OUP user menu

Natural hybrids from Saccharomyces cerevisiae, Saccharomyces bayanus and Saccharomyces kudriavzevii in wine fermentations

Sara S. González, Eladio Barrio, Jürg Gafner, Amparo Querol
DOI: http://dx.doi.org/10.1111/j.1567-1364.2006.00126.x 1221-1234 First published online: 1 December 2006


Several wine isolates of Saccharomyces were analysed for six molecular markers, five nuclear and one mitochondrial, and new natural interspecific hybrids were identified. The molecular characterization of these Saccharomyces hybrids was performed based on the restriction analysis of five nuclear genes (CAT8, CYR1, GSY1, MET6 and OPY1, located in different chromosomes), the ribosomal region encompassing the 5.8S rRNA gene and the two internal transcribed spacers, and sequence analysis of the mitochondrial gene COX2. This method allowed us to identify and characterize new hybrids between Saccharomyces cerevisiae and Saccharomyces kudriavzevii, between S. cerevisiae and Saccharomyces bayanus, as well as a triple hybrid S. bayanus×S. cerevisiae×S. kudriavzevii. This is the first time that S. cerevisiae×S. kudriavzevii hybrids have been described which have been involved in wine fermentation.

  • natural hybrids
  • interspecific hybridization
  • wine fermentation
  • Saccharomyces sensu stricto
  • molecular characterization


Wine fermentation is a complex ecological and biochemical process involving the sequential contribution of different yeast species (Fleet & Heard, 1993). Although many genera and species of yeasts are found in musts, those most commonly responsible for alcoholic fermentation are species of the genus Saccharomyces (Pretorius, 2000), including Saccharomyces cerevisiae and closely related taxa. These species were formerly included in the Saccharomyces complex (Vaughan-Martini & Martini, 1998), which was recently interpreted to represent the genus level (Kurtzman & Robnett, 2003).

Although S. cerevisiae is the predominant yeast in most fermentation processes, three other species of the genus have also been described to be involved in wine fermentations. Thus, Saccharomyces bayanus var. bayanus (according to Naumov, 2000) or simply S. bayanus (Nguyen and Gaillardin, 2005) has been isolated from beer, and S. bayanus var. uvarum (Naumov, 2000) or Saccharomyces uvarum (Pulvirenti, 2000; Nguyen and Gaillardin, 2005) has been found mainly associated with winemaking (Naumov, 2000b, 2002; Demuyter, 2004) and cider production (Naumov, 2001) at low temperature in colder areas of Europe; Saccharomyces paradoxus, considered to be a natural species with a fortuitous presence in wines, has recently been described as the predominant yeast in Croatian vineyards (Redzepovic, 2002); and Saccharomyces pastorianus (syn. Saccharomyces carlsbergensis), a partial allotetraploid hybrid (Pedersen, 1986a, b; Hansen & Kielland-Brandt, 1994; Tamai, 1998; Yamagishi & Ogata, 1999), is responsible for fermentation in the production of lager beer. The genus Saccharomyces also includes three other natural species, Saccharomyces cariocanus, Saccharomyces mikatae and Saccharomyces kudriavzevii, which were described (Naumov, 2000a) based on a few strains isolated from natural habitats, the former in Brazil and the latter two in Japan.

In general, industrial Saccharomyces strains are highly specialized organisms, which have evolved to their full potential in the different environments or ecological niches provided by human activity. This process can be described as “domestication” and is responsible for the peculiar genetic characteristics of the industrial yeasts, such as the high level of chromosome length polymorphism (Bidenne, 1992; Rachidi, 1999) and the common presence of aneuploidies or polyploidies (Codón & Benítez, 1995; Naumov, 2000c). In recent years, intensive research has focused on elucidating the molecular mechanisms involved in the adaptation of yeasts to industrial processes and how genomic characteristics of industrial yeast have been reshaped as they have been selected over thousands of years (Querol, 2003).

In the case of the genus Saccharomyces, one of the most interesting mechanisms observed in its adaptation to industrial processes is the formation of interspecific hybrids (de Barros Lopes, 2002). This is thought to be an important mechanism for enhancing genetic flexibility and promoting adaptive change (Greig, 2002), which also occurs in other yeast genera of clinical (Cryptococcus, Boekhout, 2001) or industrial interest (e.g. Zygosaccharomyces, James, 2005). The best described example of hybrid yeasts are lager yeasts, included in the taxon S. pastorianus (syn. S. carlsbergensis). This yeast is a partial allotetraploid hybrid between two species of the Saccharomyces sensu stricto group, S. cerevisiae and an S. bayanus-related yeast (Nguyen, 2000; Casaregola, 2001). Chromosome sets from both parental species are present in strains of S. pastorianus (Yamagishi & Ogata, 1999), while the mtDNA was inherited from the non-S. cerevisiae parent (Piškur, 1998).

Saccharomyces sensu stricto interspecific hybrids have been found in different fermentation processes. In addition to S. pastorianus, present in lager brewing, other hybrid strains have also been described from wine (S6U) and cider (CID1, a triple hybrid) (Masneuf, 1998; Groth, 1999; Naumova, 2005). The type strain of S. bayanus, originally isolated from beer, has recently been suggested to be a hybrid between S. cerevisiae and S. bayanus based on the presence of subtelomeric repeated sequences and genes (Nguyen, 2000; de Barros Lopes, 2002; Nguyen & Gaillardin, 2005). However, the presence of certain introgressive subtelomeric sequences is not necessarily an indication of a hybrid genome (Naumova, 2005). Other potential Saccharomyces hybrids isolated from both natural habitats and fermentation processes have been postulated on the basis of their complex electrophoretic karyotypes (Naumov, 2002) or their hybridization patterns with subtelomeric and transposable repetitive elements (Liti, 2005).

