Abstract
Two molecular methods, interdelta polymorphism fingerprinting of genomic DNA and COX1 intron polymorphism fingerprinting of mitochondrial DNA, were used to characterize 52 strains of Saccharomyces identified to species by 26S rDNA sequencing only in order to assess the potential of these techniques for strain typing and genetic analysis individually and in combination. Two laboratory isolates derived from the same parentage served as a control. Forty-seven S. cerevisiae strains representing a mix of commercial strains and vineyard or winery isolates were used. Five of these strains have been given the commercial designation S. cerevisiae race bayanus. Three strains were identified as S. bayanus, S. kudriavzevii, and S. servazzii from their 26S rDNA sequence and thus may be hybrids. Forty-four genetic patterns were found among the strains by interdelta polymorphism fingerprinting and 47 genetic patterns were found by COX1 intron polymorphism fingerprinting. Each method allowed differentiation of the majority of strains but grouped strains into different clusters. Strains were clustered into eight groups using both techniques in combination. The two laboratory strains clustered together but were in a larger grouping of wine strains. Saccharomyces bayanus, S. kudriavzevii, and S. servazzii were separated from each other but each clustered with S. cerevisiae strains and could not be differentiated from S. cerevisiae by either method. Three of the S. cerevisiae race bayanus strains clustered together but the other two were in different groupings. Of the 10 commercial strains, 904 (French Red) was grouped alone, while 522 (Montrachet) and 905 (Premier Cuvee) were grouped with wild isolates from different areas. Thus both methods reveal patterns of strain similarity but neither method differentiated strains by species or by origin.
- Saccharomyces
- genetic relatedness
- interdelta polymorphism fingerprinting
- COX1 intron polymorphism fingerprinting
Precise identification of commercial and native isolates of Saccharomyces strains is very important in winemaking because of the impact of yeast strain on wine composition and fermentation progression. Use of strains selected as pure culture inocula is a widespread practice in the wine industry where strains impart different characteristics to the final product (Dequin 2001, Cocolin et al. 2004, Romano et al. 2008). Wines can also be made using autochthonous microbiota of the grapes and wineries and fermentation will eventually be dominated by S. cerevisiae strains. Often a mixture of species is found in autochthonous fermentations and each of those local strains may contribute differently to the final flavor profile of the wine (Howell et al. 2006). Autochthonous fermentations may also be conducted by multiple strains of S. cerevisiae with differing metabolic contributions (Torija et al. 2001, Wang and Liu 2013), and profiling the strains present will allow replication of the fermentation profile and identification of those strains most important to the desired character of the wine. Furthermore, strain typing of Saccharomyces is a prerequisite for the study of yeast strain population dynamics during the wine fermentation process (Nadal et al. 1999, Lopes et al. 2002, Cappello et al. 2004).
Many methods have been developed for and applied to the identification of Saccharomyces strains. These methods assess differences in the sequence and adjacencies of genetic information of either genomic (nuclear) or mitochondrial DNA. Genomic and mitochondrial DNA represent unique heritable lineages and are subject to different mutational and recombinational forces. Thus it is not clear which of these DNA-containing elements best reflects true strain relatedness and diversity. One goal of this study was to compare differences in genomic and mitochondrial DNA to determine if strains would appear in similar or different groupings by the two approaches. If strains grouped similarly by both approaches, then the mutagenic forces operative during strain evolution would be expected to impact both genomes equivalently. If there is not good agreement among groupings, then the adaptive forces impacting the two genomes would be more independent.
