Screening and typing of Patagonian wine yeasts for glycosidase activities


  • M.E. Rodríguez,

    1. Laboratorio de Microbiología y Biotecnología, Departamento de Química, Facultad de Ingeniería, Universidad Nacional del Comahue, Neuquén, Argentina
    Search for more papers by this author
    • *

      Consejo Nacional de Investigaciones Científicas y Técnicas de la República Argentina (CONICET).

  • C.A. Lopes,

    1. Laboratorio de Microbiología y Biotecnología, Departamento de Química, Facultad de Ingeniería, Universidad Nacional del Comahue, Neuquén, Argentina
    Search for more papers by this author
    • *

      Consejo Nacional de Investigaciones Científicas y Técnicas de la República Argentina (CONICET).

  • M. van Broock,

    1. Laboratorio de Microbiología y Biotecnología, Departamento de Biología, Centro Regional Universitario Bariloche, Universidad Nacional del Comahue, Bariloche, Río Negro, Argentina
    Search for more papers by this author
    • *

      Consejo Nacional de Investigaciones Científicas y Técnicas de la República Argentina (CONICET).

  • S. Valles,

    1. Departamento de Biotecnología, Instituto de Agroquímica y Tecnología de Alimentos (C.S.I.C.), Paterna, Valencia, Spain
    Search for more papers by this author
  • D. Ramón,

    1. Departamento de Biotecnología, Instituto de Agroquímica y Tecnología de Alimentos (C.S.I.C.), Paterna, Valencia, Spain
    Search for more papers by this author
  • A.C. Caballero

    1. Laboratorio de Microbiología y Biotecnología, Departamento de Química, Facultad de Ingeniería, Universidad Nacional del Comahue, Neuquén, Argentina
    Search for more papers by this author

Adriana C. Caballero, Laboratorio de Microbiología y Biotecnología, Departamento de Química, Facultad de Ingeniería, Universidad Nacional del Comahue, Neuquén, Argentina (e-mail: .


Aims:  The purpose of this study was to select autochthonous glycosidase producer yeasts with potential use in industrial production of Patagonian red wines.

Methods and Results:  The study was carried out in oenological autochthonous yeasts from Comahue region (Argentinean North Patagonia). A set of screenable yeast phenotypic characteristics indicative of their potential usefulness in more aromatic red wine production was defined and tested in both, Saccharomyces and non-Saccharomyces populations. Twelve isolates showing six different glycosidase phenotypes were selected and they were characterized at species and strain levels using molecular methods. A close correlation between molecular and phenotypic characteristics was observed. Five strains belonging to Candida guilliermondii, C. pulcherrima and Kloeckera apiculata with highest constitutive β-glucosidase activity levels without anthocyanase activity were discriminated. Some of them also showed constitutive β-xylosidase and inductive α-rhamnosidase activities.

Conclusions:  The extension of the selection of oenological yeast to non-Saccharomyces species provided strains possessing novel and interesting oenological characteristics which could have significant implications in the production of more aromatic young red wine.

Significance and Impact of the Study:  As these non-Saccharomyces are indigenous to wine, they can be used in mixed starters at the beginning or as pure cultures at the end fermentation to contribute in enhancing the wine nuance that is typical of this specific area.


Flavour (aroma and taste) is the most important distinguishing characteristic of wine (Lambrechts and Pretorius 2000). Wine flavour is classified according to the sources of the different compounds contributing to it, this includes varietal, prefermentative, fermentative and postfermentative flavours (Schreier 1979; Rapp 1998). It is well known that several secondary metabolites accumulated in grapes are responsible for providing the basis of the ‘varietal character’ of the wine (Rapp and Versini 1991). These compounds can be found in grapes and musts as free, volatile and odorous forms as well as in flavourless, nonvolatile forms β-glycosidically bound to disaccharide molecules such as 6-O-α, l-arabinofuranosyl-, 6-O-α, l-rhamnopyranosyl- and 6-O-β-d-apiofuranosil-β-d-glucose (Vasserot et al. 1995; Winterhalter and Skouroumounis 1997). These flavourless glycosidic complexes are generally more abundant than free odorous forms and they represent an important potential source of the fragrant compounds in wine requiring enzymatic or acidic hydrolysis for the liberation of their fragrances (Günata et al. 1985; Williams et al. 1989). The enzymatic hydrolysis of these glycosidic complexes is a two-step sequential reaction: first, an α-l-arabinofuranosidase, an α-l-rhamnosidase or a β-d-apiofuranosidase activity cleaves the (1→6) osidic linkage, and secondly, the flavour compounds are liberated from the monoglucosides by the action of a β-d-glucosidase activity (Günata et al. 1988). Unlike acidic hydrolysis, enzymatic hydrolysis does not result in modifications of the aromatic character (Günata et al. 1990a). Because of their characteristics (glucose inhibition, poor stability at the low pH), grape glycosidase activities have a minimal effect on enhancing varietal aroma during winemaking (Canal-Llauberes 1993). As an alternative to the inefficient grape glycosidases, the presence of these activities has been investigated in several fungal and yeast species and it has been found that some of them possess interesting properties for practical uses (Günata et al. 1985, 1990a,b; Aryan et al. 1987; Vasserot et al. 1991; Rosi et al. 1994; Rogerson et al. 1995; Riou et al. 1998; Riccio et al. 1999; Spagna et al. 2002).

