Applied Microbiology and Biotechnology

, Volume 97, Issue 1, pp 223–235

Yeast genes involved in sulfur and nitrogen metabolism affect the production of volatile thiols from Sauvignon Blanc musts

Authors

    • School of Biological SciencesUniversity of Auckland
    • Centre de Recherche Pernod Ricard
  • Richard C. Gardner
    • School of Biological SciencesUniversity of Auckland
Applied genetics and molecular biotechnology

DOI: 10.1007/s00253-012-4198-6

Cite this article as:
Harsch, M.J. & Gardner, R.C. Appl Microbiol Biotechnol (2013) 97: 223. doi:10.1007/s00253-012-4198-6

Abstract

Two volatile thiols, 3-mercaptohexan-1-ol (3MH), and 3-mercaptohexyl-acetate (3MHA), reminiscent of grapefruit and passion fruit respectively, are critical varietal aroma compounds in Sauvignon Blanc (SB) wines. These aromatic thiols are not present in the grape juice but are synthesized and released by the yeast during alcoholic fermentation. Single deletion mutants of 67 candidate genes in a laboratory strain of Saccharomyces cerevisiae were screened using gas chromatography mass spectrometry for their thiol production after fermentation of SB grape juice. None of the deletions abolished production of the two volatile thiols. However, deletion of 17 genes caused increases or decreases in production by as much as twofold. These 17 genes, mostly related to sulfur and nitrogen metabolism in yeast, may act by altering the regulation of the pathway(s) of thiol production or altering substrate supply. Deleting subsets of these genes in a wine yeast strain gave similar results to the laboratory strain for sulfur pathway genes but showed strain differences for genes involved in nitrogen metabolism. The addition of two nitrogen sources, urea and di-ammonium phosphate, as well as two sulfur compounds, cysteine and S-ethyl-L-cysteine, increased 3MH and 3MHA concentrations in the final wines. Collectively these results suggest that sulfur and nitrogen metabolism are important in regulating the synthesis of 3MH and 3MHA during yeast fermentation of grape juice.

Keywords

Saccharomyces cerevisiaeSauvignon BlancSingle-gene deletionAroma compoundsWineVarietal thiols

Introduction

Volatile thiols are important aroma compounds in foods and beverages and are major contributors to the aroma of red and white wines. Thiols typically have very low perception thresholds, in some cases down to the parts-per-trillion level (Mestres et al. 2000). Although the majority of volatile thiols in wine are associated with unpleasant odors (i.e., ethanethiol, hydrogen sulfide), a few contribute pleasant fragrances reminiscent of tropical fruits and boxwood (Mestres et al. 2000; Swiegers and Pretorius 2007b). In particular, it is now well established that the distinctive sensory characteristics of wines made from the grape variety Sauvignon Blanc (SB) are mainly due to three volatile thiols: 3-mercaptohexan-1-ol (3MH), 3-mercaptohexyl-acetate (3MHA), and 4-mercapto-4-methylpentan-2-one (4MMP) (Tominaga et al. 1996, 1998a). Strong correlations of all three volatile thiols with aroma descriptors such as passion fruit, grape fruit, and sweaty have been demonstrated in SB wine (Lund et al. 2009; Swiegers et al. 2007a).

3MH and 4MMP are not detectable in grape juice and are formed by yeast during alcoholic fermentation by enzymatic degradation of odorless precursors, like conjugates with cysteine (Peyrot des Gachons et al. 2000; Tominaga et al. 1995, 1998b) or glutathione (GSH) (Fedrizzi et al. 2009; Grant-Preece et al. 2010; Peyrot des Gachons et al. 2002; Roland et al. 2010a) and through alternative biogenetic pathways involving aromatic unsaturated carbonyl compounds such as (E)-2-hexenal (Schneider et al. 2006). The common model for 3MHA synthesis during must fermentation is the acetylation of the volatile thiol 3MH, which is mediated by alcohol acetyltransferase(s) produced by yeast (Swiegers et al. 2006). However, alternative non-3MH precursors cannot be categorically ruled out.

Molar conversion of the cysteinylated and glutathionylated precursors into their corresponding thiol is generally very low, typically less than 10 % and sometimes less than 1 % (see recent reviews by Coetzee et al. 2012; Peña-Gallego et al. 2012; Roland et al. 2011). To date, estimates of the combined contribution of these various possible pathways made using labeled precursors can only explain around 10 % of the total thiols produced (Roland et al. 2010a; Schneider et al. 2006; Subileau et al. 2008a). Currently, it is unclear the degree to which the various proposed pathways contribute to thiol production or whether there are additional unidentified pathways. Recent data suggest that grape harvesting practices and treatments can have a significant influence on thiol yields (Allen et al. 2011; Capone et al. 2011).

