Yeast biodiversity and dynamics during sweet wine production as determined by molecular methods

Authors


  • Editor: Graham Fleet

Correspondence: Luca Cocolin, Dipartimento di Valorizzazione e Protezione delle Risorse Agroforestali, Università degli studi di Torino, via Leonardo da Vinci 44, 10095 Grugliasco, Turin, Italy. Tel.: +39 011 670 8553; fax: +39 011 670 8549; e-mail: lucasimone.cocolin@unito.it

Abstract

In this study we investigated yeast biodiversity and dynamics during the production of a sweet wine obtained from dried grapes. Two wineries were selected in the Collio region and grapes, grape juices and wines during fermentations were analyzed by culture-dependent methods (plating on WLN medium) and culture-independent methods (PCR-DGGE). Moreover, the capability of the Saccharomyces cerevisiae starter cultures to take over the fermentation was assessed by RAPD-PCR. On WLN agar several species of non-Saccharomyces yeasts (Hanseniaspora, Metschnikowia, Pichia, Candida, Torulaspora and Debaryomyces), but also strains of S. cerevisiae, were isolated. After inoculation of the starter cultures, only colonies typical of S. cerevisiae were observed. Using PCR-DGGE, the great biodiversity of moulds on the grapes was underlined, both at the DNA and RNA level, while the yeast contribution started to become important only in the musts. Here, bands belonging to species of Candida zemplinina and Hanseniaspora uvarum were visible. Lastly, when the S. cerevisiae isolates were compared by RAPD-PCR, it was determined that only in one of the fermentations followed, the inoculated strain conducted the alcoholic fermentation. In the second fermentation, the starter culture was not able to promptly implant and other populations of S. cerevisiae could be isolated, most likely contributing to the final characteristics of the sweet wine produced.

Introduction

The composition of the yeast microbial communities within grapes contributes significantly to the sensory characteristics of wine produced, as different genera, species and strains, and their metabolic activity, influence the sensory quality and organoleptic characteristics. Many studies have found that species of Kloeckera, Metschnikowia, Candida, Hanseniaspora, Rhodotorula, Pichia, Schizosaccaromyces, Kluyveromyces, Hansenula, Saccharomyces, Zygosaccaromyces and Debaryomyces are predominant in the initial stage of spontaneous fermentation (Fleet, 2003). Non-Saccharomyces yeasts generally grow during the first stage of fermentation; when the ethanol concentration increases, the more ethanol-tolerant Saccharomyces species complete the fermentation (Fleet, 1998). The growth of indigenous yeasts in must can significantly impact the efficacy of the inoculated Saccharomyces and even inhibit normal Saccharomyces growth and function (Bisson, 1999). However, the presence of non-Saccharomyces yeast populations can contribute to the final taste and flavour of wines.

Rapid information regarding the composition and dynamics of yeast flora occurring throughout the vinification process will help to control fermentation and consequently wine quality. Both traditional (Fleet & Heard, 1993; Fleet, 1998; Kurtzman & Fell, 1998) and molecular methods (Lieckfeldt et al., 1993; Schütz & Gafner, 1994; Baleiras Couto et al., 1995, 1996; de Barros Lopez et al., 1996; Smole-Monzina et al., 1997; Guillamón et al., 1998; Esteve-Zarzoso et al., 1999; Férnandez et al., 1999; Granchi et al., 1999; Cocolin et al., 2000, 2001; Caruso et al., 2002) have been used to study the dynamics of indigenous yeast populations during wine fermentations, allowing for a better understanding of the relations and interactions between different species of yeasts involved in the fermentation process. Moreover, a recent study has employed denaturing gradient gel electrophoresis (DGGE) to create an inventory and to monitor the microbial consortia of wines (Renouf et al., 2007).

Picolit is an ancient sweet wine typical of the north-east part of Italy, Friuli Venezia Giulia region, already established during Roman times. Picolit grapes are characterized by the development of just a few berries in each cluster because of a defect in pollination, and the resulting juice is thus very rich in sugars. Moreover, for production of Picolit wine, the grapes are dried either on the plants or in controlled chambers, allowing for an even greater concentration of sugars. After grape harvesting, the whole clusters, or selected berries, are subjected to soft-pressing and the juice is cold-decanted. Fermentation is carried out either naturally or with the addition of a starter culture. After the fermentation, ageing of the wine takes place in barrique (small barrels) for a period of 12–24 months.

