, Germany) was used. All measurementswere taken in duplicate.Identification and Quantification of Phenolics. We used a

Thermo Finnigan Surveyor HPLC system (Les Ulis, France)comprised of an autosampler (Surveyor Autosampler Plus), aquaternary pump (Surveyor LC Pump Plus), and an ultraviolet−visible (UV−vis) detector (Surveyor PDA Plus) hyphenated to afluorescence detector (Finnigan FLD Plus) and a Thermo-FinniganLCQ Advantage MS spectrometer equipped with an ion trap massanalyzer, all controlled by the Xcalibur data treatment system.Photodiode array (PDA) detection was configured to read from 200to 600 nm, while for the fluorescence detection, λex = 280 nm and λem= 320 nm [superhigh photomultiplier tube (PMT) voltage].Electrospray ionization mass spectrometry was performed in negativeand positive modes, reading m/z between 80.00 and 1200.00 with thefollowing optimized parameters: capillary temperature, 300 °C;capillary voltage, −5 V; nebulizer gas flow, 1.75 L/min; desolvationgas flow, 1 L/min; and spray voltage, 5 kV.The analyses were carried out in a Nucleosil 100-5 C18 column (250

× 4 mm, 5 μm, Agilent, France) at room temperature, and the samples

were kept at 8 °C in the sample tray. The following eluents were used:solvent A, water/acetic acid (99:1, v/v); and solvent B, acetonitrile/acetic acid (99:1, v/v). The elution gradient was as follows: 0 min, 3%B; 24 min, 6% B; 30 min, 6% B; 38 min, 12% B; 46 min, 20% B; 50min, 20%; 58 min, 60% B; 63 min, 80% B; and 68 min, 80% B. Thesystem was re-equilibrated for 10 min before the next injection.

For identification proposes (Table 2), 5 mL of sparkling winesamples was evaporated under vacuum (below 30 °C) and the residue

was redissolved in 500 μL of solvent A from the HPLC methodology(concentration of 10:1). The concentrated fraction was used forcompound identification by high-performance liquid chromatogra-phy−photodiode array−fluorometric detection−mass spectrometry(HPLC−PDA−FLD−MS), by comparing the retention times to co-injected commercial standards (when available), UV spectra, and massspectrometry (MS) spectra (in positive and negative modes).

For the quantification, recently filtered (0.45 μm) wine samples,with no prior treatment, were injected directly (20 μL) into thesystem. Each sample was injected in duplicate. Each compound wasquantified using an external calibration curve obtained from thecorresponding standard measured at the maximum wavelength, exceptfor caftaric, grape reaction product or 2-S-glutathyonilcaftaric (GRP),cis-/trans-coutaric, and fertaric acids, in which the respective free non-esterified standards (caffeic, p-coumaric, and ferulic acids) were used.Quantification was performed by PDA detection at the maximumwavelength of each compound, except for catechin, for whichfluorometric detection (FLD) was used.

5-HMF Quantification. The method described by Serra-Cayuelaet al.18 was used for the determination of this compound. Briefly, anultra-high-performance liquid chromatography (UHPLC) Acquitychromatographic system (Waters, Milford, MA) with a binary pump,an autosampler, and a photodiode array detector (DAD, Waters 2996)was employed. The system was controlled by Empower Pro 2 (Waters,Milford, MA). Chromatographic separation was carried out using aBEH C18 column (50 × 1.0 mm inner diameter) with a particle size of1.7 μm (Waters, Milford, MA), kept at 40 °C. The mobile phaseconsisted of eluent A as a solution of acetonitrile/water/formic acid(2:97:1, v/v/v) and eluent B as a solution of acetonitrile/water/formicacid (80:19:1, v/v/v). The mobile phase was passed through a 0.22μm polytetrafluoroethylene (Acrodisc, CR PTFE, Waters, Milford,MA) filter. Sample vials in the injector were maintained at 10 °C, andthe injected volume was 5 μL. The linear gradient conditions were asfollows: 0 min, 0% B, 3 min, 20% B; 3.5 min, 100% B; 4 min, 0% B;and 5 min, 0% B, at a flow rate of 0.800 mL/min. Data were acquired

