The aptitude of commercial yeast strains for lowering the ethanol content of wine

Abstract The high alcohol content in wine usually has a negative impact on its sensory properties, but can also affect the general health of the consumers. The possibility to reduce ethanol production in wines during fermentation involves the use of different yeast strains characterized by the increased production of fermentation by‐products (glycerol, 2,3‐butanediol, etc.) from the available sugar. The activity of these strains should not impair the sensory properties of the wine. In general, the use of genetically and evolutionarily (non‐GM) engineered Saccharomyces cerevisiae strains is still not close enough to commercial application, and therefore, it is unavailable for wine producers. Thus, the aim of this study was to examine the possibility of reducing the production of ethanol in wines using different selected yeast strains (S. cerevisiae, Saccharomyces bayanus, Torulaspora delbrueckii, and Metschnikowia pulcherrima) available at the market. The application of individual yeast and sequential inoculation for wine alcoholic fermentation was examined. The achieved effects were evaluated by determining the content of ethanol, as well as fermentation by‐products (glycerol and volatile acids) and aromatic components in wine samples. Depending on the strain/s used, a decrease in ethanol content of up to 0.9% v/v was recorded in comparison with fermentation by S. cerevisiae alone. The sensory analysis of produced wine showed significant differences in taste and flavor. The results of the experiment conducted at the laboratory level and with the use of sterile must were compared to the ones from the scale‐up experiment in real vinification conditions. The observed differences in the alcohol content of produced wines were significantly lower.

on the sensory properties, which increases the perception of the heat and alters the perception of wine aroma complexity (Goldner, Zamora, Di Leo Lira, Gianninoto, & Bandoni, 2009). Also, excessive alcohol intake through the consumption of wines with higher ethanol levels is often associated with undesirable implications on human health. Furthermore, higher alcohol in wine may increase costs in countries where taxes are levied according to alcohol concentration. Thus, the combination of quality, health, and economic issues associated with high-alcohol wines has created significant interest in the development of technologies for the production of reduced ethanol wines.
Different approaches to reduce alcohol levels in wines have been proposed at all stages of the winemaking process. These mainly fit into four basic groups as viticultural, prefermentation, fermentation, and postfermentation strategies . Viticulture strategies, as promising but long-term techniques, are based on the selection of new grape varieties with low sugar accumulation, viticultural practices adapted to unripe grapes, and different agronomical methods (Olego et al., 2016). On the other hand, postfermentation strategies such as reverse osmosis, nanofiltration, and distillation represent a short-term perspective dependent on the current EU and OIV regulations. Moreover, these procedures may increase production costs and also can compromise the wine organoleptic quality due to the elimination of volatile compounds (Schmidtke, Blackman, & Agboola, 2012). Considering possible approaches, the application of yeast strains characterized by lower sugar-to-ethanol transformation rates has been imposed as an attractive way to deal with the problem of high-alcohol wines (Kutyna, Varela, Henschke, Chambers, & Stanley, 2010). Lower ethanol-producing yeast strains could be isolated and characterized from spontaneous wine alcoholic fermentations or obtained through the application of adaptive evolution (development of the low-alcohol variants of existing Saccharomyces cerevisiae strains) and genetic modification techniques (GMO approaches) (Ozturk & Anli, 2014;Varela et al., 2015). Due to poor consumer acceptance of GMO foods and beverages, there is a need to investigate and develop the non-GMO approaches for the generation of wine yeasts that produce less ethanol (Kutyna et al., 2010).
Nonconventional yeasts such as Kloeckera, Pichia, Candida, Metschnikowia, Schizosaccharomyces, and Torulaspora species are among the main representatives of grape natural microbiota. In general, their pronounced sensitivity to antimicrobial agents (e.g., SO 2 ) and higher alcohol contents prevent the complete transformation of grape sugars into ethanol during alcoholic fermentation. Therefore, their application in co-inoculation or sequential inoculation with S. cerevisiae is increasingly getting popular especially regarding their potential positive effects on wine flavor . On the other hand, species, such as Saccharomyces bayanus, are associated with spontaneous fermentation of must and have been shown to be of oenological interest (González, Barrio, Gafner, & Querol, 2006). The use of mixed cultures of selected Saccharomyces and non-Saccharomyces yeasts for wine fermentation can result in the formation of higher amounts of undesired compounds (e.g., volatile phenols and ethyl acetate) which can affect both structure and the aromatic profile of the wines. Therefore, ensuring the expression of appropriate metabolic characteristics of different yeast strains individually or in mixed cultures might serve as an efficient mechanism for reducing ethanol production in wines (Ciani & Comitini, 2015).
The application of non-Saccharomyces species for decreased alcohol production is possible through both aerobic (respiration) and anaerobic (fermentation) metabolism .
The aim of this research was to examine the possibility of reducing ethanol production in wines using different selected yeast strains (Saccharomyces cerevisiae, Saccharomyces bayanus, Torulaspora delbrueckii, and Metschnikowia pulcherrima) which are currently available at the market. The experiments implied a series of wine fermentations carried out both by the activity of chosen strains individually and in the form of mixed cultures applied through sequential inoculation.

