TY - JOUR
AU - Lievens, Bart
AB - Abstract Brettanomyces (Dekkera) bruxellensis is an ascomycetous yeast of major importance in the food, beverage and biofuel industry. It has been isolated from various man-made ecological niches that are typically characterized by harsh environmental conditions such as wine, beer, soft drink, etc. Recent comparative genomics studies revealed an immense intraspecific diversity, but it is still unclear whether this genetic diversity also leads to systematic differences in fermentation performance and (off-)flavor production, and to what extent strains have evolved to match their ecological niche. Here, we present an evaluation of the fermentation properties of eight genetically diverse B. bruxellensis strains originating from beer, wine and soft drinks. We show that sugar consumption and aroma production during fermentation are determined by both the yeast strain and composition of the medium. Furthermore, our results indicate a strong niche adaptation of B. bruxellensis, most clearly for wine strains. For example, only strains originally isolated from wine were able to thrive well and produce the typical Brettanomyces-related phenolic off-flavors 4-ethylguaiacol and 4-ethylphenol when inoculated in red wine. Sulfite tolerance was found as a key factor explaining the observed differences in fermentation performance and off-flavor production. Sequence analysis of genes related to phenolic off-flavor production, however, revealed only marginal differences between the isolates tested, especially at the amino acid level. Altogether, our study provides novel insights in the Brettanomyces metabolism of flavor production, and is highly relevant for both the wine and beer industry. (Off-)flavors, ecological niche, phenolic acid decarboxylase, vinylphenol reductase, volatile phenols INTRODUCTION The yeast Brettanomyces bruxellensis (teleomorph Dekkera bruxellensis) is of major importance in the fermentation industry, where it can coexist with, or sometimes even outcompete, the traditional fermentation yeast Saccharomyces cerevisiae. Despite that it is mainly known for its spoilage capacity in wine, it has been isolated from various man-made ecological niches, including (spontaneous) alcoholic fermentation processes (wine, beer, cider, bioethanol, etc.), soft drinks, dairy products, kombucha tea and sourdough (Crauwels et al.2015a; Steensels et al.2015; Smith and Divol 2016). The only source from which B. bruxellensis has been isolated that is not associated with industrial settings is grape berries (Renouf and Lonvaud-Funel 2007), illustrating its close association with man-made ecological niches. A common thread in these niches is the harsh environmental conditions that are detrimental for many microbes, such as high ethanol concentrations, low pH, the absence of readily fermentable nitrogen and carbon sources, and low oxygen (Smith and Divol 2016). While resistance to these stressors is not uncommon in microbes, there are only a few species that combine resistance to all of these stressors, which explains their high competitiveness in these niches (Smith and Divol 2016). Interestingly, the role of Brettanomyces in the food and beverage industry is very ambiguous (Crauwels et al.2015a; Steensels et al.2015). Brettanomyces bruxellensis is generally reported as a spoilage yeast responsible for off-flavor production in wine, beer, cider or dairy products, leading to huge economic losses. However, in some specialty beers such as lambic, Berliner Weiβe and traditional acidic ale beers, the presence of Brettanomyces is required to obtain the characteristic and complex ‘Brett flavor’, which is (among others) often described as clovy, spicy, floral and/or smoky (Licker, Acree and Heninck-Kling 1999). Notably, while volatile compounds such as 4-ethylguaiacol (4-EG) and 4-ethylphenol (4-EP) strongly contribute to off-flavors in, for example, wine, the same compounds are considered essential in the production of the specialty beers mentioned above. This dichotomy is in part due to the difference in relative concentration of these volatile phenols: beer generally contains higher concentrations of 4-EG (clove-like or spicy aroma), while wine contains more 4-EP (medicinal aroma) (Curtin et al.2012). The production of these phenolic compounds involves the sequential activity of two enzymes: the first is a phenolic acid decarboxylase, which decarboxylates hydroxycinnamic acids into the corresponding vinylphenol (4-vinylphenol (4-VP) from p-coumaric acid, or 4-vinylguaiacol (4-VG) from ferulic acid), and the second is a vinylphenol reductase, which reduces the vinylphenol into the corresponding ethylphenol (4-EP from 4-VP and 4-EG from 4-VG) (Godoy et al.2009; Laforgue and Lonvaud-Funel 2012). Surprisingly, the genes encoding these enzymes have only been recently identified. Godoy et al. (2014) described a gene encoding a phenolic acid decarboxylase in B. bruxellensis (DbPAD gene), whose function was verified by heterologous expression in S. cerevisiae. The vinyl phenol reductase enzyme was described and characterized by Granato et al. (2015), and found to have both vinyl phenol reductase and superoxide dismutase activity (further referred to as DbVPR/SOD). However, so far, no information is available about the genetic diversity of these genes in B. bruxellensis and how different gene variants may affect production of these phenolic compounds. Apart from heavily influencing the aroma profile of various foods, production of ethyl phenols may also have an ecological role. Indeed, it seems to pose a clever strategy of Brettanomyces to travel to new environments, as it was recently shown that these compounds can serve as an attractant for Drosophila fruit flies (Dweck et al.2015), and can therefore play a crucial role in the dispersal of the yeast through insect vectors, a mechanism which has also been described for S. cerevisiae (Christiaens et al.2014). Previous studies have shown a huge genetic and phenotypic diversity of B. bruxellensis yeasts (e.g. Conterno et al.2006; Martorell et al.2006; Miot-Sertier and Lonvaud-Funel 2007; Vigentini et al.2008, 2012; Borneman et al.2014; Crauwels et al.2014, 2015b). Interestingly, much like in S. cerevisiae (Gallone et al.2016), some of these studies reported a correlation between genotype and source of isolation (Conterno et al.2006; Vigentini et al.2012; Crauwels et al.2014), or even described links between specific features and their niche. For example, in Crauwels et al. (2015b), the assimilation of particular α- and β-glycosides by B. bruxellensis was found to be linked with strain origin. Despite knowledge about these genetic and phenotypic differences, only little is known about the behavior of different B. bruxellensis strains in different ecological niches and whether this impacts the production of the typical ‘Brett’ flavors. Therefore, we aim to assess differences in sugar consumption and (off-) flavor development by genetically distinct B. bruxellensis strains isolated from various ecological niches in different media known to be Brettanomyces habitats. More specifically, eight