Abstract
The production of volatile metabolic products was determined for five Brettanomyces bruxellensis isolates from wine when grown in a defined medium with and without the hydroxycinnamic acids caffeic, coumaric, and ferulic or the aromatic amino acids phenylalanine, tryptophan, and tyrosine. The aim of the study was to determine the relationship between substrates and end products and to define strain differences in the production of volatile compounds. In the presence of coumaric and ferulic acids, all strains produced very similar metabolic products, primarily 4-ethylphenol (4EP) and 4-ethylguaiacol (4EG), respectively. There was a more pronounced effect of strain with the other substrates, and a variety of compounds with the potential to impact wine aroma were detected. Growth of Brettanomyces in the defined medium, with and without coumaric acid, under varying oxygen concentrations was also studied to determine the effect of this compound on growth parameters. The highest concentrations of 4EP were found under anaerobic conditions. Coumaric acid also had a significant positive affect on growth of Brettanomyces at 25% air saturation levels. At full aeration, coumaric acid addition showed little to no impact on growth or 4EP formation. Significantly higher concentrations of acetic acid were formed in the presence of coumaric acid, suggesting that 4EP formation may aid in the recycling of oxidized cofactor NAD+.
Dekkera/Brettanomyces bruxellensis has long been recognized as a common contaminant in wine (Custers 1940, van der Walt and van Kerken 1958). Brettanomyces can produce very potent off-aromas described as horse sweat, band aid, barnyard, and burnt plastic, which devalue the quality of wines, particularly when production occurs after bottling (Coulon et al. 2010). Not all characteristics produced by Brettanomyces are universally disliked. There are anecdotal reports of Brettanomyces strains that grow faster or produce less acetic acid or even of strains that only produce “good Brettanomyces” characteristics, but only a few compounds have been identified as being produced by Brettanomyces from specific substrates (Chatonnet et al. 1995, 1997, Hesford et al. 2004). Factors controlling the production of known chemicals that can produce the characteristic flavors associated with Brettanomyces have been investigated (Heresztyn 1986a,b, Ciani and Ferraro 1997, Rodrigues et al. 2001, Romano et al. 2008, 2009). The effect of media components has also been investigated (Rodrigues et al. 2001). Brettanomyces produces a wide range of metabolites, the most significant of which appear to be acetic acid and ethylphenols, such as 4-ethylphenol (4EP), 4-ethylguaiacol (4EG), and 4-ethylcatechol (4EC) (Chatonnet et al. 1995, 1997, Hesford et al. 2004), as well as several other known and unknown compounds (Licker 1998). As many as 10 compounds are denoted as being related to the “Brett” defect in wines, and only a few have defined substrate association (Heresztyn 1986a,b, Chatonnet et al. 1997).
Solid-phase microextraction (SPME) gas chromatography with mass spectrometry (GC-MS) detection has been used to define the composition of volatile compounds that may contribute to aroma in the headspace of a sample, including wine (Howard et al. 2005). This technique has been used extensively in food chemistry studies and is applied here to characterize the compounds produced by Brettanomyces under controlled nonwine conditions in the presence of specific precursors or substrates. This analysis was undertaken in order to begin to define specific relationships between putative aromatic compound precursors and the appearance of the aromatic compounds and to define strain differences in the appearance of these characters. Cinnamic acids were chosen due to their previously defined role as precursors for odor active compound production in Brettanomyces contaminated wines (Chatonnet et al. 1995, 1997, Hesford et al. 2004). Aromatic amino acids were chosen because of their known conversion to odor active compounds in wine during production and the ease of detection of the known products. Preliminary bench trials were conducted using several amino acids as potential substrates, and the aromatic amino acids were deemed as having an impact on aroma and thus were selected for further analysis. Other amino acids also showed variable impacts on aroma but not to the same extent as those described for the aromatic amino acids. Finally the benefit to the cell of formation of vinylphenols was investigated using coumaric acid and controlled oxygen additions in defined batch fermentations. It was hypothesized that the reduction of coumaric acid would assist the cell in recycling of the oxidized cofactor, NAD+.
Materials and Methods
Chemicals and reagents.
Chemical standards were purchased from VWR International (Radnor, PA) with the following exceptions. Ethyl acetate, butanol, 3-methyl-1-butanol, and the hydroxycinnamic acids, coumaric, ferulic, and caffeic were purchased from Fisher Scientific (Pittsburg, PA; hydroxycinnamic acids are from MP Biomedicals, Solon, OH). All standards were >98% purity. Amino acids were purchased from Sigma-Aldrich (St. Louis, MO). Ethanol was purchased through the University of California, Davis (UC Davis) from Gold Shield (Hayward, CA).
Brettanomyces growth.