Although species of the genus Saccharomyces are difficult to distinguish on the basis of conventional morphology and assimilation tests (i.e. phenetic identification criteria), these strains can be unequivocally assigned to one of the four sibling species on the basis of genetic compatibility (Naumov, 1993) or molecular methods, such as karyotyping (Naumov, 1996; Vaughan-Martini, 1993); nDNA/nDNA reassociation experiments (Rosini, 1982; Vaughan-Martini & Kurtzman, 1985); hybridization to species-specific probes (Naumov, 1992); random amplified polymorphic DNA (RAPD) PCR (Fernández-Espinar, 2003); or sequencing (Belloch, 2000; Kurtzman & Robnett, 2003). However, strains assigned to formerly recognized species have occasionally been transferred to the present Saccharomyces species on the basis of a single molecular method, which makes it difficult to identify natural hybrids that could potentially be present in fermentation processes.

Hybrids were identified using different methods. DNA–DNA hybridization experiments demonstrated the hybrid nature of S. pastorianus (Vaughan-Martini & Kurtzman, 1985), which was later confirmed by the analysis of nuclear genes, for example MET2 (Hansen & Kielland-Brandt, 1994). However, this method failed to detect certain hybrids, such as the type strain of the former species Saccharomyces monacensis, now included in S. pastorianus. The combined analysis of nuclear (MET2) and mitochondrial (ATP8, ATP9 or SSU) genes proved to be very helpful in characterizing double and triple hybrids (Masneuf et al., 1998; Groth, 1999). The analysis of multilocus markers, such as amplified fragment length polymorphism (AFLP) (de Barros Lopes, 2002) or RAPD (Fernández-Espinar, 2003), enabled potential hybrids to be identified and discriminated from nonhybrid strains. Finally, PCR amplification and sequencing of several nuclear regions, used to characterize beer yeasts, was very effective in unveiling the complex introgression between Saccharomyces species (Casaregola, 2001).

The aim of the present study was to identify and characterize new natural hybrids present in wine fermentations. To do this, we genetically characterized different wine yeasts by combining the RFLP analysis of five different nuclear gene regions located in different chromosomes (i.e. CAT8, CYR1, GSY1, MET6 and OPY1), the 5.8S internal transcribed spacer (ITS) region and sequencing of the mitochondrial gene COX2.

Materials and methods

Yeast strains

Forty-five Saccharomyces strains were used in this study (Table 1). Forty-three of them were isolated from wine, some are used as commercial starters, and the hybrids CID1 (CBS 8614) and S6U (Lallemand, Montreal, Canada) correspond to reference hybrids.

View this table:

List of Saccharomyces strains used in the present study

Strain designationOriginal epithetIsolation sourceOrigin isolationNew characterization
CECT 1479S. cerevisiaeWineHungaryS. cerevisiae
CECT 1882S. cerevisiaeSherry wineHuelva (Spain)S. cerevisiae
CECT 1885S. pastorianusWineValladolid (Spain)S. bayanus×S. cerevisiae
CECT 11757S. cerevisiaeWineSpainS. cerevisiae
CECT 11761S. cerevisiaeWineSpainS. cerevisiae
CECT 12600S. bayanusSweet wineAlicante (Spain)S. bayanus
CECT 12627S. pastorianusWineValladolid (Spain)S. bayanus
CECT 12629Saccharomyces sp.MustZaragoza (Spain)S. bayanus
CECT 12638S. pastorianusMustCádiz (Spain)S. bayanus
CECT 12669S. pastorianusGrapesLa Rioja (Spain)S. bayanus
CECT 12681S. bayanusMustValladolid (Spain)S. cerevisiae
CECT 12682S. bayanusMustValladolid (Spain)S. cerevisiae
CECT 12686S. bayanusMustAlicante (Spain)S. cerevisiae
CECT 12687S. bayanusMustAlicante (Spain)S. cerevisiae
CECT 12688S. bayanusMustValencia (Spain)S. cerevisiae
CECT 12738S. bayanusMustMadrid (Spain)S. cerevisiae
CECT 12739S. bayanusMustToledo (Spain)S. cerevisiae
CECT 12740S. bayanusMustMadrid (Spain)S. cerevisiae
CECT 12741S. bayanusMustToledo (Spain)S. cerevisiae
CECT 12742S. bayanusMustToledo (Spain)S. cerevisiae
CECT 12743S. bayanusMustMadrid (Spain)S. cerevisiae
CECT 12744S. bayanusMustZaragoza (Spain)S. cerevisiae
CECT 12904S. bayanusMustToledo (Spain)S. cerevisiae
CECT 12905S. bayanusGrapesValladolid (Spain)S. cerevisiae
CECT 12922S. pastorianusWineValladolid (Spain)S. bayanus
CECT 12923S. pastorianusWineValladolid (Spain)S. cerevisiae
CBS 423S. cerevisiaeWineRiddes, SwitzerlandS. cerevisiae
CBS 1175S. cerevisiae or paradoxusWineMontibeux, SwitzerlandS. cerevisiae
CBS 1399S. cerevisiae or paradoxusWineValais, SwitzerlandS. cerevisiae
CBS 2834S. cerevisiae or paradoxusWineWädenswil, SwitzerlandS. bayanus×S. cerevisiae ×S. kudriavzevii
CBS 2897S. cerevisiae or paradoxusWineLausanne, SwitzerlandS. cerevisiae
CBS 2898S. bayanus or pastorianusWineWädenswil, SwitzerlandS. bayanus
CBS 2983S. cerevisiae or paradoxusWineWädenswil, SwitzerlandS. cerevisiae
CBS 2984S. cerevisiae or paradoxusWineWädenswil, SwitzerlandS. cerevisiae
CBS 2986S. bayanus or pastorianusWineWädenswil, SwitzerlandS. bayanus
W27S. cerevisiaeWineWädenswil, SwitzerlandS. cerevisiae×S. kudriavzevii
W46S. cerevisiaeWineWädenswil, SwitzerlandS. cerevisiae×S. kudriavzevii
SPG 14-91S. cerevisiaeWineWädenswil, SwitzerlandS. cerevisiae×S. kudriavzevii
SPG 16-91S. cerevisiaeWineWädenswil, SwitzerlandS. cerevisiae×S. kudriavzevii
126S. cerevisiaeWineWädenswil, SwitzerlandS. cerevisiae×S. kudriavzevii
172S. cerevisiaeWineWädenswil, SwitzerlandS. cerevisiae×S. kudriavzevii
319S. cerevisiaeWineWädenswil, SwitzerlandS. cerevisiae×S. kudriavzevii
441S. cerevisiaeWineWädenswil, SwitzerlandS. cerevisiae×S. kudriavzevii
S6US. bayanus×S. cerevisiaeWineItalyS. bayanus×S. cerevisiae
CID1S. bayanus×S. cerevisiae×S. kudriavzeviiCiderBrittany, FranceS. bayanus×S. cerevisiae×S. kudriavzevii
  • * Commercialized by Lallemand Inc., Montreal, Canada.

