Analysis of Saccharomyces During Normal and Problem Fermentations

This is the second year of a three year proposal. In this past year, the comparative analysis of the microarray and proteome technologies was completed. For several reasons, the proteome analysis was found to be ideal for profiling the physiological status of yeast during enological fermentations. The microarray analysis will not allow comparison of expression patterns across yeast strains and growth conditions as most of the genes and proteins of interest can not be quantified by this technique. However, the microarray analysis of the French White strain under nitrogen sufficient and limiting conditions yielded valuable insights into the metabolic state of the strains. Contrary to our initial hypothesis, the low nitrogen culture does not readily enter a non-proliferative phase and rather shows a failure to adapt to grape juice conditions. In contrast, metabolic pathways associated with ethanol tolerance, are expressed in the nitrogen sufficient culture before the onset of maximal fermentation rate. During this period, the proteome gel protocols have been optimized. Comparison of the proteomes of French White and Montrachet under nutrient sufficient conditions has revealed many differences in protein patterns. For the second objective, we have experimentally verified the mechanistic model that we developed in the previous year. We have also examined the effects of juice, yeast, and initial oxygen content on maximum cell concentration attained and on fermentation rate. In all cases, cell concentration was critical. In fact, we have built a correlation between maximum viable cell concentrations (mvcc) (reached at 24 – 48 hours) and fermentation rate. Under a certain threshold of mvcc, all of our fermentations stuck, indicating that this parameter may be useful in the early prediction of sluggish or stuck fermentations. The technology for using artificial neural networks to predict kinetics has also been largely developed. We have established that neural networks show promise for prediction of wine fermentation kinetics, except in cases of extrapolation to conditions outside of the training data set when the accuracy of prediction fails. Much progress was made in the last year on methods development for direct analysis of microorganisms in wine and musts. An evaluation of the 26S ribosomal RNA gene demonstrated this locus to be more discriminatory than the 18S gene for differentiation of various wine yeasts. In addition, denaturing gradient gel electrophoresis using chemical denaturants instead of temperature (“TGGE”) was shown to provide better separation of PCR products amplified from wine yeasts. A significant achievement this year was the optimization of methods for isolation of microbial DNA directly from fermenting wine. This accomplishment made it possible to use PCR-DGGE to profile the yeast successions that occurred during a fermentation at a local winery. Moreover, sequence analysis of the DGGE bands obtained in this profile allowed identification of the yeasts involved to the genus and/or species level. Additional progress was made in laboratory fermentations where DGGE profiling was demonstrated to accurately depict the presence or absence of specific yeast strains above a threshold level (~104 cells per mL) from within a mixed culture. In addition, methods were developed for isolation of microbes from the grape surface. Initial PCR-DGGE analysis revealed different populations of fungal species present on grapes which had undergone different viticultural treatments. Methods were also optimized for discrimination of bacterial strains by PCR-DGGE and work is underway to optimize bacterial DNA purification from musts and wine. Finally methods are being developed to selectively identify and differentiate Saccharomyces species from within wine fermentations.

Impact of Fermentation Rate Changes on Hydrogen Sulfide Concentration in

The correlation between alcoholic fermentation rate, measured as carbon dioxide (CO2) evolution, and the rate of hydrogen sulfide (H2S) formation during wine production has a significant impact on the H2S content of a finished wine. Both rates and the resulting concentration peaks in fermentor headspace H2S are strongly impacted by yeast assimilable nitrogenous compounds in the grape juice. We have conducted a series of model fermentations in temperature-controlled and stirred fermentors using a complex model juice with a defined combination of ammonium ions and/or amino acids. Fermentation rate was measured indirectly by weighing the fermentors on a laboratory scale. This assumes that once CO2 saturation of the juice is reached, weight loss corresponds to CO2 evolution which in return is proportional to ethanol formation. H2S production was measured using a calibrated transparent tube packed with color-indicating metal acetate. The tube was inserted into a fermentor port instead of a regular gas lock, and provided quantitative trapping of H2S formed over time. Fermentation rates for CO2 and H2S as well as the relative ratios between them were calculated. The fermentations confirmed observations that high concentrations of yeast assimilable nitrogen do not necessarily protect against elevated H2S formation. High initial concentrations of ammonium ions via addition of diammonium phosphate can cause a higher evolution of H2S as in a non-supplemented but non-deficient juice. We observed that the availability of yeast assimilable amino acids, particularly arginine, can results in a more evenly distributed CO2 production throughout the alcoholic fermentation. In addition, relative maximum H2S evolution rates can be observed earlier in the fermentation, and CO2 produced during the remainder of the fermentation may sufficiently strip out initial sulfides.