Strains were obtained from the UC Davis Wine Microbe Collection in the Department of Viticulture and Enology (Table 1). These strains were grown in defined Brettanomyces medium with 1% glucose and 10 μg/L thiamine (Conterno et al. 2006) for 14 days at 30°C with and without the addition of phenolic acids and amino acids (Table 2). Cultures were inoculated at 0.01 absorbance units at 600 nm and were not aerated but grown as still cultures with loose caps. Cultures were vortexed briefly at inoculation, one week, and before centrifugation on day 14.
Dekkera/Brettanomyces isolates from the University of California, Davis, Wine Microbe Collection selected for analysis (from Conterno et al. 2006, Wirz 2005).
Substrates and concentration added to defined medium.
SPME sample preparation.
Five Brettanomyces strains were analyzed under seven conditions: with no addition and with the addition of six different substrates: caffeic acid, coumaric acid, ferulic acid, phenylalanine, tryptophan, and tyrosine. Brettanomyces cultures were centrifuged to remove cells in a clinical centrifuge at 2,000 g for 10 min. Ten mL of the supernatant was transferred to a sterile 20 mL brown vial, and 10 μL of a 1 g/L 2-nonanone internal standard in 50% ethanol was added to give a final concentration in the vial of 1 mg/L 2-nonanone. Vials were sealed with a screwcap and a silicone septum until sampling. Samples were refrigerated overnight and brought to room temperature for 30 min before gas chromatography (GC) analysis.
HS-SPME GC-MS analysis.
The analysis was done by GC using an Agilent 6890K gas chromatograph (Santa Clara, CA) paired with an Agilent 5973 mass selective detector with an electron impact (70 eV) source and a Gerstel MPS2 autosampler (Gerstel, Mülheim, Germany). The GC injector was equipped with a SPME inlet liner, 0.75 mm i.d. (Supelco, St. Louis, MO). The oven parameters were as follows: initial temperature 40°C and held for 5 min, then increased to 280°C at a rate of 8°C/min (total cycle time including oven cool down 42 min). The injector was held at 260°C and the temperature of the transfer line to the MS detector was 280°C. Helium was the carrier gas with a total flow of 0.9 mL/min at the start and maintaining a constant pressure of 5.95 psi throughout the run. A HP-5 MS column (30 m × 0.25 mm i.d., 0.25 μm film thickness) (J&W Scientific, Folsom, CA) was used for all analyses. A PDMS-DVB SPME fiber (65 μm thickness; Supelco) was conditioned for 30 min in an injection port at 260°C before first use. The MS detector was operated in scan mode with a mass range of m/z 50 to 550. Samples were heated to 40°C for 30 min to sorb volatiles to the SPME fiber. Following extraction, the fiber was inserted into the injection port for 10 min to desorb volatiles from the fiber. The injection was splitless for the first 5 min, after which the split flow was turned on (20 mL/min) for the remainder of the GC run. Enhanced MSD ChemStation software (G1701DA, ver. D.00.01.27; Hewlett-Packard, Santa Clara, CA) was used for instrument control and data analysis. Peak identities were confirmed by comparison of retention times and mass spectra with those of authentic standards except for ethyl 9-hexadecenoate and β-farnesene, which were identified by comparison to NIST library spectra only. β-Farnesene was available only in a mixed sample with other farnesene isomers. Three biological replicates were analyzed for each treatment.
Data analysis.
SPME samples were analyzed as three instrument and three biological replicates in the MetaboAnalyst online software (Xia et. al. 2009). The peak area of the internal standard, 2-nonanone, was given a value of 1 (1 mg/L) in each case, and all values for the other compounds were based upon the relative peak areas compared to that internal standard. Analysis of variance (ANOVA) with a Fisher’s LSD and p < 0.05 was used (Supplemental Data) to compare sample treatments. Cluster analysis was done using Pearson correlation coefficient and the complete (Figure 1) or Ward’s (Figure 2) linkage method. The combined values for each substrate, regardless of strain, were used to perform a partial least squares discriminant analysis (PLS-DA) to determine variable importance using weighted sum of absolute regression coefficients (Figure 3). The validity of the PLS-DA was determined using Q2, R2, accuracy, and a permutation analysis. All of these measures indicate that the analysis is valid, indicating that a five-component-model is best and that the accuracy using the top five components is ~70%. The permutation analysis indicates p < 0.01 for 100 permutations. The PLS-DA was used to show a pattern across all strain for the importance of specific substrates in determining the presence of certain volatile products. Further, the analysis was used to determine if there was a general trend for the cinnamic acids as opposed to the aromatic amino acid substrates.
