Wine Aroma Characterization of the Two Main Fermentation Yeast Species of the Apiculate Genus Hanseniaspora

: Hanseniaspora species are the main yeasts isolated from grapes and grape musts. Regarding genetic and phenotypical characterization, especially fermentative behavior, they can be classiﬁed in two technological clusters: the fruit group and the fermentation group. Among the species belonging to the last group, Hanseniaspora osmophila and Hanseniaspora vineae have been previously isolated in spontaneous fermentations of grape must. In this work, the oenological aptitudes of the two species of the fermentation group were compared with Saccharomyces cerevisiae and the main species of the fruit group, Hanseniaspora uvarum . Both H. osmophila and H. vineae conferred a positive aroma to ﬁnal wines and no sensory defects were detected. Wines fermented with H. vineae presented signiﬁcantly higher concentrations of 2-phenylethyl, tryptophol and tyrosol acetates, acetoin, mevalonolactone, and benzyl alcohol compared to H. osmophila . Sensorial analysis showed increased intensity of fruity and ﬂowery notes in wines viniﬁcated with H. vineae . In an evolutionary context, the detoxiﬁcation of alcohols through a highly acetylation capacity might explain an adaption to fermentative environments. It was concluded that, although H. vineae show close alcohol fermentation adaptations to H. osmophila , the increased activation of phenylpropanoid metabolic pathway is a particular characteristic of H. vineae within this important apiculate genus.


Introduction
The development of commercial wine yeast cultures in the last 50 years has caused most winemakers to use conventional fermentation technology based on Saccharomyces cerevisiae strains. However, in the last decade, some exceptional non-Saccharomyces strains commercially available for oenology have appeared. These strains have been developed as a reaction of the producers, searching for aroma complexity and flavor diversity as demanded by the consumers [1,2].
Among non-Saccharomyces yeast, Hanseniaspora are the main species isolated from mature grapes, as was reported in most wine regions worldwide [3,4]. However, strains of apiculate genus are not currently available for winemaking in an easy way. Traditionally, in oenological environments, the presence of apiculate yeasts such as Hanseniaspora genus has been considered undesirable, mainly because of the increased acetic acid production and their competition capacity for nutrients with S. cerevisiae which could cause stuck or sluggish fermentations [5]. Additionally, most Hanseniaspora species are sensitive to ethanol [6], although some strains belonging to this genus have been detected throughout the fermentation [7,8]. Due to the increased fermentation capacity of Saccharomyces cerevisiae and, therefore, the production of ethanol in a few hours, this species dominates the process spontaneously up to complete sugars depletion. In addition, some flavor compounds, such as higher alcohols (2-phenylethanol and tryptophol) and medium chain fatty acids highly produced by S. cerevisiae, are considered inhibitors for the growth of other yeasts [9,10]. However, some Hanseniaspora species present positive characteristics that confer pleasant aromas and sensory complexity to the final wines [11,12], which makes them valuable for the winemaking industry and deserve to be analyzed in depth.
Hanseniaspora genus is a heterogeneous group that present several new species identified in the last decades, today including about twenty species [13]. Not all of them have been found in winemaking environments, but on other fruits, plants, and fermented foods. The categorization performed by us regarding the genetic [12] and oenological characterization of this genus [14] evidenced the existence of two separate groupings. Attending especially to their fermentative behavior, Hanseniaspora species can be classified in two clear technological clusters: the fruit group and the fermentation group [14]. Interestingly, an evolutionary study performed by Steenwyk et al. [15] also resulted in two clades with a similar composition among the species belonging to the Hanseniaspora genus, depending on characteristics related with cell cycle regulation and DNA repair.
Hanseniaspora. uvarum is the main representative of the fruit group, frequently found on grapes and early fermentation stages of wine elaboration [4]. Selected H. uvarum strains might enhance tropical fruity and floral aromas due to its high volatile phenols production and high glucosidase activity compared with S. cerevisiae, but they also produce a polishlike odor probably related with the increased production of ethyl acetate and acetic acid [16]. In fact, other Hanseniaspora species are able to synthesize a high concentration of acetate esters, commonly ethyl acetate [17,18].
Among Hanseniaspora species belonging to the fermentation group, Hanseniaspora osmophila and Hanseniaspora vineae have been previously described in spontaneous fermentations of grape, but not so frequently on fruits. These yeasts are characterized by their increased capacity to ferment and resist higher concentrations of ethanol when compared to other apiculate species [14]. However, there are some differences between these two species that make them interesting from an oenological point of view, which have not carefully been compared yet.
H. vineae is well known by its ability for improving wine flavors. Genomic, transcriptomic, and metabolomic studies in H. vineae have enhanced our understanding of its value within the wine industry [7,12,14,19,20]. The main characteristic of the H. vineae exometabolome is its increased production of acetate esters, such as 2-phenylethanol acetate, and other benzenoids compared with S. cerevisiae, which is a desirable trait, because esters present a lower sensorial threshold than their respective alcohols [7,12]. The reason for this difference has been argued to be due to the higher copy number of proteins with alcohol acetyltransferases (AATase) domains [20]. Therefore, H. vineae is characterized by intense fruity and flowery aromas and reduced levels of volatile acidity in white wines [7,8,19].
H. osmophila has been less characterized for wine production, although its capacity to enhance pleasant aromas in red wines was reported [21,22]. Higher concentrations of acetate esters, with the exception of isoamyl acetate, were found in mixed cultures with S. cerevisiae [21,22]. However, due to the methods performed for yeast identification at that time, H. osmophila was frequently confused with H. vineae; therefore, these findings are not so clear now. Today, we have the complete genome sequenced of 32 strains of Hanseniaspora available in public databases, and metabolic analysis can be correlated with genomic and transcriptomic information.
In the present work, we use genomic data of two strains of H. vineae [23] and one of H. osmophila to correlate with the vinification phenotype exhibited in the production of a Chardonnay wine. Wines elaborated with H. uvarum and conventional S. cerevisiae inoculation were used as controls. The objective was to evaluate differences in wine fermentation and flavor performance between Hanseniaspora strains belonging to the two Fermentation 2021, 7, 162 3 of 12 species that are classified within the fermentation group, also considering the genetic basis that could explain their different behavior during the process.

