Fermentation of natural sweet wines has always been a big challenge in winemaking, but the global changes in the climate influence the parameters of the raw grape juice, and as a consequence the quality of the dry or semi-sweet wines. The higher initial sugar concentration is one of the most important factors involved, which could be a considerable challenge for the yeasts. Moreover, not only the fermentation dynamics and ethanol concentration but also the whole metabolite-profile could be modified remarkably (1, 2). In making some special sweet wines like botrytized or straw wines, sugar content of the grapes may reach an extremely high level, multiplying these effects.
Using starter cultures for wine fermentation originally meant exclusively single-strain (Saccharomyces cerevisiae or occasionally Saccharomyces uvarum) products. During the last decade, the industrial exploitation of selected non-Saccharomyces yeasts in combination with Saccharomyces strains as oligo-starter cultures has become an emerging trend (3-5). One of the promising candidates is Starmerella bacillaris (syn. Candida zemplinina) (2).
S. bacillaris was originally described in Tokaj wine region in Hungary (6) as Candida zemplinina and renamed by Duarte et al. (7). This species is particularly associated with sweet, botrytized wine fermentations (8, 9), where it spontaneously co-exists with S. cerevisiae and S. uvarum, but it is frequently isolated from other wine-related sources (10, 11). From oenological aspect, this species has several valuable properties like high glycerol production, fructophilic character and outstanding osmotolerance, reviewed by Englezos et al. (12).
The major metabolite pattern, in terms of particular fermentation by-products, plays a significant role in forming the final wine quality, while a moderate nitrogen demand, osmotolerance and fructophilic behaviour could help to avoid sluggish or stuck fermentation. These properties are highly varying on interspecific level, although there could be a considerable difference on intraspecific level too. The conventional wine yeasts, S. cerevisiae and S. uvarum are well-characterized from this aspect (13, 14), but S. bacillaris along with other non-conventional yeasts have only recently been in the focus of wine research (e.g. 15,16). Going one step further, a relevant question is how an extreme sugar concentration can modify the above-described character of different wine yeasts.
The aim of this study is to compare and contrast the changes in some major metabolite products and nitrogen utilization induced by extremely high (320 g/L) initial sugar concentration with a normal, but still considerable (220 g/L), sugar level, of the two most important wine yeasts Saccharomyces cerevisiae and Saccharomyces uvarum, and Starmerella bacillaris. Although this last species has been thoroughly studied in recent years at moderate sugar concentrations, to our knowledge, only a few studies investigated high sugar concentrations (17-19). In our work, we performed a direct comparison among the three species. The metabolites and substrates analysed in this paper were chosen on the basis of their outstanding oenological importance and included the ethanol, glycerol, volatile acidity, l-malic acid, l-succinic acid, yeast assimilable nitrogen and proline. The emphasis was on the effects of extremely high sugar concentration on these non-volatile compounds, but the behaviour of the different yeast species was also compared.
MATERIALS AND METHODS
Strains of three yeast species used in this study are as follows: three Starmerella bacillaris strains: Y1667 (CBS9494 type strain) and Y1756 from the National Collection of Agricultural and Industrial Microorganisms (NCAIM), Budapest, Hungary, isolated from botrytized Tokaj grape, and strain MLO from the Department of Oenology, Szent István University (DO-SZIU) culture collection, Budapest, Hungary, isolated from wine, six Saccharomyces cerevisiae strains: UVAFERMPM and UVAFERM228 commercial wine yeasts, strains SC57, RA100 and SB12 from DO-SZIU, isolated from Tokaj Aszú wine, and strain S701 also from DO-SZIU, isolated from the wine from Somló wine region, and three Saccharomyces uvarum strains: CBS395T, type strain from NCAIM, SB42 from DO-SZIU, both isolated from Tokaj Aszú wine, and S103 from DO-SZIU, isolated from the wine from Somló wine region. The natural isolates were previously identified at the Department of Microbiology and Biotechnology, Szent István University (20).
