In vitro removal of ochratoxin A by two strains of Saccharomyces cerevisiae and their performances under fermentative and stressing conditions




The aim of this research was to study the effect of time, temperature, sugar content and addition of diammonium phosphate (DAP) on ochratoxin A (OTA) removal by two strains of Saccharomyces cerevisiae using a completely randomized design.

Methods and Results

The strains were grown in a medium containing OTA (2 μg l−1), two sugar levels (200 and 250 g l−1), with or without DAP (300 mg l−1), and incubated at 25–30°C. The yeasts were able to decrease the toxin amount by c. 70%, with the highest removing effect observed after 3 days at 30°C in the presence of 250 g l−1 of sugars and with DAP; after 10 days, the toxin was partially released into the medium. The strains produced high ethanol and glycerol contents, showed high tolerance to single/combined stress conditions and possessed β-d-glucosidase, pectinase and xylanase activities.


Ochratoxin A removal was affected by time, temperature, sugar and addition of DAP. Moreover, the phenomenon was reversible.

Significance and Impact of the Study

Ochratoxin A removal could be an interesting trait for the selection of promising strains; however, the strains removing efficiently the toxin could release it back; thus, the selection of the starter should take into account both the removal and the binding ability of OTA.


The key role of yeasts in conducting the alcoholic fermentation has long been recognized (Fleet 1999; Mortimer and Polsinelli 1999), and many factors have been found to influence their growth and metabolism and, as a consequence, the quality of wine (Charoenchai et al. 1998; Torija et al. 2003; Mendes-Ferreira et al. 2004). The temperature is a significant parameter for the course of wine fermentation (Pizarro et al. 2008), as it can strongly impact on yeast growth (Charoenchai et al. 1998), sugar utilization and ethanol production (Torija et al. 2003). Another significant variable is the concentration of fermentable sugars in grape musts (Charoenchai et al. 1998); concentrations from 200 to 300 g l−1 hinder yeast growth and decrease both the maximum population and ethanol yield (D'Amato et al. 2006). Furthermore, nitrogen can often be a limiting factor for yeast growth and sugar attenuation (Mendes-Ferreira et al. 2004). A value <150 mg l−1 of yeast assimilable nitrogen (YAN) in the must is associated with a higher probability of fermentation problems such as sluggish or stuck fermentations (Henschke and Jiranek 1993). In these cases, the addition of diammonium phosphate (DAP) or, to a lesser extent, organic yeast nutrients to nitrogen-deficient musts is a widely used practice (Barbosa et al. 2012).

Ochratoxin A (OTA) is a potent nephrotoxic and hepatocarcinogenic mycotoxin, recognized as a possible human carcinogen by the International Agency for Research on Cancer (Cecchini et al. 2006). The presence of the toxin in musts and wines is due to fungal contamination of the grapes by Aspergillus and Penicillium genera that may develop in both pre- and postharvest or during the phases before the winemaking (Esti et al. 2012).

Decontamination methods based on physical and chemical mechanisms were proposed in the past (Patharajan et al. 2011), but biological removal is considered the best solution, because it is possible to remove mycotoxins without using harmful chemicals and without losses in nutrient value or palatability of decontaminated food (Anly and Bayram 2009). In particular, some authors reported that yeasts were able to remove OTA during winemaking (Caridi et al. 2006; Cecchini et al. 2006; Angioni et al. 2007; Meca et al. 2010; Piotrowska et al. 2013).

Although many papers focused on the role of temperature, sugar and nitrogen on the course of fermentation, few data are available on the effects of DAP, as well as on its combination with temperature and sugar, on the removal of OTA by yeasts (Petruzzi et al. 2014).

Therefore, the main aim of this research was to study the OTA-removing ability of two strains of Saccharomyces cerevisiae and its correlation with temperature, sugar and addition of DAP; the planning of the research, as well as data modelling, was based on a completely randomized design with a factorial arrangement of treatments. As an additional goal, the oenological aptitude of yeasts (fermentative performances and technological and qualitative traits) was assessed to explore the possibility of a selection programme focusing on yeast cultures acting as biological tools to remove the toxin.