Yeasts and Fermentation in Natural Grape Must
Chardonnay grape must, containing 300 mg N/L of assimilable nitrogen and 200 g/L of sugars at pH 3.5, was treated with 200 mg/L dimethyldicarbonate to prevent microorganism growth. Pre-cultures of pure H. vineae Hv025 and H. uvarum AWRI1280, both isolated from Uruguayan vineyards, H. osmophila AWRI3579 from Australian grapes, and commercial S. cerevisiae ALG804 (Oenobrands, Montpellier, France) were inoculated in Chardonnay grape must and incubated at 25 • C for 12 h in a rotary shaker at 150 rpm.
Then, 125-mL Erlenmeyer flasks containing 90 mL of the grape that were closed with cotton plugs used to simulate microaerobic conditions [24] were inoculated with 10 5 cells/mL. Static batch fermentations were conducted at 20 • C to simulate winemaking conditions. Fermentation kinetics was controlled as CO 2 liberated by daily weighting. Cell growth was followed under the microscope using a Neubauer chamber and plating on WLN agar medium (Oxoid, Hampshire, UK).
After 12 days of fermentation, samples were centrifuged at 3500 rpm for 10 min. Supernatants were used for aroma characterization and sensory analysis. Residual sugars were quantified by Near Infrared Spectroscopy using a FOSS WineScan FT 120 (HilleroedDenmark).

Aroma Characterization
Aroma compounds from wine samples (50 mL) were extracted by adsorption and separate elution from an Isolute ENV+ cartridge (International Sorbent Technology Ltd., Mid Glamorgan, UK) packed with 1 g of a highly cross-linked styrene-divinylbenzene (SDVB) polymer, as described previously [25]. Samples eluted were concentrated with N 2 and Gas Chromatography Wine aromas were identified by comparing their linear retention indices with pure standards (Sigma-Aldrich Inc., St. Louis, MO, USA). Additionally, mass spectral fragmentation patterns were compared with those stored in commercial and our own databases. Quantification by GC-MS was performed using 1-heptanol and 2-octanol 1:1 (w/w) as internal standards.