The aim of the present study was to describe the yeast microbial communities during production of Picolit wines. Its grapes are characterized by a high sugar content, owing to the limited number of berries in the clusters as well as the drying process, and by a significant mould load. As a consequence, the resulting must may significantly affect the microbial communities, determining yeast dynamics that are different from ‘normal’ must fermentations. For this reason the fermentation dynamics of two Picolit sweet wines, in two different wineries in the Friuli Venezia Giulia region, north-east Italy, were studied. During fermentations, microbiological analyses were carried out to count yeasts, moulds, lactic acid bacteria (LAB) and acetic acid bacteria (AAB). Moreover, a total of 483 yeast strains were isolated and identified by considering their morphology on Wallerstein laboratory nutrient (WLN) medium and by molecular methods. In parallel, total DNA and RNA were directly extracted from the samples and PCR-DGGE was performed using universal primers for yeasts. Finally, in order to assess the ability of the inoculated strains to take over the natural yeast microbial communities and conduct the fermentations, Saccharomyces cerevisiae isolated during fermentation and from the starter cultures were subjected to random amplification of polymorphic DNA (RAPD)-PCR.

Materials and methods

Wine fermentations and sampling procedures

Picolit wine fermentations coming from two different wineries in the Friuli Venezia Giulia region were studied. Here, they are called fermentations B and F. Grapes were in both cases dried on the plant for about 60 days and musts were inoculated with a starter culture of S. cerevisiae. Only the grape juice coming from winery F was clarified by cold treatment at 4 °C for 24 h. Before inoculation, the grape juices were supplemented with SO2 to reach about 80 mg L−1. Fermentation F was carried at ambient temperature (15–20 °C), while fermentation B was conducted at low temperatures (below 10 °C).

The samples were analysed in triplicate using both microbiological and molecular methods. Samplings were performed on grapes and grape juices, and at 1, 3, 5, 7, 14, 30 and 60 days of fermentation for winery F and at 3, 5, 7, 14, 30 and 60 days of fermentation for winery B.

Traditional microbiological analysis

For all samples, decimal dilutions in saline-peptone water (8 g L−1 NaCl, 1 g L−1 bacteriological peptone; Oxoid, Milan, Italy) were made. Grapes (about 50 g) were placed in a sterile stomacher bag and pressed manually to produce juice for analysis. The following microbiological analyses were performed in duplicate agar plates: (1) yeasts on WLN agar (Oxoid) incubated at 25 °C for 3–5 days; (2) moulds on malt extract agar supplemented with tetracycline (1 μg mL−1; Sigma, Milan, Italy) and incubated at 25 °C for 3 days; (3) LAB on double-layer MRS agar (Oxoid) supplemented with delvocid (25 μg mL−1; DSM Specialties, Heerlen, The Netherlands) incubated at 30 °C for 48 h; (4) AAB on ethanol agar [10 g L−1 yeast extract (Oxoid), 20 g L−1 CaCO3 (Sigma), 20 g L−1 agar (Oxoid), 20 mL L−1 ethanol (Sigma)], incubated at 30 °C for 48 h. After counting, means and SDs were calculated. For each sampling, a total of 30 colonies of yeasts with different colour and morphology were isolated from the WLN agar and where possible 15 colonies of moulds were randomly selected. To isolate representatives of the starter cultures used in the fermentation, 1 g of dried yeast was resuspended in sterile water, and after incubation at 30 °C for 1 h, it was plated onto WLN medium. After incubation at 30 °C for 3–5 days, 18 colonies of the starter culture were isolated for both wineries B and F. All the isolates were streaked on the corresponding fresh medium and then stored at −20 °C in YPD broth (20 g L−1 glucose, 20 g L−1 bacteriological peptone, 10 g L−1 yeast extract, all from Oxoid), supplemented with glycerol (30% final concentration; Sigma). The isolates were then identified via molecular analysis.

Chemical analysis

Sugars (glucose and fructose), acetic acid, glycerol and ethanol were determined by HPLC according to Schneider et al. (1987). Each sample was analysed three times.