Materials and methods


Two wild strains (W28 and W46) of S. cerevisiae were used throughout this study. The strains were previously isolated from a typical grapevine cultivar (Uva di Troia) of the Apulian region (southern Italy) and identified by polymerase chain reaction–restriction fragment length polymorphism (PCR-RFLP) of internal transcribed spacers (ITS), as described in Esteve-Zarzoso et al. (1999). To confirm the identification of the strains, 5.8S-ITS PCR fragments, generated by amplification with ITS1/ITS4 primers (White et al. 1990), were sequenced by Primm Biotech (Milan, Italy). The sequences obtained were compared with those available in the EMBL database using the Basic Local Alignment Search Tool version 2.2.27 ( The sequences are freely available with the following accession numbers: KC588952 (S. cerevisiae W28) and KC588953 (S. cerevisiae W46).

Removal assay and ELISA determination of OTA

Small-scale fermentation experiments were carried out in duplicate using the method proposed by Lopes et al. (2007), modified as follows. A semi-synthetic medium with the following composition was used: 100 g l−1 glucose (C. Erba, Milan, Italy), 100 g l−1 fructose (Sigma-Aldrich, Milan, Italy), 10 g l−1 yeast extract (Oxoid, Milan, Italy), 1 g l−1 ammonium sulfate (J.T. Baker, Milan, Italy), 1 g l−1 potassium phosphate (C. Erba) and 1 g l−1 magnesium sulfate (J.T. Baker). A second test was run using 250 g l−1 of sugars (glucose/fructose, 1 : 1). Each lot with the same sugar concentration was divided into two aliquots: the control and a sample supplemented with 300 mg l−1 of DAP (Sigma-Aldrich). When DAP was added to the medium, the YAN (yeast assimilable nitrogen) content increased from 250 to 310 mg l−1. YAN was evaluated through K-LARGE enzymatic kit (Megazyme, Bray, Ireland) according to the manufacturer's instructions. By combining sugar amount and DAP addition, four different samples were analysed: (i) sugar 200 g l−1; (ii) sugar 200 g l−1 + DAP; (iii) sugar 250 g l−1 and (iv) sugar 250 g l−1 + DAP.

The fermentation was carried out in 150-ml flasks containing 100 ml of medium. After the sterilization, the pH of the medium was decreased to 3·5 through a solution of citric acid (10 g l−1) (Sigma-Aldrich); then, the medium was inoculated with yeasts to 6 log CFU ml−1 and added with OTA (2 μg l−1) (Sigma-Aldrich), and the surface was covered with a thin layer of sterilized paraffin oil (10 ml per flask) (C. Erba) to avoid air contact. A control sample was prepared with media containing OTA, but without yeast. All the samples were incubated at 25 and 30°C without shaking and collected immediately after yeast inoculation and after 3 and 10 days of fermentation. Before sampling, the flasks were shaken to thoroughly mix the contents and obtain all cells in suspension. The quantitative analysis of OTA was assessed on the fermented media after centrifugation at 2100 g for 15 min at 4°C to remove yeast cells. The Ridascreen® Ochratoxin A 30/15 ELISA kit (R-Biopharm; Darmstadt, Germany; Art. No. R1311) was used according to the manufacturer's instructions. The decrease in OTA in the medium in relation to the initial concentration was expressed as removal percentage.

Fermentation process and yeast growth

The fermentation process was monitored daily by weight loss as a result of CO2 escaping from the system, until the weight was constant. Aliquots of samples were collected immediately after yeast inoculation (0) and after 3 and 10 days of fermentation to determine fermentation products and the number of cells. Analytical determinations (residual sugars, ethanol and glycerol) were performed by Fourier transform infrared spectroscopy (FTIR) by employing the WineScan FT120 instrument (FOSS Analytical, Hillerod, Denmark; software version 2.2.1); for these analyses, the samples were centrifuged at 2100 g for 15 min at 4°C to remove yeast cells and then analysed following the supplier's instructions. For the determination of the number of cells, a standard plate method on YPD agar (yeast extract 10 g l−1; bacteriological peptone 20 g l−1; glucose 20 g l−1; 12 g l−1 agar technical no. 3) was carried out. The plates were incubated at 30°C for 2–5 days.

