The vinification of partially dried grapes: a comparative fermentation study of Saccharomyces cerevisiae strains under high sugar stress

Authors


G. Zapparoli, Dipartimento Scientifico e Tecnologico, Università degli Studi di Verona, Strada Le Grazie, 15, 37134 Verona, Italy (e-mail: giacomo.zapparoli@univr.it).

Abstract

Aims:  The study of the fermentation performance of Saccharomyces cerevisiae strains under high sugar stress during the vinification of partially dried grapes.

Methods and Results:  Microvinification of partially dried grape must with sugar concentration of 35° Brix was performed using four commercial strains to carry out alcoholic fermentation. A traditional red vinification without nutrients addition was applied. Yeasts displayed different efficiency to convert sugar in ethanol and varied in glycerol yield. Sugar consumption and ethanol level were attested at 80–87% and 143·5–158·0 g l−1 respectively. High correlation between sugar and assimilable nitrogen consumption rate was observed. Statistical treatment of data by principal component analysis highlighted the different behaviours that strains exhibited in regard to the production of higher alcohols and other compounds important to wine quality.

Conclusions: Saccharomyces cerevisiae strains displayed appreciable capability to overcome osmotic stress and to yield ethanol fermenting high sugar concentration grape must in winemaking condition.

Significance and Impact of the Study:  The results provided insights on the strain contribution to wine quality subordinate to stress condition. This investigation is of applicative interest for winemaking and processing industry that use high sugar concentration musts.

Introduction

The fermentation of high sugar grape musts could occur in winemaking for wine production from dried, botritized or late-harvest grapes, for ice-wine production or in processing industry that used grape juice concentrate. Using must with high sugar concentrations stuck or sluggish fermentations are probable because of the high osmotic pressure and ethanol toxicity for yeast cells (Bisson 1999; Bisson and Butzke 2000). Moreover, an early arrest of fermentation in these musts produces wines of low quality and stability favouring the production of high volatile acidity and the growth of spoilage micro-organisms (Caridi et al. 1999).

Saccharomyces cerevisiae is routinely used as starter culture for the wine production from grape musts containing usually 16–26% (w/v) sugars. In dried and late-harvest grape musts sugar concentrations may be easily over 30% (w/v) and in frozen grape musts as much as 50% (w/v). Grape juice concentrates with sugar level of 35° Brix are used for high alcohol wine or ice wine production (Buescher et al. 2001). The high sugar concentration can affect the fermentative capacity of the yeasts (Dubois et al. 1996). Winemakers fermenting high sugar musts have to be sure that S. cerevisiae strains have high fermentation efficiency under stressful conditions and the use of osmo-ethanol-tolerant strains is recommended. Saccharomyces cerevisiae strains display different fermentation behaviours under stress conditions (Carrasco et al. 2001; Zuzuarregui and Del Olmo 2004a). Several commercial yeasts were studied to optimize addition of nitrogen and oxygen in order to prevent stuck fermentation (Blateyron et al. 2003). Recently, Zuzuarregui and Del Olmo (2004b) described a system of wine yeast selection based on the resistance to the stress conditions taking place during wine production. Gene expression patterns of commercial S. cerevisiae strains under sugar stress and nitrogen limitation were investigated (Backhus et al. 2001; Erasmus et al. 2003). Actually, the wide range of commercial active dry yeasts available for enological use allows meeting different winemaker needs in normal fermenting conditions. Nevertheless, in order to manage fermentation of high sugar concentration musts further investigations are necessary.

The aim of this study was to evaluate the fermentability of high sugar grape must using different commercial S. cerevisiae strains. The must was obtained from natural partially dried grapes to concentrate the sugars. Ethanol production and total assimilable nitrogen (TAN) consumption rates were measured in relation to sugar depletion. In order to evaluate the impact of high sugar must fermentation to wine quality, secondary metabolic products were analysed and statistical treatment of data was carried out.

Materials and methods

Yeast strains

Commercial S. cerevisiae strains utilized in this study were kindly provided by Lallemand Italia (Lallemand Inc., Montreal, Que., Canada) for BM45, EC1118 and S6U and by Enologica Vason (Verona, Italy) for Zinfandel.

