Molecular and physiological comparison of spoilage wine yeasts

Authors

  • M.P. Sangorrín,

    1. Grupo de Biodiversidad y Biotecnología de Levaduras, Instituto Multidisciplinario de Investigación y Desarrollo de la Patagonia Norte (IDEPA CONICET-UNCo), Dpto. de Química, Facultad de Ingeniería, Universidad Nacional del Comahue, Buenos Aires, Argentina
    Search for more papers by this author
  • V. García,

    1. Laboratorio de Biotecnología y Microbiología Aplicada, Departamento en Ciencia y Tecnología de los Alimentos, Universidad de Santiago de Chile (USACH), Santiago, Chile
    Search for more papers by this author
  • C.A. Lopes,

    1. Grupo de Biodiversidad y Biotecnología de Levaduras, Instituto Multidisciplinario de Investigación y Desarrollo de la Patagonia Norte (IDEPA CONICET-UNCo), Dpto. de Química, Facultad de Ingeniería, Universidad Nacional del Comahue, Buenos Aires, Argentina
    2. Facultad de Ciencias Agrarias, Universidad Nacional del Comahue, Río Negro, Argentina
    Search for more papers by this author
  • J.S. Sáez,

    1. Grupo de Biodiversidad y Biotecnología de Levaduras, Instituto Multidisciplinario de Investigación y Desarrollo de la Patagonia Norte (IDEPA CONICET-UNCo), Dpto. de Química, Facultad de Ingeniería, Universidad Nacional del Comahue, Buenos Aires, Argentina
    2. Facultad de Ciencias Agrarias, Universidad Nacional del Comahue, Río Negro, Argentina
    Search for more papers by this author
  • C. Martínez,

    1. Laboratorio de Biotecnología y Microbiología Aplicada, Departamento en Ciencia y Tecnología de los Alimentos, Universidad de Santiago de Chile (USACH), Santiago, Chile
    2. Centro de Estudios en Ciencia y Tecnología de los Alimentos (CECTA), Universidad de Santiago de Chile (USACH), Santiago, Chile
    Search for more papers by this author
  • M.A. Ganga

    Corresponding author
    1. Laboratorio de Biotecnología y Microbiología Aplicada, Departamento en Ciencia y Tecnología de los Alimentos, Universidad de Santiago de Chile (USACH), Santiago, Chile
    • Grupo de Biodiversidad y Biotecnología de Levaduras, Instituto Multidisciplinario de Investigación y Desarrollo de la Patagonia Norte (IDEPA CONICET-UNCo), Dpto. de Química, Facultad de Ingeniería, Universidad Nacional del Comahue, Buenos Aires, Argentina
    Search for more papers by this author

Correspondence

M. Angelica Ganga, Laboratorio de Biotecnología y Microbiología Aplicada, Departamento en Ciencia y Tecnología de los Alimentos, Universidad de Santiago de Chile (USACH), Obispo Manuel Umaña 050, Estación Central, Santiago, Chile. E-mail: angelica.ganga@usach.cl

Abstract

Aims

Dekkera bruxellensis and Pichia guilliermondii are contaminating yeasts in wine due to the production of phenolic aromas. Although the degradation pathway of cinnamic acids, precursors of these phenolic compounds has been described in D. bruxellensis, no such pathway has been described in P. guilliermondii.

Methods and Results

A molecular and physiological characterization of 14 D. bruxellensis and 15 P. guilliermondii phenol-producing strains was carried out. Both p-coumarate decarboxylase (CD) and vinyl reductase (VR) activities, responsible for the production of volatile phenols, were quantified and the production of 4-vinylphenol and 4-ethylphenol were measured. All D. bruxellensis and some P. guilliermondii strains showed the two enzymatic activities, whilst 11 of the 15 strains of this latter species showed only CD activity and did not produce 4-EP in the assay conditions. Furthermore, PCR products obtained with degenerated primers showed a low homology with the sequence of the gene for a phenyl acrylic acid decarboxylase activity described in Saccharomyces cerevisiae.