Analysis of Saccharomyces During Normal and Problem Fermentations

The goal of this first year of the Long-Term Research Project was to compare and develop methodologies in three key areas: analysis of global gene expression in Saccharomyces in its native habitat of grape juice; refinement of the neural network technology for prediction of problem fermentations; development of a rapid method allowing profiling of the microbial composition of industrial samples. The RNA-based microarray methodologies worked well for samples of cells grown in juice-like synthetic media, but did not work well for samples prepared from cells grown in actual juices. Further, there does not appear to be a strong correlation between actual mRNA levels and protein content in the cells. Thus the proteome analysis seems to be most useful for profiling gene expression in industrial samples. However, the microarray data provided a wealth of information on which proteins to examine in the proteome gels and has given new insights into the physiological activities of the cells under stressful conditions leading to arrest of fermentation. Neural network training methods have been established for using historical fermentation kinetics data to predict sugar utilization rates based on juice characteristics and intended processing. Small-scale fermentations have been completed to find which critical inputs to use for the neural network prediction, as well as to validate a physical and mathematical model for cell growth and sugar utilization that is likely to direct future experimentation. Temperature gradient gel electrophoresis (TGGE) works well to differentiate yeast genera using the primers that we developed and can be used to assess the microbial purity of industrial samples to be used in the proteome analyses.

Development of Conjugal Systems for Gene Transfer and Allelic Exchange in

Oenococcus oeni (formerly Leuconostoc oenos) is a common starter culture used to achieve the malolactic fermentation (MLF) in commercial wine production. To date, basic research on O. oeni, especially molecular analysis of genes involved in growth on grape juice, has been hampered by the lack of efficient gene transfer and allelic exchange systems. One potential mechanism for genetic manipulation of starter cultures is the use of conjugal systems indigenous to lactic acid bacteria (LAB). Previous work on the conjugal element pRSOl from the dairy starter Lactococcus lactis identified several regions involved in conjugative transfer and localized the conjugative origin of transfer (oriT). In an effort to develop a generally applicable system for gene transfer, we examined the host range of pRSOl, and several oriT-containing plasmid derivatives, with a range of LAB recipients. Conjugal transfer was demonstrated into Leuconostoc mesenteriodes subsp. cremoris, Enterococcus faecalis, Streptococcus thermophilus, Pediococcus cerevisiae and 0. oeni. Broad host range transfer of pRSOl and mobilization of oriT-containing plasmid derivatives provides a novel means for molecular analysis of a variety of LAB, such as O. oeni, that are resistant to current methods of genetic manipulation. Current work is focused on the development of the appropriate mobilization and allelic exchange vectors for genetic analysis of O. oeni.

Factors Affecting Sugar Utilization and Rate of Fermentation During Vinification

The HXT Display technique has been successfully applied to cells isolated from white grape juice, and a preliminary analysis of the HXT genes expressed during grape juice fermentation has been completed. Of the 14 HXT genes that we were able to evaluate, 11 were found to be expressed in grape juice. This technique did not allow quantitation of the relative abundance of the messages for the expressed HXT genes, so we are developing an alternative method that will allow quantitation, based upon the unique sequences identified in the untranslated regions of the messages identified in the original HXT display analysis. The four genes which were not amplified in this analysis are being cloned in order to characterize the 3′ untranslated regions. The pathway for the internalization and turnover of the HXT2 protein has been evaluated. It was originally thought that if this pathway could be blocked, fermentations would not arrest prematurely. This does not appear to be a viable option for the elimination of fermentation problems. Cells not turning over Hxt2p appear to be sluggish in conducting the fermentation from the beginning. Further, blocking the pathway for internalization blocks internalization of other factors as well and seems to make the cells more sensitive to ethanol. Finally, the role of fatty acids in ethanol tolerance and stuck and sluggish fermentations has been evaluated. The requirement for unsaturated fatty acids has been demonstrated under enological conditions. Saturated fatty acids offer little benefit to the organisms. Fatty acid limitation appears to affect consumption of fructose more severely than that of glucose for reasons that are as yet obscure. This may be related to the type of HXT genes that are expressed upon fatty acid limitation. Surprisingly, fatty acid limited cultures maintained viability while stuck as long as sufficient glucose was available.

