Real-Time PCR Assay for Detection and Enumeration of Dekkera bruxellensis in Wine

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Dec. 2003, p. 7430 7434 Vol. 69, No. 12 0099-2240/03/$08.00 0 DOI: 10.1128/AEM.69.12.7430 7434.2003 Copyright 2003, American Society for Microbiology. All Rights Reserved. Real-Time PCR Assay for Detection and Enumeration of Dekkera bruxellensis in Wine Trevor G. Phister and David A. Mills* Department of Viticulture and Enology, University of California, Davis, California 95616 Received 23 June 2003/Accepted 12 September 2003 Traditional methods to detect the spoilage yeast Dekkera bruxellensis from wine involve lengthy enrichments. To overcome this difficulty, we developed a quantitative real-time PCR method to directly detect and enumerate D. bruxellensis in wine. Specific PCR primers to D. bruxellensis were designed to the 26S rrna gene, and nontarget yeast and bacteria common to the winery environment were not amplified. The assay was linear over a range of cell concentrations (6 log units) and could detect as little as 1 cell per ml in wine. The addition of large amounts of nontarget yeasts did not impact the efficiency of the assay. This method will be helpful to identify possible routes of D. bruxellensis infection in winery environments. Moreover, the time involved in performing the assay (3 h) should enable winemakers to more quickly make wine processing decisions in order to reduce the threat of spoilage by D. bruxellensis. Dekkera bruxellensis (anamorph Brettanomyces bruxellensis) is a spoilage yeast in the beverage industry that is found in soft drinks and alcoholic beverages (10). It is especially important to the wine industry, where it has been shown to produce phenolic taints (4-ethylphenol [4EP] and 4-ethylguaiacol) and to contribute to the production of biogenic amines in red wine (5 7). While Dekkera sp. has been shown to produce high levels of cinnamate decarboxylase, the enzyme responsible for the production of 4EP from phenolic acids, it has been difficult to precisely correlate the amount of 4EP in red wine to the growth of Dekkera populations in the wine. Often 4EP is present at high levels in wine when Dekkera populations are not detected at significant levels (6, 7, 21). Rodrigues et al. (21) speculated that this result is due to a combination of the low proportion of Dekkera species in the total contaminating floras as well as to the use of inadequate culture media to enumerate those populations. Traditional methods to identify spoilage yeasts in wine rely on culturing (12). In the case of Dekkera or Brettanomyces species, culturing usually involves selective media containing cycloheximide and typically takes 1 to 2 weeks to perform (3). Advances in molecular typing have dramatically enhanced the ability to differentiate Dekkera colonies once they are isolated from wine. Mitrakul et al. (17) used a randomly amplified polymorphic DNA-PCR assay to discriminate strains of D. bruxellensis in three different Cabernet Sauvignon vintages. Egli and Henick-Kling (11) used a PCR assay based on the rrna internal transcribed spacer region to differentiate six wine isolates. Stender et al. (22) developed a peptide nucleic acid probe to the D. bruxellensis 26S rrna gene and examined isolates from three wines by fluorescence in situ hybridization. While these methods employ novel approaches for the differentiation of strains, they all relied on microbial enrichment. * Corresponding author. Mailing address: Department of Viticulture and Enology, University of California, One Shields Ave., Davis, CA 95616. Phone: (530) 754-7821. Fax: (530) 752-0382. E-mail: damills @ucdavis.edu. From the winemaker s perspective, this delay is problematic since decisions on wine processing (antimicrobial additions, filtering, etc.) are similarly delayed. Few researchers have employed methods without any enrichment steps to directly identify yeasts from wine. Cocolin et al. (9) directly differentiated yeasts in wine by PCR and denaturing gradient gel electrophoresis. Ibeas et al. (14) developed a two-step PCR which could detect as few as 10 intact Dekkera cells in contaminated sherry. There are two principal advantages of the direct characterization of wine microbial DNA as opposed to yeast enrichment and plating. The first is the fact that many microbial populations