Grape sampling and spontaneous alcoholic fermentation

Undamaged and healthy grapes of different varieties, Malvazija (syn. Malvasia), Merlot, Refošk (syn. Refosco), Chardonnay, Šipon (syn. Furmint), Zweigelt, Modra frankinja (syn. Blaufränkisch) and Kerner were aseptically collected on different sampling dates in September 2016 (Table 1) in sterile plastic bags in three replicates (approx. 2 kg per replicate) from the vine-growing regions of Primorska (3 varieties×3=9 samples), Posavje (2 varieties×3=6 samples) and Podravje (4 varieties×3=12 samples), and processed separately in the laboratory.

The must was treated with Redox Arom, a mixture of antioxidants: l-ascorbic acid 35%, K-metabisulphite 55% and purified gallotannins 10% (DAL CIN GILDO S.p.A., Concorezzo, Italy) to protect it from the action of oxygen in the must. The mixture was added into the bags in the amount of 0.2 g/kg of grapes. The grape juice was obtained after destemming, crushing and squeezing the grapes in a sterile inox container.

The must was poured into 1-litre sterile fermentors. Spontaneous alcoholic fermentations were performed at room temperature (21-23 °C), monitored by weighing the fermentors and calculating the amount of released CO₂.

Measurement of must parameters

Sugar concentration, total acidity and pH value were determined in the musts. The sugar concentration was measured by WineLab touch (CDR s.r.l., Ginestra Fiorentina, Florence, Italy). The total acidity and pH values were determined by the methods accredited in the laboratory (OIV-MA-AS313-01:R2015 (23) and OIV-MA-AS313-15:R2011 (24)).

Yeast sampling, isolation and identification

During the spontaneous alcoholic fermentation, must aliquots were aseptically sampled and expected countable dilutions were plated on Wallerstein Laboratory (WL) nutrient agar (Merck, Darmstadt, Germany) in two replicates, in order to determine colony counts and to morphologically distinguish between non-Saccharomyces and Saccharomyces yeasts. The first sampling was carried out on the fifth day of spontaneous alcoholic fermentation, in order to obtain a first count of the non-Saccharomyces yeast population. Further samplings followed the dynamics of fermentations, and were taken in the middle and at the end of the fermentations.

After incubation at 26 °C for 2-3 days, the colonies were counted, and grouped depending on their morphology (OIV-MA-AS4-01:R2010) (25). Selected representatives of different morphological groups of yeasts were purified on yeast-malt (YM) agar plates (yeast extract 3.0 g/L (Biolife, Milano, Italy), malt extract 3.0 g/L (Biolife), peptone 5.0 g/L (Biolife), glucose 30.0 g/L (Merck), agar 20.0 g/L (Biolife)), with 0.01% chloramfenicol (Merck) added. The strains were cryopreserved in 10% glycerol (Merck) at −80 °C.

Molecular identification of yeast species

The total DNA was isolated using the MasterPure™ Yeast DNA Purification Kit (Illumina, San Diego, CA, USA). The primers used for amplification of Iinternal transcribed spacer (ITS) regions and D1/D2 region of large subunit (LSU) rDNA were ITS1 (5` TCCGTAGGTGAACCTGCGG) and NL4 (5` GGTCCGTGTTTCAAGACGG) as described, respectively, by White et al. (26) and Kurtzman and Robnett (27). The final volume of the polymerase chain reaction (PCR) on the mixture was 50 µL containing 100 ng genomic DNA, 1×standard buffer Mg²⁺ free, 1.5 mM MgCl₂, 2 mM of each dNTP, 50 pM of each of a pair of primers and 1 U Taq DNA polymerase (Promega, Madison, WI, USA).

For amplification of ITS rDNA, the PCR conditions were as follows: an initial denaturing step of 5 min at 94 °C was followed by 35 cycles of 40 s at 94 °C, 40 s at 56 °C and 30 s at 72 °C and terminated with a final extension step of 7 min at 72 °C and cooling down to 4 °C. The amplicon was sequenced by the commercial sequencing facilities (Macrogen Inc., Amsterdam, The Netherlands). The sequences were aligned and trimmed using BioNumerics v. 7.6 (28). For the molecular identification of yeast isolates, BLAST tools against GenBank recordings (29) of the rDNA sequences of the reference/type strains were used. All different sequences of the isolates were deposited in the GenBank and their accession numbers are listed in Table S1.

