OUP user menu

Novel wine yeast with mutations in YAP1 that produce less acetic acid during fermentation

Antonio G. Cordente, Gustavo Cordero-Bueso, Isak S. Pretorius, Christopher D. Curtin
DOI: http://dx.doi.org/10.1111/1567-1364.12010 62-73 First published online: 1 February 2013


Acetic acid, a byproduct formed during yeast alcoholic fermentation, is the main component of volatile acidity (VA). When present in high concentrations in wine, acetic acid imparts an undesirable ‘vinegary’ character that results in a significant reduction in quality and sales. Previously, it has been shown that saké yeast strains resistant to the antifungal cerulenin produce significantly lower levels of VA. In this study, we used a classical mutagenesis method to isolate a series of cerulenin-resistant strains, derived from a commercial diploid wine yeast. Four of the selected strains showed a consistent low-VA production phenotype after small-scale fermentation of different white and red grape musts. Specific mutations in YAP1, a gene encoding a transcription factor required for oxidative stress tolerance, were found in three of the four low-VA strains. When integrated into the genome of a haploid wine strain, the mutated YAP1 alleles partially reproduced the low-VA production phenotype of the diploid cerulenin-resistant strains, suggesting that YAP1 might play a role in (regulating) acetic acid production during fermentation. This study offers prospects for the development of low-VA wine yeast starter strains that could assist winemakers in their effort to consistently produce wine to definable quality specifications.

  • mutagenesis
  • acetic acid
  • wine
  • YAP1
  • volatile acidity


During fermentation of grape must to wine, yeast produce an array of metabolites that affect the aroma properties of the final product. Acetic acid is the main component of volatile acidity (VA), which imparts a vinegar-like aroma and is therefore considered a wine fault when present in excessive amounts. Although the aroma threshold for acetic acid depends on the wine variety and style, a concentration of at least 0.9 g L−1 produces a noticeable bitter, sour aftertaste in wine (Ribéreau-Gayon et al., 2006a). According to current legislation, the maximum acceptable limit for VA in most wines is 1.2 g L−1 of acetic acid (OIV, 2010).

VA can be contributed to by several yeast and bacterial species associated with wine, most notably by Acetobacter spp. that if allowed to grow in finished wine will spoil the product by turning it into vinegar. Acetic acid produced by Saccharomyces cerevisiae rapidly forms during fermentation of the first 50–100 g L−1 of sugar, but later some is metabolized. During fermentation, the factors that affect the production of acetic acid include yeast strain (Patel & Shibamoto, 2002; Torrens et al., 2008), nitrogen content (Vilanova et al., 2007), pH, and initial sugar content of the grape must (Ribéreau-Gayon et al., 2006b). The latter is particularly significant given the trend in recent years to harvest grapes at maximum flavor, ripeness. Excessive VA can be removed by reverse osmosis and ion exchange (Jones, 1997) or by re-inoculation (re-fermentation) with appropriate S. cerevisiae wine yeast (Vilela-Moura et al., 2008; Vasserot et al., 2010). Best-practice winemaking avoids such spoilage issues; nonetheless, that under certain conditions can be negative to wine quality. A more elegant solution would be to modify wine yeast to produce less VA, a technological target identified by Pretorius (2000).

Acetic acid metabolism in S. cerevisiae is complex. During alcoholic fermentation, acetate is produced predominantly as an intermediate of the cytosolic pyruvate dehydrogenase (PDH) bypass. This involves conversion of pyruvate to acetaldehyde by pyruvate decarboxylase, which is in turn oxidized to acetate by aldehyde dehydrogenase (ALD). Acetate is then converted into the integral fatty acid building block, acetyl-CoA, through the action of acetyl-CoA synthetase (ACS). Therefore, the production of acetate by yeast during fermentation represents the balance between activity of ALD and ACS; yeast with the lowest ALD activity and highest ACS activity produce the least amount of acetic acid (Verduyn et al., 1990). Indeed, it has been shown that the genetic modification of yeast strains to either overexpress the cytosolic ACS2 (Akamatsu et al., 2000) or delete ALD6, the main cytosolic ALD (Remize et al., 2000; Eglinton et al., 2002; Pretorius et al., 2012), decreases acetate formation by yeast. In addition to being a substrate for ACS, a physiological role of acetate formation may be the regeneration of reducing equivalents for maintaining the redox balance (Van Dijken & Scheffers, 1986). In this regard, a mutant strain with defective NADH dehydrogenase activity produced higher levels of acetate (Kurita et al., 2003).

Developing nongenetically modified strains with reduced acetate production would therefore be of high value in enology. Previously, mutants of saké yeast with resistance to the fatty acid synthase (FAS) 2 inhibitor cerulenin (Goto-Yamamoto et al., 2000), and a brewer's yeast resistant to 2-deoxyglucose (Mizuno et al., 2006), were demonstrated to produce significantly lower concentrations of acetic acid. In the former case, the low-acetic acid production phenotype was linked to a higher ACS activity, while in the second, to a lower ALD activity.

