Improved Saccharomyces cerevisiae Strain in Pure and Sequential Fermentation with Torulaspora delbrueckii for the Production of Verdicchio Wine with Reduced Sulfites

The application of yeast strains that are low producers of sulfur compounds is actually required by winemakers for the production of organic wine. This purpose could be satisfied using a native Saccharomyces cerevisiae strain improved for oenological aptitudes. Moreover, to improve the aromatic complexity of wines, sequential fermentations carried out with S. cerevisiae/non-Saccharomyces yeast is widely used. For these reasons, in the present work an improved native S. cerevisiae low producer of sulfite and sulfide compounds was evaluated in pure and in sequential fermentation with a selected Torulaspora delbrueckii. Additionally, the influence of grape juices coming from three different vintages under winery conditions was evaluated. In pure fermentation, improved native S. cerevisiae strain exhibited a behavior related to vintage, highlighting that the composition of grape juice affects the fermentation process. In particular, an increase in ethyl octanoate (vintage 2017) and phenyl ethyl acetate (vintage 2018) was detected. Moreover, isoamyl acetate was highly consistent and could be a distinctive aroma of the strain. The sequential fermentation T. delbrueckii/S. cerevisiae determined an increase in aroma compounds such as phenyl ethyl acetate and ethyl hexanoate. In this way, it was possible to produce Verdicchio wine with reduced sulfites and characterized by a peculiar aromatic taste.


Introduction
The use of Saccharomyces cerevisiae as a starter culture significantly improves the control fermentation process, avoiding negative repercussions in winemaking. On the other hand, this biotechnological practice may determine a standardization of wine flavor [1]. Different approaches and alternative strategies to improve the analytical and aroma profile, enhancing the complexity of wines, were proposed. In this regard, several research studies evaluated the positive influence of non-Saccharomyces yeasts in mixed fermentation on the flavor complexity [2][3][4][5][6][7][8][9][10]. In addition, non-Saccharomyces yeasts exhibited other features such as bio-protective action, antimicrobial activity, and variations on the production of some compounds such as ethanol, glycerol, lactic acid in mixed

Analytical Procedures
Total acidity, volatile acidity, pH, ethanol content and total SO 2 were determined according to the Official European Union Methods (EC Regulation No. 2870/00) [26]. Enzymatic kits (Megazyme International Ireland) were used to measure the amounts of glucose and fructose (K-FRUGL), glycerol (K-GCROL), and malic acid (K-DMAL) according to the manufacturer instructions. Acetate strips (CARLO ERBA Reagents S.r.l., Milan, Italy) were used to evaluate the H 2 S production during the fermentation process. The production of H 2 S was detected by the increase in color of acetate strips from white to black in function of H 2 S during the fermentation. The semi-quantitative evaluation was expressed using an arbitrary scoring scale from zero (no H 2 S production, white strip) to 5 (maximum level of H 2 S production, black strip).
A specific enzymatic kit (kit no. 112732; Roche Diagnostics, Germany) was used to determine the ammonium content. The free α-amino acids were evaluated following Dukes and Butzke protocol [27].

Sensory Analysis
At the end of the fermentation, the wines obtained were transferred into completely filled 750 mL bottles, closed with the crown cap and maintained at 4 • C until sensory analysis. After storage for 3 months, wines were subjected to sensory analysis on the basis of smell and taste. A group of ten testers (80% expert and 20% non-expert), using a score scale of 1 to 10, expressed their opinion regarding smell and taste of each wine tested. The data obtained were used to compare the wines and provide information regarding the organoleptic quality and probable consumers' acceptability of the wines obtained.

Statistical Analysis
Analysis of variance (ANOVA) was used to elaborate the analytical character data of wine. The means were analyzed using the statistical software package JMP ® 11. Duncan tests were used to detect the significant differences. The experimental data were significant with associated p-values < 0.05. Principal Component Analysis (PCA) carried out using JMP 11 ® statistical software (Statistical discovery from SAS, New York, NY, USA) was used to analyze the mean values of each volatile compound and the main by-products of fermentation. The mean data were normalized to eliminate the influence of hidden factors.

