A wide variety of yeast species have been implicated in wine spoilage. Of the spoilage yeasts, species of
Brettanomyces (imperfect state,
Dekkera) are probably the most serious (
21) and controversial. These organisms are most frequently found in red wine 6 to 10 months after barrelling (
4), but they have also been isolated from other fermented beverages, such as beer and cider.
Brettanomyces species can survive, multiply, and contaminate wines from transfer piping or cooperage that has been insufficiently cleaned and disinfected after use. These potential spoilage yeasts have been identified in almost every wine-producing area of the world (
12). Wines infected with
Brettanomyces/Dekkera yeasts develop off-flavors, which are described as animal, stable, barnyard, horse blanket, and burnt plastic (
5,
12,
13,
16,
18).
Brettanomyces can produce a distinct haziness when the concentration of cells present in the wine is <3 × 10
3 cells/ml (
4) or higher (
13).
Brettanomyces species can synthesize volatile phenolic compounds, including phenol, syringol (
16), and several ethylphenols (
6). Control of these organisms is difficult due to their relative resistance to normally used concentrations of sulfur dioxide (
15). The physiology and ecology of this spoilage genus are still unclear, and little is known about it (
14).
In this paper we describe the development of a PCR-restriction enzyme (RE) analysis protocol to detect and identify, directly in wine samples, B. bruxellensis and B. anomalus. Moreover, other molecular techniques, such as PCR-DGGE, reverse transcription (RT)-PCR-DGGE, and hybridization with a B. bruxellensis-specific probe, were used as confirmatory tests for detection of spoilage yeasts in wine samples.
DISCUSSION
Brettanomyces is one of the most complex and controversial yeast issues that a winemaker encounters when making wine. This genus is usually not included in the genera of yeasts found on the grape surface (
11); however, it is quite common to find
Brettanomyces in a winery (
1,
4,
12).The use of different selective and/or differential media (
1,
5,
9,
12,
24) is time-consuming and/or very expensive. Rapid and reliable detection, quantification, and characterization methods are required for a successful wine microbiological analysis. A novel approach involves application of molecular techniques. Among these techniques, the PCR-based methods are attractive because of their rapidity.
Previous studies have described detection and discrimination of
Brettanomyces/Dekkera species by different PCR methods. Mitrakul et al. (
23) used PCR-randomly amplified polymorphic DNA analysis for discriminating different strains of
B. bruxellensis. A new fluorescence in situ hybridization method, in which peptide nucleic acid probes are used for identification of
Brettanomyces, was proposed by Stender et al. (
26). This method is easily adapted to microscopic techniques currently used in wine laboratories, but a fluorescence microscope is required. Identification of
Brettanomyces/Dekkera species based on polymorphism in the rRNA internal transcribed spacer region has been reported (
10). In this study the authors used four primers to identify the species of the genus
Brettanomyces/Dekkera. They used the discriminatory potential of the internal transcribed spacer regions located between the rRNA genes. However, for the analyses described in the papers mentioned above the workers used isolated strains and classical techniques and thereby included biases inherent in traditional plating or enrichment. The use of culture-independent methods has repeatedly demonstrated that there is tremendous variance between cultivated and naturally occurring species. These approaches have been used recently to study the ecology of different ecosystems, including wine fermentations (
8,
22). Only Ibeas et al. (
18) described a nested PCR method for identification of
Brettanomyces/Dekkera strains directly from sherry wine. The protocol described in this paper was found to be specific for
D. intermedia,
B. bruxellensis, and
B. lambicus, and no PCR product was obtained for
B. anomalus and
Brettanomyces claussiensis, thus eliminating the possibility that the technique could be used to detect one species considered to be a wine spoilage organism,
B. anomalus.
The goal of this study was to optimize a culture-independent, molecular technique-based protocol that would allow detection and differentiation of B. bruxellensis and B. anomalus, the main agents of wine spoilage. The protocol which we developed is based on PCR amplification and RE analysis and can be used for suspected colonies isolated from spoiled wine for identification purposes, but more significantly, it can be used directly with wine samples. DNA and/or RNA may be extracted from spoiled wine, and by using PCR-RE analysis the presence of B. bruxellensis and/or B. anomalus can be determined without prior traditional, culture-dependent isolation. Moreover, since the method could be also applied to directly extracted RNA, studies of the activity and viability of B. bruxellensis and B. anomalus can be performed.
Primers DB90F and DB394R were designed on the basis of the differences between 26S rDNA sequences belonging to
Brettanomyces spp. and yeast species involved in wine fermentations (Fig.
1), and this was possible because of the extensive database for the D1-D2 loop created in the last few years (
20). The primers selected in this study had a high specificity for
B. bruxellensis and
B. anomalus. This specificity made it possible to amplify DNA extracted directly from wine for detection of
B. bruxellensis and
B. anomalus. The protocol described here had a high detection limit. When the DNA and RNA were extracted from wine, a visible signal was not obtained for concentrations less than 10
4 to 10
5 cells/ml. This finding could be explained by considering different factors. The first factor is the possible presence of residual inhibitory substances that are present in the nucleic acid preparations that are not completely removed with the specific extraction kit and interfere with the amplification step. As a matter of fact, other authors have reported a substantial difference in the detection limit for
Dekkera cells when the PCR was performed with DNA extracted from pure cultures or directly with wine (
18). Moreover, the use of degenerate primers for simultaneous amplification of
B. bruxellensis and
B. anomalus may affect the stability of the primer match (defined as the measure of how tightly the primer and target are bound), as well as the primability of the primer match (which indicates how easily the DNA polymerase is able to extend the sequence). Detection of
B. bruxellensis and
B. anomalus in wines containing lower numbers of cells could be partially overcome by using a volume of wine larger than 1 ml, as suggested by other authors (
22), in a way that more cells could be collected and processed. On the other hand, overloading of the kit used for extraction of the nucleic acids should be avoided, to prevent false-negative results due to inhibition of DNA polymerase by wine compounds, such as polysaccharides and polyphenols.