In this work, we investigated the presence of several glycosidase activities in oenological indigenous yeasts from Comahue region (north-western Patagonia, Argentina) in order to use these yeasts themselves, or their enzymes, to increase the varietal character of the regional red wines. To do this, a large number of Saccharomyces and non-Saccharomyces yeasts isolated from grape surfaces and fermentation musts were screened twice to determine the presence of β-glucosidase, β-xylosidase, α-l-arabinofuranosidase and α-l-rhamnosidase activities and their anthocyanase capabilities. The taxonomic identity of 12 indigenous yeast isolates with highest β-glucosidase activity was confirmed by molecular methods at species and strain levels, and their killer phenotype was characterized. On the basis of the relationships between their phenotypic and genotypic characteristics, these indigenous yeast isolates were clustered in six different strains, some of which showed interesting characteristics for winemaking.

Materials and methods


A total of 180 indigenous yeast isolates belonging to the genera Candida (74 isolates), Torulaspora (six isolates), Kloeckera (25 isolates) and Saccharomyces (75 isolates) were analysed. The yeast isolates were obtained in our laboratory from Merlot, Malbec and Trousseau type grapes (van Broock et al. 1996; Lopes 1999) and from three different fermentation stages of their musts spontaneously fermented at laboratory and industrial scales (Rodríguez 1999). Yeasts have already been identified at genus and species level according to conventional physiological and morphological tests and keys proposed by Kreger-van Rij (1984) and Kurtzman and Fell (1998) (van Broock et al. 1996; Lopes 1999; Rodríguez 1999).

Glycosidase screening methods

Screening for detecting β-glucosidase activity was carried out on agar plates with arbutin as substrate. The medium, containing 6·7 g yeast nitrogen base (YNB; Difco), 5 g arbutin, 20 g agar per litre and pH 5·0, was autoclaved at 120°C for 15 min. Immediately after sterilization 2 ml of a sterile 1% (w/v) ferric ammonium citrate solution was added to 100 ml melted medium (Rosi et al. 1994). Each plate was inoculated in radial streaks with 24-h-old yeast cultures, incubated at 25°C and examined daily during a week. An uninoculated plate served as the control. Yeast isolates with β-glucosidase activity hydrolyse the substrate and a dark brown halo develops in the agar medium.

Screening for detecting all β-glucosidase, β-xylosidase, α-rhamnosidase and α-arabinofuranosidase activities was carried out on agar plates containing 1·7 g YNB (without amino acids and ammonium sulphate; Difco), 5 g ammonium sulphate, 5 g source carbon and 20 g agar per litre. The pH was adjusted to 5·5. The carbon sources assayed were either arabinose, cellobiose, glucose, rhamnose and xylose (Sigma). Glycosidase activities were determined by using the appropriate 4-methylumbelliferyl glycoside (Sigma) as substrate, as described by Manzanares et al. (1999). The presence of the enzymatic activity was visualized as a fluorescent halo surrounding yeast growth by plate exposition to u.v. light.

β-Glucosidase activity

Liquid culture conditions. A medium containing 1·7 g YNB, 5·0 g (NH4)2SO4, 5·0 g yeast extract, 5·0 g peptone, 5·0 g glucose per litre and pH 5·0 was used to examine yeasts showing β-glucosidase activity and to localize the enzymatic activity. Inocula were prepared by growing a loopful of cells from GPY plate (5·0 g yeast extract, 5·0 g peptone, 40·0 g dextrose, 20 g agar per litre; pH 5·0) in 10 ml of liquid medium. After 24 h, a 2% volume of the inoculum was added to 100 ml Erlenmeyer flasks filled to 20% of their volume and shaken at 180 rev min−1 in a Vicking shaker at 25°C for 24 h (aerobic conditions). In studies on the influence of anaerobic growth conditions, the yeasts were grown in 20 ml screw-capped tubes totally filled and incubated without shaking at 25°C for 3 days. Growth control was checked by monitoring the culture absorbance at 600 nm and tubes or Erlenmeyer flasks containing uninoculated media were used as the blank. Once the assays were finished, cultures were centrifuged (5000 g, 10 min, 4°C) and the cells and culture supernatant were assayed for enzymatic activity and the dry weight was determined.