Karyotyping based on chromosome length polymorphism or mitochondrial DNA polymorphism are more widely used because of their accuracy (Querol et al. 1992, Versavaud et al. 1995, Nadal et al. 1996, Zerva et al. 1996, McCullough et al. 1998, Comi et al. 2000, López et al. 2001), but these can be cumbersome and time-consuming. Thus, several polymerase chain reaction (PCR)-based techniques have been proposed with the advantage of simplicity, rapidity, and availability. These techniques include random amplified polymorphic DNA (RAPD) (McGrath et al 1998, Martínez et al. 2007, Romano et al. 2008, Bovo et al. 2009), COX1 intron polymorphism fingerprinting (de Barros Lopes et al. 1996, López et al. 2003), interdelta polymorphism fingerprinting (Ness et al. 1993, Legras and Karst 2003), and application of microsatellites (Legras et al. 2005, Jubany et al. 2008, Richards et al. 2009). Interdelta polymorphism fingerprinting was developed based on the variation of the number and position of the delta element, a repeated sequence which flanks the Ty1/Ty2 retrotransposon (Ness et al. 1993), and has been commonly applied for monitoring strain diversity of S. cerevisiae as a method with high differentiation level (Schuller et al. 2004, Ayoub et al. 2006, Charpentier et al. 2009, Mercado et al. 2010, Xufre et al. 2011, Wang and Liu 2013). COX1 intron polymorphism fingerprinting is based on the variation in the number and position of mtDNA COX1 introns, and has the potential of discriminating strains of more species (López et al. 2003). The variation of COX1 introns is not limited to S. cerevisiae and is also present in strains of some non-Saccharomyces yeast (Foury et al. 1998). However, this technique has not been widely applied to the analysis of strain diversity in Saccharomyces, and no study has been conducted to compare the accuracy of the interdelta polymorphism fingerprinting and COX1 intron polymorphism fingerprinting on the discrimination of Saccharomyces yeast at strain level.
In the present study, interdelta polymorphism fingerprinting of nuclear DNA and COX1 intron polymorphism fingerprinting were used to perform strain typing of S. cerevisiae in the University of California (UC) Davis Department of Viticulture and Enology Culture Collection. Dendrograms were constructed based on similarity among different patterns of bands and the genetic relationships of all strains were evaluated by combination of the two techniques. The aim of this work was to examine the differentiation accuracy of interdelta polymorphism fingerprinting and COX1 intron polymorphism fingerprinting and to provide available methodologies for wine strain diversity analysis of commercial and native isolates of S. cerevisiae. Since one method is reliant on differences in nuclear genetic information and the other on differences in mitochondrial genomes, the combined use of both methods was expected to allow the greatest discrimination of strains.
Materials and Methods
The yeast strains used in this study (Table 1) were obtained from the UC Davis Department of Viticulture and Enology Culture Collection. Wild or noncommercial isolates of S. cerevisiae comprised the bulk of the strains tested; however, two common laboratory strains, 10 commercial strains (including five that are marketed as S. cerevisiae race bayanus), and three non-cerevisiae strains were also tested. Strains were maintained as glycerol stocks stored at −80°C. The species identity of all strains was determined by sequence analysis of the NL1 and NL4 region of 26S rDNA (Kurtzman and Robnett 1998). Use of 26SrDNA sequence analysis does not unequivocally allow differentiation of pure and hybrid strains, and thus some of these strains may be hybrids of one or more species of Saccharomyces.
DNA isolation.
Yeast cells were cultured in 3 mL YPD (10 g/L yeast extract, 20 g/L peptone, and 20 g/L glucose) (Difco, Thermo Fisher Scientific, Waltham, MA) for 40 hr at 28°C with shaking at 160 rpm, and total DNA isolation was performed using the Master Pure Yeast Purification kit (Epicentre Biotechnologies, Madison, WI). DNA was quantified using a DN-1000 Spectrophotometer (NanoDrop, Thermo Scientific, West Palm Beach, FL).
Interdelta polymorphism fingerprinting.
PCR amplification was carried out in 25 μL reaction volumes containing PCR buffer (10 mmol/L Tris pH 8.4, 50 mmol/L KCl), 30 to 100 ng yeast DNA, 200 μmol/L dNTPs, 2.5 mmol/L MgCl2, 0.5 μmol/L of each oligonucleotide primer δ12 (5′-TCAACAATGGAATCCCAAC-3′) and δ21 (5′-CATCTTAACACCGTATATGA-3′) (Legras and Karst 2003), and 2.0 U Taq polymerase. Amplification reactions were performed in a Peltier Thermal Cycler (PTC-200; MJ Research, Reno, NV) as described by Legras and Karst (2003).