Table 1. Commercial Sparkling Wine Characteristics at theInitial Sampling Point

samplingwineseries pH

freeSO2

(mg/L)

totalSO2

(mg/L)

total sugarcontent(g/L)

alcoholcontent(vol %)

CO2pressure(bar)

A 3.02 18 132 9.8 11.6 6.2B 3.03 18 116 10.2 11.7 6.2C 3.03 22 131 11 10.6 4.9D 3.03 22 116 10.8 10.4 5.1E 2.98 21 126 9.5 11.7 6.0F 2.98 19 104 9.3 11.8 6.1

Table 2. Identification and Quantification Parameters of thePhenolics in the Sparkling Wine Samplesa

compoundRt

(min) λ (nm) std add ESI−MS % CV

gallic acid 6.6 272 + (−) 169 5protocatechuic acid 11.4 260 + (−) 153 5tyrosol 16.0 257 + (+) 139 4caftaric acid 18.1 328 + (−) 311 5caffeic acid 19.2 323 + (−) 179 6GRP 22.8 365 (−) 616 7cis-coutaric acid 23.3 308 (−) 295 6trans-coutaric acid 27.8 308 (−) 295 5catechin 29.3 280/320b + (−) 289 6fertaric acid 37.4 329 (−) 325 5p-coumaric acid 43.9 308 + nf 6HMF 0.7 284 + (+) 127 4

aThe retention times, maximum wavelength, and m/z used for theidentification as well as confirmation by a standard addition (std add)method, when available, and the coefficient of variation (% CV) of themethod are shown. Rt, retention time; λ, wavelength; std add, standardaddition; ESI−MS, positive or negative and its m/z; nf, not found.bExcitation and emission.

Journal of Agricultural and Food Chemistry Article

dx.doi.org/10.1021/jf403281y | J. Agric. Food Chem. 2014, 62, 1159−11661160

with the Empower Pro 2 software. Chromatograms were monitored at284 nm. Sparkling wine samples and the 5-HMF standard were passedthrough a Waters (Waters, Milford, MA) 13 mm polypropylene (PP)of 0.22 μm pore size filter before injection. All samples were analyzedin duplicate using external calibration curves.Statistical Analysis. The analytical replicates were averaged to

produce the sample (bottle) value and then the variance within thetwo bottles; finally, the variances within the wines series (A−F) wereused to calculate the overall variance and 95% confidence limits.Principal component analysis (PCA) was employed to observe

trends in the data. The first two principal components were used toplot the scores and loadings separately.Analysis of variance (ANOVA) was applied for the parameters

analyzed (general parameters, phenolics, A420, and 5-HMF) to observewhether the factors time and temperature were significant as well astheir interaction with the series factor. To show the compoundbehaviors along shelf life, the coefficients of variation (% CV) betweeninitial and final concentrations were calculated as well as the meansand the 95% confidence intervals of each series and time−temperaturepoints. When there was not interaction time series, all of the serieswere joined, at each temperature, and the statistical differencesbetween time points were carried out.Linear regression was used to determinate the relationships between

the compounds with time at each temperature and also to calculate thevelocity constants (k). To check whether there were differencesbetween the slopes (velocity constants), we performed the Tukeyhonest significant difference (HSD) multi-range test.22 To check thegoodness of fit of the models, the observed values were plotted againstthe predicted values.A box-and-whisker plot was performed as an exploratory data

analysis to show the distribution of 5-HMF values and to establish theinterval covered by the middle 50%, interquartile range, of the data(box Q25−Q75). Whiskers are drawn from the edges of the box to thelargest and smallest values, unless there are concentrations unusuallyfar away from the box (outliers). The outliers are considered whentheir values are higher than 1.5-fold the interquartile range (boxheight) above or below the box, and they are represented by circles.Only the threshold of outliers above the interquartile range (Q75−Q25)was considered (upper inner fence). With the difference between thisupper limit and the median, the maximum variation that a wine canundergo before it has overbrowned was calculated.

Statistical significances were set at p < 0.05. Version 2.15 of the R(Foundation for Statistical Computing, Vienna, Austria), with theFactoMineR and Hmisc packages, was used for all statistical andgraphical analyses.