Biomass Evolution and Sugar Consumption
Growth kinetics of the pure and sequential fermentations carried out during the vintages 2016-2018 are reported in Figure 1. The results showed that improved S. cerevisiae DiSVA 708 exhibited the same growth kinetics of OKAY ® starter strain in vintage 2016 (Figure 1a The sugar consumption ( Figure 2) exhibited a similar trend independently by fermentation trials and vintages, completing the fermentation process on the 10th day. Only S. cerevisiae OKAY ® showed a faster rate of sugar consumption in vintage 2016 ( Figure 2).

Statistical Analysis
Analysis of variance (ANOVA) was used to elaborate the analytical character data of wine. The means were analyzed using the statistical software package JMP ® 11. Duncan tests were used to detect the significant differences. The experimental data were significant with associated p-values < 0.05.
Principal Component Analysis (PCA) carried out using JMP 11 ® statistical software (Statistical discovery from SAS, New York, NY, USA) was used to analyze the mean values of each volatile compound and the main by-products of fermentation. The mean data were normalized to eliminate the influence of hidden factors.

Biomass Evolution and Sugar Consumption
Growth kinetics of the pure and sequential fermentations carried out during the vintages 2016-2018 are reported in Figure 1. The results showed that improved S. cerevisiae DiSVA 708 exhibited the same growth kinetics of OKAY ® starter strain in vintage 2016 (Figure 1a

Main Oenological Characters and Volatile Compounds of S. cerevisiae Pure Fermentations
The main analytical characters of the fermentation trials carried out during the three consecutive vintages are reported in Table 2. With regard to the ethanol content in the resulting wines carried out by the S. cerevisiae pure fermentations (vintages 2017-2018), the two strains exhibited a comparable alcohol content consuming all sugars. Only in vintage 2016, DiSVA 708 showed a lower ethanol content than that exhibited by OKAY ® due to a slight amount of residual sugars. On the other hand, the differences in the final ethanol content among the three vintages were due to different initial sugar content in the grape juice. OKAY ® fermentation trials showed a constantly lower total acidity than the DiSVA 708 pure fermentation trials through the vintages, while the production of volatile acidity showed more variation among the vintages although it was produced in low quantities. With regard to the glycerol content, S. cerevisiae fermentations did not show significant differences, with the exception of T. delbrueckii sequential fermentation (vintage 2017). A variable trend was exhibited for malic acid content.
The starter OKAY ® exhibited the lowest SO 2 production in all vintages, despite the variability found among the vintages. However, the SO 2 content of the other fermentation trials was in any case low in relation to the initial SO 2 content of grape juice. The score of H 2 S production was the same for both S. cerevisiae pure fermentation in vintages 2017-2018 while in 2016 it was significantly different.
The results of volatile compounds are reported in Table 3. DiSVA 708 led a greater general increase in esters content than did OKAY ® , with a different trend among the vintages. In particular, DiSVA 708 significantly increased ethyl acetate (2016); ethyl hexanoate (2017); phenyl ethyl acetate, ethyl octanoate and isoamyl acetate (2018). On the other hand, S. cerevisiae OKAY ® showed a significantly high production in ethyl butyrate in all vintages, a compound characterizing the volatile profile of this strain. With regard to higher alcohols, the variations found were more dependent on the harvest than on the inoculated strains. The production of n-propanol was an exception. Indeed, OKAY ® showed a production of n-propanol consistently and was significantly higher than the other trials in all vintages even when at different concentrations. Notably, an increase in acetaldehyde content in pure fermentation in vintage 2016 and a reduction in vintage 2018 by DiSVA 708 were observed.    0.02 ± 0.001 b,c 0.08 ± 0.02 a Geraniol 0.01 ± 0.01 b 0.00 ± 0.00 c 0.00 ± 0.00 c 0.01 ± 0.01 b 0.09 ± 0.02 a 0.00 ± 0.00 c 0.00 ± 0.00 c 0.00 ± 0.00 c