Analysis of microsatellite repeats

To evaluate the genetic diversity of S. cerevisiae strains, their microsatellite regions were analysed according to Legras et al. (30). Two multiplex PCR reactions for amplification of loci C5, C3, C8, C11, SCYOR267c and YKL172w, ScAAT1, C4, SCAAT5, C6, YPL009c by using Qiagen Multiplex Master Mix (Qiagen, Hilden, Germany) were performed. The size of the fluorescently labelled PCR products was determined by capillary gel electrophoresis in a commercial laboratory (Macrogen, Seoul, South Korea). The number of repeats in the microsatellite loci was automatically determined by a multiple-locus variable-number tandem repeat analysis (MLVA) module of the BioNumerics v. 7.6 software (28). The results were then manually curated. A dendrogram of similarity was created using the Bay-Curtis algorithm and clustering unweighted pair group method with arithmetic mean analysis (UPGMA). All isolates were genotyped using eleven microsatellite loci.

Measurement of chemical parameters of wine

After the completion of alcoholic fermentation, 2 mL/L of 5-6% sulphuric acid (Agrolit, Litija, Slovenia) were added to each fermentor, and later, the wine was transferred into bottles and placed in cold storage for clarification. The principle chemical parameters of wine (alcohol, glycerol, acetic acid, total and free SO₂) were measured after one month of storage by WineLab Touch (CDR s.r.l.), where the measurements are based on enzymatic reactions for a single parameter according to the manufacturer´s instructions.

Mixed alcoholic fermentation using indigenous strains

In the following experiment, we tested the fermentation capacity and efficiency of three previously isolated indigenous strains of Saccharomyces cerevisiae (RM1), Hanseniaspora uvarum (RM2) and Starmerella bacillaris (RM3). They showed a good potential to form volatile compounds (thiols, ethyl acetate, acetaldehyde and higher alcohols) during spontaneous alcoholic fermentation of Moscato bianco must, from which they were isolated and identified (9). The fermentation performance of these three indigenous strains was tested in Sauvignon must, in the following experiments: experiment A (33% S. cerevisiae, 33% H. uvarum, 33% S. bacillaris), experiment B (20% S. cerevisiae, 40% H. uvarum, 40% S. bacillaris) and experiment C (80% S. cerevisiae, 10% H. uvarum, 10% S. bacillaris). The fermentations were performed in 1000 mL glass fermentors in two replicates at (17.5±0.5) °C. The must had the following parameters: sugar concentration 207 g/L, total acidity 6.2 g/L, yeast assimilable nitrogen concentration 208 mg/L, and pH=3.21. The alcoholic fermentations were monitored by measuring mass loss. Must aliquots for plating were aseptically taken at the beginning, in the middle, and at the end of fermentation, and then plated on WL nutrient agar (Merck KGaA), in order to morphologically distinguish between inoculated non-Saccharomyces and Saccharomyces yeasts.

Measurement of aromatic compounds in wine

At the end of the alcoholic fermentation, the chemical parameters and the content of the aromatic compounds were measured in the wines. The basic chemical parameters of wine were measured by WineLab Touch (CDR s.r.l.), where measurement is based on enzymatic reactions. The concentrations of volatile thiols (4-mercapto-4-methyl-pentan-2-one (4MMP), 3-mercaptohexan-1-ol (3MH) and 3-mercaptohexyl acetate (3MHA)) were measured two weeks after the completion of the fermentation on the GC-MS system. Hydroxy mercury benzoate (5 mL of a 2-mM solution) and butylated hydroxylanisole (0.5 mL of a 0.02-mM solution) were added to 50 mL of the wine sample. After mixing for 1 min, internal standards 4-methoxy-2-methyl-2-mercaptobutane (4M2M2MB), deuterated 3-mercaptohexan-1-ol (d3MH) and deuterated 3-mercaptohexyl acetate (d3MHA) were added, and the procedure continued according to the method described in Jenko et al. (31).

