Functional divergence in the proteins encoded by ARO80 from S. uvarum, S. kudriavzevii and S. cerevisiae explain differences in the aroma production during wine fermentation.

Phenylethanol (PE) and phenylethyl acetate (PEA) are commonly desired compounds in wine because of their rose-like aroma. The yeast S. cerevisiae produces the PE either through de novo biosynthesis by shikimate pathway followed by the Ehrlich pathway or the direct phenylalanine catabolism via Ehrlich pathway, and then converted into PEA. Previous work demonstrated that, compared to S. cerevisiae, other Saccharomyces species, such as S. kudriavzevii and S. uvarum, produce higher concentrations of PE and PEA from the precursor phenylalanine, which indicates differential activities of the biosynthetic-involved enzymes. A previous in-silico analysis suggested that the transcriptional activator Aro80p is one of the best candidates to explain these differences. An improved functional analysis identified significant radical amino acid changes in the S. uvarum and S. kudriavzevii Aro80p that could impact the expression of the catabolic genes ARO9 and ARO10, and hence, the production of PE from phenylalanine. Indeed, wine S. cerevisiae strains carrying the S. uvarum and S. kudriavzevii ARO80 alleles increased the production of both compounds in the presence of phenylalanine by increasing the expression of ARO9 and ARO10. This study provides novel insights of the unidentified Aro80p regulatory region and the potential usage of alternatives ARO80 alleles to enhance the PE and PEA concentration in wine.

Introduction

Saccharomyces cerevisiae is frequently used as a starter culture in winemaking since it is adapted to a variety of stresses that occur during fermentation (Pretorius, 2000; Querol et al., 2018). Nowadays, to solve the problems of the production of higher alcohol wines, as a consequence of the adverse effects of climate change and to adjust to the current consumer preference trends, other species of the genus, such as Saccharomyces uvarum and Saccharomyces kudriavzevii, as well as their interspecific hybrids with S. cerevisiae, can be used as suitable alternative starter yeasts (Querol et al., 2018). These yeasts have been associated with wine fermentation and display desired enological traits of interest to solve the global warming effect on the wine production, such as lower ethanol yields, higher glycerol synthesis, good fermentation performance at low temperatures and the generation of interesting aroma profiles (González et al., 2007; Gamero et al., 2013; Pérez‐Torrado et al., 2015). The most important aroma compounds produced by yeast during fermentation are higher alcohols and acetate esters which highly impact wine’s flavour and aroma (Ugliano and Henschke, 2009). Higher (or fusel) alcohols are synthesized by yeast from the catabolism of aromatic, branched and sulphur‐containing amino acids through Ehrlich pathway (Hazelwood et al., 2008; Ugliano and Henschke, 2009; Cordente et al., 2012). These alcohols can subsequently be transformed into their acetate esters by the action of acetyltransferases encoded by the genes ATF1 and ATF2 which, together with the esterase Iah1p, modulate their final concentration (Ugliano and Henschke, 2009; Cordente et al., 2012). Among these compounds, 2‐phenylethanol (PE) and 2‐phenylethyl acetate (PEA) are commonly desired in the wine because of their pleasant rose‐like aroma that positively contributes to the wine’s aromatic profile (Ugliano and Henschke, 2009). PE biosynthesis via Ehrlich pathway (Kim et al., 2014; Qian et al., 2019) begins with the transamination of the aromatic amino acid phenylalanine into phenylpyruvate by the aromatic aminotransferase II Aro9p. Part of the phenylpyruvate also proceeds from the sugar catabolism (Rollero et al., 2019), in which, the glycolytic flux enters to the shikimate pathway to generate chorismite (Gientka and Duszkiewicz‐Reinhard, 2009), and then, transformed into phenylpyruvate that finally enters to the Ehrlich pathway. Then, the phenylpyruvate is decarboxylated to phenylacetaldehyde by the broad‐substrate‐specificity 2‐keto acid decarboxylase Aro10p. Finally, phenylacetaldehyde is reduced into PE through the alcohol dehydrogenases Adh1p to Adh7p together with the formaldehyde dehydrogenase Sfa1p (Cordente et al., 2012). PE is then either excreted to the medium or converted into PEA by acetyltransferases Atf1p and Atf2p. (Stribny et al., 2015) reported that S. kudriavzevii and S. uvarum species produce higher concentrations of PE and PEA, respectively, from the aromatic amino acid phenylalanine than S. cerevisiae. Their results suggest differential activities of the enzymes involved in the aromatic compound biosynthetic pathways. To determine the molecular foundations that explain these differences, an in‐silico analysis based on Grantham’s score (Grantham, 1974) determined that Aro10p, Atf1p and Atf2p were good candidates because of their high scores (Stribny, 2016). This method quantifies how similar or dissimilar are two amino acids residues based on their physicochemical properties such as composition, polarity and molecular volume to predict their evolutionary distance. High values implies that both amino acids are evolutionary unlikely to be substituted with one another. Such replacements in a protein could potentially generate functional changes in their activity.

