Physiological and oenological traits of different Dekkera/Brettanomyces bruxellensis strains under wine-model conditions

Yeast strains

The yeasts compared in this study are all classified as D. bruxellensis strains and belong to the International CBS collection. Seventeen strains were screened under aerobic conditions in the synthetic medium reported below: D.b.CBS73, D.b.CBS1940, D.b.CBS1941, D.b.CBS1942, D.b.CBS1943, D.b.CBS2336, D.b.CBS2499, D.b.CBS2547, D.b.CBS2796, D.b.CBS2797, D.b.CBS4459, D.b.CBS4601, D.b.CBS4602, D.b.CBS4480, D.b.CBS4481, D.b.CBS4482, D.b.CBS5206. They have various geographical and ecological origins as they were isolated from grape musts or wines of different countries (Table 1). A wine strain of Saccharomyces cerevisiae (L288) belonging to the E.S.A.V.E collection (Tebano, Italy) was used as a reference strain. All strains were maintained at −80 °C in YPD medium [1% yeast extract, 2% peptone, 2% glucose (w/v)] with 20% (v/v) glycerol.

Five strains (D.b.CBS4459, D.b.CBS4481, D.b.CBS4601, D.b.CBS2499 and D.b.CBS2336) producing different amounts of volatile phenols were then analysed in more detail.

Growth media and conditions

All strains were revitalized on YPD agar plates adjusted at pH 5.6. Experiments were carried out in duplicate using a synthetic medium similar in composition to a wine with the following formulation: 6.7 g L⁻¹ yeast nitrogen base (Difco), 1 g L⁻¹ fructose, 5 g L⁻¹ glycerol, 5 g L⁻¹ tartaric acid, 0.5 g L⁻¹l-malic acid, 0.2 g L⁻¹ citric acid, 4 g L⁻¹l-lactic acid, 0.12 g L⁻¹ NH₄Cl, 0.02 g L⁻¹ uracil, 5 mg L⁻¹ oleic acid, 0.5 mL L⁻¹ Tween80 and 15 mg L⁻¹ ergosterol, 0.18 g L⁻¹ peptone. The latter allowed an amino acid content close to must and wine to be achieved (Usseglio-Tomasset & Bosia, 1990). The concentration of fructose used can be easily found in a dry wine, because a similar amount of residual sugar is on an average left over at the end of alcoholic fermentation (Boulton et al., 1996). The medium was adjusted to pH 3.3, autoclaved and finally added with ethanol 11.5% (v/v). Ten milligrams per litre each of p-coumaric and ferulic acid were added.

Two parallel experiments called Experiment WithOut Adaptation (EWOA) and Experiment With Adaptation (EWA), respectively, were performed. Cells from YPD agar plates were inoculated and grew for 72 h in YPD liquid medium containing (EWA) or not containing (EWOA) ethanol at 10% (v/v). The cells were harvested after 72-h growth, washed with sterile water and transferred to the synthetic medium, yielding a final yeast concentration of 10⁵ cells mL⁻¹ (corresponding to an OD_640 nm between 0.025 and 0.070). The culture was divided into 10-mL aliquots and cultivated at 27 °C in hermetic and static tubes with no headspace volume. To avoid any influence of the sampling for the analytical techniques, each aliquot was discarded after the analysis. Experiments lasted up until 3 months. The dissolved oxygen concentration was measured in both the experiments EWOA and EWA using a Mettler–Toledo polarographic oxygen probe: the dissolved oxygen concentration was about 98% of air saturation at the beginning whereas it reduced to below 30% of air saturation (corresponding to a value <1.8 mg L⁻¹) after 3 months, simulating a semi-anaerobic environment as in wine maturation. To minimize oxygen diffusion during measurements, the determination was conducted by flushing certified pure nitrogen gas (SAPIO Srl, Italy) containing <5 μg mL⁻¹ of oxygen on the culture surface at a flow rate of 0.5 L min⁻¹.

