OUP user menu

Nitrogen catabolic repression controls the release of volatile thiols by Saccharomyces cerevisiae during wine fermentation

Cécile Thibon, Philippe Marullo, Olivier Claisse, Christophe Cullin, Denis Dubourdieu, Takatoshi Tominaga
DOI: http://dx.doi.org/10.1111/j.1567-1364.2008.00381.x 1076-1086 First published online: 1 November 2008


Volatile thiols such as 4-methyl-4-sulfanylpentan-2-one (4MSP) and 3-sulfanylhexan-1-ol (3SH) are aromatic molecules having an important organoleptic impact on white wines. These components are produced from inodorous nonvolatile cysteinylated precursors by Saccharomyces cerevisiae metabolic activity during alcoholic fermentation. Here we provide a new insight into the genetic determinism of the production of volatile thiols by yeast. Using a gene deletion approach, we investigated the role of three yeast β-lyases and demonstrate that Irc7p, a putative cystathionine β-lyase, is one of the main proteins catalyzing the 4MSP and 3SH release under enological conditions. Moreover, we demonstrate that Ure2p/Gln3p proteins mainly control the bioconversion of volatile thiols by the transcriptional regulation of the IRC7 gene through the general mechanism of nitrogen catabolic repression. Finally, our findings suggest that the enantiomer balance of 3SH may be modulated by activating specifically stereoselective enzymes such as Irc7p.

  • URE2
  • IRC7
  • wine yeast
  • S-cysteine conjugate
  • aroma release
  • nitrogen catabolic repression


Some volatile thiols play a decisive role in contributing to the aroma of wines made from the Sauvignon blanc grape variety (Tominaga, 1998b). Their aromatic contribution was also successively demonstrated in several other grape varieties such as Gewürztraminer, botrytized Sémillon (Sarrazin et al., 2007), Colombard, Petit Manseng (Tominaga, 2000a), Merlot (Bouchilloux et al., 1998; Aznar et al., 2001) and Cabernet Sauvignon (Blanchard et al., 2004). Among these compounds, 4-methyl-4-sulfanylpentan-2-one (4MSP) (Darriet et al., 1995; Tominaga, 1998c) and 3-sulfanyhexan-1-ol (3SH) (Tominaga, 1998c) are principally formed during alcoholic fermentation from specific S-cysteine conjugate nonvolatile precursors present in the must (Tominaga et al., 1995, 1998a). The transformation of such precursors into volatile thiols is achieved during the course of yeast metabolism and involves a carbon–sulfur β-lyase activity (Tominaga, 1998c). The 3-sulfanylhexyl acetate (3SHA) resulting from the acetylation of 3SH by yeast also contributes to wine aroma (Tominaga et al., 1996).

4MSP, 3SHA and 3SH are characterized by fruity descriptors, i.e. box tree, passion fruit and grapefruit, and present very low perception thresholds, i.e. 0.8, 4.2 and 60 ng L−1, respectively (Tominaga, 1998a). The powerful aromatic properties of these molecules have prompted recent studies to investigate how their production can be enhanced and modulated in winemaking. Viticultural practices such as nitrogen feeds (Choné et al., 2006) and prefermentation operations such as skin contact (Peyrot des Gachons et al., 2002) modulate the amount of aromatic precursors in grape must. Fermentation conditions such as temperature (Howell et al., 2004; Masneuf-Pomarède et al., 2006) also influence the final concentration of these aromas in wine. However, the most important factor in the liberation of thiols is undoubtedly the yeast strain (Dubourdieu et al., 2006). Indeed, a strong variability in the liberation of thiols (about 10-fold) was found among Saccharomyces cerevisiae strains (Murat, 2001a; Howell et al., 2004; Masneuf-Pomarède et al., 2006), Saccharomyces uvarum strains and related interspecific hybrids (Masneuf et al., 2002). This variability incited strain screening and breeding programs to obtain new strains characterized by the liberation of higher thiols (Murat, 2001a; Masneuf et al., 2002). However, the genetic determinism and molecular mechanisms of thiols liberation by yeast remain unclear.

Recently, Howell (2005) identified three genes, BNA3, CYS3 and IRC7, that encode putative β-lyase in the S. cerevisiae genome. Using gene deletion experiments, they suggested that these enzymes are able to cleave the 4MSP-precursor (P-4MSP) into the related volatile aroma during fermentation in synthetic grape juice. Although these enzymes have been identified, the mechanisms of precursor assimilation and cleavage by yeast during fermentation are far from being understood. Understanding the regulation of the release of volatile thiols during alcoholic fermentation motivated the present study.