Previous studies demonstrated that S. kudriavzevii Aro10p and both S. kudriavzevii/S. uvarum Atf1p and Atf2p versions showed differential substrate preferences and activities compared to those S. cerevisiae counterparts (Stribny et al., 2016a,2016b). Both S. kudriavzevii and S. uvarum Atf2p increased twofold activity against the PE compared to the S. cerevisae Atfp2. In addition, the Km of S. kudriavzevii and S. uvarum Atfp1 are twofold and threefold higher than S. cerevisiae Atf1p using isoamyl alcohol as substrate respectively. Besides, The Vmax value of both Atf1p non‐cerevisiae species was two times lower than S. cerevisiae Atf1p. Moreover, these alleles changed the final aromatic compound composition in synthetic wine fermentations when they were expressed into an S. cerevisiae background (Stribny et al., 2016a,2016b). Another protein that showed a high Grantham’s score in both S. uvarum and S. kudriavzevii species was Aro80p (Stribny, 2016), which belongs to the zinc binuclear proteins family (Iraqui et al., 1999; MacPherson et al., 2006). Aro80p is constitutively bound to the cis UASARO elements of both ARO9 and ARO10 genes (Lee and Hahn, 2013) and specifically induces their expression in the presence of aromatic amino acids (Iraqui et al., 1999; Godard et al., 2007).

Since Aro80p is directly related to the Ehrlich pathway, this study aimed to test the effect of ARO80 S. kudriavzevii/S. uvarum alleles on the production of PE and PEA from phenylalanine compared to the S. cerevisiae allele. To attain this, we improved the functional divergence analysis based on the method developed by Macías et al. (2019), who refined the functional divergence method developed by Toft et al. (2009) by quantifying divergences according to Grantham’s scores. This method allowed us to identify candidate amino acid changes in Aro80p that might explain the observed differences in the synthesis of PE and PEA. To demonstrate these observations, we generated S. cerevisiae strains carrying the ARO80 S. kudriavzevii/S. uvarum alleles and we cultured them in the presence of precursor phenylalanine to decipher their impact on the expression of the genes ARO9 and ARO10 regulated by Aro80p.

Experimental procedures

Aro80p protein functional divergence analysis

ARO80 gene sequences from representative Saccharomyces isolates (Table 1) were used for the functional divergence analysis. The method, described elsewhere (Macías et al., 2019), was used to identify amino acids in the predicted Aro80p sequences from S. uvarum (Aro80p‐Su) and S. kudriavzevii (Aro80p‐Sk) that have diverged significantly from the Torulaspora delbrueckii output orthologue in these two species with respect to the homologous site in the S. cerevisiae sequences (Aro80p‐Sc). Once all divergent amino acid sites were obtained, results were filtered by Grantham’s scores (Grantham, 1974), to quantify the biochemical divergence between S. uvarum‐S. kudriavzevii and S. cerevisiae amino acids. Those showing scores equal to or higher than 120 were considered as radical changes (Stribny, et al., 2016; Macías et al., 2019).

Table 1

Strains used for the bioinformatics analysis and sources of the genomic sequences.

Strain	Specie	References
T73	Saccharomyces cerevisiae	Morard et al. (2019)
S288C	Saccharomyces cerevisiae	Goffeau et al. (1996)
Y9	Saccharomyces cerevisiae	Liti et al. (2009)
YPS128	Saccharomyces cerevisiae	Liti et al. (2009)
BMV58	Saccharomyces uvarum	Macías et al. (2021)
NPCC1290	Saccharomyces uvarum	Macías et al. (unpublished)
CECT12600	Saccharomyces uvarum	Macías et al. (2021)
CBS7001	Saccharomyces uvarum	Scannell et al. (2011)
CR85	Saccharomyces kudriavzevii	Macías et al. (2019)
ZP591	Saccharomyces kudriavzevii	Scannell et al. (2011)
CA111	Saccharomyces kudriavzevii	Macías et al. (2019)
IFO1802	Saccharomyces kudriavzevii	Scannell et al. (2011)
CBS1164	Torulaspora delbrueckii	Gordon et al. (2011)

Yeast strains and growth conditions

The parental and engineered strains used in this study are listed in Table 2. The transformant aro80 mutant and ARO80 recombinant strains were grown at 30°C on selective YPD solid media (1% yeast extract, 2% peptone, 2% glucose, 2% agar) containing, respectively, 200 µg ml‐1 G418 and 100 µg ml‐1 nourseothricin. For aroma compound determination and the gene expression experiments, the cell was grown in YNB liquid media (0.17% yeast nitrogen base without amino acid and ammonium, 2% glucose) containing the desired amino acid.

Table 2

Strains used in this study.

Strain	Genotype	References
AQ2775	T73 MATalpha	This study
AQ2901	BMV58 MATa	This study
AQ4013	CR85 MATalpha ho::MX4dsdA	This study
ST44	MATalpha, aro80::KanMX	This study
ST44‐Sc	MATalpha, aro80::kanmx::ARO80(T73)	This study
ST44‐Su	MATalpha, aro80::kanmx::ARO80(BMV58)	This study
ST44‐Sk	MATalpha, aro80::kanmx::ARO80(CR85)	This study

Strains constructions

The deletion of the ARO80 open reading frame (ORF) in the haploid strain AQ2775, that derives from the wine strain T73 (Querol et al., 1992), was carried out through PCR‐mediated gene disruption using KanMX cassette as a selection marker (Baudin et al., 1993), which was PCR amplified from the pUG6 plasmid (Güldener et al., 1996) using NZYTaq II DNA Polymerase (NZYTech, Lisbon, Portugal) following the provided instructions. The strains were transformed through the lithium acetate method (Gietz and Schiestl, 2007) and deletions were confirmed by PCR using the diagnostic primers (Table 3). The aro80 mutant strain was used as the parental strain to generate the recombinant ARO80 strains (Fig. 1, Table 2) through allele swapping by CRISPR‐Cas9‐mediated gene disruption (Stovicek et al., 2017). The protospacer sequence against the KanMX cassette was designed according to Doench et al. (2014) using the T73 genome sequence as reference. Then, the entire plasmid pRCC‐N was amplified with primers carrying the protospacer sequence at their 5’ ends by PCR (Generoso et al., 2016) which was carried out with PhusionTM High‐Fidelity Polymerase (Thermo Fisher Scientific, Vilnius, Lithuania) following the provided instructions. Before the addition to the transformation mix, 30 µl of the PCR product were treated with 10 U of DpnI (Thermo Fisher Scientific, Vilnius, Lithuania) for 3 h to guarantee the degradation of the pRCC‐N plasmid template. The KanMX cassette at ARO80 locus was swapped with the ARO80 ORF alleles from the S. uvarum strain AQ2901, S. kudriavzevii strain AQ4013 and S. cerevisiae strain AQ2775, which were amplified by PCR using PhusionTM High‐Fidelity Polymerase. The ORF allele swapping was confirmed by PCR analysis of total DNA extracted from nourseothricin‐resistant transformant strains whose resistance against G418 antibiotic was lost and whose ability to grow in the presence of tryptophan as the sole nitrogen source was recovered (Iraqui et al., 1999). The reinserted ARO80 alleles sequences were checked through Sanger sequencing from Eurofins Genomics Mix2Seq service.

Table 3

Primers used in this study.

Primer	Sequence (5’ to 3’)
Primers for ARO80 gene disruption
aro80Δ‐Fw(1)	GCATAATAAGGTTACATTAAGCACTGCTTTATCCTCTATGTAGAGATCTGTTTAGCTTGCCT
aro80Δ‐Rv(1)	GCGGTTGTCTTGGTTGATGACGTAATTCTTTGATATCTACGTTTTCGACACTGGATGGC
Primers for ARO80 allele swapping
Protospacer incorporation in pRRC‐N plasmid
gRNA‐KanMX‐Fw(2)	TGTTTTGCCGGGGATCGCAGGAATTCTTTGATATCTACTTTTAGAGCTAGAAATAGCAAGTTAAAATAAGG
gRNA‐KanMX‐Rv(2)	CTGCGATCCCCGGCAAAACAGATCATTTATCTTTCACTGCGGAG
Primer for ARO80 alleles amplification as donor DNA
ARO80‐Sc‐Fw	TCCACGCATAATAAGGTTACAT
ARO80‐Sc‐Rv	ATTTTTACGAATAGTGCGGTTG
ARO80‐Sk‐Fw(3)	GCATAATAAGGTTACATTAAGCACTGCTTTATCCTCTATGTCTCCTAAGAGAAGATCC
ARO80‐Sk‐Rv(3)	GCGGTTGTCTTGGTTGATGACGTAATTCTTTGATATCTACTTATTTACGTGTTACTGGCC
ARO80‐Su‐Fw(3)	GCATAATAAGGTTACATTAAGCACTGCTTTATCCTCTATGTCTACTACCAAAAGAGGATCC
ARO80‐Su‐Rv(3)	GCGGTTGTCTTGGTTGATGACGTAATTCTTTGATATCTACTTATTGGTGTGCAGCTGGC
continuation
Diagnostic primers
ARO80‐Sc‐test‐Fw	TTTCGGAATCAACGAGAGTACA
ARO80‐Sc‐test‐Rv	TTTTCAGTGGTTTCGGTGTT
ARO80‐Sk‐test‐Rv	ATCTGCTGTTCACTTTTGCT
ARO80‐Su‐test‐Rv	CTGTTGGAACTAAAGAGACATC
K2	GGGACAATTCAACGCGTCTG
Primers for relative quantification through qPCR
ACT1‐F	CTTACAACTCCATCATGAAGTGTGA
ACT1‐R	ATTTCCTTTTGCATTCTTTCGGC
18S‐F(4)	TTGCGATAACGAACGAGACC
18S‐R(4)	CATCGGCTTGAAACCGATAG
ARO9‐F	CCGTGTCATCCGTTTGGAGA
ARO9‐R	TGGACTCAGCCATTGCCTTT
ARO10‐F	GATTTCGCGTTTCCTTCGCA
ARO10‐R	AATTCCAACACCTAGGGCGG
Sequencing primers
ARO80‐Sc‐test‐Rv2	TTTTGATTCCTATGGCTCCTAG
ARO80‐Sc‐Seq_1	TAATGGAAGAAATCGGGAAAGT
ARO80‐Sc‐Seq_2	AAACGATGGAAATGAAAGCAAT
ARO80‐Sc‐Seq_3	CGAGGTATGTGGAATTAGCATA
ARO80‐Sc‐Seq_4	CCCCTGTTATTCTGTCTTTGTA
ARO80‐Su‐Seq_1	TAGCACAGAACAAAATGGGATA
ARO80‐Su‐Seq_2	TAAAGAAGGTGCGAAGGAAATA
ARO80‐Su‐Seq_3	AAACCTTTCATTCCCGAGATAA
ARO80‐Su‐Seq_4	TCTCGGGAATGAAAGGTTTTAT
ARO80‐Su‐Seq_5	ATCAACAGATATTGCACTCAGT
ARO80‐Su‐Seq_6	TCTTGGTTGATGACGTAATTCT
ARO80‐Sk‐Seq_1	CTGCTTTATCCTCTATGTCTCC
ARO80‐Sk‐Seq_2	ACCCATTTTCTCCGTTATACAA
ARO80‐Sk‐Seq_3	TTGTGCAAACGAGAATCTTAAC
ARO80‐Sk‐Seq_4	ACAGATATTGCACTCAGTTTTA
ARO80‐Sk‐Seq_5	CCCCTGTTATTCTGTCTTTGTA

(1) Homologous sequences to KanMX from the pUG6 plasmid are underlined. (2) The underlined sequences stand for the protospacers. (3) Underlined sequences stand for the specific ARO80 alleles sequences. (4) Primer sequences from (Pérez‐Torrado et al., 2016).

Fig. 1

Two‐step ARO80 allele swapping approach used for recombinant strain generation. (1) The ARO80 open reading frame was deleted by PCR‐mediated gene disruption using KanMX cassette. (2) The marker was replaced by CRISPR‐Cas9 mediated gene disruption and the amplified ARO80 alleles were used as donor DNA to ensure homologous recombination.

Quantification of 2‐phenylethanol and 2‐phenylethyl acetate production

To test PE and PEA production, the cells were incubated overnight in YNB liquid media containing 5 g l‐1 ammonium sulphate as nitrogen source and then inoculated in 80 ml of YNB liquid media containing either 12.5 g l‐1 phenylalanine or 5 g l‐1 ammonium sulphate as the sole nitrogen source (Stribny et al., 2015) starting with 1 × 106 cells ml‐1. Medium samples were collected after 48 h of cell growth when all glucose is depleted. To quantify the PE and PEA production in each sample, 5 ml of sample was mixed with 4.95 ml of 303 g l‐1 NaCl solution and 50 µl of 3‐octanol as internal standard. Then passed it through a TRACETM GC Ultra gas chromatograph (Thermo Fisher Scientific, Waltham, MA, USA) coupled with a flame ionization detector (FID), equipped with a 30 m × 0.25 mm × 0.25 µm HP‐INNOWax capillary column coated with a layer of cross‐linked polyethylene glycol (Agilent Technologies, Santa Clara, USA) at carrier gas helium flow rate of 1 ml min‐1. The oven temperature programme was: (i) five minutes at 50°C, (ii) temperature raised to 100°C at an increasing rate of 1.5°C min‐1. (iii) then up to 215°C at a rate of 3°C min‐1 and (iv) was kept for 2 min more. The FID detector temperature was 280°C and the aromatic compounds were identified by their retention time. Quantification was made by using calibration plots of the corresponding compounds.

ARO9 and ARO10 gene expression analysis by real‐time qPCR

To monitor the expression pattern of both ARO9 and ARO10 genes in the recombinant ARO80 strains, samples of cells growing in YNB 12.5 g l‐1 phenylalanine liquid media were taken at different times (10, 24, 30 and 48 h), frozen in liquid nitrogen and stored at −80°C until RNA extraction. The aro80 mutant strain ST44 was used as a negative expression control. Total RNA was extracted using a Qiagen RNA extraction kit (Qiagen, Hilden, Germany) following the provided instructions. The RNA samples were treated with 10 U of DNAse I (Roche, Mannheim, Germany) and 1 µg was used to generate cDNA using NZY First‐Strand cDNA synthesis kit (NYZTech, Lisbon, Portugal). The qPCR experiments were carried in the LightCycler® 480 Instrument (Roche,Mannheim, Germany). Expression of both ARO9 and ARO10 genes in every sample was normalised against the average expression of the housekeeping genes ACT1 and ribosomal 18S rRNA. Afterwards, fold change expression in each sample value was determined as the binary logarithm (log‐2) of expression in one sample respect to the average expression of all samples at time 10 h.

Statistical analyses

All One‐way ANOVA and Tukey’s multiple comparisons analyses were performed using GraphPad Prism version 8.01 for Windows 10, GraphPad Software, La Jolla California USA (www.graphpad.com).

Results

S. uvarum and S. kudriavzevii show a high number of radical amino acid changes in Aro80p

A previous study demonstrated high functional divergence in Aro80p amino acid sequences in the S. uvarum and S. kudriavzevii species compared to S. cerevisiae based on Grantham’s scoring (Stribny, 2016). In this study, we have used a different approach taking advantage of a new method, recently described (Macías et al., 2019), to identify both evolutionary and functional significant radical amino acid changes in the Aro80p. These results (Fig. 2, Table S1) highlight the change D45A45/56 which is shared by both species and located at the C6 zinc‐finger of the DNA‐binding domain region. Furthermore, the Aro80p‐Sk exhibits four changes in the middle region where the change N502Y506 is at a similar position to the N502C517 showed by Aro80p‐Su, although the function of that region remains unclear. Other changes in Aro80p‐Su are widespread along the protein sequence. The changes R6G17, S114L125 and N170C181 are located at the N‐terminal end of the zinc finger region, at the linker and the dimerization domain of the protein respectively. The change N939I954 is located between the C‐terminal end and the acidic region, essential for recruiting the transcriptional machinery. Finally, the amino changes Y532C367 and G690C705 are in regions of unknown functions. Altogether, these changes could explain differences in the function of Aro80p‐Sk and Aro80p‐Su with respect to the S. cerevisiae protein.

Fig. 2

Radical amino acids changes were observed in Aro80p comparing different Saccharomyces species.

