Utilizing Malolactic Fermentation as a Tool to Prevent Brettanomyces Bruxellensis Wine Spoilage

Brettanomyces bruxellensis is considered one of the most problematic wine spoilage yeasts due to the difficulty of controlling it, the potential significant financial losses due to loss of wine quality, and the cost of prevention and remediation measures. Wine is particularly vulnerable to B. bruxellensis infection during and shortly after the malolactic fermentation (MLF) as SO2 cannot be added until this process is complete. It has been suggested that conducting a rapid MLF initiated by inoculation of Oenococcus oeni is a useful strategy to prevent B. bruxellensis spoilage as this minimizes the length of time the wine is not protected by SO2. This project investigates an additional benefit of conducting a rapid MLF, the prevention of B. bruxellensis growth due to inhibitory interactions with O. oeni. Pinot noir wine (no SO2 additions, no MLF) was produced and used to test the ability of a large number of commercial O. oeni strains to inhibit B. bruxellensis growth at the end of MLF. Sterile filtered wine was inoculated with one of eleven commercial O. oeni strains and growth and malic acid monitored. When MLF was complete, wines were inoculated with a select strain of B. bruxellensis and growth and volatile phenol production monitored.

All O. oeni strains tested inhibited the growth of B. bruxellensis UCD2049 in Pinot noir wine with O. oeni strain variation observed. O. oeni strains Alpha, 350, VP41, MBR31 and PN4 most strongly inhibited growth of B. bruxellensis UCD2049, while strains CH11, Omega, Beta, and VFO 2.0 inhibited B. bruxellensis to a lesser extent. The potential mechanism of this inhibition was investigated by using a dialysis membrane to physically separate O. oeni and B. bruxellensis cells but allow free movement of nutrients and other potential inhibitory compounds. The physical separation of O. oeni from B. bruxellensis relieved the inhibition of B. bruxellensis by O. oeni that occurred when the two microorganism were in present together. These results indicate that inhibition is not due to nutrient depletion by O. oeni as nutrients could flow freely across the dialysis membrane. It is also unlikely that B. bruxellensis inhibition was due to the production of an inhibitory compound by O. oeni as any potential inhibitory compound would also have passed through the dialysis membrane. Instead, these results provide strong evidence that the inhibition of B. bruxellensis by O. oeni is due to cell-cell contact.

The sensitivity of additional B. bruxellensis strains to O. oeni was also determined. While B. bruxellensis UCD2049 populations declined rapidly when inoculated into Pinot noir wine that had just completed MLF with O. oeni Alpha, growth of the other B. bruxellensis strains tested was not impacted. Why B. bruxellensis strain UCD2049 was inhibited by O. oeni while the other B. bruxellensis strains were not was subsequently investigated. Initial experiments considered whether ethanol tolerance between B. bruxellensis strains impacted inhibition by O. oeni. Given that earlier experiments had been conducted in 13% (v/v) wine, wines were instead adjusted to 12.5% or 14% (v/v) ethanol. In low (12.5%) ethanol wine that had undergone MLF, B. bruxellensis UCD2049 grew well, in contrast to what was observed in 13% wine where growth was inhibited. B. bruxellensis strains AWRI-1499 and Copper Mountain also grew well in low ethanol wine with no difference between treatments. In higher ethanol wine, B. bruxellensis UCD2049 struggled to grow whether the wine had undergone MLF or not. In contrast, B. bruxellensis strains AWRI-1499 and Cooper Mountain grew well in the higher ethanol wine. B. bruxellensis strains AWRI-1499 populations recovered slower in wine that had undergone MLF while the opposite occurred for strain Copper Mountain. These results demonstrate that ethanol tolerance differences between B. bruxellensis strains impact their inhibition by O. oeni. For example, strain UCD2049 was not inhibited by O. oeni in wine at 12.5% ethanol but was inhibited in 13% and 14% ethanol wine. Additional experiments will be conducted where pH will also be considered as tolerance to this factor is known to differ between B. bruxellensis strains. Experiments are also underway exploring how long MLF induced B. bruxellensis inhibition last as well as whether B. bruxellensis inhibition occurs if infection happens at the beginning or mid-point of MLF.