COX1 intron polymorphism fingerprinting.
The COX1 introns were amplified using primers of 3L (5′-GCTTTAATTGGWGGWTTTGG-3′), 3R (5′-ATTGTCATACCATTTGTYCTYAT-3′), 4L (5′-GAAGTAGCAGGWGGWGGWGA-3′), and 5R (5′-GTTAGCTAAGGCWACWCCWGT-3′) (López et al. 2003). PCR reactions were run in a final volume of 25 μL containing 10 to 250 ng DNA, 0.1 μM of each oligonucleotide primer, 80 μM dNTPs, and 0.5 U Taq DNA polymerase, Taq buffer (10 mM Tris pH 8.4, 50 mM KCl; Life Technologies, Invitrogen, Grand Island, NY), 0.1% Triton X-100, and 1.5 mM MgCl2. All the primers were run in a same reaction. Amplifications were carried out using a Peltier PTC-200 Thermal Cycler with thermal cycling parameters described by López et al. (2003).
Electrophoresis and cluster analysis of strains.
The DNA fragments were separated on a 1.5% (w/v) agarose gel or 1.5% agarose and Synergel (BioAmerica Inc., Miami, FL) (1:1) containing ethidium bromide, at 100 V for 3 hr in 0.5 × TBE buffer, visualized and photographed. Electrophoresis patterns of both methods were analyzed by GeneTools software (SynGene, Frederick, MD). Dendrograms were constructed using unweighted pair group method with arithmetic mean (UPGMA) methods based on Euclidean distance by DPS 7.05 software (Huaiyin Normal University, Huaian Jiangsu, and Nanjing University of Science and Technology, Jiangsu Nanjing, China).
Results
Characterization of Saccharomyces strains.
Forty-four genetic patterns were found among 52 strains by interdelta polymorphism fingerprinting (Figure 1A). Most strains displayed unique interdelta patterns with the exception of six pair of S. cerevisiae strains and a set of three commercial S. cerevisiae race bayanus strains that exhibited the same patterns. These patterns were divided into seven clusters (Figure 2). Diversity in the COX1 region of the mitochondria was slightly more discriminatory, and 47 genetic patterns were found (Figure 1B). Five pair of S. cerevisiae strains gave identical patterns. Strains were divided into four clusters (Figure 2). Strains 685 (S. bayanus), 855 (S. kudriavzevii), and 2211 (S. servazzii) exhibited unique interdelta and COX1 patterns, such as the interdelta pattern of 685 which had no bands greater than 900 bp. However, this kind of band pattern was not sufficient to indicate distinct species and separation from the other S. cerevisiae strains in this analysis. These strains may be hybrids containing S. cerevisiae DNA and thus not distinguishable by these methods as distinct species. Alternately these methods may be too sensitive and overemphasize strain differences, thus obscuring species identity.
A comparison of electrophoresis patterns from both methods showed that variation in bands less than 300 bp by COX1 intron polymorphism fingerprinting was less than that shown by interdelta polymorphism fingerprinting. Two sets of S. cerevisiae strains, 49 and 76 and 933 and 936, exhibited the same patterns by both methods. Other strains with the same pattern by one method were differentiated by the other method, suggesting that the nuclear and mitochondrial genomes evolved separately.
Genetic relatedness assessment.