DNA extraction from pure cultures

Two millilitres of an overnight culture was centrifuged at 14 000 g for 10 min at 4 °C to pellet the cells, which were subjected to DNA extraction using a bead beater treatment as suggested by Cocolin et al. (2000).

Direct nucleic acid extraction from Picolit samples

The protocols described by Mills et al. (2002) were used. Briefly, after each sample had settled for 5 min, 2 mL each for DNA and RNA extraction was transferred into a screw-cap tube containing 0.3 g of glass beads with a diameter of 0.5 mm and they were centrifuged at 12 000 g for 10 min at 4 °C. The cell pellet was resuspended in 300 μL of breaking buffer [2% Triton X-100, 1% sodium dodecyl sulphate, 100 mM NaCl, 10 mM Tris (pH 8), 1 mM EDTA (pH 8)]. Subsequently, 300 μL of phenol/chlorophorm/isoamyl alcohol (25 : 24 : 1, pH 6.7; Sigma) was added for the DNA extraction, while 300 μL of phenol/chlorophorm (5 : 1, pH 4.7; Sigma) was added for the RNA extraction. The cells were homogenized in a bead beater instrument (Fast Prep 24, Bio 101, Vista, CA) three times each for 30 s, with an interval of 10 s between each treatment. The mixture was then centrifuged at 14 000 g for 10 min at 4 °C and the aqueous phase was then purified using the DNeasy Plant Mini kit and the RNeasy Plant Mini kit (Qiagen, Valencia, CA) to purify the DNA and RNA, respectively. The instructions of the kit manufacturer were followed.

PCR and reverse transcriptase (RT)-PCR

DNA extracted directly from the samples was amplified with primers NL1 (5′-GCCATATCAATAAGCGGAGGAAAAG-3′) and LS2 (5′-ATTCCCAAACAACTCGACTC-3′) as previously described (Cocolin et al., 2000). A GC clamp (5′-GCCCGCCGCGCCCCGCGCCCGTCCCGCCGCCCCCGCCCG-3′) was added to primer NL1 when employed for DGGE analysis (Sheffield et al., 1989). RT-PCR was performed with the M-MLV reverse transcriptase of Promega (Milan, Italy). One microgram of RNA was mixed with 10 μM of primer LS2 in a final volume of 10 μL, made up with the addition of sterile water. The mix was incubated at 70 °C for 5 min. After being chilled on ice, the reverse transcription reaction mix was added. The final concentrations in the 25 μL RT reaction were: 50 mM Tris-HCl (pH 8.3), 75 mM KCl, 3 mM MgCl2, 10 mM dithiothreitol (DTT), 2 mM dNTPs, 4 μM primer, 200 U M-MLV and 0.48–0.96 U RNasin RNAse inhibitor (Promega). The reverse transcription was carried out at 42 °C for 1 h and 1 μL of the synthesized cDNA was used for the regular PCR reaction as described previously.

DGGE analysis

The Dcode Universal Mutation Detection System (BioRad, Hercules, CA) was used for DGGE analysis and electrophoreses were performed in a 0.8-mm polyacrylamide gel (8%, w/v, acrylamide/bisacrylamide 37.5 : 1). A denaturing gradient from 30 to 60% was used. The electrophoretic runs were carried out at a constant voltage of 120 V for 4 h with a constant temperature of 60 °C. DGGE gels were stained for 20 min in 1.25 × Tris-acetate-EDTA containing 1 × SYBR Green (Molecular Probes, Eugene, OR) and visualized under UV light at 520 nm. Images were digitally captured and analysed using the GeneGenius Bio Imaging System (SynGene, Cambridge, UK). DGGE analysis was performed at least twice. Bands of interest were excised from the gel as described by Cocolin et al. (2001) and after cloning in pGEM-T Easy vector they were checked by performing a new PCR-DGGE and sequenced by a commercial facility (MWG Biotech, Edersberg, Germany). Sequence comparisons were performed using the blast program (Altschul et al., 1997).

Molecular identification of the isolates

Yeast and mould isolates were subjected to PCR-DGGE in order to group them based on their DGGE mobility. Sequencing of representative strains of each group was then performed. DNA from yeast and moulds was amplified with primer set NL1GC/LS2 and the PCR products were analysed by DGGE as described above. At least five representatives for each migration group, where possible, were amplified with the primer set NL1/NL4 as previously reported (Kurtzman & Robnett, 1997) and sent for sequencing to MWG Biotech. The resultant sequences were aligned in GenBank using the blast program for their identification.

