Isolation and Characterization of Cryotolerant Yeasts from Fiano di Avellino Grapes Fermented at Low Temperatures

A fermentation of Fiano di Avellino grape must was carried out at 9°C with the aim of selecting cryotolerant yeast strains and testing their fermentative performances and volatile production following molecular characterization. A total of 20 yeast cultures were isolated at different fermentation stages. Based on molecular identification and characterization, Metschnikowia (M.) pulcherrima, Hanseniaspora (H.) uvarum, Staremerella (St.) bacillaris, Saccharomyces (S.) cerevisiae, S. kudriavzevii, and S. paradoxus were found to be the yeast species dominating the fermentation. S. paradoxus has been rarely isolated in vineyards and never in the cellar environment. Moreover, in this study, S. kudriavzevii is detected for the first time in vine-wine environments. Both S. kudriavzevii and S. paradoxus co-occurred with S. cerevisiae when grapes were micro-fermented at low temperatures. The growth kinetics of the three species were greatly affected by the fermentation temperature. As a consequence, Fiano wines obtained with S. kudriavzevii and S. paradoxus significantly differed from those made by S. cerevisiae in terms of chemical and volatile composition.


Introduction
Low-temperature wine fermentation enhances the freshness and fruity notes of the final product by preserving both the varietal and fermentative aromas to a greater extent [1,2]. Despite the quality improvement, low fermentation temperature has some negative implications: (1) longer fermentation time; (2) higher risks of oxidation; (3) major management and energetic costs; (4) increased persistence of non-Saccharomyces yeast. At low temperatures, the rate of ethanol production, as well as the growth of Saccharomyces (S.) cerevisiae, is lower, whereas the ethanol resistance of non-Saccharomyces is higher. As a consequence, the probability of stuck and sluggish fermentations increases. Therefore, the wine industry is clearly interested in using yeast strains with an enhanced capability to ferment at low temperatures. In the Saccharomyces sensu stricto complex, some cryotolerant strains of S. kudriavzevii, S. uvarum, or S. eubayanus were so far isolated and characterized. However, due to the greater ethanol sensitivity than S. cerevisiae, their application in the cold winemaking process is limited [3]. By contrast, in wine environments, cryotolerant natural hybrids (S. uvarum × S. cerevisiae, S. cerevisiae, and S. kudriavzevii) have been described [3,4]. As summarized by García-Ríos [5], these hybrids obtain their physiological capability from both parents. Hybrids might have inherited the ability to grow at high temperatures (30-37 • C) and the ethanol tolerance from S. cerevisiae as well as the ability to grow at low temperatures (10-16 • C) from the other parent (S. kudriavzevii, S. uvarum, and S. eubayanus). Artificial interspecific hybrids into the Saccharomyces genus were also proposed as a possible biotechnological solution for improving cryotolerance in wine yeasts [5].
Given these facts, the industry is driven to select yeast strains with improved lowtemperature fermenting abilities. Naturally, cold-tolerant strains of the Saccharomyces genus, such as S. kudriavzevii, S. uvarum, or S. eubayanus, may be employed for low-temperature fermentations. However, they might not be as efficient for alcoholic fermentation because they typically have higher ethanol sensitivity than S. cerevisiae. In this study, natural cryotolerant yeasts were isolated and characterized from Fiano white grapes, one of the most significant vine cultivars in the Irpinia wine district (Avellino Province, Campania Region, Italy), used to make the well-known Fiano di Avellino DOCG (Appellation of Controlled and Guaranteed Origin) wine. Grapes were harvested and crushed in sterile conditions and allowed to ferment in an old cellar at a temperature that was naturally set at 9-10 • C in order to focus the search on yeast strains that are well acclimated to low temperatures. The isolated yeast cultures were identified and characterized by DNA-based techniques and evaluated for features of technological interest. Selected cultures were employed for microfermentation trials.