A systematic screening of the physicochemical factors influencing the release of volatile thiols shed light on assimilable nitrogen, which was responsible for 35% of the observed variation in the release of volatile thiols (Thibon et al., 2006). However, the pleiotropic effect of nitrogen in yeast metabolism during fermentation (Bely et al., 1990; Julien et al., 2000) prevents us from assessing the eventual regulation effect of nitrogen in the release of volatile thiols. We favor the hypothesis that volatile thiols release might in part be controlled by nitrogen catabolic repression (NCR). This physiological response involves the transcription factors Gln3p and Gat1p (ter Schure et al., 2000; Scherens et al., 2006), which are inactivated by Ure2p protein in the presence of nitrogen-rich sources. Besides experimental data, our hypothesis was based on the fact that many yeast catabolic enzymes, having a putative role in precursor cleavage, might be controlled by nitrogen sensing. For example, BNA3 or IRC7 expression was increased following rapamycin treatment (Hardwick et al., 1999; Huang et al., 2004) and in a URE2-deleted background (Scherens et al., 2006). Moreover, numerous nitrogen transporters have been described to be transcriptionally repressed by NCR (Cooper & Sumrada et al., 1983). As their structure is similar to that of amino acids, uptake of aroma precursors from must into yeast cytoplasm might be driven by amino acid transporters. Finally, during wine fermentation, NCR seems to be effective (Rossignol et al., 2003; Beltran et al., 2004) and can be silenced by deleting the URE2 gene (Salmon & Barre et al., 1998). Such a deletion was described to improve the nitrogen uptake, leading to an improvement in fermentation kinetics under enological conditions.

In the present study, we investigated the role of NCR in the release of volatile thiols in an enological context. On combining genetic and analytical chemistry approaches, we observed how this regulation pathway influences precursor transport and cleavage, shedding light on the most important actors involved in the release of volatile thiols.

Materials and methods

Yeast strain culture and construction

Yeast cells were grown at 30 °C on a complete 1% yeast extract, 1% peptone, 2% dextrose medium (YPD) solidified with 2% agar if necessary. G418 sulfate (Geneticin) (100 μg mL−1) was added to YPD for selection. Sporulation was induced on acetate medium (1% potassium acetate, 2% agar) for 3 days at 24 °C. The VL3-1D S. cerevisiae strain used was derived from a commercial starter VL3c (Laffort Enologie, Bordeaux, France) described previously for its strong capacity to reveal volatile thiols (Murat, 2001a; Howell et al., 2004, 2005; Masneuf-Pomarède et al., 2006). VL3-1D, a homothallic diploid spore clone (HO/HO), was obtained by tetrad microdissection, and was assumed to be totally homozygous (Mortimer et al., 1994). All the mutant strains were constructed from the VL3-1D strain using a PCR deletion strategy (Baudin et al., 1993; Wach et al., 1994). Deletion cassettes were obtained by PCR amplification of the genomic DNA of appropriate deleted strains (BY4742 background) from the Euroscarf collection (Frankfurt, Germany). The deletion cassettes were amplified using the following primers: ure2::KanMx4: p59: CAAATGCCGAGAAAAATACCGC and p60: AAACGAACGCCGAAACACATA; gln3::KanMx4: p318: CAGCTTTTCAACCTTCCATTC and p319: GATAGACAAAAGAAACAGGGGAG; irc7::KanMx4: p314: TGATAACGATTTTATTGTCGCCTC and p315: TGATACAGCTAGAAAATTGAACCA; cys3::KanMx4: p316: GACCCCATACCACTTCTTTTTGTT and p317: TGATCTCGTTCTAGTTCTGGAAGC; bna3::KanMx4: p1: GAGCAGATTGTTTTGAGTAGG and p2: TTCCTAAGCAACTCATCGTG. Cells were chemically transformed (Puig et al., 1998), positively selected on YPD-G418, sporulated and microdissected in order to obtain fully homozygous spore clones carrying two deleted gene copies. For each construction, the correct deletion was verified by PCR using appropriate primers. All the strains are described in Table 1. The Δirc7Δure2 double mutant was obtained by spore-to-spore mating and segregation analysis using conventional yeast genetic methods (Wickner et al., 1991).