Regardless of the pathway(s) used, the absolute requirement for yeast activity in the generation of volatile thiols is clear. There have been some success in identifying yeast genes involved, including those encoding the carbon–sulfur β-lyase activity responsible for cleavage of cysteinylated precursors (Swiegers et al. 2007a; Tominaga et al. 1998b). Howell et al. (2005) found that deletions of BNA3, CYS3, GLO1, or IRC7 exhibited at least a 40 % reduction in 4MMP synthesis, after fermentation in chemically defined medium spiked with synthetic Cys-4MMP. However, Thibon et al. (2008) showed that a BNA3 deletion affected thiol synthesis only when Cys-4MMP was spiked at very high concentrations and that a CYS3 deletion exhibited no influence on 4MMP and 3MH synthesis. Only the knockout of the gene IRC7 resulted in significant reductions in both 4MMP (~96 %) and 3MH (~40 %) independent of the initial cysteinylated precursor concentration. Recently, the IRC7 gene has been confirmed to have β-lyase activity and to be responsible for 4MMP production in SB (Roncoroni et al. 2011). However, deleting this gene did not affect production of either 3MH or 3MHA in grape juice. Moreover, most strains (including the main laboratory strain) have a truncated form of IRC7, which may be inactive for 4MMP production (Roncoroni et al. 2011). Thibon et al. (2008) further demonstrated that the β-lyase activity of IRC7 was transcriptionally regulated by the proteins Ure2p and Gln3p, which are core enzymes of nitrogen catabolic repression (NCR) in yeast (Cooper 2002). The release of NCR, either genetically induced (Thibon et al. 2008) or chemically provoked (i.e., by the use of urea as a sole nitrogen source (Subileau et al. 2008b)), resulted in an increase in thiol synthesis by yeast. The Saccharomyces cerevisiae STR3 gene also exhibits a broad β-lyase activity, which was recently shown to liberate 3MH and 4MMP from their corresponding cysteinylated precursors in vitro (Holt et al. 2011). However, this side activity of Str3p was about 99.4 % lower than against its physiological substrate L-cystathionine, and overexpression of the STR3 gene in commercial wine yeast VIN13 led only to a moderate increase of 3MH during SB fermentations. The knockout phenotype of STR3 was not examined (Holt et al. 2011).

The only information about the genes regulating the uptake of potential 3MH precursors was provided by Subileau et al. (2008a, b). When the general amino acid permease GAP1 was deleted in a laboratory strain, 3MH synthesis decreased significantly in synthetic media spiked with Cys-3MH (Subileau et al. 2008b). However, the gap1 deletion strain showed no effect on volatile thiol production in SB grape must. The same group showed that deletion of the glutathione transporter, OPT1, gave a 50 % reduction in 3MH and 3MHA in grape must (Subileau et al. 2008a), which indicated that GSH–3MH could be a potential 3MH precursor.

Here we describe a screening of mutants of the laboratory yeast strain deleted in genes potentially influencing volatile thiol synthesis. Because of the time and cost of thiol analysis, a set of strains individually deleted for 67 candidate genes was analyzed, rather than the entire deletion collection. The seven major genes that influenced thiol production in the laboratory strain were then deleted in a wine yeast to confirm their role in varietal thiol synthesis during fermentation in SB must. The major target of the work was to identify β-lyase gene(s) responsible for cleaving cys-3MH. Although we were not aware of it when we initiated the work, all three yeast genetic backgrounds that were used in this study contain versions of the IRC7 β-lyase gene that are deleted for the C-terminal 60 amino acids and are likely to be inactive (Roncoroni et al. 2011). Considering that 3MH may be derived from different, partially unknown precursors, this screening was entirely carried out using fermentation in grape juice and in particular in SB musts from the Marlborough region of New Zealand, which are known to have high-thiol potential (Benkwitz et al. 2012; Lund et al. 2009). This screen ensured a more oenologically relevant approach than previous studies with synthetic media to which possible precursors were added.

Material and methods

Yeast strains

Yeast strains used in this work belong to the species S. cerevisiae and are listed in Table 1. Single-gene deletion strains were purchased from EUROSCARF (http://web.uni-frankfurt.de/fb15/mikro/euroscarf). Commercial wine yeast strains F15 (Zymaflore F15, Laffort, France) and GRE (Lalvin ICV-GRE, Lallemand, Canada) were kindly provided by Andy Frost of Pernod Ricard NZ, Blenheim. Homothallic spores of these two wine strains were derived by two rounds of tetrad dissection. These spores were then used to create the corresponding haploid and uracil–auxotrophic strains F15-h and GRE-h (see Table 1) by performing two standard homologous replacements. Firstly, the HO gene was replaced with the hygromycin-resistant hphMX cassette, followed by the replacement of the URA3 gene with the ura3∆0 mutation from BY4743 (Heather Niederer, this laboratory). F15 and GRE were chosen as model wine yeast strains because they showed homozygous microsatellite profiles paired with good spore viability and transformation efficiency (Bradbury et al. 2006). Haploid progenies of F15-h and GRE-h showed no difference in fermentation performance to their diploid parents and were used for the constructions of the single-gene deletions and over expression studies.
Table 1

Yeast strains used in this study

Strain

Genotype

Comment

BY4743

MATa/α, his3∆1/h is3∆1, leu2∆0/leu2∆0, LYS2/lys2∆0, met15∆0/MET15, ura3∆0 /ura3∆0