Origin of Grape Samples and Fermentation Conditions
Fiano grapes were collected from a vineyard (N 40 • 58 16,59716 , E 14 • 44 28,34808 ) located in Sant'Angelo a Scala (Irpinia district) in the first decade of November 2021. Grapes (about 5 kg) were harvested using sterile gloves and plastic bags (previously weighed) and immediately transferred to the laboratory. Grapes were aseptically crushed in the collecting bags, weighed, and transferred in a sterile 10 L steel container. After the addition of potassium metabisulfite (100 mg/kg), ammonium phosphate (200 mg/kg), and thiamine (0.6 mg/kg) [6,7], the must batch was left to ferment in a historic cellar at 9-10 • C.

Fermentation Monitoring and Yeast Isolation
At the beginning (T0 day) and during fermentation (T7, T14, T21, and T28 days), the main oenochemical parameters ( • Babo, pH, total acidity, and alcoholic degree), as well as yeast populations, were monitored. Sugar content was determined by using Babo mustimeter (Klosterneuburg). pH meter Basic 20 (Crison) for pH and total acidity determinations. Total acidity was expressed as g/L of tartaric acid (mL of 0.25 N NaOH to neutralize 25 mL of must/wine × 0.75). The alcoholic degree was evaluated by using the Malligand Ebulliometer. All analyses were performed in triplicate.
Colonies showing different morphology and/or color were all selected independently by their number. Cultures were purified by repetitive streaking on WL-nutrient agar. Yeast cultures were preserved on WL-nutrient agar slants and freeze-dried on malt extract broth (Oxoid) containing 20% (w/v) of glycerol (Fluka). Before each test, strains were cultured twice in YPD (yeast extract 10 g/L, peptone 20 g/L, dextrose 20 g/L, Oxoid).

Yeast Molecular Identification and Typing
DNA was isolated by using the method reported by Aponte and Blaiotta [6]. Preliminary identification was achieved by ITS (Internal Transcribed Spacer) rDNA region analysis coupled to ITS-RFLP (ITS-Restriction Fragment Length Polymorphism) by using HaeIII as a restriction enzyme. The taxonomic affiliation of representative strains of each ITS-RFLP group was determined by ITS region sequencing. All isolates belonging to the genus Saccharomyces spp. were typed by Interdelta [8] and DAN4 [9] analyses. Moreover, selected Saccharomyces spp. strains were characterized by RFLP analysis or/and sequencing of nuclear genes CAT8, CYR1, GSY1, MET6, and OPY1, as described by Gonzalez et al. [4]. Specifically, for RFLP analyses, the following restriction endonucleases were used: CfoI for CAT8 gene; HaeIII for CYR1 and OPY1 genes; MspI for GSY1 gene; Hinf I for MET6 gene. DNA from commercial strain EC1118 (LALVIN) was also analyzed in this phase.
Growth kinetics at different temperatures (10, 14, 18, and 28 • C) were evaluated in YPD broth at pH 3.2. At 10 • C, the commercial strain EC1118 (LALVIN) was used as control. The growth was monitored by spectrophotometric determinations at 600 nm (OD600).

Microfermentation Trials
The standardized procedure proposed by Romano et al. [15] was followed for microfermentation experiments. Strains cultured twice in YPD medium were used to inoculate (about 6 Log CFU/mL) Fiano must (22 • Brix, total acidity 6.36-6.42 pH 3.17-3.20) sterilized by tyndallization (100 • C for 3 min, 3 times) in 250 mL Erlenmeyer flasks closed with a Müller valve filled with sulphuric acid. The must was supplemented with potassium metabisulfite (100 mg/kg) and ammonium phosphate (200 mg/kg) before inoculum. During incubation, at 9-10 • C, flasks were handle stirred for 30 s every 12 h. Weight loss due to CO 2 escaping from the system was quantified for the fermentation kinetics monitoring. Fermentation was considered concluded when no weight loss was recorded within 24 h. Fermentation vigor (FV) was expressed as grams of CO 2 produced in 100 mL of Fiano must during the first 5 days of fermentation, while fermentation power (FP) was expressed as the alcoholic degree (% v/v = g CO 2 produced/100 mL × 1.25) reached at the end of fermentation. Each trial was performed in triplicate. At the end of fermentation, total sugars (glucose plus fructose), acetic acid, total acidity (expressed in grams/L of tartaric acid), and total and free SO 2 were quantified by using a Dionysos 100 multiparametric analyzer (Sinatech, Fermo, Italy). For each parameter, a specific kit (Sinatech, Fermo, Italy) was used. Moreover, concentrations of glucose, fructose, glycerol, ethanol, as well as tartaric, malic, and succinic acids, were evaluated by HPLC (High-Performance Liquid Chromatography) as previously described [10].