In this study, we have used a classical mutagenesis approach to isolate cerulenin-resistant mutants from the diploid commercial wine yeast strain Maurivin PDM that produce significantly lower amounts of acetic acid during fermentation. Three of the four isolated strains showed mutations in the transcription factor YAP1, essential for the normal response of cell to oxidative stress. The relationship between mutations in YAP1 and the low-VA phenotype of the cerulenin-resistant strains was also examined.

Materials and methods

Microorganisms and culture conditions

All yeast strains were obtained from The Australian Wine Research Institute (AWRI) culture collection, with the exception of Maurivin PDM, hereafter referred to as PDM, which was acquired as a pure strain on agar slopes from Mauri Yeast Australia (AB Mauri). Yeast cultures were maintained on solid YPD agar plates (2% glucose, 2% peptone, 1% yeast extract and 2% agar). YPD and YPG agar plates (3% glycerol, 2% peptone, 1% yeast extract and 2% agar) supplemented with 8 mg L−1 cerulenin (Sigma, C2389) were used to isolate cerulenin-resistant strains. Yeast strains were transformed with the plasmids using the lithium acetate procedure (Schiestl & Gietz, 1989). Grape juice used for fermentations trial was cold-settled at 4 °C for 48 h and filtered through 0.22-μm Stericup filters (Millipore). Minimal synthetic medium (MSM) was used, containing 0.67% yeast nitrogen base without amino acids (Difco) and 8% glucose, and filtered though through 0.22-μm Stericup filters.

Mutagenesis and strain isolation

Mutants were generated using a classical mutagenesis technique (Lawrence, 1991) applied to the PDM strain. A concentration of 6% of the alkylating agent ethylmethane sulfonate (EMS) was used to mutagenize PDM, as described by Cordente et al. (2009). After stopping the mutagenesis with sodium thiosulfate, cells were then washed twice with sterile water, grown for 3 h in liquid YPD, and 1.5 × 108 cells spread on either YPD or YPG agar plates containing 8 mg L−1 of cerulenin. Mutants appearing after 3 days of incubation at 30 °C were restreaked three times to isolate stable cerulenin-resistant colonies.

Sequence analysis and plasmid construction

Standard procedures for the isolation and manipulation of DNA were used (Ausubel et al., 1994). The sequence analysis of YAP1 and FAS2 was performed in the parental PDM strain and four of the cerulenin-resistant strains, as follows. Genomic DNA was isolated, and amplification of both genes was carried out using PfuTurbo® DNA polymerase (Stratagene). Primers Fas2(-900).for (5′-gcagccataagcgatgtaggg-3′) and Fas2(3utr350).rev (5′-ctgaccctttcccctcctcac-3′) were used to amplify a fragment of 6.9 kb, encompassing the FAS2 coding region, plus a fragment of 900 bp upstream and 350 bp downstream of the coding region. Primers Yap1-p(-730).for (5′-aagagagcagcagaggatagg-3′) and Yap1-t(+300).rev (5′-ggataaatctcagcgttgtg-3′) were used to amplify a fragment of 3 kb, encompassing the complete YAP1 coding region, plus a fragment of 730 bp upstream and 320 bp downstream of the coding region. Both YAP1 and FAS2 amplicons were purified (UltraClean PCR Clean-Up kit; Mo-Bio) and sequenced using an ABI PRISM 3730xl DNA Analyzer (Applied Biosystems), by the Australian Genome Research Facility Ltd (Adelaide). The PDM YAP1 sequences were submitted to GenBank with the following accession numbers: JQ433656 and JQ433657, for alleles 1 and 2, respectively. All the allelic variants in YAP1 of the DC mutant strains are relative to the allele 1. The 3-kb YAP1 amplicon from the parental PDM strain and the cerulenin-resistant strains DC12, DC23, and DC49 was subcloned into the pGEM-T-easy vector (Promega), yielding pGEM-T-easy-YAP1.

Chromosomal integration of mutant YAP1 alleles in AWRI 1631

Each of the YAP1 mutant alleles was integrated at the chromosomal location of the YAP1 gene in the haploid wine yeast strain AWRI 1631 via homologous recombination (Ausubel et al., 1994). Briefly, each of the pGEM-T-easy-YAP1 plasmids, containing the YAP1 mutated alleles from strains DC12, DC23 and DC49, was linearized with HpaI in the position 466 within the YAP1 coding region to increase the frequency of the integrative transformation. Approximately 2 μg of the linearized plasmid was then transformed, and YPD plates supplemented with 4 mg L−1 cerulenin were used to select the transformants.

Fermentation conditions

High throughput fermentation experiments

The initial screening for fermentation performance and acetic acid production by the isolated cerulenin-resistant strains was performed in triplicate in 96-well microplates sealed with Breathe-Easy membranes (Astral Scientific). Fermentations were carried out in either 600 μL of filter-sterilized MSM or four different white grape musts. Fermentations were carried out at 23 °C in an anaerobic hood with no agitation. Sacrificial plates were used to assess when the control strain PDM had fermented to dryness, at which point all plates were removed, centrifuged at 1600 g, and the supernatant was analyzed for acetic acid production. Residual sugar was estimated using Clinitest reagent tablets (Bayer Australia), and ferments were considered finished when the results were negative (< 2 g L−1).