View this table:

Yeast strains

Δure2HO/HO; ure2::KanMx4/ure2::KanMx4
Δgln3HO/HO, gln3::KanMx4/gln3::KanMx4
Δcys3HO/HO; cys3::KanMx4/cys3::KanMx4
Δbna3HO/HO; bna3::KanMx4/bna3::KanMx4
Δirc7HO/HO; irc7::KanMx4/irc7::KanMx4
Δirc7Δure2HO/HO; ure2::KanMx4/ure2::KanMx4; irc7::KanMx4/irc7::KanMx4

Precursor synthesis

S-4-(4-methylpentan-2-one)-l-cysteine (P-4MSP) was synthesized by addition of mesityl oxide (22 mmol) in l-cysteine (20 mmol) and purified according to Shinkaruk (2008). S-3-(hexan-1-ol)-l-cysteine (P-3SH) was synthesized according to Thibon (2008) as follows: trans-2-hexen-1-al (5 mmol) was added to a stirred solution of N-(tert-butoxycarbonyl)-l-cysteine (4.5 mmol) and cesium carbonate (2.25 mmol) in acetonitrile and incubated overnight. After elimination of acetonitrile by rotary evaporation, sodium borohydride (2.5 mmol) in 2 mL methanol was added drop wise and stirred overnight. The deprotection of 0.5 mL of N-(tert-butoxycarbonyl)-S-3-(hexan-1-ol)-l-cysteine was carried out by the addition of HCl 10 M (10 mmol). The crude product was concentrated and purified on an ion exchange column (DOWEX 50 WX8).

Precursor assay

The cysteine-S-conjugates were purified from the must using the method described by Murat (2001b), modified as described by Thibon (2008). The pH of the Chelating Sepharose column, containing immobilized copper (Porath et al., 1975), was adjusted to 10 by percolation of 3 mL sodium carbonate buffer (Na2CO3, 50 mM, pH 10). Samples (500 μL) were brought to pH 10 and directly loaded onto the column, which was further washed with 2 mL potassium phosphate buffer (50 mM, pH 7) and eluted by percolation with 3 mL hydrochloric acid solution (25 mM). The eluate containing cysteine-S-conjugates was evaporated to dryness and dispersed in a 500-μL volume of an ethanol/acetone mix (50 : 50). The soluble fraction was again evaporated dry under vacuum in a 2-mL vial.

The dry residue in a vial capped under a nitrogen stream was resuspended in heptafluorobutyric anhydride (HFBA) (40 μL) and heptafluorobutanol (HFOH) (10 μL) and the mixture was incubated for 15 min at 70 °C. After cooling, excess HFBA and HFOH was eliminated by evaporation at room temperature using a nitrogen stream for 10 min. The dry residue obtained was resuspended in 50 μL of anhydrous ethyl acetate and a 2-μL aliquot was analyzed by GC-MS.

Volatile thiol assay

Volatile thiols were extracted at the end of the fermentation process from 250 mL synthetic medium using the method of Tominaga (1998b) as modified by Tominaga (2000b). Amounts of 4MSP, 3SH and 3SHA were quantified by GC-MS according to the methods described by Tominaga (1998b). The separation of two 3SH enantiomers [(R)/(S)-3SH] was carried out by injecting the extracted volatile thiols into a Lipodex C (50 m × 0.25 mm) chiral column (Interchim, Montluçon, France). The chromatography conditions were identical to those described by Tominaga (2006).

Thiol production assay in a fermented synthetic medium

The release of volatile thiols was measured at the end of the fermentation process, which was carried out in 350-mL bottles at 24 °C in synthetic grape juice described previously as KP medium (Marullo et al., 2006). This medium was buffered to pH 3.3 and contained 80 g L−1 glucose, 80 g L−1 fructose and 190 mg L−1 available nitrogen. Aroma precursors (P-4MSP and P-3SH) were added to the medium before yeast inoculation at different concentrations indicated in the text. Yeast strains [wild-type (WT) or mutants] were inoculated at 1 × 106 cells mL−1 from an overnight preculture. The CO2 release was measured by estimation of weight loss after complete fermentation. All experiments were carried out in triplicate and independently repeated two or three times.

