**3. Application of mtDNA-RFLP as a rapid method for monitoring the inoculated yeast strains in wine fermentations**

Although PFGE has been reported to be the most efficient in discriminating between different strains of *S. cerevisiae*, the mtDNA-RFLP technique is frequently used to differentiate between yeast isolates of the same species (González et al., 2007) because it enables a larger number of strains to be analyzed in a shorter time; it is a fast, simple, reliable and economic method, which does not require sophisticated material or specialized personnel (Fernández-Espinar et al., 2006). For these reasons, it is a very suitable technique for use by industry.

Most of the mitochondrial DNA in yeasts does not code proteins, and contains a high proportion of AT bases. Analysts can take advantage of this characteristic to characterize yeasts; it involves measuring variation in sequences in the mtDNA affecting the restriction sites of several endonucleases. Endonucleases such as *Alu*I, *Hinf*I or *Rsa*I, recognise the very frequent restrictions in the chromosomal DNA but not in the mitochondrial DNA, leading to a total cleavage of the chromosomal DNA in small pieces. These pieces can be easily differentiated from the mitochondrial fragments, which appear as bands with an electrophoretic mobility corresponding to molecules greater than 2 kb, generating polymorphisms that allow the characterization between yeast strains.

In previous studies applying this technique it has been demonstrated that the population of a fermentation vessel is "taken over" by wild yeasts, which displace the inoculated yeast strain, reducing it to a minority presence (Esteve-Zarzoso et al., 2000; Lopes et al., 2007; Raspor et al., 2002). In our research, when we have analyzed the inoculated fermentation of white wine as described above, we have found several examples of real situations that led to a significant decrease in the proportion of the inoculated strain (pattern V) and, in consequence, the quality of the wine was reduced. In order to minimize the impact of unwanted yeasts, wineries need a simple method for rapid diagnosis of the degree of dominance of inoculated strains, a method that could be performed routinely during the fermentation process (Ambrona et al., 2006, López et al., 2003). With this object we have used RFLP analysis of mtDNA for the rapid monitoring of the dominance, or otherwise, of inoculated yeast strains in industrial fermentations of white and red wines in a winery in southern Spain (Rodriguez et al., 2011).

We apply this technique directly to samples of fermenting wine without previously isolating yeast colonies. For white wine fermentations, a rapid assay is performed consisting of taking a sample of fermenting must, purifying the DNA from harvested cells, and obtaining the restriction patterns by digestion with endonuclease *Hinf*I. The same protocol is applied to red wine fermentation, but an overnight cultivation step is added before purification of the DNA (Figure 5).

The criterion for considering the result of the rapid test to be positive was obtaining restriction patterns of mtDNA that were identical for the total cells and the inoculated strain; when this is the case, the starter yeast can be taken as being dominant in the fermentation. The result was considered negative when additional bands, or absence of bands, were observed in the patterns; in this case neither the dominance, nor even the presence, of the inoculated yeast strain can be assured.

Application of Gel Electrophoresis Techniques to

displacing other wild yeasts.

**MW** P5

lambda-*Hind*III molecular marker.

wild yeast population was dominant.

for the inoculation of other vessels.

**A**

2027 2322

the Study of Wine Yeast and to Improve Winemaking 13

positive or negative. Figure 6 shows examples of positive and negative results for the last two years (2009 and 2010) for the dominance and non-dominance respectively of the

For the 2009 vintage, we tested several different vessels at the initial phase of the fermentation, for white wine (Panel A). The results were positive for all cases after applying the rapid test, i.e. the restriction patterns of the samples and the inoculated strain were identical, and the strain P5 was responsible at the beginning of the fermentation process

**MW** P5 F G B

**B**

2027 2322

inoculated strain P5 in white wine of the same winery described above.