Whole cell preparation. Cells harvested from 1 ml of culture as indicated above, were washed twice with cold sterile distilled water, centrifuged and the pellet was resuspended in 0·5 ml of 100 mmol l−1 citrate–phosphate buffer, pH 5·0, and used for enzymatic assay.

Permeabilized cell preparation. The procedure of Salmon (1984) with some modifications was used: 5 ml of liquid culture were centrifuged and washed with 5 ml of cold sterile distilled water. The pellet was resuspended in 1 ml of 75 mmol l−1 imidazole buffer pH 7·5, and added quickly to 50 μl of 0·3 mmol l−1 glutathion, 10 μl of 10% Triton X-100 and 50 μl of toluene/ethanol (1 : 4 v/v). The suspension was shaken vigorously for 5 min and then centrifuged. The yeast pellet was suspended in 5 ml cold sterile water: 1 ml of this suspension was centrifuged, washed with cold sterile distilled water and the pellet was resuspended in 0·5 ml citrate–phosphate buffer and used for enzymatic assay.

Enzyme assay.β-Glucosidase activity was assayed by measuring the amount of p-nitrophenol (pNP) released from p-nitrophenyl-β-d-glucoside (pNPG) used as substrate. Enzyme solution (0·1 ml) was mixed with 0·4 ml of a 2 mmol l−1 solution of pNPG in 100 mmol l−1 citrate–phosphate buffer, pH 5·0. The reaction mixture was incubated at 30°C for 5–15 min and subsequently, 2·0 ml of 250 mmol l−1 Na2CO3 was added to stop the reaction. The released pNP in this mixture was measured spectrophotometrically at 405 nm in a Shimadzu UV–V spectrophotometer (Shimadzu Corporation, Kyoto, Japan). The pNP molar extinction coefficient used was ε = 18 300 mol−1 l cm−1. One unit (U) of enzyme activity was defined as the quantity of enzyme that released 1 μmol of pNP per minute under the above experimental conditions. The values of intracellular activity were obtained subtracting the whole cell enzymatic activity values from the permeabilized cell values. All assays were performed in duplicate.

Dry weight. Ten millilitres of culture samples were centrifuged at 10 000 g 5 min. The pellets were washed twice with 5 ml of cold sterile distilled water and resuspended in sterile distilled water. These cell suspensions were placed on preweighed dishes and dried at 105°C until constant weight.

PCR–RFLP analysis of rDNA

The 5·8S-ITS region was amplified in a Progene thermocycler (Techne, Cambridge, UK) using ITS1 (5′-TCCGTAGGTGAACCTGCGG-3′) and ITS4 (5′-TCCTCCGCTTATTGATATGC-3′) primers already described (White et al. 1990). PCR conditions were identical to those described by Esteve-Zarzoso et al. (1999). The amplified DNAs (0·5–10 μg) were digested without further purification with the restriction endonucleases CfoI, HaeIII, HinfI and DdeI (Roche Molecular Biochemicals, Mannheim, Germany) according to the supplier's instructions. PCR products and their restriction fragments were separated on 1·4% (w/v)and 3% (w/v) agarose gels, respectively, with 1x TAE buffer (40 mmol l−1 Tris–acetate, 1 mmol l−1 EDTA, pH 8). After electrophoresis, gels were stained with ethidium bromide (5 μg ml−1) and visualized under u.v. light. A 100-bp DNA ladder marker (Gibco BRL, Gaithersburg, MD, USA) served as size standard.

The taxonomic identity of indigenous S. cerevisiae isolates tested in this study has already been confirmed by us using the same molecular method (Lopes et al. 2002).

Karyotype analysis

DNA was prepared in agarose plugs as described by Carle and Olson (1985). Chromosomal profiles were determined using the CHEF technique with a CHEF-DRIII (Bio-Rad, Richmond CA, USA). The yeast chromosomes were separated on 1% (w/v) chromosomal grade agarose gels (Bio-Rad) under the following running conditions: electric field, 6 V cm−1; angle, 120°; temperature, 14°C; block 1, 60 s pulse time for 15 h; block 2, 120 s pulse time for 13 h; buffer 0·5 × TBE (45 mmol l−1 Tris–borate pH 7·5, 1 mmol l−1 EDTA pH 8·0). After electrophoresis, gels were stained with ethidium bromide (5 μg ml−1) and visualized under u.v. light.

mtDNA-RFLP analysis

Total DNA extraction and mtDNA restriction analysis were performed by the method of Querol et al. (1992) modified by López et al. (2001). Yeast DNA was digested with HinfI restriction enzyme (Roche Molecular Biochemicals, Mannheim, Germany) and the fragments were separated in TAE buffer containing 1% (w/v) agarose gels (Fernández-Espinar et al. 2000).