Ethyl acetate, acetaldehyde and higher alcohols were measured using the methods described by Bavčar et al. (32). The wine samples were diluted (1:4) with water (Milli-Q, Milipore, Billerica, MA, USA) to achieve a 1:3 ratio between the liquid and the headspace of a 20-mL solid space microextraction (SPME) vial. The samples were incubated at 40 °C for 1 h, and adsorbed to a polydimethylsiloxane/divinylbenzene (PDMS/DVB) fibre (Supelco, Bellefonte, PA, USA). The compounds were identified and quantified with a gas chromatograph (Agilent 7890A; Agilent Technologies, Palo Alto, CA, USA), equipped with an automatic multipurpose sampler MPS 2 (Gerstel, Mülheim and der Ruhr, Germany) and coupled with a mass spectrometer (Agilent 5975C; Agilent Technologies).

Statistical analysis

The results were tested for normality distribution by Shapiro-Wilk test and statistically analysed using ANOVA in Statgraphics® Centurion XVI software (33).

Chemical composition of the musts

During the harvest, grape samples of the representative grape varieties for each vine-growing region in Slovenia were collected and processed. As detailed in Table 1, the chemical parameters of the musts differed depending on grape variety and vine-growing region. The lowest concentration of total acids in musts from white varieties and the highest concentrations of sugars were found in musts from the Primorska region, which has a Mediterranean climate. However, generally the highest total acid concentrations and the lowest densities were characteristic of the Podravje region musts, which have a cooler climate. If we take a closer look at the must parameters, they are mainly linked to the varietal characteristics, e.g. Refosco in Primorska, with a higher value for total acidity, and Kerner, Chardonnay and Zweigelt, with lower values in regions Posavje and Podravje.

Fermentation kinetics of the musts

Using the collected grape samples, 27 spontaneous alcoholic fermentations were conducted. The amount of exhausted CO₂ was more intense in the fermentors with varieties from the Primorska vine-growing region, on average 12.49 g per 100 mL of must (Fig. 1a), which is also a consequence of higher initial sugar concentration in the must (Table 1). All spontaneously fermented wine samples from the Primorska vine-growing region had a content of reducing sugars below 42 g/L when the fermentation stopped. In the fermentors with the must from the Posavje vine-growing region (Fig. 1b), the average final mass of exhausted CO₂ was 9.6 g and in the fermentors from the Podravje vine-growing region 8.7 g per 100 mL of must (Fig. 1c). The eight samples (two fermentors containing Blaufränkisch (MF) variety from the Posavje vine-growing region, and three fermentors with Blaufränkisch (MF) variety, two fermentors with Furmint (Š) variety and one fermentor with Zweigelt (ZW) variety from the Podravje vine-growing region) did not reach the stationary phase after 45 days of spontaneous alcoholic fermentation, because the fermentation got stuck.

Fig. 1

Mass concentration of exhausted CO₂ in spontaneous alcoholic fermentation of: a) Refosco, Malvasia and Merlot (Primorska vine-growing region) (average values±S.D. are shown) b) Kerner (average values±S.D. are shown) and Blaufränkisch (one fermentation completed) (Posavje vine-growing region), c) Chardonnay (three fermentations completed), Zweigelt (two fermentations completed), Furmint (one fermentation completed), and Blaufränkisch (none fermentation completed) (Podravje vine-growing region) (average values±S.D. are shown for Chardonnay and Zweigelt)

The nutritional status of must is an important parameter that influences the alcoholic fermentation (10, 21). In our case, we noticed the poor kinetics of alcoholic fermentations and yeast efficiency, probably due to the poor nutritional status of the must together with a specific yeast population that developed in each fermentor and led to the stuck fermentation.

Abundance of yeasts and their identification

We isolated altogether 64 isolates for further identification: 46 Saccharomyces and 18 non-Saccharomyces yeast (Table S1) and determined their macro- and micromorphology. The count of yeasts was determined at different sampling points; however, none of the Saccharomyces and non-Saccharomyces yeasts were isolated from the third sampling (Table 2).

The percentage of similarity between the ITS and D1/D2 sequences of the species type strains and our isolates (representatives of each group) with the GenBank accession numbers is shown in Table S1. The identified non-Saccharomyces yeast species were Hanseniaspora uvarum (five strains), Pichia kudriavzevii (three strains), Saturnispora diversa (two strains), and Starmerella bacillaris (syn. Candida zemplinina) (eight strains). H. uvarum is one of the prevailing apiculate yeast species in wine grapes from warm vine-growing regions, and was isolated only from Merlot must at the beginning of alcoholic fermentation. Yanagida et al. (34) also found the species to be a typical constituent of the yeast microbiota under mild climatic conditions.