RAPD analysis

One hundred nanograms of the DNA extracted from S. cerevisiae isolated from the starter cultures and during the Picolit fermentations was subjected to RAPD-PCR with primer M13 (5′-GAGGGTGGCGGTTCT-3′) (Huey & Hall, 1989) as previously reported (Andrighetto et al., 2000). Images of the gels were captured digitally by BioImaging System GeneGenius (SynGene), and Gel Compare, version 4.1 (Applied Maths, Kortrijk, Belgium), was used for pattern analysis. The calculation of similarities in the profiles of bands was based on the Pearson product-moment correlation coefficient. Dendrograms were obtained by means of the unweighted pair group method using arithmetic average (UPGMA) clustering algorithm (Vauterin & Vauterin, 1992). A coefficient of correlation of 70% was arbitrarily selected to distinguish the clusters. Yeast isolates were subjected to RAPD-PCR analysis at least twice.

Results

Microbiological counts and yeast morphology on WLN plates

The results of the plate counts obtained from the two wine fermentations monitored in this study are reported in Table 1. Yeast counts varied in grapes and grape juices between the two fermentations. In fermentation B, values were c. 106 CFU mL−1, and in fermentation F c. 104 CFU mL−1. Yeasts then increased to 107 CFU mL−1 until the day 30 in the two fermentations. At day 60, yeast counts decreased to 104 and 106 CFU mL−1 in fermentations B and F, respectively. Moulds were present only in grapes and grape juices, with counts being comparable in B and F. LAB load was 103 CFU mL−1 in grapes B and 60 CFU mL−1 in grapes F. LAB were present until the end of fermentation B until day 30 with counts of about 102–103 CFU mL−1, decreasing to 20 CFU mL−1 at the end of the fermentation. In fermentation F, LAB were detected on plates only until day 14. AAB were found only in grape juice F with values of 102 CFU mL−1.

Table 1.   Microbial counts, as determined by plating, of the wine fermentations monitored in this study
Fermentation B
Microbiological parameterGrapesGrape juiceDay of fermentation
357143060 
 
 Yeast9 × 1051 × 1062 × 1074 × 1072 × 1074 × 1071 × 1072 × 104 
 Moulds2 × 1042 × 104<10<10<10<10<10<10 
 Lactic acid bacteria2 × 1033 × 1034 × 1027 × 1021 × 1023 × 1021 × 10220 
 Acetic acid bacteria<101 × 102<10<10<10<10<10<10 
Fermentation F
Microbiological parameterGrapesGrape juiceDay of fermentation
1357143060
  1. Values are expressed as CFU mL−1.

 
 Yeast1 × 1048 × 1032 × 1075 × 1073 × 1073 × 1072 × 1072 × 1072 × 106
 Moulds3 × 1052 × 103<10<10<10<10<10<10<10
 Lactic acid bacteria602 × 10340502 × 1021 × 1021 × 102<1<1
 Acetic acid bacteria<10<10<10<10<10<10<10<10<10

On WLN agar plates, five different yeast morphotypes were observed. Colony types A and B were both characterized by a smooth and shiny surface, and they were convex with creamy consistency. Type A was cream to brown, while type B was cream coloured. From the bottom of the plate the colony aspect was brown and cream for type A and B, respectively. Colony types C and D both had a green colour with shiny surface. Whereas colony D was totally green, type C had a characteristic white outline. Both colony types appeared green from the bottom of the plate. Lastly, colony type E was wrinkled and convex, with a grey colour. It was also grey as observed from the bottom of the plate. Grapes and grape juices were characterized by a high degree of biodiversity, as all five morphotypes were observed. A high percentage of yeast type C was found in grapes B, while yeasts with colony type D were found mainly in grape juice B and grapes F. After inoculation of the starter cultures, only yeast type B was found on all sampling days (data not shown).