■ RESULTS AND DISCUSSION

Six series of commercial sparkling wines were stored at threetemperatures (fridge, 4 °C; cellar, 16 °C; and room, 20 °C) forover 2 years. Table 1 shows the main enological parameters ofthe series at the initial sampling point (time zero). Allparameters remained constant during storage, except for freeSO2. For all series, the concentration of free SO2 fell below thelimit of quantification (LOQ; 6 mg/L) within the first 4months and until the end of sampling (data not shown). Inmodel wines, it has been observed that this compound can reactwith aldehydes, such as 5-HMF, and flavonoids, such ascatechin.23,24 However, the lowest concentration of free SO2(which is the reactive form) seems to not have a relevantimpact on the compounds studied in comparison to the effectof the temperature and time factors throughout the study.Further work would be required to specifically assess the impactof SO2 and catechin on the kinetics of 5-HMF accumulation insparkling wines.PCA was used to observe the trends of the compounds

during storage (Figure 1). Scores were affected by the factorstime, temperature, and sparkling wine series (Figure 1a). Thefirst component showed that the scores for the sparkling winesamples tended to change positively with time. 5-HMF, trans-coutaric acid, p-coumaric acid, caffeic acid, catechin, and theA420 parameter were the variables that contributed greatly tothis first component (Figure 1b) and went in the samedirection as time, except catechin, which went in the oppositedirection. Thus, the more aged wines have higher amounts of 5-HMF, trans-coutaric acid, and p-coumaric acid, while they haveless catechin (p < 0.05). In contrast, the scores for the secondcomponent were mainly distributed by sparkling wine series,determining the gallic acid, GRP, cis-coutaric acid, and caffeicacid amounts.

Figure 1. PCA: (a) scores and (b) loadings plot of the six series of sparkling wines (A, B, C, D, E, and F). Scores are coded by the series (letter) andlength (number) and temperature (color) of storage.



Table

3.MeanCon

tent

ofPheno

lics,5-HMF,

andBrowning

attheThree

Tem

peraturesa

AB

CD

EF

compound

T(°C)

i(m

g/L)

((Δf−

i)/i)(%

)i(m

g/L)

((Δf−

i)/i)(%

)i(m

g/L)

((Δf−

i)/i)(%

)i(m

g/L)

((Δf−

i)/i)(%

)i(m

g/L)

((Δf−

i)/i)(%

)i(m

g/L)

((Δf−

i)/i)(%

)

gallicacid

40.89

b1.06

b0.62

b0.66

80.81

b0.98

−5

16b

−16

bb

b−15

20−27

−28

bb

b−9

protocatechuicacid

44.54

91.51

b3.97

b1.17

121.38

b1.68

b16

24b

4812

−8

b20

25b

55b

bb

tyrosol

411.68

b3.85

b3.54

b13.25

74.14

b7.42

b16

bb

bb

bb

204

b−18

bb

bcaftaric

acidc

44.98

b4.73

b3.82

b4.04

b5.19

b3.95

b16

bb

−10

−6

−20

−15

20−14

−23

−15

−15

−23

−23

caffeicacid

40.72

b0.59

b0.48

b0.42

b0.87

b0.76

b16

26b

4077

2116

2024

2345

7629

23GRP

41.96

b13.19

b2.25

b3.02

b8.49

b10.91

b16

b−50

−27

−38

−39

−42

20−34

−36

−34

−44

−48

−48

cis-coutaricacidd

41.00

b2.13

b0.85

b1.1

b1.53

−6

1.83

−9

16−34

−51

−36

−30

−41

−50

20−43

−43

−42

−38

−48

−47

trans-coutaricacidd

41.31

131.12

b1.08

b1.01

b1.47

b1.14

b16

4457

48105

3071

2044

3543

9140

60catechin

40.44

−20

0.48

b0.54

−19

0.49

−15

0.44

b0.47

b16

b−22

−22

−16

−7

−16

20b

−29

−29

−32

b−17

fertaricacide

40.94

b1.22

b0.82

b1.12

b1.00

b1.21

b16

bb

bb

bb

20b

−20

−9

b−16

−15

p-coum

aric

40.53

140.36

b0.44

b0.32

b0.66

b0.45

b16

5197

47141

4468

2055

6257

131

6395

5-HMF

40.46

150.46

170.34

220.32

300.78

210.73

2116

289

279

393

410

208

212

20373

423

576

603

304

352

A420f

472

b77

b58

b51

b79

b69

b16

1915

2729

2625

2026

1737

4635

32aThe

percentage

ofvariatio

n((Δf−

i)/i)betweenthefinalandinitialvalues

with

ineach

seriesisalso

show

n.bNodifferenceswerefoundby

theANOVAtest(p

>0.05).c U

sing

caffeicacid

asthe

standard.dUsing

p-coum

aricacid

asthestandard.eUsing

ferulic

acid

asthestandard.fThe

meanisexpressedas

mAU.