Conclusions

D. bruxellensis and P. guilliermondii may share a similar metabolic pathway for the degradation of cinnamic acids.

Significance and Impact of the Study

This is the first work that analyses the CD and VR activities in P. guilliermondii, and the results suggest that within this species, there are differences in the metabolization of cinnamic acids.

Introduction

The presence of phenolic aromas is associated to microbiological contamination by Dekkera bruxellensis or its anamorph Brettanomyces bruxellensis (Chatonnet et al. 1992; Curtin et al. 2007; Suárez et al. 2007). This yeast produces aromas described as ‘phenolic’, ‘animal’, ‘horse sweat’ and ‘stable’ which are collectively known as ‘Brett character’ (Chatonnet et al. 1992; Suárez et al. 2007). Pizarro et al. (2007) chemically characterized this defect describing it as being composed of a series of volatile phenols, including 4-ethylguaiacol (4-EG), 4-ethylphenol (4-EP), 4-vinylguaiacol (4-VG) and 4-vinylphenol (4-VP). A concentration >425 μg l−1 of a mixture of 1 : 10 of 4-EG/4-EP produces an animal aroma, whilst a concentration >770 μg l−1 of a 1 : 1 mixture of 4-VG/4-VP produces a strong pharmaceutical aroma (Chatonnet et al. 1995).

Although at a global scale, Dbruxellensis would be the main producer of these volatile phenols in wine, other species have also been described (Dias et al. 2003a; Martorell et al. 2006; Lopes et al. 2009a). Dias et al. (2003a) showed that Candida winckerhamii, Candida cantarelli, Lactococcus lactis, Debaryomyces hansenii and Pichia guilliermondii are also able to produce volatile phenols with a similar efficiency to D. bruxellensis. This observation was confirmed by Barata et al. (2006) who showed the capacity of P. guilliermondii to produce volatile phenols in wine must. Similar results were obtained by Lopes et al. (2009a) who detected the production of phenolic compounds, particularly 4-VP in must inoculated with different P. guilliermondii isolates. Likewise, Martorell et al. (2006) showed that different isolates of this species produce different concentrations of phenolic compounds in synthetic culture media. Additionally, mixed cultures of the spoilage yeasts D. bruxellensis and P. guilliermondii produced an increase in the total volatile phenols concentration regarding as compared with pure cultures of the same species (Sáez et al. 2010).

Chatonnet et al. (1992) proposed a biosynthetic pathway for the synthesis of volatile phenols from cinnamic acids in D. bruxellensis. This yeast transforms these organic acids, naturally present in the must, into vinyl and ethyl derivatives through a cinnamate decarboxylase (CD) and a vinylphenol reductase (VR). These enzymes have been purified and characterized in D. bruxellensis (Godoy et al. 2008; Tchobanov et al. 2008). Until now, a similar biosynthetic pathway has not been described for P. guilliermondii.

The fact that different strains belonging to a same species can produce significantly different amounts of volatile phenols or other metabolic products makes the use of molecular techniques necessary to allow the intraspecific differentiation of the yeasts. This characterization could be also very helpful in establishing the origin of wine spoilage yeasts, their routes of contamination and the association with different metabolic capacities (Suárez et al. 2007; Lopes et al. 2009a,b). Different techniques like RAPD-PCR, RFLP's of mtDNA restriction analysis, AFLP and fluorescence in situ hybridization were applied for strain discrimination in this species (Mitrakul et al. 1999; Dias et al. 2003a; Cocolin et al. 2004; Agnolucci et al. 2009; Godoy et al. 2009). On the other hand, the characterization at intraspecific level in P. guilliermondii has been less studied; mtDNA restriction analysis, RAPD-PCR and physiological (killer biotype) characterization methods have been already evaluated (Martorell et al. 2006; Lopes et al. 2009a).

In this work, both the production of volatile phenols and the associated CD and VR activities (described in D. bruxellensis) were evaluated in vitroin D. bruxellensis and P. guilliermondii strains collected in different wine-producing areas. Molecular relationships among strains based on RAPD patterns, and its association with the capacity to produce volatile phenols was also analysed.