B. bruxellensis strains were subjected to five different fermentation conditions by inoculating them in blond ale (Maredsous Blond), strong golden pale ale (Duvel), red wine (Chinon, Domaine des hardonnieres 2011), white wine (Muscadet, Comtesse du val, 2011) and soft drink (Canada Dry), supplemented with a mixture of malto oligosaccharides. Following an incubation period of 31 days, the sugar and aromatic profiles were evaluated. Additionally, for each strain, the genes encoding the two enzymes involved in the biosynthesis of the typical Brettanomyces-related ethylphenols were sequenced and compared, and tolerance to the common antifungal agent sulfite was assessed. These analyses provide novel insights in Brettanomyces evolution and flavor metabolism, and can aid in selection of Brettanomyces starter cultures for bioflavoring and the production of novel beverages. MATERIAL AND METHODS Study strains Eight Brettanomyces bruxellensis isolates were used in this study. These isolates have been characterized genotypically and phenotypically in previous studies and represent different genotypes representative of their source of isolation (Crauwels et al.2014, 2015b). More specifically, studied strains included three isolates from lambic beer (ST05.12/22 (VIB X9085), ST05.12/26 (MUCL 49865) and ST05.12/53), three isolates from wine (ST05.12/56 (CBS 2499), ST05.12/62 (AWRI 1499) and ST05.12/63 (MUCL 54014)), and two isolates from soft drink (ST05.12/30 (CBS 8027) and ST05.12/59 (CBS 6055)) (Table S1, Supplementary Information). All isolates were stored at –80°C in glycerol-based standard storage medium (yeast extract peptone dextrose (YPD) broth containing 26.1% glycerol). Fermentation tests Isolates were subjected to small-scale fermentations in media in which B. bruxellensis is known to occur. These included (i) wine (red (Chinon, Domaine des hardonnieres, 2011) and white wine (Muscadet, Comtesse du val, 2011)), (ii) beer (blond ale (Maredsous Blond) and strong golden pale ale (Duvel)) and (iii) soft drink (Canada Dry). Test media were supplemented with malto oligosaccharides (Belgosuc) to obtain a final sugar content of at least 10 °P and were filtered through a 0.45 μm filter (bottle-top vacuum filtration). Subsequently, filtered media were degassed by creating a vacuum in the headspace for 15 min followed by overnight (12 h) shaking at 200 rpm and, afterwards again by creating a vacuum in the headspace for 15 min. Main characteristics (SO2, total proteins, total ammonia, alpha-amino nitrogen, ethanol and pH) of the fermentation media are given in Table S2 (Supplementary Information). Isolates were precultured for 5 days at 25°C on yeast peptone glucose (YPG; 10 g/l glucose, 10 g/l peptone and 5 g/l yeast extract) agar, followed by restreaking a single colony on YPG and incubation for 96 h under the same conditions. Next, each strain was picked up with a sterile cotton swab and suspended in 20 ml of saline water (0.8% NaCl) until an optical cell density of 0.85 was reached, measured using a spectrophotometer at a wavelength of 600 nm (corresponding to ∼1 × 106 cells/ml). Subsequently, 2 ml of this solution was inoculated in 50 ml conical flasks with 45 ml test medium (resulting in a final cell concentration of ∼4 × 104 cells per ml), followed by static incubation for 31 days at 25°C. Each fermentation was performed twice. Non-inoculated media were included as controls. After incubation, a 10-fold dilution series of each fermentation was plated on YPG agar (100 μl/plate). Plates were incubated at 25°C for 7 days and subsequently yeast colonies were counted. The identity of the recovered yeasts was confirmed by sequencing part of the large subunit of the ribosomal RNA gene as described previously (Crauwels et al.2014). At the same time, all media were filtered using a 0.45 μm filter (bottle-top vacuum filtration) to obtain cell free cultures. The cell-free media of both replicates were then combined and stored in sterile dark glass vials at –80°C until further analysis. Determination of sugar profiles Sugar profiles before and after yeast inoculation and incubation were determined using high-pressure liquid chromatography with refractive index detector (HPLC-RI), as described in De Rouck et al. (2013). More in particular, the following sugars were measured: fructose, glucose, saccharose, maltose, maltotriose, maltotetraose, maltopentaose (degree of polymerization (DP) 5), maltohexaose (DP6), maltoheptaose (DP7) and maltooctaose (DP8). Prior to HPLC analysis, proteins were removed from the samples by precipitation with Carrez reagent. Therefore, subsamples of 2 ml (both undiluted and 1:40 diluted) were mixed with 200 μl of Carrez-1 (106 g K4Fe(CN)6.3H2O dissolved in 1 l of demineralized water) and 200 μl of Carrez-2 (220 g Zn(CH3COO)2.2H2O and 30 ml of glacial acetic acid, made up to 1 l with demineralized water). After centrifugation at 11 000 × g for 5 min, the samples were ready for manual injection into the HPLC apparatus through a Rheodyne 7125 injection valve, equipped with a 20 μl sample loop. The HPLC sugar profiles of the subsamples were obtained using a Nucleosil NH2 100–5 column (250 × 4 mm i.d.; Macherey-Nagel, Düren, Germany), which was kept at ambient temperature. A mixture of water/acetonitrile (33/67; v/v) was used as mobile phase and delivered by a Shimadzu LC-6A pump at a flow rate of 1 ml/min. Components were detected by an RI detector (Erma HRG-7152; Erma Inc., Tokyo, Japan). Determination of aromatic profiles Aromatic profiles were determined as described in Van Opstaele et al. (2012). Briefly, first headspace solid-phase micro extractions (SPME) were performed using a CombiPal autosampler (CTC Analytics, Zwingen, Switzerland). Next, extracted compounds were separated and detected using gas chromatography-mass spectrometry (GC-MS). Prior to extraction, 1.5 g NaCl was brought into the SPME extraction vials (20 ml). Next, subsamples of 5 ml were brought to a total volume of 7.5 ml, by adding ultrapure water (for the wine samples) or absolute ethanol (for the Canada Dry and beer samples) leading to an alcohol concentration of 8% (v/v) in each of the samples. Extraction vials were closed with a polytetrafluoroethylene-coated septum. For the extraction itself, the extraction fiber, coated with polydimethylsiloxane (100 μm), was exposed into the headspace of the vial (25 mm); extraction time and temperature were set at 30 min and 40°C for extraction. Before the actual extraction, samples were preincubated at 40°C for 5 min. During preincubation and extraction, samples were stirred at 500 rpm. Gas chromatographic operating conditions were as follows. Extracted volatiles were thermally desorbed in the heated inlet (split/splitless injector, 250°C) of the Ultra Trace gas chromatograph (Thermo Fisher Scientific, Austin, TX, USA) for 3 min. Helium (Alphagaz 2, Air Liquide, Luik, Belgium) was used as a carrier gas at a constant flow of 1.0 ml/min. Injection was performed in the split mode (split ratio 1/10) for 3 min at 250°C. Separation of the injected compounds was performed on a 40 m × 0.18 mm i.d. × 0.20 μm film thickness