Correspondence: Matthias Sipiczki, Department of Genetics and Applied Microbiology, University of Debrecen, PO Box 56, 4010 Debrecen, Hungary. Tel.: +36 52 518 600; fax: +36 52 533 690; e-mail: firstname.lastname@example.org
Yeasts identified as Candida stellata are frequently associated with overripe and botrytized grapes and can survive in the fermenting must until the completion of vinification. The molecular taxonomic examination of 41 strains deposited in six culture collections or described in the literature as C. stellata revealed that most of those isolated from grapes or wines belonged to Candida zemplinina and related species. This confusion around the taxonomic position of the strains may account for the rather controversial descriptions of the oenological properties of C. stellata in the literature. Because the authors did not find it among strains newly isolated from botrytized grapes and wines, it was proposed that it is usually C. zemplinina rather than C. stellata that occurs on grapes and in wine fermentation.
Candida stellata is an imperfect yeast with a taxonomic history characterized by numerous changes of names and definition (for reviews see Lodder & Kreger-van Rij, 1952; Kreger-van Rij, 1984; Kurtzman & Fell, 1998). Originally, two types of this organism were isolated from a must made in Germany from overripe, somewhat shrivelled grapes (‘Trocken-Beerenauslese’) with a very high sugar concentration (ca. 60%). One type had elongated cells and was therefore named Saccharomyces bacillaris. The other type was designated Saccharomyces stellatus because of the star-like chains of its more spherical cells observed in liquid media. Both species were later placed in the genus Torulopsis because of the lack of spore formation. In an independent line of taxonomic research, a third type was isolated from grapes in Italy and described under the name Brettanomyces italicus. Finally, all these taxa were united in a single species named C. stellata, for which the strain originally described as Saccharomyces stellatus was chosen as type strain (CBS 157).
However, the presumed role of C. stellata in quality determination appears to be rather controversial because contradictory features have been attributed to it by different laboratories. For example, some authors reported on the production of high levels of acetic acid (Soles et al., 1982; Soden et al., 2000), glycerol (Ciani & Ferraro, 1996; Soden et al., 2000) and succinic acid (Ciani & Maccarelli, 1998), whereas other works found low acetic acid production (e.g. Povhe Jemec & Raspor, 2005) and low glycerol production (Clemente-Jimenez et al., 2004). The controversial results indicate that C. stellata is either a heterogeneous species or easily confused with other yeast species that colonize the same substrates. The recent finding that the strain DBVPG 3827, frequently used to investigate the oenological properties of C. stellata, belongs to Starmerella bombicola (Sipiczki et al., 2005), and the recent identification of a new species (C. zemplinina), that can easily be confused with C. stellata when conventional taxonomic tests and routine PCR-restriction fragment length polymorphism (RFLP) analysis of the internal transcribed spacer (ITS) region are used for identification (Sipiczki, 2003; Sipiczki et al., 2005), support the latter possibility and raise doubts about the precise taxonomic position of the oenological C. stellata strains described in the literature. These considerations prompted the authors to reinvestigate the taxonomic position of certain C. stellata strains deposited in culture collections or described in recent reports. This study shows that most of the strains tested do not belong to C. stellata but to C. zemplinina or other related species. The molecular examination of the yeast populations of botrytized grapes and wines in the Tokaj wine region did not identify C. stellata either, although previous works using conventional taxonomic methods found it characteristic of this region (Minarik & Hanikova, 1982; Antunovics et al., 2003).
Materials and methods
The yeast strains used in this study and their origin are listed in Table 1. All CBS and DBPVG strains were purchased from Centralbureau voor Scimmelcultures, Delft, the Netherlands (CBS) or from the Industrial Yeast Collection, Dipartimento di Biologia Vegetale, Perugia, Italy (DBPVG). The other strains were kindly provided by the curators of the culture collections CCY (Culture Collection of Yeasts, Bratislava, Slovakia) and NCAIM (National Collection of Agricultural and Industrial Microorganisms, Budapest, Hungary), the researchers who isolated and described them in publications or were isolated in this study. Saccharomyces cerevisiae S288c was from the Yeast Genetic Stock Center (Berkeley, CA).