Materials and methods

Yeast strains and culture conditions

Yeast strains used in this study included 14 isolates of D. bruxellensis and 15 isolates of Pichia guilliermondii obtained from the collection of the Biotechnology and Applied Microbiology Laboratory of the Universidad de Santiago de Chile (LAMAP) and North Patagonia Culture Collection (NPCC), Neuquén, Argentina (Table 1).

Table 1. Geographical and molecular pattern of Dekkera bruxellensis and Pichia guilliermondii
StrainDesignationSourcePattern types obtained by RAPD-PCR composite patterns by OPA primerHaplotype (RAPDs)
E09E12E15R08R14S05S11
  1. In code: D, D. bruxellensis; P, P. guilliermondii; Designation: Only number, Martorell et al. (2006); CECT, Collection Spain Culture Type; T, Type strain; L, Laboratory of Biotechnology and Applied Microbiology of the Universidad de Santiago de Chile (LAMAP); NPCC, Noth Patagonia Collection Culture; In source: G, grape; Dr, Drosophila; M, fresh must; FM, fermenting must; RW, red wine; A, Argentina; C, Chile; P, Portugal. In number subindice vineyard or winery source. OPA primers, E09, E12, E15, R08, R14, S05, S11.

D1CECT 1451TCollectionCDDFIFK1
D2L2480RW1 CAAAABAB2
D3L2652RW2 CJKDHDBA3
D4L2676RW3 CEEFDHEC4
D5L2686RW4 CCDECEGH5
D6400RW6 PBCEBFHG6
D7419RW6 PIHGGMDI7
D81327RW7 PHJCIACO8
D91600RW7 PDFFDHGD9
D101700RW8 PEGFDJGF10
D111701RW5 PCIDDCHE11
D121791RW9 PCDDDLIN12
D132101RW6 PGBEEGHM13
D142113RW11 PFBBDJHL14
P12105G6 PRRTURRS15
P22136G6 PRRTURRT16
P32141Dr10 PRRTURRU17
P42131RW6 PRRTURRS15
P5NPCC1051G1 APPPPPPP18
P6NPCC1052G2 APPPPPPP18
P7NPCC1053G3 ASPQQPPP19
P8NPCC1055M4 APPPPPPP18
P9NPCC1056M4 ATPRRPPQ20
P10NPCC1057FM5 APPPPPPP18
P11NPCC1058FM5 APPPPPPP18
P12NPCC1061FM5 AQQSTQQR21
P13NPCC1063M7 APPPPPPP18
P14NPCC1067M8 APPPPPPP18
P15NPCC1071M8 APSPSPPP22

The yeast were grown in YPD medium (yeast extract 5 g l−1, peptone 5 g l−1 and glucose 20 g l−1) for 4 days. These cultures were used to inoculate 100 ml of YNB medium (yeast nitrogen base 6·7 g l−1 and glucose 20 g l−1) supplemented with ethanol 5% and p-coumaric acid 0·6 mmol l−1 as previously described by Godoy et al. (2008).

All cultures were incubated at 28°C in an orbital shaker (150 rev min−1) for 5 days. The cultures were controlled by cell count using a Neubauer chamber until a concentration of 1 × 108 cells ml−1 was obtained.

RAPD-PCR analysis

Intraspecific characterization was carried out by RAPD analysis according to the methodology described by Martorell et al. (2006). For the preparation of DNA samples, the yeasts were grown for 72 h in YPD medium. DNA was obtained by the Wizard® Genomic DNA Purification kit (Promega, Madison, WI, USA). RAPD-PCR analysis was carried out using seven primers from the series OPA (Operon Technologies Inc., Alameda, CA, USA); OPAE-09, OPAE-12, OPAE-15, OPAR-08, OPAR-14, OPAS-05 and OPAS-11. DNA amplification was carried out in a PTC-100 Peltier Thermal Cycler, MJ Research (Galenica, Santiago, Chile) in a final volume of 25 μl. The reaction mixture contained 2·5 μl Taq Buffer 10× with KCl (Fermentas, Hanover, MD, USA), 0·1 mmol l−1 dNTPs, 2 mmol l−1 MgCl2, 3 μmol l−1 primer, 500 ng de DNA molde y 1 U Taq polimerase. The RAPD-PCR products were visualized on a 2% agarose gel. The band patterns were analysed using the Quantity One software (BioRad, Hercules, CA, USA).

