Wine Fermentation Performance of Indigenous Saccharomyces cerevisiae and Saccharomyces paradoxus Strains Isolated in a Piedmont Vineyard

: The role of yeast in wine quality is very important. The use of selected autochthonous yeasts is becoming more and more frequent in enology, not only to obtain a diversiﬁcation of wines, but also as a link between the wine and its territory of origin. The objectives of this work were to test two indigenous yeasts in a cellar on a pilot scale. The yeasts were a strain of Saccharomyces cerevisiae and a strain of Saccharomyces paradoxus previously isolated in a vineyard in Piedmont (Italy). Studying the oenological characteristics of S. paradoxus is of particular interest, as it is rarely found in the cellar–vineyard environment. Molecular biology methods conﬁrmed the predominance of the strain inoculated in the various fermentation tests. Additionally, products of yeast metabolism, including volatile compounds, were quantiﬁed at the end of the alcoholic fermentation and sensory proﬁle of wines was tested by a trained panel of tasters. Our results indicated that both strains have good characteristics to be used as starter in winemaking; S. paradoxus was characterized by a high production of glycerol and the ability to degrade malic acid, together with a lower production of ethanol and a low volatile acidity, while S. cerevisiae conferred to the wine a pleasant smell of rose, as highlighted in the sessions of sensory analysis.


Introduction
Wine quality depends mainly on the characteristics of the grapes and the diversity of microorganisms (yeasts and bacteria) present during winemaking. During alcoholic fermentation (AF) yeasts of the genus Saccharomyces have a predominant role in completing the fermentation of the grape sugars to ethanol. At the end of the fermentation only the species best adapted to the high ethanol content, in many instances S. cerevisiae and S. bayanus, are found [1].
However, it is known that the grape-wine ecological habitat has a much more complex microbial biodiversity; in fact, wine fermentation is not a 'single species' fermentation, but it is the result of a composite community where the indigenous yeasts also play an important role in the final wine complexity [2,3].
In this contest, these yeasts, Saccharomyces and non-Saccharomyces, are interesting in oenology, because together with the soil or the microclimate they represent the territory of origin of the wine [4]. Actually, their use is growing in response to the ever-increasing need for wine personalization, which is becoming an important aspect of winemaking [5,6]. Nowadays, wineries are dealing with new challenges due to market demands and climate Beverages 2021, 7, 30 2 of 10 change and the selection and the use as starter of non-conventional yeasts can be beneficial since they represent an important resource of biodiversity.
S. cerevisiae is one of the most extensively studied yeast species because it is the main actor of industrial fermentations, such as wine, beer, and bread production [7]. S. paradoxus is phylogenetically closed related to S. cerevisiae; it is frequently found in association with oak trees [8][9][10], but rarely with vineyards. S. paradoxus, being the subject of studies on ecology and evolution, was the first Saccharomyces yeast to be acknowledged as a nondomesticated species [11,12]. Majdak et al. [13] and Orlić et al. [14] reported the possibility of using S. paradoxus strains as starter in fermentation because of their contribution to the aroma of the wines. Alonso del Real et al. [15] tested one S. paradoxus strain in a mixed co-culture with one S. cerevisiae but in their results they found that during the fermentation S. cerevisiae dominated over S. paradoxus. Due to these divergent data, further studies must be conducted to increase knowledge about the behavior of S. paradoxus in wine.
A very important aspect of quality is determined by the aroma components of the wine: Varietal aromas (originating from the grapes) and fermentative aromas (originating during alcoholic and malolactic fermentations). The contribution of yeast fermentation metabolites to the aromatic profile of wine is well documented [16,17], but yeast can also contribute to wine aroma by other mechanisms, the de novo biosynthesis of volatile compounds and the transformation of neutral grape compounds into flavor-active components [18,19].
Recently, we isolated and characterized the Saccharomyces and non-Saccharomyces yeasts present in the grapes of a new implantation vineyard of Grignolino in Piedmont (Italy). Grignolino is a red variety cultivated in Piedmont in the northwest of Italy to produce three DOC (Denomination of Controlled Origin) wines. Some of the characterized strains showed good technological basic features such as fermentative vigor and low volatile acidity, as tested in laboratory fermentations [20].
In this work, the specific enological characteristics of two indigenous strains (S. paradoxus and S. cerevisiae) isolated in this vineyard were determined by comparing their wine fermentation performances in a pilot scale in real cellar conditions. Their dominance was ascertained at the end of alcoholic fermentation (AF) and their impact on the final wines was investigated by quantifying volatile compounds and by performing sensory analysis test.