VOCs Analysis
For VOCs extraction, 100 mL of wine was extracted by applying a liquid-liquid extraction method described by De Filippis et al. [16]. Each extraction was carried out in triplicate. For High-Resolution Gas-Chromatography/Mass Spectrometry (HRGC/MS) analysis, 2 µL of the organic extract was injected in splitless mode, while the injection port of a GC/MS-QP2010 quadrupole mass spectrometer (Shimadzu, Shimadzu Corp., Kyoto, Japan) was maintained at 250 • C. The GC/MS was equipped with a DB-WAX column (60 m, 0.25 mm i.d., 0.25 µm film thickness; J&W Scientific Inc., Folsom, CA 95360, USA). The carrier gas was helium (1.3 mL/min), and the temperature program used was 0 • C for 5 min, raised to 220 • C at a rate of 2 • C/min, and held for 20 min at the maximum temperature. Electron impact mass spectra were recorded with ion source energy of 70 eV, while the temperature was kept at 230 • C. The peak areas were measured using a GC/MS solution program Shimadzu version 2.30 (Shimadzu Corp., Kyoto, Japan). Compounds concentrations (semiquantitative) and identification were computed and performed as previously reported [17]. Concentrations of isolated compounds were expressed as a ratio of the response of each compound against the response of the internal standard.
In a few cases, the pure chemical standard was not available, and the identified compounds were labeled as tentative (t).

Statistical Analysis
Significant differences among data were tested by ANOVA (Tukey). Principal component analysis (PCA) was employed to test the relationship (Pearson correlations) between volatiles and wine samples. The significance level was p ≤ 0.05 throughout the analyses. Data elaboration was carried out using XLStat (Addinsoft Corp., Paris, France, version 2012.6.02), an add-in software package for Microsoft Excel (Excel 2013).

Fermentation Monitoring and Yeast Isolation
The main oenochemical parameters detected during the fermentation of Fiano grape must at 9-10 • C are reported in Table 1. Fermentation started with a natural yeast microflora of about 10 6 CFU/mL. Yeast loads reached the maximum (about 10 8 CFU/mL) after seven days and then slowly declined. After 28 days, sugars were completely metabolized (0 Babo). The wine showed an alcoholic degree of 13.6% v/v, a little lower than expected, a pH of 3.88, and a total acidity of 6.04 g/L. In order to characterize the dominant yeast microflora, colonies were collected at each fermentation phase from countable agar plates: two from must (F0), five at 7 days of fermentation (F7), four at 14 days (F14), four at 21 days (F21) and five at 28 days (F28).