Background strain of single-gene deletion library, laboratory yeast

EC1118

Heterozygous, wild type

Wine yeast isolated in Champagne; Lallemand, Canada

F15

Heterozygous, wild type

Zymaflore F15, Laffort, France

F15-h

MATα, ho::hphMX, ura3∆0

Haploid auxotroph derived from F15a

GRE

Heterozygous, wild type

Lalvin ICV-GRE, Lallemand, Canada

GRE-h

MATα, ho::hphMX, ura3∆0

Haploid auxotroph derived from GREa

aProvided by Heather Niederer

Grape juice and fermentation conditions

SB grape juices from 2005 and 2006 harvests were kindly supplied by Andy Frost of Pernod Ricard NZ, Blenheim. Both juices were mechanically harvested from the Squire vineyard located in the Rapaura region of Marlborough, South Island, New Zealand and were stored at −20 °C prior to use. The properties of both juices are given in Table 2. Both juices were sterilized using dimethyl dicarbonate (DMDC, 200 μL/L) and incubated at 25 °C with 100 rpm shaking for 12 h prior to start of vinification. The weak fermentation of BY4743, which is the background laboratory strain of the single-gene deletion library (http://www-sequence.stanford.edu/group/yeast_deletion_project/deletions3.html) in grape must, was overcome by applying measures described in detail by Harsch et al. (2010): (a) grape juice was diluted with 25 % synthetic grape media without sugars and (b) auxotrophies were compensated for by supplementing with elevated amounts of histidine (200 mg/L), leucine (300 mg/L), and uracil (100 mg/L). Medium with these adjustments is referred to as “optimized juice.” Auxotrophies arising from a particular deleted candidate gene during the screen were accounted for as described in the text. Fermentations inoculated with wine yeast-derived strains were conducted in undiluted grape juice, with uracil (100 mg/L) added when necessary. All ferments were carried out in 250-mL conical flasks sealed with fermentation locks. Single yeast colonies were used to inoculate standard yeast peptone dextrose (YPD) pre-culture medium, which was incubated at 28 °C with 150 rpm shaking. After 24 h pre-culture, yeast cells were harvested via centrifugation (2,800 × g, 5 min), resuspended in water and used to inoculate a fermentation volume of 200 mL grape juice at a concentration of 2.5 × 106 cells per milliliter. Non-inoculated juice controls were included in all microfermentation experiments. All fermentations were conducted at 25 °C with 100 rpm shaking in triplicate unless otherwise stated. Fermentation progress was monitored by weight loss every 12 h during the logarithmic growth of the ferment and every 24 h thereafter.
Table 2

Properties of SB grape juices from the Squire vineyard (Marlborough, NZ)

Juice properties

2005

2006

Degree Brix

22.2

22.8

Yeast available nitrogen (mg/L)

330

194

Titratable/total acidity (g/L)

9.9

10.5

Volatile acidity (VA) (g/L)

0.17

0.3

Tartaric acid (g/L)

5.8

4.7

Malic acid (g/L)

4.2

3.7

pH

3.1

3.1

L-cysteine (mg/L)

1

2

L-histidine (mg/L)

27

28

L-leucine (mg/L)

29

24

L-methionine (mg/L)

6

6

L-serine (mg/L)

63

79

Manipulation of DNA

All oligonucleotides used for the deletion of genes in wine yeast and the overexpression of genes in wine and laboratory yeast are listed in Table S1 and Table S2, respectively. Polymerase chain reactions (PCR) were conducted using standard molecular procedures described in the supplementary material deposited online.

Construction of single-gene deletions in wine yeast

In order to knock out genes in wine yeast, a PCR-based deletion strategy was chosen that is based on Wach et al. (1994, 1996) but also takes advantage of the existing deletion mutant collection in the BY4743 yeast background. DNA was extracted from the appropriate BY4743 single-gene knockout strain using Chelex (Field et al. 1996; Gietz et al. 1995). Primer pair GeneX-F + R (Table S1) was used to amplify the KanMX4 deletion cassette including 300-bp regions of homology 5´ and 3´ to the wild-type gene. After DNA purification (High Pure PCR Product Purification Kit, Roche) and fragment size verification via electrophoresis, haploid wine yeasts F15-h and GRE-h were transformed by a lithium acetate–polyethylene glycol method (Field et al. 1998). To allow recuperation, cells were centrifuged at 2,000 × g for 2 min, the supernatant discarded, yeast cells resuspended in 2 mL of YPD and incubated for 3 h at 28 °C with gentle shaking (100 rpm). Cells were then plated on YPD containing the antibiotic G418 (Sigma Aldrich) and incubated at 28 °C. Colonies were selected, replica plated onto YPD-G418 and single colonies tested via PCR with three test primer pairs: GeneX-F/R-test (Table S1), GeneX-F-test/5´kanI-R, and GeneX-R-test/3´kanI-F (Table S1 and S2).

Overexpression of genes in yeast

Overexpression of genes was achieved by using a method based on homologous recombination, similar to the one used for the creation of the single-gene deletions described in the previous paragraph and references therein. In this approach, the high-level, constitutive promoter from the yeast phosphoglycerate kinase gene PGK1 was integrated immediately upstream of the chromosomal copy of the targeted gene. The overexpression cassette contained two marker genes encoding for URA3 and kanMX4 as well as the PGK1 promoter (UKP cassette). This UKP cassette was previously engineered, transformed, and successfully expressed in haploid wine yeast F15-h (Heather Niederer and Soon Lee, this laboratory). This transgenic wine yeast strain was used in this work to extract total genomic DNA (Hanna and Xiao 2006). The UKP cassette was then amplified from the template DNA with the appropriate UKP-GeneX-F/R primer pair (Table S2), which was designed with 45 bp of gene-specific homology to either side of the start codon of the target gene. After checking the size of the amplicon via gel electrophoresis and purification, the amplicon was inserted via transformation into diploid BY4743 and haploid F15-h strains, which were then plated onto YPD containing G418. Replica plating of putative transformants onto G418 medium lacking uracil and subsequent amplification of the UKP module with UKP-GeneX-test-F/R and 3´kanI-F/UKP-GeneX-test-R primers ensured that the cassette had integrated at the right locus. Figure S1 illustrates this overexpression strategy.