The sequence divergence in the fragment amplified was subsequently exploited to identify the two species considered by means of RE analysis. The
DdeI restriction reaction could be used to differentiate the two species. As shown in Fig.
3, distinct restriction patterns were obtained, which allowed straightforward identification.
Twelve samples of wine that were suspected to be spoiled by
Brettanomyces spp. because of their odor characteristics were analyzed by culture-dependent and culture-independent methods. When plating on WLN medium (a culture-dependent method) was used, six samples were positive for the presence of
Brettanomyces spp. The counts (Table
2) refer to the
Brettanomyces population, based on the specific characteristics of morphology and the color on WLN medium. The concentrations were about 10
4 CFU/ml only for samples 11 and 12, whereas for the other positive samples concentrations of about 10
3 CFU/ml were obtained. The remaining six samples exhibited no growth on the plates, underlining the conclusion that the concentration was less than 10 CFU/ml. Sample 12 was the only sample that produced a mixed population on the WLN plates. Another yeast population was observed along with
Brettanomyces spp. All the other positive samples had a single colony morphology and color characteristic of
Brettanomyces spp. At least five suspected colonies from the positive samples were isolated and subjected to molecular identification with the protocol developed in this study. After PCR amplification and RE analysis, all the isolates produced a profile identical to that of
B. bruxellensis (Fig.
4).
Interesting results were obtained when culture-independent methods were used to detect and identify the two species considered in the wine samples. Because of the high detection limit of the method, DNA was extracted from 50 ml of wine. When the specific PCR was used, a larger number of positive samples were detected than when the traditional plating method was used. All the samples that produced colonies on the plates also gave the specific PCR product when the concentration was not greater than 10
3 CFU/ml, but samples 5 and 6, containing a concentration of <10 CFU/ml, also produced the 305-bp amplicon (Fig.
5). This evidence could be explained by the presence of nonculturable or dead cells of
Brettanomyces spp. This issue was immediately addressed when the RNA extracted from wine was subjected to specific RT-PCR. One important aspect is that RNA was extracted from 1 ml of wine. When higher volumes were used for RNA preparation, the pellets always were pink to dark brown, indicating that the purity of the nucleic acid obtained was poor. No positive results were obtained by specific RT-PCR (Table
2), even for the samples that gave positive PCR results. A possible explanation for this inconsistency is that different volumes of wine samples were processed. On the basis of these results, samples 5 and 6 were considered to contain only dead cells of
Brettanomyces spp. The specific PCR products obtained from the wine samples were then subjected to RE analysis with the
DdeI restriction endonuclease. All the amplicons were cut, and the patterns obtained were identical to the
B. bruxellensis pattern. These results were in agreement with the identification of the isolates from the samples that exhibited growth on WLN medium.
The results of
Brettanomyces detection by specific PCR were confirmed by DGGE analysis of PCR and RT-PCR products generated with universal primers and RNA hybridization with a
B. bruxellensis-specific probe. In the DNA DGGE gels, all positive samples, as determined by specific PCR, produced a band that comigrated with the
B. bruxellensis band (Fig.
7). Samples 5, 6, and 12 also produced a second band that was referable to
S. cerevisiae. The specific
B. bruxellensis band in sample 12 was faint, probably because of a masking effect due to the presence of a high concentration of
S. cerevisiae, as demonstrated by plate counting (data not shown). After DGGE RNA analysis, samples 11 and 12 produced the cognate
B. bruxellensis band, while the other samples contained different yeast populations (Fig.
8). Finally, RNA hybridization with probe BRE26S14 specific for
B. bruxellensis was carried out. The detection limit of the assay was determined to be 10
4 cells of
B. bruxellensis (Fig.
9), and it was 10-fold lower than the level of specific PCR amplification. A positive spot was observed only for samples 11 and 12, which were characterized by high counts of
Brettanomyces spp. on the plates.
The results obtained by the multiphasic approach used allowed us to study the microbial ecology of the samples considered, with the specific aim of detecting Brettanomyces spp. Only samples 11 and 12 contained an active B. bruxellensis population consisting of at least 104 cells/ml. For the rest of the positive samples the size of the active population was 103 CFU/ml, as determined by plate counting. For samples 5 and 6, characterized by positive signals at the DNA level but not from the RNA, we speculate that there was a dead population of B. bruxellensis. Moreover, if the results obtained by PCR and RT-PCR with both specific and universal primers were combined, it could be possible to determine that spoilage by B. bruxellensis in samples 11 and 12 occurred more recently than spoilage in sample 5 or 6, in which only dead cells were detected by molecular methods.
Since the time required for extraction of DNA from cells and for differentiation of
Brettanomyces strains by RE analysis is only approximately 8 h, this method could also be used for fast identification of
B. bruxellensis and
B. anomalus strains isolated from wine. Thus, this method is faster than traditional methods, which take 1 to 2 weeks. The availability of a rapid technique permits easy identification of
Brettanomyces species during wine maturation. The disadvantage of the protocol described here is the high detection limit. However, this deficiency is less important if it is considered that the typical unpleasant odor appears when the
Brettanomyces concentration reaches 10
5 CFU/ml or higher (
5).