Cultures were maintained on YPGA (2% glucose, 2% agar, 1% yeast extract and 1% peptone; all w/v). Cultures for electrophoretic karyotyping and DNA isolation were grown in YPGL (YPGA without agar) at 26 °C in a gyratory shaker overnight. The culture media used for taxonomic tests are described in van der Walt & Yarrow (1984).
Yeast isolation and taxonomic tests
Noble rotted grapes and wine samples were collected in wineries located in the Hungarian part of the Tokaj wine-growing region (Table 1). Individual botrytized grape berries were collected aseptically from 26 bunches, homogenized in 1 mL sterile water, and samples of the paste obtained were streaked onto YPGA plates. From fermenting wines, 1-mL samples were taken in six wineries and streaked onto YPGA plates. Individual colonies of white colour and smooth surface were isolated from the plates after incubation at 25 °C for 5 days. Morphological observation, fermentation, assimilation and other growth tests standard to yeast taxonomy were performed as described by van der Walt & Yarrow (1984).
PCR amplification, restriction analysis and sequence comparison
The D1/D2 domains of the large-subunit (LSU) rRNA genes and the ITS1–5.8S rRNA gene-ITS2 sequences were amplified from genomic DNA using the PCR conditions described previously (Sipiczki, 2003). The D1/D2 domains were amplified with primers NL-1 and NL-4 and sequenced. The blast network service of the NCBI database (http://ncbi.nlm.nih.gov/blast) was used for DNA sequence similarity searches. The sequences of the amplified fragments were also compared to the sequences of the type strains obtained from the CBS database using the blast 2 Sequences tool of NCBI (http://www.ncbi.nlm.nih.gov/blast/bl2seq/wblast2.cgi). The ITS1–5.8S–ITS2 regions were amplified with primers ITS1 and ITS4. The amplified DNA was digested with the restriction endonucleases DraI, MboI and HaeIII.
Chromosomal DNA for karyotyping was prepared by the method described in Sipiczki et al. (2001). Electrophoresis was performed with the CHEF-DRIII system (Bio-Rad Laboratories, Hercules, CA) in 1% (v/w) agarose gel and 0.5% (v/w) TBE buffer. The running protocol was 1.5 V cm−1 with 100–4400 s switch time ramp at an angle of 120° for 96 or 120 h. The system was cooled to 14 °C using a chiller water bath and a pump.
Taxonomic reidentification of C. stellata strains
Thirty-three strains deposited in six culture collections and eight strains provided by four laboratories were tested for the utilization of a selected subset of carbon and nitrogen sources routinely used in conventional taxonomy. Eight strains (C2, CBS 1779, CCY 26-10-7, CCY 26-13-1, DBVPG 3826, DBVPG 4171, FAW70 and RIVE 3-16-1) showed unique taxonomic features. The rest of the strains comprised groups with properties characteristic of the species Starmerella bombicola, C. lactis-cendensi or the species pair C. stellata-C. zemplinina.
All strains were then subjected to PCR-RFLP analysis of the chromosomal ITS1-5.8S rRNA gene-ITS2 region. The amplified fragments were digested with MboI, DraI and HaeIII separately. Digestion with MboI is known to generate specific patterns for each of C. stellata, C. zemplinina and Starmerella bombicola (Sipiczki et al., 2005). DraI distinguishes C. stellata from C. zemplinina (Sipiczki, 2004). Digestion of the fragments of all strains gave three (DraI: lanes 1, 2 and 7; HaeIII: lanes 5, 19 and 24) or four (MboI: lanes 3, 4, 13 and 15) patterns (Fig. 1). Thirty-nine out of the 41 strains examined showed combinations of patterns different from that of the type strain of C. stellata (Table 1), indicating that only two of the investigated strains might belong to C. stellata.
