Selection of Botrytis cinerea and Saccharomyces cerevisiae strains for the improvement and valorization of Italian passito style wines

Authors


Correspondence: Giacomo Zapparoli, Dipartimento di Biotecnologie, Università degli Studi di Verona, strada Le Grazie 15, 37134 Verona, Italy. Tel.: +39 0458027047; fax: +39 0458027051; e-mail: giacomo.zapparoli@univr.it

Abstract

In order to improve the quality of Italian passito wine, produced from withered grapes that can be naturally infected by noble rot, in this study, a novel protocol was developed to select suitable cultures of both Botrytis cinerea to infect grapes (as noble rot) and of Saccharomyces cerevisiae to ferment grapes. A total of 16 B. cinerea isolated from withered grapes were typified by RAPD-PCR, and three representative strains were selected for physiological characterization. The strains showed different mycelial growth and enzymatic activities (i.e. polygalacturonase, protease, and laccase). A total of 15 yeasts were isolated from spontaneous fermented wines, these were identified as S. cerevisiae, and typified at strain level. Seven strains were selected according to RAPD-PCR profiles and tested for their fermentation performances. The effects of B. cinerea and S. cerevisiae cultures on the aroma profile of sweet style wine were preliminary evaluated fermenting artificially botrytized grapes induced with B. cinerea infection. The combination of selected fungi affected the aroma profile of wine according to the variation of the content of important molecules (i.e. alcohols, esters, and lactones). This study has provided valuable information to develop new natural cultures destined to induce grape botrytization and manage fermentation in passito winemaking.

Introduction

The term ‘passito’ generally refers to an array of Italian wines (e.g. Recioto, Vin Santo, Zibibbo, Amarone, and Sfurzat) made from withered grapes using ancient, traditional, and artisan winemaking procedures (Domizio & Lencioni, 2011; Paronetto & Dellaglio, 2011). Various types of passito wines may possess marked differences in character, depending on grape variety used, degree of grape drying, vinification technology, and length and method of aging. Those produced in north-east Italy have a common key winemaking practice that is the postharvest drying of grapes, lasting several weeks, in dedicated fruit drying rooms (called ‘fruttaio’) during autumn–winter. During this period, grapes loose moisture thus reducing total weight, and the concentration of sugars and other substances increases (Ribéreau-Gayon et al., 2006). In addition, grapes are susceptible to Botrytis cinerea infection that can develop as noble rot under favorable conditions, thus affecting the saprophytic microbiota (Nisiotou et al., 2007) and the chemical composition of grapes (Ribéreau-Gayon et al., 2006).

As grapes are slowly dehydrated in fruttaio without environmental conditioning in the traditional passito wine production, the incidence of noble rot on the grapes can be extremely variable according to weather conditions. To increase the control of noble rot development, artificial induction of B. cinerea strains in harvested grapes has been tested since the middle of 1900 (Nelson & Amerine, 1956), giving promising results. However, the method has yet to be adopted significantly and, to the best of our knowledge, characterization and selection programs for B. cinerea strains suitable for these peculiar wines have never been carried out.

In the production of passito wines, grape juice fermentation is a critical step. Indeed, both the high sugar concentration in juice and low cellar temperatures (the juice is usually fermented in winter) can hinder yeast activity. This makes the onset of alcoholic fermentation particularly difficult and/or leading to sluggish or stuck fermentations. Most small wineries rely on spontaneous fermentation, although the use of starter cultures is increasing, and is a standard procedure in larger cellars. However, fermentation problems can often occur even when selected yeast cultures are added. In fact, using yeast cultures isolated from fermentation processes that have similar characteristics and are specifically selected to overcome environmental stress conditions found in passito wine production, could better assure the occurrence of alcoholic fermentation and preserve the typical quality of such wines (Domizio et al., 2007; Sipiczki, 2011).

There is considerable literature concerning searching and developing new wine yeasts for various types of wine (Lopes et al., 2006; Capece et al., 2011), included passito wines (Torriani et al., 1999; Dellaglio et al., 2003; Agnolucci et al., 2007). However up to now, no research has been inspired by selection programs for typical strains of B. cinerea and yeasts (e.g. Saccharomyces cerevisiae) for the production of these peculiar kinds of wines.

The concurrent isolation from a particular ecological niche and the selection of both types of fungi for technological purposes can be a new approach to improve and appraise the overall quality of passito style wines. Indeed, the microbial management of grape-drying and alcoholic fermentation by the appropriate use of selected fungal strains could ensure better control of these two critical processes in passito winemaking. In this manner, the goal to obtain wines with valuable and reproducible quality would be more easily reachable by winemakers, because traditional processes, which conducted with poor control of process variables, may still lead to the production of excellent wines, even if their characteristics may vary considerably from year to year.

Therefore, this study is aimed at developing a novel protocol for the selection and characterization of indigenous molds and yeasts to use as starter cultures in the production of passito style wines. The procedure consisted of the following main phases: (1) isolation of B. cinerea and S. cerevisiae from the same ecological niche, that is, withered grapes; (2), identification and typing of isolates at the strain level; (3) strain evaluation by physiological and technological tests; (4) examining of the effect of selected B. cinerea and S. cerevisiae cultures on the aroma profile of a passito wine.

Materials and methods

Isolation of molds and yeasts and growth conditions

Twenty-three samples of naturally withered grapes of Garganega and Corvina varieties were collected from a same number of fruit drying rooms (fruttaio) (one grape sample, amounted to four or five bunches, for each room) located in the Verona and Vicenza winemaking area (Italy) and used for fungal isolation.