Extraction of volatile thiols and their quantification via GC-MS

The quantification of volatile thiols is based on the method developed by Tominaga et al. (1998a, 2006) which involves derivatisation of the volatile thiols with p-hydroxymercuribenzoate, column chromatography, and extraction with dichloromethane. The full method, including run conditions for gas chromatography mass spectrometry (GC-MS) analysis used in this work, is described in detail by Anfang et al. (2008).

Compiling the list of candidate genes

The search strategies employed for the generation of the list of candidate genes (Table S4) are described in detail in the supplementary material.

Statistical analysis

The p values were derived by applying a two-tailed Student’s t test (Student 1908). Variances of this pairwise comparison were estimated with Fisher’s F test. However, the approach of null hypothesis significance testing using p values does not assess the magnitude of an effect or the precision of the effect size estimate (Nakagawa et al. 2007). Consequently, a statistically significant effect does not necessarily relate back to biological and practical importance. In an attempt to assess the practical importance, here called biological relevance, of a statistically significant effect, we also applied effect size statistics. Formulas 1–5 (see Table S3) were used to calculate Hedges’ d (d), a parameter for the standardized bias-corrected effect size, which estimates the magnitude of the difference between two measurements along with an associated 95 % confidence interval. The absolute value of d describes the magnitude of the difference of the effect (d = 0 means no difference). The sign of the number indicates the direction of the effect compared to the control group. Cohen (1988) proposed benchmarks for small (d < 0.2), medium (d = 0.2–0.5) and large (d > 0.8) biological effects. Here we applied a much more stringent cutoff point of d ≥ |4| for a result to have biological relevance.

Results

Deletion of CYS3, CYS4, or MET17 increases volatile thiol synthesis in the laboratory strain BY4743

The first screen for thiol production in optimized grape juice included 12 different single-gene deletion strains in the diploid BY4743 laboratory background (Fig. 1a). Two control strains were used: the wild-type parental BY4743 strain and BY4743-yal068c. The latter consists of an insertion of the kanamycin resistance cassette (KanMX4) into the last gene on chromosome 1, YAL068C, which encodes PAU8, a member of a multigene family. Deletion of this gene was expected not to influence thiol synthesis and was used to determine if the KanMX4-gene-deletion-cassette itself was causing a change in thiol synthesis. As shown in Fig. 1a, yal068c displayed no significant difference in 3MH and 3MHA compared to the wild-type BY4743 strain.
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-012-4198-6/MediaObjects/253_2012_4198_Fig1_HTML.gif
Fig. 1

Volatile thiol production of 12 (a) and 40 (b) different single-gene knockouts in the BY4743 yeast background after fermentation in optimized grape juice (2005). Black dashed lines indicate volatile thiol levels of yeast controls BY4743-wt and BY4743-yal068c; bolded strains exhibited changes of biological relevance (d ≥ |4|); error bars = SEM; *p < 0.01, **p < 0.001; numbers above bars = effect size d; an = 3–5; bn = 2, optimized juice supplemented with 7 mg/L cysteine. c Sulfur metabolism of S. cerevisiae. Tested single-gene knockouts are in bold. Plus sign single-gene knockouts causing biologically relevant increases in volatile thiol(s); minus sign single-gene knockouts causing biologically relevant decreases in volatile thiol(s); c is based on Ganguli et al. 2007; Kumar et al. 2003; Melcher et al. 1992; Momose et al. 2001; Ono et al. 1999; Spiropoulos et al. 2000; Tehlivets et al. 2004; Wang et al. 2003

No reduction in 3MH and 3MHA was observed for any of the other deletion strains. However, increases in both thiols were seen in gly1 and cyc3 and most notably cys3 and cys4 mutants, whereas the deletion of CYT2 caused an increase of 3MH only. Although all these increases were statistically significant in the experiment shown (Fig. 1a, p < 0.01), the biological relevance of most of these changes was small. Throughout this work, the criterion was applied that only changes with an effect size of d ≥ |4| were considered to be important. Therefore, only the elevated levels produced by cys3 (3MH) and cys4 (3MH and 3MHA), which in both cases more than doubled the volatile thiol production, were regarded as biologically relevant in this experiment.

A second screen of 40 additional candidate genes was undertaken (Fig. 1b). The largest change in thiol production was an increase by the met17 deletion mutant. Smaller increases were seen for aat2, bio3, car2 mutants, while shm2 and ser1 strains showed decreases in one or both thiols. It is striking that the three most biologically relevant changes came from deletion of MET17, CYS3, and CYS4, which encode enzymes that perform three consecutive steps in the sulfur amino acid biosynthesis pathway. Collectively they convert hydrogen sulfide into cysteine (see Fig. 1c).

The two deleted genes causing major decreases in thiol synthesis are both involved in serine biosynthesis but play opposite roles; SHM2 reversibly converts serine to glycine (Fig. 1c), whereas SER1 is part of an alternative serine biosynthesis pathway (Melcher et al. 1992). In the sulfur amino acid pathway, serine is required by CYS4 to form cystathionine and ultimately cysteine (see Fig. 1c).