Laboratory-scale fermentation in Chardonnay juice

Fermentations were performed in triplicate in 250-mL conical flasks fitted with an air lock and side-arm port sealed with a rubber septum for sampling. Fermentations were incubated at 18 °C with shaking (100 r.p.m.), and their progress was monitored by CO2 loss. Yeast starter cultures were prepared in YPD medium and incubated aerobically at 30 °C with shaking for 24 h to stationary phase. Cultures were then centrifuged and washed with sterilized water and used to inoculate 200 mL of a Chardonnay grape juice at a density of 1 × 106 cells mL−1. Samples were taken periodically from each flask to determine acetic acid concentration. At the end of the fermentation, samples were centrifuged for 5 min at 5000 g and the cell-free supernatants were stored at −20 °C for the analysis of the concentrations of residual sugars, ethanol, glycerol, acetic acid, and organic acids.

Small-scale fermentation in Shiraz grapes

Four-liter glass fermentation vessels fitted with an air lock (filled with 10 mg L−1 potassium metabisulfite solution) and ports for sampling, in addition to a stainless steel screen for submerging the fermentation cap, were used, similar to as previously described (Sampaio et al., 2007). For each treatment, 2 kg of randomized lots of berries were crushed in zip lock bags, pH was adjusted to pH 3.5, and yeast assimilable nitrogen content was also adjusted to 250 mg L−1 Yeast strains were inoculated at a density of 1 × 106 cells mL−1, and fermentations were carried out at 22 °C in triplicate. Fermentation progress was followed by weighing the ferments. After 13 days, the wines were pressed, approximately 1.2 L of the free run was transferred to Schott bottles, and fermentation was continued until all the sugar has been consumed (20 days). Then, the wines were racked to remove the yeast lees and 50 mg L−1 of SO2 added to each bottle. A sample of the finished wine was centrifuged for 5 min at 5000 g, and the supernatants were stored at −20 °C for analysis.

Analytical methods

The concentrations of sugars, ethanol, glycerol, and organic acids were measured by high-performance liquid chromatography using a Bio-Rad HPX-87H column, as described previously (Nissen et al., 1997). Acetic acid was determined using a spectrophotometric commercial kit (R-Biopharm).

Minimum inhibitory concentration (MIC) of drugs

Strains were cultured in liquid YPD medium overnight at 30 °C. An aliquot of 5 μL of the culture was then spotted in YPD agar plates containing the indicated drugs, and growth of the cells was monitored after incubation for 2 days at 30 °C.

Concentrations of H2O2 (from 6 to 14 mM), diazaborine (from 25 to 100 mg L−1), cerulenin (from 0.5 to 8 mg L−1), and cycloheximide (from 0.1 to 2 mg L−1) were assessed.

Enzyme analysis

Cell-free extracts for enzymatic analysis were prepared as follows: 5 mL of each sample was collected from the fermentation flasks at different time points during fermentation, and cells were washed twice with 10 mM potassium phosphate buffer (pH 7.5). Cells were then resuspended in 100 mM potassium phosphate buffer (pH 7.5) containing 2 mM MgCl2 and 1 mM dithiothreitol. Yeast cells were then disrupted by vortexing with glass beads for 5 × 1 min, with cooling intervals in between. After centrifugation for 30 min (16 000 g) at 4 °C, the supernatant was collected and protein concentration was measured using the Bradford assay (Bio-Rad) with bovine serum albumin as standard. Enzyme activities were assayed immediately in a 96-well spectrophotometer. ACS activity was carried out with the same commercial kit used for the determination of acetic acid, according to the study by Postma et al. (1989). The kit ACS enzyme was omitted, and a volume of cell-free extract containing 10 μg of protein was used instead. The reaction was carried out in a total volume of 200 μL and was started with the addition of 10 mM potassium acetate. Absorbance at 340 nm was recorded after 15 min. Alcohol dehydrogenase (ADH) and ALD enzymatic activities were determined according to the study by Mizuno et al. (2006). NAD+ was used as cofactor for the determination of ADH activity, while NADP+ was used for ALD activity. In both cases, 10 μg of protein was used, and absorbance at 340 nm was recorded after 60 and 10 min, respectively.

Volatile ester analysis

Ester analysis was performed as described by Siebert et al. (2005) with the following modifications. The Agilent Technologies 6890 gas chromatograph was fitted with a new column and a guard column, with the following specifications: ~60 × 0.25 mm Agilent J&W fused silica capillary column DB-WAX, 0.25 μm film thickness, with a ~ 0.5 m Restek Siltek deactivated silica guard column. The linear velocity and flow rate of the carrier gas (helium) were changed to 33 cm s−1 and 1.4 mL min−1, respectively. The oven temperature was started at 40 °C, held at this temperature for 4 min, then increased to 130 °C at 5 °C min−1, and finally increased to 240 °C at 40 °C min−1 and held at this temperature for 5 min. The inlet was held at 220 °C instead of 200 °C. A Supelco polydimethylsiloxane 100-μm fiber was used instead of the Carbowax/divinylbenzene 65-μm fiber.