Killer behaviour

Killer behaviour was assayed using the seeded agar-plate technique as described by Sangorrín et al. (2002). The killer sensitivity of the indigenous yeasts were determined contrasting them to reference panel of the 10 collection well-known killer strains and their killer character was evaluated against two collection sensitive strains. The collection killer strains used were S. cerevisiae YAT 679 (K1 type), S. cerevisiae NCYC 738 (K2 type), S. capensis NCYC 671 (K3 type), C. glabrata NCYC 388 (K4 type), Hansenula anomala NCYC 434 (K5 type), Kluyveromyces fragilis NCYC 587 (K6 type), C. valida NCYC 327 (K7 type), H. anomala NCYC 435 (K8 type) and H. mrakii NCYC 500 (K9 type), K. drosophilarum NCYC 575 (K10 type), provided by Professor Isato Kono (Industrial Technology Center of Okayama Prefecture, Japan), and S. cerevisiae P352 (K1 type) from PROIMI yeast collection from Tucumán (Argentina). The collection-sensitive strains used were S. cerevisiae P351 (PROIMI yeast collection) and C. glabrata NCYC 388 above cited.

Anthocyanase activity

Anthocyanase activity was measured using red wine as substrate. Aerobically grown YNB-cellobiose or rhamnose cells of 1.5 ml was collected by centrifugation and resuspended in 1·5 ml of red wine and anaerobically incubated at 4°C during a week. Wine decolourization was monitored measuring the absorbance at 520 nm in 2 mm light path cuvettes. Glycosidase activity was also assayed using the appropriate pNP glycoside (Sigma) as substrate.


Arbutin, pNPG, pNP, MUG, MUX, MUA, MUR, glutathion and Triton X-100 were supplied by Sigma, and culture medium constituents by Difco. All other chemicals used were of reagent grade.

Data analysis

Significant differences between experimental figures were estimated using Student's test. Dendrograms were generated using an unweighed pair group with arithmetic average (UPGMA) algorithm and the concordance between individuals was (1-SM) matching.


Glycosidase screenings and characterization

Patagonian Saccharomyces and non-Saccharomyces yeast isolates, all of oenological origin, were screened for β-glucosidase activity on agar plates containing arbutin as substrate. Forty-five of 180 analysed isolates showed enzymatic activity and they were characterized as strong (11 isolates), moderate (eight isolates) or weak (26 isolates) according to the colour intensity (dark, fair or light brown) of the halo (Table 1). Among these producer isolates, 44 belonged to C. guilliermondii, C. pulcherrima and Kloeckera apiculata species and only one belonged to S. cerevisiae; this isolate came from grape berry surface and it was a weak producer (Table 1).

Table 1.  Indigenous yeast isolates tested for β-glucosidase activity
SpeciesSourceTested isolate numberActivity* Isolate identity†
  1. *Activity is expressed as hydrolysis of arbutin. The figures between brackets indicate the number of yeast isolates with β-glucosidase activity.

  2. †The capital letters indicate the grape variety: A, Merlot; M, Malbec; T, Trousseau. The first digit indicates the source: 1, grape; 2, must completely processed in laboratory; 3, must partially processed in laboratory; 4, must completely processed in winery and the last two digits represent the isolate number within each series. The superscript letters s, m and w indicate strong, moderate or weak β-glucosidase production, respectively.

Saccharomyces cerevisiaeGrape1+(1)A102w
Saccharomyces kluyveriMust2 
Kloeckera apiculataGrape4+(3)A117w and 119w; T124m
Must21+(15)M201m, 204m, 313w, 315w, 321s, 325s, 326m, 327s; 402w, 407w, 411w, 412w and 414w T202w and 206w
Torulaspora delbrueckiiMust4
Torulaspora pretoriensisMust2
Candida colliculosaMust7
Candida dattilaMust16
Candida guilliermondiiGrape13+(13)M102s and 119s; A108w, 109w, 110w, 111w, 114w, 131w, 132m, 135s; 142w, 146w and 150w
Must3+(3)M229s, 231s and 425s
Candida parapsilosisGrape2
Candida pulcherrimaGrape10+(6)T140w, 149w and 152s A123w, 128w and 162m
Must17+(4)M227s T205m; 218m and 220w
Candida stellataMust6
Total 180+(45) 