RTX-1 capillary column (Restek Corp., Bellefonte, PA, USA). The oven temperature program for separation of the volatiles was as follows: 3 min at 35°C, followed by a temperature increase of 5°C/min to 250°C (1 min isotherm). Mass spectrometric detection of volatiles was performed by a dual stage quadrupole MS (DSQ II, Thermo Fisher Scientific) operating in the electron ionization mode (EI, 70 eV). The ion source temperature was set at 240°C, and the electron multiplier voltage was 1445 V. Analyses were performed in the full scan operating mode (m/z 40–400). The identity of extracted volatiles was confirmed by mass spectral comparison via the ‘NIST98’ and ‘Flavor MS Library for Xcalibur, 2003’ spectral libraries using the Xcalibur software (v.1.4 SR1, Thermo Fisher Scientific), retention times of authentic reference compounds and calculation of retention indices (KI). Retention indices were determined using a homologous series of normal alkanes (C8–C18; Sigma-Aldrich, St. Louis, MO, USA). Extracted volatiles were quantified by adding dodecane (C12 (≥99%; Sigma-Aldrich)) as internal standard prior to SPME extraction (10 μl of internal standard solution (23.9 μg C12/ml ethanol) to the samples). The concentration of the compound of interest was then calculated (in arbitrary units) on the basis of the ratio of the peak area of the compound to the peak area of the internal standard and to the concentration of added internal standard. Sequence analysis For each strain investigated, the complete open reading frames (ORFs) for both the DbPAD gene (666 bp), encoding a phenolic acid decarboxylase (Godoy et al.2014), and the DbVPR/SOD gene (465 bp), encoding an enzyme with both vinyl phenol reductase and superoxide dismutase activities (Granato et al.2015), were sequenced and investigated. To this end, DNA amplicons were produced using the primer sets PadF1-3 (5΄-ATG TAT ACG AAT GTT CTA ATA TTT-3΄) / PadR1-2 (5΄-CTA AAA GGT AAT TGC ATC AGG-3΄) and VprF1-2 (5΄-ATG GTT AAA GCA GTT GCA GTT-3΄) / VprR1-1 (5΄-TTA TGC AGA CAA GCC AAT GAC-3΄), respectively, developed based on the whole-genome sequence of B. bruxellensis AWRI 1499 (Curtin et al.2012). Amplification was performed in a reaction volume of 20 μl containing 312.5 μM of each deoxynucleoside triphosphate (dNTP), 1.0 μM of each primer, 1.25 units TaKaRa Ex Taq polymerase, 1× Ex Taq buffer (Clontech Laboratories, Palo Alto, CA) and 1 μl (10 ng) genomic DNA. Amplification was performed using a Bio-Rad T100 thermal cycler according to the following thermal profile: initial denaturation at 95°C for 2 min, followed by 30 cycles of 95°C for 1 min, 57°C for 45 s and 72°C for 1 min. A final 10-min extension step at 72°C concluded the protocol. Sequencing was performed using the same primers used for the amplification. Amplification and sequencing were performed in triplicate, based on which consensus sequences of the complete ORFs were assembled. Heterozygous sites were determined by identifying the presence of a double peak on the Sanger sequencing traces. Assessment of sulfite tolerance For each strain, sulfite tolerance was determined using the microplate method described by Curtin, Kennedy and Henschke (2012). Briefly, yeasts grown on YPD agar were transferred to 15-ml test tubes containing 5 ml Yeast Nitrogen Base (YNB; Difco, USA) and were incubated at 25°C for 72 h. Subsequently, a 96-well plate (Sterilin Limited, UK) was filled with 100 μl 2x YNB (40 g/l glucose; pH 3.5) containing different sodium metabisulfite (SMBS) concentrations and 100 μl inoculum having an OD600 of 0.5. The same medium without SMBS was used as a control. Furthermore, the microplate was sealed with an air permeable membrane (Thermo Scientific Nunc, USA) and incubated for 7 days at 25°C. Following incubation, absorbance was measured at 600 nm using a spectrophotometer (Multiskan GO microplate spectrophotometer, Thermo Fisher Scientific). Following background subtraction, absorbance values were corrected and maximal molar sulfite tolerance was determined. The experiment was performed three times, and averages of the three experiments were taken into account. Statistical analysis Both for sugar profiles and aromatic profiles, non-metric multidimensional scaling (nMDS) plots based on Bray-Curtis coefficient similarities were constructed using the Vegan package in R v12.2.1 (Development core team R 2006; Oksanen et al.2012) to create a 2D representation of the similarities between the different samples. Using nMDS, a ‘map’ is constructed on which ‘similar’ samples cluster closely together and ‘dissimilar’ samples plot far apart: the greater the distance between two data points, the more dissimilar the sugar or aromatic profiles of the samples. All ordinations were computed following 10 000 random starts. RESULTS Sugar profiles of Brettanomyces bruxellensis inoculated in different test media Table S3 (Supplementary Information) summarizes the initial sugar concentration in the different test media. Main differences were observed between the Canada Dry-based medium and the other test media for fructose, glucose and saccharose (higher concentrations in the Canada Dry-based medium). Further, the two tested beer-based media differed from the other media for maltotriose, maltotetraose and the DP5 to DP7 sugars (higher for the beer-based media). Following an incubation period of 31 days, all yeast strains were recovered from the beer-based and Canada Dry-based media, reaching densities between 105 and 107 colony forming units (cfu)/ml. In contrast, when yeasts were inoculated in red or white wine, densities dropped to levels below the detection limit (10 cfu/ml), except for the three wine isolates. While the pH of the different media remained fairly constant compared to the initial situation (data not shown), the sugar content and composition had altered during incubation, with most evident changes for soft drink and beer (Fig. 1 and Table S4, Supplementary Information). These changes were particularly due to consumption of the sugars with a lower degree of polymerization, including maltose, maltotriose, maltotetraose and maltopentaose (DP5), while sugars with higher degrees of polymerization, i.e. starting from DP6, were typically less or even not consumed (Table S4, Supplementary Information). Further, sugar consumption was shown to be both strain and medium dependent (Fig. 2). For example, maltose consumption in red wine was highly variable between strains (left-over concentrations ranged between 0.50 and 4.63 g/l (Table S4, Supplementary Information), while this variability was not observed in other media (Fig. 2A). Furthermore, in line with the overall limited growth of the tested strains in the wine media, the overall sugar consumption was low in these media (Fig. 2H). In contrast, all strains, with exception of the soft drink strain ST05.12/30, were able to efficiently consume maltose in the tested beer media (detected average concentration of ∼1 g/l) (Fig. 2A). Similar trends were observed for consumption of maltotriose (Fig. 2B), maltotetraose (Fig. 2C) and DP5 (Fig. 2D). It can also be seen from Fig. 2 that especially in