Random mutagenesis

To isolate low-VA-producing yeast derived from the PDM commercial wine yeast, a classical mutagenesis technique was used, based on the incubation of yeast cells with the alkylating agent EMS. A concentration of 6% EMS, which produced a survival rate of 57%, was used. A total of 64 and 67 cerulenin-resistant colonies were isolated in YP agar plates containing glucose (DC colonies) and glycerol (GC colonies) as carbon source, respectively, in the presence of 8 mg L−1 cerulenin. The frequency of induction of cerulenin-resistant mutants was approximately 4 × 10−7 in both selection media.

High throughput screening of the cerulenin-resistant strains

Screening of the cerulenin-resistant strains in MSM

All 131 mutants were used in a series of microfermentations in MSM containing 8% glucose in anaerobic conditions (Table 1). The nonmutagenized (parental) PDM strain and the low-VA-producing wine yeast strain AWRI 1375 (Eglinton et al., 2000) were also included. Acetic acid was determined at the end of the alcoholic fermentation, after 6 days. Overall, both DC and GC colonies showed a wide distribution; the average of acetic acid production by the DC colonies was significantly lower than that of the GC colonies, but not significantly different from the control PDM (Table 1). A total of 20 colonies (3 GC and 17 DC) were selected as they fermented to dryness (< 2 g L−1 residual sugar) and produced at least 25% lower levels of acetic acid compared with the parental strain PDM, similar to the low-VA control AWRI 1375.

View this table:
Table 1

Acetic acid production after fermentation of four different white grape musts and minimal synthetic must (MSM) in a 96-well microplate format

Yeast strain(s)Fermentation mustAverage*
MSMJuice 1Juice 2Juice 3Juice 4
PDM1ab (0.43 g L−1)1a (0.38 g L−1)1a (0.55 g L−1)1a (0.57 g L−1)1a (0.53 g L−1)1a (0.49 g L−1)
AWRI 13750.71bcd0.32cd0.69b0.66b0.42c0.52bc
DC strains (n = 64)0.85bc
GC strains (n = 67)1.13a
  • Results are shown as the average of three replicates and relative to the amount of acetic acid produced by the control strain PDM. The absolute acetic acid production of the strain PDM is indicated in brackets (g L−1).

  • Standard deviations were typically about 10% and never exceeded 20%. Means with the same letter are not significantly different from each other (Tukey's test, P < 0.05).

  • * Average of acetic acid production in the four grape musts.

  • DC and GC strains: isolated in YPD and YPG plates supplemented with 8 mg L−1 cerulenin, respectively.

Screening of 20 cerulenin-resistant strains in four different white grape musts

The 20 selected strains were then screened under the same conditions, but this time in 4 different white grape musts, including two Sauvignon Blanc and two Chardonnay musts (Table 1). The control strain, PDM, produced between 0.38 and 0.57 g L−1 of acetic acid depending on the grape juice. Four strains were selected (DC12, DC23, DC48 and DC49) as they produced, in average, 39–66% less acetic acid than the control PDM (Table 1) and were able to ferment to dryness in at least 3 of the 4 juices.

Laboratory-scale fermentation in Chardonnay juice

Performance of the selected four DC mutants was analyzed in a Chardonnay must, as described in Materials and methods. The PDM control fermented to dryness after 14 days, as did the strains DC12, DC23, and DC49 (Fig. 1a). Strain DC48 displayed a slower fermentation rate than the parental strain, achieving dryness after 17 days. All four DC strains produced significantly less acetic acid than PDM throughout fermentation (Fig. 1b). Strains DC48 and DC49 produced 78% and 90% less acetic acid, respectively, compared with the control strain PDM (Table 2). Regarding the production of other nonvolatile fermentation products, strain DC48 showed an increase in lactic acid production and a decrease in glycerol. Volatile esters were also analyzed in the ferments, and interestingly three of the DC mutants (DC12, DC23, and DC48) produced significantly higher concentrations of acetate esters (ethyl acetate, 3-methylbutyl acetate, and 2-phenylethyl acetate) than the PDM control (Table 2).

View this table:
Table 2

Main fermentation products (g L−1) and volatile ester compounds (μg L−1) produced at the end of fermentation (17 days) of 200 mL of sterile Chardonnay grape juice (Juice 4)

MetaboliteYeast strain
Main fermentation products (g L−1)
Citric acid0.
Tartaric acid1.371.401.231.501.30
Malic acid2.37d2.50cd2.60bc2.91a2.69b
Succinic acid2.232.432.232.162.33
Lactic acid0.08c0.08c0.12b0.17a0.14b
Acetic acid0.15a0.09b0.06b0.03c0.02c
Residual sugar1.
Ethanol (%, v/v)12.612.412.212.612.4
Volatile compounds (μg L−1)
Ethyl acetate51393d96233a72731bc88982ab55312d
2-Methylpropyl acetate69d186a138b93cd103c
3-Methylbutyl acetate1408c4458a3103b2775b1407c
Hexyl acetate297bc453a267bc372ab199c
2-Phenylethyl acetate282c522a457a362b364b
Ethyl propanoate164c213bc242bc362a414a
Ethyl butanoate416b418b306c505ab512ab
Ethyl hexanoate979ab965ab711b1134a1177a
Ethyl octanoate14151279106012651244
Ethyl decanoate302a90bc61c78c170b
  • Average values of three independent repeats. Standard deviations were typically about 10% and never exceeded 20%. Means with the same letter are not significantly different from each other (Tukey's test, P < 0.05).