Technological and qualitative traits of the yeasts

Analysis of yeast tolerance to several stress conditions

The yeasts were studied in relation to their ability to grow in presence of ethanol (6–24%), sulfur dioxide (150–1000 mg l−1), glucose (200, 250 and 300 g l−1), fructose (200, 250 and 300 g l−1), sucrose (200, 250 and 300 g l−1), copper (100–1000 mg l−1), acetic acid (3·5 g l−1) and cycloheximide (0·01 and 0·05%), at two different temperatures (37 and 42°C); moreover, some combined stress was evaluated, that is, 10% ethanol +150 mg l−1 sulfur dioxide, 250 g l−1 sucrose + incubation at 37°C, 10% ethanol + incubation at 37°C. YPD agar, buffered to pH 3·5 and supplemented with the adequate amounts of ethanol, sulfur dioxide, glucose, fructose, sucrose, copper, acetic acid and cycloheximide, was inoculated with the strains and incubated at 30°C or at 37–42°C (temperature stress) for 1–5 days. For each test, a positive control was used, that is, YPD agar buffered to pH 3·5, inoculated with yeasts and incubated at 30°C. The experiments were performed in duplicate over two different batches.

Enzymatic activities

The following enzymatic activities were investigated: β-d-glucosidase (Pérez et al. 2011); pectinase and xylanase (Bevilacqua et al. 2009); protease (Strauss et al. 2001); ester hydrolase and glycosidase (Comitini et al. 2011); and β-d-xylosidase (Fiore et al. 2005). β-d-glucosidase, pectinase and xylanase were also studied at wine pH and wine ethanol concentration. Three different combinations were tested: (i) 10% ethanol; (ii) pH 3·5; (iii) pH 3·5 and 10% ethanol. The tests were carried out in duplicate; noninoculated plates were used as negative controls.

Other characteristics

The ability to produce hydrogen sulfide was tested using a qualitative method based on colony appearance on BIGGY agar (Orlić et al. 2010). Formation of biogenic amines was determined using the method of Nikolaou et al. (2006); histidine, tyrosine, phenylalanine, tryptophan, lysine, leucine, ornithine and arginine were used as precursor amino acids. The assay was performed under aerobic and anaerobic conditions using an anaerobic jar system form AnaeroGen (Oxoid). Interaction with phenolic compounds was determined using the methodology of Caridi et al. (2002). For all the assays, a negative control was carried out with uninoculated plates. The tests were performed in duplicate.

Statistical analysis

The results of the quantitative assays (OTA reduction, ethanol, glycerol, residual sugars and cell count) were analysed through one-way analysis of variance and Tukey's test as the post hoc (P < 0·05) comparison test; the mean values were put in a cross-table. In each column, there are the data of the combinations sugar/time of sampling for the two strains (3 days – 200 g l−1 of sugar for the strain W28; 3 days – 200 g l−1 of sugar for the strain W46, etc.), whereas in each row, there are the results referred to the combination DAP and temperature for each strain individually (control-25°C; DAP-25°C, etc.). The statistical analysis was applied to each row and each column separately to highlight which sample was different.

Thereafter, the mean values of OTA removal were used as the input data for a DoE analysis through the software Statistica for Windows (StatSoft, Inc., Tulsa, OK, USA; software version 10.0.1011.0). Temperature (25 and 30°C), sugar (200 and 250 g l−1) and DAP (0 and 300 mg l−1) were used as independent variables or factors; each of them was set at two different levels (‘−1’ and ‘+1’, respectively, the minimum and the maximum levels of each variable). The analysis was performed through the option DoE/2(kp) standard designs/2-way interactions; OTA reductions after 3 and 10 days were used as dependent variables.

The software uses a multiple regression approach/intercept option to highlight the significant effect of each factor individually, as well as of their two-way interaction (i.e. the interaction between two factors); the output is a table reporting for each factor or interaction the mathematical coefficient (i.e. the strength), the standard error of the coefficient, the 95% confidence interval and the significance (t-test, evaluated as the ratio of the mathematical coefficient vs its standard error). The significance is used to build a Pareto chart, whilst the mathematical coefficients were used to predict the value of the dependent variable for many combinations of the design and draw the contour plots.


OTA removal

The removal of OTA was evaluated after 3 and 10 days of fermentation for two wild strains of S. cerevisiae. The experiment was designed to be carried out at 25–30°C to fit better with the real conditions encountered in Apulian region; the fermentation of grape must is traditionally performed at the end of summer, during the warm days of September in the Mediterranean countries at average temperatures between 25–30°C. These temperatures are also used by some producers to enhance the extraction of anthocyanin pigments (Belloch et al. 2008). Moreover, the experiments were performed in media taking into account the typical composition of grape musts containing equal amounts of fructose and glucose in a range between 160 and 300 g l−1 (Tronchoni et al. 2009) and in YAN content ranging from 50 to 500 mg l−1 (Bely et al. 2003). Accordingly, eight combinations were analysed by combining temperature (25 and 30°C), different levels of sugar (200 and 250 g l−1) and DAP addition.