  • CECT, Colección Española de Cultivos Tipo, Valencia, Spain; CBS, Centraalbureau voor Schimmelcultures, Utrecht, the Netherlands.

PCR amplification and DNA sequencing reactions

DNA was isolated according to Querol (1992). Oligonucleotide primers for PCR amplification of gene regions are listed in Table 2. PCR was performed in a mixture containing 1 × Taq polymerase buffer (BioTools, B&M Labs S.A., Madrid, Spain), 100 μM deoxynucleotides, 1 μM of each primer and 2 U of Taq polymerase (BioTools, B&M Labs S.A.). A volume of 4 μL of DNA, diluted to 1–50 ng μL, was transferred to a PCR tube before adding the reaction mixture, to a final volume of 100 μL. PCR amplifications were carried out in a Techgene thermocycler (Techne, Cambridge, UK). PCR amplification was performed as follows: initial denaturing at 95°C for 5 min, then 40 PCR cycles of three steps (denaturing at 94°C for 1 min, annealing at 55.5°C for 2 min and extension at 72°C for 2 min), followed by final extension at 72°C for 10 min.

View this table:

PCR primers designed in the present study to amplify five nuclear gene regions


PCR products were separated on a 1.4% agarose (Pronadisa, Laboratorios Conda S.A., Madrid, Spain) gel in 0.5 × TBE (4.5 mM Tris-borate, 1 mM EDTA, pH 8) buffer. After electrophoresis, gels were stained with ethidium bromide (0.5 μg mL) (AppliChem, Darmstadt, Germany) and visualized under UV light. A 100-bp DNA ladder marker (Roche Molecular Biochemicals, Mannheim, Germany) served as the size standard.

The 5.8S-ITS region was amplified with the general primers its1 (5′-TCCGTAGGTGAACCTGCGG) and its4 (5′-TCCTCCGCTTATTGATATGC), described elsewhere (White, 1990). Primers used to amplify other nuclear gene regions (CAT8, CYR1, GSY1, MET6 and OPY1) were designed by comparing the corresponding genome sequences of the reference strains of the species S. bayanus var. uvarum (MCYC 623=CBS 7001), S. cerevisiae (S288C), S. kudriavzevii (IFO 1802T=CBS 8840T), S. mikatae (IFO 1815T=CBS 8839T) and S. paradoxus (CBS 432NT), available in the budding yeasts genome comparison (http://db.yeastgenome.org/cgi-bin/FUNGI/FungiMap) and fungal sequence alignment (http://db.yeastgenome.org/cgi-bin/FUNGI/showAlign) sections of the Saccharomyces Genome Database (SGD). These sequences were aligned with the CLUSTAL × program (a Windows version of the Clustal W program, Thompson, 1994) and analysed with the GeneDoc program (Nicholas, 1997). Saccharomyces cariocanus was not included because it is the only species whose complete genome has not yet been sequenced. Primers are universal for the amplification of each gene of S. bayanus, S. cerevisiae, S. kudriavzevii, S. mikatae and S. paradoxus, except in the case of MET6-3K, which is a specific reverse primer for MET6 from S. kudriavzevii, owing to difficulties in amplifying this gene region with the general reverse primer.

Restriction analysis

Restriction digestions of the PCR products were carried out with 15 μL of amplified DNA in a final volume of 20 μL. The restriction endonucleases used were: AccI, HaeIII and ScrFI to digest the 5.8S-ITS amplified region; CfoI and MspI for CAT8; HaeIII and MspI for CYR1; MspI for GSY1; HaeIII, HinfI and ScrFI for MET6; and ScrfI for OPY1 PCR products (all from Roche Molecular Biochemicals) following the supplier's instructions. Restriction fragments were separated on 3% agarose (Pronadisa, Laboratorios Conda S.A.) gels in 0.5 × TBE buffer. A mixture of 50- and 100-bp DNA ladder markers (Roche Molecular Biochemicals) was used for size standards.

Amplification, sequencing and phylogenetic analysis of MET6 and COX2 genes

The nuclear MET6 and the mitochondrial COX2 genes were sequenced for those strains identified as hybrids. The mitochondrial gene was analysed through sequencing, given the absence of diagnostic restriction sites that allow easy differentiation of some Saccharomyces sensu stricto species.

COX2 was amplified by PCR, using the primers and conditions described in Belloch (2000). MET6 was amplified under the conditions described above; however, the S. bayanus-like and S. cerevisiae-like alleles were amplified with primers MET6-5 and MET6-3, and the S. kudriavzevii-like alleles were specifically amplified with primers MET6-5 and MET6-3K. In the case of hybrid strains S6U, CBS2834 and CID1, which have S. bayanus-like and S. cerevisiae-like MET6 alleles, the PCR products amplified with primers MET6-5 and MET6-3 contained both alleles. Although the nucleotide sequence of each allele can be deduced easily from sequencing the PCR product directly, this product was cloned with the pGEM®-T Easy vector system II (Promega, Madison, WI). Several clones were then sequenced to confirm the nucleotide sequence of each allele.

MET6 and COX2 PCR products were cleaned with the Perfectprep® Gel Cleanup kit (Eppendorf, Hamburg, Germany) and both strands of the DNA were directly sequenced using the BigDye Terminator V3.0 Cycle Sequencing Kit (Applied Biosystems, Warrington, UK), following the manufacturer's instructions, in an Applied Biosystems automatic DNA sequencer Model ABI 3730 × l (Applied Biosystems).