Determination of protein and enzymatic assays

The protein extracts were prepared as indicated by Godoy et al. (2008), and the proteins were quantified according to the method described by Bradford (1976) using bovine serum albumin as standard. The determination of CD activity was based on the methodology described by Godoy et al. (2008) with some modifications. The reaction mixture contained 200 μl of protein extract, 50 mmol l−1 phosphate buffer pH 6·0 and 12·2 mmol l−1 of p-coumaric acid, and incubated at 40°C for 40 min. Subsequently, the mixture was diluted 50 times with bidistilled water to avoid interference with the proteins. CD activity was monitored by the decrease in absorbance at 350 nm. One unit (U) of enzymatic activity was defined as the amount of enzyme that consumes 1 μmol of p-coumaric acid per minute. The VR activity was performed as described by Godoy et al. (2009).The reaction mixture contained 200 μl protein extract, 50 mmol l−1 phosphate buffer pH 6·0, 0·15 mmol l−1 NADPH and 2 mmol l−1 4-vinylphenol. It was incubated at 20°C for 60 min, and the reaction was stopped with 25 mmol l−1 Tris–HCl, 0·3% SDS. The VR activity was monitored by the decrease in absorbance at 340 nm of NADPH, an oxidizable cofactor present in the reaction. One unit (U) of enzymatic activity was defined as the amount of enzyme that consumes 1 μmol of NADPH per minute.

Volatile phenol production

The transformation of p-coumaric acid into 4-VP and 4-EP was monitored using the method described by Ross et al. (2009). Chromatographic separation was performed with a waters HPLC equipped (Waters Corporation, Milford, MA, USA) with a 600 pumps, an UV visible detector. Millenium software (Waters, Oshawa, Canada) was employed for chromatographic control. The separations were performed on a reversed phase column C18 (150 × 4·6 mm). Chromatographic separation was carried out using a gradient elution of (A) H2O: acetic acid 90 : 10 and (B) Methanol as follows: 0–10 min A 100%, 10–15 min A 70%, B 30%, 15–30 min A 30%, B 70%, 30–35 min A 100%. The flow rate was 1·0 ml min−1, and the temperature was set at 45°C. Quantification was performed by comparison against an external standard of p-coumaric acid (range, 0–112 mg l−1), 4-VP (range, 0–100 mg l−1) and 4-EP (range, 0–112 mg l−1).

Amplification of the putative CD gene from spoilage yeast

The Saccharomyces cerevisiae PAD1 gene (access number NM_001180846) encodes for a protein with phenyl acrylic acid decarboxylase activity (Clausen et al. 1994), which metabolizes the cinnamic acids to vinyl derivatives. We used the sequence of putative PAD gen of Debaromyces hansenii CBS767, Candida albicans WO-1 and Giberella zeae PH-1 (www.ncbi.nlm.nih.gov). The resulting alignment showed highly conserved areas from which the degenerate primers 1F (5′-ATTACTGGTGCNACAGGT-3′) and 3R (5′-CATGGAACAGGGCACRACAATC-3′) were designed. The PCR reaction was carried out in a thermal cycler (PTC-100 Peltier Thermal Cycler; MJ Research) using the following programme: one initial step of 95°C for 5 min and 35 cycles consisting of 95°C for 1 min, an annealing of 50°C for 1 min and a final extension of 72°C for 1 min. The resulting amplicons were analysed by electrophoresis in 1% agarose gels. Those products with the highest intensity were extracted and purified using the FavorPrep GEL/pCR Purification Mini Kit (Biotech Corp., Kuala Lumpur, Malaysia). Subsequently, the eluted DNA fragments were cloned in pGEM ®-T and sequenced in Macrogen (Seul, Korea).

Statistical analysis

All experiments were performed in duplicate, and the results were analysed using Statgraphics Plus 5.0 (Manugistics, Inc., Rockville, MD, USA). Differences between treatment means were compared using the Tukey's test at P < 0·05.