A GH I KQR T F

Fig. 6. Rapid test based on mtDNA-RFLP with *Hinf*I of samples from white wine

fermentations which were inoculated with autochthonous yeast strain P5. Panel A shows results of samples taken in 2009 from vessels A, F, G, H, I, K, Q, R and T at the initial fermentation phase (11-7 ºBé). Panel B shows results of samples taken in the 2010 vintage from vessels F, G and B during the main phase of the fermentation (5-3 ºBé). MW is the

Only the fermentation vessel reference T was considered negative for the RFLP test at the beginning of the process. In this vessel evidence was observed of spontaneous fermentation before the inoculation, due to the conditions in which the must was stored. When this must was inoculated, the strain P5 did not implant successfully and it was concluded that another

In situations like this, the winemaker can take the decision not to use the fermenting must

Panel B (Figure 6) shows examples of the negative results of the RFLP test in three vessels mid-way during the fermentation process. In these deposits the population was perhaps similar because the restriction pattern of the total cells in each sample was similar. After the vintage, 20 colonies were isolated from the same sample previously analyzed by rapid test in the vessel G (Figure 6, panel B) in order to confirm the composition of the yeast population by karyotype. Surprisingly, all the clones show the same pattern as a commercial yeast strain used for the inoculation in the fermentation of another type of white wine in the previous year. It is assumed that this commercial yeast was also dominant in the fermentations sampled in vessels F and B. In previous studies researchers have reported the risks in using commercial yeast, because they can become part of the microbiota of the

bp bp

Fig. 5. Methodology for the rapid test assays by restriction analysis of the mtDNA.

This methodology of the RFLP test has been applied since 2005 in a winery of SW Spain and the results were obtained 11 and 23 hours after sampling, for white and red wine respectively (Rodríguez et al., 2011). If the wine-producer knows whether or not the presence of the inoculated yeasts has suffered a sudden decrease, and the inoculated strain is no longer dominant, in any phase of the fermentation process, a rapid intervention can be made. Since this year the winery has only used the selected autochthonous strain P5 for the inoculation of its white wine fermentations; this strain was in a clear majority in spontaneous fermentations, as reported above, and it shows karyotype pattern V. The peculiarities of the inoculation of industrial vessels for this wine have been described above (section 2). After applying the RFLP test, the correct course of the fermentation of all starters was assured before the inoculation of the industrial vessels. Generally, all the fermentations are tested in at least two different phases of the process: first after refills with fresh must and again when the fermentations are finished.

In addition, the results were checked using PFGE to validate the previous results obtained from the RFLP test. For this validation, 34 samples tested by RFLP were analyzed by electrophoretic karyotype of 323 colonies for white wine. The results indicated that when RFLP test was positive, the inoculated strain was present in the fermentation at 64% or more. When the RFLP tests were negative it was confirmed by PFGE that the starter yeast was present at only 60% or less.

The red wine of the winery was fermented in stainless steel vessels of 27.000 l and no refills of must were carried out. The fermentations were inoculated with several commercial active dried wine yeasts (ADWY) by hydration following manufacturer's instructions. For this wine, for 331 colonies analyzed by PFGE, the results of the RFLP test correctly predicted the results obtained later by applying PFGE. However, in this case, the limit found for white wine cannot be established because, from the positive results obtained by applying the RFLP test, the presence of the inoculated strain was greater than 75%, and all the negative results were at 55% or less (Rodríguez et al., 2011). Nevertheless, further experiments will be necessary to confirm these correlations because the RFLP test shows qualitative results and the actual percentage implantation of the starter yeast cannot be known when the results are

White wine

Red wine

Clean biomass

Washing

Collected cells

DNA purification

Centrifugation (650x g, 5 min) and collected cells

250 mL fermentation sample

again when the fermentations are finished.

was present at only 60% or less.

Fermenter

Agarose gel electrophoresis

+ + + +

Fig. 5. Methodology for the rapid test assays by restriction analysis of the mtDNA.