Genomic Analysis
Genomic data from the different yeast species were obtained from NCBI databases. H. uvarum AWRI13580; H. osmophila AWRI13579; H. vineae T02/19AF; S. cerevisiae S288c genome sequences were used for comparison purposes.
Protein domains predictions were carried out with Pfam protein families database [26]. Additionally, the analyzed predicted protein sequences were aligned and compared in a dendrogram that was constructed by Neighbor-Joining method using MEGA version 4 software [27].

Sensory Analysis
Sensory analysis was performed by a panel of three experts who were asked to report any defect in the wines and give an overall qualification from 0 to 10 for all the samples in triplicates, with 10 being the best and 0 the worst punctuation.
An initial approach to flavor description of tryptophol and tyrosol acetates was performed by an extended panel of six tasters. Pure standards of tyrosol acetate (Santa Cruz Biotechnology, Dallas, TX, USA) and tryptophol acetate (Angene International Ltd., Nanjing, China) were added to hydroalcoholic solutions containing 12% of ethanol to reach concentrations of 5 mg/L and 10 mg/L, respectively, considering our quantification Fermentation 2021, 7, 162 4 of 12 levels in the experimental wines. Normalized tasting glasses at 20 • C were used for all these analyses.

Statistical Analysis
All the fermentations were performed in triplicate and the statistical error for fermentation kinetics, sugar concentration, and aroma quantification were calculated as the standard deviation. To compare the concentration of volatiles quantified in each wine, variance comparison was performed by the ANOVA test carried out with STATISTICA 7.0 software. Mean rating and Tukey significant differences were calculated.

Fermentation and Aroma Profile of Wines Elaborated with Hanseniaspora Spp.
Chardonnay musts inoculated using the four selected strains were analyzed for sugar consumption after 12 days. S. cerevisiae ALG804 was able to complete the fermentation and less than 2 g/L of residual sugars was detected (Table 1). However, Hanseniaspora strains were not able to completely deplete the sugars. Table 1. Residual sugars in wines after 12 days of fermentation (g/L).