Yeast Molecular Identification and Typing of Isolates
All isolates were identified and characterized by means of molecular tools. As Table 2 shows, only four different ITS profiles were identified: 380 bp (showed by two isolates), 480 bp (one isolate), 750 bp (two isolates), and 850 bp (two isolates). Based on ITS-RFLP analyses and ITS sequencing, the two isolates with ITS of 380 bp were identified as Metschnikowia (M.). pulcherrima, the two isolates with ITS of 750 bp as Hanseniaspora (H.) uvarum, and the isolate with ITS of 480 bp as Staremerella (St.) bacillaris (Tables 2 and 3). Isolates with ITS of 850 bp (Saccharomyces spp.) showed the same ITS-RFLP patterns with both CfoI and Hinf I restriction enzymes (Table 2). By contrast, with HaeIII, two isolates (F4-72 and F14-72) showed a profile with only three bands (500, 230, and 150 bp, respectively) ( Table 2). All Saccharomyces spp. isolates were analyzed by means of molecular markers Interdelta and DAN4. By Interdelta typing, only three different profiles were retrieved: pattern reported as A characterized 11 isolates, and pattern B and pattern C were exhibited by only two isolates: F7-72 and F14-62, and F21-62 and F28-52, respectively (Table 2). These four isolates did not produce any pattern when analyzed by DAN4 molecular marker (Table 2). Therefore, based on ITS-RFLP, Intedelta, and DAN4, Saccharomyces spp. strains could be gathered in three different biotypes: (i) F7-72 and F14-72; (ii) F21-62 and F28-52; (iii) all the remaining. One isolate of each group (F4-72, F21-62, and F14-62, respectively) was further analyzed by other molecular markers. By means of ITS sequencing, the three strains were identified as S. kudriavzevii, S. paradoxus, and S. cerevisiae, respectively (Tables 2 and 3). Moreover, to exclude/confirm their hybrid nature, RFLP analysis and sequencing of some nuclear genes were performed. Data were compared by those obtained by S. cerevisiae EC1118 or available in open-source databases.  Table 3). 4 no result. S. cerevisiae F14-62 showed an RFLP profile identical to that of EC1118 ( Table 2). The other two strains (F7-72 and F21-62) showed RFLP profiles similar to those reported for S. kudriavzevii and S. paradoxus by Gonzalez et al. [4], respectively. All patterns were straightforward, and no mixed profile could be seen, allowing interspecific hybrid strains to be ruled out. To clear up any confusion, all the genes of the strains S. kudriavzevii F7-72 and S. paradoxus F21-62 were sequenced. Sequencing results of nuclear genes CAT8, CYR1, MET6, GSY1, and OPY1 supported the identification of strains F7-72 and F21-62 as S. kudriavzevii and S. paradoxus, respectively (Table 3).

Technological Characterization of Saccharomyces spp. Yeast Strains
The Saccharomyces spp. strains S. kudriavzevii F7-72, S. cerevisiae F14-62, and S. paradoxus F21-62 were selected as the population's representatives. In YPD broth with a pH of 3.20, 100 mg/L of potassium meta-bisulfite, and 13% (v/v) ethanol, all strains could thrive. Moreover, on Biggy agar, all of them proved to be low H 2 S producers (Data not shown). On media containing 4-methylumbelliferyl-b-Dglucopyranoside, arbutin, esculin, or cellobiose, no strain displayed beta-glucosidase activity, and none of them expressed lipolytic or proteolytic activity (Data not shown). On the other hand, by comparing growth kinetics at various temperatures, particularly when higher than 14 • C, variations across strains were revealed. All strains displayed similar trends at 10 • C, but as the fermentation temperature was raised, S. cerevisiae's development was noticeably higher (Figure 1).

VOCs Analysis
Forty-three VOCs were identified by HRGC/MS: 15 esters, 11 alcohols, nine acids, and eight miscellaneous compounds (Table 5). According to the ANOVA (Tukey, p < 0.05), S. kudriavzevii wine samples were characterized by an overall higher VOC production. Wines produced with S. cerevisiae were the richest in 3-+ 2-methyl-1-butanol and isobutyl alcohol. They had a significantly higher level of isoamyl acetate, an ester that is the most powerful wine fruity aroma with a banana note. Different letters in the row of each compound refer to significant differences (Tukey, p ≤ 0.05); ns = not significant.
Moreover, S. cerevisiae wines were the samples showing the lowest levels of some volatile acids. Differently, S. kudriavzevii samples showed significantly higher amounts of some volatile acids (isobutyric, butyric, nonanoic, and decanoic acids) and important contributors to the fruity aroma of wine, such as ethyl butyrate, hexanoate, and decanoate. S. kudriavzevii wines were also the richest in beta-phenylethyl acetate and beta-phenyl ethanol. S. paradoxus showed an intermediate behavior between S. kudriavzevii and S. cerevisiae. Indeed, for very few VOCs, significant differences were found between the three wines. Amongst these significant differences, two compounds (i.e., 1-butanol and 3-methyl-1-pentanol) showed the highest concentrations in S. paradoxus samples, and three miscellaneous VOCs (2-pentanone, beta-citronellol, and benzothiazole) showed the lowest concentrations in S. paradoxus wines. These wines contain similar levels of the smoky 4-vinyl guaiacol to that of S. kudriavzevii and significantly higher compared to S. cerevisiae, which produced more methionol, while S. kudriavzevii more acetoin (Table 5).
A principal component analysis was carried out considering the fermentative replications for each wine (observations) and the volatile compounds (variables) (Figure 3). The first two components, accounting for 72.15% of the variance (50.45% and 21.70% on F1 and F2, respectively), clearly distinguish the different wines in three different regions of the biplot. The biplot mainly opposes samples produced by the fermentation with S. cerevisiae