The genetic relationship among strains was evaluated using Principal Coordinates Analysis (PCoA) and depicted in a 2D scatter plot by the Numerical Taxonomy System program (NTSYS) (Rohlf 2005). A matrix value of 1 or 0 was assigned to denote the presence or absence of a particular band according to the RAPD-PCR profiles for each strain (qualitative variables). A minimum-length spanning tree was calculated from the simple matching coefficient matrices and superimposed on PCo plots to help detect local distortions. Principal Component Analysis (PCA) on the centred and standardized quantitative variables (4-VP, 4-EP, CD and VR activities) was also performed using the same software.

Results

Molecular characterization of the isolates

The intraspecific characterization of the isolates was carried out by RAPD-PCR analysis with OPA series primers. In the case of D. bruxellensis 14 different haplotypes were identified, whilst eight haplotypes were obtained for Pichia guilliermondii (Table 1, Fig. 1). To simplify the profile analysis and detect potential relationships among isolates, RAPD-PCR patterns were converted into binary data matrices and subjected to PCoA (Fig. 2). The first two axes of the plot obtained after PCoA explained 50 and 83% of the total variability detected among D. bruxellensis and P. guilliermondii, respectively. Scatter-plots resulting from these two axes demonstrated different strain clusters for each species (Fig. 2a,b). For D. bruxellensis, one cluster groups the majority of the studied strains and only strains D2 and D3 do not fall within this cluster (Fig. 2a). On the other hand, the P. guilliermondii strains formed two clusters: cluster I contains strains P1–P4 and cluster II contains the rest of the strains. Additionally, cluster II is not as homogenous as cluster I since isolates P7, P9 and P12 are separated from the rest of the strains.

Figure 1.

Molecular patterns detected among the Dekkera bruxellensis and Pichia guilliermondii strain using RAPD analysis. (a) Analysis of D. bruxellensis strains with primer OPA-E09. Lane 1: λEcoRI/HindIII molecular weight standard, Lane 2: 100 bp molecular weight standard, Lanes 3–16: D9, D7, D13, D6, D1, D2, D11, D10, D12, D8, D14, D4, D5 and D3, respectively. (b) P. guilliermondii strain with primer OPA-E09. Lanes 1–6: strains P5, P7, P1, P9, P2, P5, Lane 7: 100 bp molecular weight standard, Lanes 8–16: strains P6, P7, P8, P3, P12, P10, P11, P13 and P4, respectively. (c) D. bruxellensis strains with primer OPA-E12. Lanes 1 and 18: λEcoRI/HindIII molecular weight standard, Lane 2: 100 bp molecular weight standard, Lanes 3–17: strains D9, D3, D7, D13, D6, D1, D2, D11, D10, D12, D8, D14, D10, D4 and D5, respectively. (d) P. guilliermondii strains with primer OPA-E12. Lanes 1–15: Strains P1–P15.

Figure 2.

Principal coordinates analysis of the individual RAPD-PCR profiles for 14 strains of Dekkera bruxellensis (a) and 15 strains of Pichia guilliermondii (b) D. bruxellensis (D); P. guilliermondii (P).

Quantification of volatile phenols and enzymatic activity

D. bruxellensis and P. guilliermondii strains were grown in medium supplemented with p-coumaric acid. The production of volatile phenols, non-metabolized p-coumaric acid and enzymatic activities of CD and VR were determined (Table 2). All the D. bruxellensis strains produced more than 0·44 mg l−1 of 4-EP, which is considered the sensory threshold for this compound (Ferreira et al. 2002). The majority of the P. guilliermondii strains produced high levels of 4-VP except strains P1–P4. Likewise, all the strains studied produced 4-VP at levels which exceeded the sensory threshold of 0·18 mg l−1(Ferreira et al. 2002).

Table 2. Quantification of precursor acid, volatile phenols and enzymatic activities (CD and VR) for the micro-organisms of this study
Strainp-coumaric acid (mg l−1)CD activity (U mg−1)4-VP (mg l−1)VR activity (mU mg−1)4-EP (mg l−1)
  1. 4-VP, 4-vinylphenol; 4-EP, 4 ethylphenol; nd, not detected; CD, cinnamate decarboxylase; VR, vinyl reductase.

  2. Values within a column followed by the same letter are not significantly different according to Tukey's tests (P > 0·05).

