Base Wine and Traditional Sparkling Wine Making Using Torulaspora delbrueckii Killer Yeasts

: The killer strains of Torulaspora delbrueckii can be used to improve the dominance of this yeast during must fermentation. The present work analyzes its usefulness for traditional sparkling wine making. T. delbrueckii killer strain dominated base wine fermentation better than non-killer strains and produced dried wines. The foam ability of T. delbrueckii base wines was very low compared to that of Saccharomyces cerevisiae . Significant positive correlations of foam parameters were found with some amounts of C 4 –C 16 ethyl esters and proteins, and negative correlations with some antifoam alcohols. The organoleptic quality of T. delbrueckii base wines was considered un-usual for cava making. While S. cerevisiae (single or mixed with T. delbrueckii ) completed the second fermentation to produce dry sparkling wines with high CO 2 pressure, single T. delbrueckii did not complete this fermentation, leaving sweet wines with low CO 2 pressure. Death due to CO 2 pressure was much higher in T. delbrueckii than in S. cerevisiae , making any killer effect of S. cerevisiae on T. delbrueckii irrelevant. However, the organoleptic quality of cava inoculated with mixtures of the two yeast species was better than that of wine inoculated exclusively with S. cerevisiae , and no deterioration in the quality of the foam was observed.


Introduction
The use of non-Saccharomyces yeasts is being highly recommended for winemaking because it can improve their organoleptic complexity. Specifically, Torulaspora delbrueckii is the most non-Saccharomyces yeast species used after Saccharomyces cerevisiae in the wine industry. It is a torula-shaped yeast that is smaller and grows somewhat slower than S. cerevisiae, but it has interesting technological advantages for making still wines. Its usefulness has been confirmed in several published works, it can reduce volatile acidity and acetaldehyde levels in wines and increase some interesting dried fruit and pastry aromas [1]. Furthermore, it has recently been verified that the sequential inoculation of T. delbrueckii and S. cerevisiae increases the concentration of glycerol, reduces volatile acidity, and exerts a positive effect on the foam properties of base wines to produce sparkling wines [2].
Foam formation and its stability are very important organoleptic characteristics valued by consumers in sparkling wines such as "cava" (closed-bottle-fermented sparkling wine). It has been described that the foam of cava, mainly the foam stability, de-  pends to a great extent on its content of proteins and mannoproteins [3,4]. It has also been reported that the foam maximum height (HM) correlates negatively with C8, C10, and C12 fatty acids, and positively with the ethyl esters of C6, C8, and C10 fatty acids [5]. These studies have been carried out with cava made entirely with Saccharomyces yeasts. Few studies have been carried out with cava made with non-Saccharomyces yeasts such as Torulaspora.
An alternative to accelerate the yeast autolysis is to use mixtures of killer and sensitive yeasts as inocula in the second cava fermentation. Killer toxins can kill sensitive cells and accelerate their autolysis [6]. This strategy has not been tested at the winery level until very recently. In this work, it was demonstrated that inoculation with mixed cultures of S. cerevisiae killer yeast caused cell death and early autolysis of sensitive yeasts during cava-winemaking, without negatively affecting fermentation kinetics or the consequent increase in pressure, improving the cava foam andits organoleptic quality [7]. To complement these results, it is of interest to analyze the utility of killer T. delbrueckii yeast strains, which can dominate must fermentation [1,8], in producing base wine and cava. Furthermore, given the killer effect, it can enhance yeast autolysis and cava quality.It is also necessary to analyze the usefulness of T. delbrueckii sensitive strains. This work analyzes the capacity of T. delbrueckii (killer and sensitive) to dominate and complete the fermentation in basewinemaking, to carry out the second fermentation at high CO2 pressure, and the aromatic and sparkling wine quality of the base wine and cava made with T. delbrueckii with respect to S. cerevisiae.

Materials and Methods
For base wine making, a cold-settled Macabeo grape must was used, inoculating with two T. delbrueckii: EX1180-11C4 (killer Kbarr-1 and resistant to cycloheximide, cyh R ), and EX1180-2K − (no-killer, cyh R ); and two S. cerevisiae: E7AR1 (killer K2, cyh R ) and EX85R (no-killer, cyh R ), yeast strains. For cavawinemaking, a base wine blend of S. cerevisiae was used. Before inoculating the wine, the yeasts were adapted to growth in this medium as previously described [9] and 2.4% sucrose and 0.02% diammonium phosphate were added. Subsequently, the base wine, single (with S. cerevisiae EX229, killer Klus and sensitive to cycloheximide, cyh S ; or T. delbrueckii EX1180-2K − ), and mixed (EX229+EX1180-2K − ), wasinoculated in 0.75 L cava bottles, inoculating about 1-4×10 6 cells/mL forS. cerevisiae or2-4×10 7 for T. delbrueckii and incubating at 18-19 °Cfor 15 days, to enhance the killer effect, which is more effective at this temperature, and then at 12-14 °C for up to 9 months. During the first and second fermentation,the yeast population was monitored by analyzing its resistance to cycloheximide (cyh R ) by replica-plating on YEPD (yeast extract peptone dextrose) (1% Bacto yeast extract, 2% Bacto-peptone, 2% glucose, 2% Bacto-agar) plates supplemented with cycloheximide. For the first fermentation, must density was monitored every day; and for the second fermentation, the pressure was measured (expressed in atm at 20°C) using an aphrometer. Cell death was counted by methylene blue staining, mannoprotein, and protein content as previously described [9]. The wine aroma compounds were measured by GC-MS (Gas Chromatography-Mas Spectrometry), and the foaming parameters using a Mosalux system as described previously [9]. The principal analytical parameters were determined according to EC (Amending regulation EEC Nº 2676/90) (Com-recommended methods and the organoleptic analysis was carried out by a wine-tastingexpert as described previously [9]. The statistical analysis of the data was performed with the parametric ANOVA test (p<0.05), Pearson's correlation, and Duncan's test, using SPSS software version 20.0 for Windows (Chicago, IL).