Determination of biomass and culturability

Cell growth was evaluated by OD determination at 640 nm. Cellular culturability (Millet & Lonvaud-Funel, 2000) was carried out in duplicate on YPD agar plates maintained at 25 °C for 5 days. Results from preliminary tests (56 samples) were subjected to analysis of covariance by Statgraphics Plus V. 4. No significant differences were observed between two replicates (P<0.05) in both the determinations.

Analysis of extracellular metabolites

d-Fructose, ethanol, glycerol, acetic acid, l-malic acid, citric acid and d-/l-lactic acid concentrations were determined using R-Biopharm enzymatic kits (Roche, Mannheim, Germany: 10139106035, 10176290035, 10148270035, 10148261035, 10139068035, 10139076035, 11112821035, respectively). The metabolic determinations were analysed in duplicate and the data did not differ by more than 3%.

HPLC analysis

All the HPLC determinations were performed using a Waters 2695 Alliance HPLC module equipped with a Waters 2487 double-wavelength UV detector and a Novapak C18 (4 μm, 3.9 mm × 150 mm) column (Waters, Milford, MA).

Vinyl- and ethyl-phenols were evaluated by injecting 50 μL of culture media previously filtered through a 0.22-μm polyvinylidene difluoride membrane (Millipore, Bedford). Column temperature was set to 30 °C and UV detection was performed at 260 and 280 nm for vinyl- and ethyl-phenols, respectively. HPLC solvents were water/formic acid 0.2% (v/v) (eluent A) and acetonitrile/formic acid 0.2% (v/v) (eluent B). Elution was achieved at 1 mL min⁻¹ flow increasing eluent B from 10% to 35% in 20 min and from 35% to 75% in 6 min. The separation column was finally rinsed with 100% eluent B for 2 min. Run-to-run time was 40 min.

Biogenic amine analysis was performed according to the manual derivatization procedure described by Krause et al. (1995) with some modifications. Aliquots of 20 μL of the growth media previously filtered through 0.2-μm pore size membranes were transferred into a 2-mL vial and then added with 180 μL of a 0.15 mM NaHCO₃ pH 8.6 buffer solution. Two hundred microlitres of a dabsyl-chloride solution prepared by dissolving 40 mg into 10 mL acetone was added to the diluted growth media then the vial was stoppered and incubated at 70 °C for 60 min. Derivatization was stopped by cooling with cold water and the sample was diluted with 400 μL of acetonitrile/ethanol/HPLC elution buffer A solution (50/25/25, v/v/v). Samples were filtered through a 0.2-μm pore size filter before HPLC injection.

RP-HPLC separation was performed adopting the separation conditions described by Krause et al. (1995) for a 150 mm × 3.9 mm Novapack C₁₈ column, 4 μm (Waters). Dabsylated amines were detected using a UV–Vis spectrophotometer set at 436 nm. The method allowed for the determination of ethanolamine, phenylethylamine, isoamylamine, putrescine, hexylamine, cadaverine, histamine, heptylamine, tyramine, spermidine, octopamine and synephrine.

All reagents and standards used in HPLC analysis were of the highest available purity grade. Chromatograms were recorded and processed using millennium software v.4 (Waters). The SE of the method was calculated as 5%.

Analysis of D. bruxellensis growth and culturability on synthetic medium

Two experimental conditions called EWOA and EWA were independently tested. EWOA was performed using cells that had not been adapted to ethanol, while in the EWA experiments cells previously grown on YPD medium plus ethanol 10% (v/v) were used as an inoculum. These conditions were adopted in order to study the D. bruxellensis spoilage character: in detail, EWOA was performed to simulate a condition in which yeast develops at the stage of finished wine and EWA to simulate yeast adaptation during grape must fermentation. A partial condition of anaerobiosis was reached (dissolved oxygen concentration below 1.8 mg L⁻¹ of air saturation) at the end of the incubation.

A collection of 17 strains of D. bruxellensis was preliminary screened, under aerobic conditions on synthetic medium, for their ability to produce off-flavours. The strains were tested and clustered into three groups according to the amount of ethyl derivative produced from the corresponding hydroxyl-cynamic acids (Table 1). Five strains producing different amounts of volatile phenols were then selected for a further, more detailed analysis. A wine strain of S. cerevisiae (L288) was used for comparison. The physiology of growth was monitored by assaying culturability, biomass production, consumption of carbon sources (fructose, ethanol, glycerol, citrate, l-lactate and l-malate) and production of metabolites strictly correlated with growth, such as ethanol, acetate and glycerol.