The dendrogram deduced by interdelta polymorphism fingerprinting (Dendro-InterD) and COX1 intron polymorphism fingerprinting (Dendro-COX1) (Figure 2) revealed seven and four groupings, respectively, when the U distance was 3.75. Each technique contained one large grouping: 34 strains for interdelta polymorphism and 40 for COX1 intron polymorphism, representing 65% and 77% of the total number of strains evaluated, respectively. Strains were not grouped identically by the two methods. For example, the three putative non-cerevisiae strains grouped together in group 1 of Dendro-InterD, while two, 685 (S. bayanus) and 855 (S. kudriavzevii), were still clustered together in group 2 of Dendro-COX1 but 2211 (S. servazzii) was placed in group 1. Similarly, strains in group 2 of Dendro-COX1 were scattered among groups in Dendro-InterD but did show some cluster similarity, such as with strains 522, 957, 952, 956, 514, and 518.
The combination dendrogram of the two techniques allowed for differentiation of nearly all strains evaluated (Figure 3), with the largest group containing only 18 strains (35% of the total number of strains tested). When the U distance was 5.05, eight groups were clustered, containing 6, 3, 7, 18, 14, 2, 1, and 1 strain(s), respectively. To assess the ability of strains to be grouped by geographic origin, the set of tested strains consisted of subgroups of isolates from the same region. Four strains were noncommercial isolates from California, six were vineyard or winery isolates from Spain, and 27 were from different vineyards and wineries in Italy. Although some strains in each of these geographic sets clustered by both methods, others were in different groupings. For the four wild S. cerevisiae isolates from California, 49 (citrus fermentation isolate) and 76 (unknown isolates from Mrak) were clustered in group I together with S. servazzii 2211 from California; 587 (a winery isolate) was clustered in group III together with the two common lab strains (2515 and 2516); and 2118 (wine isolate) was included in group VI. Of the six S. cerevisiae isolates from Spanish wine, five were in the group IV cluster. The other isolate, 542, was in group V. For the large group of S. cerevisiae isolates from Italy, 11 isolates from six different varieties were contained in group V with commercial S. cerevisiae race bayanus 905 and putative S. kudriavzevii 855 (both from France); 10 isolates from five different varieties were included in group IV with early commercial S. cerevisiae 522 from France and S. bayanus 685 from northern Europe; 943 and 944 isolated from Trebbiano were clustered in group I together with commercial S. cerevisiae race bayanus 2032 from France; 933 and 936 were clustered in group III together with two commercial S. cerevisiae strains from France (2031 and 813); 948 isolated from Gutturnio was included in group VI with French commercial S. cerevisiae 713; and 935 isolated from Lumbrusco in group VII alone. Commercial S. cerevisiae 904 from France was in group VIII alone, and group II consisted of the other three commercial S. cerevisiae race bayanus from France (819, 777, and 969). Thus, there was evidence of clustering by region of isolation, but that was not absolute; region of origin could not be defined by the patterns obtained using either method alone or in combination.
Discussion
Accurately defining strain relatedness can provide important information on expected phenotypes and strain evolution as well as identify the possible roles of regional wine isolates in regional wine character. The choice of heritable element to be evaluated can impact the assessment of strain similarity. In Saccharomyces, either nuclear (genomic) DNA or mitochondrial DNA can be used to interpret strain similarities and evolutionary or adaptive distance. Mitochondrial and nuclear genomic elements are subject to differences in frequency of genetic change and in the mechanisms that can induce changes in primary sequence and respond to different adaptive forces in the native environment. The robustness of use of each of these genomes for detection of strain differences was compared with the utility of combining both sets of sequence divergence information in the statistical analysis of relatedness.
Interdelta polymorphism fingerprinting with improved primers (Legras and Karst 2003) has been commonly used to monitor strain diversity of S. cerevisiae (Schuller et al. 2004, Charpentier et al. 2009, Mercado et al. 2010, Xufre et al. 2011, Wang and Liu 2013) because it is rapid, simple, and of high discrimination ability (Ayoub et al. 2006). In response to two disadvantages of interdelta polymorphism fingerprinting, the poor reproducibility of band patterns with primary primers (Ness et al. 1993) and the limitation of discriminating only some Saccharomyces strains (Legras and Karst 2003), COX1 intron polymorphism fingerprinting was developed by López et al. (2003). However, this technique has not been widely used in corresponding analysis and no comparison study on the two techniques has been reported. In our study, the two techniques proved to be effective PCR-based methods and also gave results that complemented each other. Some strains were only distinguished by one method while others were easily distinguished with either method.