Ochratoxin A reduction after 3 days was used as the input to build a simple polynomial model, able to describe the effects of sugar, nitrogen and temperature; the approach was significant for both the strains, with an adjusted regression coefficients ranging from 0·813 to 0·858. The Pareto chart highlights that the addition of DAP and the content of sugar influenced significantly the removal of OTA mediated by both the strains, whilst the effect of the temperature was strain dependent, as it was found only for the strain W28, but not for the isolate W46 (Fig. 1a,b).

Figure 1.

Pareto chart for the effects of temperature (1), supplementation with diammonium phosphate (DAP) (2) and sugar (3) on the removal of ochratoxin A (OTA) by Saccharomyces cerevisiae W28 (a) and W46 (b) after 3 days. 1 by 2, interactive effect of temperature/DAP; 1 by 3, temperature/sugar; 2 by 3, DAP/sugar.

Figure 2 shows the contour plots for the strain W28; OTA reduction was maximum when temperature, sugar and DAP were set at the highest levels. For example, OTA was reduced by c. 70% at 30°C with either 300 mg l−1 of DAP or 250 g l−1 of sugar.

Figure 2.

Contour plots for the interactions [diammonium phosphate (DAP)] x [temperature], [sugar] x [temperature] and [sugar] x [DAP] on the removal of ochratoxin A (OTA) by Saccharomyces cerevisiae W28 after 3 days. (a) (■) >68; (image_n/jam12350-gra-0001.png) <68; (image_n/jam12350-gra-0002.png) <66; (image_n/jam12350-gra-0003.png) <64; (image_n/jam12350-gra-0004.png) <62; (image_n/jam12350-gra-0005.png) <60 and (□) <58. (b) (■) >68; (image_n/jam12350-gra-0001.png) <67; (image_n/jam12350-gra-0002.png) <65; (image_n/jam12350-gra-0003.png) <63; (image_n/jam12350-gra-0002.png) <61; (image_n/jam12350-gra-0005.png) <59 and (□) <57. (c) (■) >70; (image_n/jam12350-gra-0001.png) <69; (image_n/jam12350-gra-0002.png) <67; (image_n/jam12350-gra-0003.png) <65; (image_n/jam12350-gra-0004.png) <63; (image_n/jam12350-gra-0006.png) <61; (image_n/jam12350-gra-0005.png) <59 and (□) <57.

Similar results were found for the strain W46, as OTA was reduced by c. 70% with 250 g l−1 of sugar and 300 mg l−1 of DAP (Fig. 3).

Figure 3.

Contour plot of the interaction of [sugar] x [diammonium phosphate (DAP)] on the removal of ochratoxin A (OTA) by Saccharomyces cerevisiae W46 after 3 days. (■) >65; (image_n/jam12350-gra-0001.png) <63; (image_n/jam12350-gra-0002.png) <58; (image_n/jam12350-gra-0004.png) <53; (image_n/jam12350-gra-0006.png) <48 and (image_n/jam12350-gra-0007.png)43.

The yeasts released OTA throughout the fermentation, as shown in Fig. 4; however, this process was random, and the statistical analysis showed that the factors of the design (sugar, DAP supplementation and temperature) did not play any significant role (data not shown).

Figure 4.

Ochratoxin A (OTA) removal by Saccharomyces cerevisiae W28 and W46 after 3 and 10 days. Mean values ± standard deviation. S, sugar (200 or 250 g l−1); DAP, diammonium phosphate. (image_n/jam12350-gra-0008.png) 3 day and (image_n/jam12350-gra-0009.png) 10 day.

Fermentation performances of yeasts

Table 1 shows the fermentation products of the strains under different conditions. As expected, ethanol yield was higher at 30°C than at 25°C, whilst the increase in sugar (250 vs 200 g l−1) exerted a negative effect on this trait. Glycerol increased within the running time, without significant differences amongst the different combinations, whilst DAP addition enhanced the consumption of sugar.