Main Oenological Characters and Volatile Compounds of T. delbrueckii/S. cerevisiae DiSVA 708 Sequential Fermentations
The results of the main oenological characters (Table 2) of T. delbrueckii/S. cerevisiae DiSVA 708 sequential fermentation highlighted a comparable trend with regard to ethanol, residual sugar, total and volatile acidity and malic acid in comparison to DiSVA 708 pure fermentation. Meanwhile, significant differences among the vintages were found to highlight the differences in grape juice composition between one vintage and the other. The use of T. delbrueckii in sequential fermentation allowed a greater reduction in SO 2 content than in DiSVA 708 pure fermentation in both vintages (significant only vintage 2017) ( Table 2). Moreover, T. delbrueckii sequential fermentations exhibited the same H 2 S production in vintages 2017-2018, but a higher score in comparison with S. cerevisiae pure cultures. T. delbrueckii sequential fermentations showed a significant reduction in glycerol content in comparison with the other trials while a variable trend of malic acid content was observed among the fermentation trials independently by strains and vintages.
The main by-products and the main volatile compounds are reported in Table 3. Sequential fermentation T. delbrueckii/S. cerevisiae DiSVA 708 showed a number of esters during vintage 2017 comparable with DiSVA 708 pure fermentation. In vintage 2018, the presence of T. delbrueckii led to an increase in phenyl ethyl acetate and ethyl hexanoate when compared with DiSVA 708 pure fermentation. Regarding the higher alcohol content, the use of T. delbrueckii led to wines with an increase in isoamyl alcohol and to a lesser extent of β-phenyl ethanol (vintage 2017). Monoterpenes compounds were generally improved by the presence and fermentation activity of T. delbrueckii, even if the geraniol in 2018 vintage was not detected. A significant reduction in acetaldehyde content was exhibited in vintage 2018.

Principal Component Analysis of the By-Products and Main Volatile Compounds
The PCA of the main volatile compounds and all by-products was used to assess the effect of vintage and the inoculated pure and sequential fermentations (Figure 3). The graphical representation indicated that the fermentations were grouped according to both vintage and inoculated S. cerevisiae. Indeed, it is possible to group for the vintages (bottom-right 2016; bottom-left 2017; upper 2018). Strain effect was also detected. S. cerevisiae OKAY ® trials were grouped together in bottom quadrants. The sequential fermentations carried out with T. delbrueckii were grouped together in the upper-left quadrant and closed to DiSVA 708 pure fermentations of the same vintage that showed some variations. DiSVA 708 2016 trials were different from the other trials (2017-2018), probably caused by an incomplete fermentation (residual sugar). The results obtained by PCA analysis showed that the final composition of wine was affected at different levels by the vintage, inoculated S. cerevisiae and non-Saccharomyces used in sequential fermentation.

Sensory Analysis
The wines obtained by pure (OKAY ® and S. cerevisiae DiSVA 708) and sequential fermentations (T. delbruekii DiSVA 130/S. cerevisiae DiSVA 708) underwent sensory analysis, and the results are reported in Table 4. All the wines obtained did not show significant differences regarding smell and taste. However, the testers expressed a positive judgement regarding each wine, characterized by specific aromatic notes and without defects.

Discussion
The wide use of a limited number of S. cerevisiae strains as starter cultures of fermentation could standardize the aroma and analytical profile of wine [1,29]. To overcome this problem, the use of selected native S. cerevisiae strains as starter cultures of fermentation could be a suitable strategy [21,22]. Another strategy to enhance the wine aroma complexity is the use of non-Saccharomyces yeasts in sequential fermentations. Indeed, several studies emphasize the use of non-Saccharomyces yeasts in winemaking for their attitudes to produce aromatic compounds (such as esters, terpenes, thiols) and modify some structural compounds as polysaccharides, volatile acidity and ethanol content [30][31][32][33][34][35][36].
In this work, we evaluated the fermentation performance of a native strain of S. cerevisiae (DiSVA 708) improved for the absence of H2S production and reduced production of SO2 [21] to be used in the production of organic wines with a low SO2 content. In pure fermentation, DiSVA 708 strain exhibited a behavior related to vintage, highlighting that the composition of grape juice can affect the fermentation performance of yeast strain. In particular, the increase in different aromatic compounds was detected among the vintages: in 2017 vintage ethyl octanoate, responsible for fruity flavors such as apple, pear and peach [37][38][39] and in 2018 vintage phenyl ethyl acetate, responsible for floral and