Figure 1

Fermentation profile of PDM (○) and the cerulenin-resistant strains DC12 (●), DC23 (▲), DC48 (Δ), and DC49 (♢) in 200 mL of sterile Chardonnay grape must (Juice 4). (a) Fermentation kinetics depicted as cumulative weight loss because of CO2 release, and (b) acetic acid production (g L−1) at different time points during fermentation. Standard deviations of the three replicates did not exceed 10%. Fermentations were stopped after 17 days.

Small-scale fermentation in Shiraz grapes

Performance of the selected four DC mutants was also evaluated in red grape must fermentations. Strain DC49 fermented at a similar rate as the PDM control strain, and both strains were dry after 15 days. DC12 and DC48 fermented slightly slower and were dry after 18 days, whereas DC23 was only dry after 20 days (data not shown). All the DC strains produced significantly lower levels of acetic acid when compared with the parental PDM strain (Table 3). Again, DC48 and DC49 strains produced the least acetic acid, 70% less than the control strain. As in the Chardonnay ferment, DC48 produced significantly higher levels of lactic acid than the parental strain and lower levels of glycerol (Table 3). The strains DC12 and DC23 produced higher levels of 3-methylbutyl acetate and 2-phenylethyl acetate, as seen in the Chardonnay ferments. On the other hand, this time only DC12 produced more ethyl acetate than the control PDM, while DC49 produced less. At the end of fermentation, yeast lees were spread in YPD plates supplemented with 8 mg L−1 of cerulenin, to confirm that the inoculated strains were able to finish the fermentation. For DC12 and DC49, 100% of the recovered colonies were cerulenin resistant, while for DC23 and DC48, 91% of the colonies were resistant to cerulenin.

View this table:
Table 3

Main fermentation products (g L−1) and volatile ester compounds (μg L−1) produced at the end of fermentations (20 days) of 2 kg of Shiraz grapes

MetaboliteYeast strain
Main fermentation products (g L−1)
Citric acid0.
Tartaric acid2.
Malic acid2.87bc3.17b3.23ab3.63a2.67c
Succinic acid3.87ab3.93a4.00a3.33c3.67c
Lactic acid0.09cd0.08d0.13b0.17a0.11bc
Acetic acid0.089a0.027b0.023b0.009c0.012c
Residual sugar0.
Ethanol (%, v/v)14.614.714.214.614.5
Volatile ester compounds (μg L−1)
Ethyl acetate41880b49030a42040b38720bc34430c
3-Methylbutyl acetate1790d3410b5640a2470c1900cd
Hexyl acetate4.7c5.7bc7.6a6.3b5.0bc
2-Phenylethyl acetate48c79b100a60c88ab
Ethyl propanoate247d333c465b668a474b
Ethyl butanoate193b178b185b233a249a
Ethyl hexanoate417b260c287c450ab500a
Ethyl octanoate440a313b307b430a400a
Ethyl decanoate4238364340
  • Average values of three independent repeats. Standard deviations were typically about 10% and never exceeded 20%. Means with the same letter are not significantly different from each other (Tukey's test, P < 0.05).

Sequence analysis of the cerulenin-resistant strains with a low-VA phenotype

It has been described that cerulenin resistance is caused by gain-of-function mutations in either FAS2 (Fas2pG1250S) or the transcription factor YAP1 (Yap1pC620F) (Inokoshi et al., 1994; Jungwirth et al., 2000). Therefore, we sequenced both FAS2 and YAP1 genes, with their respective promoter regions, in the parental strain PDM and the four selected DC mutants, to look for mutations.

No mutations were found in the FAS2 gene in any of the DC mutants, so this gene was not investigated further. In the case of YAP1, the diploid parental strain PDM displays several allelic variants with respect to the haploid reference strain S288c. The most noticeable variant is the insertion of 9 and 18 bp in each of the PDM alleles: 1203insGATAGCACT for allele 1 and 1203insGATAGCACTGGTAGCACT for allele 2. In the case of YAP1, three of the four sequenced strains (DC12, DC23, and DC49) showed different mutations, all of them in allele 1. In both DC23 and DC49, the mutations are C-T transitions (1621C>T and 1717C>T, respectively), while in the DC12 strain, a deletion of a single nucleotide (1579delG) was found. In all three cases, the mutations result in a stop codon in the C-terminus of Yap1p (A527fsX537, Q541X, and Q573X for DC12, DC23, and DC49, respectively), before its cysteine-rich domain (CRD). As it has been reported that either missense or nonsense mutations in the C-terminus of Yap1p confer multiple drug resistance (Jungwirth et al., 2000) and hypersensitivity to H2O2 (Kuge et al., 1997; Takeuchi et al., 1997; Coleman et al., 1999), MIC were determined for different drugs (Table 4). All four DC strains were more resistant to cerulenin, cycloheximide, and diazaborine than the parental strain. On the other hand, all the strains behaved as the control PDM in terms of sensitivity to H2O2. Even though the DC48 strain does not have a mutation in YAP1, it exhibited a similar behavior to the other strains in terms of drug resistance.