The fermentation was carried out in chemically defined grape juice medium at 220 and 320 g/L of sugar. Based on Henschke and Jiranek (21), the medium included (per litre of distilled water) d-glucose 110 or 160 g, d-fructose 110 or 160 g, KOH 4 mL to adjust pH to 3.3, yeast carbon base 11.7 g, Tween 80 0.5 mL, l-tartaric acid 2 g and l-malic acid 3 g. To obtain the yeast assimilable nitrogen (YAN) level of 300 mg/L, the following mixture of single amino acids and diammonium phosphate was used (in mg/L in the final medium): arginine 657.0, asparagine 131.0, aspartic acid 263.0, glutamine 175.0, glutamic acid 438.0, histidine 8.8, isoleucine 20.5, leucine 67.0, methionine 44.0, phenylalanine 17.5, threonine 43.9, tryptophan 17.5, tyrosine 17.5, valine 87.7 and proline 438.5, which is not part of YAN but present in every grape juice, and diammonium phosphate 136.0. All chemicals were obtained from Sigma-Aldrich Chemie GmbH, Merck, Munich, Germany.
Yeast culture maintenance
Yeast cultures were maintained on YEPD (20 g/L glucose, 10 g/L peptone, 10 g/L yeast extract and 15 g/L agar-agar) agar slants and stored at 6 °C. For inoculum preparation, an inoculation loop of cells from fresh agar slants was transferred into 20 mL of YEPD broth in 100-mL flasks and incubated at 25 °C for 48 h without agitation.
Fermentations were carried out under semi-anaerobic conditions, in 200-mL flasks containing 180-mL aliquots of culture media, without shaking, in triplicate. The flasks were inoculated with 106 cell/mL with 3% of 48-hour-old yeast cultures grown in YEPD broth. The fermentation temperature was kept at 20 °C. The course of fermentation was monitored by measuring the total soluble solids content (Brix values) with an Atago RX-5000 CX digital refractometer (Atago Co. Ltd., Fukaya-shi, Japan) after filtration, by sampling on days 0, 2, 4, 7, 9, 14, 21 and 28. Biomass concentration was measured by Bürker chamber cell counting after methylene blue staining. After the monitoring period of 28 days, the samples for chemical analysis were membrane filtered (0.45 μm) and stored at -18 °C until analysis.
Chemical and statistical analysis
Ethanol concentrations were measured by distillation and determination of specific gravity of the distillate (22), volatile acidity (expressed as acetic acid) by steam distillation followed by titration (23) and reducing sugars by iodometric titration (24) using the official methods of the International Organisation of Vine and Wine. d-glucose/d-fructose ratio (K-FRUGL), glycerol (K-GCROL), l-succinic acid (K-SUCC), l-malic acid (K-LMAL) and YAN (K-LARGE and K-PANOPA) concentrations were measured with Megazyme enzymatic kits (Bray, Ireland). Proline concentrations were measured by 1H NMR technique on a Bruker AVANCE 400 spectrometer and 400’54 ASCEND magnet system (Bruker, Karlsruhe, Germany) (25).
Data were evaluated with analysis of variance one way (ANOVA) and multivariate (MANOVA) and discriminant analysis (DA), after checking the assumptions, using the statistical package IBM SPSS Statistics for Windows v. 23.0 (26).
RESULTS AND DISCUSSION
Sugar consumption dynamics
Sugar consumption curves of the different species are shown in Fig. 1. At the end of the monitored period of fermentation with the lower initial sugar concentration (220 g/L), only S. cerevisiae strains UVAFERMPM, SC57 and S701 completed the fermentation. After fermentation with S. cerevisiae strains RA100, UVAFERM228 and SB12, smaller amounts of sugar remained (13.6-22.2 g/L), while with all the S. uvarum and S. bacillaris strains, considerable amounts of sugar remained (45.1-75.3 g/L and 117.5-125.0 g/L respectively), which is in accordance with the earlier described behaviour of the given species (9).
The most striking observation is that the extreme sugar concentration (320 g/L) did not influence significantly the rate of sugar consumption by S. uvarum, and in particular by S. bacillaris strains. Only a shift can be noticed in the sugar consumption curves, according to the elevated initial sugar content of the juice, resulting in higher residual sugar concentrations (Fig. 1).
Effect of extreme sugar concentration on the major metabolites
Our focus was on the major by-products of the alcoholic fermentation present in the chemically defined grape juice at the end of the monitored period (28 days) (Table 1). This experiment confirmed the earlier described strong fructophilic nature of S. bacillaris (8, 9) (Table 1), and this characteristic was not influenced by the high sugar concentration, therefore, it was excluded from further analysis and the sum of the consumed glucose and fructose was used for the yields, if applicable (Table 2 and Table 3).