Factors Affecting Sugar Utilization and Rate of Fermentation During Vinification

In this grant year we have: 1) successfully developed the technology for the analysis of HXT gene expression; 2) developed methodology yielding high quantities of high quality mRNA from strains of Saccharomyces during grape juice fermentation; 3) begun profiling of H’XT gene expression during grape juice fermentation conducted by commercial strains; 4) confirmed the critical role of HXT2 in non-proliferative phase fermentation of glucose and fructose in grape juice; 5) defined the pathway by which the HXT2 transporter is degraded in response to nitrogen limitations. Further, we have shown that HXTl and HXT3 are expressed early during the fermentation of grape juice, with a shift to HXT2 and HXT3 upon entry into stationary phase of growth during the fermentation. We are now poised to undertake a more thorough investigation of expression of the remainder of the HXT genes during grape juice fermentation.

Studies on Stuck Fermentations and on Factors that Control the Rates of

We have found that the factor(s) that determine whether or not a fermentation will stick are vineyard-determined. If grapes from a given vineyard are crushed and the resultant must is placed in more than one fermentation tank, if one tank sticks all tanks from the same must also will stick. Stuck fermentations occur for all types of must and also occur for both inoculated and natural fermentations. We have found that viabilities of Saccharomyces cerevisiae cells in stuck fermentations are very low and this appears to be the reason the fermentations have stopped. We have also found that dilution of a stuck wine with water or even with a dry wine followed by addition of live yeast results in a reinitiation of fermentation which continues to dryness. We have considered six models that have been proposed to explain stuck fermentations. These models are: 1) nutrient depletion, 2) excess temperature during fermentation, 3) killer yeast, 4) toxin introduced with the grapes, 5) toxin produced by another microorganism and 6) toxin produced by the wine yeast. The dilution experiments eliminate model 1 and occurrence of stuck fermentations in temperature-controlled fermenters eliminates model 2. Experiments done by others eliminate model 3. The dilution experiments and the low viabilities of the wine yeast cells point to the presence of a toxin in the stuck wine. This toxin could be a fungicide brought in with grapes (model 4) or a toxin produced by another microorganism (model 5) or by the wine yeast (model 6). We believe model 5 may occur but is exceptional. Some fermentations become infected with acetobacter and the acetic acid kills or inhibits the wine yeast. We saw no other microorganisms in the 14 fermentations we studied and volatile acidity levels were near the normal range. We considered model 4 to be unlikely because one would expect the fungicide to act from the beginning of fermentation and this does not appear to be the case. Fermentation kinetics are normal until the time of sticking. This leaves model 6, a toxin produced by the wine yeast. This model was proposed several years ago by a group in Bordeaux and the toxin was proposed to be medium chain fatty acids, octanoic, decanoic and dodecanoic acid. These compounds are extremely toxic and are produced during fermentation. Several experiments done by others seem to lend strong support to this model. The goal of the study supported by AVF was to test this specific model. We studied 4 stuck fermentations and did gas chromatographic analyses of the stuck wine. We found that the concentrations of octanoic and decanoic acid were at or below normal levels in these stuck wines. This eliminates this specific model. We now believe that there may be another toxic compound produced by wine yeast and that this kills the cells and stops the fermentation. We are also now reconsidering model 4. If the fungicide brought in with the grapes needs to be modified during the fermentation process to be toxic to wine yeast this would explain the lag in onset of the stuck state. Glenn Andrade of Sutter Home Winery (personal communication) has found a correlation between spraying of the grapes with a fungicide and the occurrence of stuck fermentations.

The Extraction of Condensed Tannins in Red Wine Fermentations

The first year of this project was designed to survey the effects of a large number of different pomace maceration practices in commercial-scale fermentations on the extraction of anthocyanins and tannins. This year of the project focused on Pinot Noir and observed variables included cold soak, manual vs. mechanical punch down, and pump over vs. punch down. The wines were prepared at a cooperating winery according to an agreed-to protocol. The phenolic composition of the musts/wines were analyzed using both spectral (at the cooperating winery) and chromatographic methods (at UC Davis and ETS), including a new Silica gel based procedure which separates phenolics based on size. Five analyses were carried out on 9 wines although the focus was on 5 Pinot Noir wines from the Widoe’s Vineyard. The major difference was that the treatments of both mechanical punch down and pump over increased both colored and uncolored tannin, while the cold soak decreased these components. The pump over regime also increased monomelic color. The goals in this project included the following: 1) Develop a method for sample preparation and an HPLC method for the determination of condensed tannins. 2) Assist in the production of four lots of wine in each of two wineries. Production lots will include four fermentations each of a single vineyard Pinot Noir and Cabernet Sauvignon (fermentation conditions to be decided at a later date and with input from each winery.) 3) Through the course of each fermentation, pull samples and analyze condensed tannins by HPLC and information on color colorimetrically. Additional analyses (acetaldehyde, sugar, SO2) will be performed at the principal investigators’ discretion. 4) Conduct tastings of finished wines with production staff to associate style preferences with chemical data.