Chemical analysis

The results of chemical analyses of grapes, juice and fermenting juice are reported in Table 2. Fermentation B was characterized by a higher content of glucose and fructose, 234.6 and 244.0 g L−1, respectively, as compared with fermentation F, 179.5 and 189.2 g L−1, respectively. During fermentations, yeasts progressively consumed the sugars, showing a glucophylic trend. At 60 days of fermentation, about one-third of the glucose and more than half of the fructose were left. Consumption of the sugars led to the production of about 11% ethanol (v/v) in both fermentations. In the first days of fermentation B, the acetic acid content was higher than in fermentation F, but at the end of the period followed the amounts measured were comparable. The glycerol content was appreciably different in the fermentations followed. Whereas in fermentation B the initial and final glycerol content was 2.31 and 18.79 g L−1, respectively, in fermentation F it started from 0.68 and increased to 12.37 g L−1.

Table 2.   Results of the chemical analysis performed by HPLC on the fermentation samples
Fermentation BGrapesGrape juiceDay of fermentation
357143060
MeanSDMeanSDMeanSDMeanSDMeanSDMeanSDMeanSDMeanSD
Acetic acid (g L−1)0.150.020.170.020.950.031.240.021.340.031.480.031.680.021.830.02
Ethanol (%, v)ND0.120.036.000.206.880.237.370.188.250.249.730.2310.660.30
Glycerol (g L−1)2.310.515.050.4214.710.6215.610.5516.090.7216.460.6517.721.0218.791.24
Glucose (g L−1)234.64.6221.85.2140.66.4124.04.2120.35.0105.23.889.95.287.94.8
Fructose (g L−1)244.04.4229.44.7184.44.8174.05.0173.74.2164.04.0158.04.9161.33.6
Glucose/fructose ratio0.960.010.970.010.760.020.710.010.690.010.640.020.570.020.540.02
Fermentation FGrapesGrape juiceDay of fermentation
157143060
MeanSDMeanSDMeanSDMeanSDMeanSDMeanSDMeanSDMeanSD
  1. ND, not determined.

Acetic acid (g L−1)0.140.010.080.020.420.021.070.021.130.021.180.011.250.021.310.04
Ethanol (%, v)ND0.170.021.650.216.690.147.410.248.680.3210.240.311.110.35
Glycerol (g L−1)0.680.072.340.175.310.3210.860.1811.090.2410.490.2811.730.2512.370.31
Glucose (g L−1)179.52.3185.43.5165.5492.72.987.93.575.24.262.33.857.24.0
Fructose (g L−1)189.23.3191.74.0182.83.2142.14.6140.63.9132.94.1125.95.1122.82.9
Glucose/fructose ratio0.950.010.970.010.910.020.650.020.630.010.570.010.500.020.470.02

Molecular identification of isolates

A total of 483 yeast isolates, of which 130 were isolated before addition of the starter culture and 353 after inoculation, previously characterized on WLN, were identified by PCR-DGGE analysis and 26S rRNA gene sequencing. Migration profiles of the colony types from WLN via DGGE are reported in Fig. 1. Yeasts characterized as colony types A, C and E had only one DGGE profile, whereas yeast types B and D had seven and three different DGGE profiles, respectively. After sequencing of representatives of each group, the results reported in Table 3 were obtained. Fourteen different yeast species were identified. Most of the isolates from grapes and musts belonged to species of Metschnikowia, Hanseniaspora and Candida, and a smaller number of species of Pichia, Torulaspora, Debaryomyces, Zygosaccaromyces and Saccharomyces were found, as well. After inoculation of the starter cultures, only the presence of S. cerevisiae was observed. Fermentation B was characterized by the presence of 12 yeast species. Prior to inoculation of the starter culture, Candida zemplinina and Hanseniaspora clermontiae/uvarum were the most abundant species. Concerning fermentation F, a smaller number of species were found. Before the addition of the starter, H. clermontiae/uvarum was the most common, followed by Metschnikowia sp. and C. zemplinina (Table 3).

Figure 1.

 DGGE migration profiles of the yeasts with different morphotype isolated during the two Picolit wine fermentations. Lane 1, Metschnikowia sp.; lane 2, Debaryomyces hansenii; lane 3, Pichia anomala; lane 4, Pichia guilliermondii; lane 5, Torulaspora delbrueckii; lane 6, Zygosaccharomyces bailii; lane 7, Saccharomyces cerevisiae; lane 8, Candida riodocensis/stellata; lane 9, Candida zemplinina; lane 10, Hanseniaspora clermontiae/uvarum; lane 11, Candida oleophila; lane 12, Hanseniaspora osmophila; lane 13, Hanseniaspora vinae; lane 14: Pichia membranifaciens. A, B, C, D, E: different colony type detected on WLN.