H. vineae 025
Residual sugars (g/L) 1.6 ± 0.6 115.6 ± 11.8 86.8 ± 6.2 69.1 ± 12.1 Volatile acidity (g/L) 0.45 ± 0.05 0.91 ± 0.12 0.44 ± 0.14 0.34 ± 0.02 Among Hanseniaspora species, H. vineae was the one that showed the fastest fermentation kinetics (Figure 1), similar to H. osmophila. Conversely, H. uvarum presented the highest concentration of residual sugars after 12 days.  These results are in agreement to the lower ethanol resistance presented by H. uvarum and H. osmophila compared with H. vineae [14], probably due to their reduced copy number of alcohol dehydrogenases genes (ADH) as shown in Figure 2C, where H. uvarum presents only four copies of ADH sequences, H. osmophila six copies, and H. vineae eight copies. This fact might also explain the higher acetic acid levels of H. uvarum compared to H. vineae, an important technological difference within this yeast genus. Table 2. Aroma compounds (μg/L) detected by GC-MS in wines produced by the selected S. cerevisiae and Hanseniaspora species. Linear retention index based on a series of n-hydrocarbons reported These results are in agreement to the lower ethanol resistance presented by H. uvarum and H. osmophila compared with H. vineae [14], probably due to their reduced copy number of alcohol dehydrogenases genes (ADH) as shown in Figure 2C, where H. uvarum presents only four copies of ADH sequences, H. osmophila six copies, and H. vineae eight copies. This mation in Figure 2B. Other authors have described that H. osmophila presented high pro-duction of 2-phenylethyl acetate (up to 15 mg/L) compared with S. cerevisiae. In fact, H. vineae and H. osmophila present a higher copy number of ARO8, ARO9, and ARO10 genes compared with H. uvarum and S. cerevisiae ( Figure 2C), which are involved in the biosynthesis of aromatic alcohols [18]. However, the enhanced production of these compounds was shown uniquely by H. vineae in our results. Further research should be done to confirm this characteristic, because in old culture collections, there were some strains of both Hanseniaspora species that might be confused according to the identification method that was available at that time.  Figure 2B shows that 2-phenylethyl acetate is produced in significantly higher amounts in H. vineae than in the other species of Hanseniaspora or Saccharomyces, presenting 10 to 20 times more concentration after 12 days of fermentation. However, a significant  Wine aroma was analyzed by GC-MS, showing differences between the three Hanseniaspora species and S. cerevisiae. The main compounds produced with significant differences were acetate esters, such as 2-phenylethyl acetate, tyrosol acetate, and tryptophol acetate ( Table 2). It is well-known that ethyl acetate is the main acetate produced in quantity by these species and with consistent concentration differences ranging between 15 and 60 mg/L [7, 11,19,21]. Reports showed that H. osmophila synthesizes the highest concentration compared to H. vineae and S. cerevisiae, but all produce moderate levels that could impact positively the aroma and fruity character of wines (threshold aroma value 12.3 mg/L). However, ethyl acetate was not quantified in this study due to the small sample volumes utilized and the analytical methodology performed avoiding distillation. Although in the case of H. vineae, it is well known that this species is able to acetylate high amount of higher alcohols to esters [20] compared with Saccharomyces, these results showed for the first time significant differences in tyrosol and tryptophol acetates formation in Figure 2B. Other authors have described that H. osmophila presented high production of 2-phenylethyl acetate (up to 15 mg/L) compared with S. cerevisiae. In fact, H. vineae and H. osmophila present a higher copy number of ARO8, ARO9, and ARO10 genes compared with H. uvarum and S. cerevisiae ( Figure 2C), which are involved in the biosynthesis of aromatic alcohols [18]. However, the enhanced production of these compounds was shown uniquely by H. vineae in our results. Further research should be done to confirm this characteristic, because in old culture collections, there were some strains of both Hanseniaspora species that might be confused according to the identification method that was available at that time. Figure 2B shows that 2-phenylethyl acetate is produced in significantly higher amounts in H. vineae than in the other species of Hanseniaspora or Saccharomyces, presenting 10 to 20 times more concentration after 12 days of fermentation. However, a significant higher production was also found for the acetate esters derived from tyrosol and tryptophol compared to the other species. These two esters were not detected in wines fermented with S. cerevisiae, and neither were reported in other studies in the literature. In contrast, in this study, ethyl 2-hydroxypropanoate (ethyl lactate) was produced at significantly higher levels by H. osmophila ( Table 2). The aroma for this compound is described as strawberry and raspberry, but its odor threshold is higher than the concentration detected in our results (60 mg/L) [28]. These acetates clearly increase when low nitrogen levels are used, and they are generally related to the composition of the must, mainly the amino acid composition and the nitrogen content. The addition of diammonium phosphate (DAP) during fermentation is known to decrease the synthesis of higher alcohols precursors [29,30]. The aroma of 2phenylethyl acetate is well known and described as having rose, honey or tobacco notes [7]; however, to our knowledge, there are no reports about the sensory descriptions or the odor thresholds for the tyrosol and tryptophol acetates.
Acetate esters present, in general, lower odor thresholds that their corresponding alcohols. Thus, the presence of increased acetates levels is a desirable characteristic in order to improve the aroma perception of wine flavor. In a first approach, our sensory panel of experts characterized the aroma of tyrosol and tryptophol acetates, utilizing 5 mg/L and 10 mg/L, respectively, using as concentration guide the values found in this work ( Table 2). Pure compounds were added to an hydroalcoholic solution of 12% by volume. Fruity and flowery primary descriptors were obtained by the six tasters, with a higher intensity impact for tyrosol acetate than tryptophol acetate. Further studies for determining the sensory threshold of these esters might be carried out, which according to our previous experience, should be higher than the value reported for 2-phenylethanol acetate (200 µg/L) but below the results obtained in this work [30].
Other distinctive characteristic is the production of benzyl alcohol, which was detected in wines fermented by H. vineae, but not in those obtained with the other species that were evaluated ( Table 2). Its aroma was described as being floral, rose or almond [31], but the odor threshold in wine is quite high (200 mg/L) [32].
3-hydroxy-2-butanone (acetoin) was highly produced by H. vineae and detected in H. uvarum, but was not produced by H. osmophila fermentation ( Table 2). Sensory descriptors for this compound are sour yogurt and sour milk [31]. Although its odor threshold is 30 mg/L [32], it is known that it might contribute to the palate and flavor complexity of the final wines at lower concentrations [33]. Another highly produced compound by H. vineae was the lactone mevalonolactone [(±)-β-hydroxy-β-methyl-δ-valerolactone, (±)-3-hydroxy-3-methyl δ-valerolactone], which is a precursor in cholesterol biosynthesis [34], being used in cosmetic applications as a skin antiaging conditioner and humectant that works at dermis and epidermis [35]. Interestingly, no data are found about the sensory potential effect of this volatile compound, but it is known that many other lactones contribute to wine aroma characteristics.
Nonetheless, the aroma profile exhibited by these yeast species was sensory analyzed by an expert panel of tasters. The overall punctuation was lower for wines inoculated with H. uvarum AWRI1280 that were described as possessing some defects as acetic acid and nail polish odor, results which agree to the increased volatile acidity values detected (Table 1). H. vineae and H. osmophila produced floral and fruit aromas, and no defects were detected (Figure 3).