Values are mean of triplicate fermentations for each strain, bold values are species mean values of strain performance. According to the Games-Howell post hoc comparison test, the values in the same column are statistically different at p<0.05; different lower-case letters indicate difference at γ(sugar)=220 g/L, capital letters at γ(sugar)=320 g/L, and asterisks denotes the change induced by initial sugar concentration in the grape. LSD5% values are the least significant differences between the mean values of species: LSD5%S.c-other between S. cerevisiae and the other two species, and LSD5%S.u.-S.b. between S. uvarum and S. bacillaris. n.s.=not significant
Values are mean of triplicate fermentations for each strain, bold values are species mean values of strain performance. After the Games-Howell post hoc comparison, the mean values in the same column are statistically different at p<0.05; different lower-case letters indicate difference at γ(sugar)=220 g/L, capital letters at γ(sugar)=320 g/L, and asterisks denotes the change induced by initial sugar concentration in the grape. LSD5% values are the least significant differences between the mean values of species: LSD5%S.c-other between S. cerevisiae and the other two species, and LSD5%S.u.-S.b. between S. uvarum and S. bacillaris. n.s.=not significant
Values are mean of triplicate fermentations for each strain, bold values are species mean values of strain performance. After the Games-Howell post hoc comparison, the mean values in the same column are statistically different at p<0.05; different lower-case letters indicate difference at γ(sugar)=220 g/L, capital letters at γ(sugar)=320 g/L, and asterisks denotes the change induced by initial sugar concentration in the grape. LSD5% values are the least significant differences between the mean values of species: LSD5%S.c-other between S. cerevisiae and the other two species, and LSD5%S.u.-S.b. between S. uvarum and S. bacillaris. n.s.=not significant. YAN=yeast assimilable nitrogen
S. bacillaris strains used the highest concentration of sugars to produce 1% ethanol, the Saccharomyces strains used considerably less (Table 2). Evaluating the change, only S. cerevisiae S701 was able to produce slightly increased alcohol yield from the excess sugar. The sugar-induced change in the ethanol yield was only significant when using S. bacillaris Y1667 and MLO (reduction from 0.34 to 0.27-0.30 g/g), which implies that these strains changed their main metabolite ratio due to the higher osmotic pressure.
The glycerol yield of S. bacillaris strains was considerably higher than of the investigated Saccharomyces species at both sugar levels (Table 2). The strong positive effect of the increasing sugar concentration was confirmed previously for the S. cerevisiae and S. uvarum strains and was verified also for S. bacillaris in this study, in accordance with the findings of Rantsiou et al. (18). The high glycerol production of the S. bacilllaris strains is highly valuable and it could shape considerably the body of the wine regardless of the sugar level.
The glycerol/ethanol ratio also showed an alteration in the main balance of the alcoholic fermentation, which was 2 to 4 times higher when using S. bacillaris than Saccharomyces species and clearly distinct at both sugar levels. Consequently, reduced alcohol content and a fuller wine texture could be reached employing S. bacillaris strains, preferably in combination with Saccharomyces yeasts due to their limited fermentation ability.
It has long been known that the increasing osmotic stress increases the acetic acid production of Saccharomyces species and we found similar trend for S. bacillaris. S. uvarum strains SB42 and S103 had the smallest increase, while CBS395 was similar to S. bacillaris strains Y1667, Y1756 and MLO, followed by S. cerevisiae strains UVAFERMPM, SC57, RA100 and UVAFERM228 (Table 1). In contrast to these results, Rantsiou et al. (18) found a sugar-independent trend in the acetic acid production by S. bacillaris that was mainly time-dependent. The absolute amount of acetic acid is comparable with that produced by Saccharomyces species, which is in accordance with some earlier results (9). Considering the limited fermentation ability of S. bacillaris, the acetic acid yield is not problematic since this species is not used in pure culture but in mixed fermentation (15, 16).
All species consumed l-malic acid and produced l-succinic acid, at both sugar concentrations. The effect of sugar concentration on the consumption was significant in the case of S. uvarum CBS395 and SB42, and S. bacillaris Y1756 and MLO, resulting in a reduction of the consumed malic acid. The negative initial sugar dependence of l-malic acid utilisation is in accordance with an earlier study (18). The l-succinic acid production was generally low in oenological terms (usual range 0.5-2.0 g/L (27)), but the yields were higher in the case of S. bacillaris than that of the other strains at both sugar concentrations (Table 2). l-succinic acid production by S. bacillaris Y1667 and MLO, S. cerevisiae UVAFERMPM and S. uvarum CBS395 seems to be influenced negatively by the initial sugar content. In certain vintages, the capability of a positive net organic acid production of a yeast is highly appreciated.