Enhance of Killer T. delbrueckii Yeasts on the First Fermentation and Quality of Base Wine
Fermentation kinetics inoculated with T. delbrueckii strains were generally slower than those of S. cerevisiae. However, base wines inoculated with killerT. delbrueckii dominated fermentation more easily than non-killer T. delbrueckii and left the wines dried ( Figure 1). In the descriptive organoleptic analysis, T. delbrueckii wines were clearly different from those of S. cerevisiae. Wine tasters appreciated the latter as they were more intense and fruitier, although the differences in valuation were not statistically significant. S. cerevisiae wines were foamier, had more protein, and better foamability (HM) and stable foam (HS). T. delbrueckii wines were spicier, with more aging notes, more polysaccharides, and better foam stability time (TS). The concentration of ethyl esters, acetate esters, furans, volatile phenols, and organic acids was higher in the S. cerevisiae wines, which would explain their greater aromatic intensity and more fruity character. The higher quantity of proteins could also explain its greater foamability, and its higher quantity of glycerol could explain the lower stability of the foam. The higher amount of alcohol in T. delbrueckii wines can explain its lower foamability, and its higher amount of polysaccharides can explain that the little foam that is formed is more stable ( Table 1).  In general, considering all the wines together, there was a significant positive correlation of HM and HS with proteins and 31 aromatic compounds, mainly C4-C16 ethyl esters; and TS with various alcohols. The correlation of HM and HS with polysaccharides was negative, as was that of TS with other 35 compounds, mainly alcohols ( Figure 2). Some of these foam correlations with aromatic compounds have already been previously described for sparkling wines, especially the positive correlations with C4-C16 ethyl esters [5,7], indicating that wine compounds other than polysaccharides and proteins may be importantly implicated in the wine's foaming quality. To continue with the elaboration of cavawinemaking, a base wine blend of S. cerevisiae was used for this objective. The organoleptic properties of the base wine of T. delbrueckii were considered anomalous for this purpose, nonetheless, these wines were considered of good quality and without defects.

Utility ofT. delbrueckii on the Second Fermentation and the Quality of the Sparkling Wine
Fermentation with S. cerevisiae (single or mixed with T. delbrueckii) was very efficient, reaching 6 or more atm of pressure at 60 days. In contrast, single yeast T. delbrueckii showed little viability and did not complete the second fermentation under these conditions. The percentage of dead cells was always higher in T. delbrueckii fermentation, single or mixed, and S. cerevisiaetotally replacedT. delbrueckii at 60 days (not shown). The S. cerevisiae and S. cerevisiae + T. delbrueckii cava wines were of good quality, as indicated by the physical-chemical parameters and the organoleptic analysis ( Table 2). The wines with mixtures of S. cerevisiae + T. delbrueckii were also the most valued for their complexity, better mouthfeel, notes of dried fruit, and pleasing aged character. On the contrary, the T. delbrueckii cava wines presented low levels of pressure, alcohol, and total acidity, and higher levels of volatile acidity, reducing sugars, and pH, which explain its low score in the organoleptic analysis (Table 2). In general, the foam parameters of cava wines were worse than those of base wines ( Figure 3A). S. cerevisiaecava wines (single or mixed with T. delbrueckii) had the best HM, and those of T. delbrueckii (single or mixed with S. cerevisiae) had the best TS and greater amount of total polysaccharides and mannan ( Figure 3A, B). Although there were no differences in the amount of protein between the three types of cava wines, in all of them, it increased by 30% compared to the base wine ( Figure 3B). These results suggest that the amount of these compounds is less relevant than previously thought [3,4], at least in our working conditions. Nor was any correlation found between the foam properties and the aromatic compounds, probably because the differences in these parameters in these cava wines were relatively small as they all came from the same base wineblend. On the other hand, there were significant differences in 15 of the 75 volatile compounds analyzed: seven compounds more abundant in T. delbrueckii cava wines and eight more abundant in S. cerevisiae and S. cerevisiae + T. delbrueckii, mainly ethyl esters responsible for fruity aromas, and with a relevant odor activity value (OAV), such as ethyl hexanoate, ethyl octanoate, and β-damascenone ( Figure 3C). These results are like what was previously observed for still wines [1,8].

Conclusions
The killer phenotype permitted T. delbrueckii to reduce the presence of wild yeasts during must fermentation. Nonetheless, the lower aromatic quality and lower capacity to form foam in their base wines make this yeast unsuitable for cava winemaking, although it could be interesting to produce other types of wines. Furthermore, the exclusive inoculation of T. delbrueckii did not complete the second fermentation, which also discourages its use for this purpose. Nevertheless, the mixed inoculation of S. cerevisiae + T. delbrueckii in the second fermentation proved to be a good strategy to enhance the organoleptic quality of the cava wine, mainly because T. delbrueckii enhanced a greater amount of some compounds of interest and improved the foam stability. Funding: This research was funded by Extremadura Regional Government and Spanish Ministry of Economy, Industry and Competitiveness (co-financed with FEDER funds), grant number GR18117 and AGL2017-87635-R".

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.