might not respond to enrichment due to injury, lack of appropriate nutrients, or persistence in a viable but nonculturable state. For example, approaches relying on PCR and denaturing gradient gel electrophoresis have identified nonculturable yeast populations in commercial wine fermentations (8, 16). The second advantage is that direct analyses take less time than enrichment methods. This advantage may, in turn, allow winemakers to use microbial detection data in a prophylactic fashion, avoiding spoilage problems before they arise. Moreover, the logistics of DNA analysis allow larger numbers of samples to be processed than would be operable for plating studies. Real-time or quantitative PCR (QPCR) assays have been developed for the detection and enumeration of a number of fungi and food-borne pathogens (2, 4, 13, 15, 18). QPCR offers significant advantages over other molecular methods in terms of the speed by which assays are performed and the ability to quantify the target microbial population. In this study we developed a QPCR method for the detection and quantification of D. bruxellensis in wine. This method will enable a more comprehensive determination of D. bruxellensis in wine, thereby facilitating a better understanding of its origin in wineries as well as aiding studies of the interactions between D. bruxellensis and the normal wine flora. Finally, this method should also allow winemakers to more quickly assess the spoilage potential of D. bruxellensis in various juices and wines during vinification. 7430

VOL. 69, 2003 PCR ASSAY FOR DETECTION OF D. BRUXELLENSIS IN WINE 7431 TABLE 1. Specificity of the DBRUXF-DBRUXR primer pair Microorganism Strain no. a PCR result D. bruxellensis or synonyms D. bruxellensis 82-26 D. bruxellensis UCD2050 D. bruxellensis NRRLY-12961 Brettanomyces abstinens 82-24 B. lambicus 89-4 B. intermedius 82-30 Other yeasts D. anomala NRRLY-17522 Brettanomyces custersianus NRRLY-6653 B. nanus NRRLY-17527 B. naardenensis NRRLY-17526 Candida apicola 71-17 Candida kefyr 76-22 Candida krusei 94-157.1 Candida lambica 72-232 Candida mesenterica 67-35 Candida sake 51-18 Candida sorboxylosa 91-491.3 Candida stellata 72-1034 Candida veronae 86-32 Candida vini 40-36 Debaryomyces hansenii 74-86 Debaryomyces carsonii 66-20 Hanseniaspora osmophila 40-417 Hanseniaspora uvarum 54-192 Hanseniaspora valbyensis 68-28 Issatchekia terricola 91-489.3 Issatchekia orientalis 75-62 Kloeckera japonica 94-153.2 Kluyveromyces marxianus 55-82 Metschnikowia pulcherrima 40-214 Pichia anomala 76-71T Pichia membranaefaciens 57-22 Saccharomyces bayananus 01-159 S. cerevisiae NRRLY-12632 Schizosaccharomyces japonicus 71-26 Zygosaccharomyces bailii 68-113 Bacteria Acetobacter aceti UCD114 Lactobacillus plantarum UCD1 Oenococcus oeni UCD154 a Strains are from the following sources: Agricultural Research Service culture collection (NRRLY) and the University of California, Davis, Department of Viticulture and Enology culture collection (UCD). All other strains are from the Herman J. Phaff Yeast Culture Collection, University of California, Davis. MATERIALS AND METHODS Yeast strains and propagation. The strains used in this study are listed in Table 1. All yeasts were grown in YM broth (3 g of yeast extract, 3gofmalt extract, 3gofpeptone, 10 g of dextrose, and 1 liter of H 2 O) (Becton Dickinson, Sparks, Md.) at 25 C. Yeasts were obtained from the U.S. Department of Agriculture Agricultural Research Service culture collection (Peoria, Ill.), the Herman J. Phaff Yeast Culture Collection (University of California-Davis, Davis, Calif.) or the Department of Viticulture and Enology culture collection (University of California-Davis). Primer design. Sequence analysis was performed with the Seqweb GCG version 2 sequence analysis program (Accelrys, Inc., San Diego, Calif.). The D1-D2 domain of the large subunit domain of the rrna gene from Dekkera anomala, D. bruxellensis, Brettanomyces custersii, Brettanomyces naardensis, Brettanomyces nanus, and Saccharomyces cerevisiae were aligned, and primers DBRUXF 5 -G GATGGGTGCACCTGGTTTACAC-3 and DBRUXR 5 -GAAGGGCCACA TTCACGAACCCCG-3 were selected to produce a 79-bp fragment specifically from D. bruxellensis. Specificity of PCR assays. DNA samples from all yeasts were isolated as described previously (16). PCRs were performed at a final volume of 50 l. All PCR reagents