View this table:
Table 4

MIC of different drugs for the cerulenin-resistant strains (DC strains)

DrugYeast strain
Cerulenin (mg L−1)1> 8> 8> 8> 8
Cycloheximide (mg L−1)
Diazaborine (mg L−1)50> 100> 100> 100> 100
H2O2 (mM)1212121212
  • Strains were grown overnight in liquid YPD, and a volume of 5 μL was spotted on YPD solid medium containing various concentrations of the drugs and incubated at 30 °C for 48 h.

Effect of YAP1 mutations in VA production

To examine whether the YAP1 mutations observed in the DC12, DC23, and DC49 strains were responsible for their low-VA production phenotype, each of the mutated alleles was individually inserted in the haploid wine strain AWRI 1631 in the locus of the YAP1 gene, as explained in Material and methods. A total of three cerulenin-resistant integrants were selected for each of the mutated alleles, and the correct integration of the mutated YAP1 allele was confirmed by PCR and sequencing (data not shown). Interestingly, we found that for each of the introduced alleles, some of the integrants had, as expected, one copy of the mutated allele and another of the wild-type gene in tandem, whereas in others, two copies of the mutated allele were found (Table 5). Microfermentations were carried out to determine the acetic acid production of each of the integrants in four different white musts, with their respective controls. Both control strains PDM and AWRI 1631 produced similar levels of acetic acid (0.32 vs. 0.36 g L−1, respectively), in the average of the four grape musts. As expected, the low-VA mutants DC12, DC23, and DC49 produced significantly lower levels of VA when compared with the parental PDM strain (data not shown). We observed a trend between the number of copies of the mutated YAP1 allele in the 1631 integrants and VA production. Irrespective of the introduced allele, strains with two copies of the mutated YAP1 gene produced significantly lower acetic acid concentrations when compared with the wild-type strain, AWRI 1631. In the case of the DC12-5 and DC12-8 integrants, the decrease in VA was equivalent to the PDM-mutated strain DC12, whereas in the DC23-3 and DC49-5 integrants, the decrease was not as dramatic as for their respective PDM-mutated strains. On the other hand, strains with one copy of the wild-type gene and another of the mutated allele produced the same or higher amounts of VA than the parental AWRI 1631.

View this table:
Table 5

Characteristics of the strains derived from the haploid strain AWRI 1631 with chromosomal integration of the mutated YAP1 alleles

Yeast strainAcetic acid production*MIC H2O2 (mM)Number of YAP1 alleles and origin
AWRI 16311 (0.36 g L−1)c 141 (1631wt)
DC12-50.80d 102 (DC12mut/DC12mut)
DC12-80.78d 102 (DC12mut/DC12mut)
DC12-111.55a 142 (1631wt/DC12mut)
DC23-11.29b 122 (1631wt/DC23mut)
DC23-30.77de 102 (DC23mut/DC23mut)
DC23-71.26b 122 (1631wt/DC23mut)
DC49-11.21b 142 (1631wt/DC49mut)
DC49-31.06c 142 (1631wt/DC49mut)
DC49-50.67e 102 (DC49mut/DC49mut)
  • Standard deviations were typically 10% and never exceeded 20%. Means with the same letter are not significantly different from each other (Tukey's test, P < 0.05).

  • * Results are shown as the average after fermentation of four white grape musts in triplicate, and relative to the amount of acetic acid produced by the control strain AWRI 1631. The absolute acetic acid production of the strain AWRI 1631 is indicated in brackets (g L−1).

  • DC12mut, DC23mut and DC49mut depict the YAP1 alleles of the cerulenin-resistant strains DC12, DC23 and DC49, respectively. These alleles were integrated in the genomic DNA of the haploid AWRI 1631 in tandem with the wild type copy of YAP1 (1631wt), as explained in Materials and methods.

The drug resistance phenotype also differed according to the result of the integration event. Integrants with two copies of the mutated YAP1 were more sensitive to H2O2 than the control 1631 strain, whereas integrants with one copy of the mutated YAP1 showed the same resistance to H2O2 as the 1631 control (DC12-11, DC49-1, and DC49-3 integrants) or were slightly more sensitive (DC23-1 and DC23-7 integrants).

Enzymatic activities of the low-VA strains during fermentation

We measured the activities of the enzymes involved in the production of acetic acid during fermentation of a Chardonnay must, to explain the mechanism behind the lower productivity of this metabolite by the selected DC strains.

ACS activity was measured at different time points during fermentation; however, no increase in activity was observed for the DC strains (data not shown). We estimated ALD activity of the DC strains in the presence of NADP+ (Fig. 2a), because the major isoforms involved in acetate formation during anaerobic growth in glucose (Ald6p and Ald5p) are NADP+ dependent (Saint-Prix et al., 2004). We observed an overall decrease in NADP+-dependent activity for the DC12 and DC49 mutants, and strain DC23 also showed a lower ALD activity at the beginning of the fermentation (days 3 and 5), when most acetate is produced. On the other hand, NADP+-dependent activities for strain DC48 were not statistically different to those of the control PDM. Regarding ADH activity (Fig. 2b), two of the strains DC12 and DC23 had a higher activity than PDM at days 5 and 9, but no differences were observed for the other DC48 and DC49.