Fermentation experiment

Healthy grapes of Corvina variety was used (Valpolicella, Italy), partially dried for 4 months in attics, not affected by Botrytis or other moulds as revealed by gluconic acid content determined in the must (<0·1 g l−1) (Perez et al. 1991) added in each tank. In order to prepare the homogenized fractions to ferment, at the grape crushing the must was separated from the skins. Then, the must was mixed vigorously and, without a previous clarification, it was divided in equal amounts (11 l) and collected in 12 tanks of 15 l capacity. The must composition was as follows: sugars 35° Brix, total acidity 5·65 g l−1 of tartaric acid, pH 3·1, l-malic acid 0·71 g l−1, TAN 325 mg l−1. The amount of grape skins was divided in 12 fractions and added in each tank. The inoculation of active dry yeasts was carried out according to manufacture instructions performing a rehydration step for 20 min in tap water at 36–38°C. The size of the inoculum for each tank was about 3–5 × 106 cells ml−1 of must. The fermentations were carried out in the winery. The temperature, monitored daily during the fermentation, ranged between 12 and 15°C both inside and outside of each tank. Wine samples were collected during fermentation and immediately analysed for sugar, ethanol and TAN content. Fermentations were performed in triplicate.

Analytical determinations

Analysis of the wines was carried out when sugar consumption stopped. Total and volatile acidity, sugars, ethanol, sulfite and pH were determined using standard methods for wine analysis. Organic acids and glycerol were quantified using enzyme kits (La Roche, Basel, Switzerland).

TAN was determined by formol titration (Grump et al. 2000). Total polyphenols, total antocyanins and colour parameters were determined by Glories method (Glories 1979).

Volatile compounds were gas chromatographically analysed by a Vega series 6000 gas chromatograph (Carlo Erba Instruments, Milan, Italy). A capillary column CB AW/6·6%, 200 cm × 2 mm i.d. (Supelco, Bellefonte, PA, USA) with FID detector. Column temperature varied from 80 to 180°C (4°C min−1); injector and detector temperature was 200°C and gas carrier (N2) flow rate was 20 ml min−1. For each analysis 2 μl of wine sample preloaded with 100 mg l−1 2-butanol (as internal standard) were used.

Trans-resveratrol were quantified by HPLC analysed according to Pezet and Cuenat (1996), using a HPLC Agilent Technologies Serie1100 (Palo Alto, CA, USA), equipped with column ODS-Hypersil C18 (5 μm, 200 mm × 2·1 mm). HP-Chemstation software (Hewlett-Packard, Wilmington, DE, USA) was used to control injector, pump and UV-VIS DAD.

For analysis of each compound at least two determinations were carried out.

Statistical analysis

Analysis of variance (anova) and t-test were carried out for each fermentation parameter and wine compound to test differences among yeast strains on fermentation capability and on wine quality. Principal component analysis (PCA) was used to establish the relationship among variables represented by wine compounds. Kaiser criterion was used as extraction method, retaining only factors with eigenvalues >1. Varimax rotation was used as rotation type that maximizes the variability of the component and minimizes the variance around the component (Dillon and Goldstein 1984). All statistical analyses were performed using SPSS software program (SPSS Inc., Chicago, IL, USA).

Results

Fermentation kinetics

In all the microvinification trials the alcoholic fermentation resulted incomplete. The consumption of sugars arrested when their content in the wines ranged between 48·4 and 74·9 g l−1 (Table 1). The wine drawing off was performed when sugars depletion resulted negligible after 33 d in tanks inoculated with S6U and EC1118 strains, and after 36 d in tanks inoculated with Zinfandel and BM45 strains. No further sugar consumption was determined 10 d later when lees separation was carried out. Table 1 shows sugar, ethanol and glycerol content measured in the wines after lees separation. Zinfandel and EC1118 strains displayed higher capacity to utilize fermentable sugars, than S6U and BM45 strains, consuming 86 and 87% of sugar amount presents in grape must respectively. Nevertheless, results on the coefficient of sugar conversion to ethanol were similar for the four strains. S6U strain produced wines with the highest amount of glycerol to the detriment of ethanol. However, ethanol yield (YE/S) resulted similar for all the strains, while S6U displayed a significant (P < 0·001) higher glycerol yield (YG/S) than the other strains.