were obtained from Applied Biosystems, Foster City, Calif. Each reaction mixture contained 5 l of AmpliTaq Gold buffer; 2.0 mm MgCl 2 ; 0.2 mm (each) datp, dctp, dgtp, and dttp; 0.2 mm primers; 1.25 U of Ampli- Taq Gold; and 2 l (approximately 20 ng) of extracted DNA. The reactions were run for 40 cycles on a GeneAmp 2700 thermal cycler (Applied Biosystems), with denaturation at 95 C for 60 s, annealing at 69 C for 45 s, and extension at 72 C for 7 s. An initial 5-min denaturing step at 95 C and a final 7-min extension step at 72 C were used. The products were analyzed by agarose gel electrophoresis on a 3% gel and stained with 0.5 mg of ethidium bromide per ml (1). The gels were visualized under UV transillumination with a Multimage light cabinet (Alpha Innotech Corporation, San Leandro, Calif.). QPCR. QPCR were performed on an Applied Biosystems Prism 7700 sequence detection system. SybrGreen master mix was used according to the manufacturer s instructions (Applied Biosystems). Optimized reactions were performed in 0.5-ml MicroAmp optical tubes or plates, and each 50- l reaction mixture contained the following: 1 SybrGreen master mix, 900 nm DBRUXF, 300 nm DBRUXR, and 2 l of purified DNA. Each reaction was performed in triplicate. The reactions were run for 40 cycles, with denaturation at 95 C for 60 s, annealing at 69 C for 45 s, and extension at 72 C for 7 s. An initial 5-min denaturing step at 95 C was used. The cycle threshold (C T ), or the PCR cycle where fluorescence first occurred, was determined automatically by using sequence detector software (version 1.7; Applied Biosystems). Artificial contamination of wine. D. bruxellensis UCD2050 (5.8 10 7 CFU per ml) was serially diluted in sterile peptone water, plated on YM agar, and incubated for 1 to 2 weeks at 25 C to obtain the number of CFU per milliliter. For QPCR analysis, this same original culture was serially diluted in filter-sterilized wine (Cabernet Sauvignon) and wine containing approximately 10 7 S. cerevisiae cells. DNA was isolated from 700 l of sample by using a MasterPure yeast DNA purification kit (Epicentre Technologies, Madison, Wis.) and then diluted 10- fold in sterile water. This DNA was then used in the QPCR reactions described above. Standard curves for quantification of unknown samples and determination of amplification efficiency were generated by plotting the C T values of QPCR performed on the DNA from these dilution series against the log input cells (ABI PRISM 7700 sequence detection system, user bulletin 2). Analysis of true wine samples. Three Cabernet Sauvignon samples and a Merlot sample known to be contaminated with Dekkera were provided by a local winery. Each sample was serially diluted in sterile peptone water. These dilutions were plated in triplicate on WL nutrient agar (Becton Dickinson) containing 10 mg of cycloheximide (Sigma-Aldrich, St. Louis, Mo.) per liter and incubated at 25 C for 14 days. DNA was isolated from 700 l of the sample and then diluted 10-fold. This DNA was then quantified by QPCR as described above. Each sample was analyzed in triplicate. RESULTS Primer design and specificity. 26S ribosomal DNA (rdna) gene sequences for the five Dekkera and Brettanomyces species plus S. cerevisiae were aligned, and regions specific to D. bruxellensis were used to create primers DBRUXF and DBRUXR (Fig. 1). These primer sequences were then checked against both the GenBank and EMBL databases. Both sequences exhibited significant homology only to D. bruxellensis and, in the case of DBRUXR, to a D. anomala entry; however, no other significant hits were identified. The primers were then empirically tested by PCR against various yeasts and bacteria known to have been isolated from wine. Only D. bruxellensis and its synonyms produced PCR products (Table 1). QPCR detection limits. The QPCR assay was carried out on D. bruxellensis cells contained within rich medium (RM), wine, and wine supplemented with S. cerevisiae, which was used to gauge the impact of a large amount of nontarget DNA on the QPCR assay. D. bruxellensis cells were serially diluted in RM, and DNA isolated from each dilution was used to construct a standard curve. The same original culture was also serially diluted in wine or wine supplemented with S. cerevisiae in order to determine the effects of this matrix on the QPCR assay (Fig.