Discussion
Grape musts are characterized by a complex microbial ecology, including filamentous fungi, yeasts, and bacteria. Some microbial species are found only in musts before the onset of fermentation, while other species, such as yeast, lactic, and acetic acid bacteria, may survive and/or grow during the winemaking process and, due to their different physiological characteristics, mightily affect the wine quality. In must, these microorganisms are subjected to a selective pressure exerted by several factors, including high sugar content, high acidity, nutrient availability, low oxygen tension, increasing ethanol concentrations, and the presence of specific inhibitors such as SO 2 , botriticin, and medium chain fatty acids [18]. Moreover, among abiotic factors, fermentation temperature plays a crucial role in the growth and fermentation performances of different yeast species [19]. Fermentation at low temperatures is becoming a trend for the enhancement of wine aroma 'freshness' [1,2,20]. In this study, the yeast population dynamics during cryo-fermentation (9 • C) of a Fiano grape must were evaluated. Following a starting phase that was dominated by M. pulcherrima and H. uvarum, a mixed population of Saccharomyces spp., comprising S. kudriavzevii, S. paradoxus, and S. cerevisiae, drove the fermentation. S. paradoxus is frequently found in association with oak trees but has been rarely isolated from fruits or fermentations [21,22], and it has only sporadically been associated with wine production [23,24]. S. kudriavzevii was originally isolated from decaying leaves in Japan [25] and had not been found in grape fermentations or any other type of fermentative process. However, hybrids S. cerevisiae x S. kudriavzevii were often isolated in winemaking and brewing environments [4,[26][27][28]. According to González et al. [27], it is unknown whether such hybrids are the result of events that happened in the production environment or nature. Despite not naturally occurring in wine settings, S. kudriavzevii can conduct wine fermentation and produce the proper amounts of ethanol and glycerol by consuming the entire sugar content [28]. For instance, Arroyo-Lopez et al. [29] questioned (a) why S. kudriavzevii is absent from wine fermentations and (b) where their hybrids originated if S. cerevisiae and S. kudiravzevii do not coexist in wine environments. So, in a wine-model environment, it was examined how the two species competed and likely produced hybrids. They demonstrated that S. kudriavzevii's lack of competitiveness, particularly at high temperatures (over 17 • C), may account for the absence of this species in wine fermentations [29]. As a result, the authors hypothesized that natural S. cerevisiae x S. kudriavzevii hybrids were more likely to have evolved in wild conditions than in industrial fermentations.
Findings from this study are the first to describe the presence of S. kudriavzevii in a wine environment. This species was detected, most likely because of the low fermentation temperature (9-10 • C). In fact, by comparing growth kinetics at various temperatures, it was demonstrated that only at very low temperatures (9 • C) did the strains of S. kudriavzevii, S. paradoxus, and S. cerevisiae exhibit comparable growth kinetics; at temperatures higher than 14 • C, S. cerevisiae displayed better performances. These findings support those of Arroyo-Lopez et al. [29], who found that fermentation temperature plays a key role in the competitive exclusion of S. kudriavzevii by both S. paradoxus and S. cerevisiae.
Microfermentation experiments performed at 9 • C revealed that the three strains have comparable fermentation capabilities and partially supported data previously reported by other authors [30,31] that S. kudriavzevii and S. paradoxus yield less ethanol and more glycerol compared to S. cerevisiae.