Cluster diagram for B. bruxellensis wine strains UCD VEN 2058, 2091, 2082, 615, and 2077 grown in defined medium and analyzed by SPME-GC-MS for volatile metabolite production.
Cluster diagram for B. bruxellensis wine strains UCD VEN 615, 2058, 2077, 2082, and 2091 grown in defined media alone (Media) or supplemented with caffeic acid (CAF), coumaric acid (COU), ferulic acid (FER), phenylalanine (PHE), tryptophan (TRP), or tyrosine (TYR).
Partial least square analysis indication of the top 15 features that differentiate the samples in Figure 2 based on the weighted sum of absolute regression coefficients (x axis). Samples were unsupplemented (None), supplemented with caffeic acid (Caffeic), coumaric acid (Coumaric), ferulic acid (Ferulic), phenylalanine (Phe), tryptophan (Trp), or tyrosine (Tyr).
For growth data, ANOVA was performed using SAS software (SAS Institute, Cary, NC). Single-variance, three-way ANOVAs were performed with a confidence limit of 95%. Mean separation analysis was carried out via Tukey’s honestly significant difference (HSD) test. For acetic acid data, analysis of covariance was performed using SAS software. Mean separation analysis was carried out via Tukey’s HSD test.
Growth of Brettanomyces in bioreactors.
Yeast strains were maintained in a liquid defined medium (Conterno et al. 2006) with 10 μg thiamin HCl/L. Prior to the inoculum preparation, strains were grown without the addition of trace elements ZnSO4 • H2O, H3BO3, and FeCl3 • 6H2O to the medium for at least five days until turbidity was present. These trace elements were excluded so that the subsequent inoculum (used for the fermentation) would not contain surplus trace elements.
Determination of coumaric acid level.
The effect of the coumaric acid level on growth was initially tested at concentrations of 0, 10, 15, and 20 mg/L for each strain. Each strain was tested in triplicate under each coumaric acid level. Forty mL of minimal media was transferred to a sterile large culture tube and inoculated to achieve an optical density of 0.01 at 650 nm. The concentration of coumaric acid in each tube was adjusted via addition of 16 g/L coumaric acid stock solution in 95% ethanol. Culture tubes were incubated at room temperature on a roller drum using moderate rotation speed. Optical density was monitored daily at 650 nm for 10 days.
Determination of air saturation level.
Bioreactors were programmed to maintain 0%, 25%, or 50% air saturation. These air percentages were chosen to result in ~0, 2, and 4 mg/L, respectively, of dissolved oxygen in saturated media. Industrial nitrogen and/or air were sparged into the bioreactor to maintain the desired concentration of oxygen. The bioreactor was inoculated to a final optical density of 0.01 absorbance units at 600 nm. Twenty mL of either 95% ethanol or 4250 mg/L p-coumaric acid (98+%; MP Biomedicals) in a 95% ethanol matrix was added to its respective bioreactor. The coumaric acid addition resulted in a final concentration of ~20 mg/L within the bioreactor. The resulting final ethanol concentration from the addition was less than 0.50%. After all bioreactors were inoculated, air saturation percent and the temperature of each bioreactor were recorded every 12 min for the entire 240 hr of fermentation. Beginning at time 0 and in 24-hr intervals, two subsamples were removed from each bioreactor for growth measurements. The optical density of each subsample was analyzed in triplicate using a UV160U UV-Visible Recording Spectrophotomer (Shimadzu, Kyoto, Japan) set to 600 nm. Samples were serially diluted and plated onto WL agar to determine colony forming units/mL (cfu/mL) for each strain; this information was used to normalize the compound production data to yeast biomass. The experimental design was a complete randomized factorial. All 18 combinations of the three factors (air saturation, strain, and coumaric acid) and their levels (3, 3, and 2, respectively) were randomized using the random number function of Excel (Microsoft). Fermentations within each replication were carried out between 5 and 7 at a time due to a limiting number of large-scale bioreactors. Three replicates were performed for each combination of factors.
Quantification of acetic acid.
Acetic acid concentrations were determined by Acetic Acid Enzyme Assays (Boehringer Mannheim/R-Biopharm, Darmstadt, Germany). Samples were centrifuged and frozen at day 9 and the results of the two subsamples were averaged. For statistical analysis, concentrations were normalized with respect to biomass. A Student’s t test was performed on the means in every strain comparing the acetic acid production at each air level with and without coumaric acid.
Results
Volatile compound production in the absence of precursors.