Figure 2

(a) NADP+-dependent ALD and (b) NAD+-dependent ADH activities (U mg−1 protein) of cell-free extracts from the control strain PDM (○), and strains DC12 (●), DC23 (▲), DC48 (Δ), and DC49 (♢) during fermentation of 200 mL of a Chardonnay juice. Standard deviations were typically 10% and did not exceed 20%. *Significantly different (Tukey's test, P < 0.05) from the control strain PDM.


In this study, we applied a classical mutagenesis technique in combination with a screening approach based upon microfermentations in multiwell plates. Our success in isolating variants of a commercial wine yeast strain that produce less acetic acid during fermentation highlights that chemical mutagenesis can be efficiently used for diploid industrial yeast strain development (Cordente et al., 2009). Harnessing the microfermentation platform described by Liccioli et al. (2011), we evaluated 131 cerulenin-resistant colonies for fermentation performance and acetic acid production in synthetic media and several white grape juices.

The isolation of cerulenin-resistant S. cerevisiae mutants that produced less acetic acid during saké fermentation was described by Goto-Yamamoto et al. (2000); however, no links were made between the low acetic acid production phenotype and mutations in the yeast genome. Sequence analysis of the four cerulenin-resistant strains selected in our study showed that three (DC12, DC23, and DC49) harbored heterozygous mutations in the transcription factor YAP1. These mutations cause a premature truncation of the protein, before its C-terminal CRD. The transcriptional activity of Yap1p is regulated by its subcellular location, upon induction by oxidative stress Yap1p relocalizes from the cytoplasm to the nucleus (Kuge et al., 97). The CRD domain of Yap1p mediates this relocalization, because its removal or mutation results in a constitutively nuclear and active protein under both stressed and unstressed conditions (Kuge et al., 1997; Coleman et al., 1999). Cells expressing the constitutively active protein are hypersensitive to H2O2 (Kuge et al., 1997; Coleman et al., 1999) and exhibit pleiotropic resistance to drugs such as cycloheximide, cerulenin, and diazaborine (Jungwirth et al., 2000). Pleiotropic resistance was evident for strains DC12, DC23, and DC49, and also for strain DC48 that does not contain mutations in YAP1 or FAS2. On the other hand, the DC strains were not hypersensitive to H2O2. Phenotypic similarity of DC48 may be explained by mutations that increase the levels of Yap1p, because overexpression of YAP1 confers pleiotropic drug resistance (Oskouian & Saba, 1999; Jungwirth et al., 2000; Akada et al., 2002). Alternatively, DC48 may have mutations in Crm1p, responsible for the export of Yap1p from the nucleus to the cytosol, which results in the accumulation of Yap1p in the nucleus in the absence of oxidative stress (Bourens et al., 2008).

The mutant YAP1 allele of DC49 (Yap1pQ573X) is almost identical to the Yap1p1–571 variant (Kuge et al., 1997), the latter showing constitutive nuclear localization, and sensitivity to H2O2. This differential response to oxidative stress could be due to the fact that the Yap1p1–571 allele was present in a haploid strain, while the Yap1pQ573X mutation in DC49 is heterozygous. Supporting this, when the DC mutations were introduced in the haploid strain AWRI 1631, hypersensitivity to H2O2 was only observed in strains with two copies of the mutated YAP1 in tandem, but not in strains where one mutant allele was present together with one wild-type allele. We also observed differential behavior in terms of acetic acid production, according to the number of mutant vs. wild-type alleles in AWRI 1631. Surprisingly, the low-VA phenotype was only observed when the result of integration was that both copies of YAP1 contained the DC mutations, while the combination of a mutant and wild-type alleles (analogous to heterozygosity in DC mutant strains) caused AWRI 1631 to overproduce acetic acid. Therefore, these results imply that regarding the acetic acid production phenotype, the YAP1 gain-of-function mutations present in the PDM-derived strains DC12, DC23, and DC49 are not dominant. We can speculate that differential behavior in terms of acetic acid production for AWRI 1631 integrants nominally containing similar heterozygous YAP1 alleles to those found in DC mutants may be due to other mutations in the DC strain genomes, or that in the AWRI 1631 integrants, the YAP1 alleles are in tandem. It was not possible to study the segregation pattern of the low-acetic acid phenotype because of low sporulation frequency and viability observed for all DC strains.

There are several possible explanations for the gain-of-function mutations in YAP1 affecting acetate production during fermentation (Fig. 3). Previously, low-acetic-acid-producing strains isolated from cerulenin-resistant mutants of sake yeast (Goto-Yamamoto et al., 2000) displayed an increased ACS activity during the late period of growth, which the authors linked to low acetate productivity. We did not, however, observe any change in ACS activity for our DC mutants. Mizuno et al. (2006), hypothesized that the combination of decreased ALD activity (principally) and increased ADH activity would explain the low acetic acid production phenotype of their 2-deoxyglucose-resistant mutant. We observed a significant decrease in NADP+-dependent ALD activity for all three DC strains with a mutation in YAP1, but not for DC48, and in addition, two of the YAP1 mutants (DC12 and DC23) also showed a significantly higher ADH activity. Variation in the amount of acetic acid produced by fermenting cells of different wine yeast strains was recently shown to correlate with ALD6 expression levels (Rossouw et al., 2008).