Within one set of experiments, the thiol assay typically gave relative standard deviations (RSD) of less than 10 %. If larger RSDs were encountered, up to five thiol extractions per strain were performed, to decrease the risk of analytical artefacts. However, the absolute concentration of thiols produced and the 3MH–3MHA conversion rate varied significantly between some experiments fermented in subsamples from the same batch of juice using the same strain (e.g., compare BY4743 of Fig. 1a, b). Part of this between-ferment variability may be because thiol extractions and GC-MS analysis for both screens were performed about six months apart using different calibration curves. Differences between juice samples from the same vineyard could also have contributed to the observed discrepancy, since sub-blocks of the same vineyard can vary in volatile thiol yield (Lee et al. 2008). In any case, an internal control strain (BY4743) and repeated screens of the key deletion strains in each different grape must (data not shown) were used to validate the results.

Some NCR-related and GSH-degradation genes influence volatile thiol production

Twelve strains deleted for genes involved in the NCR regulation and GSH degradation were tested for their effect on volatile thiol production. Nine mutants caused changes, mainly decreases in 3MHA (Fig. 2). The only biologically relevant change for 3MH was a decrease in the mks1 mutant. BY4743-gcn4 was the only strain, which caused a biologically relevant increase in 3MHA. Decreases in 3MHA with d ≥ |4| could be observed for six strains: gln3, car1, tor1, mks1, and most notably for ure2 and npr1. The role, interaction, and thiol phenotype of all NCR-related genes tested are illustrated in Fig. 2b. No clear pattern could be observed between the changes of the thiol concentrations and the reported effect of deletion mutants on NCR in the laboratory yeast BY4743. For example, the main transcriptional regulator of NCR, URE2, abolishes NCR when deleted and genes under the control of NCR are expressed constitutively (Courchesne et al. 1988). Conversely, the deletion of GLN3, which is normally inactivated by Ure2p protein in the presence of optimal nitrogen sources, is unable to upregulate NCR-regulated genes (Courchesne et al. 1988). However, in this study both mutants showed the same effect—a reduction in 3MHA. Of mutants in the alternative GSH degradation pathway, only BY4743-dug2 caused a biologically relevant change, a decrease in 3MHA.
https://static-content.springer.com/image/art%3A10.1007%2Fs00253-012-4198-6/MediaObjects/253_2012_4198_Fig2_HTML.gif
Fig. 2

a Volatile thiol synthesis of BY4743 strains deleted in key genes of the NCR and GSH degradation pathway after fermentation in optimized grape juice (2006). Black dashed lines indicate volatile thiols levels of BY4743-wt; bold indicates strains showing biologically relevant changes (d ≥ |4|); n = 5; error bars = SEM; *p < 0.01, **p < 0.001; numbers above bars = effect size d. b Nitrogen catabolite repression (NCR) in S. cerevisiae. Tested single-gene knockouts in laboratory yeast BY4743 are in bold. Plus sign single-gene knockouts causing biologically relevant increases in volatile thiol(s); minus sign single-gene knockouts causing biologically relevant decreases in volatile thiol(s); b is based on Cardenas et al. 1999; Cooper et al. 2002; Crespo et al. 2004; Edskes et al. 1999; Feller et al. 2006; Sosa et al. 2003

Deletion of candidate genes in a wine yeast background also affects thiol production

To verify the results obtained in the laboratory yeast, eight candidate genes were chosen, deleted in a wine yeast background, and were tested for their impact on volatile thiol release during fermentation in SB grape juice. This confirmation was undertaken for two reasons. Firstly, data obtained in a laboratory background do not necessarily translate to data generated in wine yeast. Large genetic differences between laboratory yeast and wine yeast have recently been reported (Bornemann et al. 2011; Liti et al. 2009; Schacherer et al. 2009). Secondly, the optimized grape juice employed for the laboratory strain, with its lower sugar content and increased yeast available nitrogen (YAN) values, resulting from the amino acid supplementation, could potentially alter or disguise changes in thiol release caused by deleted genes. Confirming the results in full grape juice would demonstrate that the data are enologically relevant.

Four genes involved in sulfur metabolism and three NCR-related genes, which had shown the largest biological effects in the laboratory strain, were deleted in the wine yeast F15-h, along with the IRC7 gene as a control. Table 3 compares the differences in thiol release for all eight deletion knockouts between the laboratory yeast BY4743 and the wine yeast F15-h. Seven of the eight mutants in the wine yeast showed significant differences in thiol production compared to the wild-type strain, confirming the involvement of these genes in modifying this pathway. The exception was the irc7 mutant, which was unaffected in both genetic backgrounds. In general, deletion of the sulfur-related genes showed similar effects in F15-h to those initially obtained in BY4743, with cys3, cys4, and met17 mutants showing elevated total thiol production and the shm2 mutant showing reduced concentrations of thiols. In contrast, the three NCR-related genes all showed strain differences in the mutant phenotypes. The F15-h mutants in general showed changes in both 3MH and 3MHA production, in contrast to the BY mutants which were affected mostly in 3MHA. Additionally, in four of the six comparisons, the changes in production of the two thiols were in the opposite direction in the different genetic backgrounds: for 3MH (npr1, gln3) and 3MHA (ure2, npr1).
Table 3

Comparison of volatile thiol release (3MH + 3MHA) of single-gene deletions knocked out in laboratory yeast BY4743 and wine yeast F15-h

 

% Change to wt (3MH)

% Change to wt (3MHA)

Strain

Pathway

BY4743

 