D1 6·37ab13·37a 90·23de21·50e22·13abc
D252·44e 8·06a 21·83ab31·80e22·11abc
D3nd15·03a 1·50a27·50e69·28hi
D4nd39·96b 1·08a24·20e73·08hi
D511·64abcd 9·33a 49·79bc24·10e48·33defg
D6 7·34abc 9·30a 13·23a21·29de54·73efgh
D7nd 9·65a nd21·60e67·88ghi
D8nd11·54a 95·13e20·70cde12·88ab
D9 0·34a 8·66a 93·85e25·80e29·72bcd
D10nd10·16a 6·25a17·60abcde61·71fghi
D11 0·42a 7·80a 62·41cd20·80cde39·45cde
D12 0·07a 8·92a 3·25a20·30bcde76·43i
D13 8·39abc 6·51a 13·66a26·20e58·29efghi
D14 0·05a15·71a nd11·50a72·12hi
P146·67e 6·95a 2·20a 2·15ab28·30bcd
P274·89f 9·54a nd 2·82abc 7·80a
P324·12cd 9·60a 1·70a 0·18a38·20cde
P415·38abcd 4·95a 1·72a 2·88abc44·50def
P5 2·01ab10·42a105·70endnd
P6 0·79a 3·82a107·41endnd
P7 1·57ab 1·74a112·00cendnd
P8 2·03ab 2·18a107·30endnd
P9 5·33ab 4·75a104·70endnd
P10 0·09a11·26a 99·66endnd
P11 0·37a 8·29a109·42endnd
P12nd16·46a 95·61endnd
P1318·12bcd 2·92a 89·86bcndnd
P1416·17abcd 5·31a 85·01dendnd
P1528·29d15·09a 85·07dendnd

As shown in Table 2, there are significant differences in the production of 4-EP among the strains studied. Strains D3, D4, D7, D10, D12 and D14 produced concentrations >60 mg l−1, whilst for strains D1, D2, D8 and P2 <25 mg l−1 of 4-EP were detected in the culture media. The rest of the strains produced intermediate values of this compound. In the conditions assayed, strains P5–P15 of P. guilliermondii (Table 2) did not produce 4-EP, whilst P1–P4 did. However, these strains produced on average a lower concentration of 4-EP than D. bruxellensis (Table 2). We can observe that all the D. bruxellensis strains and strains P1–P4 of P. guilliermondii showed CD and VR activities. In the case of CD activity, except for isolate D4 (which showed the greatest specific activity), there were no significant differences among the strains studied (Table 2). When analysing strains D2 and P3, we observed that on average, D2 had the greatest VR activity, whilst P3 on average had the lowest out of the strains studied. However, it was not possible to correlate the VR activity and the production of 4-EP.

Principal component analysis was used to cluster the strains showing similar production values of 4-VP and 4-EP as well as enzymatic activities (Fig. 3). The PCA plot explained 75% of the total variability in the data in the first two dimensions and four clusters were identified: cluster I contained P. guilliermondii P5–P15 strains and was characterized by a high production of 4-VP and a low concentration of residual p-coumaric acid; cluster II grouped the P. guilliermondii P1–P4 strains and D. bruxellensis D2, which showed intermediate levels of this enzymatic activity; cluster III grouped those D. bruxellensis strains that showed the highest values of 4-EP and enzymatic activities. These left a low concentration of residual p-coumaric acid in the culture medium, similarly to cluster I. On the other hand, cluster IV contained those D. bruxellensis strains, which produced the highest amount of 4-EP and enzymatic activities and left a low concentration of p-coumaric acid. The D. bruxellensis strains D3 and D4 are the greatest producers of 4-EP, with a high CD and VR activity.

Figure 3.

Principal component analysis of 4-vinylphenol, 4-ethylphenol cinnamate decarboxylase and vinyl reductase activity levels obtained from 14 strains of D. bruxellensis (D) and 15 strains of Pichia guilliermondii (P).