Yeast Strains
Saccharomyces cerevisiae and Saccharomyces paradoxus strains were previously isolated in a Grignolino Vineyard in Nord Italy [20]. They are included in the Microbial Culture Collection of Oenological and Viticultural Environment (CREA-CMVE) of the Center with the references: S. cerevisiae (ISE1567) and S. paradoxus (ISE1618). A yeast pre-inoculum, previously grown in YEPG broth, was prepared in commercial grape juice (Bravo, Rauch, Austria) and then inoculated (5 × 10 6 cells/mL) in Grignolino must. Each strain was tested in triplicate.

Vinification
The trials were run during the 2018 vintage. Homogeneous samples of Grignolino grapes were harvested in crates. Grapes were divided into six homogeneous lots of 130 kg each, containing equally distributed grapes of each row of the vineyard. After destemming and crushing, 80 mg/kg of potassium metabisulfite was added to all six trials.
AF was carried out at a temperature of 24-25 • C. In the first four days of fermentation, a pump down was carried out in the morning and a pumping over of about 20% of the volume in the evening. From the fifth day onwards, two pump-overs in the air for half the volume were done twice a day. Racking off was carried out after 20 days of maceration. At the end of AF, approximately 70 L of wine was obtained in each fermentation. Wine fermented with S. paradoxus is termed SpW, and the one with S. cerevisiae ScW.

Yeast Dominance Analysis
At the end of fermentation, yeasts were isolated by dilution and spreading on WL (Wallerstein Laboratory) agar. After growth, 24 colonies from each sample were randomly collected from plates. Dominance analysis was performed by microsatellite multiplex PCR (MM-PCR) to distinguish S. cerevisiae strains [21].
MM-PCR data were managed using Bionumerics software (Applied Maths, Belgium). The band pattern profile obtained on the colonies isolated at the end of fermentation was compared with the profile of the inoculated strain.
Since microsatellite loci amplification is not possible in S. paradoxus, the species was assessed by amplifying the D1-D2 domain with primers NL1-NL4 [22] and sequencing. For this, the NS1⁄ITS2 primer pair was used to amplify the ITS1 region of the 18S rDNA; PCR products were digested with MspI, and separated by electrophoresis [8]. Gels were processed using Bionumerics software as above.

Chemical Analysis
Density, volatile acidity, titratable acidity, and ethanol content were analysed according to the methods of the OIV (International Organisation of Vine and Wine). Residual sugar and glycerol were quantified using an HPLC with a refractometric detector using the following conditions: Rezex RCM-Monosaccharide column (300 × 7.8 mm, 8 µm, Phenomenex, Torrance, CA, USA), water as eluent with a flow of 0.35 mL/min, column temperature 85 • C, and injection volume 20 µL. Organic acids were quantified by HPLC Agilent 1100 as described [23], and YAN (Yeast Assimilable Nitrogen) was determined by formol titration [24].

Sensory Analysis
The wine sensory descriptive analysis (sensory profile) was conducted by a trained panel (6 males and 7 females) following a methodology deriving from the ISO norms [25]similarly to other procedures [26,27]-using ISO (3591-1977) approved glasses in an ISO (8589-2007) tasting room.
In all the sensory sessions, 4 wines were served (50 mL) in a randomized order and identified with a three-digit code. All the wines were tasted in a preliminary tasting session to define the odor descriptors with the help of a predefined odor list [28]. The choice of descriptors was made on the identification frequencies. The second-level descriptors (fresh herbaceous, dry herbaceous, and balsamic/resinous) were chosen when their frequency of identification was higher than 39 (13 assessors × 6 wines/2), and the third level descriptors (e.g., rose, geranium flower, pepper, cloves, raspberry, cherry and jam/marmalade) when their frequency was higher than 19.5 (13 assessors × 6 wines/4). The taste and mouth-feel attributes evaluated were acidity, bitterness, astringency, body (structure) and taste-olfactory persistence. The chosen attributes were confirmed by presenting to the panel appropriate standards, and measured twice in the wines in two different tasting sessions. Qualitative and quantitative sensory analyses were performed by FIZZ (Biosystems, Couternon, France). The intensity of the wine sensory attributes measures was acquired in two repetitions using a non-structured scale (0-100).
The sensory profile of each wine obtained from the average of the two-tasting session of each of the 3 wines produced with the same yeast is represented with radar diagrams. The quantitative sensory results (sensory profiles) were processed with ANOVA and Tukey's test (p = 95%).