Discussion
The wide use of a limited number of S. cerevisiae strains as starter cultures of fermentation could standardize the aroma and analytical profile of wine [1,29]. To overcome this problem, the use of selected native S. cerevisiae strains as starter cultures of fermentation could be a suitable strategy [21,22]. Another strategy to enhance the wine aroma complexity is the use of non-Saccharomyces yeasts in sequential fermentations. Indeed, several studies emphasize the use of non-Saccharomyces yeasts in winemaking for their attitudes to produce aromatic compounds (such as esters, terpenes, thiols) and modify some structural compounds as polysaccharides, volatile acidity and ethanol content [30][31][32][33][34][35][36].
In this work, we evaluated the fermentation performance of a native strain of S. cerevisiae (DiSVA 708) improved for the absence of H 2 S production and reduced production of SO 2 [21] to be used in the production of organic wines with a low SO 2 content. In pure fermentation, DiSVA 708 strain exhibited a behavior related to vintage, highlighting that the composition of grape juice can affect the fermentation performance of yeast strain. In particular, the increase in different aromatic compounds was detected among the vintages: in 2017 vintage ethyl octanoate, responsible for fruity flavors such as apple, pear and peach [37][38][39] and in 2018 vintage phenyl ethyl acetate, responsible for floral and honey aroma [38]. On the other hand, isoamyl acetate (banana flavors) was consistently highly produced by DiSVA 708 strain which can be considered a distinctive aroma of the strain. On the contrary, OKAY ® showed a more homogenous wine aroma across all three vintages with ethyl butyrate and n-propanol as a distinctive aroma compound [40,41].
In sequential fermentation, the results showed an influence on analytical profiles of the Verdicchio wines by T. delbrueckii selected strain. In particular, the sequential fermentation carried out during the 2017 vintage showed a reduced content of volatile acidity than the wine obtained by the pure fermentation of the native S. cerevisiae DiSVA 708. The ability to reduce the volatile acidity by T. delbrueckii is a feature well documented [2,42,43]. Moreover, the use of T. delbrueckii in sequential fermentation led to a reduction in total SO 2 content. Actually, there are no studies that directly correlate low sulfur dioxide production with the use of selected non-Saccharomyces yeasts. However, Bely and co-workers [42] highlighted the ability of T. delbrueckii to reduce SO 2 -binding compounds production. We found that T. delbrueckii in sequential fermentation reduced acetaldehyde content.
The sequential fermentation T. delbrueckii/S. cerevisiae determined a general increase in some aroma compounds affecting the aroma complexity of the wine, such as phenyl ethyl acetate, ethyl hexanoate and β-phenyl ethanol, even if not at significant levels. Several previous works highlighted the tendency of T. delbrueckii to form higher levels of acetate compounds, β-phenyl ethanol and a general improvement of the quality of wine aroma increasing the perception of varietal and fruity characters [15,44,45].
On the other hand, confirming a previous work [15], the aroma profile of resulting wines was affected not only by non-Saccharomyces/S. cerevisiae used in sequential fermentation, but also by the vintage.
Overall data emphasized the use of selected non-Saccharomyces yeasts, such as T. delbrueckii, in sequential fermentation with selected native S. cerevisiae strains for Verdicchio wine production. Indeed, on one side, the use of native S. cerevisiae reinforced the distinctive aromatic characteristics of a particular enological area, usually conferred by native yeast strains. On the other side, non-Saccharomyces yeast produced peculiar fermentation by-products that positively influenced the analytical profile of Verdicchio wine, characterized by a well-structured style. Furthermore, a suitable strategy to obtain wine with reduced sulfites content is the use of improved native S. cerevisiae DiSVA 708 (low sulfite producer) in pure and mixed fermentation with a selected T. delbrueckii strain.
Author Contributions: A.A., L.C., F.C. and M.C. contributed equally to the manuscript. All the authors participated in the design and discussion of the research. A.A. carried out the experimental part of the work. A.A. and L.C. carried out the analysis of the data. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding