FEMS Yeast Research, 15, 2015, fov034 doi: 10.1093/femsyr/fov034 Advance Access Publication Date: 2 June 2015 Research Article

RESEARCH ARTICLE

Yeast genes required for conversion of grape precursors to varietal thiols in wine Margarita Santiago∗ and Richard C. Gardner Wine Science Programme, School of Biological Sciences, University of Auckland, Private Bag 92019, Auckland 1010, New Zealand ∗ Corresponding author: School of Biological Sciences, University of Auckland, 3A Symonds Street Auckland, Private Bag 92019, Auckland 1010, New Zealand. Tel: +64-9-3737599 ext 87338; E-mail: [email protected] One sentence summary: Identification of yeast genes involved in transport and cleavage of cysteinylated and glutathionylated thiol precursors to release 3MH and 4MMP during wine fermentation. Editor: Isak Pretorius

ABSTRACT Three varietal thiols are important for the tropical fruit aromas of Sauvignon blanc: 4-mercapto-4-methylpentan-2-one (4MMP), 3-mercaptohexanol (3MH) and its acetylated derivative 3-mercaptohexyl acetate (3MHA). These thiols are produced by yeast during fermentation from precursors in grape juice. Here we identify genes in Saccharomyces cerevisiae that are required for the transport and cleavage of two thiol precursors: cysteine-4MMP and glutathione-3MH. A full-length copy of IRC7 is absolutely required for the cleavage of both precursors in the tested strains; the deleted form of the enzyme found in most yeast strains is incapable of converting these compounds into detectable thiols. By using strains that overexpress full-length IRC7, we further show that the glutathione transporter OPT1 and the transpeptidase CIS2 are also required for conversion of glut-3MH to its varietal thiol. No transporter for cys-4MMP was identified: a strain deleted for all nine known cysteine transport genes was still capable of converting cys-4MMP to its varietal thiol, and was also able to take up cysteine at high concentrations. Based on these results, we conclude that cysteine and glutathione precursors make a relatively minor contribution to 3MH production from most grape juices. Keywords: Saccharomyces cerevisiae; Sauvignon blanc; cysteine conjugates; glutathione conjugates; IRC7; OPT1; CIS2; DUG1

INTRODUCTION Varietal thiols are known to play a key role as odorants in Sauvignon blanc wines (Tominaga et al. 1998; Dubourdieu et al. 2006; Lund et al. 2009; Benkwitz et al. 2012). In particular, three thiols contribute to the tropical fruit aromas of this wine: 4mercapto-4-methylpentan-2-one (4MMP, with aromas of black currant/cat pee), 3-mercaptohexanol (3MH, grapefruit) and 3mercaptohexyl acetate (3MHA, passion fruit). Yeast is required for the production of these aromatic compounds during wine fermentation from precursors found in grape juice. So far, cysteinylated (cys) and glutathionylated (glut) conjugates have been proposed as thiol precursors (Fig. 1). There is good experimental support for a biochemical pathway for the formation of 4MMP and 3MH proposed by Dubourdieu and co-workers: cys and/or

glut precursors in grape juice are taken up by yeast and converted to their respective thiols. Yeast then acetylates a fraction of 3MH to yield 3MHA (Dubourdieu et al. 2006; Coetzee and du Toit 2012). However, the contribution to total thiols made from cys and glut precursors has come under increasing scrutiny. Firstly, the yield from the pathway is low—the concentrations of cys-4MMP, cys-3MH and glut-3MH in grape juice are high, but conversion rates of precursor added to synthetic medium range from less than 0.1% (Wakabayashi et al. 2004; Swiegers et al. 2007) up to 4% (Kobayashi et al. 2010), with most estimates in the range from 0.2 to 0.6% (Tominaga, Peyrot des Gachons and Dubourdieu 1998; ` Murat et al. 2001; Howell et al. 2005; Masneuf-Pomarede et al. 2006; Subileau et al. 2008a,b; Roland et al. 2010a; Zott et al. 2011).

Received: 23 January 2015; Accepted: 29 May 2015  C FEMS 2015. All rights reserved. For permissions, please e-mail: [email protected]

1

2

FEMS Yeast Research, 2015, Vol. 15, No. 0

Figure 1. Cysteinylated and glutathionylated precursors. Cys conjugates have been shown to release thiols through the action of β-lyase enzymes (thick black arrows). Evidence suggests that glut conjugates are converted into cys conjugates (thin black arrows). However, the conversion of glut precursors directly into thiols has not been proven (dotted black arrows).

Moreover, the concentration of 3MH precursors in juice does not correlate with the concentration of thiols in the final wine (Pinu et al. 2012). Experimental estimates of the contribution to total thiols from this pathway to date range from 10 to 15% (Schneider et al. 2006; Subileau et al. 2008a; Roland et al. 2010a). An alternative pathway for the formation of thiols was proposed (Schneider et al. 2006) in which 3MH is derived from E2-hexenal, a ‘green leaf volatile’ formed in damaged plant cells and found in grape juice. E-2-hexenal could react directly with H2 S to form 3MH or react with cysteine and/or glutathione to provide precursors which yeast then biochemically converts to thiols. There is some evidence that E-2-hexenal may be reacting with glutathione in juice to increase thiols (Roland et al. 2010b). More recently, supplying H2 S to grape juice was shown to produce very high concentrations of 3MH, and that both E-2hexenal and (E)-2-hexen-1-ol can act as precursors (Harsch et al. 2013). However, the evidence to date suggests that in most grape juices this pathway may also be responsible for only a minor proportion of thiols (Schneider et al. 2006; Subileau et al. 2008a; Roland et al. 2010a). Hence, additional pathways likely contribute to varietal thiol formation. The details of the yeast genes involved in the pathway from cysteinylated and glutathionylated precursors to thiols are not yet clearly established. A key step in this pathway was clarified by the demonstration that expression in yeast of a bacterial enzyme with β-lyase activity (tnaA) results in very high levels of thiol production (Swiegers et al. 2007). A yeast β-lyase gene, IRC7, has been shown to be capable of performing this reaction in yeast cells (Thibon et al. 2008; Roncoroni et al. 2011). Fulllength IRC7 is both necessary and sufficient for 4MMP production in juice (Roncoroni et al. 2011). However, its contribution to the formation of 3MH is less clear. Deletion of full-length IRC7 only partly blocked conversion of cys-3MH to 3MH (Thibon et al. 2008) and did not affect thiol yields from grape juice (Roncoroni et al. 2011). Nitrogen catabolite repression (NCR) has been shown to affect thiol yields in some yeast strains (Subileau et al. 2008b; Thibon et al. 2008; Winter et al. 2011a), but has no effect in others (Deed, Van Vuuren and Gardner 2011; Harsch and Gardner 2013). Dufour et al. (2013) have clarified that release of NCR (in ure2 strains) has a positive impact on thiol release only when the strain also possess a full-length IRC7 gene. Other genes have been proposed to be involved in thiols release, the glutathione transporter OPT1 was suggested to be required for uptake of glut-3MH precursor, based on the observa-

tion that a yeast strain deleted for OPT1 gave about 50% reduction in thiol yields from grape juice (Subileau et al. 2008a). In addition, deletion of GAP1 reduced conversion of cys-3MH to 3MH in synthetic medium (Subileau et al. 2008b). This work focused on defining yeast genes required for the uptake and conversion of cys-4MMP and glut-3MH precursors to varietal thiols. In particular, we sought to define the role of IRC7 in the formation of 3MH from its glut conjugate. The studies were undertaken in defined synthetic media resembling grape juice, with added precursors, in order to remove contributions to thiols from other pathways that might operate in grape juice. The work utilized previous observations (Swiegers et al. 2007; Roncoroni et al. 2011) that overexpression of β-lyase genes in yeast greatly increases thiol yields, which allowed the effects of gene deletions in the pathway to be seen in much greater contrast.

MATERIALS AND METHODS Yeast strains and growth Yeast strains and deletion mutants used in this work are shown in Table 1. We distinguish between the full-length version of IRC7 found in a few yeast strains such as YJM789, Zymaflore X5 and EC1118, which will be referred to as IRC7F , and the naturally deleted form found in the majority of yeast strains including the lab strain S288c (IRC7S ). The latter has a 38-bp deletion that results in a 340-amino acid protein, compared to the 400-amino acid full-length protein (Roncoroni et al. 2011). Overexpression of IRC7F was achieved as described (Roncoroni et al. 2011): in brief, the IRC7F coding region from commercial wine yeast Zymaflore X5 was inserted into an expression cassette driven by the constitutive PGK1 promoter and integrated into the HO locus of yeast strains by homologous recombination using selection for the antibiotic ClonNat. Yeast strains were grown on standard rich (YPD) or minimal (SD) medium, supplemented to overcome auxotrophies as required (Huang, Roncoroni and Gardner 2014). Modified SD plates (MSD, corresponding to ‘B’ medium in Cherest and SurdinKerjan 1992) containing low sulfate (0.3 mM), were used to test toxicity to sulfur analogs; prior to streaking yeast onto these plates, cells were sulfur-starved by growing them on MSD plates lacking sulfate.

Santiago and Gardner

3

Table 1. Yeast strains used in this work. Name BY4741 IRC7F ox BY4742 IRC7S ox BY4743 IRC7F ox BY4743 IRC7F/S ox BY4743 cis2 IRC7F ox BY4743 dug1 IRC7F ox BY4743 gnp1 IRC7F ox BY4743 mup1 IRC7F ox BY4743 opt1 IRC7F ox F15 Zymaflore F15 IRC7F ox F15 IRC7S ox F15 IRC7F/S ox F15 IRC7F ox opt1 M4054 IRC7F ox M4584 IRC7F ox MS2 M4238 IRC7F ox X5 Zymaflore

Genotype and strain information PPGK -IRC7F recombined into HO locus in BY4741(MATa; his3-1, leu2-0, met15-0, ura3-0) PPGK -IRC7S recombined into HO locus in BY4742 (MATα, his3- 1, leu2- 0, lys2- 0, ura3- 0) PPGK -IRC7F recombined into HO locus in BY4743 (MATa/α, his3-1/his3-1, leu2-0/leu2-0, LYS2/lys2-0, met15-0/MET15, ura3-0/ura3-0) Cross between BY4741 IRC7F ox and BY4742 IRC7S ox PPGK -IRC7F recombined into HO locus in BY4743 cis2 PPGK -IRC7F recombined into HO locus in BY4743 dug11 PPGK -IRC7F recombined into HO locus in BY4743 gnp1 PPGK -IRC7F recombined into HO locus in BY4743 mup1 PPGK -IRC7F recombined into HO locus in BY4743 opt1 Wine yeast, diploid PPGK -IRC7F recombined into HO locus in F15 h(α) PPGK -IRC7F recombined into HO locus in F15 (a), ura3 0 Cross between F15 IRC7F ox and F15 IRC7S ox opt1::hphMX recombined in F15 IRC7F ox PPGK -IRC7F recombined into HO locus in M4054 (MATa gal2 ura3 gap1-101, isogenic to S288c, Birgitte Regenberg) PPGK -IRC7F recombined into HO locus of M4584 (MATa ura3 gap1 agp1 gnp1 (bap2-tat1) bap3 tat2, Birgitte Regenberg) yct1::URA3, mup1::hphMX in M4584 IRC7F ox PPGK -IRC7F recombined into HO locus of M4238 (MATa gal2 ura3 gap1-101 ssy1-1, Birgitte Regenberg) Wine yeast

Origin This work This work This work This work This work This work This work This work This work Laffort Oenologie This work This work This work This work This work This work This work This work Laffort Oenologie

ox = overexpression.

Yeast growth rates were monitored using the Bioscreen C MBR plate incubator/reader/shaker controlled by EZExperiment software (Oy Growth Curves Ab Ltd, Helsinki, Finland). Precultures were grown in YPD to stationary phase, diluted to 3 × 104 cells mL−1 in water, and 5 μL of this yeast suspension was used to inoculate 150 μL of growth media in a 100-well honeycomb plate. Plates were incubated at 25◦ C for 72 or 144 h. The turbidity was measured at OD480–560 nm every 15 min, with 7.5 min shaking prior to measurement. Cultures were grown in SGM medium (see below) with 21% sugars to provide fermentation-like conditions. For testing growth on various sulfur sources, SGM without sulfur was made using magnesium chloride (rather than sulfate) and with methionine, cysteine and glutathione omitted from the supplements (the minor salts still contained 1.4 μM sulfate). For growth on cysteine as a nitrogen source, diammonium phosphate and the amino acids were omitted from SGM.

tent of the original SGM recipe (Harsch et al. 2010) was modified to reflect New Zealand grape juice measurements more accurately as follows: Ala (100 mg L−1 final concentration), Arg (400), Asp (50), Asn (10), Cys (5), Glu (100), Gln (125), Gly (5), His (20), Ile (25), Leu (25), Lys (5), Met (10), Phe (40), Pro (300), Ser (60), Thr (75), Trp (10), Tyr (10), Val (30). Fermentation was performed at 25◦ C, with shaking at 100 rpm and was monitored by daily weighing. Ferments were considered finished when weight loss was ≤0.1 g per 24 h. Wine was then separated from yeast cells and solids by centrifugation at 6000 g for 10 min, treated by adding 12 μL of dimethyl dicarbonate and stirring at 25◦ C for 4 h to inactivate remaining yeast cells, and stored frozen at −20◦ C. Thiols were purified using either extraction with phydroxymercuribenzoate or derivatization with ethyl propiolate (as detailed in Herbst-Johnstone et al. 2013) and analyzed using gas chromatography followed by mass spectrometry (GC-MS) using the methods described in Herbst-Johnstone et al. (2013).

Fermentation and thiol analysis

RESULTS

Yeast cells were precultured in YPD to stationary phase, washed in water and inoculated into fermentation media at 2 × 106 cells mL−1 . Fermentations were carried out in duplicate in 250mL conical flasks sealed with airlocks and containing 200 mL of a modified version of SGM, a synthetic media designed to resemble grape juice (Harsch et al. 2010) with added thiol precursors: glut-3MH (500 μg L−1 , Buchem BV, Apeldoorn, the Netherlands) and cys-4MMP (50 μg L−1 , Hebditch, Nicolau and Brimble 2007). SGM contains minerals, vitamins, organic acids, a mixture of amino acids (225 mg L−1 N total) and diammonium phosphate (75 mg L−1 total N) as nitrogen source, and 21% sugar (1:1 glucose:fructose), pH 3.2. For laboratory strains fermentation, SGM was three-quarters strength in sugars and supplemented with corresponding auxotrophies requirements. The amino acid con-

Full-length IRC7 is required for conversion of glutathione or cysteine precursors to thiols The production of varietal thiols was examined after fermentation of synthetic grape media supplemented with two precursors, cys-4MMP (50 μg L−1 ) and glut-3MH (500 μg L−1 ). Table 2 summarizes thiol yields from each precursor for five different yeast strains. The highest conversion of precursors to thiols occurred in strains that overexpressed a full-length copy of the IRC7 gene derived from X5 (Roncoroni et al. 2011). Conversion of cys-4MMP to 4MMP exceeded 50% in one experiment using the wine yeast F15 overexpressing, while conversion of glut-3MH to 3MH + 3MHA approached 10% in this strain. Yields from IRC7F overexpression in a laboratory strain derived from S288c were

4

FEMS Yeast Research, 2015, Vol. 15, No. 0

Table 2. Summary of thiol yields by different yeast genetic backgrounds in synthetic medium with added precursors.

Table 4. OPT1 and CIS2 deletion affects 3MH release. Strain

Strain (IRC7 genotype)

4MMP yield (%)

3MH + 3MHA

4MMP

3MH + 3MHA yield (%)

Glutathione transporter F15 (IRC7S /IRC7S ) IRC7F ox BY4743 (IRC7S /IRC7S ) IRC7F ox X5 (IRC7F /IRC7F ) EC1118 (IRC7F /IRC7S ) F15 (IRC7S /IRC7S )

34%, 67% 5%,11% 1% 0.2% nd

8%, 10% 5% 0.3% 0.2% nd

Results of two independent experiments are shown for the BY and F15 overexpression lines; nd = not detected; ox = overexpression.

Table 3. Overexpression of IRC7S is insufficient for thiol production.

BY IRC7F ox BY IRC7F ox opt1 F15 IRC7F ox F15 IRC7F ox opt1

BY IRC7F ox BY IRC7S ox BY IRC7F/S ox F15 IRC7F ox F15 IRC7S ox F15 IRC7F/S ox

4MMP yield (%) 5 0 5 44 0 41

± ± ± ± ± ±

0.4a 0b 0.3a 0.3a 0b 5.1a

3MH + 3MHA yield (%) 5 0 2 6 0 4

± ± ± ± ± ±

0.1a 0b 0c 0.2a 0b 0.3c

ox = overexpression. Different letter labels in a column (within a strain) mean that samples are significantly different (P < 0.001, ANOVA, Tukey’s HSD).

lower, but still convert 5–10% of cys-4MMP into 4MMP and ∼5% of glut-3MH into 3MH + 3MHA. Strains overexpressing IRC7F hugely overcome thiol release by two commercial wine yeast strains that were homozygous (X5) or heterozygous (EC1118) for alleles of the full-length IRC7 gene (under the control of their own promoters). The F15 strain containing only the shortened form of IRC7 did not produce detectable thiols. To confirm the dependence of thiol production from these two precursors on a full-length copy of IRC7, we also overexpressed the short form of the enzyme (derived from S288c), in both F15 and the lab strain backgrounds. In addition, because Irc7F p from yeast is tetrameric (Santiago and Gardner 2015) as other β-lyase enzymes (Clausen et al. 1996; Breitinger et al. 2001; Ku, Yip and Howell 2006), we tested hybrid strains that overexpressed both the full length and the short form of IRC7, to screen for dominant negative interference by the short form. Table 3 confirms that overexpression of the shortened form gave no detectable production of either 4MMP or 3MH from their respective precursors. In addition, overexpression of the shortened IRC7 gene did not affect conversion of cys-4MMP by the full-length enzyme, suggesting that the short version of the protein does not act as a dominant negative for this reaction. However, the production of 3MH was reduced somewhat in both strain backgrounds by expression of the shorter form of IRC7. These results demonstrate that conversion of cys-4MMP and glut-3MH to their respective thiols is both dependent on the presence of a full-length copy of IRC7. The high levels of thiol production as a result of overexpression also encouraged us to use deletion strains of yeast to identify other genes that are required for the uptake or conversion of precursors to active thiols.

OPT1 is required for uptake of glutathione precursors in F15 (with a second transporter in BY strains) Subileau et al. (2008a) demonstrated that a yeast strain deleted for the glutathione transporter, OPT1, gave a reduced yield of thiols from grape juice. We therefore tested the effect in our synthetic system of an OPT1 deletion in two yeast backgrounds,

± ± ± ±

0.4 0.4 0.3 2.3

5 2 6 0

± ± ± ±

0.1 0.1∗ 0.2 0

Glutathione degradation genes BY IRC7F ox BY IRC7F ox cis2 BY IRC7F ox dug1 ∗

Strain

5 5 44 46

11 ± 0.3 8 ± 0.6 8 ± 0

5 ± 0.2 1 ± 0.1∗ 4 ± 0.1

Significant differences compared to wt IRC7F ox strain (∗ P < 0.01, ANOVA).

into which an IRC7F overexpression construct was introduced. Table 4 (upper part) shows that while conversion of cys-4MMP to 4MMP was unaffected, the conversion of glut-3MH to 3MH was completely abolished in the wine yeast background, and reduced significantly in the lab strain. To confirm the lack of functional glutathione transport in these opt1 mutants, their growth was assessed on glutathione as a sole sulfur source. Fig. 2A shows that the deletion of OPT1 in the F15 strain background completely abolished growth on glutathione, whereas the same deletion in the BY4743 genetic background only partly abolished growth. These growth results reflected the relative conversion yields by each strain for the glutathionylated precursor in Table 4. It was concluded that the OPT1 gene is the only transporter present in F15 that is able to take up either glutathione or the glut-3MH precursor. However, the BY4743 genetic background may include a second transporter for both substrates.

CIS2, but not DUG1, is required for conversion of glutathione precursors to thiols There has been good recent progress in identifying genes in the degradation pathways for glutathione and its conjugates. Glutathione conjugates are broken down via the action of the CIS2 gene, encoding γ -glutamyl transpeptidase along with two other vacuolar peptidases (Wunschmann et al. 2009). In contrast, glutathione turnover in vivo appears not to involve CIS2, but rather utilizes the DUG1–3 peptidase genes and occurs in the cytoplasm (Baudouin-Cornu et al. 2012). Table 4 (lower part) shows the results of testing deletion mutations of CIS2 and DUG1 in the lab strain for their ability to convert thiol precursors in synthetic medium resembling grape juice. The cis2 deletion reduced conversion of glut-3MH by ∼80%, while there was no significant change in the dug1 deletion strain. Neither deletion affected yields of 4MMP from its cysteinylated precursor, as expected. Fig. 2B shows that deletion of DUG1, but not CIS2, reduced growth of the lab strain on glutathione as a sole sulfur source, consistent with previous findings (Baudouin-Cornu et al. 2012) for these mutants. The ability of strains to utilize glut-3MH as a sulfur source was also tested. Fig. 2B (right side) shows that the control BY4743 strain could grow on this substrate, although more slowly than on glutathione, and with a prolonged lag phase. Growth on glut-3MH as a sulfur source was delayed in the cis2 mutant and abolished in the dug1 deletion mutant. The dug1 result is interpreted as indicating that in order for the glut-3MH conjugate to be used as an S source, it must first be

Santiago and Gardner

5

Figure 2. OPT1 and DUG1 deletion mutants showed reduced growth on glutathione and glut-3MH as a sulfur source. (A) opt1 deletion mutants and (B) glutathione degradation mutants were grown using glutathione and glut-3MH as sulfur source (0.5 mM). Absorbance was measured every 15 min. Average curves were plotted, n = 2. The spiked growth curve of F15 in G-3MH is caused by the ‘clumping’ phenotype showed by haploid F15 due to its active form of the AMN1 gene, enhanced at low OD readings.

converted to glutathione, and then degraded by yeast to produce cysteine.

Cysteine transporters are not involved in cysteine precursor uptake We next attempted to define the transporter involved in uptake of cysteinylated precursors into yeast cells. Amino acid uptake into yeast involves two large families of genes (the AAP and DAL families), which encode transporters with overlapping substrate specificities and wide-ranging Km values, and which show complex regulation. In total, nine transporters have thus far been shown to be capable of transporting cysteine: GAP1, AGP1, GNP1, BAP2, BAP3, TAT1, TAT2, YCT1, MUP1 (During-Olsen et al. 1999; Kosugi et al. 2001; Kaur and Bachhawat 2007). As an initial screen, we tested the ability of single deletions of the nine cysteine transporters in BY4743 for their resistance to two cysteine analogs whose transport genes have not been defined. Fig. 3A shows that deletion mutants in mup1 and gnp1 conferred resistance to ethionine and S-ethyl-L-cysteine, respectively. Encouraged by this result, we overexpressed IRC7F in gnp1 and mup1 mutants and tested the resulting strains for their ability to convert cys-4MMP to its corresponding thiol. However, none of the mutants showed a significant reduction (P < 0.01) in 4MMP production from cys-4MMP (Fig. 3B). In addition, IRC7F was overexpressed in other six cysteine transporter mutants (agp1, bap2, bap3, tat1, tat2 and yct1) and also no significant difference in 4MMP release was found after fermenting synthetic media with added cys-4MMP (not shown). BAP2 deletion gave the lowest 4MMP release (P = 0.068, ANOVA) compared to the BY4743 wt strain, but it also showed a slower fermentation rate compared to the rest of the strains (not shown). In addition, none of the mutants affected conversion of glut3MH to thiols (data not shown).

A strain in which all nine of the known cysteine transporters had been deleted was assessed for thiols conversion. Fig. 4A shows that this multiple deleted strain was also able to convert both precursors into thiols, with no significant reductions in yield. The result suggests that none of the deleted genes is responsible for uptake of cys-4MMP. In addition, we tested this multiply deleted strain for their ability to grow in fermentation conditions using only cysteine as a nitrogen source—we have shown elsewhere (Santiago and Gardner 2015) that overexpression of IRC7F in yeast is sufficient to confer the ability to utilize cysteine as a nitrogen source, via its action of cleaving cysteine to yield pyruvate, ammonia and H2 S. This test also served to assay for functional overexpression of the IRC7F gene in each mutant line. Growth on high cysteine (15 mM) is quite toxic for yeast, and growth on this nitrogen source is significantly delayed and at a much slow maximum rate compared to that on ammonium (not shown). However, the multiply deleted strain was still able to grow on cysteine (Fig. 4B), suggesting that it still possess at least one intact transport gene that is able to take up cysteine. We also examined the effect of high nitrogen conditions, with the aim of identify if the cys-4MMP permease is affected by NCR. The experiment was carried out in the strain M4238 gap1 ssy1 (Didion et al. 1998), which contains deletions of the general amino acid permease GAP1 and of the extracellular amino acid sensor protein SSY1; thus, it has impaired transport of multiple amino acids due to the downregulation of specific aa permeases. This strain was chosen to try to reduce the transport background. In addition, IRC7F has been overexpressed in M4238 to observed thiol release. Fig. 5 shows that the addition of very high levels of ammonium to synthetic medium reduced the conversion of cys-4MMP to 4MMP by around 10-fold—from around 4000 ng L−1 to around 400 ng L−1 —without affecting the conversion of glutathionylated precursor to 3MH. This result strongly

6

FEMS Yeast Research, 2015, Vol. 15, No. 0

Figure 3. MUP1 and GNP1 deletion mutants show resistance to ethionine and S-ethyl cysteine, but no effect on thiol release. (A) BY4743 strains deleted for individual permease genes were screened for ethionine or SEC resistance. (B) The gnp1 and mup1 mutants were transformed with the ho::PPGK -IRC7F overexpression cassette and used to ferment SGM with added precursors. The labels above the bars show the percentage of thiol conversion from the added precursor. n = 2; error bars = SE; no significant differences compared to BY4743 IRC7Fox strain (P < 0.01, ANOVA).

suggests that the Cys-4MMP transporter is regulated by NCR, in a GAP1-independent way.

DISCUSSION The formation of varietal thiols such as 3MH and 4MMP requires the action of yeast on precursors in grape juice. This work assessed the role of candidate yeast genes in the conversion of cysteinylated (cys-4MMP) and glutathionylated (glut-3MH) precursors to thiols. A major finding from the work is that the yeast IRC7F gene is essential for the conversion of both of these thiol precursors in the wine yeast strain F15 and some lab strains. Also, it was confirmed that IRC7F was also needed for the conversion of cys-3MH into 3MH in F15 (not shown). In addition, OPT1 and CIS2 were shown to be important for the uptake and degradation, respectively, of the glutathionylated precursor. However, none of the known cysteine uptake genes are essential for transport of the cysteinylated precursor. The conclusion that there is a requirement for full-length IRC7 gene activity for the generation of 4MMP is consistent with published data. The IRC7 gene has previously been shown to be required for the conversion of cys-4MMP to 4MMP in synthetic media (Thibon et al. 2008). Initially, it was suggested that deletion of BNA3 in VL3 reduced thiol yields (Howell et al. 2005), but this result has been shown to be an artifact due to the very high concentrations of precursors used (Thibon et al. 2008). IRC7 has also been shown to be required for the production of 4MMP in wine made from grape juice (Roncoroni et al. 2011). Additionally, several authors have reported 4MMP release from cys-4MMP when using IRC7 full-length strains [VL3 (Murat et al. 2001; Masneuf` Pomarede et al. 2006), VL1 (Murat et al. 2001), VIN13 (Masneuf-

` et al. 2006, Swiegers et al. 2007) and X5 (Zott et al. 2011)]. Pomarede Also some strains with unknown IRC7 allele have shown to release 4MMP in synthetic conditions [EG8, 522d (Murat et al. 2001) ` and NW3 (Masneuf-Pomarede et al. 2006)]. However, the strain F10 (IRC7S , unpublished data) has been proven to produce 4MMP from its cysteinylated precursor (Zott et al. 2011), being so far the only exception that contrast our results that a full-length IRC7 is required for the release of 4MMP from cys-4MMP. Our conclusion that there is a requirement for an active copy of IRC7F in the production of 3MH from cys and glut-3MH is also supported by other authors that have successfully converted synthetic precursors to thiols using different yeast strains with full-length IRC7 (Roncoroni et al. 2011; M. Santiago, unpublished), including VL3 (Kobayashi et al. 2010; Winter et al. 2011b), X5 (Zott et al. 2011) and Vin13 (Swiegers et al. 2007; Kobayashi et al. 2010; Roland et al. 2010a). In addition, strains with unknown IRC7 alleles, Siha8 (Wakabayashi et al. 2004), ES1 and ES2 (Subileau et al. 2008a), have also been reported to release 3MH from precursors. However, our result contrasts with three earlier reports. First, deletion of IRC7F in the VL3 strain only partly eliminated the conversion of cys-3MH to 3MH in synthetic medium (Thibon et al. 2008). Secondly, two strains Sigma1278b (Subileau et al. 2008b) and Zymaflore F10 (Zott et al. 2011) that have only IRC7S (Sigma 1278b has been sequenced, and we have genotyped F10 at this locus) have shown to release 3MH from cys-3MH. Thus, these three reports provide exceptions to our conclusion that strains with a shortened version of IRC7 are incapable of converting precursors into thiols. Currently, we are unable to reconcile these divergent published results with our findings, but it could be possible that some strains (e.g. VL3, F10, Sigma 1278b) also possess a gene

Santiago and Gardner

Figure 4. Deletion of all nine cysteine transporters does not affect 4MMP release. (A) Mutant (MS2) with nine deleted cysteine transporters was transformed with the ho::PPGK -IRC7F overexpression cassette and used to ferment SGM with added precursors. The labels above the bars show the percentage of thiol conversion from the added precursor. n = 2; error bars = SE; no significant differences compared to gap1 strain (∗ P < 0.01, ANOVA). (B) MS2 mutant growth on L-cysteine as a nitrogen source. Average curves were plotted. n = 2.

different from IRC7F capable of cleaving the cys-3MH and possibly cys-4MMP precursors. Our conclusion that a full-length copy of IRC7 is absolutely required for conversion of glutathionylated precursors to 3MH depends on two lines of evidence. First, we consistently failed to detect 3MH peaks when IRC7S -containing strains such as F15 or BY4743 were fermented (e.g. Table 2). However, it must be noted that peak sizes for strains with wildtype full-length IRC7 alleles were relatively small, and that both strain differences and experimental variations in yield are common. Thus, it is possible that we missed some small peaks. Second, overexpression of an S288c-derived IRC7S allele gave no detectable thiols in the same two parent strains. In this case, the absence of any peak (with a GC-MS detection limit of around 10–20 ng L−1 ) contrasts with very much higher values of around 10 000 ng L−1 produced with overexpression of the full-length gene in the background of the wine strain F15. Thus, the overexpression strategy produced very large differences between producers and non-producers. However, we currently have only very limited evidence that our overexpression construct for the short version of IRC7 was functionally expressed. The construct did give one altered phenotype—a reduction of conversion of glut-3MH in the F1 hybrid with the overexpressing long form. However, there was no corresponding reduction in 4MMP yields, which might have been anticipated. We also assessed the copper tolerance on YPD media of both various IRC7-deleted strains and overexpressing strains, in both the F15 and BY genetic backgrounds, based on earlier reports of a difference for the IRC7S deletion strain (van Bakel et al. 2005); however, no differences were noted for any of the strains compared to their parent (including the IRC7S deletion; data not shown). Thus, it remains