Wines produced with S. cerevisiae showed a significantly higher isoamyl acetate content and were richest in 3-+ 2-methyl-1-butanol and isobutyl alcohol that, together with ethanol and even ethyl acetate, could globally act as aroma suppressors [32]. The amount of several volatile acids in S. cerevisiae and S. kudriavzevii wines were also demonstrated to vary. Higher amounts of some volatile acids were found in S. kudriavzevii wine, which may contribute to the overall aroma of the wine and, at low concentrations, have the potential to enhance the perception of fruitiness in wine due to esters or, at concentrations higher than 100 mg/L, have the opposite effect by masking fruity and varietal aromas [33,34]. Moreover, S. kudriavzevii wines were also the richest in beta-phenylethyl acetate and beta-phenyl ethanol, both involved in the expression of floral aromas in wines [34]. S. paradoxus showed an intermediate behavior between S. kudriavzevii and S. cerevisiae. This intermediate behavior showed by S. paradoxus is in line with data recently reported by Costantini et al. [31]. In fact, comparing the fermentation performances of S. paradoxus and S. cerevisiae on Grignolino grapes, the authors observed that the two yeast strains did not show great differences in terms of free volatile compounds in the wines at the end of the fermentation. Additionally, in partial agreement with Majdak et al. [35], S. paradoxus wines had a higher concentration of several volatile esters than S. cerevisiae samples, including ethyl butyrate, ethyl hexanoate, and beta-phenethyl acetate.
Results reveal that even though no significant differences were discovered, S. kudriavzevii samples always displayed consistently greater levels for the three main chemical VOC classes linked to the fermentation (i.e., total esters, alcohols, and acids).
The trend to a higher production of total esters could be explained by the fact that S. kudriavzevii is defined as a cryotolerant yeast and, therefore, better adapted to lower temperatures [36]. Fermentations at 9 • C could have enhanced esters production. Other studies have already investigated the ability of S. kudriavzevii and hybrids of S. kudriavzevii x S. cerevisiae to produce pleasant aromas in wine. It has been reported that there are many variables that can influence this process, such as nitrogen availability [37], temperature [38], fast sugar consumption during fermentation [39], and enzymatic properties [40]. In accordance with these results, other authors have observed that S. kudriavzevii and its hybrids produce higher amounts of fusel alcohols if compared with other yeasts. According to research by Stribny et al. [37], S. kudriavzevii is more efficient than other yeasts at converting phenylalanine, which may be the cause of the enhanced generation of beta-phenyl ethanol. The fusel alcohols and acetate esters content could result from the different enzymatic properties of acetyltransferase 1 and/or acetyltransferase 2 enzymes, as reported by Stribny et al. [40].

Conclusions
These findings provide information that makes S. paradoxus and S. kudriavzevii interesting yeasts for enological usage. The former is because of cryotolerance, while the latter is due to the actual trend of consumers for low-alcohol wines, as well as for its ability to produce glycerol amounts close to its taste threshold (5.2 g/L in white wine) and likely impacting sweetness and body in the mouth, low volatile acidity, and the malic acid degradation, as noted by Costantini et al. [31] and here confirmed. In this regard, additional research is required, above all in relation to varietal enology. For instance, it is reported that S. kudriavzevii can incorporate polyfunctional mercaptans precursor inside the cells with the potential enhancement of tropical fruit aromas [41]. Due to the significant S. cerevisiae competition, techniques like sequential inoculum, differential inoculum, and controlled aeration should be used when using Saccharomyces non-cerevisiae yeasts as an alternative to non-Saccharomyces for ethanol reduction and unique and interesting aroma development [30].