Sequences from the hybrids characterized in this work were deposited in the EMBL Sequence Database under accession numbers AJ973286AJ973296, AJ973304AJ973316 and AJ973299 for MET6 sequences, and AJ938037–AJ938048 for COX2 genes. MET6 sequences from the reference or type strains of S. bayanus var. uvarum (MCYC 623), S. cerevisiae (S288C), S. kudriavzevii (IFO 1802T), S. mikatae (IFO 1815T) and S. paradoxus (CBS 432NT) were retrieved from the fungal sequence alignment section (http://db.yeastgenome.org/cgi-bin/FUNGI/showAlignfrom) of the SGD, with the exception of MET6 from the type strain of S. pastorianus (CBS 1538NT), which we sequenced (accession number AJ973300). Other COX2 sequences from reference and type strains were retrieved from the sequence databases, where they are accessible under the following numbers: AF442210 and AY130329 for S. kudriavzevii strains IFO 1802T and IFO 1803, respectively; AF442207 and AY130331 for S. cariocanus UFRJ 50816T and UFRJ 50791, respectively; AF442209 and AY130330 for S. mikatae IFO 1815T and IFO 1816, respectively; AF442208 for S. paradoxus CBS 432NT; AY244992 for S. cerevisiae CBS 1171NT; AF442211 for S. bayanus CBS 380T; and AF442212 for S. pastorianus CBS 1538NT.

Each set of homologous sequences was aligned with the Clustal × program (Thompson, 1994). The sequence evolution model that fits our sequence data best was optimized using the hierarchical model comparison by likelihood ratio tests, implemented in Modeltest version 3.6 (Posada & Crandall, 1998). The best fitting model for MET6 sequences was the HKY model (Hasegawa, 1985) with a gamma distribution (G) of substitution rates with a shape parameter α=0.2452, and for COX2 gene sequences it was the general time reversible model (GTR; Tavaré, 1986) with a gamma distribution (G) of substitution rates with a shape parameter α=0.1347% and 81.74% of invariable sites (I). The parameters of each model, estimated in the previous analysis, were used to obtain the best trees under the optimality criterion of maximum likelihood (ML). Tree reliability was assessed using nonparametric bootstrap resampling of 100 replicates. All these phylogenetic analyses were performed using PAUP* version 4.0b10 (Swofford, 2002).

Pulsed-field electrophoresis

DNA for electrophoretic karyotyping was included in agarose plugs as described by Carle & Olson (1985). Chromosomal profiles were determined by the CHEF technique using a BioRad DRIII system (Bio-Rad Laboratories, Hercules, CA), using S. cerevisiae (strain YNN295) chromosomes (Bio-Rad Laboratories) as standard marker. Yeast chromosomes were separated on 1% agarose gels in two steps as follows: 60 s of pulse time for 14 h, and then 120 s of pulse time for 10 h, both at 6 V cm−1 with an angle of 120°. Running buffer used was 0.5 × TBE (45 mM Tris-borate, 1 mM EDTA) cooled at 14°C.


A procedure to differentiate the species of the genus Saccharomyces and their hybrids based on the restriction analysis of different gene regions

Given the advantages afforded by a combined analysis of different gene regions in order to identify and characterize natural hybrids, we have developed a simple method based on PCR amplification and restriction analysis of five nuclear gene regions (CAT8, CYR1, GSY1, MET6 and OPY1, located in chromosomes XIII, X, VI, V and II, respectively) and the 5.8S-ITS ribosomal region (located in chromosome XII). These gene regions were selected, from the complete sequences available of the genome of representative and type strains of the Saccharomyces sensu stricto species indicated in Table 3, according to three criteria: (1) their location in different chromosomes to avoid linkage, (2) high interspecific variability to obtain diagnostic restriction patterns for each species and (3) the presence of conserved flanking regions where general primers could be designed to amplify those regions from any strain of the Saccharomyces sensu stricto complex.

View this table:

Composite restriction patterns deduced from the gene region sequences of the reference and type strains of the Saccharomycessensu stricto species for which the complete genome sequence is available

Gene regionRestriction enzymesMCYC 623 (S. bayanus)S288C (S. cerevisiae)IFO 1802T (S. kudriavzevii)CBS 432NT (S. paradoxus)IFO 1815T (S. mikatae)Alternative patterns
5.8S-ITSAccI850850730 110850850
HaeIII500 230 125325 230 170 125500 230 125325 230 170 125500 230 125
ScrFI400 320 120400 320 120400 320 120520 320520 320
CAT8CfoI810810560 175810390 245 100810
MspI280 250 200690 50450 205 80810450 205 80360 280 90
CYR1HaeIII560560295 155 80 30365 165380 130 50295 155 80 30
MspI560400 170560400 170490 70460 100
GSY1MspI380 340 50610 160340 270 160430 340770
MET6HaeIII480 200680680635 50680
HinfI625 60450 160 60625 60625 60625 60
ScrFI680680680680615 70
OPY1HaeIII405 345750505 245505 245345 195 160 50750
ScrFI565 185450 300750450 300490 180 80315 250 185
  • These composite patterns for each gene region have been named after the initial of the specific name followed by the order numeral 1. Alternative patterns found in hybrids are also described and named using the same criterion but followed by the subsequent order numeral. These alternative patterns differ from those found in the reference strains by one or two restriction gains/losses.

Table 3 shows the diagnostic restriction patterns of the six nuclear PCR regions for the reference and type strains of the Saccharomyces s.s. species deduced from their genome sequence. The combined profiles (name based on the initial of the species name followed by 1) of restriction patterns, given by digestion of the amplified gene regions with two or three endonucleases, enabled them to be unequivocally assigned to the nonhybrid species of the genus Saccharomyces, with the exception of S. cariocanus, because its complete genome sequence is not available. All these patterns were confirmed by PCR amplification using DNA from the type and reference strains, and subsequent restriction analysis with the appropriate endonucleases.

Characterization of wine strains based on restriction analysis of the different nuclear gene regions

Following the procedures and using the primers described before, we analysed the 43 wine strains and two reference hybrid strains (S6U and CID1; Masneuf, 1998) listed in Table 1. These strains were isolated from different wine regions in Spain, mainly characterized by a Mediterranean climate, and in Switzerland, characterized by a continental climate.

The restriction patterns of the six gene regions confirmed that most strains had been incorrectly identified originally, especially those included in the taxa S. bayanus and S. pastorianus. The new classification of the strains according to the present study is given in Table 1.