7

possible that our IRC7S overexpression construct was not functional in some way. On balance, then, this issue will require further experiments to resolve. By overexpressing IRC7F in a range of mutant strains, we demonstrated that both OPT1 and CIS2 are important for conversion of glut-3MH to its corresponding thiols 3MH and 3MHA. In the case of OPT1, we did see some strain dependence that suggested that some yeast strains have more than one transporter for both glutathione and the glutathionylated precursors of thiols. In the case of CIS2, deletion of the gene greatly reduced but did not completely abolish thiol production—similar results were found for degradation of a glutathione-bimane conjugate (Wunschmann et al. 2009). Fig. 6 shows a speculative model that outlines possible steps and yeast genes involved in the uptake and conversion of glutathione-3MH precursors in yeast. Knowledge of yeast genes required for conversion of cysteinylated and/or glutathionylated precursors will provide a research tool to distinguish these sources of thiols in wines made from grape juice, provided that the role of IRC7 can be clarified. For example, deletion of OPT1 reduced 3MH + 3MHA yields from one grape juice by around 50% (Subileau et al. 2008a), and we have repeated this observation in two New Zealand grape juices comparing F15 and F15 opt1 strains (data not shown). However, while this reduction may have occurred due to reduced uptake of the glutathionylated precursor, as suggested by Subileau et al. (2008a), these experiments used strains that lack a full-length copy of IRC7. Therefore, the effect on thiols may also be via another pathway, perhaps by reducing the uptake of glutathione and thereby altering sulfur metabolism in yeast. It has been shown that perturbations to sulfur and nitrogen metabolism alter thiol yields from grape juices using IRC7S yeast strains (Harsch and Gardner 2013), and that addition of glutathione to grape juice affects thiol yields (Patel et al. 2010). One possible conclusion from this work is that cysteinylated and glutathionylated precursors may make only a small contribution to 3MH + 3MHA production in most grape juices. Previously, we showed that deletion of IRC7 did not affect 3MH + 3MHA production from grape juice (Roncoroni et al. 2011). Indeed, it is clearly possible to get good 3MH yields from grape juice using yeast strains possessing only the short form of the gene (e.g. BY4743) (Subileau et al. 2008a; Harsch and Gardner 2013), M2 (Deed, Van Vuuren and Gardner 2011) and F15 (Harsch and Gardner 2013). In our hands, commercial yeast strains typically vary less than 2-fold in their yields of 3MH + 3MHA from grape juice, regardless of the allele(s) of IRC7 they contain (S. Lee and R. Gardner, unpublished results). Thus, if a full-length IRC7 is required for conversion of both the cysteinylated and glutathionylated precursor, then these particular precursors cannot be contributing the majority of 3MH + 3MHA in wine. This conclusion is in agreement with the low estimates based on conversion of labeled precursors added to juice (Schneider et al. 2006; Subileau et al. 2008a; Roland et al. 2010a), as well as the finding that there is no correlation between 3MH + 3MHA yields and the concentrations of the cysteinylated and glutathionylated precursors (Pinu et al. 2012). Fig. 2 showed that an OPT1-deleted derivative of the S288Cderived laboratory strain BY4743 was able to grow on glutathione, and it is therefore suggested that this strain may have an additional transporter for glutathione, such as ADP1 (Gustafsson et al. 2014), or alternatively that it may have an extracellular system to degrade glutathione. In contrast, deletion of OPT1 in both F15 and another S288c-derived strain M4238 (not shown) led to their being unable to grow. Previously, OPT1 was found to

8

FEMS Yeast Research, 2015, Vol. 15, No. 0

Figure 5. High nitrogen strongly reduced cys-4MMP conversion to 4MMP. Strains were fermented in SGM with added precursors. Diammonium phosphate (DAP) was added to provide 77 mg L−1 of nitrogen (standard SGM) or 950 mg L−1 of nitrogen (SGM also contains 250 mg L−1 of amino nitrogen as a mixture of amino acids, giving total YAN of 327 mg L−1 for SGM or 1200 mg L−1 for high DAP treatment). The labels above the bars show the percentage of thiol conversion from the added precursor. n = 2; error bars = SE; asterisks above bars represent significant differences compared to gap1 strain (∗ P < 0.01, ANOVA).

Figure 6. Model for yeast processing of glutathionylated precursors of 3mercaptohexanol. Transport of glut-3MH into the cell occurs through Opt1p. Then, a fraction is cleaved by γ -glutamyltranspeptidase Cis2p and subsequently by beta lyase, Irc7F p to release 3MH. In some strains it seems to be an alternative unknown gene(s) coding for the transporter and the β-lyase. Alternatively, part of glut-3MH is degraded through the action of Dug1 and used as a sulfur source.

be the only glutathione transporter (Bourbouloux et al. 2000) in the YPH499 strain derived from S288C (Sikorski and Hieter 1989). The genetic basis for these strain differences is currently unclear. We identified MUP1 and GNP1 as the likely transporters for ethionine and S-ethyl-L-cysteine, respectively (Fig. 3), but we were not successful in identifying a major transport gene for cys-4MMP. However, the nine known cysteine transport genes are unlikely to play a major role, since strains deleted for all of these transporters were capable of uptaking both cysteine and cys-4MMP in fermentation medium. However, we did find a large effect of nitrogen on thiol yields, consistent with previous work. Fig. 5 shows that increasing the nitrogen content of synthetic fermentation medium to very high concentrations reduced yields of 4MMP from cys-4MMP by around 10-fold, without affecting glut-3MH conversion. It is possible that this reduction occurred because of a release of NCR on a transporter that takes up the cys-4MMP precursor; a number of candidate transporters are regulated in these conditions including DIP5, UGA4, PUT4, DAL4, DAL5 (see Deed, Van Vuuren and Gardner 2011). Our data are not consistent with a previous report (Subileau et al. 2008b) that deletion of GAP1 reduces thiols. Fermenting gap1 strains in our experimental conditions (e.g. M4054 and M4238, in Figs 4 and 5, respectively) showed ∼14% of 4MMP conversion yield, which is similar to other wild-type (wt) laboratory strains. However, our experiments used different strains, different fermentation media and cys-4MMP rather than cys-3MH. We confirmed high 4MMP yields from cys-4MMP in strains overexpress-