All isolates characterized as enzyme producers on plate assay were grown in liquid media under aerobic conditions and they were tested for β-glucosidase activity using p-nitrophenyl-β-d-glucoside as substrate (in vitro assays). In these assays, whole cells and culture media were used as enzyme source. Under assay conditions, none of the yeast isolates showed any capacity to excrete the enzyme into the medium. Analysis of whole cell activity data, using the box-and-whiskers statistic method, revealed that within C. guilliermondii, C. pulcherrima and K. apiculata producer species, the activity levels found in must isolates (median values of 13·00, 1·00 and 0·07 U g−1 dry weight, respectively) were greater than in grape isolates (median values of 2·1, 0·40 and 0·05 U g−1 dry weight, respectively). These results also evidence that C. guilliermondii isolates from must (M229, 231 and 425) were the greatest β-glucosidase producers. However, some single grape isolates of this species (A135, M102 and M119) as well as some single isolates of C. pulcherrima (T152 and M227) and K. apiculata (M321, 325 and 328) also showed very high enzyme activity levels (outlier values), between four to 40 times higher than the medians of their populations. These 11 best β-glucosidase producers, characterized as strong on plate assay, were selected to continue the study. The moderate to weak producer K. apiculata M326 was also included in order to continue checking the screening and selection method and the 12 isolates were renamed as V1–V12 (see Table 2).

Table 2. β-Glucosidase activity in indigenous yeasts under anaerobic and aerobic conditions
SourceSpeciesIsolate identity (original name)Enzymatic activity†
  1. †Enzymatic activity is expressed as U g−1 dry weight. Values without brackets indicate the enzymatic activity after 72 h under anaerobic conditions and values in italics and within brackets indicate the enzymatic activity after 24 h under aerobic conditions. ND, not detected.

  2. *P < 0·01, **P < 0·001 Student's test, n = 2.

GrapesC. guilliermondiiV1 (A135)9·90 ± 0·30 (33·7 ± 3·10)*0·34 ± 0·150 (ND)9·56 ± 0·15 (24·12 ± 1·24)*ND (9·57 ± 1·86)
V2 (M102)12·90 ± 1·52 (60·00 ± 1·70)*0·35 ± 0·09 (ND)11·54 ± 2·51 (38·30 ± 3·92)*1·01 ± 0·90 (21·24 ± 2·12)**
V3 (M103)41·75 ± 3·00 (33·33 ± 5·15)0·29 ± 0·00 (ND)24·87 ± 5·14 (20·47 ± 0·91)16·63 ± 2·14 (12·82 ± 6·06)
C. pulcherrimaV4 (T152)27·67 ± 2·75 (30·98 ± 4·00)0·22 ± 0·12 (ND)1·63 ± 0·06 (6·20 ± 2·98)25·89 ± 2·93 (24·80 ± 1·02)
MustC. guilliermondiiV5 (M425)27·50 ± 3·81 (35·83 ± 8·21)0·14 ± 0·06 (ND)17·11 ± 2·11 (18·87 ± 4·36)10·25 ± 5·98 (16·93 ± 3·85)
V7 (M229)36·60 ± 6·53 (51·86 ± 8·93)0·44 ± 0·12 (ND)20·06 ± 9·12 (33·22 ± 6·42)16·10 ± 2·47 (18·64 ± 2·51)
V8 (M231)43·4 ± 1·88 (37·9 ± 3·27)0·34 ± 0·00 (0·45 ± 0·01)27·04 ± 3·89 (25·01 ± 0·78)16·02 ± 2·01. (12·43 ± 2·48)
C. pulcherrimaV6 (M227)25·15 ± 4·50 (33·7 ± 5·13)ND (ND)5·23 ± 1·63 (4·71 ± 1·72)19·92 ± 6·13 (28·98 ± 6·85)
K. apiculataV9 (M321)5·01 ± 0·28 (22·08 ± 1·61)*0·15 ± 0·01 (ND)4·85 ± 0·29 (16·70 ± 1·42)*ND (5·39 ± 3·03)
V10 (M325)3·42 ± 0·25 (15·56 ± 0·42)*0·09 ± 0·07 (ND)3·31 ± 0·18 (13·59 ± 0·72)*ND (2·01 ± 0·30)
V11 (M326)1·4 ± 0·08 (9·1 ± 0·32)*0·01 ± 0·02 (ND)1·39 ± 0·05 (7·06 ± 0·61)*ND (2·03 ± 0·29)
V12 (M328)3·96 ± 0·30 (12·82 ± 0·80)*0·08 ± 0·11 (ND)3·92 ± 0·19 (11·28 ± 1·00)*ND (1·52 ± 1·80)