Table 3.   Results of the molecular identification of yeasts isolated during Picolit wine fermentations: (a) prior to inoculation of the starter culture; (b) after addition of the starter culture
SpeciesNumber
of
strains
From
fermentation
B
From
fermentation
F
(a) Before inoculation
 Candida oleophila66/
 Candida riodocensis/stellata2/2
 Candida zemplinina23203
 Debaryomyces hansenii11/
 Hanseniaspora clermontiae/uvarum471829
 Hanseniaspora osmophila33/
 Metschnikowia sp.201010
 Pichia anomala2/2
 Pichia guilliermondii22/
 Pichia membranifaciens853
 Saccharomyces cerevisiae734
 Torulaspora delbrueckii963
 Zygosaccharomyces bailii11/
Total1307555
(b) After addition
 Saccharomyces cerevisiae353175178
Total353175178

A total of 35 mould isolates were identified by sequencing. Species of Penicillium and Cladosporium, Fusarium acutotum/proliferatum and Microdiploidia hawaiiensis were the main representatives of the mould microbial communities in grapes and grape juices. No differences between the two fermentations followed could be observed in terms of species composition (data not shown).

Analysis by DGGE of nucleic acids extracted directly from the samples

The PCR-DGGE and RT-PCR-DGGE profiles obtained from the DNA and the RNA extracted directly from the two fermentations are shown in Fig. 2, whereas the results of the sequencing of the bands excised from the gel are reported in Table 4. No differences were observed in the DGGE profiles when the triplicate samples were analysed (data not shown). Grapes and grape juices had a complex picture. The great biodiversity of moulds on the grapes was underlined, both at DNA and RNA level, while the yeast contribution started to become important only in the musts, where bands belonging to C. zemplinina, H. uvarum and S. cerevisiae were visible. Debaryomyces sp./Candida sp. was found after 3 days in fermentation B. As for the plating analysis, also at DGGE level, after inoculation of the starter cultures, only the band specific for S. cerevisiae was detected in both fermentations.

Figure 2.

 Results obtained by DGGE analysis of the DNA (a) and RNA (b) directly extracted from the fermentation samples. Line designations: G, grapes; J, grape juice; 1, 3, 7, 14, 30 and 60 refer to days of fermentation. Species identification for band numbers are given in Table 4.

Table 4.   Identification, based on blast comparison in GenBank, of the bands obtained by PCR-DGGE using universal primers NL1/LS2
 IdentificationHomology
(%)
Band no. from DNA
 1Penicillium sp.97
 2-8-13Aureobasidium pullulans99
 3-12Candida zemplinina98
 4-9Botryotinia fuckeliana99
 5-14-15Saccharomyces cerevisiae99
 6Dichomera saubinetii/Botryosphaeria dothidea97
 7Cladosporium sp98
 10Heteroduplex 
 11Hanseniaspora uvarum99
Band no. from RNA
 16Lemonniera pseudofloscula/Alternaria sp.100
 17-27Dichomera saubinetii/Botryosphaeria dothidea96
 18Fusicoccum mangiferum/F. arbuti/Botryosphaeria ribis99
 19-26-28-32Saccharomyces cerevisiae99
 20-29Cladosporium sp.98
 21Candida zemplinina99
 22-30Aureobasidium pullulans99
 23Vuilleminia comedens99
 24Leptosphaerulina trifolii99
 25Debaryomyces sp./Candida sp.99
 31Botryotinia fuckeliana99