Identification of the putative CD gene in Pichia guilliermondii

In S. cerevisiae the phenyl acrylic acid decarboxylase is encoded by the PAD1 gene (Clausen et al. 1994). This enzymatic activity is responsible for transforming cinnamic acids in vinylderivatives, similarly to what the CD activity carries out in D. bruxellensis. To identify the possible CD gene of D. bruxellensis and P. guilliermondii, the PAD gene sequences of S. cerevisiae YJM789, Debaryomyces hansenii CBS767, Candida albicans WO-1 and Gibberellazeae PH-1 were aligned (www.ncbi.nlm.nih.gov), and primers 1F and 3R were designed (see 'Methods and Results'). Subsequently, the DNA of D. bruxellensis D2 and P. guilliermondii P1 and P14 strains was amplified using these primers. In Dbruxellensis, a 200 bp fragment was obtained, which had scarce homology with the PAD1 gene of S. cerevisiae. In the case of P. guilliermondii isolates P1 and P14, the amplicons were of 900 and 1100 bp, respectively. The homology detected between these two sequences obtained from different P. guilliermondii strains studied was very low (25%).

Discussion

In the present study, we evidenced a high degree of molecular polymorphism in Dekkera bruxellensis and Pichia guilliermondii. Using seven OPA primers, we differentiated 14 isolates of D. bruxellensis by the combined haplotype of the RAPD-PCR composite patterns (Table 1, Fig. 1). Using three different OPA primers (OPA-2, OPA-3 and OPA-9) Martorell et al. (2006) did not show genetic differences among the D6–D14 D. bruxellensis strains tested in this work; however, by increasing the number of primers, we obtained a greater differentiation at the genetic level. In fact, we were able to differentiate strains D7, D9 and D12, which in the study by Martorell et al. (2006) showed a different mtDNA restriction pattern but a similar RAPD-PCR pattern. Agnolucci et al. (2009), using three different OPA primers only, obtained six different profiles when analysing 84 D. bruxellensis isolates. By sequencing the 26S rDNA segment, Conterno et al. (2006) grouped 47 D. bruxellensis strains into six clusters. Curtin et al. (2007) analysed 244 Australian isolates and found only eight different genotypes. On the other hand, Oelofse et al. (2009) used different molecular methods (PCR-DGG; ISS-PCR; REA-PFGE) to analyse D. bruxellensis strains and concluded that the species has a low genetic variability.

The PCoA carried out with the RAPD-PCR data for the D. bruxellensis strains showed a large cluster that grouped most of the strains analysed, whilst strains D2 and D3 were grouped apart (Fig. 2a).

The production of volatile phenols (4-VP and 4-EP) and the enzymatic activities (CD and VR) showed very heterogeneous values (Table 2, Fig. 3). A similar observation was made by Conterno et al. (2006) and Oelofse et al. (2009) who stated that isolates from the same or different geographical zones show differences in the production of volatile phenols even though they are genetically close.

The use of RAPD-PCR with eight primers resulted in six different haplotypes within the 15 P. guilliermondii isolates (Table 1). This technique showed a lower capacity to differentiate strains within this species than that obtained in D. bruxellensis. The P. guilliermondii strains were grouped in two clusters: cluster I grouped strains that were very homogenous, whilst the strains of cluster II showed differences between them. Martorell et al. (2006) analysed 32 isolates of this species from Alentejo (Portugal) using mtDNA-RFLP and found a great genetic variability, whereby seven different profiles were detected. Four of these isolates showing a greater production of 4-EP were selected and included in our study. Although we differentiated these isolates using RAPD-PCR, the PCoA confirmed a high genetic relationship between them (Fig. 2b). This genetic relationship was also evident in their physiological evaluation where they were found in cluster II in the PCA as a result of their production of volatile phenols as well as the enzymatic activities analysed (Fig. 3). On the other hand, the P. guilliermondii P5–P15 strains used in this study have been previously identified as different strains by molecular and physiological techniques (Lopes et al. 2009a,b). In the present study, we confirmed the genetic differentiation between these strains using eight OPA primers, which grouped the strains in cluster II. Within this cluster isolates P7, P9 and P12, which came from different samples, were clearly distinguished.

