Mannoprotein Content and Volatile Molecule Profiles of Trebbiano Wines Obtained by Saccharomyces cerevisiae and Saccharomyces bayanus Strains

The production of volatile compounds has become one of the major technological features for yeast selection. In fact, although the aromatic profile of the wine is the sum of varietal-, pre-, post-, and fermentative-aroma compound, yeasts affect the quality of the grape from maturation throughout fermentation, metabolizing sugars and other components into alcohols, esters, organic acids, and aldehydes. Among the new technological features, the production of mannoproteins has gained interest. From this perspective, the main aim of this work was to characterize 9 strains of Saccharomyces cerevisiae and 1 of Saccharomyces bayanus for their volatile profiles and the release of mannoproteins. The strains were inoculated in Trebbiano musts and incubated at 15 ◦C; at the end of fermentation the wines were evaluated by GC/MS/SPME for their volatile profiles and mannoprotein content by enzymatic assay. The strains were inoculated at level ranging between 4.9 and 6.3 log CFU/mL but only the strains L318 and 12233X6167 were able to reach values of 7.5 log CFU/mL. The aromatic profiles resulted in a strain-specific fingerprinting. According to the principal component analysis, the wines produced by the strains L288, L234, and L318 were characterized by the presence of propanoic acid, butanol, octanoic acid, and 3 methyl pentanol while the wine obtained by the strain 12233x35G2 was characterized by the presence of propanoic acid, butanol, octanoic acid and 3 methyl pentanol while the strain 12233x35G2 was characterized by the presence of decanoic acid ethyl ester, heptanoic acid ethyl ester, and acetic acid 2 phenetyl ester. Regarding mannoproteins, the highest concentration was achieved by strain12233x6167 (104 mg/L). The data allowed to select the strains endowed with the best fermentation performances in terms of aroma and mannoproteins release.


Introduction
Saccharomyces cerevisiae and the other strictly related species like Saccharomyces bayanus are considered the "wine yeasts" with the highest oenological potential and are commonly used in the wine making [1]. Although, they are not the numerically prevalent species on the raw material, their strong alcohol tolerance and fermentative aptitude represent a great ecological advantage allowing these species to take over quickly on the grape yeast microbiota [1]. In addition to their efficient conversion of fermentable sugars into ethanol and carbon dioxide during alcoholic fermentation, these species produce many secondary metabolites (alcohols, aldehydes, ketones, esters, carbonyl compounds, organic acids, and volatile sulfur molecules) which specifically contribute to the volatile

Yeast Growth and Fermentation Kinetics
In Table 1, the yeast cell load (log CFU/mL) evolution during the fermentation of Trebbiano must is reported. Among the tested strains, only S. cerevisiae L318 and hybrid S. cerevisiae × S. cerevisiae 12233x6167 were able to reach cell loads higher than 7 log CFU/mL within 4 days of incubation at 15 • C. Otherwise, the remaining strains reached values no higher than 6.5 log CFU/mL in the same time considered. However, within 11 days of incubation, most of the tested strains were able to reach values higher than 7 log CFU/mL and to maintain high cell load values until the end of the fermentation. After 26 days of incubation at 15 • C, the strains considered decreased their cell loads. Table 1. Evolution of yeast cell loads (CFU/mL) during alcoholic fermentation at 15 • C of Trebbiano musts in relation to the fermenting agent used.
The fermentation kinetic data indicate that the greatest reduction in weight (%) were recorded for the L288, L318, and 12233 strains ( Figure 1) which, at the end of fermentation, recorded the highest alcoholic degrees but not the lowest sugar residues (Table 2). values higher than 7 log CFU/mL and to maintain high cell load values until the end of the fermentation. After 26 days of incubation at 15 °C, the strains considered decreased their cell loads. Data are expressed as mean of three repetition. The coefficient of variability ranges between 5% and 10%.
The fermentation kinetic data indicate that the greatest reduction in weight (%) were recorded for the L288, L318, and 12233 strains ( Figure 1) which, at the end of fermentation, recorded the highest alcoholic degrees but not the lowest sugar residues (Table 2).

Total Alcoholometric Volumic Title and Residual Sugar
As reported in Table 2, the strains L12233, L288, and VELO 5+ originated wines having a TAV higher than 11 with residual sugar level ranging between 1 and 2 g/L. Also the strains 6075 and L318 originated wine having TAV higher than 11 but with residual sugars of 4 and 2 g/L, respectively. The strain L234, L404, and L951 presented the highest residual sugar levels.