For isolation of molds, berries randomly selected from grape bunches were plated on MEA agar (20 g L−1 malt extract, 10 g L−1 peptone, 20 g L−1 dextrose, 15 g L−1 agar). After the incubation at 26 °C for 4 days, individual colonies were isolated, purified, and stored on MEA slants at 4 °C.

The isolation of yeasts was carried out from the wines produced by spontaneous fermentation of the same withered grape samples used to isolate the molds. A total of about 100 g of berries from each withered grape sample was crushed and, without SO2 addition, spontaneously fermented in 250-mL flasks capped with a silicon cap with glass tube to allow the CO2 release during the fermentation. The flasks were kept in a local winery (the temperature ranged between 17 and 22 °C), and the course of fermentation was followed by measuring the weight loss. At the end of spontaneous fermentation (when weight loss ceased), 10-fold dilutions were spread on WL medium (Oxoid Ltd, London, UK) containing 0.1% (w/v) chloramphenicol to inhibit the bacterial growth. After incubation at 28 °C for 3 days, colonies were randomly selected and purified in the same medium. Pure cultures were grown in YPD broth (10 g L−1 yeast extract, 20 g L−1 peptone, 20 g L−1 dextrose), and portions supplemented with 25% (v/v) glycerol were stored at −80 °C.

Molecular analysis

Genomic DNA of yeasts was extracted from overnight YPD cultures applying the protocol of Cocolin et al. (2000). The same protocol was used to extract DNA from molds with some modifications, mainly to improve DNA purification. The mycelium of 10-day-old cultures was disrupted with glass beads (425–600 μm, Sigma, St. Louis, MO) in the breaking buffer, as indicated in the protocol. After centrifugation, the aqueous phase was added with an equal volume of 2× CTAB solution (2% w/v cetyltrimethylammonium bromide, 1% w/v polyvinylpyrrolidone, 100 mM Tris pH 8.0, 20 mM EDTA pH 8.0, 1.4 M NaCl) and incubated at 65 °C for 5 min. Then, two consecutive purification treatments using an equal volume of chloroform/isoamyl alcohol (24 : 1) were performed. Precipitation and resuspension of DNA were performed in ethanol and in bidistilled water, respectively.

Species-specific PCRs for the identification of B. cinerea and S. cerevisiae were carried out as reported by Rigotti et al. (2002) and Torriani et al. (2004), respectively.

Random amplified polymorphic DNA (RAPD)-PCR analysis of B. cinerea isolates was carried out using primer OPD2 (GGACCCAACC) according to Alfonso et al. (2000). The strain B. cinerea DSM 877 (Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures) was used as reference. Data analysis of RAPD-PCR patterns was carried out with the GelCompar 4.0 software (Applied Maths) using the unweighted pair group method with arithmetic means (UPGMA). Flipper and Boty transposable elements were identified in B. cinerea by PCR according to Rigotti et al. (2006) to recognize transposa and vacuma phenotype.

Strain discrimination among S. cerevisiae isolates was carried out by microsatellite multiplex PCR as described by Vaudano & Garcia-Moruno (2008), and by δ-sequence amplification according to Ness et al. (1993). UPGMA dendrogram was obtained by the combination of microsatellite and δ-sequence data. The commercial strain S. cerevisiae EC1118 (Lallemand Inc., Montreal, Canada) was included in these analysis.

Karyotype pulsed-field gel electrophoresis (PFGE) analysis was carried out to discriminate the reference strain EC1118 and Sc49 and Sc81 using the protocol described by Tosi et al. (2009).

Morphology, growth assay, and infection test of B. cinerea

Colony morphology and sclerotia production were examined visually as described by Martinez et al. (2003). Radial growth rate was determined in MEA and in other two media composed by agarized (20 g L−1 agar) red (RGJ) and white (WGJ) grape juices which pH was adjusted at pH 4.0 with NaOH to favor the solidification. An agar plug (about 5 mm2), cut from 10-day-old culture, was placed at the middle of the Petri dish. Radial growth rate was calculated by measuring daily the average of two perpendicular diameters per colony. The trials were carried out in triplicate.

For the infection test, a total of 75 Garganega berries, surface-sterilized by immersion in 2% (v/v) sodium hypochlorite for 10 min and rinsed twice with sterile distilled water, were inoculated with a conidial suspension by puncturing it 2–3 mm deep with a tip. Conidia were harvested from 10-day-old cultures in sterile distilled water, and the suspension was adjusted to a density of 104 conidia mL−1. Control berries were inoculated in the same way with 5 μL of sterile distilled water. The covered plates (Sterilin, Barloworld Scientific, Staffordshire, UK), containing 25 berries each (three replicates for a total of 75), were incubated at 25 °C for 10 days. The degree of mycelial growth on berry surface was expressed (in percentage) on a 0–5 scale, where 0 was no visible mycelium and 5 was two-thirds or more of the berry surface covered by mycelium.

Enzymatic activities

Esterase, glucosidase, protease, and laccase activities were assayed on mycelial suspension, whole yeast cells, and cell-free growth medium. Polygalacturonase (PG) and glucanase activities were determined on mycelial-free extracts.

Mycelial suspension were prepared from 10-day-old cultures in MEA by harvesting with a sterile blade, and the mycelium that was washed twice in sterile distilled water and resuspended in 50 mM potassium phosphate buffer (PPB) pH 7.0.

Mycelial-free extracts were prepared by resuspending the washed mycelium in PPB containing 5 mM phenylmethylsulfonyl fluoride and glass beads for the mycelial disintegration by shaking at high speed.