Utilizing malolactic fermentation as a tool to prevent Brettanomyces bruxellensis wine spoilage

Brettanomyces bruxellensis is considered one of the most problematic wine spoilage yeasts due to the difficulty of controlling it, the potential significant financial losses due to loss of wine quality, and the cost of prevention and remediation measures. Wine is particularly vulnerable to B. bruxellensis infection during and shortly after the malolactic fermentation (MLF) as SO2 cannot be added until this process is complete. It has been suggested that conducting a rapid MLF initiated by inoculation of Oenococcus oeni is a useful strategy to prevent B. bruxellensis spoilage as this minimizes the length of time the wine is not protected by SO2. This project investigates an additional benefit of conducting a rapid MLF, the prevention of B. bruxellensis growth due to inhibitory interactions with O. oeni. Pinot noir wine (no SO2 additions, no MLF) was produced and used to test the ability of a large number of commercial O. oeni strains to inhibit B. bruxellensis growth at the end of MLF. Sterile filtered wine was inoculated with one of eleven commercial O. oeni strains and growth and malic acid monitored. When MLF was complete, wines were inoculated with a select strain of B. bruxellensis and growth and volatile phenol production monitored.

All O. oeni strains tested inhibited the growth of B. bruxellensis UCD2049 in Pinot noir wine with O. oeni strain variation observed. O. oeni strains Alpha, 350, VP41, MBR31 and PN4 most strongly inhibited growth of B. bruxellensis UCD2049, while strains CH11, Omega, Beta, and VFO 2.0 inhibited B. bruxellensis to a lesser extent. The sensitivity of a range of B. bruxellensis strains to O. oeni was also determined in 2018 Pinot noir using O. oeni strain Alpha. While B. bruxellensis UCD2049 populations declined rapidly when inoculated into wine that had just completed MLF with O. oeni Alpha, growth of the other B. bruxellensis strains tested was not impacted. Why B. bruxellensis strain UCD2049 is sensitive to O. oeni while the other B. bruxellensis strain were not is unknown at this point but is being investigated in ongoing experiments. Additional experiments are underway exploring how long MLF induced B. bruxellensis inhibition last as well as whether B. bruxellensis inhibition occurs if infection happens at the beginning or mid-point of MLF. The mechanism of inhibition is also being investigated to determine if inhibition occurs via cell to cell contact, nutrient depletion, and/or production of an inhibitory compound by O. oeni. While wineries must continue to use sound winemaking practices to prevent the growth of Brettanomyces, results from this study may provide winemakers with an additional strategy/tool to help prevent wine spoilage by Brettanomyces.

Production and Management of Aroma Compounds by

Summary of Major Research Accomplishments and Results: The primary objective has yielded some good preliminary results that indicate that it will be possible to do GC-TOF on extracts of cells grown in wine. As many of the constituents in wine can be harsh to sensitive equipment, this is an important component of the proposed work. Cells are washed prior to preparation for analysis but there are still phenolics present in the cell pellets as indicated by the red color in the extracts. The analysis of the samples revealed some expected as well as unexpected results (Table 1). Table 1 includes all of the metabolites that showed a significant difference between the wine and media grown cells in either of the two strains assayed. NS indicates no significant difference between the amounts of the metabolite in the sample. Units are relative and are based upon peak area. When cells are grown in wine there is a higher level of lipid associated metabolites, as would be expected as a reaction to the alcohol in the wine. Other differences are less easy to understand and less predictable. Many of the sugar metabolites are elevated in the wine but fructose metabolites are the same in the wine and the medium. Amino acid metabolites are variable with some higher in wine and some in medium and some showing no difference. Interestingly, urea is elevated in the wine. There are also strain differences indicated in the data. With one exception, these are differences only in whether or not there are significant differences in certain metabolites and not in the general trend of metabolite levels in media versus wine. This data is very preliminary and we should avoid trying to over analyze the data at this early stage. The main thing we can say is that the methods are working and we are able to see significant differences in the data (Figure 1). The metabolites that were identified but showed no statistically significant differences in either strain are shown in Table 2.

Quantification of Microbial Rot in Wine

Summary of Major Research Accomplishments and Results (by Objective):

Objective 1: (a) UV-VIS spectra of uninfected grapes and grapes with 0.5,1, 2, 3, 4, and 5%mold can be used to distinguish between moldy and non-moldy grapes. However, it was not possible to distinguish between the different amounts of mold. (b) The initial pass/fail FTIR calibration for mold in Chardonnay and Zinfandel grapes established in the previous funding cycle was strengthened. (c) Differences in the FTIR spectra of infected and uninfected grapes were detected. Analysis using several different approaches including principal component analysis (PCA) and partial least squares (PLS) regression gave excellent prediction values. (d) Differences in the Raman spectra of infected and uninfected grapes were detected. Analysis using several different approaches including principal component analysis (PCA) and partial least squares (PLS) gave excellent prediction values. (e) Changes in methodology to take into account bunch to bunch variation, and the use of fresh or frozen berries did not significantly affect regression and prediction values. (f) Analysis to determine if the FTIR data for Chardonnay and Zinfandel grapes can be combined to give a single quantitative model for rot rather than one for each grape variety looks very promising.