Genetic Analysis of putative Acetyltransferases in Hanseniaspora spp.
To understand the significant increase of acetate esters produced by H. vineae by comparison to the other species evaluated, we made a genomic search of acetyltransferases in the four species of this work. In S. cerevisiae, ATF1 and ATF2 are genes codifying alcohol O-acetyl transferases that can mainly acetylate branched higher alcohols into their corresponding esters. Additionally, SLI1, which codifies in S. cerevisiae for N-acetyl transferase related to sphingolipid biosynthesis, presents AATase domains according to the Pfam database. Therefore, it is possible that SLI1 homologous as those found in H. vineae [20] could participate in the biosynthesis of acetate esters because of its high sequence homology [36].
Regarding the results obtained from the direct comparison of sequences with AA-Tase domains in Pfam, all the Hanseniaspora strains studied presented these domains throughout their genomes in different quantities. H. vineae presents four genes homologous to SLI1 of S. cerevisiae and one of ATF2, as previously described by Giorello et al. (2019) [20], and in the genome of H. osmophila, only two were annotated as SLI1 and one as ATF2, while H. uvarum has neither a SLI1 nor a ATF2 homologous sequence annotated in the databases (Table 3). Nevertheless, there are other two sequences with AATase domains in H. vineae, four in H. osmophila, and two in H. uvarum. The similarities found in the predicted amino acid sequences of these genes are shown in the dendrogram (Figure 4).