Table 1.  Sugars, ethanol and glycerol metabolism from the fermentation of high sugar concentration grape must carried out by four wine yeast
 ZinfandelS6UEC1118BM45
  1. Values ± standard deviation are average of three independent trials.

  2. †Ethanol yield (g ethanol g sugar−1).

  3. ‡Coefficient sugar–ethanol conversion (% Vol ethanol produced/% w sugar consumed).

  4. §Glycerol yield (g glycerol g sugar−1).

  5. Numbers with different letters differ at *P < 0·001; ns, not significant.

Residual sugars* (g l−1)50·8 ± 5·37 a74·9 ± 2·16 b48·4 ± 1·01 a63·4 ± 1·33 c
Sugar consumed* (%)86 ± 1·15 a80 ± 0·58 b87 ± 0·00 a83 ± 0·00 c
Ethanol* (g l−1)158·0 ± 1·39 a143·5 ± 0·66 b157·5 ± 0·80 a149·8 ± 0·87 c
YE/S0·486 ± 0·006 ns0·477 ± 0·004 ns0·481 ± 0·003 ns0·480 ± 0·002 ns
C‡0·616 ± 0·005 ns0·604 ± 0·003 ns0·609 ± 0·003 ns0·607 ± 0·003 ns
Glycerol* (g l−1)14·0 ± 0·26 a15·9 ± 0·04 b14·1 ± 0·14 a14·0 ± 0·23 a
YG/S§*0·043 ± 0·002 a0·053 ± 0·001 b0·043 ± 0·001 a0·045 ± 0·001 a

Ethanol production rate

Figure 1 shows the specific ethanol production rate measured during fermentation. Ethanol production started quickly in trials inoculated with EC1118, S6U and BM45 strains reaching maximum production rate (14 g l−1 d−1) after 12 d, while in tanks inoculated with Zinfandel strain the maximum ethanol production rate (9 gl−1 d−1) was obtained 3 d later. Drastic reduction of ethanol production was observed after 30 d in all the trials.

Figure 1.

Specific ethanol production rate (g l−1 d−1) determined during the fermentation of high sugar concentration grape must carried out by four wine yeasts (bsl00046 Zinfandel, bsl00001 S6U, bsl00066 EC1118, • BM45). Bars are standard deviation

Total assimilable nitrogen and sugar consumption rate

Figure 2 shows TAN and sugars consumption kinetics during alcoholic fermentation. Most of TAN present in the must was utilized at the beginning of alcoholic fermentation in correspondence with the highest rate of sugar consumption. In all the trials at the end of fermentation c. 50% of TAN was depleted; EC1118 and BM45 strains consumed more nitrogen than Zinfandel and S6U strains (Table 2). High correlation between sugar and TAN contents determined during alcoholic fermentation was observed. Strains displayed significant (P < 0·001) differences for sugar and TAN consumption rates.

Figure 2.

Total assimilable nitrogen (empty symbol) and fermentable sugars (fill symbol) depletion during fermentation of high sugar concentration grape must carried out by four wine yeasts (bsl00046bsl00067 Zinfandel, bsl00001bsl00000 S6U, bsl00066bsl00084 EC1118, •○ BM45). Bars are standard deviation

Table 2.  Specific sugars and total assimilable nitrogen consumption (TAN) rate and TAN content in the wines obtained from the fermentation of high sugar concentration grape must carried out by four wine yeasts
 ZinfandelS6UEC1118BM45
  1. Values ± standard deviation are average of three independent trials.

  2. qinline image, qS, maximum and average sugar consumption rate (g l−1 d−1) respectively.

  3. qinline image, qN, maximum and average TAN consumption rate (mg l−1 d−1) respectively.

  4. r2, Correlation coefficient between sugar and TAN content determined during the alcoholic fermentation.