Whole yeast cells were harvested from YPD stationary-phase cultures, washed twice in 9 g L−1 NaCl and resuspended in the appropriate reaction buffer. Aliquots of cell-free medium, obtained after centrifugation, were also used for the assays.

Esterase, β-glucosidase, and laccase activities were determined as previously reported (Tosi et al., 2012), while protease activity was determined according to Bossi et al. (2006).

The PG and β-1,3-glucanase activities were determined in 50 mM potassium phosphate buffer pH 6.0 and 0.2% (w/v) polygalacturonic acid and laminarine, respectively, as substrate. After an incubation of 20 min at 28 °C, aliquots were taken for measuring the amount of released reducing groups using the Nelson-Somogyi method (Nelson, 1944). PG and β-1,3-glucanase activities were expressed in μmole of galacturonic acid or glucose released h−1 mg−1 of total proteins measured by the Bradford method (Sigma) using bovine serum albumin as the standard, according to the manufacturer's instructions.

All the enzyme assays were carried out in triplicate.

Vinifications

A laboratory microvinification was carried out to assay the fermentation performances of 15 indigenous yeasts and the S. cerevisiae EC1118 used as a control, using 1.5 L of pasteurized (80 °C, 5 min) grape juice (Garganega variety, pH 3.24, 240.0 g L−1 reducing sugars, and 5.8 g L−1 titratable acidity as tartaric acid), inoculated with c. 106 CFU mL−1 cells, harvested from 4-day-old YPD cultures. Alcoholic fermentation, conducted at 20 °C, was monitored by ethanol production.

Two vinifications were carried out in winery in order to produce Amarone and Recioto wines (the former is dry, obtained with the completion of fermentation, the second is sweet). Generally, for both Valpolicella and Soave winemakers, Recioto is the minor production. Because the major concern for Amarone vinification is to avoid sluggish or stuck fermentation, caused by the high ethanol content, the selection of yeasts, suitable for both types of passito wines, should be addressed to those with high ethanol yields.

The Amarone vinification was carried out to test two indigenous yeast strains that were selected from laboratory microvinification on the basis of fermentation performances, and EC1118, preparing trials of 50 kg of withered grapes, that, after crushing, was fractioned in 100-L stainless steel tank. A total amount of 50 mg L−1 SO2 was added in each tank before the yeast inoculation to eliminate most of indigenous microbiota. The juice characteristics were: pH 3.43, 335.0 g L−1 reducing sugars, and 6.0 g L−1 titratable acidity as tartaric acid. Alcoholic fermentation was induced by adding 2 L of fermenting pasteurized juice (pied de cuve) inoculated with 4-day-old yeast culture to reach a concentration of 4–6 × 106 CFU mL−1. The tanks (three for each strain) were kept on a cellar at 15–18 °C.

The Recioto di Soave vinification was prepared using Garganega withered grape berries infected by B. cinerea Bc5 (or not infected in the case of the control) prepared as described in the infection test (see above). The berries (about 200) were incubated at 25, 16, 12, and 8 °C for 2, 4, 5, and 10 days, respectively, at c. 80% relative humidity, to obtain a dehydration rate that mimics the natural withering system. At the end of withering (about 30% of initial weight was loss), c. 70 mL of juices, obtained by crushing separately the botrytized and healthy berries, were inoculated with the indigenous yeast Sc77, that was selected according to its high ethanol yield obtained in the laboratory microvinification and the results of the correspondence analysis of Amarone wine data, or the commercial strain EC1118 at a concentration of about 2–3 × 106 CFU mL−1. Due to the low juice volume, no replicates were prepared for each mold/yeast couple. Alcoholic fermentation was performed in laboratory at 18 °C and monitored by determining the sugar consumption.

The implantation of the inoculated yeasts in the Amarone and Recioto wine vinifications was assessed by plate counts, and molecular analysis carried out on isolates randomly taken from the plates.

Grape juice and wine analysis

Ethanol was analyzed by NIR spectroscopy using Alcolyzer Wine apparatus (Anton Paar GmbH, Graz, Austria). Sugar content in wines during fermentation was determined by the Fehling method with automatic titration (Crison, Allela, Spain). Organic acids and glycerol were quantified using enzyme kits (La Roche, Basel, Switzerland).

Volatile compounds (higher alcohols and acetaldehyde) of Amarone wines were analyzed by a gas chromatograpy (Carlo Erba Instruments, Milan, Italy) equipped with a 200 cm × 2 mm ID glass-packed column filled with 80/20 Carbopack B AW/6.6% and PEG 20M (Supelco, Bellefonte, PA), and with a flame ionization detector (FID). The operative conditions were the same reported by Tosi et al. (2009). Values of compounds were expressed as average (± standard deviation) of three determinations, one for each independent trial (replicates).

Aroma compounds were analyzed by gas chromatography–mass spectrometry (GC-MS) after solid-phase extraction (SPE) using ENV+ cartridge (1 g × 40–140 μm; Isolute, IST Ltd, Mid Glamorgan, UK). The process was performed by an Aspec XL Sample Processor for SPE (Gilson Inc., Middleton). GC-MS analysis was performed with a 6890N Network GC System coupled with a 5978B inert XL EI/CI MS (Agilent Technologies, Milan, Italy), equipped with and a HP-WAX Bonded PEG fused silica capillary column (60 m × 320 μm i.d. × 0.25 μm film thickness; Agilent Technologies). The cartridge conditioning, sample loading, free aroma elution, and instrumental analysis were carried out as previously described (Tosi et al., 2012).