the strong golden pale ale medium variation between strains was high for consumption of different maltooligosaccharides, except for maltose. Figure 1. View largeDownload slide nMDS plot (stress of plot = 0.08) based on Bray-Curtis coefficient similarities of the sugar profiles of the investigated test media (beverages supplemented with malto oligosaccharides) after inoculation with different B. bruxellensis strains and 31 days of static incubation at 25°C. Profiles for red wine, white wine, soft drink, strong golden pale ale and blond ale are represented by triangles, reverse triangles, circles, diamonds and squares, respectively (each data point represents a sugar profile). The origins of the different B. bruxellensis strains, i.e. beer, soft drink and wine, are highlighted in green, orange and red, respectively. Uninoculated media are represented by black symbols (sugar profiles determined at the start of the experiment). nMDS constructs a ‘map’ on which ‘similar’ samples cluster closely together and ‘dissimilar’ samples plot far apart: the greater the distance between two data points, the more dissimilar the samples (here the sugar profiles of the samples). Figure 1. View largeDownload slide nMDS plot (stress of plot = 0.08) based on Bray-Curtis coefficient similarities of the sugar profiles of the investigated test media (beverages supplemented with malto oligosaccharides) after inoculation with different B. bruxellensis strains and 31 days of static incubation at 25°C. Profiles for red wine, white wine, soft drink, strong golden pale ale and blond ale are represented by triangles, reverse triangles, circles, diamonds and squares, respectively (each data point represents a sugar profile). The origins of the different B. bruxellensis strains, i.e. beer, soft drink and wine, are highlighted in green, orange and red, respectively. Uninoculated media are represented by black symbols (sugar profiles determined at the start of the experiment). nMDS constructs a ‘map’ on which ‘similar’ samples cluster closely together and ‘dissimilar’ samples plot far apart: the greater the distance between two data points, the more dissimilar the samples (here the sugar profiles of the samples). Figure 2. View largeDownload slide Differential ability of B. bruxellensis strains (represented by box plots) to consume maltose (A), maltotriose (B), maltotetraose (C), DP5 (D), DP6 (E), DP7 (F) and DP8 (G) sugars, as well as the total of these sugars in general (H) in the different test media (beverages supplemented with a mixture of malto oligosaccharides) compared with the original content in the non-inoculated media (red line). Boxplots are a graphical representation of the five-number summary, the bottom and top of the box are the 25th and 75th percentile (the lower and upper quartiles, respectively), and the band near the middle of the box is the 50th percentile, i.e. the median. The whiskers at both ends of the box extend to the most extreme data point which is no more than 1.5 times the interquartile range. Outliers of the dataset are represented by points. Extensive substrate use by the yeast strains is noticeable when the median concentration (black line) is substantially lower than the concentration observed in the non-inoculated medium (red line). The height of the box indicates variation among strains: the larger the box, the more variation. Figure 2. View largeDownload slide Differential ability of B. bruxellensis strains (represented by box plots) to consume maltose (A), maltotriose (B), maltotetraose (C), DP5 (D), DP6 (E), DP7 (F) and DP8 (G) sugars, as well as the total of these sugars in general (H) in the different test media (beverages supplemented with a mixture of malto oligosaccharides) compared with the original content in the non-inoculated media (red line). Boxplots are a graphical representation of the five-number summary, the bottom and top of the box are the 25th and 75th percentile (the lower and upper quartiles, respectively), and the band near the middle of the box is the 50th percentile, i.e. the median. The whiskers at both ends of the box extend to the most extreme data point which is no more than 1.5 times the interquartile range. Outliers of the dataset are represented by points. Extensive substrate use by the yeast strains is noticeable when the median concentration (black line) is substantially lower than the concentration observed in the non-inoculated medium (red line). The height of the box indicates variation among strains: the larger the box, the more variation. Volatile compounds measured in the different test media Prior to yeast inoculation, 84 volatiles could be assigned to the different test media. After inoculation with the different yeast strains, 12 additional volatiles could be measured, bringing the total of assigned constituents to 96. Seventy-six volatiles were classified into nine pre-defined chemical compound classes, including ethyl esters (20), monoterpenes (16), higher alcohols (10), acetate esters (8), aldehydes (8), volatile phenols (5), organic acids (3), sequiterpenes (3) and ketones (1); 22 volatiles were classified as miscellaneous (for more details, see Table S5, Supplementary Information). It becomes clear that often differences occur depending on the strain and test medium used. For example, increased amounts of ethyl esters (Fig. 3A) and monoterpenes (Fig. 3B) were found in only strong golden ale and blond ale, respectively, while acetate ester concentration was reduced only in strong golden ale (Fig. 3D). Furthermore, higher concentrations of aldehydes were detected in wine after yeast inoculation, especially in white wine (Fig. 3E). However, other trends seemed to be consistent in all media; the concentration of volatile phenols (Fig. 3H) and ketones (Fig. 3I) increased, while the amount of organic acids (Fig. 3F) decreased. Fermentation had no or negligible influence on the concentration of higher alcohols and sequiterpenes (Fig. 3C and G). Figure 3. View largeDownload slide Volatile composition in the different test media (beverages supplemented with a mixture of malto oligosaccharides) inoculated with B. bruxellensis strains (represented by box plots) compared with the original status in the non-inoculated media (red line). Results are shown for ethyl esters (A), monoterpenes (B), higher alcohols (C), acetate esters (D), aldehydes (E), organic acids (F; observed concentration in non-inoculated white wine was 86.56 a.u.), sequiterpenes (G), volatile phenols (H) and ketones (I). Boxplots are a graphical representation of the five-number summary, the bottom and top of the box are the 25th and 75th percentile (the lower and upper quartiles, respectively), and the band near the middle of the box is the 50th percentile, i.e. the median. The whiskers at both ends of the box extend to the most extreme data point which is no more than 1.5 times the interquartile range. Outliers of the dataset are represented by points. Extensive production by the yeast strains is noticeable when the average concentration (black line) is substantially higher than the concentration observed in the non-inoculated