Table 1. Fermentation products (mean values ± standard deviation) of Saccharomyces cerevisiae W28 and W46 as a function of time, temperature, sugar concentration and supplementation with diammonium phosphate (DAP) (300 mg l−1)
 Time(d)Sugar (g l−1)W28W46
  1. a

    The letters indicate the significant differences for each compound (one-way anova and Tukey's test); the small letters (a,b,c) indicate the difference in a column, whilst the capital ones highlight the differences in a row.

  2. b

    Below the detection limit.

Ethanol (g l−1)32003·91 ± 0·09aAa4·22 ± 0·15aA4·20 ± 0·03aA4·72 ± 0·04aB4·56 ± 0·13aA4·65 ± 0·14aA5·46 ± 0·16aC5·47 ± 0·01aC
2502·99 ± 0·16bA3·00 ± 0·04bA3·99 ± 0·02aA4·06 ± 0·06bA4·19 ± 0·05aB4·39 ± 0·02aB4·96 ± 0·13bB4·93 ± 0·21bB
102008·49 ± 0·09cA8·67 ± 0·01cA8·92 ± 0·04bA9·04 ± 0·09cB9·28 ± 0·07bB9·41 ± 0·08bB9·7 ± 0·09cC9·80 ± 0·08cC
2507·93 ± 0·06dA8·05 ± 0·05dA8·52 ± 0·09cA8·51 ± 0·02cA9·04 ± 0·02bB9·22 ± 0·07bB9·57 ± 0·02cB9·74 ± 0·04cB
Residual sugars (g l−1)320089·20 ± 2·54aA72·65 ± 1·06aB80·15 ± 4·31aA64·41 ± 0·86aB70·32 ± 2·36aB56·50 ± 2·12aC64·00 ± 2·82aB49·00 ± 1·41aC
250135·05 ± 5·16bA120·05 ± 1·62bB120·05 ± 1·62bA108·05 ± 4·45bB88·80 ± 1·69bC75·95 ± 0·28bD81·10 ± 3·88bC62·85 ± 1·06bD
Glycerol (g l−1)32002·49 ± 0·09aA2·63 ± 0·12aA2·87 ± 0·15aA2·92 ± 0·15aA3·45 ± 0·02aB3·53 ± 0·12aB3·90 ± 0·07aB3·92 ± 0·07aB
2502·81 ± 0·06aA2·80 ± 0·15aA3·30 ± 0·04aA3·19 ± 0·02aA3·71 ± 0·09aB3·79 ± 0·10aB4·18 ± 0·13aB4·18 ± 0·04aB
102005·69 ± 0·03bA5·80 ± 0·12bA6·04 ± 0·09bA6·12 ± 0·14bA7·01 ± 0·17bB7·08 ± 0·04bB7·37 ± 0·05bB7·44 ± 0·09bB
2506·08 ± 0·07bA6·22 ± 0·15bA6·58 ± 0·03cA6·61 ± 0·09cA7·35 ± 0·16bB7·39 ± 0·12bB7·63 ± 0·07bB7·75 ± 0·17bB

Technological and qualitative traits of yeasts

The technological and qualitative traits of yeasts were reported in Table 2. Concerning the single fermentation stressors, the strain W28 showed a better tolerance to ethanol than the isolate W46 (20 vs to 18%) and a worse resistance to SO2 (300 vs 700 mg l−1). The trends in the presence of sugars (growth at 300 g l−1), growth at high temperatures (37 and 42°C) and resistance to cycloheximide (0·01%), copper (600 mg l−1) and acetic acid (3·5 g l−1) were similar; moreover, both the strains were able to grow under combined stress conditions.

Table 2. Technological and qualitative traits of Saccharomyces cerevisiae W28 and W46
Resistance to single and combined oenological stressW28W46
  1. a

    Minimal inhibitory concentration (MIC).

  2. b

    Diameter of halo: weak activity (between 14–17 mm); medium activity (18–22 mm), strong activity (≥23 mm);

  3. c

    Diameter of the growth zone: ‘–’ (no growth); ‘+’ (between 2–5·5 mm); ‘++’ (>5·5 mm);

  4. d

    Diameter of the halo: ‘–’ (no halo); ‘+’ (between 1–3 mm); ‘++’ (>3 mm);

  5. e

    Colour of biomass: ‘0’ (white; no hydrogen sulfide production); ‘1’ (cream; low production); ‘2’ (light-brown; medium production); ‘3’ (dark-brown; high production);

  6. f

    The test was negative both under aerobic and anaerobic conditions for all the amino acids (arginine, histidine, leucine, lysine, ornithine, phenylalanine, tyrosine, tryptophan);

  7. g

    Colour of biomass: ‘0’ (white/pale grey; low adsorption of phenolic compounds); ‘1’ (pale; medium adsorption); ‘2’ (dark hazel; high adsorption).