| Inoculation strategies for the fermentation of experimental wines
In order to investigate the possibility of reducing the production of ethanol in wine using different yeasts as producing microorganisms, the following commercial strains were used: 1. Saccharomyces cerevisiae Oenoferm Bouquet (Erbslöh, Germany), shortened CER.

| Laboratory-scale experiment
Wines were produced from the Serbian white grape variety Sila (Vitis vinifera L.), from grapes originating from the experimental vineyards of the Faculty of Agriculture, University of Novi Sad, located in Sremski Karlovci. Grapes were harvested at the stage of technological maturity (optimal ratio between sugar and acids, phenolic and aromatic maturity also ensured), at the end of September 2016. The processing of grapes included crushing and destemming (Zambelli Gamma 30), followed by pressing in classical vertical basket press (capacity 100 kg). The sugar content in the must was 19.5%, total acidity 5.1 g/L (as tartaric acid), and assimilable nitrogen 225 mg/L. The must was sulfited by the addi-

| Scale-up experiment
The experiments conducted at laboratory level and with the use of pasteurized must were followed by the scale-up trials. The goal was to evaluate the previously obtained results in real winemaking conditions which do not imply the pasteurization of the must. For this purpose, experiments were carried out in 200-liter stainless still tanks. Grape processing and wine fermentation were conducted in an identical way as in the case of the laboratory experiment. The only difference was the use of fresh unpasteurized grape must. After the end of fermentation, the content of ethanol, glycerol, and volatile acids was determined.

| Analyses
Total sugars, total acidity (expressed as tartaric acid), volatile acidity (expressed as acetic acid), and ethanol content (hydrostatic balance Densi Alcomat; Gibertini) of the must and produced wines were determined using official OIV methods (OIV, 2016). The content of yeast assimilable nitrogen in the must was determined by formol titration method (Zoecklein, Fugelsang, Gump, & Nury, 1999). The content of glycerol was determined by a commercial enzyme test (Megazyme, CO).

| Wine sensory evaluation
Buxbaum model of positive ranking, described in the paper of Amerine and Roessler (1983) Wines were presented to panelists in ISO standard wine glasses, in isolated booths, and under daylight-type lighting. All wine samples were evaluated by the panel (seven trial sets, Table 1) with randomized presentation order. Three sessions were employed in three consecutive days where the panelist evaluated all seven individual wines each day in two sets. Each replicate presented on three consecutive days of tasting was poured from a separate bottle.

| Statistical analysis
Statistical analysis in the present study was performed using Statistica 12.0 (StatSoft). The statistical difference between mean values of parameters was estimated by analyses of variance (ANOVA), at the 95% confidence level. Values detected as significantly different by the use of Duncan multiple range test were marked with different letters (a, b, c …).

| The influence of different yeast strains' metabolic activity on ethanol production
Different selected yeast strains were used in wine fermentation trials, and the effect of their metabolic activity on the content of most important fermentation products was assessed. At the same time, the activity of autochthonous microbiota was suppressed by pasteurization of the must.
Sugar consumption profiles during the fermentation of must inoculated according to the defined inoculation plan are shown in Figure 1. Glycerol is nonvolatile three-hydroxy alcohol which indirectly contributes to the sensory character of a wine. At high concentration, it contributes significantly to the sweetness, body, and fullness of wines. For these reasons, glycerol production is one of the desirable features during grape must fermentation. Furthermore, of all the microbiological strategies to reduce ethanol yield, glycerol overproduction was proven to be the most effective (Varela et al., 2012).
Among experimental wines produced in this study, the highest glycerol content (6.99 g/L) was obtained during sequential fermentation