Five Brettanomyces isolates—two from France, two from California, and one from Chile—were grown in synthetic wine media with or without caffeic acid, coumaric acid, ferulic acid, phenylalanine, tryptophan, or tyrosine. In the base medium without added substrates, all strains produced the ester ethyl acetate, the alcohol 3-methyl-1-butanol, the fatty acids octanoic and decanoic acid, and the ethyl esters of the fatty acids hexadecanoic, dodecanoic, decanoic, and octanoic acid (Supplemental Data). Four out of five strains also produced acetic acid, 3-methylbutanoic acid, 2-methylbutanoic acid, 2-ethyl-1-hexanol, phenethyl alcohol, and a compound tentatively identified as ethyl 9-hexadecenoate. Compound identification was made by comparison to a mass spectral library match, which included a match to the retention index but no standard was available for confirmation. Strain 2082 also produced the ethyl ester of 2-methyl butyrate. Strains 615 and 2091 produced 2-methyl-1-butanol and strains 615 and 2077 produced 1-decanol. Strain 2058 produced some unusual compounds in medium alone, including 1-octanol, phenethyl propionate, and possibly the terpene β-farnesene. Filamentous fungi are known to produce terpenes and other yeasts have been reported to produce terpenes (Drawert and Barton 1978, Carrau et al. 2005) but that has not been shown for B. bruxellensis. While this identification was tentative, the only other similar mass spectral matches in the compound library were also terpenoid compounds. Thus, the strains produced a spectrum of compounds that were similar but also showed variation in both the levels and types of compounds produced.
Impact of cinnamic acid precursor on volatile compound production.
In the presence of the cinnamic acids, as expected 4-ethylguaiacol (4EG) was produced primarily when ferulic acid was present (Chatonnet et al. 1995). When caffeic acid was added to the medium a small amount of 4EG was produced by all of the strains, which may have been due to a contamination of the caffeic acid with a small amount of ferulic acid (complete purity information was not available for caffeic acid) or a conversion of 4-ethylcatechol (4EC) to 4EG by methylation of the hydroxyl group. If rates of conversion to 4EG were consistent over the range of concentrations involved, then the amount of contamination with ferulic acid needed to account for these 4EG values would range from 0.25 to 0.5%. 4-Ethylphenol (4EP) was only detected when coumaric acid was added to the medium. The cinnamic acid additions tended to increase production of the ethyl esters of fatty acids in most of the strains. The production of the fatty acid esters ethyl dodecanoate, ethyl tetradecanoate, and ethyl hexadecanoate generally increased across all strains (Supplemental Data). Production of other compounds in the presence of caffeic acid was typically strain dependent. The addition of any of the supplements increased concentrations of acetic acid in strain 2077 and 2-methyl-1-butanol in 2058.
In the presence of coumaric acid, concentrations of fatty acids and their esters tended to increase, especially ethyl dodecanoate, ethyl tetradecanoate, and ethyl hexadecanoate (Supplemental Data). Concentrations of all of the other products were strain dependent. Acetic acid production was significantly increased in strains 615 and 2077.
In the presence of ferulic acid, strain 615 produced detectable amounts of 3-methylbutanoic acid. In contrast, the other strains produced 3-methylbutanoic acid in the absence of supplementation. Changes in the production of other compounds were strain dependent.
Impact of amino acid precursor on volatile compound production.
The aromatic amino acids, phenylalanine, tyrosine, and tryptophan, can also affect the volatile compounds that could be produced in wine. The addition of amino acids increased concentrations of 2-methyl-1-butanol, 2- and 3-methylbutanoic acid, isobutyric acid, phenethyl alcohol, and phenethyl acetate for most of the strains. Strain 2058 produced phenylacetaldehyde in the presence of any of the amino acids. The addition of phenylalanine resulted in the production of higher amounts of phenethyl alcohol in all strains, but particularly in strain 2058.
When phenylalanine was added to the medium all strains produced phenylethyl acetate. Strain 615 also produced phenethyl alcohol at detectable levels under these conditions (Supplemental Data). All of these compounds have sweet floral aromas (Aldrich Flavors and Fragrances, Milwaukee, WI). Strain 615 produced detectable amounts of 3-methylbutanoic acid, 2-methylbutanoic acid, and the terpenoid β-farnesene as well as increased levels of ethyl octanoate, ethyl decanoate, and ethyl dodecanoate. Strain 2082 produced increased levels of the fruity esters, ethyl decanoate and ethyl octanoate, and strain 2091 produced ethyl isobutyrate.
When tryptophan was added to the medium no specific product was predominant in all strains. All strains showed a significant increase in 3-methylbutanoic acid production and a decrease in decanoic acid production. Strain 2058 produced phenylacetaldehyde at detectable levels (Supplemental Data).