Table 3 shows the initial values of each compound and theA420 parameter. GRP, caftaric acid (both quantified as caffeicacid equivalent), tyrosol, and protocatechuic acid were thecompounds with the highest concentration. In contrast, flavan-3-ol catechin and the hydroxycinnamic p-coumaric acid werethe phenolics with the lowest content. Table 3 also shows thepercentage of variation between the final and initial point foreach wine series and each storage temperature. Thesepercentages of variation are only shown when the differencewas statistically significant by the ANOVA test. The compound5-HMF had by far the greatest positive variation during shelflife. Thus, this compound underwent considerable changesduring storage at the temperatures studied. To a lesser extent,caffeic acid, trans-coutaric acid, p-coumaric acid, and A420 alsounderwent positive variation during storage. A compound mustmeet a certain requirement to be used as a marker; it shouldbehave in the same way in all series. Consequently, gallic acid,tyrosol, and fertaric acid were discarded as markers. Thesecompounds, along with protocatechuic acid, were rejectedbecause they increased or decreased depending upon the wineseries. Finally, caftaric acid, GRP, cis-coutaric acid, and catechinwere not further considered, because all of these compoundsalways decreased with time. A marker cannot disappear over the

shelf life of a product, because this would make it difficult todetermine the storage time.Figure 2 shows the correlation of the compounds that had

the highest coefficient of variation and always increased overtime (5-HMF, caffeic acid, trans-coutaric acid, and p-coumaricacid), as well as the classic browning parameter (A420). Allcompounds had a high correlation with A420, with caffeic acidbeing the compound with highest correlation. The correlationwith caffeic and p-coumaric acids could be due to the fact thatboth arise from the hydrolysis of their respective esters (caftaticand cis-coumaric acids). Caffeic, trans-coutaric, and p-coumaricacids did not change significantly at low temperatures (ANOVAtest), and no difference was observed between 16 and 20 °C(Figure 2 and Table 3). Thus, only 5-HMF changedsignificantly (p < 0.05) at all temperatures and wine series.Therefore, this compound had the most suitable characteristicsto be a time−temperature marker, which led us to study itskinetics. Moreover, 5-HMF was correlated (p < 0.05) with theA420 parameter, which is usually monitored in wineries bymeasuring browning3,25 (Figures 1b and 2).

5-HMF and A420 Models. The kinetics of 5-HMF (Figure3a) and A420 (Figure 3b) were studied and compared. Todeterminate the reaction order, the integrated kinetic equations

Figure 2. Pearson correlation between selected phenolics, 5-HMF, A420, and time at each temperature and the overall value.



of zero, first, and second order were tested to find the reactionorder that best fit the experimental data (data not shown). Both5-HMF and A420 showed zero-order kinetics (eq 1), increasinglinearly over time

= +Y Y kti (1)

where Y is the 5-HMF content (mg/L) or absorbance at 420nm (mAU), Yi is the initial value of the 5-HMF content (mg/L) or initial absorbance (mAU), k is the velocity constant(expressed as mg L−1 month−1 for 5-HMF or mAU/month forA420), and t is time (in months).The velocity constants (k) were calculated at each temper-

ature (kR, room; kC, cellar; and kF, fridge. 5-HMF (Figure 3a)showed higher differentiation than the A420 parameter betweenthe velocity constants at various temperatures (Tukey HSDtest) because 5-HMF has lower variability than A420 (Figure3b). Moreover, the A420 parameter at the fridge temperature didnot change significantly with time (Figure 3b and Table 3).According to the results indicated above, we will focus only on5-HMF kinetics to construct a prediction model because thiscompound has less variability, is more sensitive, and can beused in a larger range of temperatures than the A420 parameter.We used the Arrhenius equation (eq 2) to predict k at anytemperature