F15-ha

BY4743

 

F15-ha

cys3

Sulfur

+95b**

+69**

+123b**

+56**

cys4

Sulfur

+104b**

+51**

+217b**

+46**

met17

Sulfur

+57b**

+26**

−1b

+15*

shm2

Sulfur

−32b**

−8*

−77b**

−32**

irc7

Not known

−15c

−3

−6c

−6

gln3

NCR

−6c

−29**

−26c**

−24**

npr1

NCR

+4c

+33**

−65c**

+35**

ure2

NCR

+9c*

+57**

−33c**

+81**

The approximately equal sign denotes a similar trend, the not equal sign denotes a different trend, when comparing the effect of the same deletion in BY4743 and F15-h yeast backgrounds

YAN yeast available nitrogen

*p < 0.01, d ≤ |1.9|; **p < 0.001, d > |4|, significantly different from the corresponding wild-type strain

aUndiluted juice (2006), YAN = 219

bOptimized juice (2005), YAN = 441

cOptimized juice (2006), YAN = 305

Supplementing grape juice with sulfur and nitrogen compounds alters volatile thiol release

To study the impact of additional sulfur and nitrogen on the release of volatile thiols during fermentation of SB grape juice, compounds that could act as potential sources of sulfur or nitrogen were added to juice. Three potential sulfur sources (each added to approximately 1.5 mM) were used: cysteine, GSH, and S-ethyl-L-cysteine (SEC). The former two are found in grape juice while the latter was chosen because of its resemblance to the cysteinylated precursor of 3MH. Two nitrogen sources, di-ammonium phosphate (DAP) and urea, were added to provide additional 200 mg/L nitrogen to the SB grape juice, which resulted in a total YAN of 394 mg N/L in the juice. Both are good nitrogen sources for yeast, but DAP is known to induce NCR while urea does not (Magasanik and Kaiser 2002). Note that the three sulfur compounds added also contain nitrogen: the amount used corresponds to an additional 21 and 23 mg/L (for SEC and SEC) or 63 mg/L (for GSH) of nitrogen. GSH can be used as a nitrogen source by yeast (Penninckx 2000), cysteine cannot (Chen and Kaiser 2002), and no data is available for SEC, although it is known to be taken up by yeast (Maw 1963) and to be toxic in some media (Maw 1961).

Table 4 summarizes thiol production following addition of these five compounds to grape juice. Four different strains showed a moderate impact of added cysteine on thiol production, but directions of the changes were different. The laboratory strain BY4743 and the most widely used commercial strain EC1118 showed decreases in both 3MH and 3MHA production when fermented in optimized grape juice (d values ranging from −4.8 to −12.7). In contrast, two other derivatives of commercial strains, F15-h and GRE-h, showed increases in 3MH production with added cysteine in undiluted grape juice (d = +12.5 and +29), but with no biologically relevant effect on 3MHA. Table 4 also summarizes the impact of four added compounds on volatile thiol release in full SB grape juice (2006) inoculated with commercially sold wine yeast F15. Both ammonia and urea caused increases in 3MH and 3MHA in the F15 strain (d = 9–18.5), as did supplementation with 1.5 mM SEC (d ≈ 7.5 for both thiols). GSH had no biologically relevant effect on thiol levels in the wines. Vmax values were highly increased in ferments supplemented with DAP and urea (d = 14.2 and 17.1, respectively), whereas the addition of GSH and SEC did not influence fermentation kinetics compared to the control fermentation (data not shown). This increase in fermentation rate indicates that the nitrogen status of this grape juice (initial YAN = 194 mg/L) was limiting for the growth of the F15 strain.
Table 4

Impact of additional nitrogen and sulfur sources on volatile thiol release

Compound

Strain

Concentration added

Extra N

Extra S

3MH

3MHA

mg/L

mg/L

mg/L

% Change vs. no addition

Cysteine

BY4743a (his3, leu2, ura3)

200

23

53

−34**

-41**

EC1118a (commercial)

200

23

53

−33**

-43**

F15b (ho::hphMX, ura3)

200

23

53

+33**

+10**

GREb (ho::hphMX, ura3)

200

23

53

+36**

+3

SEC

F15c (commercial)

223

21

48

+59**

+81**

GSH

F15c (commercial)

460

63

48

−8

+14

DAP

F15c (commercial)

943

200

+45**

+106**

Urea

F15c (commercial)

429

200

+45**

+109**

The results are shown for three separate experiments, using 2006 Sauvignon Blanc juice

**p < 0.001

aOptimized juice–diluted juice with elevated levels of histidine, leucine, and uracil

bUndiluted grape juice with elevated levels of uracil

cUndiluted grape juice

These results confirm the relevance of sulfur and nitrogen compounds to thiol production and also confirm that there are differences between yeast strains. The similar increases in thiol concentrations when urea and ammonium was added to SB must suggest that NCR is of only limited importance for the thiol response under enological conditions.

Volatile thiol release in yeast overexpressed in five candidate genes

To further elucidate the influence of certain candidate genes on volatile thiol release, a subset of five genes—CYS3, CYS4, GLN3, IRC7, and SHM2—were overexpressed in both the laboratory yeast BY4743 and the wine yeast F15-h. After fermentation of the BY4743 strains in optimized grape juice and the F15 strains in undiluted grape juice, none of the overexpressed genes showed any biologically relevant differences in total thiol (3MH + 3MHA) production compared to their corresponding wild type strain (data not shown). The only phenotypic difference observed was in strains carrying the overexpression cassette oxGLN3, which caused a longer lag phase accompanied by a decrease in 3MHA (d = −5.5 BY4743, d = −13.5 F15). BY4743-wt, which was fermented in grape juice diluted with 25 % SGM without sugars, released 28 % less thiols (3MH + 3MHA) than its wine yeast counterpart, indicating that the precursor thiol conversion rate at 25 °C is similar in the two strains.