From this analysis, we confirmed the predominance of S. cerevisiae among the wine strains isolated in Spain, although the cryophilic S. bayanus has been isolated even in the southern wine regions of Alicante and Cádiz, where spontaneous wine fermentations can reach temperatures above 30°C. Moreover, it appeared that hybrid strains were rare in Spanish wines. Saccharomyces pastorianus strain CECT 1885 is the only strain that was shown to be a true S. bayanus×S. cerevisiae hybrid. The other five strains originally identified as S. pastorianus correspond to either S. bayanus (CECT 12627, 12638, 12669 and 12922) or to S. cerevisiae (CECT 12923).

Most Swiss strains (12) correspond to isolates from the Wädenswil wine region of eastern Switzerland, although a few of them were isolated from the Valais (three strains) and Vaud (one strain) wine regions of south-west Switzerland. The characterization of these strains (Table 1) gave contrasting results: the four strains from the neighbouring wine areas of Valais and Vaud correspond to S. cerevisiae, whereas a novel hybrid strain predominates in the Wädenswil region. Eight wine strains from Wädenswil, including the commercial strains W27 and W46, exhibited a mixture of restriction patterns for the protein-coding gene regions, due to the presence of two different copies of each region (Table 4). One copy of each gene exhibits the typical restriction pattern of S. cerevisiae (C1, Table 3) and the other the same restriction pattern as for S. kudriavzevii (K1), with the exception of CYR1, for which the second copy shows a restriction pattern differing from that of S. kudriavzevii in one single restriction site (Table 3). Three of these strains (W27, W46 and 441) contain one single 5.8S-ITS region type with an identical pattern to that of S. kudriavzevii; the other five (SPG14-91, SPG16-91, 126, 172 and 319) exhibited the S. kudriavzevii 5.8S-ITS pattern and additional faint bands corresponding to S. cerevisiae (Table 4).

View this table:

Conformation of the hybrids for each gene region according to the composite restriction patterns exhibited

Hybrid strainsNuclear genesmtDNA COX2
S. cerevisiae × S. kudriavzevii
W27 and 441K1C1K1C1K2C1K1C1K1C1K1K2
SPG 14-91, SPG 16-91 126. 172 and 319C1K1C1K1C1K2C1K1C1K1C1K1K2
S. bayanus × S. cerevisiae
CECT 1885B1C1B2C1B1B1B1C1B2B2
S. bayanus × S. cerevisiae × S. kudriavzevii
CBS 2834B1C1B1K1C1B1K2B1K1C1B1K1C1B1K1K4
  • For a description of the composite restriction patterns, see Table 3.

  • S., Saccharomyces.

Sequences of the 5.8S-ITS region from some of these strains (W27, W46, SPG14-91 and SPG16-91) had already been obtained at the Swiss Federal Research Station in Wädenswil and are available in the sequence databases under accession numbers AJ295627, AJ295628, AJ295632 and AJ295630, respectively. These 5.8S-ITS sequences display very high similarity levels (>99.3%) with the S. kudriavzevii type strain, forming a significant group [bootstrap value (BV) 93%] in the phylogenetic analysis of these sequences, as well as with those belonging to the type strains of the Saccharomyces species (results not shown).

However, the most striking results were obtained with wine strain CBS 2834 (Table 4), also isolated from Wädenswil. This strain contained three copies of gene regions CAT8, CYR1, MET6 and OPY1, one with the S. cerevisiae restriction pattern (C1), another with the S. bayanus pattern (B1) and the third with the S. kudriavzevii (K1), with the exception of CYR1, which exhibited the alternative S. kudriavzevii pattern (K2) found in the other Swiss hybrid strains. In the case of GSY1, strain CBS 2834 only contained two copies, one from S. bayanus and the other from S. kudriavzevii. Finally, the 5.8S-ITS region only exhibited an S. bayanus restriction pattern.

This is the first time that hybrids from S. kudriavzevii have been found among wine strains. The only clear precedent from another fermentation process is the cider strain CID1 (Masneuf, 1998), a hybrid containing mtDNA from S. kudriavzevii (Groth, 1999). The nuclear genome of this strain was characterized by only one single nuclear gene (MET2). For this gene, CID1 contains two different copies, one from S. bayanus and the other from S. cerevisiae, but none from S. kudriavzevii. A subsequent AFLP analysis by de Barros Lopes (2002) showed that not only did S. bayanus and S. cerevisiae contribute to the genome of CID1, but so too did a third species, presumably S. kudriavzevii. However, Naumova (2005) recently demonstrated that CID1 is a triploid strain containing three copies of the ACT1 gene, each one similar to the genes from S. cerevisiae, S. bayanus var. uvarum and S. kudriavzevii. These results clearly support the triple hybrid nature of this strain. To compare these two triple hybrids isolated from different sources and origins, we performed the same combined restriction analysis with CID1 (Table 1). This analysis enabled us to confirm the triple hybrid nature of CID1, and also corroborated the contribution of the parental species S. kudriavzevii to the nuclear genome of CID1. This strain contained three copies of the protein-coding gene regions under analysis, coming from the three parental species, and the 5.8S-ITS region of S. bayanus. Therefore, the difference between CID1 and CBS 2834 was the presence of an S. cerevisiae copy of GSY1 in the former, which is absent in the latter.

Direct sequencing of some of the PCR products confirmed the presence of two or three gene copies, coming from different species, in the hybrid strains. The corresponding chromatograms showed that hybrids contain ambiguous nucleotide sequences in those sites that are variable among the species. At these nucleotide sites, hybrids exhibited more than one peak (up to three in the triple hybrids CBS 2834 and CID1), corresponding to those differential nucleotides present in the parental species.

Phylogenetic analysis of MET6 gene sequences from hybrids

In the case of MET6, the hybrid nature of the wine strains was determined by PCR amplification with two sets of primers. One set, MET6-5 and MET6-3K, was used specifically to amplify the gene copy coming from S. kudriavzevii, and the other, MET6-5 and MET6-3K, to amplify the gene copies coming either from S. bayanus or from S. cerevisiae, which were differentiated by restriction analysis.