Total Alcoholometric Volumic Title and Residual Sugar
As reported in Table 2, the strains L12233, L288, and VELO 5+ originated wines having a TAV higher than 11 with residual sugar level ranging between 1 and 2 g/L. Also the strains 6075 and L318 originated wine having TAV higher than 11 but with residual sugars of 4 and 2 g/L, respectively. The strain L234, L404, and L951 presented the highest residual sugar levels.
To highlight differences among the samples, a principal component analysis (PCA) was carried out taking into consideration all the molecules detected since the final wine volatile aroma profile is the result of the interaction of different molecules. PCA mapped wine samples in the space described by the first two principal component PC1 versus PC2 (Figures 3 and 4). PC1 accounted the 20.28% of the total variability, while the PC2 for the 20.48%. Figure 3 describes the projection of samples in the space and the clusterization of the samples. On the basis of PCA analysis, samples were clustered into 4 different groups and were divided along the PC1 and PC2. Cluster 1 obtained from S. cerevisiae L 288, L 234, and L 318 strains was separated from other groups both along PC1 and PC2 as well as the Cluster 2 formed by the wine volatile profiles from the S. cerevisiae 12233×35G2 hybrid. Cluster 3 grouped samples from S. cerevisiae 12233 and S. bayanus L 951 strains. Cluster 4 was formed by S. cerevisiae strains L 404, velo 5+, 12233x6167 and 6075 samples ( Figure 3). According to the detected data, wine samples showed both qualitative (number of detected compounds) and quantitative (ppm equivalent) differences in volatile molecule profiles.
To highlight differences among the samples, a principal component analysis (PCA) was carried out taking into consideration all the molecules detected since the final wine volatile aroma profile is the result of the interaction of different molecules. PCA mapped wine samples in the space described by the first two principal component PC1 versus PC2 (Figures 3 and 4). PC1 accounted the 20.28% of the total variability, while the PC2 for the 20.48%. Figure 3 describes the projection of samples in the space and the clusterization of the samples. On the basis of PCA analysis, samples were clustered into 4 different groups and were divided along the PC1 and PC2. Cluster 1 obtained from S. cerevisiae L 288, L 234, and L 318 strains was separated from other groups both along PC1 and PC2 as well as the Cluster 2 formed by the wine volatile profiles from the S. cerevisiae 12233×35G2 hybrid. Cluster 3 grouped samples from S. cerevisiae 12233 and S. bayanus L 951 strains. Cluster 4 was formed by S. cerevisiae strains L 404, velo 5+, 12233x6167 and 6075 samples ( Figure 3). The projection of the variables which affected the clusterization is reported in Figure 4. In particular, cluster 1 was affected by butanol, propanoic acid, 3 ethoxy propanol, 3 methyl butanol, nonanol; for the second group mainly ethyl or methyl esters like heptanoic acid ethyl ester, acetic acid 2 phenyl ethyl ester, decanoic acid ethyl ester, and benzenacetic acid ethyl ester affected the clusterization. The third cluster differed for a greater production of benzenacetaldehyde, octanol, hexanol and hexanoic acid ethyl ester; while 2-ethyl hexanol, octanoic acid, 4-methyl-3-penten-2-one contributed to the grouping of samples 6075, 12233x6167, velo 5+ and L404 in cluster 4.
In order to increase the significance of the variability among the samples, raw data concerning alcohols and esters were analyzed by PCA separately. In Figure 5, the projection of the samples, according to the used strains and the production of alcohols is reported. The PC1 and PC2 explained more than 50% of the variability among the samples. On the other side, the PC1 and PC2, considering the detection of esters, were able to explain more than 60% of variability among the obtained wines ( Figure 6). The projection of the variables which affected the clusterization is reported in Figure 4. In particular, cluster 1 was affected by butanol, propanoic acid, 3 ethoxy propanol, 3 methyl butanol, nonanol; for the second group mainly ethyl or methyl esters like heptanoic acid ethyl ester, acetic acid 2 phenyl ethyl ester, decanoic acid ethyl ester, and benzenacetic acid ethyl ester affected the clusterization. The third cluster differed for a greater production of benzenacetaldehyde, octanol, hexanol and hexanoic acid ethyl ester; while 2-ethyl hexanol, octanoic acid, 4-methyl-3-penten-2-one contributed to the grouping of samples 6075, 12233x6167, velo 5+ and L404 in cluster 4.
In order to increase the significance of the variability among the samples, raw data concerning alcohols and esters were analyzed by PCA separately. In Figure 5, the projection of the samples, according to the used strains and the production of alcohols is reported. The PC1 and PC2 explained more than 50% of the variability among the samples. On the other side, the PC1 and PC2, considering the detection of esters, were able to explain more than 60% of variability among the obtained wines ( Figure 6).      The concentrations of mannoproteins (MPs) detected by the enzymatic method at the end of fermentation in relation to the starter agent are reported in Table 3. The strains that allowed the highest concentrations of MPs at the end of the fermentation were both the S. cerevisiae hybrids 12233x6167, 12233x35G2 and the 12233, L951 strains (83-104 mg/L) On the other hand, Trebbiano wines obtained using S. cerevisiae 6075 and S. cerevisiae L234 as fermenting agent, resulted with the lowest amounts of mannoproteins respectively of 56 and 57 mg/L.