Single stress
Glucose (g l−1)300300
Fructose (g l−1)300300
Sucrose (g l−1)300300
Growth at 37°C++
Growth at 42°C++
Ethanol (%)a2018
Acetic acid++
Sulphur dioxide (mg l−1)a300700
Copper (mg l−1)a600600
Cycloheximide (%)0·010·01
Combined stress
250 g l−1 sucrose × 37°C++
10% ethanol × 150 mg l−1 SO2++
10% ethanol × 37°C++
Enzymatic activities
pH 6·0MediumMedium
pH 6·0 + 10% of ethanolMediumMedium
pH 3·5
pH 3·5 + 10% of ethanol
pH 4·0++
pH 4·0 + 10% of ethanol++
pH 3·5++
pH 3·5 + 10% of ethanol++
pH 6·0+++
pH 6·0 + 10% of ethanol++
pH 3·5++
pH 3·5 + 10% of ethanol++
Ester hydrolase
Production of undesirable compounds
Hydrogen sulfidee22
Amino acid decarboxylationf
Interaction with phenolic compoundsg
Anthocyanin adsorption11
Tannin adsorption00

Concerning the enzymatic traits, both the strains were positive to β-d-glucosidase, pectonase and xylanase; the latter two activities were retained by yeasts under winemaking conditions, whilst β-d-glucosidase did not. In addition, both the strains produced medium amounts of hydrogen sulfide, had a low-to-medium parietal interaction with phenolic compounds and were not able to decarboxylate amino acids.


The level of OTA in wine is markedly lower than that found in the grapes, due to toxin removal throughout the vinification, and especially during the alcoholic fermentation (Paster 2008), as a result of yeast activity (Caridi et al. 2006; Cecchini et al. 2006; Angioni et al. 2007; Meca et al. 2010; Piotrowska et al. 2013) and/or the interaction between OTA and grape constituents (Paster 2008). The adsorption to cell wall is suggested as a mechanism for OTA removal by yeasts (Bejaoui et al. 2004), because harbouring polysaccharides (glucan, mannan), proteins and lipids exhibits numerous different and easy accessible adsorption sites (Cecchini et al. 2006). Recently, Petruzzi et al. (2014) studied the effect of the fermentative conditions on the removal of OTA mediated by two commercial yeasts as well as by a potential starter strain (S. cerevisiae W13). This research focused on the influence of the most important environmental parameter (i.e. temperature) (Pizarro et al. 2008), as well as on the effect of two levels of sugar (200 and 250 g l−1), and the addition of one rate of DAP (300 mg l−1) on OTA-removing ability of two wild S. cerevisiae strains (W28 and W46) using a completely randomized design with factorial arrangements of the treatments to find a possible general trend on the toxin removal. With this scope, a semi-synthetic medium was used to avoid the interference of must components towards OTA.

After 3 days, the strains W28 and W46 were able to remove OTA by c. 70%; this result confirmed the ability of S. cerevisiae to reduce OTA content in different matrices, including synthetic and natural grape juice (Bejaoui et al. 2004), white and red must (Caridi et al. 2006; Cecchini et al. 2006) and model wine (Nunez et al. 2008). Cecchini et al. (2006) found that yeasts removed OTA by c. 50–70% in a red grape must after 36 days of fermentation. Caridi et al. (2006) obtained similar results, as they observed that yeasts reduced OTA by c. 70–80% after 90 days of fermentation. However, both these papers were in vivo studies, whilst our research was performed in vitro, and OTA removal in wine could be attributed to a synergistic effect between yeasts and phenolic compounds, as suggested by the cited authors. On the other hand, Bejaoui et al. (2004) and Nunez et al. (2008) reported higher removal indices (c. 90%), but they used heat-treated cells. Heating may cause changes in the surface properties of cells, for example denaturation of proteins or formation of Maillard reaction products. These products could harbour higher adsorption sites than viable cells with an enhancement of OTA removal (Piotrowska et al. 2013).