Precursors and thiols monitoring during alcoholic fermentation

Precursor uptake and volatile thiol production during the fermentation were monitored using 1-L bioreactors in the same synthetic juice. This apparatus (described previously by Marullo et al., 2006) allows calculation of the CO2 production rate dCO2/dt by monitoring weight loss every 20 min. Culture growth and cell numbers were monitored using an electronic particle counter (ZII; Coulter-Counter Coultronics, Margency, France) and analyzed using accucomp® software (Coulter-Counter Coultronics). To compare ure2 and WT strains under the same physiological conditions, we withdrew must samples at points in the reaction advancement (expressed as percentage of total CO2 release). Precursors (P-3SH and P-4MSP) were measured from 0.5 mL of fermenting medium sample collected when 0, 5, 10, 20, 30, 40, 50, 60, 75 and 100% of the total CO2 had been released. P-3SH and P-4MSP were added at 15 μM. For analytical convenience and only for this assay, the P-4MSP concentration was 750-fold higher than that in grape juice. The volatile thiols [(R)/(S)-3SH and 4MSP] were measured for 0%, 10%, 20%, 30%, 40%, 60% and 100% of the total CO2 released. Data were expressed in specific activity taking into account the cell number for each strain tested. Both measures were carried out in duplicate on two independent preparations.

RNA extraction and cDNA synthesis

Biomass samples for expression assays were collected for 10%, 20%, 40% and 60% of the total CO2 released. Ten-milliliter samples were centrifuged (5 min, 10 000 g) to pellet the cells. Lysis treatment was performed using a Fastprep FP120 (MP Biomedicals, Ohio) apparatus with the following parameters: 45 s at 65 m s−1 applied twice. The two cycles were separated by a 5 min incubation on ice. DNA contamination was treated using a DNA-free Kit (Ambion Inc., Austin, TX), according to the manufacturer's instructions. The absence of contaminating DNA was checked by PCR directly on the RNAs. RNAs were retrotranscribed into cDNAs using the iScriptTM cDNA Synthesis Kit (Bio-Rad, Hercules, CA), according to the manufacturer's instructions. For both strains (Δure2 and WT), extractions were realized from three independent batches.

IRC7 expression analysis by real-time PCR

Quantitative real-time PCR with SYBR-Green I was performed using the iCycler iQ real-time PCR Detection System (Bio-Rad). Primers p325, TCAGCTTCTGGGCTTGGTTCT, and p326, TCAACACCGAACTTGGCCAAT, and p323, TACCGGCCAAATCGATTCTC, and p324, CACTGGTATTGTTTTGGATACC, were used to amplify 137 bp of IRC7 (interest gene) and 124 bp of ACT1 (housekeeping gene), respectively. The use of ACT1 as a housekeeping gene has been described routinely in the literature (Giulietti et al., 2001; Bleve et al., 2003) as a quantification marker for many yeast species through quantitative real-time PCR. Real-time quantitative PCRs (qPCRs) were carried out using the iQ SYBR Green Super Mix (Bio-Rad). Primers were added at a concentration of 0.3 mM each. The PCR program was as follows: 3 min at 95 °C for initial denaturation, then 40 cycles of 30 s at 95 °C, 30 s at 58 °C and 30 s at 72 °C. A final melt curve was carried out by 51 cycles of 10 s starting at 65 °C, with increasing steps of 0.5 °C at each cycle. The efficiency was 97.9 and 102.8% for IRC7 and ACT1, respectively. A standard curve was determined for each gene, where x is the threshold cycle and y is the log value of the starting quantity (ng): (IRC7 y=−3.374x +19.132, R2=0.997 and ACT1 y=−3.257x+20.308, R2=0.993). Standard curves were obtained from eight points and linearity was observed from 0.15 pg to 63 ng of DNA. A threshold value for the fluorescence of all samples was set manually, to maintain the same value in each experiment. For strain comparison, relative amounts of IRC7 and ACT1 were calculated from the standard curve. Results are given as the relative fold expression of (IRC7/ACT1) taking as the WT value at 10% of CO2 produced a reference (=1).