With regard to the physiological characteristics of the P. guilliermondii P1–P4 strains, Martorell et al. (2006) found that this species was capable of producing levels of 4-EP close to 65 mg l−1 in synthetic medium without ethanol. On the other hand, in a study by Barata et al. (2006), the same strains produced approximately 4–12 mg l−1 4-EP in grape juice prior to the onset of fermentation by S. cerevisiae. In our study, the production of 4-EP was carried out in synthetic medium with 5% ethanol and intermediate levels of 4-EP were obtained when compared with those reported by these authors (8–44 mg l−1). This inhibition of 4-EP production in wine spoilage yeasts by increasing concentrations of ethanol has been previously demonstrated (Dias et al. 2003b; Silva et al. 2004). It is important to note that in the conditions assayed in this work only strains P1–P4 were able to produce 4-EP and had VR activity indicating a large phenotypic diversity between the P. guilliermondii strains isolated in different wine-producing areas. Strains P5–P15 used in this study have been previously described as producers of low levels of 4-EP (0·1 mg l−1) in must (Lopes et al. 2009a), this metabolite was not detected in the conditions assayed and with the method of detection used. This difference could be related to the presence of ethanol in the culture media used in this work, the use of different detection methods: gas chromatography in Lopes et al. (2009a) and HPLC in this work or the presence of certain compounds in the must that could induce the enzymatic activities responsible for the production of these volatile phenols. The low production of this compound and the lack of VR activity observed in the P5–P15 strains suggest that the yeast growth conditions for the production of 4-EP is different to that required by the other strains. These differences could be explained by the low sequence similarity of the putative CD gene, only 25% between strains P1 and P14 from cluster I and II, respectively (Fig. 2b).

Based on the proposed stoichiometry for the production of 4-VP and 4-EP by D. bruxellensis (Suárez et al. 2007) and from the analysis of Table 2, it is possible to observe that in strains D3, D4, D6, D7, D10, D12–14 and P1–P4, the sum of 4-VP + 4-EP + residual p-coumaric acid is less than the initial concentration of the cinnamic acid added to the culture medium (100 mg l−1). The production of ethoxyphenols through the interaction of ethanol with p-coumaric acid and phenols may explain the lack of stoichiometry of the enzymatic reaction (Dugelay et al. 1995). An alternative explanation is that the p-coumaric acid may be absorbed by the yeast wall (Salameh et al. 2008) and therefore not be completely available for its metabolization by the yeast.

On the other hand, the comparison of the putative segment for the D. bruxellensis CD gene shows a low homology with the PAD1 gene of S. cerevisiae. This could be due to the phylogenetic distance between both species. Woolfit et al. (2007) carried out a partial sequence of the D. bruxellesis genome and calculated the identity between amino acids with orthologous proteins from both micro-organisms and showed that there is only a 49·9% identity between S. cerevisiae and D. bruxellensis. Likewise, the homology detected between the possible CD gene of P. guilliermondii and the PAD1 gene was very low. Similar percentages were obtained in our study between the 1100 and 900 pb amplicons of P. guilliermondii and the PAD1 gene of S. cerevisiae (42·6 and 30%, respectively). In a recent study, Huang et al. (2012) identified the CgPAD gene of Candida guilliermondii (anamorph of P. guilliermondii); the sequence of this gene also exhibited very low sequence similarity to our amplicons (37%). Since degenerate primers were employed in our study based on other known PAD genes that encode for the decarboxylase activity domain of the yeast, it is possible that we amplified the gene of another enzyme responsible for the decarboxylation of p-coumaric acid.

The P. guilliermondii P1–P4 strains do not consume all the p-coumaric acid and produce 4-EP, whilst the other strains consume the precursor in greater quantity and produce high levels of 4-VP, which could indicate that the VR activity is limiting for the metabolization of p-coumaric acid. This is the first study that detects and evaluates the CD and VR enzymatic activities in P. guilliermondii. These enzymes could be part of an equivalent two step biosynthesis pathway for the production of volatile phenols in D. bruxellensis.

Acknowledgements

This work was supported by FONDECYT grant number 1080376, ProyectoApoyo de VisitasExtranjero USACH and Programa de Cooperación Científico-Tecnológica between MINCyT and CONICYT CH/09/12.

Ancillary