Non-Saccharomyces strains of oenological interest were isolated from the must of Furmint variety. Although the fermentation was stuck in two fermentors, we obtained three representatives of H. uvarum and three of P. kudriavzevii. Nemcová et al. (35) reported that in their research, P. kudriavzewii isolated from Blaufränkisch variety, was more associated with damaged than with intact grapes. The third most common non-Saccharomyces yeast was S. bacillaris and was isolated as a representative from Kerner must, three from Blaufränkisch must from Posavje and two from Podravje. We should stress that our intention was to get a diverse population of S. cerevisiae strains, and not to get an excess population of non-Saccharomyces isolates.

Chemical composition of wines

When the concentration of reducing sugars dropped below 45 g/L, we stopped spontaneous alcoholic fermentation by adding sulphur to the wine (varieties from Primorska on the 34th day, varieties from Posavje on the 48th day and varieties from Podravje on the 46th day). Later on, some chemical components in the wines were measured. Table 3 shows the differences among the wines of different grape varieties, as a result of Saccharomyces and non-Saccharomyces yeasts present in the fermenting must. Interestingly, for all wine samples, the volatile acid concentration was below 1 g/L.

Genetic diversity of S. cerevisiae populations

In order to investigate the genetic diversity of S. cerevisiae, polymorphisms in eleven microsatellite regions of 46 strains were determined, and are shown in the dendrogram of similarity (Fig. 2). The dendrogram clustered S. cerevisiae strains into 14 groups at a similarity level of 99%. As reported by Schuller et al. (12), many S. cerevisiae strains were homozygous, although in our case, many were heterozygous, indicating outcrossing between different populations.

Fig. 2

Dendrogram of the similarity of 46 S.cerevisiae strains isolated from different vine-growing regions: Primorska (blue square), Podravje (green square) and Posavje (yellow square) and varieties (MAL=Malvasia, MER=Merlot, R=Refosco, K=Kerner, MF=Blaufränkisch, CH=Chardonnay, ZW=Zweigelt), further analysed with 11 microsatellite loci (YKL172w, ScAAT1, C4, SCAAT5, C6, YPL009c, C5, C3, C8, C11 and SCYOR267c). The numbers below each microsatellite locus represent strain ploidy; homozygous strain has a single allele at a single microsatellite locus, and two different alleles present a heterozygous diploid

The numbers of Saccharomyces cerevisiae strains isolated from the musts were 10 (Posavje), 11 (Podravje) and 25 (Primorska). We did not isolate S. cerevisiae strains from the spontaneous alcoholic fermentations of Furmint and Blaufränkisch must.

Genetically identical strains in all loci, S. cerevisiae R3164 and S. cerevisiae MAL383 (group 9), were isolated from two different vineyards in the Primorska vine-growing region, and were heterozygous in loci ScAAT1, C4, C6, YPL009c, C5 and C8. Interestingly, the strains S. cerevisiae Mer183 and S. cerevisiae K1162, belonging to group 9, were isolated from different vine-growing regions, and were identical in all loci. Group 9 comprised eight strains of S. cerevisiae, of which six strains were isolated from the Primorska vine-growing region. Genetically the most closely related strains from one variety in this group were isolated from the Refosco must. Otherwise, the genetically diverse group of strains (group 7), isolated from the varieties of the Primorska, Posavje and Podravje vine-growing regions, were identical in two motifs (37/37 and 21/22) of microsatellite loci ScAAT1 and C4. Genetically similar strains of group 14, S. cerevisiae K3161, S. cerevisiae K2163 and S. cerevisiae K2162, were isolated from the Posavje vine-growing region, and coincided in all microsatellite loci (K3161 and K2163), except C3 (K2162). S. cerevisiae MAL384 and S. cerevisiae CH1213 (group 12) were isolated from different vine-growing regions, and differed in C4 and C3 loci only.

Strains from the Primorska vine-growing region had characteristic microsatellite loci YKL172w and C4 with seven repeats of the motif (7 out of 13 strains). Strains from the Podravje vine-growing region also had the characteristic microsatellite loci YKL172w and ScAAT5, with eight repeats of the motif (8 out of 12 strains).

In Slovenian yeast populations, from two to 22 different alleles were detected in a single microsatellite locus. The largest number of alleles, and thus the highest diversity of S. cerevisiae, were shown by microsatellite loci C5, YPL009c, C3 and ScAAT1. As reported by other authors (20, 36), microsatellite loci C3, C5 and ScAAT1 are most commonly used to describe the genotype of S. cerevisiae present in complex samples such as must.