Objective 2: (a) A large number of VOCs produced by molds growing on grapes have been identified by Gas Chromatography ? Mass Spectrometry (GC-MS). (b) GC-MS focus is on the repeatability of detection, i.e., are these VOCs produced every time grapes are infected and are they produced in sufficient quantities that they can be used for quantification of rot.

Production and Management of Aroma Compounds by Dekkera /Brettanomyces Isolated from Wine

The yeast Dekkera/Brettanomyces is commonly found in wines and is responsible for a wide array of characteristic odors. The aim of this project is to define the role of substrate availability and physical growth factors in production of desired and undesired aroma compounds. A practical aim is to determine the impact of nutrients on growth and off odor production by Brettanomyces. We attempted to determine if the growth of Brettanomyces and the production of off odors can be controlled by controlling nutrients.

To this end we have looked at egg white fining as a means of controlling vitamin availability in wine. We are in the process of doing a descriptive analysis of egg white fining trials in wine. The isolates of Brettanomyces were also grown in a defined medium supplemented with aromatic amino acids or phenolic compounds to determine what substrates might produce specific aroma associated chemical compounds. Initially we looked at the compounds that were common to most of the five strains that were grown with the added substrates as well as those that are substrate specific. Now we have looked at the strain specific compounds as well.

In a defined medium the range of aroma-associated compounds produced by Brettanomyces was very broad. The variety of compounds produced by the Brettanomyces strains without supplementation included many of the same aroma compounds associated with Saccharomyces. The compounds produced specifically in the defined medium supplemented with the amino acids and organic acids belonged to several different classes. We found higher alcohols, fatty acids and their esters, volatile phenols, terpenes, and an aldehyde and a carboxylic acid. The volatile phenols were associated with the organic acid supplements while the fatty acids and their esters were primarily associated with amino acid supplements. Certain strains seem to be using specific substrates preferentially to produce aroma compounds. The use of these compounds may be correlated with regeneration of reducing power and could be connected to preferred metabolic pathways.

We have grown three Brettanomyces bruxellensis strains under three levels of oxygen, 0, 25, and 50%air saturation. Results indicate that there is some production of 4-ethylphenol (4-EP) from coumaric acid even under 0%air when growth is severely limited. Results indicate that growth under low oxygen is enhanced by the presence of coumaric acid in strain 2077 at 25%air saturation, in strain 2091 at 50%air saturation, and in strain 2082 at both 25%and 50%air saturation. These results are again consistent with the use of coumaric acid conversion to 4-ethylphenol to regenerate reductant under low oxygen conditions.

Quantification of microbial rot in wine grapes

In the initial few months of this project a five pronged approach was adopted. A major consideration was the potential use of instrumentation found in many wineries – Fourier Transform Infra-Red (FTIR) spectroscopy typified by the FOSS Winescan. We also had access to two types of Near Infra-Red (NIR) instruments that were being used for other winegrowing projects. The final spectrometric approach was Raman Spectroscopy. The objective of using these instruments is to find reliable differences in the spectra that could be used to distinguish sound grapes from moldy grapes. Two additional technologies available at Fresno State are Flow Cytometry and hand-held Gas Chromatography – zNose. The assay objectives are listed in the figure below.

Figure included in progress report.

Flow cytometry: a fluorescent compound that binds to chitin, a compound found in the cell wall and spores of molds (the lectin, wheat germ agglutinin) can be used in conjunction with flow cytometers’s cell counting to detect molds. Mold fragments and spores were detected using this system. The method needs further development to quantify mold material present in grape loads.

Gas chromatography: all molds produce volatile organic compounds (VOCs) when growing (like a car exhaust). The zNose hand-held gas chromatograph is versatile and can be used to concentrate on identifying VOCs produced by common grape molds. Differences have been detected in the VOC patterns produced by the three molds growing on whole grapes. Identification of these VOCs will be the basis of an assay for mold detection.

Raman Spectroscopy was eliminated as a possible method when it was determined that the spectra from moldy and non-moldy grapes showed little variation.