=−+

⎛⎝⎜

⎞⎠⎟k Ae

ER T( 273.15)

a

(2)

where k is the pseudo-zero-order rate constant (mg L−1

month−1), A is the pre-exponential constant, T is thetemperature (°C), Ea is the activation energy (kJ/mol), andR is the universal gas constant (8.314 J K−1 mol−1). Ea and Aare obtained from a representation of the natural logarithm ofeq 2. The slope multiplied by R provided Ea, while A wasobtained from the y intercept. For 5-HMF, the values were Ea =

Figure 3. Kinetics of (a) 5-HMF and (b) A420 of commercial sparkling wines during storage at three temperatures (room, cellar, and fridge). Themean, the 95% confidence interval, and the number of samples (n) for the set of all series (A−F) at each time point are plotted. The velocityconstant of formation (k) of the zero-order kinetics is also shown for each temperature.

Figure 4. Correlation between observed versus predicted values tocheck the goodness of fit of the prediction model of 5-HMF, withdashed line representing the 95% prediction interval.



125 kJ/mol and A = 2 × 1021 mg L−1 month−1. These Ea and Avalues allowed us to establish eq 3, which was obtained bysubstituting eq 2 into eq 1. This equation can be used to predictthe 5-HMF content, depending upon the length and temper-ature of storage

= + ×−+

⎛⎝⎜

⎞⎠⎟t[HMF] [HMF] 1.7 10 e T

i21

15014( 273.15) (3)

where [HMF] is the content of 5-HMF in mg/L, [HMF]i is theinitial content (mg/L), T is the temperature (°C), and t is thetime (months).Sparkling Wine Shelf Life Prediction. Figure 4 shows the

correlation between observed versus predicted values to showthe goodness of the fitted model. The predicted model could beused to assess the length of storage depending upon thetemperature conditions. At higher temperatures (outside therange studied) the behavior may be used by extrapolation. Forinstance, a sparkling wine stored at 25 °C for 15 months wouldincrease its 5-HMF content by 4.05 mg/L.The “best before” date of wine is regulatory not considered.

However, it is of great interest for winemakers to know theshelf life of a wine. To determine whether a wine hadanomalous data related to browning, we performed a box-and-whisker plot for 5-HMF (Figure 5a), which shows the medianand upper inner fence for all of the samples. The latterparameter indicates the limit from which point the sampleswere considered overdeveloped (statistical outliers) and isdefined as 1.5-fold the interquartile difference (Q75−Q25).Thus, the difference between the upper inner fence and medianwas considered the maximum variation that a wine can undergobefore it has overdeveloped. This assessment is purelymathematical and does not take into account sensory aspects.The variation was 1.81 mg/L (Figure 5a); therefore, eq 4 isobtained substituting this value in eq 3. With this equation (eq4), we can predict the maximum time of storage, dependingupon the temperature

= × − +⎛⎝⎜

⎞⎠⎟t 9.9 10 e T22

15014( 273.15) (4)

where t is the time (months) and T is the temperature (°C).Thus, this equation can be used to predict the maximum time

before a wine overdevelops at a certain temperature or viceversa. Figure 5b graphically represents eq 4. The representationis also extrapolated outside the range of temperatures studied(dashed curve), thus showing the behavior in most commonstorage conditions. Thus, a sparkling wine stored at 20 °Cwould have overdeveloped after the first year and a halfaccording to measurements of 5-HMF. At lower temperatures(4−16 °C), wines evolve slowly, with their life lengthening formany years. For instance, by extrapolation of the predictionmodel, at intermediate temperatures (16−24 °C), the shelf lifeof the sparkling wines would range from around 1 to severalyears. For storage at higher temperatures (above 27 °C), theprediction model indicates that sparkling wines should beconsumed as soon as possible (within the first few months).Notwithstanding, because sparkling wines are produced fromdifferent grape varieties (blanc de blanc or blanc de noir) andblending, we believe that each winery could calculate the shelflife of their sparkling wine using their own initial value and thelimit beyond which they consider the wine has overdeveloped.Once these values have been determined, the shelf life could becalculated by applying them to eq 3.In summary, sparkling wines are stable at regular storage

temperatures. Lower temperatures (fridge) can almostcompletely prevent browning and inhibit other reactions thatinvolve phenolics. The hydroxycinnamic acids: caffeic, trans-coutaric, and p-coumaric, were the phenolics that have thehighest correlation with browning and time, and theirconcentration increased over time. The compound 5-HMFproved to be a better time−temperature marker than A420 orany of the phenolics tested, because it increased linearly withtime at all temperatures, was more sensitive, even at lowtemperatures, and more precise (with lower variability). The

Figure 5. (a) Box-and-whisker plot of 5-HMF values for all of the series samples. The first, second, and third quartiles and the minimum andmaximum values of each category are shown. The outlier limit, defined as 1.5-fold the interquartile difference, and its distance to the median areindicated. The outliers are represented by circles. (b) Time−temperature plot for predicting the shelf life of sparkling wines. The dashed linesindicate the extrapolation of the models.



mathematical models allow us to determinate the concentrationof 5-HMF, depending upon the length and temperature ofstorage. The model could be used by wineries to predictsparkling wine shelf life, that is, to determinate how longsparkling wines can be stored at a certain temperature beforethey overdevelop. Furthermore, in the case of complaint, thewineries can use 5-HMF as a reliable marker to reveal theconditions under which the sparkling wine had been stored.

■ AUTHOR INFORMATIONCorresponding Author*Telephone: +34-934033795. Fax: +34-934035931. E-mail:[email protected] study was made possible by funding from the SpanishComision Interministerial de Ciencia y Tecnologia (CICYT)Grant AGL2011-23872, the ACC1O EVALXARTA2010(EX12) and the Generalitat de Catalunya (2009SGR-606).The Ph.D. student Arnau Serra-Cayuela was awarded a grantand scholarship from the Universitat de Barcelona (APIF-UB).NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThe authors thank Freixenet S.A. wineries for providing thesamples.

■ ABBREVIATIONS USEDLOQ, limit of quantification; GRP, grape reaction product (2-S-glutathyonilcaftaric); 5-HMF, 5-hydroxymethylfurfural; PCA,principal component analysis; k, velocity constant; A, pre-exponential constant; Ea, activation energy

■ REFERENCES(1) Maury, C.; Clark, A. C.; Scollary, G. R. Determination of theimpact of bottle colour and phenolic concentration on pigmentdevelopment in white wine stored under external conditions. Anal.Chim. Acta 2010, 660, 81−86.(2) Salacha, M.-I.; Kallithraka, S.; Tzourou, I. Browning of whitewines: Correlation with antioxidant characteristics, total polyphenoliccomposition and flavanol content. Int. J. Food Sci. Technol. 2008, 43,1073−1077.(3) Singleton, V. L.; Kramlinga, T. E. Browning of white wines andan accelerated test for browning capacity. Am. J. Enol. Vitic. 1976, 27,157−160.(4) Bosch-Fuste, J.; Sartini, E.; Flores-Rubio, C.; Caixach, J.; Lopez-Tamames, E.; Buxaderas, S. Viability of total phenol index value asquality marker of sparkling wines, “cavas”. Food Chem. 2009, 114,782−790.(5) Ibern-Gomez, M.; Andres-Lacueva, C.; Lamuela-Raventos, R. M.;Buxaderas, S.; Singleton, V. L.; de la Torre-Boronat, M. C. Browningof cava (sparkling wine) during aging in contact with lees due to thephenolic composition. Am. J. Enol. Vitic. 2000, 51, 29−36.(6) Serra-Cayuela, A.; Aguilera-Curiel, M.; Riu-Aumatell, M.;Buxaderas, S.; Lopez-Tamames, E. Browning during biological agingand commercial storage of cava sparkling wine and the use of 5-HMFas a quality marker. Food Res. Int. 2013, 53, 226−231.(7) Li, H.; Guo, A.; Wang, H. Mechanisms of oxidative browning ofwine. Food Chem. 2008, 108, 1−13.(8) Oliveira, C. M.; Ferreira, A. C. S.; De Freitas, V.; Silva, A. M. S.Oxidation mechanisms occurring in wines. Food. Res. Int. 2011, 44,1115−1126.(9) Chamkha, M.; Cathala, B.; Cheynier, V.; Douillard, R. Phenoliccomposition of champagnes from Chardonnay and Pinot Noirvintages. J. Agric. Food Chem. 2003, 51, 3179−3184.