  5. Numbers with different letters differ at *P < 0·001; ns, not significant.

qinline image*17·9 ± 0·51 a25·7 ± 0·00 b29·3 ± 0·35 c24·8 ± 0·76 b
qS*8·1 ± 0·04 a7·6 ± 0·07 b8·2 ± 0·02 a7·8 ± 0·05 c
qinline image10·4 ± 1·40 ns13·9 ± 2·54 ns14·6 ± 0·06 ns12·0 ± 1·63 ns
qN*3·9 ± 0·07 a4·0 ± 0·04 a4·4 ± 0·10 b4·3 ± 0·04 b
TAN (mg l−1)*168·00 ± 3·00 a164·43 ± 1·50 a146·53 ± 4·08 b155·00 ± 1·73 c
r20·945 ± 0·023 ns0·872 ± 0·041 ns0·892 ± 0·019 ns0·916 ± 0·024 ns

Wine analysis and statistical treatment of the data

The wines differed significantly for the contents of several compounds as shown in Table 3. High content of volatile acidity characterized the wines in particular those fermented by Zinfandel and BM45 strain where the concentration was over 1 g l−1. Five of eight higher alcohols (2-phenylethanol, 2-methyl-1-butanol, 3-methyl-1-butanol, 1-propanol and 2-methyl-1-propanol) differed significantly (P < 0·001) for their concentration in, at least, two of four wines. Zinfandel strain produced wines with the lowest higher alcohols content. The wines varied in total anthocyanin content, while colour density and hue were similar. Moreover, the wines had similar total polyphenol concentrations, but they differed for the trans-resveratrol content.

Table 3.  Wine composition determined when sugar consumption ceased. Values ± standard deviation are average of three independent trials
 ZinfandelS6UEC1118BM45
  1. †Tartaric acid (g l−1).

  2. ‡Acetic acid (g l−1).

  3. §Gallic acid (g l−1).

  4. Numbers with different letters differ at *P < 0·001 and **P < 0·05; ns, not significant.

Total acidity†** (g l−1)6·27 ± 0·03 a6·38 ± 0·08 ab6·17 ± 0·21 a6·68 ± 0·12 b
Volatile acidity‡* (g l−1)1·03 ± 0·06 a0·84 ± 0·02 b0·81 ± 0·08 b1·24 ± 0·02 c
Total SO2** (g l−1)16·00 ± 1·00 a23·67 ± 2·52 b26·67 ± 2·52 b25·67 ± 2·08 b
Free SO2 (mg l−1)3·00 ± 1·00 ns3·00 ± 1·00 ns2·67 ± 0·58 ns3·33 ± 2·31 ns
l-malic acid** (g l−1)0·41 ± 0·05 a0·52 ± 0·03 ad0·76 ± 0·11 bc0·71 ± 0·10 cd
l-lactic acid (g l−1)0·00 ± 0·00 ns0·01 ± 0·01 ns0·01 ± 0·02 ns0·02 ± 0·04 ns
d-lactic acid (g l−1)0·00 ± 0·00 ns0·08 ± 0·09 ns0·10 ± 0·05 ns0·08 ± 0·02 ns
Succinic acid* (g l−1)0·84 ± 0·01 a0·96 ± 0·02 b0·97 ± 0·01 b0·91 ± 0·01 c
Acetaldehyde* (mg l−1)35·33 ± 0·58 a36·00 ± 1·00 a46·00 ± 2·00 b52·00 ± 4·36 b
Ethyl acetate* (mg l−1)77·33 ± 4·73 a67·00 ± 1·00 b83·00 ± 5·29 ac90·67 ± 1·53 c
Ethyl lactate (mg l−1)0·00 ± 0·00 ns0·00 ± 0·00 ns0·00 ± 0·00 ns0·00 ± 0·00 ns
2-Phenyletanol* (mg l−1)26·33 ± 2·08 a39·00 ± 1·73 b25·00 ± 1·00 a24·33 ± 4·04 a
1-Butanol (mg l−1)2·33 ± 0·58 ns2·00 ± 0·00 ns3·67 ± 0·58 ns2·67 ± 1·15 ns
2-Butanol (mg l−1)1·00 ± 0·00 ns1·00 ± 0·00 ns1·00 ± 0·00 ns1·00 ± 0·00 ns
1-Hexanol (mg l−1)6·00 ± 0·00 ns5·00 ± 0·00 ns6·00 ± 0·00 ns6·00 ± 0·00 ns
2-Methyl-1-butanol* (mg l−1)29·33 ± 1·15 a43·00 ± 0·00 b40·00 ± 1·00 c36·00 ± 1·00 d
3-Methyl-1-butanol* (mg l−1)208·33 ± 5·13 a232·67 ± 2·52 b223·00 ± 3·61 b163·67 ± 3·51 c
1-Propanol* (mg l−1)66·67 ± 0·58 a72·67 ± 2·89 a137·00 ± 2·00 b179·67 ± 4·16 c
2-Methyl-1-propanol* (mg l−1)29·33 ± 1·15 a45·00 ± 0·00 b34·33 ± 2·08 c43·00 ± 1·00 b
Total higher alcohols* (mg l−1)369·33 ± 5·69 a440·33 ± 6·51 b470·00 ± 6·00 c456·33 ± 10·02 bc
Trans-resveratrol** (mg l−1)1·42 ± 0·14 a0·57 ± 0·14 b0·48 ± 0·13 b0·94 ± 0·42 ab
Total polyphenols§ (g l−1)1·26 ± 0·09 ns1·20 ± 0·10 ns1·31 ± 0·07 ns1·34 ± 0·05 ns
Total anthocyanins** (mg l−1)43·67 ± 6·03 a44·00 ± 0·00 a57·00 ± 1·73 b54·00 ± 3·61 b
Colour density2·53 ± 0·05 ns2·56 ± 0·13 ns2·95 ± 0·30 ns2·81 ± 0·29 ns
Hue1·36 ± 0·07 ns1·35 ± 0·05 ns1·37 ± 0·07 ns1·25 ± 0·11 ns