Free Volatile Compounds Analysis
The volatile compounds were extracted by solid-phase extraction (SPE) as follows: 30 mL wine was diluted threefold with water and 300 µL and added with 1-heptanol (51.43 mg/L) as internal standard; samples were loaded onto a 1 g C18-EC cartridge (Biotage AB, Uppsala, Sweden), extracted with 5 mL dichloromethane and concentrated to 100 µL under a weak nitrogen flow. After the addition of 1-pentanol and stirring, the Beverages 2021, 7, 30 4 of 10 aqueous phase was extracted with dichloromethane and concentrated to 100 µL as before. The GC-MS analysis was performed with an Agilent 7890 Series gas chromatograph with an Agilent 5975 N Mass Selective Detector. The chromatographic conditions were: Helium carrier gas with a flow of 1 mL/min; the sample (1 µL) was injected in splitless mode on a Zebron ZB-WAX column (60 m × 0.25 mm, 0.25 µm, Phenomenex, Torrance, CA, USA); the source and the transfer line were kept at 230 • C and the injector at 250 • C [29]. Data were acquired in TIC mode (Total Ion Current) and processed with the ChemStation software (Agilent Technologies, Santa Clara, CA, USA). The identifications were achieved by comparing the retention times with those of pure reference compounds (when available), or comparing the LRI (linear retention index) to those reported in the literature. All high purity standards were purchased from Sigma-Aldrich (Milan, Italy).

Enological Characteristics of S. paradoxus and S. cerevisiae Strains
To overcome any inhomogeneity during ripening of the grapes that could result in differences in acidity, alcohol, color or polyphenol content of wines, the initial grape mass was collected and randomized distributed in boxes. The musts had the following mean chemical parameters: Brix 23.44 • ; titrable acidity (expressed as tartaric acid) 6.4 g/L; pH 3.49; YAN 287.7 mg/L.
Microbiological and biomolecular analyses were performed to evaluate the dominance of the yeasts at the end of fermentation. In S. cerevisiae wines, microsatellite analysis showed that the inoculated S. cerevisiae strain dominated the fermentations (Figure 1). , x FOR PEER REVIEW 4 of 10 aqueous phase was extracted with dichloromethane and concentrated to 100 μL as before.
The GC-MS analysis was performed with an Agilent 7890 Series gas chromatograph with an Agilent 5975 N Mass Selective Detector. The chromatographic conditions were: Helium carrier gas with a flow of 1 mL/min; the sample (1 μL) was injected in splitless mode on a Zebron ZB-WAX column (60 m × 0.25 mm, 0.25 μm, Phenomenex, Torrance, CA, USA); the source and the transfer line were kept at 230 °C and the injector at 250 °C [29]. Data were acquired in TIC mode (Total Ion Current) and processed with the ChemStation software (Agilent Technologies, Santa Clara, CA, USA). The identifications were achieved by comparing the retention times with those of pure reference compounds (when available), or comparing the LRI (linear retention index) to those reported in the literature. All high purity standards were purchased from Sigma-Aldrich (Milan, Italy).