This methodology of the RFLP test has been applied since 2005 in a winery of SW Spain and the results were obtained 11 and 23 hours after sampling, for white and red wine respectively (Rodríguez et al., 2011). If the wine-producer knows whether or not the presence of the inoculated yeasts has suffered a sudden decrease, and the inoculated strain is no longer dominant, in any phase of the fermentation process, a rapid intervention can be made. Since this year the winery has only used the selected autochthonous strain P5 for the inoculation of its white wine fermentations; this strain was in a clear majority in spontaneous fermentations, as reported above, and it shows karyotype pattern V. The peculiarities of the inoculation of industrial vessels for this wine have been described above (section 2). After applying the RFLP test, the correct course of the fermentation of all starters was assured before the inoculation of the industrial vessels. Generally, all the fermentations are tested in at least two different phases of the process: first after refills with fresh must and

In addition, the results were checked using PFGE to validate the previous results obtained from the RFLP test. For this validation, 34 samples tested by RFLP were analyzed by electrophoretic karyotype of 323 colonies for white wine. The results indicated that when RFLP test was positive, the inoculated strain was present in the fermentation at 64% or more. When the RFLP tests were negative it was confirmed by PFGE that the starter yeast

The red wine of the winery was fermented in stainless steel vessels of 27.000 l and no refills of must were carried out. The fermentations were inoculated with several commercial active dried wine yeasts (ADWY) by hydration following manufacturer's instructions. For this wine, for 331 colonies analyzed by PFGE, the results of the RFLP test correctly predicted the results obtained later by applying PFGE. However, in this case, the limit found for white wine cannot be established because, from the positive results obtained by applying the RFLP test, the presence of the inoculated strain was greater than 75%, and all the negative results were at 55% or less (Rodríguez et al., 2011). Nevertheless, further experiments will be necessary to confirm these correlations because the RFLP test shows qualitative results and the actual percentage implantation of the starter yeast cannot be known when the results are


\_ \_ \_ \_

Restriction (*Hinf*I)

Eliminate the must remain

Spread 300 L of sample onto YPD agar plate. Incubate 24 h, 28 ºC

positive or negative. Figure 6 shows examples of positive and negative results for the last two years (2009 and 2010) for the dominance and non-dominance respectively of the inoculated strain P5 in white wine of the same winery described above.

For the 2009 vintage, we tested several different vessels at the initial phase of the fermentation, for white wine (Panel A). The results were positive for all cases after applying the rapid test, i.e. the restriction patterns of the samples and the inoculated strain were identical, and the strain P5 was responsible at the beginning of the fermentation process displacing other wild yeasts.

Fig. 6. Rapid test based on mtDNA-RFLP with *Hinf*I of samples from white wine fermentations which were inoculated with autochthonous yeast strain P5. Panel A shows results of samples taken in 2009 from vessels A, F, G, H, I, K, Q, R and T at the initial fermentation phase (11-7 ºBé). Panel B shows results of samples taken in the 2010 vintage from vessels F, G and B during the main phase of the fermentation (5-3 ºBé). MW is the lambda-*Hind*III molecular marker.

Only the fermentation vessel reference T was considered negative for the RFLP test at the beginning of the process. In this vessel evidence was observed of spontaneous fermentation before the inoculation, due to the conditions in which the must was stored. When this must was inoculated, the strain P5 did not implant successfully and it was concluded that another wild yeast population was dominant.

In situations like this, the winemaker can take the decision not to use the fermenting must for the inoculation of other vessels.

Panel B (Figure 6) shows examples of the negative results of the RFLP test in three vessels mid-way during the fermentation process. In these deposits the population was perhaps similar because the restriction pattern of the total cells in each sample was similar. After the vintage, 20 colonies were isolated from the same sample previously analyzed by rapid test in the vessel G (Figure 6, panel B) in order to confirm the composition of the yeast population by karyotype. Surprisingly, all the clones show the same pattern as a commercial yeast strain used for the inoculation in the fermentation of another type of white wine in the previous year. It is assumed that this commercial yeast was also dominant in the fermentations sampled in vessels F and B. In previous studies researchers have reported the risks in using commercial yeast, because they can become part of the microbiota of the

Application of Gel Electrophoresis Techniques to

abundance and diversity.

phenotype.

the Study of Wine Yeast and to Improve Winemaking 15

200 different PTMs have been described (including phosphorylation, methylation, acetylation, etc.) that transform each single gene into tens or hundreds of different biological functions. Before the advances made in proteomics, the differential analysis of the genes that were expressed in different cell types and tissues in different physiological contexts was done mainly through analysis of mRNA. However, for wine yeast, it has been proved that there is no direct correlation between mRNA transcripts and protein content (Rossignol et al., 2009). It is known that mRNA is not always translated into protein, and the amount of protein produced by a given amount of mRNA depends on the physiological state of the cell. Proteomics confirms the presence of the protein and provides a direct measure of its