In order to emphasize the different contribution of each strain to wine composition PCA was carried out considering the compounds listed in Table 3, including residual sugars, ethanol, glycerol and TAN content and leaving out those variables absent (2-butanol and ethyl lactate). PCA yielded five principal components (PC); three of five PC explaining c. 75% of total data variance (Table 4). The variables which mainly contributed positively (PC loadings >0·800) to PC1 were total SO2, malic and succinic acid, to PC2 were residual sugars, glycerol, 2-phenylethanol and 2-methyl-1-propanol, whereas to PC3 were volatile and total acidity. Negative contribution (PC loadings less than −0·800) for variables as TAN to PC1, ethanol and 1-hexanol to PC2 and 3-methyl-1-butanol to PC3 was observed. As shown Fig. 3 the separation among the wines fermented by the same strain was total when PC1 was plotted to PC2 and PC3 (Fig. 3a,b). In the plot of PC1 vs PC2, the wines fermented by Zinfandel and EC1118 resulted close to each other (Fig. 3c).

Table 4.  Rotated principal component loadings resulting by principal component analysis for wines fermented by four yeast strains
 PC 1PC 2PC 3PC 4PC 5
  1. Values given in bold are PC loadings less than −0·800 and >0·800.

Residual sugars−0·0290·9700·1100·1260·061
Ethanol−0·0790·977−0·113−0·116−0·022
Glycerol−0·0150·899−0·400−0·1020·012
TAN0·8860·312−0·133−0·0580·193
Total acidity−0·0010·3930·8260·1350·140
Volatile acidity−0·328−0·1040·8970·1560·034
Total SO20·8660·2330·2420·053−0·023
Free SO20·0100·0630·1220·0340·943
l-malic acid0·858−0·1930·2310·2910·202
l-lactic acid0·2250·0540·1380·884−0·321
d-lactic acid0·7170·215−0·0870·3320·081
Succinic acid0·8510·398−0·2990·008−0·071
Acetaldehyde0·559−0·1990·746−0·002−0·137
Ethyl acetate0·323−0·5680·6840·1560·046
2-Phenylethanol−0·1490·838−0·393−0·171−0·204
1-Butanol0·552−0·516−0·059−0·1620·528
1-Hexanol0·0280·9080·3960·088−0·005
2-Methyl-1-butanol0·7100·633−0·245−0·106−0·063
3-Methyl-1-butanol0·0550·1920·906−0·295−0·081
1-Propanol0·635−0·2100·7180·1710·015
2-Methyl-1-propanol0·3510·8590·3410·040−0·014
Trans-resveratrol−0·793−0·3020·324−0·008−0·142
Total polyphenols0·183−0·3890·3450·4910·205
Total anthocyanins0·780−0·4030·337−0·1070·043
Colour density0·693−0·3330·0360·2760·435
Hue−0·045−0·090−0·3560·833−0·284
Proportion of total variance %27·88627·57020·1868·6016·903
Figure 3.