Enological Characteristics of S. paradoxus and S. cerevisiae Strains
To overcome any inhomogeneity during ripening of the grapes that could result in differences in acidity, alcohol, color or polyphenol content of wines, the initial grape mass was collected and randomized distributed in boxes. The musts had the following mean chemical parameters: Brix 23.44°; titrable acidity (expressed as tartaric acid) 6.4 g/L; pH 3.49; YAN 287.7 mg/L.
Microbiological and biomolecular analyses were performed to evaluate the dominance of the yeasts at the end of fermentation. In S. cerevisiae wines, microsatellite analysis showed that the inoculated S. cerevisiae strain dominated the fermentations (Figure 1). As expected, no amplification was achieved in S. paradoxus wine isolated colonies by MM-PCR because this method is S. cerevisiae species-specific; thus, ARDRA analysis was conducted, confirming that they belonged to the S. paradoxus species (Figure 2). As expected, no amplification was achieved in S. paradoxus wine isolated colonies by MM-PCR because this method is S. cerevisiae species-specific; thus, ARDRA analysis was conducted, confirming that they belonged to the S. paradoxus species (Figure 2). The progress of alcoholic fermentation (13 days) in the different trials was similar. At the end of the AF, significant differences (p ≤ 0.05) in glycerol (6.70 vs 8.62 g/L) and malic acid (3.27 vs 2.50 g/L) content between S. cerevisiae and S. paradoxus were recorded (Table  1). Results are the average of three independent fermentations; titratable acidity was expressed as tartaric acid. Volatile acidity was expressed as acetic acid. SpW: S. paradoxus wines; ScW: S. cerevisiae wines. Different letters "a" and "b" mean statistical differences at ANOVA and Tukey's test (p = 95%).
Remarkably, the S. paradoxus strain production of glycerol, a very important compound for wine quality that provides sweetness and fullness [30], was high, thereby confirming previous data indicating that S. paradoxus produces a higher amount of glycerol than S. cerevisiae [31]. The observed ability of S. paradoxus to degrade malic acid is an interesting property in fermentation of musts with a high acidity; several studies demonstrated that fermentations with this species lead to a degradation of malic acid [10,32,33]; therefore, our results are in agreement with these previous works. In particular, Bovo et al. [32] affirmed that S. paradoxus strains were able to degrade high amounts of malic acid in ripe grape must, i.e., high glucose and low malic acid concentration.
No significant differences (p ≤ 0.05) were found for the other parameters analyzed, even though the quantity of ethanol produced was slightly lower in the fermentations of S. paradoxus (x̄ = 12.73 g/L) than in those of S. cerevisiae (x̄ = 13.29 g/L). This result agrees with Orlic et al. [31] who reported that S. paradoxus always produced lower ethanol concentrations than S. cerevisiae.
It is important to note that this lower ethanol production was not accompanied by a higher volatile acidity. At the same time, as mentioned before, glycerol content shows significantly higher values in S. paradoxus than in S. cerevisiae fermentations. These data The progress of alcoholic fermentation (13 days) in the different trials was similar. At the end of the AF, significant differences (p ≤ 0.05) in glycerol (6.70 vs 8.62 g/L) and malic acid (3.27 vs 2.50 g/L) content between S. cerevisiae and S. paradoxus were recorded (Table 1). Results are the average of three independent fermentations; titratable acidity was expressed as tartaric acid. Volatile acidity was expressed as acetic acid. SpW: S. paradoxus wines; ScW: S. cerevisiae wines. Different letters "a" and "b" mean statistical differences at ANOVA and Tukey's test (p = 95%).
Remarkably, the S. paradoxus strain production of glycerol, a very important compound for wine quality that provides sweetness and fullness [30], was high, thereby confirming previous data indicating that S. paradoxus produces a higher amount of glycerol than S. cerevisiae [31]. The observed ability of S. paradoxus to degrade malic acid is an interesting property in fermentation of musts with a high acidity; several studies demonstrated that fermentations with this species lead to a degradation of malic acid [10,32,33]; therefore, our results are in agreement with these previous works. In particular, Bovo et al. [32] affirmed that S. paradoxus strains were able to degrade high amounts of malic acid in ripe grape must, i.e., high glucose and low malic acid concentration.
No significant differences (p ≤ 0.05) were found for the other parameters analyzed, even though the quantity of ethanol produced was slightly lower in the fermentations of S. paradoxus (x = 12.73 g/L) than in those of S. cerevisiae (x = 13.29 g/L). This result agrees with Orlic et al. [31] who reported that S. paradoxus always produced lower ethanol concentrations than S. cerevisiae.
It is important to note that this lower ethanol production was not accompanied by a higher volatile acidity. At the same time, as mentioned before, glycerol content shows significantly higher values in S. paradoxus than in S. cerevisiae fermentations. These data lead to hypothesize that in the S. paradoxus strain a carbon flow balance shifted towards glycerol Beverages 2021, 7, 30 6 of 10 to a greater extent than S. cerevisiae, thus showing interesting technological prospects for natural reduction of alcohol content in wines [32,33].

Free Volatile Compounds and Sensory Profile of the Wines
Grignolino variety is cultivated in Piedmont, northwest Italy, to produce DOC wines that are generally described as dry and slightly tannic, with a moderately bitter taste and a persistent aftertaste. In a previous study [34] carried out on a suitable number (n = 36) of commercial Grignolino wines, the olfactory descriptors were violet-rose, geranium, pepper, raspberry, straw-hay. These attributes were also present in this study, even though the odor complexity was higher, which could be due to various reasons (grape quality, evolution of grape growing and winemaking techniques, different yeasts).
Regarding the organoleptic characteristics of the under-study wines, the sensory profiles obtained with the two yeasts ( Figure 3) were different only for the odor attribute rose, which was significantly higher in the ScW wines (ANOVA and Tukey's test, p = 95%). Raspberry odor, cherry and dry herbaceous were also slightly higher in these wines, but all wines were very similar for taste and mouth-feel attributes.
Beverages 2021, 7, x FOR PEER REVIEW 6 of 10 lead to hypothesize that in the S. paradoxus strain a carbon flow balance shifted towards glycerol to a greater extent than S. cerevisiae, thus showing interesting technological prospects for natural reduction of alcohol content in wines [32,33].