Figure 1a reports the results obtained under EWOA conditions. Dekkera bruxellensis CBS4481 maintained unchanged its ability to form colonies until the end of the experiment (1.6 × 10⁶ CFU mL⁻¹ in 90 days). Furthermore, this strain showed the highest growth rate, reaching a final biomass level of 0.590 OD (corresponding to about four cell duplications) and an acetate production of 0.93 g L⁻¹. Fructose was depleted in 2 weeks as also in S. cerevisiae L288. Lactate (0.48 g L⁻¹) and malate (0.10 g L⁻¹) were used as well, the tricarboxylic acid cycle being active under condition of partial anaerobiosis. Dekkera bruxellensis CBS4601 and CBS4459 were characterized by a slower growth and slower fructose depletion rates. Dekkera bruxellensis CBS4601 was able to consume lactate (0.71 g L⁻¹) and malate (0.04 g L⁻¹) when fructose was completely consumed. Notably, this strain showed the highest acetate production (0.96 g L⁻¹), mostly produced when fructose was exhausted. Dekkera bruxellensis CBS4601 was cultivable throughout the whole experiment (reaching 2.1 × 10⁶ CFU mL⁻¹); the culturability of D. bruxellensis CBS4459 suddenly decreased when fructose was depleted.

Dekkera bruxellensis CBS2499 displayed a very limited fructose consumption during the first week and a significant ability to form colonies only for 20 days. Dekkera bruxellensis CBS2336 failed to grow under this condition.

The EWA (Fig. 1b) showed that the adaptation step, in some cases, modified the growth physiology on synthetic medium. Dekkera bruxellensis CBS4481 exhibited a metabolic behaviour similar to the one observed under EWOA conditions, the culturability being similar (1.7 × 10⁶ CFU mL⁻¹). It completely depleted fructose in 2 weeks and consumed l-lactic acid (0.72 g L⁻¹) and l-malic acid (0.08 g L⁻¹), producing 0.82 g L⁻¹ acetic acid.

Dekkera bruxellensis CBS4601 recovered a high culturability, leading to 1.6 × 10⁶ CFU mL⁻¹ and showed a delayed fructose consumption when compared with EWOA. Nevertheless, a very low consumption of l-lactic acid (0.10 g L⁻¹) and l-malic acid (0.04 g L⁻¹) was detected. Dekkera bruxellensis CBS4459 utilized the sugar during the first week of growth but within 15 days it stopped the consumption, being unable to deplete it in 3 months. In particular, this strain also showed a low culturability (3.9 × 10² CFU mL⁻¹).

Dekkera bruxellensis CBS2499 took advantage of the adaptation step: it grew (0.21 OD) and consumed fructose in 3 weeks, producing acetate (0.29 g L⁻¹) between 20 and 40 days, also maintaining good culturability for 60 days. Dekkera bruxellensis CBS2336 behaved in the same way as in EWOA, again failing to grow.

Summarizing, among the carbon sources available in the synthetic medium, fructose was first used by all the strains and the growth was mostly related to its consumption; l-lactic and l-malic acids were consumed only when fructose was exhausted. Citric acid, ethanol and glycerol were never used. All the tested strains able to grow produced about 0.2–0.3 g L⁻¹ of glycerol.

Production of volatile phenols

Volatile phenols (4-vinyl-phenol, 4-vinyl-guaiacol, 4-ethyl-phenol and 4-ethyl-guaiacol) were analysed using the RP-HPLC method.

Saccharomyces cerevisiae L288 did not produce ethyl-phenols but accumulated up to 0.56 mg L⁻¹ of 4-vinyl-phenol and 0.15 mg L⁻¹ of 4-vinyl-guaiacol after 50 days under EWOA conditions (Fig. 2a) and lower amounts of the same molecules in EWA.