7432 PHISTER AND MILLS APPL. ENVIRON. MICROBIOL. FIG. 1. Alignment of partial 26S rdna sequences with the DBRUXF and DBRUXR primers. The shaded sequences are regions of nonidentity to the D. bruxellensis 26S rdna sequence. Note that the reverse complement of DBRUXR is presented in order for homology to be viewed easily. GenBank accession numbers: for D. bruxellensis, U45738; for D. anomala, U84244; for B. naardenensis, U76200; for Brettanomyces custersianus, U76199; for S. cerevisiae, U44806; and for B. nanus, U76197. 2). In all cases, the detection limit was approximately 1 CFU per ml. The assay is linear over six orders of magnitude (Fig. 2). These results suggest that samples obtained from wine or wine containing nontarget yeasts do not significantly impact the assay. To test the reproducibility of the QPCR assay, D. bruxellensis was serially diluted and added to wine to create samples with known levels of contamination. The Dekkera population levels in these samples were then determined by QPCR and correlated to plating analysis from the same dilution (Fig. 3). Three trials were performed and in each case the relationship between the number of CFU determined by plating and that determined by QPCR produced high R 2 values (0.999, 0.986, and 0.966). Quantification of D. bruxellensis in spoiled wine. To test the accuracy of the D. bruxellensis QPCR assay on actual spoilage samples, several wines known to be contaminated with Dekkera (Brettanomyces) were obtained from a local winery. Each wine sample was serially diluted and plated in triplicate for CFU analysis. DNA was isolated from the sample, diluted 10-fold, and then quantified by QPCR (Table 2). In general, the correlation between plating and the population estimated by FIG. 2. Determination of QPCR amplification efficiency and detection limits of D. bruxellensis diluted in sterile peptone water ( ), wine ( ), and wine plus approximately 10 7 cells of S. cerevisiae ( ). The solid lines represent the regression of log cell numbers in each matrix. R 2 values are as follows: peptone water, 0.993; wine, 0.992; wine plus S. cerevisiae, 0.999. C T values are the average of three replicates. FIG. 3. Sensitivity and accuracy of QPCR assay compared to plating for the determination of cell numbers of D. bruxellensis in wine. D. bruxellensis UCD 2050 was serially diluted and plated on YM medium. The same dilution was also performed in Cabernet Sauvignon wine to provide samples with known levels of contamination. DNA was isolated from these samples and cell numbers were determined by QPCR. Three trials were preformed on three separate cultures (trial one, Œ; trial two, ; and trial three, E). The numbers of estimated cells detected by QPCR were compared to those determined by plating and after regression gave R 2 values of 0.999, 0.986, and 0.966, respectively.

VOL. 69, 2003 PCR ASSAY FOR DETECTION OF D. BRUXELLENSIS IN WINE 7433 TABLE 2. Enumeration by plating and QPCR of D. bruxellensis in wine a Sample no. Enumeration method Plating QPCR QPCR was excellent. The largest variation was found in the Merlot sample, where 1.2 10 4 CFU was determined by plating and a predicted 7.5 10 3 CFU was determined by QPCR. However, for the Merlot sample, the standard deviations for both the plating analysis and the QPCR assay were larger than for the other three contaminated wines. DISCUSSION Wine 1 598 33 609 104 Cabernet Sauvignon 2 431 24 433 67 Cabernet Sauvignon 3 100 40 291 16 Cabernet Sauvignon 4 12,700 3,055 7,533 1,856 Merlot a Values are the numbers of CFU per milliliter standard deviations. In this study, we developed a rapid QPCR-based method for the detection and enumeration of D. bruxellensis in wine. Primers were designed to the D1-D2 loop of the 26S rrna gene as this is one of the few gene sequences available for all Dekkera and Brettanomyces species and the region has been used successfully in the past to develop QPCR methods for other yeasts (4). Database searches