Direct sequencing of the PCR products amplified with each primer set, and subsequent phylogenetic analysis of these sequences, allowed us to confirm the hybrid nature of the wine strains. The only exceptions were the double hybrid S6U and the triple hybrids CBS 2834 and CID1, which contain MET6 copies coming from both S. bayanus and S. cerevisiae (Table 4). In those cases, PCR product cloning and sequencing were used to confirm the nucleotide sequence of each MET6 gene copy.

The phylogenetic tree of the MET6 sequences from hybrids and parental species, depicted in Fig. 1, was obtained by the ML method according to the best suited model of evolution for these sequences (see Materials and methods). This tree confirmed the hybrid nature of wine strains deduced from the restriction analysis of different gene regions. Double hybrids contained two divergent MET6 alleles, one (designated with a C after the strain reference) identical or similar to the S. cerevisiae S288c MET6 gene and the other clustering either with the MET6 gene from S. bayanus var. uvarum MCYC623 (indicated with a B) or with that from S. kudriavzevii IFO 1802T (denoted with a K). The only exception was strain CECT 1885, a hybrid that contained only one single MET6 gene similar to that of S. bayanus var. uvarum MCYC 623. Finally, triple hybrids CBS 2834 and CID1 possessed three copies of this gene (B, C and K alleles), each one similar to the MET6 gene from the representatives of each parental species.


Phylogenetic tree obtained with partial sequences of the nuclear MET6 gene from hybrid strains and Saccharomyces sensu stricto type and reference strains. Hybrid strains (indicated in bold) contain one, two or three different MET6 alleles, named C (cerevisiae), B (bayanus) or K (kudriavzevii) according to the closest parental relative. The tree was obtained under the optimality criterion of maximum likelihood according to the HKY model of nucleotide substitution (Hasegawa, 1985), with a gamma distribution of substitution rates with a shape parameter α=0.245. The best fitting model of evolution for these sequences was estimated by the likelihood ratio test (Posada & Crandall, 1998). Numbers at nodes correspond to bootstrap values based on 100 pseudoreplicates. The scale is given in nucleotide substitutions per site.

All MET6K alleles present in hybrids were identical and differed only in two nucleotide substitutions with respect to the S. kudriavzevii type strain MET6 gene. In the case of MET6 C alleles, two different allele types were found in the hybrids. One was identical to the MET6 gene from S. cerevisiae S288c and was present in the S. bayanus×S. cerevisiae hybrid S6U and in the triple hybrids CBS 2834 and CID1. The other differed in one single nucleotide substitution and was present in all S. cerevisiae×S. kudriavzevii hybrid wine strains. Two different “bayanusMET6 allele types were also found among hybrids, but they are highly divergent. One allele was identical to the MET6 gene from S. bayanus var. uvarum MCYC623 and the S. pastorianus neotype strain CBS 1538NT, and was present in the double hybrid S6U and the triple hybrids CBS 2834 and CID1. The other allele differed in nucleotide substitution and corresponded to the only MET6 gene found in the S. bayanus×S. cerevisiae hybrid strain CECT 1885.

Comparison of mitochondrial gene COX2 sequences from hybrids

Another interesting question was the origin of the mitochondria harboured by the interspecific hybrid strains. For this purpose, we analysed the mitochondrial gene COX2 from the hybrid wine strains. This is a highly variable gene that proved to be very informative in determining the phylogenetic relationships among species of the SaccharomycesKluyveromyces complex (Belloch, 2000; Kurtzman & Robnett, 2003). However, owing to difficulties in unveiling COX2 variability in the Saccharomyces species by restriction analysis, we performed direct sequencing analysis.

In all cases, hybrids contained a single COX2 gene (homoplasmy) coming from only one of the parental species involved in the formation of hybrids (see Table 4). This is compatible with the inheritance mtDNA observed in S. cerevisiae (Berger & Yaffe, 2000).

However, two different patterns were observed in hybrids with respect to the mitochondrial donor species. In the eight double and two triple hybrids containing the S. kudriavzevii nuclear genome, their mitochondrial COX2 sequences were always similar to, but different from, the gene sequence from this parental species (type K1). These hybrids contain four COX2 sequence types (Table 4). The most similar sequence to that from the S. kudriavzevii type strain is type K4, which differed by five nucleotide substitutions in 585 bp. Type K2, which differs from K4 in one single substitution, was the most frequent sequence observed and appeared in strains W27, SPG14-91, SPG16-91, 126, 172, 319 and 441. Type K3 differed from K2 in an additional substitution and corresponded to strain W46. Finally, type K5 was present in the cider triple hybrid CID1, and corresponded to the most divergent COX2 sequence within these groups of hybrids.

By contrast, each S. bayanus×S. cerevisiae hybrid contained a mitochondrial COX2 gene coming from a different parental species. In this way, the S6U COX2 sequence was found to be similar to, but again different from, that of S. cerevisiae, and the mitochondrial sequence from CECT 1885 was closely similar to that of S. bayanus.

To determine the phylogenetic relationships among COX2 sequences from hybrids and parental species, we obtained the ML tree depicted in Fig. 2, according to the best suited model of evolution for these sequences (see Materials and methods). This confirmed the origin of the mitochondrial gene in the hybrids. Saccharomyces cerevisiae×S. kudriavzevii hybrids and the wine triple hybrid CBS 2834 form a clear group (BV 94%), which clustered with S. kudriavzevii strains IFO 1802T (BV 96%) and IFO 1803 (BV 93%). However, CID1 clustered only loosely with them (BV 63%) and was the most divergent of the S. kudriavzevii-related sequences.


Phylogenetic tree of the COX2 sequences from hybrid strains, indicated in bold, and Saccharomyces type and reference strains. The tree was obtained under the optimality criterion of maximum likelihood according to the general time reversible model (44) with a gamma distribution of substitution rates with a shape parameter α=0.1347% and 81.74% of invariable sites. The best fitting model of evolution for these sequences was estimated by the likelihood ratio test (Posada & Crandall, 1998). Numbers at nodes correspond to bootstrap values based on 100 pseudoreplicates. The scale is given in nucleotide substitutions per site.