Next the D1/D2 domains of the LSU rRNA gene of all strains were amplified and sequenced. The blast search with the sequences identified high degrees of similarity (98–100%) with sequences of the type strains of 11 species (Table 2). Most strains originally isolated from grapes or wine fermentation belonged to C. zemplinina or to Starmerella bombicola (DBVPG strains). When the results of the taxonomic physiological tests were compared with the molecular results, all C. zemplinina strains grew in the presence of 1% acetic acid, which, however, inhibited the growth of C. stellata.
Table 2. RFLP patterns and LSU rRNA gene sequence accession numbers for the strains examined
Taxonomic identification of yeasts newly isolated from botrytized grapes and wines
Candida zemplinina was discovered among wine yeasts that showed a taxonomic profile characteristic of C. stellata (Sipiczki, 2003). Because both species grow in similar environments (e.g. overripe grapes and grape must with high sugar concentration), they presumably may form mixed populations in the colonized substrates. To find such populations, yeasts were isolated from botrytized grapes and two types of fermenting sweet botrytized wines in the Hungarian part of the Tokaj wine region (shared by Hungary and Slovakia), where C. zemplinina was identified for the first time. Twenty-two isolates were selected that appeared to belong to C. stellata in conventional (nonmolecular) taxonomic tests. The RFLP analysis of their ITS regions and the sequences of their D1/D2 domains classified them to C. zemplinina or to C. lactis-condensi. None of the isolates proved to be C. stellata in the molecular tests.
The chromosomes of many fungi and yeasts are variable in size. Both C. zemplinina and C. stellata had three chromosomes and both sets contained chromosomes that varied in size (Sipiczki, 2004). Here, selected strains of each species identified in this study were karyotyped (Fig. 2). ABT 804, the only C. stellata strain that had not been karyotyped before, also had three chromosomes, but each chromosome differed in size from those of CBS 157T and CBS 843. Because the karyotypes of the latter strains were not identical either (Sipiczki, 2004), C. stellata appears to be prone to undergo chromosomal rearrangements. In contrast, the C. zemplinina strains did not show chromosomal polymorphism. All had two bands in identical positions to the bands of the type strain. The much brighter upper band consists of two chromosomes (Sipiczki, 2004). Interestingly, C. davenportii, C. lactis-condensi and C. bombi also have three chromosomes with total genome sizes roughly identical with those of C. zemplinina and C. stellata.
The results presented in this study demonstrate that the species name C. stellata has been used for yeasts, many of which, in view of recent developments of yeast taxonomy, are not conspecific with the type strain of C. stellata. Most strains originally identified as C. stellata and examined in this study turned out to belong to species that were not known yet at the time of their isolation, such as C. zemplinina, C. lactis-condensi, C. davenportii or Starmerella bombicola.
The wine yeasts deposited in DBVPG as C. stellata strains turned out to be strains of Starmerella bombicola, a recently identified species (Rosa & Lachance, 1998), unknown at the time of their deposition. However, most wine strains preserved in CBS or described in recent publications as C. stellata proved to belong to C. zemplinina, a closely related species, which is very difficult to separate from C. stellata by means of standard taxonomical tests (Sipiczki, 2003, 2004). Candida stellata was not found among yeasts newly isolated from noble rotted grapes and botrytized wines either, although overripe grapes and fermenting grape musts with high sugar concentrations are environmental conditions in which strains identified as C. stellata were frequently detected (e.g. Minarik & Hanikova, 1982; Rosini et al., 1982; Mills et al., 2002; Divol & Lonvaud-Funel, 2005; Hierro et al., 2006). The findings indicate that C. stellata is far less widespread in grapes and natural wine fermentation than hitherto thought. Because C. stellata and C. zemplinina are almost indistinguishable in conventional taxonomical tests, it is usually believed that C. zemplinina rather than C. stellata occurs in fermenting wines, and most of the wine yeasts identified and described in the literature as C. stellata might be strains of C. zemplinina. In the case of botrytized wines, the higher occurrence of C. zemplinina might be due to its ability to withstand higher acetic acid concentrations. It has been shown that grapes infected by Botrytis cinerea have a high population of acetic acid bacteria, which can grow in grape must and accumulate gluconic and acetic acids (Barbe et al., 2001). This study found that Candida zemplinina grows in the presence of 1% acetic acid, which is inhibitory to C. stellata.