Discussion
Trebbiano is an Italian white wine, obtained from a neutral variety named Trebbiano Romagnolo, whose aroma is particularly affected by the yeasts used during the fermentation process rather than the presence of free monoterpenes [34]. For this, the selection of yeasts able to impart good volatile profiles, in addition to specific β-glucosidase activities, has become an important challenge and criterion for the strain choice to improve the final quality of this product. The strains employed in the present research fitted the primary selection criteria of yeast strain for alcoholic beverage, ending the fermentation and converting efficiently the grape sugars to alcohol. In fact, the total alcoholometric volumic title detected in the different micro-vinifications are in line with those The concentrations of mannoproteins (MPs) detected by the enzymatic method at the end of fermentation in relation to the starter agent are reported in Table 3. The strains that allowed the highest concentrations of MPs at the end of the fermentation were both the S. cerevisiae hybrids 12233x6167, 12233x35G2 and the 12233, L951 strains (83-104 mg/L) On the other hand, Trebbiano wines obtained using S. cerevisiae 6075 and S. cerevisiae L234 as fermenting agent, resulted with the lowest amounts of mannoproteins respectively of 56 and 57 mg/L.

Discussion
Trebbiano is an Italian white wine, obtained from a neutral variety named Trebbiano Romagnolo, whose aroma is particularly affected by the yeasts used during the fermentation process rather than the presence of free monoterpenes [34]. For this, the selection of yeasts able to impart good volatile profiles, in addition to specific β-glucosidase activities, has become an important challenge and criterion for the strain choice to improve the final quality of this product. The strains employed in the present research fitted the primary selection criteria of yeast strain for alcoholic beverage, ending the fermentation and converting efficiently the grape sugars to alcohol. In fact, the total alcoholometric volumic title detected in the different micro-vinifications are in line with those detected in commercial and similar products. Each strain employed was able to impart specific features in volatile molecule profiles to the wine obtained, generating a specific aromagramma which, in its turns, depends on several factors. The first ones are the molecules initially present in the must and later the yeasts employed as well as the adopted winemaking conditions. The choice of the yeast strain largely affects the array of the generated volatile substances (e.g. higher alcohols, acids, esters, carbonyls, and thiols). In our experimental conditions, no terpenic molecules were founds, leading to suppose that the S. cerevisiae strains employed are not endowed of β-glucosidase activity. On the other hand, this feature is more pronounced in non-Saccharomyces yeasts as reported by several Authors [35][36][37][38]. Regarding mannoproteins, their accumulation has been studied in recent years given several of these molecules' beneficial aspects in wine environments [39]. Thus, wine researchers have focused on finding yeast strains that have the ability to release large amounts of mannoproteins during winemaking. The experimental data have underlined that the strains 951 and 12233x6167 produced the highest amounts of mannoproteins while the strains velo 5+ and 6075 released the lowest amount. On the other hand, the quantity of mannoproteins released can vary extremely in relation to the strain and to the chemical-physical and compositional conditions of the system [33,40,41]. Their presence can also affect the release of volatile compounds, affecting the final perception of the wine [33]. In winemaking, the addition of commercial products rich in mannoproteins can be performed, but the challenge to find mannoprotein producing yeast can represent a great advantage also from an economical point of view. Generally, the use of mannoproteins and fine lees increased the levels of fruity esters such as ethyl hexanoate, methyl, and ethyl hexadecanoate probably due to the esterification of fatty acids released by yeasts during fermentation or autolysis. Particularly the strain 12233x35G2, although not producing the highest amount of mannoproteins, was characterized by the highest amount of ethyl hexadecanoate as for the strain 951. Also, Pérez-Través et al., [42] investigated the ability of natural hybrids of S. cerevisiae × S. krudriavzevii to release mannoproteins founding that the strains, at the adopted operating conditions, was able to produce higher quantity of mannoproteins with respect the only S. cerevisiae. Moreover, the author found that the genome interaction in hybrids generates a physiological environment that enhances the release of mannoproteins.