Something similar occurs with the COX2 sequence from S6U. It appeared within a group formed by COX2 sequences from the type strains of S. cerevisiae and related species, S. paradoxus, S. mikatae and S. cariocanus, but their relationships are not well supported (BV <50%).

In the case of CECT 1885, the COX2 sequence was identical to that of the type strains of S. pastorianus, a S. bayanus×S. cerevisiae hybrid originally isolated from beer (Vaughan-Martini & Martini, 1998). These two sequences grouped, with a bootstrap value of 100%, with the COX2 sequence from the type strain of S. bayanus.

Electrophoretic karyotyping of hybrid wine strains

Electrophoretic karyotyping has been widely used to characterize chromosome variability among Saccharomyces strains (Naumov, 1993; Schütz & Gafner, 1994). This technique was successfully applied to unveil interspecific chromosomal rearrangements in the genus Saccharomyces (Ryu, 1996; de Barros Lopes, 2002) and to characterize hybrid lager yeast strains (Yamagishi & Ogata, 1999). The presence of chromosome differences between S. cerevisiae and S. bayanus, resulting from translocation rearrangements, enabled S. cerevisiae×S. bayanus hybrids to be identified easily among lager yeast strains (Tamai, 1998).

We performed pulsed-field electrophoresis with the hybrid wine strains and the reference, double (S6U) and triple (CID1) hybrids to characterize their chromosomal differences. Reference and type strains, belonging to the three parental species involved in the formation of the hybrids, were also included in the analysis. As an example, Fig. 3 shows the electrophoretic karyotyping of some hybrids and their parental species. Differences in the number and intensity of the bands suggest an alloploid nature of the hybrid strains.


An example of the chromosomal profiles exhibited by some hybrid strains and representatives of their parental Saccharomyces species: S. bayanus var. uvarum CECT 1969, S. cerevisiae FY 1679 and S. kudriavzevii IFO 1802T. Commercial wine strains W27 and W46 are S. cerevisiae×S. kudriavzevii hybrids; wine strains S6U and CECT 1885 correspond to S. bayanus×S. cerevisiae hybrids; and cider strain CID1 is a triple hybrid, S. bayanus×S. cerevisiae×S. kudriavzevii. Saccharomyces bayanus diagnostic bands present in hybrids are indicated by white arrowheads. Lane m corresponds to the standard marker strain S. cerevisiae YNN295 (BioRad); chromosomal numbers corresponding to each band are indicated on the left.

Complete sequencing of the genomes from S. cerevisiae and S. bayanus revealed that they are not syntenic and differ by four reciprocal and one nonreciprocal major translocations and six subtelomeric reciprocal translocations (Kellis, 2003). Of these, translocations between chromosomes II and IV and V and VII, relative to the S. cerevisiae standard chromosome arrangement, generated clear chromosomal differences in the electrophoretic karyotypes of S. cerevisiae and S. bayanus that can be used as diagnostic characters (Fig. 3, lanes 1–3).

However, in S. cerevisiae and S. kudriavzevii, their genomes are mostly syntenic, i.e. colineal with the exception of a few subtelomeric translocations with no clear effect on their electrophoretic karyotypes. In Figure 3 (lanes 1, 2 and 4), two strains of S. cerevisiae only differed with the S. kudriavzevii type strain with respect to the third chromosomal band, a doublet in S. cerevisiae formed by chromosomes VII and XV, which were solved in two bands in the case of S. kudriavzevii IFO 1802T. Other differences appeared in the mobility of the small chromosomes, but they were similar to those differences observed among S. cerevisiae strains, and hence, they cannot be used as diagnostic characters.

The electrophoretic karyotyping analyses showed that the S. bayanus diagnostic chromosomal bands (indicated by arrows in Figure 3) were clearly present in the patterns of wine strains S6U and CECT 1885, revealed as S. bayanus×S. cerevisiae hybrids, but also in those of the triple hybrid strains CBS 2834 and CID1, confirming the contribution of this species to the generation of such hybrids. All these hybrid strains exhibit complex chromosomal patterns with more than 20 bands, differing from each other in both the mobility and the intensity of the bands.

In the case of wine strains identified as S. cerevisiae×S. kudriavzevii, their chromosomal patterns were identical, with no clear differences in the mobility of the bands, and more similar to the pattern exhibited by S. cerevisiae strains than to that of the S. kudriavzevii type strain, because their chromosomes VII and XV form a single band. This high similarity between the S. cerevisiae karyotype and those of S. cerevisiae×S. kudriavzevii hybrids is probably why these hybrids were misidentified as S. cerevisiae, and hence, were not previously detected (Schütz & Gafner, 1994).


Artificial interspecific hybridization experiments (Rainieri, 1998; Marinoni, 1999; de Barros Lopes, 2002; Sato, 2002; Antunovics, 2005), which have also been performed extensively to delimitate species of the genus Saccharomyces according to the biological species concept (reviewed in Naumov, 1996), indicated that Saccharomyces interspecific hybrids containing the genome of both parents can easily be formed, and that, although sterile, they are viable and can be maintained by asexual reproduction.

Owing to the fact that reproductive isolation in the genus Saccharomyces is postzygotic, interspecific hybridization should also be possible in natural habitats. Thus, the first natural Saccharomyces interspecific hybrid was identified by cytoduction (Nilsson-Tillgren, 1983) and by DNA–DNA hybridization (Vaughan-Martini and Kurtzman, 1985). These studies suggested that the type strains of the lager brewing yeast S. carlsbergensis (now included in S. pastorianus) could be a natural interspecific hybrid between S. cerevisiae and another species related to S. bayanus. Since then, subsequent studies have demonstrated the hybrid nature of different lager brewing yeasts (Pedersen, 1986a, b; Hansen & Kielland-Brandt, 1994; Tamai, 1998; Casaregola, 2001).

The identification of S. bayanus×S. cerevisiae hybrids was initially limited to lager brewing strains, indicating that perhaps this interspecific hybridization could be a rare event that occurred in this particular brewing environment. This hypothesis is supported by the fact that most natural Saccharomyces strains are homothallic and their asci persistent, which promotes autodiploidization, and hence prevents gene exchange both within the species (Mortimer, 1994) and between species.