In contrast to the report of Legras and Karst (2003) where no clear band patterns were shown by S. bayanus and S. servazzii using interdelta polymorphism fingerprinting, clear bands were found in all putative non-cerevisiae strains by both techniques in our study. However, the band patterns for non-cerevisiae strains in our study were not specific enough for species differentiation, consistent with their observations of the inability of interdelta polymorphism fingerprinting to identify strains by species. Our data show that the COX1 intron polymorphism fingerprinting likewise will not differentiate strains by species. Furthermore, the differentiation by both methods cannot preclude the possibility that the strains are naturally occurring hybrids between Saccharomyces sensu stricto species, as has been determined for other European fermentation isolates (Naumov et al. 2002, González et al. 2006).
By genetic relationship assessment, both techniques indicated corresponding grouping results according to the same U distance. However, the combination of the two methods seemed to be more effective than either method alone. Thirty-four strains were clustered in group 1 of Dendro-InterD and group 2 of Dendro-COX1 contained 40 strains (Figure 2), yet the largest group by the combination of techniques contained only 18 strains (Figure 3). Thus, the use of divergence based upon both the nuclear and mitochondrial genomes was more discriminatory than either method alone, which is not surprising given that the movement of transposons in the nuclear genome and genetic rearrangements in mitochondrial DNA are highly independent processes. The relationship of strains can be observed by using the combination of both techniques. For example, 49 and 76 exhibiting no difference in either grouping may be from the same genetic origin, consistent with both being isolates from California prior to the advent of commercial yeast strains.
The assessment of the genetic relationship indicated that three putative non-cerevisiae strains and five commercial S. cerevisiae race bayanus strains clustered with other S. cerevisiae strains. They were separated from each other in the cluster analysis, with the exception of 819 (Prise de Mousse), 777 (EC1118), and 969 (T73). These three commercial strains may be of similar parentage. Of the 10 commercial strains, UCD522 (Montrachet) and UCD905 (Premier Cuvee) showed a closer relationship with wild isolates from different areas than with the other commercial strains.
Geographic location and isolation are both thought to play a significant role in genetic divergence. For Saccharomyces it has been shown that this hypothesis may not hold true for regions that are fairly close geographically (Versavaud et al. 1995) but appears to be true in widely distant geographic regions (Goddard et al. 2010). However, our study showed that divergence in Saccharomyces strains may be more greatly associated with yeast propagation and selective pressures than with geographic factors. This study did not look at the possibility that some of the strains analyzed may be naturally occurring hybrids of Saccharomyces sensu stricto species. Further analysis by karyotyping would be necessary to determine the presence of hybrids in these groups.
Conclusion
The combination of interdelta polymorphism fingerprinting and COX1 intron polymorphism fingerprinting was an effective tool for typing of Saccharomyces strains and genetic diversity analysis. All 52 strains were differentiated from each other by use of these two methods simultaneously and eight genetic groups were clustered, revealing strain relatedness information. Thus, the combined analysis of genetic changes in both nuclear and mitochondrial genomes provides a stringent assessment of strain similarity.
Acknowledgments
Acknowledgments: The authors thank the National Natural Science Fund Program ( 31271917) and China Agriculture Research System (no. CARS-30-ch-03) for their financial support.
- Received May 2013.
- Revision received August 2013.
- Accepted September 2013.
- Published online February 2014
- ©2014 by the American Society for Enology and Viticulture
Literature Cited
Sign in for ASEV members
ASEV Members, please sign in at ASEV to access the journal online.
Sign in for Institutional and Non-member Subscribers
Log in using your username and password
Pay Per Article - You may access this article (from the computer you are currently using) for 2 day for US$10.00
Regain Access - You can regain access to a recent Pay per Article purchase if your access period has not yet expired.