medium (red line). The height of the box indicates variation among strains: the larger the box, the more variation. Concentrations are expressed in arbitrary units (a.u.). Figure 3. View largeDownload slide Volatile composition in the different test media (beverages supplemented with a mixture of malto oligosaccharides) inoculated with B. bruxellensis strains (represented by box plots) compared with the original status in the non-inoculated media (red line). Results are shown for ethyl esters (A), monoterpenes (B), higher alcohols (C), acetate esters (D), aldehydes (E), organic acids (F; observed concentration in non-inoculated white wine was 86.56 a.u.), sequiterpenes (G), volatile phenols (H) and ketones (I). Boxplots are a graphical representation of the five-number summary, the bottom and top of the box are the 25th and 75th percentile (the lower and upper quartiles, respectively), and the band near the middle of the box is the 50th percentile, i.e. the median. The whiskers at both ends of the box extend to the most extreme data point which is no more than 1.5 times the interquartile range. Outliers of the dataset are represented by points. Extensive production by the yeast strains is noticeable when the average concentration (black line) is substantially higher than the concentration observed in the non-inoculated medium (red line). The height of the box indicates variation among strains: the larger the box, the more variation. Concentrations are expressed in arbitrary units (a.u.). The nMDS plot based on the complete data set grouped the different aromatic profiles into five clusters, corresponding to the five media tested (Fig. 4A). Not surprisingly, the aromatic profiles of both beer varieties (blond ale and strong golden pale ale) as well as both wine varieties (red and white wine) clustered closely together (Fig. 4A). Furthermore, it becomes apparent that the aromatic profiles were altered by the fermentation process, which was most notable for the soft drink, strong golden pale ale, blond ale and red wine (Fig. 4A). Interestingly, the nMDS plot also suggests that the aromatic profiles of the red wine samples may be linked with the original source of isolation of the inoculated strains. More specifically, strains isolated from wine (ST05.12/56, ST05.12/62 and ST05.12/63) had a distinct effect on the aromatic profiles of red wine: nMDS grouped them closely together in one cluster, and at a certain distance from the other inoculated and non-inoculated red wine samples (Fig. 4B). Figure 4. View largeDownload slide (A) nMDS plot (stress of plot = 0.10) based on Bray-Curtis coefficient similarities of the aromatic profiles of the investigated test media (beverages supplemented with malto oligosaccharides) after inoculation with different B. bruxellensis strains and 31 days of static incubation at 25°C. Profiles for red wine, white wine, soft drink, strong golden pale ale and blond ale are represented by triangles, reverse triangles, circles, diamonds and squares, respectively (each data point represents an aromatic profile). The origins of the different B. bruxellensis strains, i.e. beer, soft drink and wine, are highlighted in green, orange and red, respectively. Profiles from uninoculated media are represented by black symbols (aromatic profiles determined at the start of the experiment). (B) nMDS plot (stress of plot = 0.09) based on Bray-Curtis coefficient similarities of the aromatic profiles of red wine samples inoculated with the different B. bruxellensis strains investigated in this study. The origins of the different B. bruxellensis strains, i.e. beer, soft drink and wine, are highlighted in green, orange and red, respectively. The profiles from the uninoculated medium (determined at the start of the experiment) are highlighted in black. It is clear from this panel that the three investigated wine strains are responsible for a distinct aromatic profile compared to the other strains. nMDS constructs a ‘map’ on which ‘similar’ samples cluster closely together and ‘dissimilar’ samples plot far apart: the greater the distance between two data points, the more dissimilar the samples (here the aromatic profiles of the samples). Figure 4. View largeDownload slide (A) nMDS plot (stress of plot = 0.10) based on Bray-Curtis coefficient similarities of the aromatic profiles of the investigated test media (beverages supplemented with malto oligosaccharides) after inoculation with different B. bruxellensis strains and 31 days of static incubation at 25°C. Profiles for red wine, white wine, soft drink, strong golden pale ale and blond ale are represented by triangles, reverse triangles, circles, diamonds and squares, respectively (each data point represents an aromatic profile). The origins of the different B. bruxellensis strains, i.e. beer, soft drink and wine, are highlighted in green, orange and red, respectively. Profiles from uninoculated media are represented by black symbols (aromatic profiles determined at the start of the experiment). (B) nMDS plot (stress of plot = 0.09) based on Bray-Curtis coefficient similarities of the aromatic profiles of red wine samples inoculated with the different B. bruxellensis strains investigated in this study. The origins of the different B. bruxellensis strains, i.e. beer, soft drink and wine, are highlighted in green, orange and red, respectively. The profiles from the uninoculated medium (determined at the start of the experiment) are highlighted in black. It is clear from this panel that the three investigated wine strains are responsible for a distinct aromatic profile compared to the other strains. nMDS constructs a ‘map’ on which ‘similar’ samples cluster closely together and ‘dissimilar’ samples plot far apart: the greater the distance between two data points, the more dissimilar the samples (here the aromatic profiles of the samples). Production of typical (off-)flavors by Brettanomyces bruxellensis during fermentation tests Brettanomyces bruxellensis is known for formation of specific (off-) flavors caused by volatile phenols such as 4-VG, 4-VP, 4-EG and 4-EP. Among the different media tested, surprisingly, these typical ‘Brett’ (off-) flavors were only detected in the red wine and strong golden pale ale media (Table 1 and Table S5, Supplementary Information). Furthermore, this production seems at least partly strain dependent, as only the red wine samples inoculated with the wine strains (ST05.12/56, ST05.12/62 and ST05.12/63) were found to contain 4-EG and 4-EP (Table 1), probably due to the inability of the other strains to grow in this medium. In the strong golden pale ale samples, all strains tested produced 4-EG and 4-EP. In general, the concentration of 4-EG was much higher than that of 4-EP in the strong golden pale ale samples, resulting in a ratio of 4-EP and 4-EG over 20. In red wine, the relative concentration of both compounds was equal to or below 0.6 (Table 1). Table 1. Concentration (a.u.) of volatile phenols, i.e. 4-VG, 4-VP, 4-EG and 4-EP, and the ratio of 4-EG and 4-EP in red wine and strong golden pale alea after inoculation with different B. bruxellensis strains and 31 days of static incubation at 25°C.     