Values of aroma molecules measured on Recioto wine were expressed as average of two determinations of the same sample (repetitions) and standard deviation did not exceed 15% of its average.

Statistical treatment of the data

The t-test was performed on data of growth parameters and enzymatic assays, which values were average (± standard deviation) of three independent trials. This test was also used to treat Amarone wine data. Correspondence analysis (CA) was carried out on the average of three and two determinations to discriminate Amarone and Recioto di Soave wines using the components showed in Table 4, and aroma compounds detected by SPE–GC–MS (Table 5 and Supporting information), respectively.

Results

Isolation, identification, and characterization of B. cinerea

The mold B. cinerea was isolated in 16 of 23 (c. 70%) grape samples, and one isolate for each sample was taken for molecular analysis. Species attribution was confirmed in all isolates by species-specific PCR (Rigotti et al., 2002) (data not shown). Moreover, all isolates were found to belong to the transposa subpopulation of B. cinerea (Giraud et al., 1999) due to the presence of both transposable elements Boty and Flipper, as revealed by PCR amplification of two amplicons of 604 bp (Boty) and 1250 bp (Flipper) (Supporting information).

Typification of B. cinerea isolates was carried out by RAPD-PCR, and the UPGMA dendrogram derived from profile analyses of 16 isolates and reference strain B. cinerea DSM 877 is shown in Fig. 1. Two main clusters (A and B) were obtained at a similarity level of 48%; cluster A included only three isolates and DSM 877, cluster B comprised the remaining 13 isolates with a similarity ranging from 56% to 69% and from 65% to 100%, respectively.

Figure 1.

Dendrogram derived from the RAPD-PCR patterns using OPD primer of 16 Botrytis cinerea isolates from withered grapes and of the reference strain B. cinerea DSM 877.

According to cluster analysis, three isolates representative of the B. cinerea population were selected for further physiological characterization: Bc1 and Bc5 of cluster A (despite grouping into the same cluster, they showed a low similarity level) and Bc13 representative of cluster B.

The three selected strains of B. cinerea were investigated for some morphological (mycelial aspect and sclerotia production) and phenotypic traits (mycelial growth rate) that can be related to the saprophytic ability of strains to spread and infect plant tissues.

The colonies of strains Bc1 and Bc5 on PDA plates showed rather short white mycelium, while for strain Bc13, the mycelium was thinner (data not shown). All molds were able to produce large sclerotia placed irregularly around the plate, thus could be classified as ‘sclerotial type’, while none of the isolates showed a ‘mycelial type’ with rare or no sclerotia, as described by Martinez et al. (2003).

The three strains showed different mycelial growth rates measured as radial growth in three agarized media (MEA, RGJ, and WGJ) and as percentage of infected berries (Table 1). Strains Bc1 and Bc13 displayed a similar growth rate in MEA and WGJ, while growth was significantly different in RGJ. Bc1 resulted to be the most invasive as shown by the percentage of infected berry classified in category 5 (at least two-thirds of berry surface covered by the mycelium) (Table 1).

Table 1. Mycelial growth rates and berry infection of three Botrytis cinerea strains (Bc1, Bc5, and Bc13) measured in agarized media and grape berries, respectively. Daily radial growth, measured on plates of MEA, red (RGJ) and white (WGJ) grape juices, was expressed as cm of diameters per day; berry infection, measured after 10 days of incubation at 25 °C, was expressed in percentage of berries without visible mycelium (1), covered just around the inoculum wound (2), for about one-third (3), half (4) or two-thirds, or more (5) of the berry surface
StrainRadial growth inBerry infection
MEARGJWGJ12345
  1. Different letters means that values are significant for < 0.05.

Bc11.66 ± 0.02a0.90 ± 0.02a1.19 ± 0.01a7 ± 2a0 ± 00 ± 00 ± 0a93 ± 2a
Bc51.47 ± 0.01b0.77 ± 0.06ab1.66 ± 0.02b1 ± 2b1 ± 27 ± 512 ± 4b80 ± 11ab
Bc131.65 ± 0.01a0.69 ± 0.04b1.20 ± 0.01a3 ± 5ab3 ± 58 ± 413 ± 5b73 ± 6b

Enzymatic properties, such as polygalacturonase, protease, and laccase that could be of value in the production and processing of passito wines were examined in the three strains (Table 2). The activity of all enzymes differed significantly among the strains, and of particularly, high activity was the range of protease and laccase. Bc1 displayed the highest activity in all enzyme assays, except that for laccase. Activity of laccase and protease was the lowest in Bc5, while Bc13 displayed the lowest activity of polygalacturonase, β-1,3-glucanase, esterase, and β-glucosidase.

Table 2. Enzymatic activities determined in whole cells of Botrytis cinerea strains (Bc1, Bc5, and Bc13) isolated from withered grapes
StrainPolygalacturon. (μmole h−1 mg−1 total proteins)β-1,3-Glucan. (μmole h−1 mg−1 total proteins)Protease (nmole h−1 g−1 dry weight)Esterase (nmole min−1 g−1 dry weight)β-Glucosidase (nmole min−1 g−1 dry weight)Laccase (nmole min−1 g−1 dry weight)
  1. a

    Values followed by different letters within a column are significant at < 0.05.