Tyrosine addition affected the production of a variety of volatile compounds. Butanol and isobutyric acid were only detected in the presence of tyrosine (Supplemental Data). Strains 615 and 2091 produced butanol. Strain 2091 produced detectable levels of isobutyric acid. All strains showed significant increases in production of 3-methyl-1-butanol, 2-methyl-1-butanol, 2-methylbutanoic acid, and phenethyl alcohol. Strains 615 and 2091, but not 2082, produced phenethyl acetate in the presence of tyrosine.
Analysis of strain differences in product formation.
Strain differences were apparent in the products formed with different substrates (Supplemental Data). For example, with coumaric acid four strains showed a significant increase in production of decanoic acid, ethyl tetradecanaoate, and ethyl hexadecanoate but one strain did not. Decanoic acid production did not increase in strain 615 and the concentrations of the other fatty acids did not increase in strain 2082. When ferulic acid was added, 2-methylbutanoic acid concentrations decreased significantly in three of the strains but not in strains 615 and 2082. With phenylalanine additions, only strain 615 did not increase the production of 2-methyl-1-butanol significantly; ethyl dodecanoate, ethyl tetradecanoate, ethyl 9-hexadecenoate, and ethyl hexadecanoate concentrations decreased significantly in some strains and increased in others. When tryptophan was added to the medium decanoic acid concentrations decreased in four strains but not significantly in 2082. In the presence of tyrosine, four strains showed an increase in ethyl isobutyrate, ethyl-2-methyl butyrate, and 3-methylbutanoic acid production but strain 2077 did not. Concentrations of ethyl palmitate, when tyrosine was in the medium, increased in two strains, decreased in two strains, and stayed the same in one strain.
A cluster analysis was performed on the volatile metabolite data from the five strains following growth in the defined medium in the absence of supplementation and revealed two distinct clusters (Figure 1). Strains 615, 2077, and 2082 formed a single cluster, with strains 615 and 2077 most closely related in terms of their metabolic products under these unsupplemented conditions. Strains 2082 and 615 are isolates from California and 2077 is a Chilean isolate. Interestingly, the two French strains, 2058 and 2091, formed a separate, more distantly related cluster. The volatile compound identified as the most significant in distinguishing the two clusters was phenethyl alcohol and was highest in strain 2058 (Supplemental Data). Further differentiation was due to the levels of the fatty acid ethyl esters: ethyl dodecanoate, ethyl decanoate, ethyl octanoate, and ethyl hexadecanoate. The differentiation of “new world” and “old world” isolates and strains that were entered into the collection decades apart suggests there may be regional similarities in Brettanomyces strains.
The addition of different substrates to the medium revealed several distinct patterns in the data by cluster analysis (Figure 2). The cluster analysis yielded two distinct groupings as a function of substrate addition. The first cluster was based on substrate availability alone. The presence of either coumaric or ferulic acid gave a tight clustering of all of the strains, although each substrate was distinct within this cluster. These two substrates are known to be the precursors for the 4EP (coumaric) and 4EG (ferulic) that have been identified as major components of the Brettanomyces character in wine (Chatonnet et al. 1995).
The second cluster included all of the other substrates tested (Figure 2). This cluster analysis gave groupings based on both substrate and strain. In the presence of phenylalanine, all of the strains clustered in one subgroup within this second cluster. That subgroup also contained strains 2058 and 2091 with tyrosine and tryptophan along with strain 2058 in medium alone. Phenylalanine is the precursor for phenethyl alcohol (Gopalakrishna et al. 1976), which was important in the separation of strains 2058 and 2091 from the other strains in the cluster analysis of the strains grown in medium with no supplement. Strain 2058, and to a lesser extent strain 2091, produced higher levels of phenethyl alcohol than the other strains in the absence of phenylalanine. The other subgroup within the second cluster contained strains 615, 2077, and 2082 with tryptophan, tyrosine, and caffeic acid substrates and strain 2058 and 2091 with caffeic acid. It also contained all of the strains, except 2058, in medium alone.
Compounds that were important in determining the groupings in this cluster analysis were, in order of importance, 2-methylbutanoic acid, 3-methylbutanoic acid, phenethyl alcohol, 2-methyl-1-butanol, ethyl dodecanoate, 3-methyl-1-butanol, ethyl octanoate, 4EP, octanoic acid, 4EG, isobutyric acid, ethyl-2-methyl butyrate, phenethyl acetate, ethyl-9-hexanoate, and ethyl isobutyrate (Figure 3). In addition, 2-methylbutanoic acid, 3-methylbutanoic acid, phenethyl alcohol, 2-methyl-1-butanol, 3-methyl-1-butanol, ethyl-2-methyl butyrate, and phenethyl acetate concentrations increased when amino acids were present in the medium and decreased or were at lower levels when cinnamic acids were added. Ethyl dodecanoate showed the reverse trend. Many of these compounds showed their highest levels with tyrosine in the medium, including 2-methylbutanoic acid, 3-methylbutanoic acid, 2-methyl-1-butanol, 3-methyl-1-butanol, isobutyric acid, ethyl-2-methyl butyrate, and ethyl isobutyrate. The addition of coumaric acid also induced production of the highest level of some compounds, including ethyl dodecanoate, ethyl octanoate, 4EP, and octanoic acid. Samples with added phenylalanine had the highest levels of phenethyl alcohol and phenethyl acetate (Figure 3). Caffeic acid addition resulted in the highest levels of ethyl 9-hexadecenoate, and ferulic acid addition corresponded to the highest level of 4EG. The lowest levels of the five most important compounds contributing to the cluster classes were in the unsupplemented or ferulic acid supplemented medium. Thus, the cluster analysis showed some groupings by strains and by substrate and also indicated significant differences in strain behavior as a function of the presence of specific precursor components.