The ability of the wild strains W28 and W46 to remove OTA relied upon some important parameters, namely temperature, sugar and addition of DAP, with a mathematical positive effect. Concerning the temperature, our results were in agreement with other findings. In particular, a temperature-dependent mechanism was observed for the aflatoxin B1 (AFB1) removal by Lactobacillus rhamnosus GG and LC-705 (El-Nezami et al. 1998) and for the OTA removal by Metschnikowia pulcherrima MACH1, Pichia guilliermondii M8 and Rhodococcus erythropolis AR14 (Patharajan et al. 2011). Some preliminary experiments performed showed that sugar content did not affect the removal ability of some wild strains (L. Petruzzi, unpublished results), whilst its effect was found positive for a potential starter strain of S. cerevisiae (Petruzzi et al. 2014), thus suggesting a kind of strain-specific variability. Concerning DAP supplementation, the effect of this factor was found to be not significant in S. cerevisiae W13 (Petruzzi et al. 2014), whilst it affected significantly the removal mediated by the wild strains W28 and W46 as a single term, but not in interaction with sugar and/or temperature. This different result could be due to a strong strain variability in the surface properties affecting the binding of OTA to cell wall and probably connected with the oenological performances of the strains. A future trend in this sense could be the use of a robust design (e.g. central composite design) to study the effects of different levels of DAP on both OTA removal and cell surface properties and build a general model able to predict yeast trend under different conditions.

Moreover, this research showed that OTA could be released back into the medium; this kind of phenomenon was found previously under laboratory conditions (Petruzzi et al. 2012) and for the promising functional strain S. cerevisiae W13 (Petruzzi et al. 2014); the extent of OTA release was highly variable, and in some combinations, the phenomenon was completely reversible, with a lower efficiency than that found previously (Petruzzi et al. 2012, 2014).

The release of toxins after their removal was recovered for Lactobacillus amylovorus and Lactrhamnosus towards AFB1, due to the fact that the bonds between lactobacilli/AFB1 complexes involve weak noncovalent interactions (Peltonen et al. 2001). A similar phenomenon could be associated with the release of OTA by S. cerevisiae because hydrogen bonding, ionic or hydrophobic interaction seems to be related to OTA adsorption mechanism (Cecchini et al. 2006).

Yeasts were also studied in relation to their fermentative performances under the different nutritional and environmental conditions and for their technological and qualitative traits. In particular, the strain W46 showed a higher ethanol and glycerol production and a major ability to achieve dryness than the strain W28. As expected, a high sugar concentration has a detrimental effect on ethanol production (Bely et al. 2008), whereas ethanol yield increased as the temperatures increased; this result was supported by some related findings (Fakruddin et al. 2012). For both the strains, sugars were fully consumed after 10 days of fermentation, whilst the addition of DAP determined a faster consumption of sugars, as reported by Mendes-Ferreira et al. (2004). With regard to glycerol, only the strain W28 confirmed the hypothesis of how yeast responds to increased external sugar concentration by enhanced production and intracellular accumulation of glycerol to counterbalance the osmotic pressure (Bely et al. 2008).

Finally, both the strains showed a high resistance to single and combined stressors typically encountered during alcoholic fermentation. This result was relevant because the ability to carry out vinification is largely influenced by the response of yeast cells towards stress conditions (Carrasco et al. 2001). Moreover, yeasts produced medium amounts of hydrogen sulfide, responsible for the sensorial descriptor of rotten eggs (Tristezza et al. 2012), showed a medium ability to adsorb anthocyanins, thus preserving wine colour (Caridi et al. 2002) and β-d-glucosidase, pectinase and xylanase activities, described as three of the most important extracellular enzymes of S. cerevisiae (Strauss et al. 2001); finally, they did not produce biogenic amines under in vitro conditions, but involved in toxic effects on the human organism and sensory defects in wine (Koutsoumanis et al. 2010).

In conclusion, in this article, two wild strains of S. cerevisiae (W28 and W46) were studied for their OTA-removing ability under different fermentative conditions. Both the strains were able to decrease the toxin amount by up to 70%, with the highest removing effect observed after 3 days at 30°C in the presence of 250 g l−1 sugar and with DAP supplementation (300 mg l−1). However, OTA was released back after 10 days, thus suggesting that the selection of suitable OTA-removing strains should take into account both the extent of the removed toxin and the amount of OTA released back into the medium.

Conflict of interest

No conflict of interest declared.