with the QPCR primers developed here did not reveal significant homology to any organisms other than D. bruxellensis and, in one case, a D. anomala entry. When tested empirically, the DBRUXF-DBRUXR primer pair did not produce amplicons from 36 other wine-related yeasts and three bacteria (Table 1). The detection limit for the QPCR assay, approximately 1 CFU per ml, is in line with that of other similar assays (2, 4). Perhaps most importantly, the assay is robust and functions well on DNA samples isolated from wine, a matrix known to possess various PCR inhibitors (23, 24). Moreover, the assay functions reproducibly in the presence of competing nontarget DNA templates such as S. cerevisiae, the yeast that carries out the primary alcoholic fermentation in the production of wine. In both wine and wine supplemented with S. cerevisiae, the assay exhibited an excellent correlation between the predicted number of CFU per milliliter as determined by QCPR and the number of CFU per milliliter determined by plating. Finally the QPCR assay effectively enumerated the D. bruxellensis populations present in the true spoiled wine samples, correctly estimating the size of the population determined by plating. In one case, the D. bruxellensis population estimated by plating was slightly higher than that determined by QPCR (Table 2). This result may be due to the growth of wine yeasts other than D. bruxellensis on media containing cycloheximide (19). When testing spoiled wine samples, we noted that the QPCR results on undiluted samples exhibited some variability. However, when the samples were first diluted 10-fold, this variability disappeared, and the resulting QPCR analysis showed a good correlation with plating results. The initial dilution may help remove PCR inhibitors, such as phenolics, present in the sample (23). Such inhibitors may be present at different levels in various samples because the wines were processed differently and likely exhibit slightly different chemical compositions. Regardless, we recommend diluting samples 10-fold prior to using this QPCR method to enumerate D. bruxellensis populations from actual wines. The dilution, in turn, changes the lower detection limit for the assay to 10 CFU per ml rather than 1 CFU per ml. The utility of this assay for the wine industry is easily demonstrated (Table 2). To date, few methods are available to rapidly enumerate D. bruxellensis populations in wine, and winemakers must make decisions at various processing steps without the requisite information on the potential for spoilage caused by D. bruxellensis. Currently, two methods are used to monitor Dekkera populations in wines: standard plating analysis with selective media or measurement of 4EP, a by-product of Dekkera growth in wine (12, 20). The first method is problematic since the plates must be incubated for at least 2 weeks, thereby delaying any corrective action by the winemaker until the result is obtained. The second method follows the spoilageassociated end product and, thus, only indirectly suggests the presence, or former presence, of Dekkera organisms. Few molecular methods have been developed to identify Dekkera populations directly from wine (9, 14). The QPCR method developed here takes about 3 hours and would allow winemakers to quickly ascertain Dekkera population levels in order to make efficient processing decisions to eliminate possible spoilage. ACKNOWLEDGMENTS This work was funded in part by the American Vineyard Foundation and the California Competitive Grants Program for Research in Enology and Viticulture (D.A.M.). REFERENCES 1. Ausubel, F. M., R. Brent, R. E. Kingston, D. D. Moore, J. G. Seidman, J. A. Smith, and K. Struhl. 1995. Current protocols in molecular biology. Wiley, New York, N.Y. 2. Blackstone, G. M., J. L. Nordstrom, M. C. L. Vickery, M. D. Bowen, R. F. Meyer, and A. DePaola. 2003. Detection of pathogenic Vibrio parahaemolyticus in oyster enrichments by real time PCR. J. Microbiol. Methods 53: 149 155. 