The lack of any phenotype for overexpression of the deleted form of IRC7 contrasts with published data in which a full-length copy of the gene was expressed in the same F15 background (Roncoroni et al. 2011). The difference provides experimental evidence that the deleted version, which lacks the C-terminal 60 amino acids of the mature protein, is inactive for thiol production.

Discussion

This work has identified 17 genes affecting volatile thiol release when deleted in a laboratory yeast background. These genes can be divided into two groups: (a) nine genes either directly or indirectly related to the sulfur amino acid pathway; (b) eight genes related to NCR or other amino acid pathways (summarized in Table 5). Supplementation of the grape must with additional sulfur- and nitrogen-containing compounds also modified thiol yields from SB grape must. There were significant differences between a laboratory and wine yeast, both in their deletion phenotypes of the NCR-related genes and in their responses to fermentation additions.
Table 5

Single-gene deletions showing changes of biological relevance (d ≥ |4|) in the laboratory yeast background

Deletion

Pathway

3MH

3MHA

cys3

Sulfur

+

0

cys4

Sulfur

+

+

met17

Sulfur

+

0

met16

Sulfur

0

dug2

Sulfur, GSH

0

shm2

Sulfur-related, Ser, Gly

ser1

Sulfur-related, Ser, Gly

0

aat2

Sulfur-related, Asp, Thr

+

0

bio3

Sulfur-related, biotin

+

0

car1

Arginine degradation

0

car2

Arginine degradation

+

0

mks1

NCR

gln3

NCR

0

gcn4

NCR

0

+

npr1

NCR, TOR

0

tor1

NCR, TOR

0

ure2

NCR

0

Plus signs denote an increase, minus signs denote a decrease and zeros denote no change

The most notable finding of this work was that increases of up to twofold in thiol production were found in laboratory and wine yeast strains defective for genes encoding three successive steps in sulfur amino acid biosynthesis: MET17, CYS4, and CYS3. These genes encode enzymes, which are responsible for the incorporation of inorganic H2S into O-acetyl-homoserine to form homocysteine, its conversion to cystathionine and then to cysteine (Fig. 1c). Mutants in other steps in the sulfur pathway also altered thiol yields but to a lesser extent. The addition of cysteine or a cysteine conjugate, SEC, to grape juice increased thiol production in undiluted grape juice, whereas supplementing grape juice with extra GSH had no impact on the thiol phenotype. Previous studies adding GSH have shown both small decreases or increases in thiol yields (Patel et al. 2010; Roland et al. 2010b). Collectively, these results clearly indicate the importance of the sulfur amino acid pathway for thiol production.

We suggest two general mechanisms for the interaction between the perturbation of the sulfur pathway and volatile thiol production. The interaction may be indirect, via changes to the regulation of steps in the thiol pathway(s). For example, the supply of cysteine is known to regulate enzymes in the sulfur biosynthesis pathway. Deletion of the essential genes responsible for cysteine synthesis in yeast, CYS3 and CYS4, led to large changes of this regulation (Hansen and Johannesen 2000). It is possible that such regulation could extend to transporters or enzymes involved in thiol production, such as the high-affinity GSH transporter, OPT1, which is known to be induced by sulfur starvation and repressed by cysteine through the 5´-CCG binding motif (Hiraishi et al. 2008; Wiles et al. 2006). Likewise, altering the concentration of sulfur compounds in juice could also alter regulation of the thiol pathway.

Alternatively, the sulfur pathway effects observed here could be more direct by changing metabolite flows into thiol production. For example, H2S and cysteine have been proposed as substrates that react with E-2-hexenal to produce 3MH (Schneider et al. 2006). Mutants in MET17 and CYS4 are both known to increase H2S production in the laboratory strain of yeast (Linderholm et al. 2008), so that the increases in thiols seen in these mutants might be the result of increased activity of this pathway for thiol production. A better knowledge of the pathways of thiol formation, and of their relative contributions to thiol production in the fermentation of grape juice, is needed to distinguish these alternatives.

The increased 3MH/A production in SB wine fermented with CYS3-deleted yeast strains in our work differed with results obtained by Thibon et al. (2008), who did not detect any change in synthetic medium spiked with Cys-3MH after fermentation with a CYS3-deleted wine yeast strain. This discrepancy suggests that the cys3-related thiol increase described here is not due to an increased Cys-3MH thiol conversion, but is more likely due to one or both mechanisms outlined above.

Our results also found a connection between thiol production and nitrogen metabolism in yeast. Mutation of genes involved in NCR regulation altered thiol production which extends previous observations along similar lines (Subileau et al. 2008b; Thibon et al. 2008), and we demonstrated for the first time that TOR-related genes can also affect thiol production. Additions of nitrogen to juice also altered thiol production, confirming the results of Subileau et al. (2008b). The nitrogen effect may be occurring via either of the two general mechanisms outline above. Indeed, it remains possible that these nitrogen effects are entirely mediated via the sulfur pathway, whose regulation is closely linked to nitrogen supply (Fig. 1c).