NCR controls volatile thiols release

NCR consists of the specific inhibition of the GATA transcriptional activators such as Gln3p by the cytoplasmic protein Ure2p (Courchesne & Magasanik et al., 1988). This inactivation occurred when optimal sources of nitrogen are available, reducing the transcription level of many target genes involved in poor nitrogen source utilization. Therefore, a Δure2 strain showed constitutive expression of these target genes. In contrast, a Δgln3 strain did not express these target genes even in a nitrogen-poor environment. The effect of NCR on the release of volatile thiols was assayed by measuring the production of volatile thiols in WT, Δure2 and Δgln3 strains. Fermentations were carried out in 350-mL batch cultures using a synthetic grape medium. Neither the Δure2 nor the Δgln3 allele affected the fermentation rate drastically in synthetic grape juice. All the fermentation processes were thus achieved in comparable times. Synthetic precursors were added to the synthetic juice at 20 nM for P-4MSP and 3000 nM for P-3SH in accordance with standard Sauvignon blanc must composition (Peyrot des Gachons et al., 2000). During fermentation, yeast produced 4MSP from its S-cysteine conjugate (P-4MSP). Similarly, 3SH was produced from P-3SH. A part of 3SH can also be acetylated during fermentation, producing 3SHA. At the end of the fermentation process, 4MSP, 3SH and 3SHA produced by yeast were organically extracted and measured by GC-MS (Tominaga, 1998a). The Δure2 mutant released three times more volatile thiols than the wild-type strain (Table 2). This result suggests that the derepression of nitrogen catabolism in a Δure2 background increases the ability of yeast to reveal such aromas. As Ure2p has been referenced for other cellular functions such as glutathione peroxidase (Bai et al., 2004), we investigated whether the NCR pathway was directly implicated by measuring the release of volatile thiols in a Δgln3 mutant. As shown in Table 2, the level of volatile thiols decreased dramatically in this mutant.

View this table:

Volatile thiols release is regulated by Ure2p/Gln3p couple

Strain4MSP release3SH release3SHA release
MeanCV (%)MeanCV (%)MeanCV (%)
  • * Statistically different from WT strain, anova, P=0.01.

The URE2 deletion has a stereoselective effect on the 3SH enantiomer ratio

3SH is a chiral molecule with two enantiomers, (R)-3SH and (S)-3SH. It was demonstrated recently that (R)-3SH and (S)-3SH produced different aromatic nuances, i.e. grapefruit and passion fruit, respectively, with similar perception thresholds in a hydroalcoholic model solution, i.e. 50 and 60 ng L−1, respectively (Tominaga et al., 2006). Using a specific GC-MS methodology (Tominaga et al., 2006), (R)-3SH and (S)-3SH were measured at the end of the alcoholic fermentation process in a synthetic medium containing 3000 nM P-3SH. The relative amounts of 3SH enantiomers of WT and Δure2 strains were compared (Table 3). As is observed generally in dry white wine, the medium fermented by VL3-1D (WT) yielded a racemic mix between (R)-3SH and (S)-3SH enantiomers of c. 50 : 50. Interestingly, the deletion of URE2 had an unexpected effect on this ratio. Indeed, the (R) and (S) forms increased differently, raising the (R)/(S) ratio to 78 : 22 in the Δure2 background. This (R)/(S)-3SH balance modification suggests that stereoselective mechanisms occurred in a ‘nitrogen-derepressed’ background. These findings indicated that, in addition to enhancing aroma release, Δure2 deletion might modify the sensory perception of 3SH aroma of wine by intensifying the grapefruit overtones [the main (R)-3SH descriptor].

View this table:

The Δure2 strain showed a modified 3SH enantiomers ratio

Strain(R)-3SH enantiomer(S)-3SH enantiomer(R)/(S) ratio
Mean (ng L−1)SDRelative amountMean (ng L−1)SDRelative amount
WT115978.1100130888.210053 : 47
Δure26707382.05791892108.014478 : 22
  • * Statistically different from WT strain, anova, P=0.01.