3. Boekhout, T., and V. Robert. 2003. Yeasts in food: beneficial and detrimental aspects. Chapman and Hall, New York, N.Y. 4. Brinkman, N. E., R. A. Haugland, L. J. Wymer, M. Byappanahalli, R. L. Whitman, and S. J. Vesper. 2003. Evaluation of a rapid quantitative realtime PCR method for enumeration of pathogenic Candida cells in water. Appl. Environ. Microbiol. 69:1775 1782. 5. Caruso, M., C. Fiore, M. Contursi, G. Salzano, A. Paparella, and P. Romano. 2002. Formation of biogenic amines as criteria for the selection of wine yeasts. World J. Microbiol. Biotechnol. 18:159 163. 6. Chatonnet, P., D. Dubourdieu, and J. N. Boidron. 1995. The influence of Brettanomyces/Dekkera sp. yeasts and lactic acid bacteria on the ethylphenol content of red wines. Am. J. Enol. Vitic. 46:463 468. 7. Chatonnet, P., D. Dubourdieu, J. N. Boidron, and M. Pons. 1992. The origin of ethylphenols in wines. J. Sci. Food Agric. 60:165 178. 8. Cocolin, L., and D. A. Mills. 2003. Wine yeast inhibition by sulfur dioxide: a comparison of culture-dependent and -independent methods. Am. J. Enol. Vitic. 54:125 130. 9. Cocolin, L., L. F. Bisson, and D. A. Mills. 2000. Direct profiling of the yeast dynamics in wine fermentations. FEMS Microbiol. Lett. 189:81 87. 10. Deak, T., and L. Beuchat. 1996. Handbook of food spoilage yeasts. CRC Press, New York, N.Y. 11. Egli, C. M., and T. Henick-Kling. 2001. Identification of Brettanomyces/ Dekkera species based on polymorphism in the rrna internal transcribed spacer region. Am. J. Enol. Vitic. 52:241 247. 12. Fugelsang, K. 1997. Wine microbiology. Chapman and Hall, New York, N.Y. 13. Hein, I., A. Lehner, P. Rieck, K. Klein, E. Brandl, and M. Wagner. 2001. Comparison of different approaches to quantify Staphylococcus aureus cells by real-time quantitative PCR and application of this technique for examination of cheese. Appl. Environ. Microbiol. 67:3122 3126. 14. Ibeas, J. I., I. Lozano, F. Perdigones, and J. Jimenez. 1996. Detection of Dekkera-Brettanomyces strains in sherry by a nested PCR method. Appl. Environ. Microbiol. 62:998 1003.

7434 PHISTER AND MILLS APPL. ENVIRON. MICROBIOL. 15. Jothikumar, N., and M. W. Griffiths. 2002. Rapid detection of Escherichia coli O157:H7 with multiplex real-time PCR assays. Appl. Environ. Microbiol. 68:3169 3171. 16. Mills, D. A., E. A. Johannsen, and L. Cocolin. 2002. Yeast diversity and persistence in Botrytis-affected wine fermentations. Appl. Environ. Microbiol. 68:4884 4893. 17. Mitrakul, C. M., T. Henick-Kling, and C. M. Egli. 1999. Discrimination of Brettanomyces/Dekkera yeast isolates from wine by using various DNA fingerprinting methods. Food Microbiol. 16:3 14. 18. Nogva, H., A. Bergh, A. Holck, and K. Rudi. 2000. Application of the 5 -nuclease PCR assay in evaluation and development of methods for quantitative detection of Campylobacter jejuni. Appl. Environ. Microbiol. 66: 4029 4036. 19. Pfaff, H. J., M. N. Miller, and E. M. Mrak. 1978. The life of yeasts, 2nd ed. Harvard University Press, Cambridge, Mass. 20. Pollnitz, A., K. Pardon, and M. Sefton. 2000. Quantitative analysis of 4-ethylphenol and 4-ethylguaiacol in red wine. J. Chromatogr. A 874:101 109. 21. Rodrigues, F., G. Goncalves, S. Pereira-da-Silva, M. Malfeito-Ferreira, and V. Loureiro. 2001. Development and use of a new medium to detect yeasts of the genera Dekkera/Brettanomyces. J. Appl. Microbiol. 90:588 599. 22. Stender, H., C. Kurtzman, J. J. Hyldig-Nielsen, D. Sorensen, A. Broomer, K. Oliveira, H. Perry-O Keefe, A. Sage, B. Young, and J. Coull. 2001. Identification of Dekkera bruxellensis (Brettanomyces) from wine by fluorescence in situ hybridization using peptide nucleic acid probes. Appl. Environ. Microbiol. 67:938 941. 23. Wilson, I. G. 1997. Inhibition and facilitation of nucleic acid amplification. Appl. Environ. Microbiol. 63:3741 3751. 24. Zoecklein, B. W., K. C. Fugelsang, B. H. Gump, and F. S. Nery. 1999. Wine analysis and production. Aspen Publishers, New York, N.Y.