Bc11.68 ± 0.02a1.75 ± 0.03a4.0 ± 0.3a2.4 ± 0.2aa4.1 ± 0.2a179.5 ± 12.0a
Bc51.52 ± 0.05b1.60 ± 0.03a0.3 ± 0.0b1.7 ± 0.1b2.3 ± 0.0b30.4 ± 2.1b
Bc130.37 ± 0.01c1.12 ± 0.03b1.5 ± 0.3c1.1 ± 0.1c2.1 ± 0.0c253.3 ± 13.0c

Isolation, identification, and characterization of yeasts

Spontaneous fermentation of 23 grape samples was carried out to isolate yeasts as potential candidates to drive alcoholic fermentation in the production of passito wines. A total of 15 yeasts was isolated from samples showing the best fermentation kinetics (CO2 production rate > 0.3 g L−1 day−1, ethanol yield > 15.0% v/v). All isolates were classified as S. cerevisiae due to the presence of an amplicon of 1760 bp, found under a species-specific PCR condition (Torriani et al., 2004) (data not shown).

Typification of S. cerevisiae isolates was carried out by microsatellite and δ-sequence amplifications. Microsatellite and δ-sequence PCR systems produced six different patterns (A-F and I-VI, respectively) (Fig. 2) and displayed marked differences among isolates (Table 3). Sc49 and Sc81 were discriminated by δ-sequence amplification (profile II) from Sc24, Sc28, Sc66, Sc108, and Sc162 (profile I), while all isolates displayed the same microsatellite profile (A). Microsatellite amplification distinguished Sc165 (profile C) from Sc41 and Sc75 (profile D), which showed the same δ-sequence profile (IV). PCR analysis was carried out on the reference strain EC1118, utilized also in the following microvinifications. The amplifications of microsatellite and δ-sequence did not allow the discrimination of this strain from Sc49 and Sc81 because the commercial strain displays profile A and II, respectively. However, these two indigenous yeasts were clearly distinguished from EC1118 by karyotype analysis (Supporting information). According to this DNA polymorphism data, seven combinations of microsatellite/δ-sequences profiles (A/I, A/II, BIII, C/IV, D/IV, E/V, and F/VI) were recognized. An UPGMA dendrogram derived from microsatellite and δ-sequence profiles of representative strains of each combination (in bold in Table 3) is shown in Fig. 3.

Table 3. Molecular patterns obtained analyzing the molecular fingerprinting of 15 indigenous Saccharomyces cerevisiae isolates by amplification of microsatellite (A-F) and δ-sequences (I-VI). In bold are the yeasts taken as representative isolates utilized for further analysis
 Microsatelliteδ-sequences
Sc24 AI
Sc28AI
Sc66AI
Sc108AI
Sc162AI
Sc49 AII
Sc81AII
Sc23 BIII
Sc156BIII
Sc205BIII
Sc165 CIV
Sc41 DIV
Sc75DIV
Sc67 EV
Sc77 FVI
Figure 2.

Different gel patterns obtained by amplification of microsatellite (a) and δ-sequence (b) of Saccharomyces cerevisiae isolates from withered grapes and of the reference strain EC1118. Molecular sizes (bp) are indicated with arrows.

Figure 3.

UPGMA dendrogram derived from a comparison of gel patterns obtained by the amplification of microsatellite and δ-sequences for seven Saccharomyces cerevisiae strains.

Information about fermentative capability of all selected strains was obtained by performing a laboratory-scale vinification. All strains, except Sc23, displayed similar fermentative vigor, and two of them (Sc24 and Sc77) were characterized by comparing performances with a commercial strain (Supplementary information).

Due to excellent technological performances, two indigenous strains, Sc24 and Sc77, were submitted to further characterization, evaluating their ability to produce enzymes of oenological significance. Strains Sc24 and Sc77 displayed similar esterase and β-glucosidase activity, measured in whole cells (c. 24.0 and 0.4 nmole h−1 g−1 dry weight, respectively). On the contrary, significant differences (P < 0.05) were observed in β-glucosidase and protease activity determined in the supernatant (1.3 vs. 0.7 and 1074.0 vs. 538.2 nmole h−1 mL−1, respectively).

These two strains were assayed to ferment withered grapes for Amarone wine production. The fermentation of these high sugar grapes by yeasts represents a difficult process because the reducing sugars need to be completely consumed. Microbial analysis, carried out on samples taken during fermentation, ascertained that strains, inoculated in the grape must, dominated alcoholic fermentation (data not shown), thus confirming these strains to possess the fundamental properties required for starter cultures. The three Amarone wines differed significantly in some important aroma compounds such as 2-methyl-1-butanol, 2-methyl-1-propanol, 2-phenylethanol and 1-propanol (Table 4). Wine produced by Sc77 contained almost twice of the 2-methyl-1-propanol and about half the 1-propanol than wine obtained by strain EC1118. Wine produced by Sc77 was also poorer in 2-phenylethanol.

Table 4. Composition of three Amarone wines obtained with two indigenous (Sc24 and Sc77) and one commercial (EC1118) strains of Saccharomyces cerevisiae
 Sc24Sc77EC1118
  1. a

    Values followed by different letters within a column are significant at < 0.05.