Precursor cleavage activity is the main target of Ure2p regulation

Volatile thiols precursors (P-3SH and P-4MSP), with regard to their biochemical structure, can be considered to be particular amino acids, i.e. a cysteine carrying a long S-associated residue. Although these precursors are not ‘natural’ substrates for the yeast, their uptake and cleavage should follow the same route as other amino acids. Therefore, the increase in volatile thiols observed in nitrogen-derepressed cells (Δure2) might be produced by the induction of a cleaving enzymatic activity and/or by the induction of precursor uptake through the membrane. Moreover, the change in the (R)/(S)-3SH ratio observed in Δure2 might be due to stereospecific mechanisms occurring either in transport or during the course of the metabolic process. To identify the gene category that is mainly involved in volatile thiols' release, we monitored precursor depletion and release of volatile thiols during the alcoholic fermentation for strains Δure2 and WT, respectively (Fig. 1). Both the strains stimulated fermentation, showing similar CO2 production rates (data not shown). To allow the comparison of the strains at the same physiological stages, samples were taken according to the percentage of CO2 release (as described in Materials and methods). Volatile thiols were mainly released during the first two-thirds of the fermentation process, reaching a peak production of around 40% for 3SH and 80% for 4MSP. From the first 10% of the fermentation process, the Δure2 strain produced significantly more thiols than the control strain. The utilization of precursors by yeast was estimated by measuring their respective concentrations in the fermenting medium. As depletion of precursors in a synthetic grape juice is only due to yeast metabolism, this experiment allowed us to assess the global quantity of assimilated precursors (Tominaga, 1998c). After 20% of CO2 had been produced, the Δure2 strain absorbed more than 90% of P-3SH and P-4MSP, while the control (WT) absorbed only 60–70% of these molecules. Although significant, these slight enhancements of precursors' assimilation cannot completely explain the strong discrepancy in the release of aromas observed at this time point (5.1- and 6.6-fold higher for 4MSP and 3SH, respectively). Later in the fermentation process (from 30% of CO2 released), both strains assimilated all the precursors; however, the Δure2 strain still produced three times more volatile thiols than the control strain. Finally, the (RR) and the (RS) forms of P-3SH were measured (see Materials and methods). For both strains, these diastereoisomers were assimilated in a racemic fashion during the alcoholic fermentation. As shown in Table 3, the 3SH overproduction effect of strain Δure2 was mainly due to the overproduction of the (R)-3SH enantiomer, while the (S)-3SH concentration was not modified drastically. Together, these findings demonstrated that with similar quantities of assimilated precursors, the Δure2 strain released progressively more volatile thiols [(R)-3SH as well as 4MSP] than the control strain during the fermentation process. Therefore, under enological conditions, the NCR did not seem to have a major control on assimilation of precursors but essentially on enzymatic cleavage activity.


Monitoring of precursors and thiols reveals the activation of enzymatic cleavage in the Δure2 strain. Precursor depletion and relative thiol release were monitored during the alcoholic fermentation in WT and Δure2 strains. The data presented were plotted as a function of percentage of CO2 released. Precursors: P-4MSP (●), (RR)-P-3SH (▪) and (RS)-P-3SH (◻). Thiols: 4MSP (▲), (R)-3SH (◆) and (S)-3SH (◊). Data shown are the mean of duplicate values with relative SD.

Impact of several yeast β-lyases on volatile thiol release and 4MSP precursor concentration

As hypothesized previously, cleavage of aroma precursors might be carried out by yeast enzymes bearing S-cysteine-conjugate β-lyase activity (Tominaga, 1998c). Such enzymes were investigated in S. cerevisiae using a bioinformatics approach, and deletion experiments of putative candidate genes were carried out in the VL3c strain background. Three genes, BNA3, CYS3 and IRC7, were described to contribute strongly to volatile thiol release (Howell et al., 2005). As URE2 deletion seems mainly to increase enzymatic cleavage activity, we investigated the role of these three genes and their regulation by NCR under enological conditions in depth. We deleted the two copies of BNA3, CYS3 and IRC7 genes in a VL3-1D background. The deletion effect of these genes was first assayed by measuring volatile thiol release at the end of the alcoholic fermentation process of a synthetic grape juice containing natural amounts of aroma precursors (i.e. 20 nM P-4MSP, 3000 nM P-3SH). At these concentrations, IRC7 deletion drastically reduced 3SH (−41%) and 4MSP (−96%) release (Table 4) while BNA3 and CYS3 deletion did not affect thiol release. This result partially contradicts those presented by Howell (2005) that showed a role for CYS3, BNA3 and IRC7 genes in 4MSP release. However, their conclusions were based on experimental conditions under which synthetic medium contained 0.1 g L−1 (450 μM) of P-4MSP, while in natural grape juice lower concentrations are found (0.5–20 nM) (Peyrot des Gachons et al., 2000). We supposed that these high concentrations could bias the stoichiometric conditions of the β-lyase reaction. To verify this hypothesis, we measured 4MSP release in a synthetic model medium containing 200 000 nM P-4MSP, corresponding to 10 000 times the P-4MSP concentration in grape juice. In accordance with Howell's findings, at this concentration, 4MSP release decreased c. 81 and 96% for strains Δbna3 and Δirc7, respectively (Table 4). However, deletion of the CYS3 gene did not affect this release even when using high concentrations of P-4MSP.

View this table:

The cystathionine β-lyase Irc7p was the main enzyme involved in thiols precursor cleavage

StrainPrecursor concentration
P-4MSP, 20 nM (1 ×)P-4MSP, 200 μM (10 000 ×)P-3SH 2,000 nM (1 ×)
4MSP released4MSP released3SH released
MeanCV (%)MeanCV (%)MeanCV (%)
  • * Statistically different from WT strain; anova, P=0.01.