Strains
In this study, 9 strains of S. cerevisiae and 1 strain of S. bayanus, belonging to the collection of the Department of Agricultural and Food Sciences, University of Bologna (Bologna, Italy), were used.
All the S. cerevisiae strains were isolated by Prof. Carlo Zambonelli, University of Bologna, from grapes of the Emilia-Romagna region (Bologna region, Italy). All these strains were characterized by prof. Zambonelli for the fermentation ability, SO 2 resistance, alcohol tolerance, acetic acid production, SO 2 production, glycerol production. S. bayanus 951 was isolated by Prof. Carlo Zambonelli from Lambrusco grapes [3].
All strains were grown on YPD medium (1% w/v yeast extract, 2% peptone, and 2% glucose; all chemicals provided by Oxoid, Milan, Italy) and incubated at 25 • C for 48 h before using.

Micro-Vinifications
The fermentations were carried out in flasks containing 5L of Trebbiano must vintage 2012 and inoculated with the different yeast strains at levels between 5-6 Log CFU/mL. The pre-cultures (48 hours) were used to inoculate the pasteurized must and fermentations were carried out at 15 • C. the fermentation was monitored daily by measurement of weight loss. The fermentations were considered finished when no variation of weight was observed for four consecutive days. Afterwards, the wines were clarified with bentonite (1 g/L) by stirring them very slightly for about 30 min at room temperature, then separated by filtration. Three replicates for each trial were set.

Microbiological Analysis
Viable cell counts were evaluated at different sampling times by plate counting using YPD agar (Oxoid, Basingstoke, UK)) incubated at 25 • C for 48 h. Three repetitions for each sample were considered.

Ethanol and WineRreducing Sugars
The ethanol determination was performed according to the EC2000 official EU method. Samples were considered statistically different at p < 0.05. Sugars (glucose and fructose) were detected by HPLC using a Jasco HPLC system with a pump (PU980) (Jasco Inc, Tokyo, Japan) equipped with a refractive index detector (RI830) according to the method proposed by [43]. Peak qualitative and quantitative analysis were based on retention times (Rt) and spiking technique using the external standard method. Analysis were performed in triplicates.

Mannoproteins Analysis
Polysaccharides were isolated by means of a HCl−ethanol solution and subsequent centrifugations according to [28], then mannoproteins were separated based on the OENO 26/2004 resolution [44] method and quantified by an enzymatic assay (Megazyme, Astori, Italy).

Volatile Molecule Profile
The sample volatile molecule profiles were determined by GC-MS-solid phase micro extraction (SPME). For this, five-milliliters of samples were placed in 10-mL glass vials, to which 1 g NaCl and 10 µL of 4-methyl-2-pentanol (as internal standard) were added. The samples were then equilibrated for 10 min at 45 • C, after that a 85 µm SPME fiber-polyacrylate coated fiber (Supelco Inc., Bellefonte, PA, USA) was exposed to each sample for 40 min. The equilibration and absorption phases were carried out under stirring condition. GC-MS analyses were performed using a Agilent 6890 gas chromatograph (Agilent Technologies, Palo Alto, CA, USA) equipped with an electron impact mass selective detector Agilent 5970 (ionization voltage, 70 eV) (Agilent Technologies, Palo Alto, CA, USA). CPWax52 CB capillary column (1.2 µm df; 0.32 mm i.d.; 50 m length) was used (Chrompack, Middelburg, The Netherlands). The molecule identification was performed by the comparison of NIST pure standards spectra library (version 2005) with the mass spectra obtained. The amounts of each compound identified was determined using a calibration curves method. Pure reference compounds were used to built calibration curves and the amounts of each compounds were determined by the interpolation of the relative areas versus the internal standard area. For Figure 3 the following alcohols were considered: 2 methyl propanol, 3 methyl 1 butanol, butanol, nonanol, 3 methyl pentanol, 3 etoxi propanol, (Z)-3-hexen-1-ol, 2 ethyl hexanol, (3S)-3,7-Dimethyl-6-octen-1-ol, hexanol, heptanol, phenethyl alcohol, and octanol. Also the following esters were included: butanol 3 methyl acetate, ethyl acetate, butanoic acid ethyl ester, hydroxy butyl ethyl ester, hexanoic acid ethyl ester, heptanoic acid ethyl ester, octanoic acid ethyl ester, decanoic acid ethyl ester, benzenacetic acid ethyl ester, and acetic acid 2 phenyl ethyl ester.

Statistical Analysis
Alcoholometric, residual sugars and mannoprotein data were compared using one-way ANOVA, followed by Tukey test, while raw volatile data were analyzed with a principal component analysis (PCA) using STATISTICA software (version 8.0, Statsoft, TULSA, OK, USA).

Conclusions
The present research permitted to highlight the potential of some Saccharomyces strains to produce specific volatile molecule fingerprintings and to release mannoproteins during fermentation time. Although, the production of mannoproteins need to be further investigated also from a molecular point of view, the selection of proper strains can represent a tool to increase the final quality of Trebbiano wine, especially from an aromatic point of view.