Studies on Stuck Fermentations and on Factors that Control the Rates of

We have found that the factor(s) that determine whether or not a fermentation will stick are vineyard-determined. If grapes from a given vineyard are crushed and the resultant must is placed in more than one fermentation tank, if one tank sticks all tanks from the same must also will stick. Stuck fermentations occur for all types of must and also occur for both inoculated and natural fermentations. We have found that viabilities of Saccharomyces cerevisiae cells in stuck fermentations are very low and this appears to be the reason the fermentations have stopped. We have also found that dilution of a stuck wine with water or even with a dry wine followed by addition of live yeast results in a reinitiation of fermentation which continues to dryness. We have considered six models that have been proposed to explain stuck fermentations. These models are: 1) nutrient depletion, 2) excess temperature during fermentation, 3) killer yeast, 4) toxin introduced with the grapes, 5) toxin produced by another microorganism and 6) toxin produced by the wine yeast. The dilution experiments eliminate model 1 and occurrence of stuck fermentations in temperature-controlled fermenters eliminates model 2. Experiments done by others eliminate model 3. The dilution experiments and the low viabilities of the wine yeast cells point to the presence of a toxin in the stuck wine. This toxin could be a fungicide brought in with grapes (model 4) or a toxin produced by another microorganism (model 5) or by the wine yeast (model 6). We believe model 5 may occur but is exceptional. Some fermentations become infected with acetobacter and the acetic acid kills or inhibits the wine yeast. We saw no other microorganisms in the 14 fermentations we studied and volatile acidity levels were near the normal range. We considered model 4 to be unlikely because one would expect the fungicide to act from the beginning of fermentation and this does not appear to be the case. Fermentation kinetics are normal until the time of sticking. This leaves model 6, a toxin produced by the wine yeast. This model was proposed several years ago by a group in Bordeaux and the toxin was proposed to be medium chain fatty acids, octanoic, decanoic and dodecanoic acid. These compounds are extremely toxic and are produced during fermentation. Several experiments done by others seem to lend strong support to this model. The goal of the study supported by AVF was to test this specific model. We studied 4 stuck fermentations and did gas chromatographic analyses of the stuck wine. We found that the concentrations of octanoic and decanoic acid were at or below normal levels in these stuck wines. This eliminates this specific model. We now believe that there may be another toxic compound produced by wine yeast and that this kills the cells and stops the fermentation. We are also now reconsidering model 4. If the fungicide brought in with the grapes needs to be modified during the fermentation process to be toxic to wine yeast this would explain the lag in onset of the stuck state. Glenn Andrade of Sutter Home Winery (personal communication) has found a correlation between spraying of the grapes with a fungicide and the occurrence of stuck fermentations.

The Extraction of Condensed Tannins in Red Wine Fermentations

The first year of this project was designed to survey the effects of a large number of different pomace maceration practices in commercial-scale fermentations on the extraction of anthocyanins and tannins. This year of the project focused on Pinot Noir and observed variables included cold soak, manual vs. mechanical punch down, and pump over vs. punch down. The wines were prepared at a cooperating winery according to an agreed-to protocol. The phenolic composition of the musts/wines were analyzed using both spectral (at the cooperating winery) and chromatographic methods (at UC Davis and ETS), including a new Silica gel based procedure which separates phenolics based on size. Five analyses were carried out on 9 wines although the focus was on 5 Pinot Noir wines from the Widoe’s Vineyard. The major difference was that the treatments of both mechanical punch down and pump over increased both colored and uncolored tannin, while the cold soak decreased these components. The pump over regime also increased monomelic color. The goals in this project included the following: 1) Develop a method for sample preparation and an HPLC method for the determination of condensed tannins. 2) Assist in the production of four lots of wine in each of two wineries. Production lots will include four fermentations each of a single vineyard Pinot Noir and Cabernet Sauvignon (fermentation conditions to be decided at a later date and with input from each winery.) 3) Through the course of each fermentation, pull samples and analyze condensed tannins by HPLC and information on color colorimetrically. Additional analyses (acetaldehyde, sugar, SO2) will be performed at the principal investigators’ discretion. 4) Conduct tastings of finished wines with production staff to associate style preferences with chemical data.