ing IRC7, which suggests that transport might not be a limiting factor in 4MMP release in wt strains. The yield of thiols from the two precursors studied here is typically low using standard wine yeast strains (0.2–1%, Table 2), consistent with published estimates (see the section ‘Introduction’). A possible contributing factor to the low yields is that the precursors are lost to other pathways in yeast, so that the sulfur moiety in the precursors becomes unavailable for conversion to volatile thiols. Although cys-4MMP appears not to be used as a sulfur source (data not shown), the glutathionylated 3MH precursor can be used (albeit poorly) by both the lab strain and F15 for growth, and the DUG1 gene appears to be required for this process (Fig. 2B). We therefore propose that one source of ‘thiol losses’ is that glut-3MH is converted to glutathione by the action of a reversible glutathione-S-transferase (GST) enzyme. Glutathione could then be broken down to cysteine via the DUG pathway and utilized by the cell for growth (illustrated in Fig. 6). Reversible GST enzymes have been postulated in plants, along with a range of other modifications of cysteine and glutathione conjugates that could also function as endpoints for sulfur and therefore ‘losses’ in the thiol pathway (Dixon, Skipsey and Edwards 2010).

ACKNOWLEDGEMENTS We are grateful to Birgitte Regenberg for her provision of yeast strains for this research and to Keith Richards for advice and assistance with this work.

FUNDING The research was funded by grants from the Faculty of Science at the University of Auckland, and from the New Zealand Foundation of Research Science and Technology (UOAX0404). Margarita Santiago was the recipient of a Chilean Bicentennial Scholarship from the Chilean National Scholarship Program for Graduate Students (CONICYT—Becas Chile). Conflict of interest. None declared.

REFERENCES Baudouin-Cornu P, Lagniel G, Kumar C, et al. Glutathione degradation is a key determinant of glutathione homeostasis. J Biol Chem 2012;287:4552–61. Benkwitz F, Tominaga T, Kilmartin PA, et al. Identifying the chemical composition related to the distinct aroma

Santiago and Gardner

characteristics of New Zealand Sauvignon blanc wines. Am J Enol Viticult 2012;63:62–72. Bourbouloux A, Shahi P, Chakladar A, et al. Hgt1p, a high affinity glutathione transporter from the yeast Saccharomyces cerevisiae. J Biol Chem 2000;275:13259–65. Breitinger U, Clausen T, Ehlert S, et al. The three-dimensional structure of cystathionine β-lyase from Arabidopsis and its substrate specificity. Plant Physiol 2001;126:631–42. Cherest H, Surdin-Kerjan Y. Genetic analysis of a new mutation conferring cysteine auxotrophy in Saccharomyces cerevisiae: updating of the sulfur metabolism pathway. Genetics 1992;130:51–8. Clausen T, Huber R, Laber B, et al. Crystal structure of the pyridoxal-5 -phosphate dependent cystathionine β-lyase ˚ J Mol Biol 1996;262:202–24. from Escherichia coli at 1.83 A. Coetzee C, du Toit WJ. A comprehensive review on Sauvignon blanc aroma with a focus on certain positive volatile thiols. Food Res Int 2012;45:287–98. Deed NK, Van Vuuren HJJ, Gardner RC. Effect of nitrogen catabolite repression and di-ammonium phosphate addition during wine fermentation by a commercial strain of Saccharomyces cerevisiae. Appl Microbiol Biot 2011;89:1537–49. Didion T, Regenberg B, Jorgensen MU, et al. The permease homologue Ssy1p controls the expression of amino acid and peptide transporter genes in Saccharomyces cerevisiae. Mol Microbiol 1998;27:643–50. Dixon DP, Skipsey M, Edwards R. Roles for glutathione transferases in plant secondary metabolism. Phytochemistry 2010;71:338–50. Dubourdieu D, Tominaga T, Masneuf I, et al. The role of yeasts in grape flavor development during fermentation: the example of Sauvignon blanc. Am J Enol Viticult 2006;57:81–8. Dufour M, Zimmer A, Thibon C, et al. Enhancement of volatile thiol release of Saccharomyces cerevisiae strains using molecular breeding. Appl Microb Biot 2013;97:5893–905. During-Olsen L, Regenberg B, Gjermansen C, et al. Cysteine uptake by Saccharomyces cerevisiae is accomplished by multiple permeases. Curr Genet 1999;35:609–17. Gustafsson FS, Whiteside MD, Jiranek V, et al. Development and use of a quantum dot probe to track multiple yeast strains in mixed culture. Sci Rep 2014;4:6971. Harsch MJ, Benkwitz F, Frost A, et al. New precursor of 3mercaptohexan-1-ol in grape juice: thiol-forming potential and kinetics during early stages of must fermentation. J Agr Food Chem 2013;61:3703–13. Harsch MJ, Gardner RC. Yeast genes involved in sulfur and nitrogen metabolism affect the production of volatile thiols from Sauvignon Blanc musts. Appl Microbiol Biot 2013;97:223–35. Harsch MJ, Lee SA, Goddard MR, et al. Optimized fermentation of grape juice by laboratory strains of Saccharomyces cerevisiae. FEMS Yeast Res 2010;10:72–82. Hebditch KR, Nicolau L, Brimble MA. Synthesis of isotopically labelled thiol volatiles and cysteine conjugates for quantification of Sauvignon Blanc wine. J Labelled Compd Rad 2007;50:237–43. Herbst-Johnstone M, Piano F, Duhamel N, et al. Ethyl propiolate derivatisation for the analysis of varietal thiols in wine. J Chromatogr A 2013;1312:104–10. Howell K, Klein M, Swiegers JH, et al. Genetic determinants of volatile-thiol release by Saccharomyces cerevisiae during wine fermentation. Appl Environ Microb 2005;71:5420–6. Huang C, Roncoroni M, Gardner RC. MET2 affects production of hydrogen sulfide during wine fermentation. Appl Microbiol Biot 2014;98:7125–35.