  2. b

    As tartaric acid.

pH3.65 ± 0.04aa3.58 ± 0.10ab3.58 ± 0.04b
Ethanol, % (v/v)18.29 ± 0.10a18.47 ± 0.31ab18.47 ± 0.08b
Total dry extract, g L−134.06 ± 0.47a35.10 ± 0.32b36.38 ± 1.80ab
Residual sugars, g L−15.20 ± 0.085.17 ± 0.046.41 ± 1.00
Titratable acidityb, g L−16.43 ± 0.04a6.65 ± 0.35ab6.90 ± 0.14b
Acetic acid, g L−10.40 ± 0.02a0.54 ± 0.03b0.56 ± 0.15b
Glycerol, g L−112.43 ± 0.11a12.43 ± 0.02a13.42 ± 0.15b
Total anthocyanins, mg L−1226.5 ± 14.8a250.5 ± 10.6b245.5 ± 21.9ab
Total polyphenols, mg L−11593.0 ± 70.7a1736.5 ± 92.6b1663.5 ± 109.6ab
1-Butanol, mg L−15.50 ± 0.71a4.00 ± 0.12b5.50 ± 0.71a
2-Methyl-1-butanol, mg L−160.00 ± 7.07a49.95 ± 6.29ab47.55 ± 2.62b
2-Methyl-1-propanol, mg L−127.30 ± 1.70a43.40 ± 6.08b22.95 ± 0.21c
1-Propanol, mg L−168.50 ± 13.01a44.65 ± 10.11b87.45 ± 9.40c
2-Butanol, mg L−1< 3.00< 3.00< 3.00
1-Hexanol, mg L−13.20 ± 0.714.05 ± 0.353.65 ± 0.07
3-Methyl-1-butanol, mg L−1257.90 ± 24.04a216.60 ± 17.96b216.10 ± 0.28ab
2-Phenylethanol, mg L−153.50 ± 9.19a37.75 ± 4.74b52.10 ± 4.24a
Ethyl acetate, mg L−154.55 ± 0.4954.60 ± 1.2757.20 ± 2.12
Ethyl lactate, mg L−12.65 ± 0.353.70 ± 1.703.05 ± 1.91

CA that provided information on the similarity among Amarone wines according to their composition clearly separated the three wines (Supporting information). In particular, the distance along axis 1, which explained 80% of the total extracted variance, of wines produced by Sc77 from EC1118 wines was greater than that of Sc24 wines from EC1118 wines.

An informal panel test ascertained differences among the three Amarone wines, in terms of olfactory descriptors, such as jam, almond, cherry, and cinnamon (data not shown).

Botrytis cinerea and S. cerevisiae effects on aroma of Recioto di Soave wine

A final experiment was performed to evaluate the effects of combining two selected strains on wine aroma, one being a B. cinerea strain for grape infection and the other, a strain of S. cerevisiae for fermentation. The Bc5 and Sc77 couple was chosen to carry out a preliminary pilot assay to test the feasibility of inducing grape infection to produce Recioto di Soave wine. The Bc5 strain was preferred over the other two B. cinerea strains, because of its low laccase activity. In fact, this activity can be detrimental to wine quality (browning and premature aging) (Vivas et al., 2010). In regard to yeasts, to reduce the number of the preliminary microvinification trials, only strain Sc77 was selected on the basis of CA of Amarone wine composition. Obviously, a broader and more reliable evaluation on the impact that this procedure can have on overall wine, will be necessary to select a greater number of mold-yeast strains and to test their combinations.

Most compounds detected by SPE-GC underwent marked changes in their contents in wines produced with botrytized (B) and healthy (H) grapes and fermented by Sc77 and EC1118. In particular, 38 of 61 compounds varied at least 50% between yeasts (EC1118 and Sc77) and/or grapes (H and B) (Table 5); the remaining 23 compounds are shown in Supplementary information. The sanitary state of grapes greatly affected wine aroma because more variations were observed between grapes than between yeasts. Indeed, several volatile compounds varied similarly between H and B wines, regardless of the yeast strain that carried out the fermentation. The most interesting are those molecules known to be associated with Botrytis infection, such as 1-octen-3-ol, considered a fungal marker, and benzaldehyde, that increased in content in both B wines. Other compounds of different chemical classes (e.g. endiol 3-oxo-α-ionol, 4-vinylphenols, γ-nonalactone, furfural, homo- and norfuraneol) underwent a very marked changes between H and B wines. In particular, N-(3-methylbutyl)-acetamide increased drastically in both B wines and was detected at a concentration twice as much in wines fermented by Sc77 than by EC1118.

Table 5. Aroma compounds (μg L−1) of four Recioto di Soave wines obtained with healthy (H) or botrytized (B) grapes fermented with Saccharomyces cerevisiae Sc77 or EC1118. Variations of compound content are calculated in percentage (in bold ≥ 50%) between yeasts and grapes
 HBBetween yeastsaBetween grapesa
Sc77EC1118Sc77EC1118HBSc77EC1118
  1. a

    Variation in percentage between yeasts (Sc77 vs. EC1118) calculated as [(Sc77-EC1118)/EC1118] × 100 and between grapes (B vs. H) calculated as [(B−H)/H] × 100, for concentration of < 1.0 μgL−1, it was arbitrarily attributed 0.8. Negative and positive variation indicates decrease and increase in Sc77 or B with respect to EC1118 or H, respectively.