NIR and FTIR Homogenized and sterilized grapes were inoculated with one of three molds or a combination of all three molds. Samples were taken according to a protocol, processed and run on the spectrometers . The NIR instruments distinguished moldy grapes from non-moldy grapes The FOSS FT 120 Winescan uses the whole spectrum and can distinguish moldy from nonmoldy grapes in a validation test.

The results from the FTIR are very encouraging, although there remains much testing to determine if this will be an effective system. The NIR, Flow Cytometry and zNose gas chromatography also offer very promising possibilities of quantifying grape rot. Raman Spectroscopy did not generate spectra that could be reliably used to quantify grape rot.

Further experimentation with FTIR and NIR spectroscopy, VOCs, and flow cytometry are recommended in a multipronged approach to establish a reliable method of quantifying grape rot.

Production and Management of Aroma Compounds by Dekkera/Brettanomyces Isolated from Wine

The yeast Dekkera/Brettanomyces is commonly found in wines and is responsible for a
wide array of characteristic odors. The aim of this project is to define the role of substrate
availability and physical growth factors in production of desired and undesired aroma
compounds. A practical aim is to determine the impact of nutrients, particularly those
that may be added in the wine making process that may result in off-character formation
by Brettanomyces. The isolates of Brettanomyces were grown in a defined medium
supplemented with aromatic amino acids or phenolic compounds to determine what
substrates might produce specific aroma associated chemical compounds. Initially we
looked at the compounds that were common to most of the five strains that were grown
with the added substrates.

In a defined medium the range of aroma-associated compounds produced by
Brettanomyces was very broad. The variety of compounds produced by the
Brettanomyces strains without supplementation included many of the same aroma
compounds associated with Saccharomyces. The compounds produced specifically in the
defined medium supplemented with the amino acids and organic acids belonged to
several different classes. We found higher alcohols, fatty acids and their esters, volatile
phenols, terpenes, and an aldehyde and a carboxylic acid. The volatile phenols were
associated with the organic acid supplements while the fatty acids and their esters were
primarily associated with amino acid supplements. The other compounds did not seem
to give any such specific correlations.

We have grown three Brettanomyces bruxellensis strains under three levels of oxygen, 0,
25, and 50%air saturation. Preliminary results indicate that there is some production of
4-ethylphenol (4-EP) from coumaric acid even under 0%air when growth is severely
limited. One of the strains shows a different pattern of sensitivity to oxygen than the
other two strains. While two of the strains produce the same amount of 4-EP at 25 and
50%air saturation the third strain produces only about half the 4-EP at 25%air saturation
that is does at 50%air saturation.

Metabolism of Dekkera/Brettanomyces Isolated from Wine

The aim of this one year project was to evaluate the impact of altered alleles derived from the low-sulfide producing strain UCD932 for their potential to reduce sulfide formation in high-sulfide producers. UCD932 is a native wine isolate previously shown to not produce sulfide under any enological circumstances evaluated. However, it is not as strong of a fermentor as other strains and has limited commercial applicability. The sequences of the genes of the sulfate reduction pathway were analyzed in UCD932 and genetic crosses were used to define those genes associated with the low sulfide production trait. In most crosses only one gene, subsequently identified as MET10UCD932 seemed to be required to reduce sulfide production while in other crosses it seemed that more than one gene was required. UCD932 contains mutations in CYS4HOM6MET5and MET6 in addition to the mutation in MET10. Allele swap technology, the replacement of one form of a gene by a natural variant from another strain, was used to evaluate the role of these genes in sulfide production in UCD522. UCD522 is a well known high sulfide producer. Replacement of MET10UCD522 with MET10UCD932eliminated H2S production by UCD522. Replacement of other alleles appeared to have no effect. Similarly, replacement of the MET10 genes of two native high sulfide producers, UCD940 and UCD950, with MET10UCD932also eliminated sulfide production in those two genetic backgrounds in both synthetic and actual grape juice. Thus, allele swap of the existing MET10 gene of a commercial or native strain with the MET10 gene from UCD932 is an effective strategy to reduce sulfide production. Further tests demonstrated that this allele did not affect other enological properties of the strains such as sulfite tolerance, fermentation rate or progression, or aroma profile with the exception of the loss of sulfides. A patent application has been submitted for the use of the MET10UCD932 allele to reduce sulfide formation. In one other high sulfide producing strain, UCD939, replacement of the existing MET10 gene with MET10UCD932 appeared to be a lethal event. Thus there are likely combinations of genes of this pathway that do not permit cell growth or that are toxic. An important goal for this grant was the analysis of the potential impact on strain competitiveness due to the change in allele at MET10. To test this, mixtures of equal concentrations of modified and unmodified UCD522 were inoculated into synthetic and actual juices. The two strains can be easily distinguished on BiGGY agar as one, MET10UCD932 gives white colonies on this medium and the other, MET10UCD522, gives tan colonies. At the end of fermentation, the relative percentage of the strain carrying MET10UCD932 had decreased from 50%to 20-25%indicating that there is an advantage in the wild to hydrogen sulfide formation. This is not surprising and explains why so many strains have evolved to produce sulfide. It also suggests