Figure 3

Schematic representation of the production and fate of acetate during fermentation via the cytosolic PDH bypass. Acetate is mainly produced by the NADP+-dependent cytosolic Ald6p, with a minor contribution by the mitochondrial isoforms Ald5p and Ald4p. The Ehrlich pathway for fusel alcohol production from the amino acids Leu, Phe, and Val is also shown. Genes up-regulated by Yap1p are underlined, according to the studies by DeRisi et al. (1997), Dumond et al. (2000) and Ma & Liu (2010). In bold face-type, direct evidence of interaction between YAP1 and the promoter of the genes (according to the study by Salin et al. (2008)). *Promoter of the gene contains Yap1p response element (YRE 5′-TKACTMA-3′).

A ChIP-chip genome-wide location analysis under oxidative conditions showed that Yap1p does interact directly with the promoters of the NADP+-dependent ALD6 and ALD5 genes (Salin et al., 2008). During anaerobic growth on glucose, acetate is mainly produced by the cytosolic PDH bypass via Ald6p, with a minor contribution involving mitochondrial Ald5p and Ald4p (Remize et al., 2000; Saint-Prix et al., 2004). We can speculate that deletion of the CRD in DC mutants 12, 23, and 49 could potentially alter the binding of Yap1p to the promoters of ALD5 and ALD6, and subsequently the expression of these proteins. Further experiments are needed to prove this assumption.

There are several references linking YAP1 with the regulation of the expression of ADHs. YAP1 interacts directly with the promoter of the NADPH-dependent medium chain ADH ADH6, the 3-methylbutanal reductase GRE2, and the class III ADH SFA1 (Salin et al., 2008); the latter also found to be induced in cells overexpressing YAP1 (DeRisi et al., 1997). In addition, ADH7, which belongs to the same family as ADH6, and GRE2 are highly induced by Yap1p (Ma & Liu, 2010). While the role of the ADHs ADH6, ADH7, and SFA1 might not be important in the reduction of acetaldehyde to ethanol during glucose fermentation (de Smidt et al., 2012); these enzymes are thought to participate in the synthesis of fusel alcohols from certain amino acids through the Ehrlich pathway (Larroy et al., 2002; Hazelwood et al., 2008), along with ADH1-5 and GRE2 (Fig. 3). Particularly, Sfa1p can catalyze the final reactions in phenylalanine degradation to form 2-phenylethanol (precursor of the ester 2-phenylethyl acetate), and Gre2p is thought to be involved in the production of 3-methylbutanol (precursor of 3-methylbutyl acetate) from leucine catabolism (Dickinson et al., 2003; Hauser et al., 2007). Interestingly, strains DC12 and DC23 produced significantly higher concentrations of these two acetate esters in both white and red fermentations and showed a higher ADH activity during fermentation.

Ester production rate is influenced, among other factors, by the concentrations of acyl-CoA and a higher alcohol (Verstrepen et al., 2003). Therefore, we can hypothesize the following mechanism would be responsible for the lower acetate production in strains DC12 and DC23. The gain-of-function mutations found in YAP1 in these strains could lead to the up-regulation of the expression of ADHs, which would increase the levels of fusel alcohols and therefore divert acetate, via acetyl-CoA, toward the synthesis of acetate esters. It is also possible that acetyl-CoA may be diverted via acetyl-CoA carboxylase (ACC) 1 through malonyl-CoA for de novo biosynthesis of fatty acids. Microarray analyses have demonstrated that YAP1 up-regulates ACC1 in an H2O2-independent manner (Dumond et al., 2000) (Fig. 3).

Of the four cerulenin-resistant colonies isolated, DC49 displayed the most promising characteristics from an oenological point of view, because it produced the least acetic acid while retaining fastest wild-type fermentation rate, both in red and white ferments. With the exception of acetic acid, the basic wine parameters remained largely the same for wines made with the DC strains. As described previously, some strains produced higher concentrations of aroma compounds 2-phenylethyl acetate (floral aroma) and 3-methylbutyl acetate (banana aroma), which could enhance the sensory qualities of wine. The fact that these strains show multiple drug resistance could be used as a tool to monitor their implantation in the cellar. These strains represent a viable alternative to processes aimed at reducing VA concentrations after they have reached high levels in wine and may be particularly applicable for wine styles where grapes are harvested with very high starting sugar content.


The Australian Wine Research Institute, a member of the Wine Innovation Cluster in Adelaide, is supported by Australia's grapegrowers and winemakers through their investment body, the Grape and Wine Research Development Corporation, with matching funds from the Australian Government. The authors thank Mauri Yeast Australia (AB Mauri) for funding this project and Dr. Anthony Heinrich in particular for helpful discussions throughout the project. We are grateful to Dr. Helmut Bergler from the University of Graz (Austria) for donation of diazaborine and Mark Solomon for technical assistance. The authors are also grateful to Jenny Bellon for technical assistance with sporulation and dissection. Gustavo Cordero-Bueso thanks the Office Internationale de la Vigne et du Vin (OIV) for his grant.


  • Editor: Jens Nielsen


View Abstract