Red wineb  Strong golden pale alec    Strain  Isolation source  4-VG  4-EG  4-VP  4-EP  4-EG:4-EP  4-VG  4-EG  4-VP  4-EP  4-EG:4-EP  Blankd  –  0.0  0.0  0.0  0.0  n.a.e  4.2  0.0  0.0  0.0  n.a.  ST05.12/22  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  10.2  0.0  0.5  22.7  ST05.12/26  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  17.8  0.0  0.6  28.3  ST05.12/53  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  16.3  0.0  0.6  25.4  ST05.12/30  Soft drink  0.0  0.0  0.0  0.0  n.a.  0.0  12.4  0.0  0.5  24.8  ST05.12/59  Soft drink  0.0  0.0  0.0  0.0  n.a.  0.0  23.1  0.0  0.8  28.7  ST05.12/56  Wine  0.0  1.3  0.0  2.4  0.6  0.0  26.7  0.0  1.3  21.0  ST05.12/62  Wine  0.0  1.7  0.0  6.3  0.3  0.0  24.6  0.0  1.0  25.4  ST05.12/63  Wine  0.0  2.0  0.0  4.9  0.4  0.0  16.5  0.0  0.6  29.1      Red wineb  Strong golden pale alec    Strain  Isolation source  4-VG  4-EG  4-VP  4-EP  4-EG:4-EP  4-VG  4-EG  4-VP  4-EP  4-EG:4-EP  Blankd  –  0.0  0.0  0.0  0.0  n.a.e  4.2  0.0  0.0  0.0  n.a.  ST05.12/22  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  10.2  0.0  0.5  22.7  ST05.12/26  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  17.8  0.0  0.6  28.3  ST05.12/53  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  16.3  0.0  0.6  25.4  ST05.12/30  Soft drink  0.0  0.0  0.0  0.0  n.a.  0.0  12.4  0.0  0.5  24.8  ST05.12/59  Soft drink  0.0  0.0  0.0  0.0  n.a.  0.0  23.1  0.0  0.8  28.7  ST05.12/56  Wine  0.0  1.3  0.0  2.4  0.6  0.0  26.7  0.0  1.3  21.0  ST05.12/62  Wine  0.0  1.7  0.0  6.3  0.3  0.0  24.6  0.0  1.0  25.4  ST05.12/63  Wine  0.0  2.0  0.0  4.9  0.4  0.0  16.5  0.0  0.6  29.1  a 4-VG, 4-VP, 4-EG and 4-EP were not detected in the other media (white wine, blond ale and soft drink) inoculated with the yeasts. b Chinon, Domaine des hardonnieres, 2011. c Duvel. d Non-inoculated test medium, volatiles determined at the start of the experiment. e n.a., not applicable. View Large Table 1. Concentration (a.u.) of volatile phenols, i.e. 4-VG, 4-VP, 4-EG and 4-EP, and the ratio of 4-EG and 4-EP in red wine and strong golden pale alea after inoculation with different B. bruxellensis strains and 31 days of static incubation at 25°C.     Red wineb  Strong golden pale alec    Strain  Isolation source  4-VG  4-EG  4-VP  4-EP  4-EG:4-EP  4-VG  4-EG  4-VP  4-EP  4-EG:4-EP  Blankd  –  0.0  0.0  0.0  0.0  n.a.e  4.2  0.0  0.0  0.0  n.a.  ST05.12/22  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  10.2  0.0  0.5  22.7  ST05.12/26  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  17.8  0.0  0.6  28.3  ST05.12/53  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  16.3  0.0  0.6  25.4  ST05.12/30  Soft drink  0.0  0.0  0.0  0.0  n.a.  0.0  12.4  0.0  0.5  24.8  ST05.12/59  Soft drink  0.0  0.0  0.0  0.0  n.a.  0.0  23.1  0.0  0.8  28.7  ST05.12/56  Wine  0.0  1.3  0.0  2.4  0.6  0.0  26.7  0.0  1.3  21.0  ST05.12/62  Wine  0.0  1.7  0.0  6.3  0.3  0.0  24.6  0.0  1.0  25.4  ST05.12/63  Wine  0.0  2.0  0.0  4.9  0.4  0.0  16.5  0.0  0.6  29.1      Red wineb  Strong golden pale alec    Strain  Isolation source  4-VG  4-EG  4-VP  4-EP  4-EG:4-EP  4-VG  4-EG  4-VP  4-EP  4-EG:4-EP  Blankd  –  0.0  0.0  0.0  0.0  n.a.e  4.2  0.0  0.0  0.0  n.a.  ST05.12/22  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  10.2  0.0  0.5  22.7  ST05.12/26  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  17.8  0.0  0.6  28.3  ST05.12/53  Beer  0.0  0.0  0.0  0.0  n.a.  0.0  16.3  0.0  0.6  25.4  ST05.12/30  Soft drink  0.0  0.0  0.0  0.0  n.a.  0.0  12.4  0.0  0.5  24.8  ST05.12/59  Soft drink  0.0  0.0  0.0  0.0  n.a.  0.0  23.1  0.0  0.8  28.7  ST05.12/56  Wine  0.0  1.3  0.0  2.4  0.6  0.0  26.7  0.0  1.3  21.0  ST05.12/62  Wine  0.0  1.7  0.0  6.3  0.3  0.0  24.6  0.0  1.0  25.4  ST05.12/63  Wine  0.0  2.0  0.0  4.9  0.4  0.0  16.5  0.0  0.6  29.1  a 4-VG, 4-VP, 4-EG and 4-EP were not detected in the other media (white wine, blond ale and soft drink) inoculated with the yeasts. b Chinon, Domaine des hardonnieres, 2011. c Duvel. d Non-inoculated test medium, volatiles determined at the start of the experiment. e n.a., not applicable. View Large Sequence analysis Sequence analysis of the complete ORFs of the DbPAD gene (666 bp) and DbVPR/SOD gene (465 bp) allowed us to assess the genetic diversity of these genes among the different strains as well as the level of heterozygosity within a single strain. At the nucleotide level, 35 and 18 polymorphic sites were observed for the DbPAD gene and DbVPR/SOD gene, respectively, among which 13 and 2 represented non-synomynous substitutions (Table S6, Supplementary Information). For the DbVPR/SOD gene, between 3 and 12 heterozygous sites were found (0.65%–2.58%) within a single strain; for the DbPAD gene, the number of heterozygous sites varied between 0 and 18 (0%–2.70%) (Table S6, Supplementary Information). While most heterozygous sites for the DbPAD gene were observed for the soft drink strains (ST05.12/30 (18 heterozygous positions) and ST05.12/59 (16 heterozygous positions)), no heterozygous sites were found for the investigated beer and wine strains (with exception of wine strain ST05.12/63 (8 heterozygous positions) (Table S6, Supplementary Information). Interestingly, while all investigated beer strains (ST05.12/22, ST05.12/26 and ST05.12/53) were found to have only one variant of the DbVPR/SOD protein, the investigated wine (ST05.12/56, ST05.12/62 and ST05.12/63) and soft drink strains (ST05.12/30 and ST05.12/59) contained (at least) one additional protein variant (>99% identity) (Table S6, Supplementary Information). The nucleotide sequences obtained in this study have been deposited in GenBank under the accession numbers KX752249–KX752264. Sulfite tolerance The maximal level of molar SO2 tolerated by the isolates tested in this study varied between 0.20 (for ST05.12/53) and 0.55 mg l−1 (for ST05.12/62). On average, the maximal sulfite tolerance observed was 0.26 mg l−1 for the beer and soft drink isolates, while the wine isolates tolerated up to 0.43 mg l−1, illustrating that the tested wine isolates are more tolerant to sulfite (Fig. S1, Supplementary Information). DISCUSSION Although an increasing number of studies have reported on the genotypic and phenotypic diversity of Brettanomyces bruxellensis yeasts (e.g. Conterno et al.2006; Martorell et al.2006; Miot-Sertier and Lonvaud-Funel 2007; Vigentini et al.2008, 2012; Borneman et al.2014; Crauwels et al.2014, 2015b), little is known about the behavior of these yeasts in their natural environment and to what extent strains have evolved to match their ecological niche. Here, we tested fermentation performance and aroma production of eight B. bruxellensis strains from beer, wine and soft drink in small-scale fermentation assays in five media (based on blond ale (Maredsous Blond), strong golden pale ale (Duvel)), red wine (Domaine des hardonnieres), white wine (Muscadet, Comtesse du val) and soft drink (Canada Dry)), representative for different habitats of the yeast (Crauwels et al.2015a; Smith and Divol 2016). Inoculation of the yeasts generally