Impact of coumaric acid on growth characteristics of Brettanomyces.
All strains produced 4EP in the presence of coumaric acid, but the biological benefits of doing so are unknown. There are several possible explanations for the reduction of phenolic acids by Brettanomyces. The reduction could be accidental, the enzymes involved being produced for another reaction and decarboxylation and subsequent vinylphenol reduction is a side reaction with no relevance to the biology of the cell. Alternately, the reduction of phenolic acids could serve as a critical detoxification mechanism to enable cells to thrive in an otherwise inhibitory environment. It is also possible that the cell gains some benefit from the reduction of phenolic acids and such reduction is favorable to the growth of this yeast under certain environmental conditions.
To distinguish among these possibilities, strains 2077, 2082, and 2091 were grown under fully aerobic conditions in the presence and absence of coumaric acid. Strain 2077 showed no difference in growth kinetics until roughly 250 hr, when the control, with no coumaric acid, continued to increase in cell density while any addition of coumaric acid resulted in a cessation of growth at a lower terminal density (Figure 4). In strain 2082, terminal cell density was likewise negatively correlated with initial coumaric acid concentration, but there appeared to be a slight stimulation in exponential growth phase in the presence of 10 and 15 mg/L coumaric acid. Strain 2091 showed a much stronger negative impact of coumaric acid throughout the growth cycle. There was a trend toward lower maximal cell densities under fully aerobic conditions in the presence of coumaric acid for all strains, but the sensitivity to coumaric acid varied by strain. Thus, under fully aerobic conditions, coumaric acid behaved as a simple inhibitor of growth.
Growth of B. bruxellensis strains in defined medium under aerobic conditions with 0, 10, 15, and 20 mg/L coumaric acid over a period of 350 hr: UCD VEN strains 2077 (A), 2082 (B), and 2091 (C).
In contrast, under fully anaerobic conditions, none of the strains were able to grow. Coumaric acid showed no impact on cell density or cell viability as defined by colony forming units (data not shown) under these conditions. Consequently, in the absence of molecular oxygen, coumaric acid has no discernible impact on cell survival.
Of more interest was the impact of coumaric acid on growth under conditions of limiting oxygen. With 25% air saturation, strains 2077 and 2082 demonstrated higher cell densities in the presence of coumaric acid as compared to the control during late-log growth phase, ~125 to 150 hr (Figure 5). Strain 2077 also showed significantly accelerated death kinetics in the presence of coumaric acid under these conditions. Coumaric acid had no effect on strain 2091 at 25% air saturation (data not shown). At 50% air saturation, strain 2082 showed a significantly higher level of growth during mid-log growth phase, ~75 to 100 hr (Figure 6). In contrast, at this level of air saturation, strains 2077 and 2091 did not show a significant impact of coumaric acid, although there was an increased, but not statistically significant, rate of cell death in both strains after 275 hr (data not shown). These observations suggest that coumaric acid can be both a source of yeast stress and inhibition, particularly following the growth phase, and a stimulator of cell growth under limiting aeration conditions.
Comparison of growth of Brettanomyces strains with 0 and 20 mg/L coumaric acid in defined medium at 25% air saturation over 250 hr. UCD VEN strains 2077 (A) and 2082 (B).
Growth of B. bruxellensis strain 2077 in defined medium with 50% air saturation over 250 hr with 0 and 20 mg/L coumaric acid.
The impact of coumaric acid addition on the formation of 4EP was also evaluated. In the absence of air, all three strains produced significant amounts of 4EP even though growth was not occurring (Figure 7). The strains differed in the total amount of 4EP produced, with strain 2077 displaying the highest production levels, followed by 2091 and then 2082. At 25% aeration, significantly higher 4EP levels were observed, with 2077 and 2082 producing much higher levels than 2091. At 50% air saturation, 4EP levels were only slightly elevated over that formed at 25% saturation. The greatest accumulation of 4EP was seen during stationary phase, but production of 4EP was initiated during the growth phase.