Plot of the three principal component in PCA of wines fermented by four wine yeasts (bsl00046 Zinfandel, bsl00001 S6U, bsl00066 EC1118, • BM45), (a) PC1 vs PC2, (b) PC1 vs PC2, (c) PC2 vs PC3

Discussion

The ethanol yield calculated for each strain under study resulted high and similar ethanol yield values were reported for industrial yeasts used for the production of alcohol classified as high ethanol producers (Oliveira et al. 2004). Buescher et al. (2001) improved ethanol production in wines fermenting 35° Brix sugar grape juice with moderate aeration and adding nutrient at the start of fermentation. It has been reported that in S. cerevisiae oxygen addition increase the fermentation rate in the case of stuck and sluggish fermentation (Sablayrolles et al. 1996). In our experimental condition the strains tested, in particular Zinfandel and EC1118, showed a high efficiency to produce ethanol without aeration and addition of nutrients.

The analysis of the fermentation conditions and of the fermentation profiles could help the winemaker in the identification of problem fermentations (Bisson and Butzke 2000). In our present case, the analysis of fermentation parameters indicated that ethanol toxicity was the main cause of stuck fermentation. The absence of mould infection on the grapes excluded the effects of zymostatic and zymocidal fungi toxins that are more toxic at higher ethanol level and could cause a late arrest of fermentation (Bisson 1999). The fermentation profile, obtained monitoring ethanol production, nitrogen and sugars consumption rate, displayed the ability of yeast cells to overcome rapidly the severe sugar stress and to start the metabolic activity as well. This result is in agreement with the view that higher sugar feed favours higher ethanol production regardless the existence of osmotic stress (Zhao and Lin 2003). Although, excluding bacterial contamination as resulted by l- and d-lactic acid content in the wines, high volatile acidity level indicated that osmotic stress occurred to the yeast cells, as previously observed (Caridi et al. 1999).

Often high sugar concentration juices could pose problem of fermentability because of nitrogen needs for the yeasts (Cramer et al. 2002). To overcome it, the addition of nitrogen to the juice is recommended (Bisson and Butzke 2000; Arrizon and Gschaedler 2002). Nevertheless, the amount of TAN found in the must used in this study resulted largely in excess with respect to the nitrogen requirement of the strains. It is probable that nitrogen utilization by strains ceased because of the inhibitory effect of high ethanol level for the translocation of amino acid and other nitrogen sources (Bisson 1999).

Concerning the analysis of the wine fermented, this study allows to evaluate the effects of the fermentation of high sugar concentration must on higher alcohols production. In the experimental wines produced by the strains tested here, the exhaustion of high residual TAN amount could induce further synthesis of higher alcohols negatively influencing the quality. Llauradòet al. (2002) reported higher contents of these alcohols in wine obtained from high sugar concentration musts richer in available nitrogen level.

Statistical treatment of the wine analysis data highlighted the different contribution of each strain to wine composition fermenting high sugar concentration must. It has been reported that different fermentative behaviours among S. cerevisiae strains, including commercial strains, also originated by a different expression of stress response genes during the vinification (Zuzuarregui and Del Olmo 2004b). Generally, differences of fermentative performances among the strains are attenuated in favourable winemaking conditions; on the contrary under stress the amplitude of the divergence in behaviour among strains increases. In our study testing the maximal capacity to consume must sugars in winemaking condition, we established the potentiality of four commercial strains to perform the fermentation of high sugar grape must and to produce high alcohol wines.

In conclusion, it was reconfirmed that high sugar concentration grape must could be fermented with high conversion efficiency of sugars to ethanol by commercial strain yeasts. Applying a traditional red vinification, we determined the effective ethanol yield for each strain in winemaking condition as well as we evaluated their alcohol tolerance. It is confirmed that under sugar stress the strains displayed different performance on production of principal and secondary fermentation metabolites. As a result of the increasing interest on high alcohol wine production further investigations are necessary to select strains with the best capability to produce high quality wines from high sugar concentration musts.

Ancillary