Alcohols
cis-3-Hexenol9.59.814.813.2−312 55 35
Benzyl alcohol13.312.03.63.41167371
Methionol178.493.7132.882.1 90 62 −26−12
Furfuryl alcohol7.414.88.09.250−139−38
1-Octen-3-ol1.41.44.66.16−25 217 347
Homovanillic alcohol38.736.116.927.17−3856−25
Esters
Hexyl acetate6.07.22.7< 1.0−17 232 5689
2-Phenylethyl acetate265.5325.0224.0166.9−1834−1650
Ethyl 9-decenoate32.563.922.236.850−40−32−42
Ethyl 4-hydroxybutyrate11625.529503.315299.830843.26150325
Ethyl 2-hydroxy-4-methylpentan2.7< 1.04.4< 1.0 236 450 63 0
Diethyl succinate32.017.531.622.9 83 38−131
Ethyl-isoamyl succinate< 1.0< 1.02.53.00−19 210 281
Ethyl vanillate1.21.21.71.90−1243 63
Ethyl cinnamate3.54.61.51.1–24395777
Terpenes and norisoprenoides
Citronellol4.37.26.68.9−41−25 55 23
Endiol< 1.01.95.74.65825 614 141
trans-8-Dihydroxylinalool3.710.35.34.064344362
cis-8-Dihydroxylinalool4.99.111.111.5−46−3 128 26
4-Terpineol12.818.19.35.7−29 64 −2769
3-Oxo-α-ionol20.620.846.738.5−121 127 85
Phenols
4-Vinilguaiacol66.872.72.92.1−8389697
4-Vinilfeol397.4427.327.027.5−7−29394
Eugenol2.22.0< 1.0< 1.0706360
Vanillin5.31.97.05.3 185 3132 186
Homovanillic acid2.92.5< 1.0< 1.07−3856−25
Lactones
 γ-Nonalactone4.85.212.79.3−736 165 80
 γ-Butyrolactone663.61291.2921.91709.050−463933
Sherry lactone 2214.9150.6260.8133.643 95 21−11
Carbonyl compounds
Phenylacetaldehyde21.612.123.831.5 79 −2410 160
Benzaldehyde25.521.185.871.72120 236 239
Furfural5.44.010.38.73519 90 117
5-Methyl furfural1.3< 1.01.61.1 65 392143
Furaneol5.05.69.55.7−11 67 91 2
Homofuraneol6.35.5< 1.0< 1.01508785
Norfuraneol27.820.9< 1.0< 1.03309796
Others
Isovaleric acid114.9245.6108.0261.55359−66
N-(3-Methylbutyl)-acetamide40.821.43617.51740.8 91 108 8764 8033

Other molecules (e.g. methionol, ethyl 4-hydroxybutyrate, ethyl 2-hydroxy-4-methylpentanoate and isovaleric acid) were differently constituted in both H and B wines depending on the strain that conducted alcoholic fermentation. Moreover, the comparison of data between grapes provided interesting information about the strain-specific impact on the aroma of Recioto di Soave botrytized wine. For example, endiol, cis-8-dihydroxylinalool, and furaneol increased in B wine produced by Sc77 at a greater rate than in B wine produced by EC1118, while the contrary was observed for the increase of vanillin and phenylacetaldehyde. However, in B wine fermented by EC1118, the decrease of 2-phenylethylacetate, trans-8-dihydroxylinalool, and 4-terpineol was at a higher rate than that in B wine produced by Sc77.

CA discriminated the wines according to the aroma compositions (Fig. 4). H and B wines fermented by Sc77 were well separated, especially by axis 1 that explains the highest variability (68%), while those produced by EC1118 were close to each other.

Figure 4.

Correspondence analysis bi-plot of Recioto di Soave wines (H-Sc77, H-EC1118, B-Sc77 and B-EC1118) and the most descriptive aroma compounds.

Discussion

The purpose of the present study was to explore the fungal biodiversity of withered grapes in order to select suitable indigenous starter cultures for the improvement of oenological production of typical passito style wines. Indeed, grapes are natural environments colonized by a huge number of fungal and bacterial species widely investigated as source of strains suitable for driving wine fermentations (Fleet, 2008).

Typing B. cinerea isolates, by RAPD-PCR, confirmed that gray rot populations are characterized by a high level of heterogeneity, as reported by previous investigations (Thompson & Latorre, 1999; Alfonso et al., 2000; Mirzaei et al., 2009). These strains were not related to geographic origin as evidenced also by cluster A of the UPGMA dendrogram that included indigenous isolates and the reference strain DSM 877. The results of PCR detection of Boty and Flipper transposable elements, suggested that the transposa type is dominant with respect to vacuma type, similar to that observed in other populations of different geographical origin (Esterio et al., 2011; Stylianos et al., 2012).

Regarding the phenotypic characterization, the three B. cinerea strains, selected as representative strains of all isolates according to the UPGMA dendrogram, differed in mycelial growth rates measured in laboratory (MEA) and natural (grape juice) media. Variations on growth rates among several B. cinerea isolates were previously described (Martinez et al., 2003; Mirzaei et al., 2009). Of particular importance was the pathogenicity assay carried out on berries in order to select potential strains to use for postharvest infection to obtain artificially grape botrytization. The exceptional ability to infect berries, as that observed by the three strains, can be positively evaluated because it could mean greater competition against indigenous molds (most of them unwelcome) on grape colonization during withering. However, the effects of competition among fungi strains on the settlement of B. cinerea strain inoculated to induce the noble rot infection deserve further investigation to carry out under specific conditions, like those occurring in drying fruit rooms.

The results of enzymatic assays, to understand the activities involved in the infection process, confirmed the existence of significant intraspecific variability among B. cinerea strains as previously observed (Cotoras & Silva, 2005). Of particularly interest was the data on polygalaturonase, a cell-wall-degrading enzyme involved in virulence toward pathogens (Derckel et al., 1999; Doss, 1999). The concomitance between the lowest percentage of infected berries at stage 5 (Table 1) and polygalaturonase activity in Bc13 suggests that these strains may behave differently in fruttaio infection.