The enhancing effect of URE2 deletion was mediated by Irc7p expression

The activation of IRC7 (YFR055c) transcription in a Δure2 background was demonstrated recently under laboratory conditions (Scherens et al., 2006). Using real-time PCR, we investigated this transcription activation during the alcoholic fermentation for our wine strain. During the alcoholic fermentation, the IRC7 transcription level strongly increased in the Δure2 strain as compared with the control (WT) (Table 5). Although the strains compared showed variable thiol release from 10% of CO2 production, no differences were found for the first two points (10% and 20% of CO2 released). This may be due to the mRNA detection limits in our real-time PCR assay, generating large standard deviation values in the first part of the fermentation. After 40% of CO2 release, IRC7 transcript levels were significantly higher for the Δure2 strain (one-way anova, P=0.05). Although smaller, this twofold increase in IRC7 expression in the Δure2 strain should explain the observed threefold overproduction of thiols. Nevertheless, other β-lyases, including Irc7p, should be activated by URE2 deletion and therefore may contribute to the enzymatic cleavage. We therefore investigated the Irc7p contribution in a Δure2 background by comparing thiol release in WT, Δure2, Δicr7 and the double mutant Δure2Δirc7 (Table 6). 4MSP and 3SH releases were identical in the Δirc7 and Δure2Δirc7 strains. Consequently, in the absence of the Irc7p enzyme, URE2 deletion did not increase thiol release. This demonstrated the epistasic relationship between IRC7 and URE2, suggesting that in derepressed cells, no other enzymes contributing to volatile thiol release were activated through the NCR pathway. Under the present conditions, Irc7p thus appears to be the unique enzyme, whose precursor cleavage activity was transcriptionally induced through the NCR regulation pathway. Moreover, when Irc7p was absent, the stoichiometric switch of (R)/(S)-3SH enantiomers induced by the URE2 deletion was not observed (Table 6). This suggested that the Irc7p enzyme had a better substrate affinity for (RR)-P-3SH production as a consequence of more (R)-3SH. To demonstrate this hypothesis, the (R) and the (S)-3SH production by Irc7p, in particular, was estimated in WT and Δure2 backgrounds (Table 7). In both backgrounds, these results showed a higher cleavage activity of Irc7 for (RR)-P-3SH. The fact that Irc7p was the sole β-lyase induced in a Δure2 mutant emphasizes the enantiomer shift (78 : 22) observed in a Δure2 mutant.

View this table:

IRC7 expression is enhanced during alcoholic fermentation in a Δure2 strain

StrainRelative fold expression (IRC7/ACT1) at different percentage of CO2 produced
WT1.00 ± 0.941.41 ± 0.801.64 ± 0.441.18 ± 0.340.93 ± 0.31
Δure20.67 ± 0.581.18 ± 0.063.36 ± 0.522.36 ± 0.606.40 ± 2.48
  • * Statistically different from WT strain; anova, P=0.05.

View this table:

IRC7 was the main gene activated by URE2 deletion

Strain4MSP released3SH released3SHA release
MeanCV (%)MeanCV (%)(R)-form (%)(S)-form (%)MeanCV (%)
  • * Statistically different from WT strain; anova, P=0.05.

  • ** Statistically different from WT strain; anova, P=0.01.

View this table:

The Irc7p cleaved more (R)-3SH than S-3SH enantiomer

Strain3SH released (ng L−1)
Part of P-3SH cleaved by IRC7623.1299.8
Part of P-3SH cleaved by IRC75868.41168.9
  • * Statistically different from the (S)-form quantity; t-test, P=0.05.