Growth and 4EP production of B. bruxellensis strains in defined medium with 20 mg/L coumaric acid added over 250 hr with air saturation at 0% (A), 25% (B), and 50% (C). Lines indicate cell densities and symbols indicate 4EP concentrations. Letters in parentheses indicate statistical significance.
Acetic acid concentrations were also evaluated for each strain and aeration condition and normalized to biomass (Table 3). For all strains in the absence of coumaric acid, a significantly higher concentration of acetic acid was produced at higher air saturation levels. Stains 2082 and 2091 showed significantly higher levels of acetic acid with 20 mg/L coumaric acid as compared to no coumaric acid at air saturations of 50% (2082) and 25% (2091). In strain 2077 no statistically significant difference was seen between the 0 and 20 mg/L coumaric acid samples at any air saturation level, but the trend of increasing acetic acid production in the presence of coumaric acid was observed. No significant effect of coumaric acid addition was seen at 0% air saturation for any of the three strains. This observation suggests that the reduction of coumaric acid can stimulate the production of acetic acid under some growth conditions. However, it is not clear if the stimulation of acetic acid production is to recycle oxidized cofactors or to block the toxicity of coumaric acid via reduction with the accompanied production of acetic acid as the co-reactant and oxidized substrate. Strain 2091, which showed a low production of 4EP, had among the highest levels of acetic acid produced at 25% air saturation. Taken as a whole, these data suggest that coumaric acid may be both inhibitory and stimulatory to growth and metabolic function depending on growth conditions and genetic background of the strain.
Acetic acid production normalized to biomass of B. bruxellensis individual strains grown with and without coumaric acid at 0%, 25%, and 50% air saturation.
Discussion
Brettanomyces bruxellensis has proved to be a highly diverse yeast both genetically (Hellborg and Piskur 2009, Woolfit et al. 2007) and physiologically (Conterno et al. 2006, Vigentini et al. 2008). We are interested in the metabolic diversity of Brettanomyces isolated from wines as a function of genotype. The impact of supplementation with specific cinnamic and amino acids was examined to assess the role of strain genotype on aromatic compound production profiles. In comparing the strain clusters based on metabolism (Figure 1) to those based on 26s rDNA sequencing (Table 1), the two “a” group strains clustered separately from the other strains and closer to each other. The two “b” group strains clustered with the “c” group strain; 2077, 2082, and 615 respectively. Thus, there is a correlation between genotypic differences revealed by sequencing and phenotypic differences in aromatic compound formation. There was also a correlation between the amount of 4EP and 4EG produced: 2077, 2082, and 615 produced high levels and 2058 and 2091 produced low levels in the wine tested. Lastly, a geographic correlation was also noted, as the French strains clustered together as did the strains from California and Chile.
When coumaric and ferulic acid were present in the growth medium, a number of other products in addition to 4EP and 4EG were identified that may contribute to Brettanomyces character in wine. Clustering based on metabolic products changed with the addition of hydroxycinnamic acid substrates to the medium. Other compounds that typically increased with the presence of these hydroxycinnamic acids included the ethyl esters of the C12, C14, and C16 fatty acids, all of which have a waxy, soapy character (Aldrich Flavors and Fragrances). Differences in the responses of the strains to the presence of cinnamic acids were also observed. In the presence of coumaric acid, levels of 3-methylbutanoic and octanoic acid decrease or increase, depending on the strain. These compounds have a rancid, rotten odor (Aldrich Flavors and Fragrances). While 4EP and 4EG production dominated in medium supplemented with coumaric and ferulic acid, the clustering of the strains grown with caffeic did not give as strong a pattern. However, we were unable to detect 4EC using this protocol (Hesford et al. 2004).
The aromatic amino acids also influenced the pattern of volatile compounds detected. There were two distinct subgroupings. All strains grown on phenylalanine were part of the same cluster, likely due to the production of phenethyl alcohol. Also included in this group were strain 2058 grown with tyrosine, tryptophan, and no supplementation and strain 2091 grown with tryptophan and tyrosine. The two “a” group strains therefore tended to cluster with all strains grown on phenylalanine. The other cluster consisted largely of the group “b” strains grown on caffeic acid, tyrosine, tryptophan, or no substrate. Within this group, the unsupplemented and caffeic acid supplemented samples tended to cluster, suggesting that caffeic acid had little impact on aromatic profile. Since the methodology used does not detect 4EC, this result is not surprising. The cluster subgroups indicated a slightly greater dependence on strain variability than substrate in the production of volatiles with the aromatic amino acids present. When the substrate is a preferred substrate, such as coumaric or ferulic acid or perhaps phenylalanine, the strain differences may be less important to the metabolic outcome. However, when the substrate is less dominant the strain differences seem to play a greater role. This complex mixture of compounds detected could account for the variety of aromas that are considered to constitute Brettanomyces spoilage in wines.