Because yeast strains may have exocellular, parietal (cell wall bound) and intracellular β-glycosidase activity, this activity was assayed on the culture supernatant medium (extracellular), on whole cells (parietal) and on permeabilized cells; intracellular activity was estimated from the latter (see ‘Methods’). Taking into account that winemaking is an anaerobic process, yeast cells were grown under aerobic and partially anaerobic conditions, and total β-glucosidase activity was estimated as the sum of the three activities. As a whole, the results from these assays showed two different behaviour patterns among indigenous isolates. C. pulcherrima isolates and four of the six C. guilliermondii isolates (V3, V5, V7 and V8) yielded the highest levels of enzyme activity and neither β-glucosidase production nor cellular location were significantly affected by culture aeration conditions (Table 2). Additionally, whereas in the first isolates almost all the activity was detected in permeabilized cells, indicating the presence of an intracellular enzyme, the last isolates showed important levels in whole cells and permeabilized cells, indicating the presence of both parietal and intracellular enzymes (Table 2). However, under anaerobic conditions, the other two C. guilliermondii isolates (V1 and V2) and K. apiculata isolates showed the highest levels of activity on whole cells (≥89·5 and ≥96·9% of the total activity, respectively), evidencing the presence of parietal enzyme. Aerobic growth conditions greatly stimulated β-glycosidase activity in these isolates (Table 2). Candida guilliermondii V2 shows fivefold more enzymatic activity under this condition than under anaerobic conditions, with a significant increase of both intracellular and parietal activities. Although their activity values are lower, the same trend may be observed in C. guilliermondii V1 and all K. apiculata isolates. Additionally, certain activity was observed in supernatant media, exocellular enzyme, but its level was small and, except for V8 isolate, it was only observed under anaerobic conditions (Table 2).

A second plate screening for detecting other glycosidase activities were realized on the yeast isolates. In this assay, yeast cells were grown on agar plates containing different carbon source and β-d-xylosidase, α-l-rhamnosidase and α-l-arabinofuranosidase as well as β-d-glucosidase were detected using the appropriate 4-methylumbelliferyl-glycoside as substrate. The results showed in Table 3 reveal that all C. guilliermondii and C. pulcherrima isolates were capable to produce both β-glucosidase and β-xylosidase activities on all carbon source where they can grow, indicating that the enzymes could be synthesized constitutively, whilst K. apiculata isolates only produced β-glucosidase. Candida guilliermondii V2 also showed capability to hydrolyse MUR (α-rhamnosidase activity), but this enzymatic activity was only detected using rhamnose as substrate (inducible activity) (Table 3). None of the 12 yeast strains produced α-arabinofuranosidase activity.

Table 3.  Glycosidase activity in indigenous yeasts grown on different carbon source
StrainCarbon source
  1. 1, cell growth; 2, enzymatic activity. The activity was detected on plate using MUG (β-d-glucosidase activity, BGL); MUX (β-d-xylosidase activity, BXL); MUA (α-l-arabinofuranosidase activity, AAF) and MUR (α-l-rhamnosidase activity, ARH) as substrates. ND, not detected.

C. guilliermondii
V1, V3, V5, V7 and V8(+)BGL(+)BGL(+)BGL(+)BGL(+)BGL
C. pulcherrima
V4 and V6(+)BGL(+)BGL(+)BGL(−)ND(−)ND
K. apiculata
V9, V10, V11 and V12(+)BGL(+)BGL(−)ND(−)ND(−)ND

Killer behaviour

Under assay conditions, neither C. pulcherrima, K. apiculata nor C. guilliermondii isolates were capable of producing killer toxin. However, whereas the isolates belonging to the two former species were immune against all assayed killer toxins displaying a neutral phenotype (SK), C. guilliermondii isolates showed a sensitive phenotype to K8, K9 and K10 isolates (S+K).

Antocyanase activity

In this assay yeast cells grown on cellobiose (cell bound β-glucosidase activities ≥10·00 ± 1·50 U g−1 dry weight) or rhamnose (cell bound V2α-rhamnosidase activity = 5·84 U g−1 dry weight) were used as enzyme source and the progress of decolourization of a commercial red wine was studied during a week. No significant differences in colour figures were observed between control (uninoculated red wine, A520 = 0·731 ± 0·010, n = 2) and experimental samples (V2, V5, V6, V9-cellobiose and V2-rhamnose inoculated red wine, A520 = 0·669± 0·039, 0·671 ± 0·051, 0·780 ± 0·021, 0·702 ± 0·060 and 0·739 ± 0·014, respectively; n = 2). However, low but significant residual β-glycosidase (≥100 ± 9 mU g−1 dry weight) and α-rhamnosidase activities (700 ± 23 mU g−1 dry weight) were observed at the end of the assay.

Molecular analysis and typing of the yeast isolates

The correct taxonomic identification of an isolate at species level is a previous essential step to its definitive characterization at enzymatic level.

The taxonomic species identity assigned to 12 indigenous yeast isolates by means of conventional methods was confirmed by PCR–RFLP analysis of the ribosomal internal transcribed spacers including the 5·8S rRNA gene region. Figure 1 and Table 4 show that the indigenous isolates, characterized as belonging to C. guilliermondii, C. pulcherrima and K. apiculata species present the same species-specific patterns reported in literature for type strains belonging to their teleomorph forms Pichia guilliermondii, Metschnikowia pulcherrima and Hanseniaspora uvarum (Esteve-Zarzoso et al. 1999). As the anamorph and teleomorph forms yielded the same pattern by using this method, we concluded that all selected isolates had been correctly characterized at species level.