(10) Pozo-Bayon, M. A.; Hernandez, M. T.; Martín-Alvarez, P. J.;Polo, M. C. Study of low molecular weight phenolic compoundsduring the aging of sparkling wines manufactured with red and whitegrape varieties. J. Agric. Food Chem. 2003, 51, 2089−2095.(11) Hernanz, D.; Gallo, V.; Recamales, A. F.; Melendez-Martínez, A.J.; Gonzalez-Miret, M. L.; Heredia, F. J. Effect of storage on thephenolic content, volatile composition and colour of white wines fromthe varieties Zalema and Colombard. Food Chem. 2009, 113, 530−537.(12) Berg, H. W.; Akiyoshi, M. Some factors involved in browning ofwhite wines. Am. J. Enol. Vitic. 1956, 7, 1−7.(13) Kallithraka, S.; Salacha, M. I.; Tzourou, I. Changes in phenoliccomposition and antioxidant activity of white wine during bottlestorage: Accelerated browning test versus bottle storage. Food Chem.2009, 113, 500−505.(14) Mayen, M.; Baron, R.; Merida, J.; Medina, M. Changes inphenolic compounds during accelerated browning in white wines fromcv. Pedro Ximenez and cv. Baladi grapes. Food Chem. 1997, 58, 89−95.(15) Sioumis, N.; Kallithraka, S.; Makris, D. P.; Kefalas, P. Kinetics ofbrowning onset in white wines: Influence of principal redox-activepolyphenols and impact on the reducing capacity. Food Chem. 2006,94, 98−104.(16) Delgado-Andrade, C.; Seiquer, I.; Haro, A.; Castellano, R.;Navarro, M. P. Development of the Maillard reaction in foods cookedby different techniques. Intake of Maillard-derived compounds. FoodChem. 2010, 122, 145−153.(17) Teixido, E.; Nunez, O.; Santos, F. J.; Galceran, M. T. 5-Hydroxymethylfurfural content in foodstuffs determined by micellarelectrokinetic chromatography. Food Chem. 2011, 126, 1902−1908.(18) Serra-Cayuela, A.; Castellari, M.; Bosch-Fuste, J.; Riu-Aumatell,M.; Buxaderas, S.; Lopez-Tamames, E. Identification of 5-hydrox-ymethyl-2-furfural (5-HMF) in cava sparkling wines by LC−DAD−MS/MS and NMR spectrometry. Food Chem. 2013, 141, 3373−3380.(19) Pereira, V.; Albuquerque, F. M.; Ferreira, A. C.; Cacho, J.;Marques, J. C. Evolution of 5-hydroxymethylfurfural (HMF) andfurfural (F) in fortified wines submitted to overheating conditions.Food. Res. Int. 2011, 44, 71−76.(20) Chen, L.; Huang, H.; Liu, W.; Peng, N.; Huang, X. Kinetics ofthe 5-hydroxymethylfurfural formation reaction in Chinese rice wine. J.Agric. Food Chem. 2010, 58, 3507−3511.(21) International Organisation of Vine and Wine (OIV).Compendium of International Methods of Wine and Must Analysis;OIV: Paris, France, 2004.(22) Zar, J. H. Biostatistical Analysis, 5th ed.; Pearson: London, U.K.,2009.(23) Es-Safi, N.-E.; Cheynier, V.; Moutounet, M. Study of thereactions between (+)-catechin and furfural derivatives in the presenceor absence of anthocyanins and their implication in food color change.J. Agric. Food Chem. 2000, 48, 5946−5954.(24) Nonier, M. F.; Vivas, N.; Gaulejac, N. V.; de Absalon, C.; Vitry,C. Study by LC/ESI/MSn and ESI/HR/MS of SO2 interactions inflavanols−aldehydes nucleophilic reactions. Food Chem. 2010, 122,488−494.(25) Zoecklein, B. W. Wine Analysis and Production; Chapman andHall: London, U.K., 1995.



mailto:[email protected]

Documents

Kinetics of Browning