Characterization of Volatile Phenois and Related Enzyme Activity in Brettanomyces Wine Strains

In this project we investigated the growth and off-flavor production of the potent wine spoilage yeast Brettanomyces bruxellensis. We analyzed growth and production of three dominant offflavor compounds 4-ethyl catechol, 4-ethyl guaiacol, and 4-ethyl phenol in model wine media in the presence of; 1) various concentrations of ethanol, 2) various sources of nitrogen, 3) various amounts of organic nitrogen, and 4) various types and amounts of residual sugars. This investigation was carried out with five strains of B. bruxellensis isolated from wine. The increased ethanol concentration affected the cell growth of all the strains and two strains did not show growth when the alcohol was above 12%and 13%respectively. The production of ethylphenol was affected too and diminished at the increase of ethanol concentration from 11%to 14%. Only 4-ethyl phenol was produced. The growth of B. bruxellensis was diminished in the absence of amino acids but, importantly, B. bruxellensisgrew with as little as 0.5 mg/L of diammonium phosphate (DAP) and no amino acids. Various amounts of nitrogen did affect the ethyl phenol production, yet, the production of ethyl phenols seems to be more strain dependent. More ethyl phenols were produced when only DAP was present. The presence of C6 (hexoses) and C5 (pentoses) sugars affected the yeast growth and the amounts of the ethyl phenols produced. When no C5 or C6, or only C5 sugar were present, the total amount of ethyl phenols was lower (except with strain CE116) and a lower cell density was observed. The presence of only C5 sugar sugars enhanced the production of 4-ethyl guaiacol (except for strain 2091). In the presence of only amino acids as a nitrogen source, all the strains grew when as little and 50 mg/L of total amino acids were added to the medium.

Analysis of the Genome of Oenococcus Oeni

The primary objective of this proposal is to fund gap closure of the Oenococcus oeni genome currently being sequenced at the Joint Genome Institute (JGI). Genome sequencing of O. oeni is part of a larger project undertaken at JGI to sequence the genomes of ten lactic acid bacteria. Of the ten LAB being sequenced, five are found in the wine environment (O. oeni, Pediococcus pentosaceus, Leuconostoc mesenteroides, Lactobacillus brevis, Lactobacillus casei). Two of these, P. pentosaceus and L. brevis, are prominent spoilage microorganisms in wine. The draft sequencing of the O. oeni and LAB genomes, is being done by the Joint Genome Institute at no cost to AVF/CCPRVE (a total of ~ $1.6 M in draft sequencing cost). Unfortunately, due to delays at JGI, mostly due to problems in finishing the human genome, sequencing of O. oeni and the other LAB was delayed until 2002 (actually it was first delayed from May 2001 to August, then again to Jan 2002). Currently ten LAB genomes, O. oeni included, are being sequenced and slated to be finished by summer 2002. The first completed genome of the group, Lactobacillus gasseri, is available from the JGI website (see http://www.jgi.doe.gov/JGI_microbial/html/index.html). As you might expect, a consequence of this delay is a correlating delay in my closing the O. oeni genome sequence. To date I have not hired a full time technician on the project but instead hired Lucy Joseph, our culture curator, at 25%time to work up the DNA and to generate and validate a cosmid library of O. oeni for eventual scaffolding of draft sequence. I have also purchased some computer equipment in preparation of the coming sequence analysis. To coordinate analysis of the ten LAB genomes we formed a group termed the Lactic Acid Bacterial Genome Consortium (LABGC) comprised of eleven researchers from seven universities in the US. With the expected completion of the O. oeni and other LAB sequences this summer, the LABGC is currently planning to hold a weeklong ?annotation-athon? at JGI in early October, 2002. At this meeting all of the genomes will be examined by a group of 30-40 LAB researchers from the US and assisted by bioinformatics staff at JGI. In the 2003-2003 report I expect to relay information learned at this meeting, especially as it pertains to wine-related LABs. Moreover I will report on the progress made to finish the O. oeni sequence (once the initial draft sequence is produced by JGI and we have something to work on).