Figure 1.

PCR–RFLP of indigenous yeast ITS regions. (a) Amplified products. (b, c) Restriction fragments. MW: 100-bp DNA ladder. The restriction endonucleases used in each case are indicated at the top of the pictures and indigenous isolate identity at the bottom

Table 4.  PCR–RFLP profiles of indigenous yeast isolates and collection type strains
StrainAP (bp)†Restriction frgment (bp)‡
  1. *Type strain: from the Spanish Type Culture Collection (CECT). PCR–RFLP profiles were extracted from Esteve-Zarzoso et al. 1999.

  2. †PCR amplified product size.

  3. ‡Size of restriction fragments.

V1, V2, V3, V5, V7 and V8625300 + 265 + 60400 + 115 + 90320 + 300ND
V4 and V6400205 + 100 + 95280 + 100200 + 190ND
V9, V10, V11 and V12750320 + 310 + 105750350 + 200 + 180300 + 180 + 95 + 90 + 85
Pichia guilliermondii 1019− 1021− 1438*625300 + 265 + 60400 + 115 + 90320 + 300
Metschnikowia pulcherrima 1691− 10408− 10546*400205 + 100 + 95280 + 100200 + 190
Hanseniaspora uvarum 1444− 10389*750320 + 310 + 105750350 + 200  +  180300 + 180 + 95 + 90 + 85

The population homogeneity within each species was analysed using karyotype analysis and mtDNA RFLP, two molecular methods widely used to characterize indigenous wine yeasts at strains level (Querol et al. 1992; Schütz and Gafner 1993; Nadal et al. 1996; Fernández et al. 2000). Figure 2a shows the CHEF separation of DNA from the 12 isolates. Candida guilliermondii and K. apiculata isolates displayed chromosomal-length polymorphism. Four of six isolates of C. guilliermondii, V3, V5, V7 and V8, showed a similar banding pattern, with seven bands ranging in size from ca 2200 to 590 kb, that differed from the V1 and V2 karyotype in one additional band of ca 1700 kb (Fig. 2a marked with arrow). However, two karyotype patterns were observed within K. apiculata isolates. Both patterns consisted of five chromosomal bands ranging in size from ca 2200 to 670 kb and they could be differentiated by the electrophoretic mobility of their smallest band (Fig. 2a). Under assay conditions, the chromosomes of C. pulcherrima could no be separated appropriately. Additionally, using mtDNA RFLP method we could distinguish C. guilliermondii V1 from V2 but no difference, neither between C. pulcherrima V4 and V6 isolates nor between others, could be detected (Fig. 2b).

Figure 2.

Molecular analysis of indigenous yeast isolates. (a) Electrophoretic karyotypes. MW: chromosomes of Saccharomyces cerevisiae YNN295 used as the size standard. (b) HinfI mtDNA restriction patterns. MW: weight marker-lambda DNA digest with HindIII. The codes at the tops of both figures indicate the identity of the indigenous isolates and the codes at the bottom indicate the karyotype patterns

Taking the molecular patterns as variables, we performed a cluster analysis (Fig. 3b). The isolates found with linkage distance zero show identical patterns in all used molecular methods and they can be considered as belonging to the same strain (Vezinhet et al. 1992; Fernández et al. 2000). The resulting dendrogram reveals three different strain subpopulations within C. guilliermondii species, two within K. apiculata and only one within C. pulcherrima species (Fig. 3b). Additionally, when this dendrogram was compared with that obtained from biochemical, killer and glycosidase phenotypes (Fig. 3a) a close correlation between clusters was observed. That is, from genetic and phenotypic data sets, we can discriminate exactly the same six strains among indigenous isolates.

Figure 3.

UPGMA cluster analysis of selected yeast isolates based on phenotype (a) and molecular (b) characteristics


Volatile compounds derived from grape glycosidic complexes make an important contribution to varietal flavour of the wine (Francis et al. 1992, 1999; Williams and Francis 1996; Gueguen et al. 1997). β-Glucosidase plays a key role in this aroma enhancing process and this activity has been extensively researched in wine yeast. Although some activity in different strains of S. cerevisiae has been described (Delcroix et al. 1994; Mateo and Di Stefano 1997; Hernández et al. 2002), most studies demonstrate that higher β-glucosidase producers are non-Saccharomyces species (Rosi et al. 1994; Miklósy and Pölös 1995; Charoenchai et al. 1997; Fernández et al. 2000; Mendes Ferreira et al. 2001; Strauss et al. 2001). Our screening and preliminary in vitro assay results confirm these findings. The best β-glucosidase activity producer were all non-Saccharomyces yeasts belonging to C. guilliermondii, C. pulcherrima and K. apiculata species (Table 1). This yeast isolates were preselected and they were characterized according to a set of selectable phenotypic characteristics indicative of their potential usefulness in more aromatic red wine production.