In terms of methodology, proteomics approaches are classified in two groups: (i) gel-free systems based on the use of various chromatography methods; and (ii) gel-based methods that use mainly two-dimensional polyacrylamide gel electrophoresis (2DE). This latter approach will form the focus of our discussion here, given the subject matter of this book. As a succinct summary, the typical workflow of a proteomic experiment begins with the experimental design. This must be studied in depth, and it will delimit the conclusion obtained, even more so when comparisons are made between two strains, cultures or physiological stages, among others. As an optimum, only one factor among the various different assayed conditions must change (Fernandez-Acero et al., 2007). Several biological replicates, usually from 3 to 5, will be required depending on the strategy adopted. The next key step is to obtain a protein extract of high enough quality to separate complex mixtures of proteins. Usually, the protein extraction is done in sequential steps. First the tissue, cells, etc. are ruptured using mechanical or chemical techniques. Then, proteins are precipitated and cleaned. Most existing protocols use acetone and trichloroacetic acid. In the next step the proteome is defined and visualized using electrophoretic techniques. 2DE has been widely used for this purpose. Using this technique, proteins are separated using two different parameters. In the first dimension, proteins from the purified extract are separated by their iso-electric point using an iso-electrofocusing (IEF) device. Then, the focused strips are loaded in a polyacrylamide gel where the proteins are separated by their molecular weight. This system allows the separation of hundreds of proteins from one complex mixture. The gels are visualized with unspecific (Comassie, Sypro, etc.) or specific (e.g. Phospho ProQ diamond) protein stains. The gels are digitalized and analyzed with specific software to reveal the significant spots. These spots are identified using mass spectrometry; commonly, for 2DE approaches, MALDI TOF/TOF is used. The huge list of identified proteins obtained is studied to discover the biological relevance of each identification.

In spite of the many achievements of proteomics, only a few proteomic studies have been carried out on wine yeast, whereas mRNA expression has been widely used to study a broad range of industrial conditions. However, Rossignol et al. (2009) show that substantial changes in protein levels during alcoholic fermentation are not directly associated with changes in the transcriptome; this suggests that the mRNA is selectively processed, degraded and/or translated. This conclusion is important: it is the proteome, not the genome nor the transcriptome, that is the relevant level of analysis for understanding the adaptations of wine yeasts during alcoholic fermentation, since these are responsible for the

The usual strategy for wine production is the inoculation of selected yeast strains into the must, decreasing the lag phase, a quick and complete fermentation of the must, and a high

winery, effectively creating their own ecosystem, and can subsequently be predominant in the fermentations (Santamaría et al., 2005). In our study, we think that this commercial yeast was present in equipment which was not properly cleaned. When the wine-producer was more careful in the next vintage, there were no problems of contaminations by commercial yeast, and the dominant yeast in the fermentations was the inoculated autochthonous yeast P5 (data not shown). Therefore, it was confirmed that the commercial yeast had not acquired an ecological niche because it presumably did not adapt well to the ecosystem of a properly-cleaned winery.

As stated, the results of the RFLP can be obtained 11 and 23 hours after taking the sample for white and red wine respectively. However, this time can be shortened further, because it depends on the method used to rupture cells, on the number of samples analyzed per day, and on whether the samples contain a greater amount of must residues. In the case of red wine, there was another problem in shortening the test time, because the residues were difficult to clear by centrifugation; we think that some compounds remaining in the digested DNA samples were inhibitory for the endonuclease. Therefore, a step has been added in the protocol of the rapid test (Figure 5) in which the sample of the red must is plated on YPDagar and incubated overnight at 28 ºC. Nevertheless, we think that, for red wine, the time taken to obtain the results could also be shortened further, like that for white wine, if the clean biomass can be separated from the must residues in a few minutes. To achieve this, further experiments will need to be carried out.