Strain heterogeneity in the activities of enzymes that impact wine quality may be an important aspect to determine those more suitable for botrytized wine production (Nelson & Amerine, 1956). Considering the negative effects of laccase on wine color and aroma (Vivas et al., 2010), the strains with low laccase activity, such as Bc5, should be preferred over strains with high laccase activity for the infection of grapes destined for vinification of passito wines. Moreover, in the specific case of sparkling passito wine production (like ‘Recioto spumante di Soave’ and ‘Recioto spumante di Gambellara’), the presence of Bc5 in the grapes could be preferable over the others, such as Bc1, due to its low protease activity. This latter can negatively affect wine foaming properties (Cilindre et al., 2007).

In general, the level of yeast biodiversity on Botrytis-affected grapes is higher than on healthy grapes because the mold infection alters species heterogeneity and succession during the fermentation (Domizio et al., 2007; Nisiotou et al., 2007; Bokulich et al., 2012). Yeasts, such as Zygosaccharomyces spp., Candida spp., Hanseniaspora spp., Metschnikowia spp., and other non-Saccharomyces species, are frequently found at the beginning of fermentation to produce sweet wine, despite most of them being overwhelmed during the time course by the more specialized species like Saccharomyces spp. (Nisiotou et al., 2007; Urso et al., 2008; Bokulich et al., 2012).

To individuate indigenous yeasts that potentially can be used as starters for passito wine production, selection criteria based on properties that affect the performance on the fermentation process were considered (Fleet, 2008). The isolation of yeasts from wines produced by fast and vigorous spontaneous fermentations allowed us to obtain a yeast pool characterized by a potential fermentation power of at least 15% (v/v) ethanol. This level represents the minimum content for Amarone wine, the most alcoholic among passito wines, although, often the production of this renowned dry red wine needs the completion of alcoholic fermentation to arrive at 18–19% (v/v) ethanol, as shown in Table 4. Thus, this procedure of yeast isolation favored the selection of S. cerevisiae at the expense of other indigenous species endowed with lower fermentation vigor (Fleet, 2008; Urso et al., 2008). The prevalence of isolates belonging to this species is perhaps because Saccharomyces populations can be more competitive than other species in damaged grape berries, such as those affected by Botrytis or subjected to the withering process, than in healthy grapes (Mills et al., 2002; Nisiotou et al., 2007). Finally, it is also plausible that the isolation protocol favored Saccharomyces instead of non-Saccharomyces yeasts because the latter is more frequently present in the form of viable but noncultivable cells. Nevertheless, this last occurrence is more common in aging wine rather than during alcoholic fermentation (Andorrà et al., 2010).

The relevant DNA polymorphism observed among isolates was in accordance with previous studies carried out in botrytized (Sipiczki et al., 2010) and passito wines (Torriani et al., 1999; Dellaglio et al., 2003; Agnolucci et al., 2007). Indeed, the presence of different strains isolated from fermented grapes sampled from different drying fruit rooms was largely expected, considering that the high genetic polymorphism characterized by S. cerevisiae indigenous populations analyzed in singular wine fermentation, included those inoculated with starter cultures (Urso et al., 2008; Vigentini et al., 2009). Nevertheless, the failure of strain discrimination between the two indigenous strains (Sc49 and Sc81) and EC1118, by PCR amplifications, suggests that it could be necessary to use other methods, like PFGE, to have a more reliable estimation of strain heterogeneity, as previously observed (Egli et al., 1998).

The analysis of Amarone wine produced by the indigenous yeast strains confirmed the theory of strain-specific contribution to wine aroma development (Swiegers et al., 2005; Fleet, 2008). Our results point out that phenotypic variability within S. cerevisiae can be almost as wide as that previously described among Amarone strains belonging to different species of Saccharomycessensu stricto’ (Torriani et al., 1999; Tosi et al., 2009).

Of particularly interest was the information provided by Recioto di Soave microvinifications. To the best of our knowledge, no data are published on aroma analysis of wines fermented by different strains in healthy and botrytized grapes. Although the effect of sanitary state of grapes on wine aroma was greater (as revealed by high numbers of variations between H and B grapes) than that of yeasts on aroma, the contribution of the latter also resulted significant. Indeed, beside the expected changes between grapes on the content of molecules linked to noble rot metabolism, such as 1-octen-3-ol, γ-nonlactone, and benzaldehyde (Ribéreau-Gayon et al., 2006; Cooke et al., 2009; Tosi et al., 2012), large variations between yeasts observed for several compounds (ethyl esters, lactones, phenylacetaldehyde, isovaleric acid, and N-(3-metylbutyl)-acetamide) indicate the importance of the choice of strain to modulate wine flavor. The higher divergence of wines produced by the indigenous Sc77 strain with respect to the commercial EC1118 appears very promising. Indeed, these results not only evidenced the different contribution of indigenous and commercial strains to the aroma profile of wine, but suggest that the strain-specific effects can be amplified or enhanced when fermenting infected grapes with respect to the healthy grapes.

In conclusion, this study provides a perspective in the production of special wines, like passito wine, by introducing modern technology and thus enhancing the quality of these wines. New natural sources of B. cinerea and S. cerevisiae have been individuated for the noble rot induction in grapes and fermentation starter culture development to produce boytrytized wines. The possibility to modulate wine flavor, through the artificially postharvest infection of selected B. cinerea strains, combined by the use of indigenous selected yeasts, may be promising in view to produce botrytized passito wines with desired sensory attributes and preserving, at the same time, the traditional and peculiar flavor profile. Considering the novelty of our preliminary approach based on the binomial B. cinereaS. cerevisiae, further investigations are necessary to individuate other new potential strains and to implement their use at winery conditions in order to evaluate the feasibility of this technology for the wine industry.

Ancillary