Characterization of S. cerevisiae strains by RAPD-PCR

A total of 396 S. cerevisiae strains, isolated directly from the starter cultures and during the fermentations, were subjected to RAPD-PCR followed by cluster analysis using a coefficient of similarity of 70%. Table 5a describes the cluster obtained after RAPD analysis of 196 S. cerevisiae strains from fermentation B. Twelve clusters were observed, while two strains formed single-strain clusters. The most numerous were clusters III and V with 41 and 42 strains, respectively. Clusters VII, IV, IX and I were composed of 24, 23, 19 and 14 strains, respectively. Six small clusters (II, VI, VIII, XI, XII, X) were obtained grouping three, five, two, five, six and eight strains, respectively. The strains isolated from starter culture, found only in clusters III and X, with 17 isolates and one isolate, respectively, were grouped with strains isolated only after day 5, thereby highlighting their predominance only in the latter stages of the fermentation. Of interest is the presence of autochthonous strains in the central stages of the fermentation (clusters IV, V, VII, IX). Clusters described in Table 5b were obtained for the S. cerevisiae strains isolated from fermentation F. Eight major clusters were obtained and also in this case two strains formed single-strain clusters. Cluster II was the largest, with a total of 99 strains. Clusters V, I, III, VII were composed of 33, 23, 21 and 13 strains, respectively. Three small clusters (VIII, VI, IV) were obtained grouping two or five strains. The strains isolated from starter culture were observed in clusters II, III and V, with three, five and 10 isolates, respectively, and only representatives grouping in clusters II and V conducted the alcoholic fermentation.

Table 5.   Cluster analysis of the profiles obtained by RAPD-PCR from the Saccharomyces cerevisiae strains isolated from fermentations B (a) and F (b)
Fermentation B
Cluster numberNumber of strains
isolated
from starter
Number of strains isolated from grapes, must and during the wine fermentation at specific sampling dayTotal
number
GJDay 3Day 5Day 7Day 14Day 30Day 60
 I\\\\\\14\\14
 II\3\\\\\\\3
 III17\\\\\4101041
 IV\\\149\\\\23
 V\\\121218\\\42
 VI\\\41\\\\5
 VII\\\\\2175\24
 VIII\\\\\\2\\2
 IX\\\1711\\\19
 X1\\\4\\\38
 XI\\\\\\\415
 XII\\\\\\\6\6
Total         194
Fermentation F
Cluster numberNumber of strains
isolated
from starter
Number of strains isolated from grapes, must and during the wine fermentation at specific sampling dayTotal
number
GJDay 1Day 3Day 5Day 7Day 14Day 30Day 60
I\\1121\\\1823
 II3\212212232510199
 III5\\\\\16\\\21
 IV\\\\\\\\\55
 V10\\2113112333
 VI\\\\\\\\112
 VII\\\\\\\\5813
 VIII\\\\\\2\\\2
Total          198

Discussion

We have used a multiphasic approach to investigate the ecology of yeasts during the fermentation of Picolit sweet wine from two different wineries in the Friuli Venezia Giulia region. Culture-dependent and -independent methods were used to profile the yeast communities present during the wine fermentations. Moreover, the ability to take over the fermentation of the S. cerevisiae starter cultures was assessed by RAPD-PCR. In carrying out this study we wanted to investigate the effect of the particular characteristics of the must (high sugar content and high mould load) on the yeast dynamics and biodiversity and on the performance of S. cerevisiae starter cultures.

Considering the dynamics of the main microbial groups throughout the process, from grapes to wine, we confirmed previous findings from similar studies (e.g. Mills et al., 2002). Yeast counts were different in grapes and grape juices between the two fermentations. In the grape juices, in particular, a difference of two to three orders of magnitude was found between fermentations B and F, where the loads were about 106 and 104 CFU mL−1, respectively. This result can be explained by the fact that juice in winery F was subjected to cold clarification, which may result in a decrease in yeast content due to the elimination of solid particles to which yeast cells are attached (Boulton et al., 1996), and this can also explain the reduced biodiversity in terms of yeast species found in fermentation F with respect to fermentation B (Table 3). After starter culture inoculation, yeast counts showed a rapid increase in both fermentations. Several species of non-Saccharomyces yeast were found, but strains of S. cerevisiae were also isolated. As often reported previously, most of the isolates from grapes and musts belonged to species of Metschnikowia, Hanseniaspora and Candida (Martini et al., 1996; Fleet et al., 2002; Sabate et al., 2002; Loureiro & Malfeito-Ferreira, 2003). However, smaller numbers of species of Pichia, Torulaspora, Debaryomyces, Zygosaccaromyces, Issatchenkia, Rhodotorula, Rhodosporidium, Cryptococcus and Saccharomyces can also be found (Pretorius, 2000; Prakitchaiwattana et al., 2004; Nisiotou & Nychas, 2007). After the inoculation of the starter cultures, only S. cerevisiae was observed. Yeast biodiversity in wine fermentation is an important aspect to take into consideration because it has been demonstrated that differences in wine quality are clearly related to the levels of secondary compounds, which are principally determined by the yeast species involved in the fermentation process (Romano et al., 2003).