changed the sugar content and sugar composition of the media, with most evident changes for soft drink and beer. In the tested wines, however, only strains originating from wine were able to grow, and consequently only these strains resulted in changes in sugar content and composition in these media. Likewise, yeast inoculation was found to alter the volatile composition of the media. Notably, the three B. bruxellensis strains isolated from wine similarly affected the aromatic profiles of red wine. More in particular, these strains were the only strains producing the typical Brettanomyces off-flavors 4-EG and 4-EP (Chatonnet et al.1992; Swiegers et al.2005) in red wine. Plate counting revealed that only wine yeasts could be recovered from the wine media after an incubation period of 31 days, suggesting strong niche adaptation of B. bruxellensis wine strains. Wine typically contains sulfite to prevent microbial contamination. Assessment of sulfite tolerance revealed that among the different isolates tested the wine isolates were more tolerant against sulfite compared to the other isolates (tolerating on average up to 0.43 mg l−1 molar SO2), thus explaining (at least partially) why only the wine strains were able to thrive well in the wine-based media. These results are in line with Curtin, Kennedy and Henschke (2012) who found B. bruxellensis wine isolates generally more tolerant to sulfite than B. bruxellensis strains isolated from beer or grape must. Tolerance to sulfite is a well-known domestication phenotype in Saccharomyces cerevisiae wine strains which has been shown to be the result of a chromosomal rearrangement (Pérez-Ortín et al.2002; Gallone et al.2016). It would be interesting to investigate whether similar mechanisms are at play in sulfite resistant B. bruxellenis strains. It is clear from our results that fermentation performance and aroma production are dependent on both the yeast strain and medium composition, including available sugars, nitrogen sources, volatile precursors, stressors, etc. Obvious differences between strains were observed for consumption of the maltooligosaccharides maltose, maltotriose and maltotetraose, which were not (or less efficiently) used by the soft drink strain ST05.12/30. This suggests that this strain lacks one or more essential genes involved in the assimilation of these sugars. Indeed, genome analysis of a collection of B. bruxellensis strains revealed that this strain (together with two other soft drink isolates (ST05.12/21 and ST05.12/59)) lacked a maltase gene that was suggested to play a role in the assimilation of specific α-linked sugars such as maltose (Crauwels et al.2015b). Surprisingly however, ST05.12/59 was shown to be able to use maltose, maltotriose and maltotetraose, hinting towards the presence of (a) yet unidentified enzyme(s) with maltase activity. Differences in metabolic products caused by different strains have also been observed by other research teams. For example, Joseph et al. (2013) observed pronounced differences in the production of metabolic products between B. bruxellensis strains when caffeic acids or the aromatic amino acids phenylalanine tryptophan, and tyrosine were used. In contrast, in the presence of coumaric and ferulic acids, all strains produced very similar metabolic products, primarily 4-EP and 4-EG, respectively (Joseph et al.2013). Brettanomyces bruxellensis is almost unique among other yeasts in its capacity to take up hydroxycynnamic acids and convert them into ethyl phenols, resulting in the typical ‘Brett’ taints (Crauwels et al.2015a; Steensels et al.2015). Wine spoilage by Brettanomyces is generally characterized by a ratio of 4-EG and 4-EP of <1. In contrast, in beer the ratio is generally above 20 (Curtin et al.2012). Similar ratios of both ethyl compounds were observed in our study in the cases where these compounds were formed. More specifically, in red wine we found a relative concentration of 4-EG over 4-EP of 0.6 or lower, in the strong golden pale ale samples this was above 20. PCR amplification and sequencing of the genes encoding the enzymes involved in the production of these phenolic metabolites, i.e. the DbPAD gene as well as the DbVPR/SOD gene, suggested strong sequence conservation within these genes, especially at the amino acid level for DbVPR/SOD. Interestingly, while all three investigated beer strains had only one variant of the enzyme, the investigated wine and soft drink strains contained (at least) one additional protein variant. Further research is needed to find out if this would affect production of the 4-ethyl derivatives 4-EP and 4-EG. Interestingly, from all terpenes measured, γ-terpineol (a monoterpene alcohol) was the only one that increased in quantity in the blond ale samples after yeast inoculation, reaching concentrations between 169 and 281 μg/l, while at the start of the fermentation a concentration of 105 μg/l was measured. This illustrates the yeasts’ potential to liberate compounds such as terpineol (sweet floral odor) that occur in hops or other herbs as non-volatile aroma molecules bound to glucose (Kollmannsberger, Biendl and Nitz 2006). Nevertheless, these so-called aglycons can strongly contribute to the aroma of food products when they are released from the sugar molecule, e.g. through enzymatic hydrolysis by β-glucosidases. Brettanomyces bruxellensis is known to contain at least one β-glucosidase (Crauwels et al.2014, 2015b) and has recently been shown to encompass strains with exceptionally high β-glucosidase activity, offering new possibilities for bioflavoring of food products (Vervoort et al.2016). Indeed, our results also confirm that specific strains of B. bruxellensis hold promising possibilities for bioflavoring of beers and other foods by their β-glucosidase activity, but further research is needed to find out more about these possibilities for our strains. To conclude, we have shown in this study that different B. bruxellensis strains may behave differently in various habitats with respect to sugar consumption and (off-)flavor production. Both sugar and aromatic profiles were found to be determined by both the yeast strain and composition of the medium. This knowledge may be relevant for the wine industry as well as for the beer industry, especially as brewers are more and more interested in novel beverages, e.g. by refermenting and maturing existing beer styles in oak vessels, either by spontaneous fermentation or by inoculation with a (mixed) starter culture containing non-conventional yeasts (Steensels and Verstrepen 2014; Crauwels et al.2015a). Acknowledgments We are grateful to everybody who provided us with Brettanomyces strains to perform this study. Conflict of interest. None declared. REFERENCES Borneman A , Zeppel R, Chambers Pet al.  . Insights into the Dekkera bruxellensis genomic landscape: comparative genomics reveals variations in ploidy and nutrient utilisation potential amongst wine isolates. PLoS Genet  2014; 10: e1004161 Google Scholar CrossRef Search ADS PubMed  Chatonnet P , Dubourdieu D, Boidron JNet al.  . The origin of ethylphenols in wines. J Sci Food Agr  1992; 60: 165– 78 Google Scholar CrossRef Search ADS   Christiaens J , Franco L, Cools Tet al.  . The fungal aroma gene ATF1 promotes dispersal of yeast cells through insect vectors. Cell Rep  2014; 9: 425– 32 Google Scholar CrossRef Search ADS PubMed  Conterno L , Lucy Joseph C, Arvik Tet al.  . Genetic and physiological characterization of Brettanomyces bruxellensis strains isolated from wines.. Am J Enol Viticult  2006; 57: 139– 47 Crauwels S , Steensels J, Aerts Get al.  . Brettanomyces bruxellensis, essential contributor in spontaneous beer fermentations providing novel opportunities for the brewing industry. Brewing Sci  2015a; 68: 110– 21 Crauwels S , Van Assche A, de Jonge Ret al.  . Comparative phenomics and targeted use of genomics reveals variation in carbon and nitrogen assimilation amongst different Brettanomyces bruxellensis strains. Appl Microbiol Biot  2015b; 99: 9123– 34 Google Scholar CrossRef Search ADS   Crauwels S , Zhu B, Steensels Jet al.  . Assessing genetic diversity in Brettanomyces yeasts using DNA fingerprinting and whole genome sequencing. Appl Environ Microb  2014; 80: 4398– 413 Google Scholar CrossRef Search ADS   Curtin C , Borneman A, Chambers Pet al.  . De-Novo assembly and analysis of the heterozygous triploid genome of the wine spoilage yeast Dekkera bruxellensis AWRI1499. PLoS One  2012; 7: e33840 Google Scholar CrossRef Search ADS PubMed  Curtin C , Kennedy E, Henschke PA. Genotype dependent sulfite tolerance of Australian Dekkera (Brettanomyces) bruxellensis wine isolates. Lett Appl Microbiol  2012; 55: 56– 61 Google Scholar CrossRef Search ADS PubMed  De Rouck G , Jaskula-Goiris B, De Causmaecker Bet al.  . The impact of wort production on the flavour quality and stability of pale lager beers. Brewing Sci  2013; 66: 1– 11 Dweck H , Ebrahim S, Farhan Aet al.  . Olfactory proxy detection of dietary antioxidants in Drosophila. Curr Biol  2015; 25: 455– 66 Google Scholar CrossRef Search ADS PubMed  Gallone B , Steensels J, Prahl Tet al.  . Domestication and divergence of Saccharomyces cerevisiae beer yeasts. Cell  2016; 166: 1397– 1410 Google Scholar CrossRef Search ADS PubMed  Godoy L , Garcia V, Pena Ret al.  . Identification of the Dekkera bruxellensis phenolic acid decarboxylase (PAD) gene responsible for wine spoilage. Food Control  2014; 45: 81– 6 Google Scholar CrossRef Search ADS   Godoy L , Garrido D, Martinez Cet al.  . Study of the coumarate decarboxylase and vinylphenol reductase activities of Dekkera bruxellensis (anamorph Brettanomyces bruxellensis) isolates. Lett Appl Microbiol  2009; 48: 452– 7 Google Scholar CrossRef Search ADS PubMed  Granato T , Romano D, Vigentini Iet al.  . New insights on the features of the vinyl phenol reductase from the wine-spoilage yeast Dekkera/Brettanomyces bruxellensis. Ann Microbiol  2015; 65: 321– 9 Google Scholar CrossRef Search ADS   Joseph C , Gorton L, Ebeler Set al.  . Production of volatile compounds by wine strains of Brettanomyces bruxellensis grown in the presence of different precursor substrates. Am J Enol Viticult  2013; 64: 231– 40 Google Scholar CrossRef Search ADS   Kollmannsberger H , Biendl M, Nitz S. Occurence of glycosidically bound flavour compounds in hops, hop products and beer. Brewing Sci  2006; 59: 83– 9 Laforgue R , Lonvaud-Funel A. Hydroxycinnamic acid decarboxylase activity of Brettanomyces bruxellensis involved in volatile phenol production: relationship with cell viability. Food Microbiol  2012; 32: 230– 4 Google Scholar CrossRef Search ADS PubMed  Licker J , Acree T, Heninck-Kling T. What is ‘Brett’ (Brettanomyces) flavor?: a preliminary investigation, Waterhouse AL, Ebeler SE, Chemistry of Wine Flavor  1st edn Washington, DC, ACS Symposium Series, 1999, 96– 115 Martorell P , Barata A, Malfeito-Ferreira Met al.  . Molecular typing of the yeast species Dekkera bruxellensis and Pichia guilliermondii recovered from wine related sources. Int J Food Microbiol  2006; 106: 79– 84 Google Scholar CrossRef Search ADS PubMed  Miot-Sertier C , Lonvaud-Funel A. Development of a molecular method for the typing of Brettanomyces bruxellensis (Dekkera bruxellensis) at the strain level. J Appl Microbiol  2007; 102: 555– 62 Google Scholar CrossRef Search ADS PubMed  Oksanen J , Blanchet G, Kindt Ret al.  , Vegan: Community Ecology, Package  R package version 2.0-5, 2012 https://cran.r-project.org/web/packages/vegan/index.html (30 December 2016, date last accessed) Pérez-Ortín JE , Querol A, Puig Set al.  . Molecular characterization of a chromosomal rearrangement involved in the adaptive evolution of yeast strains. Genome Res  2002; 12: 1533– 9 Google Scholar CrossRef Search ADS PubMed  Renouf V , Lonvaud-Funel A. Development of an enrichment medium to detect Dekkera/Brettanomyces bruxellensis, a spoilage wine yeast, on the surface of grape berries. Microbiol Res  2007; 162: 154– 67 Google Scholar CrossRef Search ADS PubMed  Smith BD , Divol D. Brettanomyces bruxellensis, a survivalist prepared for the wine apocalypse and other beverages. Food Microbiol  2016; 59: 161– 75 Google Scholar CrossRef Search ADS PubMed  Steensels J , Daenen L, Malcorps Pet al.  . Brettanomyces yeast - From spoilage organism to valuable contributors to industrial fermentation. Int J Food Microbiol  2015; 206: 24– 38 Google Scholar CrossRef Search ADS PubMed  Steensels J , Verstrepen KJ. Taming wild yeast: potential of conventional and nonconventional yeasts in industrial fermentations. Ann Rev Microbiol  2014; 68: 61– 80 Google Scholar CrossRef Search ADS   Swiegers J , Bartowsky E, Henschke Pet al.  . Yeast and bacterial modulation of wine aroma and flavor. Aust J Grape Wine R  2005; 11: 139– 73 Google Scholar CrossRef Search ADS   Van Opstaele F , De Causmaecker B, Aerts Get al.  . Characterization of novel varietal floral hop aromas by headspace solid phase microextraction and gas chromatography-mass spectrometry/olfactometry. J Agr Food Chem  2012; 60: 12270– 81 Google Scholar CrossRef Search ADS   Vervoort Y , Herrera-Malaver B, Mertens Set al.  . Characterization of the recombinant Brettanomyces anomalus β-glucosidase and its potential for bioflavoring. J Appl Microbiol  2016, DOI: 10.1111/jam.13200 Vigentini I , De Lorenzis G, Picozzi Cet al.  . Intraspecific variations of Dekkera/Brettanomyces bruxellensis genome studied by capillary electrophoresis separation of the intron splice site profiles. Int J Food Microbiol  2012; 157: 6– 15 Google Scholar CrossRef Search ADS PubMed  Vigentini I , Romano A, Compagno Cet al.  . Physiological and oenological traits of different Dekkera/Brettanomyces bruxellensis strains under wine-model conditions. FEMS Yeast Res  2008; 8: 1087– 96 Google Scholar CrossRef Search ADS PubMed  © FEMS 2016. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
TI - Fermentation assays reveal differences in sugar and (off-) flavor metabolism across different Brettanomyces bruxellensis strains
JF - FEMS Yeast Research
DO - 10.1093/femsyr/fow105
DA - 2017-01-01
UR - https://www.deepdyve.com/lp/oxford-university-press/fermentation-assays-reveal-differences-in-sugar-and-off-flavor-V0DnqWvJzz
SP - fow105
VL - 17
IS - 1
DP - DeepDyve
ER -