| The impact of used yeast strains on the composition of volatile compounds in wines
The aroma is one of the main characteristics that determine the qual-  Table 2.
Metabolic activity of commercial selected yeast strains during fermentation caused significant differences in the content of aromatic compounds in produced wines. As shown, ten higher alcohols and two aldehydes were detected in the analyzed wine samples; meanwhile, the compounds such as 2-propanol, 1-butanol, 2-butanol, 1-pentanol, and 1-hexanol were not detected. The produced wines were characterized by relatively uniform content of ethyl acetate, 2-methyl-1-propanol, that is, isobutanol, octanol, and decanol. The use of commercial non-Saccharomyces strains did not lead to the increase in the ethyl acetate levels, and the amounts determined (about 50 mg/L) were far below the threshold associated with a negative impact on wine sensory characteristics (above 150 mg/L).
The content of acetaldehyde was the highest in the MET+CER and MET+BAY+CER samples (15-20 mg/L); however, even the highest levels determined were significantly lower than the ones associated with oxidative faults ("over-ripe bruised apples," "sherry," and "nut-like" characters). At lower levels (below 50 mg/L), acetaldehyde can contribute pleasant fruity aromas to a wine. As for comparison, Varela et al. (2017) and Canonico et al. (2016) reported that wines produced with M. pulcherrima showed higher concentrations of ethyl acetate, total esters, total higher alcohols, and total sulfur compounds  Table 1. a,b,c different letters for the results in every specific day of alcoholic fermentation indicate significant differences between values (p < .05) and benzyl alcohol (9.7-11.2 mg/L) were detected in wines produced by the activity of S. cerevisiae, in both individual and sequential inoculations (samples CER, MET+CER, and MET+BAY+CER). The presence of furfural was detected in small amounts in all wine samples.
One of the main sources of furfural in wines is toasted oak wood.
Considering the fact that wines in this study were not in contact with wood, the possible explanation for furfural occurrence is the pasteurization of must before inoculation of yeast strains.

| Scale-up experiment
Evaluation  (Ciani & Ferraro, 1998;Ferraro, Fatichenti, & Ciani, 2000). The use of Starmerella bombicola (formerly C. stellata) and S. cerevisiae resulted in a high production of glycerol, succinic acid, and different by-products (interactions involving acetaldehyde and acetoin) with a consequent reduction of final ethanol amount. The reduction in ethanol content in these experiments varied from 0.64% v/v at pilot scale in natural grape juice to 1.60% v/v at laboratory scale using synthetic grape juice. Furthermore, different factors affecting the metabolism of M. pulcherrima AWRI1149 during fermentation of nonsterile Shiraz must were evaluated . Among different inoculation regimes which were applied, only initial inoculation with 1 × 10 6 cells/mL of M. pulcherrima AWRI1149, followed by S. cerevisiae after 50% sugar consumption, leads to a significant ethanol concentration reduction. Canonico et al. (2016) showed that application of immobilized selected strains of M. pulcherrima, followed by inoculation of free S. cerevisiae cells, resulted in a decrease in ethanol content for 1.3% v/v when synthetic grape juice was used, while in the case of natural grape juice the reduction was 1% v/v (in both trials, the beads of immobilized yeast were removed after 72 hr from inoculation).

| CON CLUS ION
In summary, this work evaluated the potential of commercial selected yeast strain application for purpose of wines with decreased alcohol content production. Sequential inoculation of the must with M. pulcherrima, S. bayanus, and S. cerevisiae resulted in the production of wines with the lowest ethanol content among experimental samples (decrease of 0.9% v/v compared to the control wine). Significant differences in the content of certain aromatic compounds, as well as in taste and flavor, were also found F I G U R E 5 Final concentrations of glycerol, volatile acids, and ethanol in wines produced by different commercial yeast strains used in scale-up conditions. Marks describing each trial in the figure legend: Saccharomyces cerevisiae-CER; Saccharomyces bayanus-BAY; Torulaspora delbrueckii-TOR; Metschnikowia pulcherrima-MET; a detailed inoculation plan is given in Table 1. a,b,c different letters indicate significant differences between values (p < .05) in produced wines. The experiment in real conditions showed that used commercially available Saccharomyces and non-Saccharomyces were not effective enough in lowering ethanol concentration in wines due to interactions and competitiveness with yeasts from autochthonous microbiota. Already evident problems of wines with very high alcohol content will force wine producers in near future to find more efficient ways of directing alcoholic fermentation to the production of wines with lower alcohol content and not worsened sensory properties. The application of Saccharomyces and non-Saccharomyces strains obtained by adaptive evolution principles (non-GM) for this purpose will need further investigation and commercialization.

ACK N OWLED G M ENT
Financial support from the Ministry of Education, Science and Technological Development of the Republic of Serbia (Project TR-31002) is greatly appreciated.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest.

E TH I C A L A PPROVA L
This study does not involve any human or animal testing.