However, new natural hybrids were found in ambits other than brewing environments. Thus, Masneuf (1998) characterized a S. bayanus×S. cerevisiae hybrid strain (S6U) isolated from Italian wine, and a striking hybrid present in a home-made French cider (CID1). This latter hybrid was found to contain two copies of the nuclear gene MET2, one originating from S. cerevisiae and the other from S. bayanus, and the mitochondrial genome originating from a third species, which was shown by Groth (1999) to correspond to the type strain of S. kudriavzevii. This was the first report indicating that a rare Saccharomyces species was involved in interspecific hybridization. From this species only two strains, isolated from tree exudates in Japan, were known (Naumov, 2000a).

In the present study, we performed a multilocus genetic analysis of wine strains to determine the incidence of hybrids. The procedure, based on RFLP analysis of five nuclear genes from different chromosomes and the ribosomal region 5.8S-ITS, allows easy identification of species of the genus Saccharomyces and their hybrids.

In our survey, the presence of S. bayanus×S. cerevisiae hybrids in wines turned out to be very limited and sporadic. Only one strain, CECT 1885, corresponded to a true S. bayanus×S. cerevisiae hybrid. However, this hybrid exhibited a very different genome structure as compared with the Italian hybrid S6U. This was also confirmed by phylogenetic analyses of the nuclear MET6 and the mitochondrial COX2 genes.

This method was used, for the first time, to characterize a new type of wine hybrid, resulting from hybridization between S. cerevisiae and S. kudriavzevii. These hybrids appeared to be very common in Wädenswil, eastern Switzerland, given that they have been isolated during different vintages from 1991 to 2000. In fact, different studies performed at the Research Station in Wädenswil have demonstrated that some of the hybrid strains analysed herein (SPG14-91 and SPG16-91), and other strains closely related to them and to W27, were predominant in spontaneous wine fermentations during two consecutive vintages (Schütz & Gafner, 1994). At that time these strains were considered to be S. cerevisiae based on their chromosome profiles. Moreover, two of these hybrids (W27 and W46) were selected at the Research Station in Wädenswil as the strains best suited for winemaking under the typical fermentation conditions of central Europe (i.e. low temperatures and low sugar content), and with the ability to produce fruity aromas. In fact, these two strains are currently commercialized as dry yeasts by Lallemand Inc. (Montreal, Canada).

In the present study, we have also found an S. bayanus×S. cerevisiae×S. kudriavzevii hybrid strain. Curiously, this wine strain (CBS 2834) was also isolated in 1951 by Torsten O. Wikén in Wädenswil, Switzerland, and at present is the oldest known strain containing DNA contributed by S. kudriavzevii. Our results indicate that the genomic structure of this strain differs from that of the other triple hybrid known thus far, CID1, which also contains nuclear DNA contributed by S. kudriavzevii. This was postulated by de Barros Lopes (2002) according to AFLP fingerprinting analysis, demonstrated by Naumova (2005) based on ACT1 gene sequencing, and corroborated in the present study with five additional genes.

Phylogenetic analyses of the nuclear MET6 and the mitochondrial COX2 genes from the hybrid strains corroborate their different origins. Although the three different MET6 alleles are identical in both triple hybrids, the mitochondrial COX2 gene from the wine triple hybrid is closely related to that of the other wine S. cerevisiae×S. kudriavzevii hybrids, and more closely related to the COX2 gene of the S. kudriavzevii type strain, than that of the cider triple hybrid. A similar conclusion can be drawn for the two S. bayanus×S. cerevisiae hybrids, as they possess a different nuclear genome structure according to the restriction analysis of nuclear genes, and also carry different MET6 alleles and mtDNA of different origin.

The diversity of Saccharomyces hybrids, together with their distinct origins and their presence in different habitats, suggests that interspecific hybridization is as rare event as previously supposed. Despite the homothallic character of most natural Saccharomyces strains and the persistence of their asci, S. bayanus and S. cerevisiae strains coexist during winemaking, brewing, cider production, etc., and hence their hybrids could be generated in these environments by rare-mating events between diploid strains, as proposed for S. pastorianus and other lager brewing hybrids (de Barros Lopes, 2002). Another possibility may be that conjugation of haploid cells, mediated by yeast-feeding invertebrates (Pulvirenti, 2000), may occur rarely in nature, and is selected in these artificial environments.

However, in the case of S. kudriavzevii hybrids, the hybridization event could possibly have occurred under natural conditions because this parental species has not so far been found in fermentation processes. In fact, the four known S. kudriavzevii strains were all isolated from natural habitats (i.e. three on decayed leaves and one from soil) in Japan.

Some of the physiological properties exhibited by S. kudriavzevii (Hittinger, 2004), which have been associated with adaptation to natural habitats under conditions that are either unsuitable or less suitable for other Saccharomyces sensu stricto species, could perhaps explain why the hybrids are better adapted to specific fermentation conditions. Hybrid yeasts acquire physiological properties from both parents, for example the alcohol and glucose tolerance of S. cerevisiae and the low-temperature tolerance of S. kudriavzevii, which confer upon them selective advantages in intermediate or fluctuating conditions with respect to their parental species (Zambonelli, 1997; Masneuf, 1998; Greig, 2002; Sato, 2002; Serra, 2005).

The present study confirms that natural hybridization among yeasts is more common than suspected previously and, hence, it should be considered as a potentially important mechanism in the evolution of yeasts, not only in the genus Saccharomyces, but also in other genera (e.g. Boekhout, 2001; James, 2005). Such a mechanism would provide new gene combinations of adaptive value (Masneuf, 1998; Greig, 2002), new or specialized functions from the divergence of redundant genes (Wolfe & Shields, 1997), or even result in new species through allopolyploid (Naumov, 2000c) or homoploid (Greig, 2002) speciation.


This work was supported by CICYT grants (ref. BIO2003-03793-CO3-01 and 02) from the Spanish Ministerio de Ciencia y Tecnología and by Conselleria de Cultura Educació I Esport, Generalitat Valenciana (Grupos03/012) to A.Q. and E.B.


  • Editor: Teun Boekhout


View Abstract