Whilst a connection to nitrogen metabolism was clear in our results, a direct connection of thiols to NCR was not. For example, mutations in GLN3 and URE2 are expected to have opposite effects on NCR, but both mutants showed a decrease in 3MHA in the BY4743 laboratory strain. Moreover, in contrast to Subileau et al. (2008b), we found that both ammonia (which induces NCR) and urea (which does not) increased thiol production substantially in commercial wine yeast F15. The stimulating effect of DAP on 3MH release in Riesling juice spiked with Cys-3MH recently reported by Winter et al. (2011) supported our findings. These differences suggest that the changes in thiol production under enological conditions are only marginally influenced by NCR. Even if the NCR effect is dependent on the activity of a full-length IRC7 as suggested by Thibon et al. (2008), the results presented here suggest that the remaining 3MH/A production is not under the control of NCR in the laboratory strain and only so to a limited extent in the wine strain F15.

We cannot totally exclude the possibility that the increased fermentation rate following addition of both urea and ammonia, and the decreased rate of fermentation in the shm2, ser1, and gln3 mutants (data not shown), contributed to the change in thiol yields. However, we note that some slower fermenting deletion mutants (such as aat2) did not show altered thiol yields and that there was no general correlation between fermentation rate in the various deletion mutants and thiol production. Furthermore, the addition of serine to the must restored the fermentation rate of ser1 almost to that of the wild-type strain but caused no major change in thiol phenotype (data not shown).

We noted larger strain differences for mutants affecting nitrogen metabolism than those affecting sulfur metabolism (Table 3). A high degree of between-strain variation for nitrogen has been noted previously for developmental processes (e.g., Strudwick et al. 2010), as well as for NCR (Deed et al. 2011; Magasanik and Kaiser 2002). In addition, there is clear evidence that wine yeasts differ hugely in both their requirements for nitrogen and in their response to added nitrogen (e.g., Taillandier et al. 2007). The effects of nitrogen on thiols for the wine strain used here differed from data obtained in other laboratories with different wine yeasts (Subileau et al. 2008b; Thibon et al. 2008), as well as data in this laboratory from wine yeast M2, which were largely unaffected (Deed et al. 2011). We therefore suggest that the different nitrogen-regulated thiol phenotypes between strains are likely the result of genetic differences between the laboratory strain and the wine strains (Liti et al. 2009; Schacherer et al. 2009). However, it remains possible that optimized grape juice, with its higher amino acid content (YAN +111), may have masked some of the effects of the NCR-related gene deletions in the laboratory yeast background.

Our screen of candidate deletion mutants was aimed at identifying genes affecting thiol yield, and in particular genes encoding β-lyase enzymes. Data from Roncoroni et al. (2011), combined with the overexpression results here, suggest that all three of the yeast strains used lack an active IRC7 gene. The results of Thibon et al. (2008) had suggested that there is more than one β-lyase involved in 3MH production, since more than half of the cys-3MH was still converted to 3MH in an IRC7-deleted strain. It was therefore unfortunate that our approach did not identify any alternative β-lyase. However, we cannot exclude the possibility that other β-lyase genes do exist, which we were not able to capture in this screen of 67 candidate genes. For example, STR3 encodes a cystathionine β-lyase that was recently shown to affect thiol production (Holt et al. 2011), but this deletion is not present in the BY4743 collection. In addition, it remains possible that in the juice we used, the cysteine and glutathione precursors make only a minor contribution to total thiol production and that either the pathway proposed by Schneider et al. (2006) is used or some other alternative pathway. Indeed, recent studies showed that the 3MH precursors, Cys-3MH and recently discovered G-3MH, can explain no more than 10 % of the total 3MH/A production in French SB wine (Roland et al. 2010a; Subileau et al. 2008a).

Our use of grape juice for the deletion strain screening had the advantage that the effects observed are much more likely to be relevant to real winemaking. This is important when researching secondary aroma compounds, which are formed by yeast from precursors present in the grape must. Possible interactions of aroma precursors with other compounds in the complex grape must matrix might have important impacts on aroma production. Furthermore, additional unknown precursors cannot be discovered by relying solely on synthetic medium.

Despite recent progress in the area of varietal thiol research (see reviews by Coetzee et al. 2012; Peña-Gallego et al. 2012; Roland et al. 2011), there is a critical need for identification of additional thiol precursors, along with a more detailed understanding of the biochemical pathways involved in the formation of varietal thiols from these precursors during fermentation. The data presented here clearly implicate the sulfur amino acid pathway as being important for the yield of varietal thiols in SB. The results presented here also reveal opportunities for regulating thiol production in a commercial winery, both by applying genetic changes to the yeast strain, and possibly by interventions in the vineyard or winery via additions to the grape juice prior to, or during fermentation (e.g., see Winter et al. 2011).

Acknowledgments

This work was funded by a grant from the Foundation of Research Science and Technology in New Zealand (contract UOAX0404) with the support of New Zealand Winegrowers. We are grateful to Andy Frost (Pernod Ricard NZ, Blenheim) for supplying grape juice and Laura Nicolau (Wine Science, University of Auckland) for access to and assistance with GC-MS analysis.

Supplementary material

253_2012_4198_MOESM1_ESM.pdf (371 kb)
ESM 1(PDF 370 kb)

Copyright information

© Springer-Verlag 2012