Impact of the β-lyases in volatile thiols release

The present study sheds new light on the genetic determinism of the release of volatile thiols. First, we have re-evaluated the role of some yeast β-lyases by investigating their potential cleavage activity under enological conditions using the same gene deletion approach as Howell (2005). The starting genetic material used was a spore clone derived from the same commercial starter, VL3c. Therefore, both data sets were obtained in strains sharing a high percentage of genome identity in order to minimize the genetic background effect. Our results confirmed that Bna3p and Irc7p proteins had β-lyase activity able to cleave P-4MSP in related volatile thiols. Nevertheless, under enological conditions, only the putative cystathionine β-lyase Irc7p achieved the P-4MSP bioconversion. The concentration of P-4MSP used to determine this effect proved to be a crucial point as it modified the stoichiometric conditions of the reaction. More generally, this underlines the importance of using relevant wine metabolite concentrations during investigations of the aromatic modulation of wine by yeast strains. The 3SH release appears to be more ‘buffered’ than that of 4MSP as only 40% of P-3SH conversion was achieved by Irc7p. The residual cleavage activity may be assigned to a small pool of other enzymes, which may include BNA3 and/or CYS3, able to carry out a β-substitution on an amino acid skeleton. Owing to their probable additive effect, tracking such enzymes should be highly complex as it requires the construction of multideleted strains. This hypothesis was reinforced by previous reports showing that yeast strains have a smaller impact on 3SH release. According to a previous study, β-lyases may have a stereoselectivity for the P-3SH diastereoisomers (Wakabayashi et al., 2004). The pool of yeast enzymes able to cleave the P-3SH in relative thiols should have these affinity and/or activity differences for (RR) and (RS)-P-3SH. Therefore, the nearly racemic ratio of 3SH enantiomers observed in wines (Tominaga et al., 2006) may be the result of these activities. In such a case, ΔIrc7 mutants were expected to produce more (S)-3SH compared with controls, as demonstrated in Table 7. These results illustrate that the activation of a specific enzyme (e.g. Irc7p quantity through URE2 deletion) can modulate the aroma composition of wines by stereoselective mechanisms, increasing the complexity level of the interaction between yeast and wine aromas.

Impact of nitrogen regulation on volatile thiol release

This study links nitrogen catabolism and volatile thiol release. Using a genetic approach, we have demonstrated that volatile thiol release from chain-cysteinylated precursors was repressed by the Ure2p regulator through the general NCR mechanism. The protein Gln3p was found to be the main transcription factor involved in this release. We also demonstrated that the NCR process did not influence precursors' uptake but mainly controlled the enzyme cleavage activity. Using a new precursor assay methodology, we showed that uptake of both volatile thiol precursors is similar in WT and derepressed strains. This similar uptake was probably due to the small concentration of precursors (a few micromolars) that can be carried by the same transporters that assimilated other amino acids in the first part of the fermentation process. Finally, we demonstrated that the Ure2p protein repressed the transcription of IRC7, which encodes the main enzyme involved in thiol bioconversion. This transcriptional control, probably mediated by Gln3/Gat1 factors, was efficient in wine fermentation as demonstrated by quantitative real-time PCR analysis.

At face value, the level of 4MSP release of the Δirc7 mutant (3.6%) suggested that the production of this compound was nearly monogenetically determined. This suggestion differs far from reality and numerous experiments have demonstrated that this trait is quantitative (Murat, 2001a; Howell et al., 2004; Masneuf-Pomarède et al., 2006). The link between β-lyase activity and NCR could in part explain this strain variation. In fact, strong differences in gene expression are commonly found between yeast strains.

Other parameters explaining yeast strain discrepancies remain to be investigated. First, the transporters involved in precursor uptake should be identified through the same gene deletion strategy as that used for β-lyase enzymes. Second, we demonstrated that URE2 deletion improved the precursor bioconversion rate from 3–5% to 7–10%. This means that a large part of the precursors are not converted to volatile thiols or that the volatile thiols are not stable (Murat, 2001b). One possible explanation may be the existence of another metabolic pathway of these precursors, such as cysteine assimilation. Conversely, the observed bioconversion rate (precursor assimilated/thiols produced) may in fact represent an apparent value. In fact, volatile thiols are highly oxidizable molecules that can react promptly with other metabolites. Understanding how volatile thiol release can be protected against oxidation inside and outside of the cell is certainly one major challenge to improve the bioconversion of volatile thiols. For this particular standpoint, the URE2 gene could have an unexpected role controlling other metabolic pathways connected to thiols. For example, it was demonstrated recently that the URE2 deletion increases the extracellular amount of reduced glutathione 20-fold (Perone et al., 2006). As this compound plays a major role in protecting volatile thiols during the aging of bottled white wines (Lavigne-Cruège & Dubourdieu et al., 2002), the URE2 deletion should also preserve volatile thiols indirectly.


We thank Johan Photsavang for deletion experiments as well as Philippe Darriet and Michel Aigle for helpful discussions. We also thank Christine Schwimmer for English review of the manuscript. C.T. and P.M. were supported by Laffort oenologie.


  • Editor: Graham Fleet


View Abstract