The response to 1% acetic acid is a physiological trait that can be exploited in routine taxonomic differentiation of the strains of these species. However, their unequivocal identification requires molecular techniques, either PCR RFLP of the ITS1–5.8S rRNA gene–ITS2 region or sequencing of the D1/D2 domain of the LSU rRNA gene. When ITS RFLP is used, it has to be kept in mind that the most frequently used restriction endonucleases (e.g. HinfI, CfoI and HaeIII) produce identical patterns for both species. Distinct, species-specific patterns can only be obtained with MboI and DraI as shown previously (Sipiczki, 2004) and confirmed here.
To test the hypothesis about the wider occurrence of C. zemplinina, the C. stellata LSU rRNA gene sequences published by other laboratories were revisited. This study found that the D1/D2 domain sequences of the C. stellata strains isolated from French cider by Coton et al. (2006) and from Spanish wine by Hierro et al. (2006) and the corresponding sequence of Candida sp. isolated from a Californian sweet botrytized wine by Mills et al. (2002) were identical with that of the type strain of C. zemplinina. Likewise, C. zemplinina but not C. stellata was found in fermenting red wine made from the Portuguese grape variety Castelao, where it was predominating for most of the fermentation (Baleiras Cuoto et al., 2005). Candida zemplinina was also identified among other Portuguese wine-related yeasts (accession number AY394855) and in Greek botrytized grapes (accession number DQ872872).
From these results and considerations it could be deduced that what has been attributed to C. stellata in oenology before could now be assigned to C. zemplinina. However, considerable amount of information published about C. stellata in wine-making came from the investigation of DBVPG strains (e.g. Ciani & Ferraro, 1996, 1998; Ciani et al., 2000; Ferraro et al., 2000), which was found to be strains of Starmerella bombicola. On the other hand, in spite of its frequent occurrence on grapes and in wines, C. zemplinina cannot be considered a wine-specific yeast. It has also been detected in yeast populations associated with Ghanaian cocoa fermentation (Nielsen et al., 2005, 2007), and two CBS strains identified in this study as C. zemplinina originate from soil (CBS 2799) and Drosophila (CBS 4729) (Phaff et al., 1956). Its association with Drosophila indicates that fruit flies might be important vectors that transport its cells from winery to ripening grapes in the vineyard. The related species C. bombi, Starmerella bombicola and C. lactis-condensi are also associated with insects (Lachance et al., 1995, 2001; Loureiro & Malfeito-Ferreira, 2003; etc.).
The genome of C. zemplinina is similar in size to the genome of C. stellata (and the genomes of the related C. bombi, C. lactis-condensi and Starmerella bombicola) but appears to differ from it in stability. The C. zemplinina chromosomes are less variable than those of C. stellata. This stability indicates that chromosome rearrangements may not be as important in this species as in Saccharomyces cerevisiae (e.g. Puig et al., 2000) for adaptation to the changing conditions during wine fermentation.
The authors thank E. Breierová (CCY, Bratislava), A. Frei (Waedenswill), J. Gafner (Waedenswill), G. Peter (NCAIM, Budapest), D.F. Rodriguez Vico (Almeria), P. Romano (Potenza) and L. Solieri (Reggio Emilia) for strains. This work was supported by the grants NKFP-4/017/2005 OM-00016/2006 and RET-06/2004 provided by NKTH and KPI.