Chemical composition of the wines produced by indigenous yeast strains

To observe the ability and efficiency of isolated indigenous yeasts to conduct wine fermentations as starter cultures, we continued our study with three previously isolated yeasts of Saccharomyces and non-Saccharomyces yeasts. Fig. 3 shows the mass fraction of exhausted CO₂ during alcoholic fermentation inoculated with different initial ratios of indigenous yeasts. As expected, the weakest fermentation kinetics in experiment B was with the lowest ratio of S. cerevisiae and vice versa, the most intensive in experiment C with the highest ratio of S. cerevisiae. Slightly less intensive was the fermentation kinetics in experiment A with an equal ratio of all three yeast species.

Fig. 3

Mass concentration of exhausted CO₂ during alcoholic fermentation inoculated with indigenous Saccharomyces cerevisiae, Hanseniaspora uvarum and Starmerella bacillaris strains at different initial ratios (in experiment A: 33:33:33, experiment B: 20:40:40 and experiment C: 80:10:10) in 1000-mL flasks (two replicates; average values±S.D. are shown)

Alcoholic fermentations with different combinations of indigenous yeast strains were stopped after 35 days, when the concentration of reducing sugars was (10.8±0.0) g/L in experiments A and B and (10.7±0.1) g/L in experiment C (Table 4). The alcohol volume fraction was significantly lower in the fermentors with the lowest initial ratio of S. cerevisiae (experiment B) ((11.4±0.0) %). The differences in glycerol concentration were not significant, while the volatile acidity was significantly higher in experiment B ((0.86±0.0) g/L) and significantly lower in experiment C ((0.57±0.0) g/L) than in experiment A ((0.65±0.0) g/L). The significantly higher concentrations of free SO₂ and total SO₂ were in experiment A ((9.0±0.4) mg/mL and (61.0±1.2) mg/mL), where the ratio of yeasts was equal.

The concentrations of volatile thiols (3MHA, 3MH and 4MMP) were determined in the wine samples 14 days after the alcoholic fermentation was completed (Table 4). The concentrations of all three volatile thiols exceeded the limits of sensory perception in wine. The significantly lowest concentration of 3MHA was measured in experiment B (169 ng/L). The combination of yeasts in experiment B released the lowest concentrations of 3MH (1344 ng/L), compared to the other two combinations (2005 and 2310 ng/L, respectively). Yeasts in experiment C released significantly higher concentrations of 3MHA (254 ng/L), with S. cerevisiae being predominant; therefore, individual yeast strains have different genetic and psychological abilities to release volatile thiols from their precursors and to convert 3MH to 3MHA (34). For the concentration of 4MMP, no significant difference was confirmed. However, in experiment B the highest concentration of 4MMP was measured (31 ng/L), and in experiment A the lowest (11 ng/L). Numerous studies have shown that the amount of released 4MMP in wine depends on the strain of wine yeasts used for alcoholic fermentation (37, 38).

The concentrations of ethyl acetate, acetaldehyde and higher alcohols in wines were measured 45 days after the completion of alcoholic fermentation (Table 4). At concentrations below 300 mg/L (8), higher alcohols gave the wines the desired complexity and improved their sensory quality, as it was noticed in all three experiments. The concentrations of ethyl acetate, 2-methylbutanol, 3-methylbutanol and 2-phenyl ethanol exceeded the sensory perception threshold in wine. Compounds 2-butanol and 2-propenyl alcohol in wines were below the limit of detection. Regardless of the composition of the yeast strains, the same amount of ethyl acetate was produced, although many studies have shown that more ethyl acetate is released in mixed alcoholic fermentations in which the ratio of non-Saccharomyces yeasts was higher (39). Various yeast strains affected the profile and concentration of the higher alcohols in the wine. The significantly higher concentrations of 2-methylpropanol, 2-methylbutanol, 3-methylbutanol and 2-phenylethanol were measured in experiment C, with the highest ratio of S. cervisiae. Moreira et al. (40) reported higher concentrations of 1-propanol, which is known for its harsh heavy odour, in mixed alcoholic fermentations with non-Saccharomyces and Saccharomyces yeasts. However, we measured concentrations of 1-propanol far below the sensory perception threshold.