The nitrogen demand of a certain species is highly important in winemaking, and it is even more highlighted in the case of non-Saccharomyces species in mixed fermentation, in terms of competition, balance or synergism (4, 28, 29). Moreover, in grape juices with high sugar concentration, the appropriate amount of YAN could be crucial to complete the fermentation (21).
The composition of the YAN mimicked the grape juice composition. YAN concentration was adjusted to 300 mg/L, well above the necessary level (150-200 mg/L; (21)), but still below the optimal concentrations (400-500 mg/L).
The YAN utilization by the strains was varying widely, between 30-63%. At the higher sugar concentration, there was a significant decrease in YAN consumption by S. cerevisiae SC57 and RA100, S. uvarum CBS395 and S103 and all three S. bacillaris strains. Although the YAN consumption is reported to increase at higher sugar concentrations, in our investigations we used extremely high sugar concentration, where the nitrogen metabolism might be repressed to some extent (Table 3). The specific YAN consumption data show (Table 3) that the sugar-induced change in its consumption by S. bacillaris strains (0.43-0.81 mg/g) was similar to S. uvarum CBS395 (0.59 mg/g) and S103 (0.76 mg/g). The absolute YAN consumption by S. bacillaris was comparable with that of Saccharomyces species at lower sugar concentration, which corresponds to an earlier experiment (30). At high sugar concentration, S. bacillaris strains showed a moderate demand for nitrogen. This behaviour could be useful in mixed fermentations in juices where the initial sugar concentration is high or even extremely high.
Proline uptake was practically undetected at 220 g/L sugar, but it was taken up poorly at the higher sugar concentration (Table 3). This trend was prevalent for all the species, but moderate and strongly strain-dependent in the Saccharomyces species, involving S. cerevisiae UVAFERMPM, UVAFERM228 and S. uvarum CBS395. However, there was a more considerable increase in the proline utilization at the higher sugar level in the case of all strains of S. bacillaris. Proline, which would be an abundant source of nitrogen in every grape juice, is not or is negligibly utilized by wine yeasts due to an inhibition under anaerobic conditions (29, 30). Our results are consistent with this view at the normal sugar concentration, but an improved proline utilization might function at extreme sugar level, particularly in S. bacillaris.
Multivariate performance assessment
The sugar-induced changes in the above-described metabolites and nitrogen utilization of the investigated yeasts should be taken into consideration in one model. Fig. S1 shows the results of discriminant function analysis, with three significant functions. The strains were clearly distinct at different sugar concentrations.
Function 1, ethanol, did not group with either factor. Function 2 represented the factors that were sugar-dependent in a positive manner, namely volatile acidity and proline utilization, while Function 3 comprised factors negatively influenced by the sugar level, namely consumption of YAN and l-malic acid, and production of l-succinic acid. The performance of S. uvarum strains was overall considerably closer to that of S. cerevisiae than of S. bacillaris strains. The S. cerevisiae strains formed the most scattered group, which indicates their high intraspecies diversity, in terms of their altered metabolite production due to extremely high sugar concentration.
The high sugar-driven changes in the major fermentation metabolite pattern and nitrogen utilization pointed out an altered behaviour of the investigated species. The overall performance of Saccharomyces uvarum strains was close to S. cerevisiae in general; however, with clear differences in certain properties, e.g. volatile acid and glycerol production, while Starmerella bacillaris strains exhibited distinct, sometimes inverse behaviour than the Saccharomyces species. The effects of the extremely high sugar concentration on the oenologically beneficial traits were the most pronounced in S. bacillaris.
Regarding the nitrogen-related features of the S. bacillaris strains, namely the moderate yeast assimilable nitrogen demand and the noticeable ability to utilize proline at higher sugar level, this species could be a promising match in a mixed fermentation with S. cerevisiae, especially for grape juices with high or even extremely high sugar concentration.
In this experiment, the adequacy of this non-Saccharomyces species for winemaking, in combination with a suitable Saccharomyces yeast, is highlighted. Moreover, the results suggest that S. bacillaris possesses a more active adaptation mechanism to extreme sugar concentration. Undoubtedly, a more detailed investigation would be useful in the future focusing on the minor and volatile metabolites with an even wider strain set.