Strain 2058 produced some unusual compounds such as β-farnesene, phenethyl propionate, 1-octanol, and phenylacetaldehyde, which were not at all or rarely detected in the other strains (Supplemental Data). Strain 615 produced detectable levels of 1-octanol and β-farnesene with tyrosine and phenylalanine, respectively. Strain 2091 also produced some unusual compounds at detectable levels, such as isobutyric acid with tyrosine and ethyl isobutyrate when phenylalanine was present, although three of the other strains also produced ethyl isobutyrate in the presence of tyrosine. These results emphasize the variability of the metabolism of Brettanomyces strains beyond the previously reported differences in ethylphenol production (Dias et al. 2003, Harris et al. 2008), confirming what many winemakers have suspected: the types of Brettanomyces spoilage are both strain and substrate dependent.
The biological benefit of the production of volatile compounds is largely unknown. To assess the impact of the presence of coumaric acid and 4EG production, growth was evaluated at differing aeration levels for three strains with and without coumaric acid supplementation. If the oxygen level is high, then coumaric acid has a primarily negative affect on growth of Brettanomyces. This observation affirms that coumaric acid can be inhibitory to the growth of Brettanomyces and that 4EP production may serve to reduce toxicity of this compound.
When the oxygen levels were lower there seemed to be some growth advantage due to the presence of coumaric acid at mid- to late-log growth phase for strains 2077 and 2082. The conversion of vinylphenols to ethylphenols results in the conversion of NADH to NAD (Tchobanov et al. 2008, Godoy et al. 2008). When oxygen is present, respiration uses this conversion for a net energy gain via the electron transport chain. However, under low oxygen the NAD must be regenerated by alternative means, and we speculate that the production of the typical Brettanomyces odor compounds in wine are the result of the Brettanomyces using available substrates to regenerate reductant for further metabolism. Similar results were seen when acetaldehyde, pyruvic acid, and other carbonyl compounds were present (Scheffers 1961). However, there were striking strain differences in the response to the presence of coumaric acid. Strain 2091 did not show any growth advantage at any time with the addition of coumaric acid. Strain 2091 also consistently showed lower levels of 4EP produced at 25% and 50% air saturation. This finding suggests several possibilities. Strain 2091 may be more reliant on other substrates for the biological benefit enabled by coumaric acid, it may simply be limited enzymatically in ability to use coumaric acid, or it may have adjusted intermediary metabolism to no longer need whatever benefit coumaric acid may be providing to other strains.
Another line of evidence that conversion of 4EP may be used by the cells to regenerate reductant can be found in the acetic acid production data. While there are some strain differences, in general, when oxygen levels increased the amount of acetic acid produced also increased. When coumaric acid was present the increased level of acetic acid produced was even more pronounced. The production of a more oxidized substrate, acetic acid, without a corresponding increase in oxygen but in the presence of coumaric acid suggests that the conversion of NADH to NAD is rate limiting under these growth conditions. The conversion of 4-vinylphenol to 4EP may therefore alter the ratio of NAD to NADH, stimulating production of acetic acid.
Conclusion
Both the Brettanomyces strain and the available substrates have an impact on the volatile metabolites produced during growth of B. bruxellensis in a defined medium. The presence of specific substrates led to the formation of a subset of identical compounds in all strains, but there were significant differences in the concentrations of the volatiles produced. In addition, strains showed variability in the production of minor components that have the potential to impact wine aroma and that are associated with more positive attributes than the cinnamic acids. There was a correlation between both the geographical and sequence-based clustering of strains and the clustering by volatile compound production. Further analysis is necessary to determine which compounds produced by Brettanomyces have an impact on the aroma produced in wine. Coumaric acid was shown to have inhibitory, neutral, or beneficial impacts on the growth and viability of Brettanomyces depending on both the level of aeration and the strain, making prediction of off-character formation challenging under production conditions.
Acknowledgments
Acknowledgments: This research was supported by grants from the American Vineyard Foundation and the California Competitive Grants Program for Research in Viticulture and Enology. The authors thank Cary Doyle for assistance in methods development and analysis and David Block and Gian Oddone for loan of and help in setting up the bioreactors.
Footnotes
-
Supplemental data is freely available with the online version of this article.
- Received July 2012.
- Revision received December 2012.
- Accepted January 2013.
- ©2013 by the American Society for Enology and Viticulture
Literature Cited
Vol 64 Issue 2