Genetic Analysis of Putative Acetyltransferases in Hanseniaspora Spp.
To understand the significant increase of acetate esters produced by H. vineae by comparison to the other species evaluated, we made a genomic search of acetyltransferases in the four species of this work. In S. cerevisiae, ATF1 and ATF2 are genes codifying alcohol O-acetyl transferases that can mainly acetylate branched higher alcohols into their corresponding esters. Additionally, SLI1, which codifies in S. cerevisiae for N-acetyl transferase related to sphingolipid biosynthesis, presents AATase domains according to the Pfam database. Therefore, it is possible that SLI1 homologous as those found in H. vineae [20] could participate in the biosynthesis of acetate esters because of its high sequence homology [36].
Regarding the results obtained from the direct comparison of sequences with AATase domains in Pfam, all the Hanseniaspora strains studied presented these domains throughout their genomes in different quantities. H. vineae presents four genes homologous to SLI1 of S. cerevisiae and one of ATF2, as previously described by Giorello et al. (2019) [20], and in the genome of H. osmophila, only two were annotated as SLI1 and one as ATF2, while H. uvarum has neither a SLI1 nor a ATF2 homologous sequence annotated in the databases (Table 3). Nevertheless, there are other two sequences with AATase domains in H. vineae, four in H. osmophila, and two in H. uvarum. The similarities found in the predicted amino acid sequences of these genes are shown in the dendrogram (Figure 4). Fermentation 2021, 7, x FOR PEER REVIEW 10 of 13 ATF1 and ATF2 synthesize the production of volatile esters in S. cerevisiae [37], although ATF1 is induced by high YAN level in musts [38]. In the case of H. osmophila and H. vineae, they present ATF2 but not ATF1, and H. uvarum does not present any of them. However, there are some conserved amino acids in the sequences of ATF2 homologous and one hypothetical protein of with AATase domains (GenBank accession number OEJ85955.1) that could reveal an acetyl transferase activity role in this species ( Figure 5). ATF1 and ATF2 synthesize the production of volatile esters in S. cerevisiae [37], although ATF1 is induced by high YAN level in musts [38]. In the case of H. osmophila and H. vineae, they present ATF2 but not ATF1, and H. uvarum does not present any of them. However, there are some conserved amino acids in the sequences of ATF2 homologous and one hypothetical protein of with AATase domains (GenBank accession number OEJ85955.1) that could reveal an acetyl transferase activity role in this species ( Figure 5). In S. cerevisiae, there are two essential regions for ATF genes. The first is the WRLICLP-motif, which is not strictly conserved throughout microorganisms [39], and the second is the H-X-X-X-D catalytic-motif, which was found in several fruit and plant AA-Tases. Catalytic residues His and Asp have been reported as crucial for AATase function as part of the active site in these enzymes [40,41].
Higher alcohols were reported as growth inhibitors [9], especially in non-Saccharo- In S. cerevisiae, there are two essential regions for ATF genes. The first is the WRLICLPmotif, which is not strictly conserved throughout microorganisms [39], and the second is the H-X-X-X-D catalytic-motif, which was found in several fruit and plant AATases. Catalytic residues His and Asp have been reported as crucial for AATase function as part of the active site in these enzymes [40,41].
Higher alcohols were reported as growth inhibitors [9], especially in non-Saccharomyces yeasts. One option to avoid this effect is the transformation of these metabolites into other metabolites that are more innocuous for the cell. Therefore, the detoxification role of AATases [42] could explain the better fermentative behavior of H. vineae and H. osmophila, and this could also be related to its presence throughout the fermentation process and its frequent absence in the fruit. It was recently proposed that some of these flavor compounds might explained the reason why some yeast behaves as more friendly in a fermentation niche, decreasing the inhibition effects of alcohols and acetic acid [2].

Conclusions
Hanseniaspora species are divided into two main technological groups according to their fermentation performance. This work showed that species from both groups are also differenced by their capacity to produce diverse flavor compounds. Within the fermentation group, the presence of different aroma profiles given in wines allowed to distinguish H. osmophila from H. vineae. Although H. osmophila and H. vineae were genetically very close, significant differences were shown by the flavor metabolites produced. Furthermore, H. vineae presented an increased acetylation capacity and intensely developed the phenylpropanoids metabolic pathway, also with increased copy numbers of ARO8, ARO9, and ARO10 genes for the synthesis of aromatic higher alcohols. H. vineae might be discriminated from the other three species evaluated in this study by the quantification of benzyl alcohol, mevalonolactone, acetoin, and the three aromatic acetates, which are highly produced by this species. Interestingly, the presence of a great increased gene copy number of AATase domains in H. vineae might explain its characteristic to synthetize aromatic alcohols acetates in high concentrations compared to the other yeast species. Saccharomyces and H. uvarum showed a moderate capacity to acetylate these higher alcohols. Further research is being carried out by our group in order to define a specific metabolic footprinting method to evaluate the contribution of H. vineae to commercial wines at the winery level. Funding: This research was funded by the Agencia Nacional de Investigación e Innovación (ANII), Application of the Hanseniaspora vineae Project ALI_2_2019_1_155314 with Lage y Cia-Lallemand, Uruguay.