The first criterion to discriminate among strains included the levels of specific β-glucosidase activity, its cellular location and the effect of aeration conditions on them. β-Glucosidase, by releasing aglicon from monoglycosidic complexes, enhances the varietal aroma of wines. The second criterion was the presence of the other glycosidases activities involved in the first step of glycosidic complex sequential hydrolysis as well as β-xylosidase activity. There are no data available about the potential of this activity to improve aroma and flavour properties of wine. However, β-xylosidase contribution to the formation of flavour and colour in other foods obtained from fermentation industries has been reported (Ohta et al. 1991; Kimura et al. 1999), so we decided to include it. The third criterion was a technological property, the killer behaviour of the isolates. Non-Saccharomyces yeasts can be employed at the begin of winemaking as mixed starter cultures in conjunction with more ethanol-tolerant S. cerevisiae strains (Rainieri and Pretorius 2000). To determine the killer behaviour is very important because different killer interactions between starters and between starters and wild yeast strains could be established during fermentation affecting its normal evolution. These interactions become relevant in areas where killer yeasts are widespread such as the north Patagonian region (Sangorrín et al. 2001). The fourth and last, unfavourable criterion, was the presence of anthocyanase activity. Anthocyanins are phenolic molecules composed of a glycosilated flavylium ion and they are responsible for wine colour. Some β-glucosidase from fungis and yeasts break the linkage between the glucose and the anthocyanidin moieties inducing loss of wine colour (Wightman et al. 1997; Sánchez-Torres et al. 1998; Manzanares et al. 2000).

On the basis of these results, we characterized six phenotypically different subpopulation among the selected isolates including the weak isolate used as control (Fig. 3a). Then, using molecular methods we confirmed their taxonomical identity at species level (Fig. 1 and Table 4) and their intraspecific variability (Figs 2 and 3b). From the ecological point of view, these results showed a great diversity of Patagonian indigenous glycosidase producer strains (Tables 2 and 3), with significantly different β-glucosidase profiles. These different glycosidase profiles seem to be more closely related to strains than to species. However, as far as is known, our study is the first one that shows a correlation between molecular polymorphism and specific phenotypic traits in wild indigenous strains. In order to simplify the selection process, the use of molecular methods before the application of oenological criteria have been proposed to discriminate among strains (Esteve-Zarzoso et al. 2000). However, more than one molecular method could be used so as not to miss any strain. With these considerations, molecular polymorphism analysis may be very useful to identify strains with specific phenotypic properties in selection protocols.

From the technological point of view, some characteristics of these Patagonian indigenous strains could justify their utilization in industrial production of more aromatic red wine. One of them is the high β-glucosidase activity levels showed by these strains under anaerobic conditions, much higher than those reported by other authors for yeast strains belonging to these same species (Rosi et al. 1994; Fernández et al. 2000; Manzanares et al. 2000; Mendes Ferreira et al. 2001). Another characteristic is the lack of killer character and their immunity to S. cerevisiae killer toxins. Although S. cerevisiae killer toxins are only active against a few yeast species including S. cerevisiae and C. glabrata, non-Saccharomyces killer toxins have a broader anti-yeast spectrum including S. cerevisiae species. We have observed that indigenous non-Saccharomyces killer yeasts resident in Patagonian winery surfaces and initial musts could hinder the implantation of commercial S. cerevisiae starters in guided fermentation, inducing stuck fermentation (I. Zajonskovsky, T.L. Lavalle, M.E. Rodríguez, Ch. Lopes, M. Sangorrín and A.C. Caballero 2002, personal communication). The last remarkable trait of these strains is the presence of β-glucosidase activities lacking anthocyanase activity and maintaining low but significant residual levels after exposition to finished wine conditions during a significant period of time.

Because some non-Saccharomyces yeasts produce undesirable concentrations of acetic acid and ethyl acetate from sugars, they had always been regarded as unsuitable for winemaking. The potential application of non-Saccharomyces yeast strains in this process has been explored only fairly recently. Due to their sensitivity to ethanol, non-Saccharomyces yeast could be employed at the beginning of winemaking as mixed cultures together with Saccharomyces strains or in sequential non-Saccharomyces–Saccharomyces inoculation protocols (Rainieri and Pretorius 2000; Soden et al. 2000). However, the results obtained in this study allow us to think in an additional and more advantageous way of using these non-Saccharomyces strains towards the end of fermentation, on the young wine.


This work was supported by the Comahue University Grant (B091). The authors wish to thank S. Genoves and E. Ibáñez for help during the work in IATA and Ph.D. S. Bramardi (U.N. Comahue) for his statistical assistance.