The use of PCR-DGGE and RT-PCR-DGGE of the nucleic acid extracted directly from the grapes and grape juices underlined the high biodiversity in these samples. A considerable number of species of moulds were identified, whereas only bands belonging to yeast species C. zemplinina, H. uvarum and S. cerevisiae were visible. Aureobasidium pullulans was found on the grapes of both wineries at DNA and RNA level. This result is in agreement with a previous study in which it was shown that A. pullulans is the main species isolated from immature, mature, and both damaged and undamaged grapes (Prakitchaiwattana et al., 2004). Of particular interest was the detection of C. zemplinina, already described as actively involved in sweet wine fermentation, and able to grow at high sugar concentrations, at low temperatures and in the presence of ethanol (Sipiczki, 2003). While C. zemplinina and Candida spp. were found to enter in a viable nonculturable state in previous studies (Mills et al., 2002; Divol & Lonvaud-Funel, 2005), here we did not see any specific RNA signature after the inoculation of S. cerevisiae. Candida zemplinina was isolated more frequently in fermentation B, where a total number of 20 strains was found (Table 3). It is interesting to note that in this fermentation a significant drop in the fructose content was recorded when analysing the grape juice and the must at 3 days of fermentation (Table 2). This evidence, in our opinion, may be correlated with the presence of C. zemplinina. However, it should be pointed out that the fructophilic nature of C. zemplinina has been demonstrated in grape juice only for strain EJ1 described by Mills et al. (2002); this specific behaviour needs to be better understood. This trend was not found in fermentation F, where C. zemplinina was isolated less frequently. Moreover, a difference in the glucose/fructose ratio was observed at the initial stage of fermentation, where a value of 0.76 and 0.91 was determined for fermentation B and F, respectively (Table 2). After the inoculation of the starter cultures, also using DGGE, dominance of S. cerevisiae was confirmed. At both DNA and RNA level only the band specific for S. cerevisiae was detected. This underlines not only the numerical importance of S. cerevisiae but also its metabolic activity.

The results obtained from our chemical analyses highlight the high concentration of sugars in fermentation B. We associate this to the production by S. cerevisiae of significant amounts of acetic acid during the fermentations. Indeed, it has been demonstrated that stress factors, such as sugar concentration, provoke an up-regulation of structural genes involved in the formation of acetic acid from acetaldehyde (Erasmus et al., 2003).

Starter cultures allow a rapid and complete fermentation with high reproducibility in the character of wines. It should be pointed out that when using commercial starter cultures, it is very important to assess their ability to take over the fermentation process. In this study, when the S. cerevisiae isolated during the fermentations were compared with those obtained from the starter culture used in the productions, by RAPD-PCR it was determined that only in fermentation F did the inoculated strain conduct the alcoholic fermentation. In fermentation B, other populations of S. cerevisiae could be isolated, most likely contributing to the final characteristics of the sweet wine produced. Previously, Torriani et al. (1999), studying the ecology of Amarone wine from the Valpolicella area (Verona, Italy), produced from partially dried grapes and fermented at low temperatures for 16–20 days, reported the copresence of Saccharomyces bayanus and S. cerevisiae. In order to verify that the clusters obtained in the RAPD-PCR analysis did not generate from different species of Saccharomyces sensu stricto, representatives of each cluster were tested by DGGE using the conditions described by Cocolin et al. (2002) that allow differentiation of S. bayanus and S. cerevisiae. For all the strains, a single migration pattern was observed and this did correspond to S. cerevisiae.

This work was performed to identify and characterize yeasts present during sweet wine production using culture-dependent and -independent methods. A wide range of microorganisms can be isolated from grapes and must and, therefore, the possibility to perform a rapid microbiological control of the yeasts becomes of great oenological importance. The combination of both traditional analysis and molecular methods resulted in a complete picture of the yeast ecology of the system investigated, efficiently and rapidly, and allowed us to understand better the relationship of yeast species during sweet wine fermentations.

Acknowledgements

We would like to thank the technical staff of the wineries that provided the Picolit samples studied.

Ancillary