9

Kaur J, Bachhawat AK. Yct1p, a novel, high-affinity, cysteinespecific transporter from the yeast Saccharomyces cerevisiae. Genetics 2007;176:877–90. Kobayashi H, Takase H, Kaneko K, et al. Analysis of S-3-(Hexan1-ol)-Glutathione and S-3-(Hexan-1-ol)-L-Cysteine in Vitis vinifera L. cv. Koshu for Aromatic Wines. Am J Enol Viticult 2010;61:176–85. Kosugi A, Koizumi Y, Yanagida F, et al. MUP1, high affinity methionine permease, is involved in cysteine uptake by Saccharomyces cerevisiae. Biosci Biotech Bioch 2001;65: 728–31. Ku SY, Yip P, Howell PL. Structure of Escherichia coli tryptophanase. Acta Crystallogr D 2006;62:814–23. Lund CM, Thompson MK, Benkwitz F, et al. New Zealand Sauvignon blanc distinct flavor characteristics: sensory, chemical, and consumer aspects. Am J Enol Viticult 2009;60:1–12. ` Masneuf-Pomarede I, Mansour C, Murat ML, et al. Influence of fermentation temperature on volatile thiols concentrations in Sauvignon blanc wines. Int J Food Microbiol 2006;108: 385–90. Murat ML, Masneuf I, Darriet P, et al. Effect of Saccharomyces cerevisiae yeast strains on the liberation of volatile thiols in Sauvignon blanc wine. Am J Enol Viticult 2001;52: 136–9. Patel P, Herbst-Johnstone M, Lee SA, et al. Influence of juice pressing conditions on polyphenols, antioxidants, and varietal aroma of sauvignon blanc microferments. J Agr Food Chem 2010;58:7280–8. Pinu FR, Jouanneau S, Nicolau L, et al. Concentrations of the volatile thiol 3-mercaptohexanol in Sauvignon blanc wines: no correlation with juice precursors. Am J Enol Viticult 2012;63:407–12. Roland A, Schneider R, Le Guerneve´ C, et al. Identification and quantification by LC–MS/MS of a new precursor of 3mercaptohexan-1-ol (3MH) using stable isotope dilution assay: Elements for understanding the 3MH production in wine. Food Chemistry 2010a;121:847–55. Roland A, Vialaret J, Razungles A, et al. Evolution of S cysteinylated and S -glutathionylated thiol precursors during oxidation of Melon B. and sauvignon blanc musts. J Agricultural and Food Chemistry 2010b;58:4406–13. Roncoroni M, Santiago M, Hooks DO, et al. The yeast IRC7 gene encodes a b-lyase responsible for production of the varietal thiol 4-mercapto-4-methylpentan-2-one in wine. Food Microbiol 2011;28:926–35. Santiago M, Gardner RC. The IRC7 gene encodes cysteine desulfhydrase activity and confers on yeast the ability to grow on cysteine as a nitrogen source. Yeast 2015, DOI: 10.1002/yea.3076. Schneider R, Charrier F, Razungles A, et al. Evidence for an alternative biogenetic pathway leading to 3-mercaptohexanol and 4-mercapto-4-methylpentan-2-one in wines. Anal Chim Acta 2006;563:58–64. Sikorski RS, Hieter P. A system of shuttle vectors and yeast host strains designed for efficient manipulation of DNA in Saccharomyces cerevisiae. Genetics 1989;122:19–27. Subileau M, Schneider R, Salmon J-M, et al. New insights on 3-mercaptohexanol (3MH) biogenesis in Sauvignon blanc wines: Cys-3MH and (E)-Hexen-2-al are not the major precursors. J Agr Food Chem 2008a;56:9230–5. Subileau M, Schneider R, Salmon J-M, et al. Nitrogen catabolite repression modulates the production of aromatic thiols characteristic of Sauvignon Blanc at the level of precursor transport. FEMS Yeast Res 2008b;8:771–80.

10

FEMS Yeast Research, 2015, Vol. 15, No. 0

Swiegers JH, Capone DL, Pardon KH, et al. Engineering volatile thiol release in Saccharomyces cerevisiae for improved wine aroma. Yeast 2007;24:561–74. Thibon C, Marullo P, Claisse O, et al. Nitrogen catabolic repression controls Saccharomyces cerevisiae volatile thiols release during wine fermentation. FEMS Yeast Res 2008;8: 1076–86. Tominaga T, Furrer A, Henry R, et al. Identification of new volatile thiols in the aroma of Vitis vinifera L. var. Sauvignon blanc wines. Flavour Frag J 1998;13:159–62. Tominaga T, Peyrot des Gachons C, Dubourdieu D. A new type of flavor precursors in Vitis vinifera L. cv. Sauvignon blanc: S-cysteine conjugates. J Agr Food Chem 1998;46:5215–9. van Bakel H, Strengman E, Wijmenga C, et al. Gene expression profiling and phenotype analyses of Saccharomyces cerevisiae in response to changing copper reveals six genes with new roles in copper and iron metabolism. Physiol Genomics 2005;22:356–67.

Wakabayashi H, Wakabayashi M, Eisenreich W, et al. Stereochemical course of the generation of 3-mercaptohexanal and 3-mercaptohexanol by b-lyase-catalyzed cleavage of cysteine conjugates. J Agr Food Chem 2004;52:110–6. Winter G, Henschke PA, Higgins VJ, et al. Effects of rehydration nutrients on H2S metabolism and formation of volatile sulfur compounds by the wine yeast VL3. AMB Express 2011a;1:36. Winter G, Van Der Westhuizen T, Higgins VJ, et al. Contribution of cysteine and glutathione conjugates to the formation of the volatile thiols 3-mercaptohexan-1-ol (3MH) and 3mercaptohexyl acetate (3MHA) during fermentation by Saccharomyces cerevisiae. Aust J Grape Wine R 2011b;17:285–90. Wunschmann J, Krajewski M, Letzel T, et al. Dissection of glutathione conjugate turnover in yeast. Phytochemistry 2009;71:54–61. Zott K, Thibon C, Bely M, et al. The grape must nonSaccharomyces microbial community: impact on volatile thiol release. Int J Food Microbiol 2011;151:210–5.