Dekkera bruxellensis CBS2499 accumulated up to 0.9 mg L⁻¹ of 4-vinyl-phenol after adaptation to ethanol and no quantifiable amounts of volatile phenols without adaptation; likewise, in S. cerevisiae, no evidence of 4-ethyl-phenol production was observed. Dekkera bruxellensis CBS4481 accumulated up to 1.7 mg L⁻¹ of 4-ethyl-phenol and 1.3 mg L⁻¹ of 4-ethyl-guaiacol under EWOA and much higher concentrations of the same compounds under EWA conditions (2.7 and 2 mg L⁻¹, respectively). Amounts up to 0.5 mg L⁻¹ of 4-vinyl-phenol were found as well.

The most striking difference between the two conditions adopted for growth was observed with D. bruxellensis CBS4601: this strain gave only sluggish amounts of volatile phenols (<0.2 mg L⁻¹) without adaptation, while after adaptation it produced nearly 1.0 mg L⁻¹ of ethyl-guaiacol and significant amounts of 4-vinyl- and 4-ethyl-phenol.

Under EWOA conditions, strain CBS4459 produced low levels of volatile phenols (<0.25 mg L⁻¹) during the first month after the inoculation, but increased its activity between 30 and 60 days, releasing about 0.8 mg L⁻¹ of ethyl-phenols and 0.5 mg L⁻¹ of 4-vinyl-phenol. Under EWA conditions, this production was anticipated, since the start of its growth.

In our condition, D. bruxellensis CBS2336, failing to grow, did not produce any trace of volatile phenols.

Biogenic amines production

The five strains of D. bruxellensis and the reference strain of S. cerevisiae were able to produce detectable amounts of putrescine, cadaverine and spermidine (Fig. 3a and b).

In the EWOA trials, two strains of D. bruxellensis (namely CBS2336 and CBS4601) synthesized relatively small (below 0.40 mg L⁻¹ overall) amounts of the three polyamines and production was limited to the first 3 weeks of incubation. A relatively high content of the same polyamine was observed during the growth of D. bruxellensis CBS4601 under EWA conditions, which was accompanied by the presence of putrescine (0.4 mg L⁻¹) and cadaverine (0.15 mg L⁻¹). Spermidine was produced up to 1.2–1.3 mg L⁻¹ by D. bruxellensis CBS4481 within 60 days, before its reduction to 0.70–0.85 mg L⁻¹ at the end of the experiments; its production was not affected by the growth conditions.

In all the cases, a maximum accumulation of the amines was observed after 40–60 days before their partial disappearance. Polyamine concentration generally increased for 40–60 days and then spermidine concentration decreased by 20 to 50%.

Statistical analysis of physiological parameters

Correlation coefficients among different variables were calculated by the general linear model (Table 2) (Camussi et al., 1986). Only values higher than zero, at least one up two variables considered, were determined. The analysis for values of ethanol, glycerol, malic acid, 4-vinyl-phenol and 4-vinyl-guaiacol was not reported because they did not show any significant interaction with any of the variables tested. OD values were positively correlated with plate counts (viable cells), confirming that the yeast population was alive and cultivable throughout the time considered; moreover, OD values presented significant interactions with acetate production and fructose and l-lactate depletion. Also, plate counts showed a negative correlation with l-lactate values, indicating that the increase in biomass level was really associated with lactate utilization. These results are validated by the significant negative interactions observed between fructose and acetate or acetate and l-lactate being evidence for the natural aptitude of Dekkera to form acetate as an end product and, noteworthy, to consume lactate as a carbon source when fructose is exhausted, with an evident cellular proliferation. 4-Ethyl-phenol and 4-ethyl-guaiacol were positively correlated to each other and to OD and acetate, while they showed a significant negative correlation with fructose.

Values of volatile phenols were subjected to the anova in order to analyse the effects of two different factors: experimental condition (EWOA, EWA) and strain. In the case of 4-ethyl-phenol values, no significant differences were found between the EWOA and EWA, whereas they were present among the strains (P<0.01). On the other hand, the anova proved that both the factors (condition and strain) significantly affected the 4-ethyl-guaiacol production (P<0.01).