Free Volatile Compounds and Sensory Profile of the Wines
Grignolino variety is cultivated in Piedmont, northwest Italy, to produce DOC wines that are generally described as dry and slightly tannic, with a moderately bitter taste and a persistent aftertaste. In a previous study [34] carried out on a suitable number (n = 36) of commercial Grignolino wines, the olfactory descriptors were violet-rose, geranium, pepper, raspberry, straw-hay. These attributes were also present in this study, even though the odor complexity was higher, which could be due to various reasons (grape quality, evolution of grape growing and winemaking techniques, different yeasts).
Regarding the organoleptic characteristics of the under-study wines, the sensory profiles obtained with the two yeasts ( Figure 3) were different only for the odor attribute rose, which was significantly higher in the ScW wines (ANOVA and Tukey's test, p = 95%). Raspberry odor, cherry and dry herbaceous were also slightly higher in these wines, but all wines were very similar for taste and mouth-feel attributes. Analysis of the aromatic compounds of wines at the end of the alcoholic fermentation ( Table 2) showed no statistical differences in alcohols and fermentation esters between the two yeasts, but statistical differences in the concentration of terpene compounds were recorded, which can explain the sensory difference for the descriptor "rose". Terpenes are characterized by floral, muscatel or fruity aromas, and their concentrations in grapes and wines depend on various factors, including cultivar, region, wine-making techniques and yeasts.  Analysis of the aromatic compounds of wines at the end of the alcoholic fermentation ( Table 2) showed no statistical differences in alcohols and fermentation esters between the two yeasts, but statistical differences in the concentration of terpene compounds were recorded, which can explain the sensory difference for the descriptor "rose". Terpenes are characterized by floral, muscatel or fruity aromas, and their concentrations in grapes and wines depend on various factors, including cultivar, region, wine-making techniques and yeasts.
In general, wines obtained in this study have good enological properties and distinctive characteristics; this is also shown by PCA analysis (Figure 4): Wines obtained by S. cerevisiae have higher malic acid, total aldehydes and ketones and ethanol; S. paradoxus wines are mainly characterized by a higher content of glycerol. In general, wines obtained in this study have good enological properties and distinctive characteristics; this is also shown by PCA analysis (Figure 4): Wines obtained by S. cerevisiae have higher malic acid, total aldehydes and ketones and ethanol; S. paradoxus wines are mainly characterized by a higher content of glycerol.

Conclusions
In conclusion, both indigenous S. cerevisiae and S. paradoxus strains isolated from vineyard possess oenological properties of interest for the wine industry. S. paradoxus is characterized by a high production of glycerol and the ability to degrade malic acid. This, together with a lower production of ethanol and a low volatile acidity, makes this S. paradoxus strain a very interesting starter from an oenological point of view, in particular for the production of low alcohol content wines. On the other hand, the strain of S. cerevisiae gives the wine a pleasant smell of rose, as highlighted in the sessions of sensory analysis. Grignolino is a neutral variety with a very low content of terpene compounds, and thus, in these kinds of varieties, the role of yeast in the production of aromatic compounds is even more important.
Further studies should be conducted to better investigate the use of S. paradoxus, in particular its use in mixed cultures with S. cerevisiae or other yeast species, or in sequential inoculation, in order to obtain specific results in terms of ethanol content and glycerol production.

Conclusions
In conclusion, both indigenous S. cerevisiae and S. paradoxus strains isolated from vineyard possess oenological properties of interest for the wine industry. S. paradoxus is characterized by a high production of glycerol and the ability to degrade malic acid. This, together with a lower production of ethanol and a low volatile acidity, makes this S. paradoxus strain a very interesting starter from an oenological point of view, in particular for the production of low alcohol content wines. On the other hand, the strain of S. cerevisiae gives the wine a pleasant smell of rose, as highlighted in the sessions of sensory analysis. Grignolino is a neutral variety with a very low content of terpene compounds, and thus, in these kinds of varieties, the role of yeast in the production of aromatic compounds is even more important.
Further studies should be conducted to better investigate the use of S. paradoxus, in particular its use in mixed cultures with S. cerevisiae or other yeast species, or in sequential inoculation, in order to obtain specific results in terms of ethanol content and glycerol production.