A. Aro80p functional domains identified in S. cerevisiae by (Schjerling and Holmberg, 1996; Iraqui et al., 1999).

B. Detected radical amino acid changes in S. uvarum (Su) and S. kudriavzevii (Sk) compared to S. cerevisiae (Sc). T. delbrueckii (Td) sequence was used as an outgroup for the functional divergence analysis. Subscript numbers indicate the original amino acid position for each species. The alignment positions are relative to Su‐Sc‐Td and Sk‐Sc‐Td alignments respectively. Grantham’s scores (Grantham, 1974) of the identified significant radical amino acid changes are shown.

Effect of ARO80 alleles on phenylethanol and phenylethyl acetate production

To determine the impact of the different ARO80 alleles on the production of PE and PEA, we generated recombinant strains carrying the S. uvarum (ST44‐Su), S. kudriavzevii (ST44‐Sk) and S. cerevisiae (ST44‐Sc) ARO80 alleles. The parental aro80 mutant strain (ST44) was used as a control. Strains were grown in a minimal medium containing either ammonium sulphate or the aromatic precursor phenylalanine and we quantified PE and PEA production after 48 h when all glucose was consumed (Fig. 3, Table S2). We observed that both recombinant strains ST44‐Su and ST44‐Sk increased the PE basal production by 11.2% and 13.4% compared to ST44‐Sc strain in the presence of ammonium sulphate as the sole nitrogen source respectively. While the strain ST44 showed a 28.3% reduction of PE production. However, no PEA production was detected by any strain in this condition. Meanwhile, the strains ST44‐Su and ST44‐Sk increased the PE production by 12.7% and 13.1% and PEA production by 29% and 32.2%, respectively, compared to ST44‐Sc in the presence of the phenylalanine as nitrogen source. In contrast, the strain ST44 showed a 63.9% and 91.3% reduction of PA and PEA production, respectively, compared to ST44‐Sc. These data confirm the importance of the ARO80 gene in PA and PEA production and demonstrate that S. uvarum and S. kudriavzevii ARO80 alleles induce both PE and PEA increased production from the precursor phenylalanine.

Fig. 3

Production of phenyl ethanol (PE) and phenylethyl acetate (PEA) by the recombinant ARO80 strains in minimal medium. Either ammonium sulphate (NH4) or phenylalanine (Phe) was used as sole nitrogen sources. Error bars represent the standard deviation from three biological replicates. Statistical differences were determined through ANOVA analysis independently for each nitrogen source. p‐values for Tukey’s comparisons test are indicated. ND, Not detected.

Effect of the ARO80 alleles in the ARO9 and ARO10 expression profiles

The results showed above might indicate differences in the regulation of the Aro80p target genes ARO9 and ARO10 by the non‐cerevisiae ARO80 alleles. Therefore, we tested the effect of the ARO80 alleles on their expression pattern profiles in the presence of phenylalanine by taking samples at different times after cells were inoculated into YNB 12.5 g l‐1 phenylalanine medium (Fig. 4, Table S3). Indeed, we observed that the expression of both ARO9 and ARO10 genes in the strain ST44‐Su and ST44‐Sk are two‐fold higher than in the strain ST44‐Sc after 10 h. However, no differences were observed between the ARO80 recombinant strains after 24 h, except for a higher expression of ARO9 in the strain ST44‐Sk compared to ST44‐Sc at 30 h. In contrast, and as expected, the mutant strain ST44 expressed both ARO9 and ARO10 genes but unable to induce their expression in the presence of phenylalanine during the whole experiment.

Fig. 4

Expression pattern of ARO9 and ARO10 genes of the ARO80 recombinant strains cultured in minimal medium containing phenylalanine as the sole nitrogen source. Fold change was determined as the log‐2 base logarithm of expression in one sample normalised against the average expression obtained at 10 h. Error bars represent the standard deviation from three biological replicates. Statistical differences were determined through ANOVA analysis independently for each time. p‐values for Tukey’s comparisons test are indicated.

Discussion

Previous studies have demonstrated that S. uvarum and S. kudriavzevii produce higher amounts of PE and PEA than S. cerevisiae from the aromatic amino acid precursor phenylalanine when it was used as the sole nitrogen source (Stribny et al., 2015), suggesting differences in the activity of the pathways involved in the production of aromatic compounds. An in‐silico analysis based on Grantham’s scoring plus experimental validation have demonstrated that S. kudriavzevii Aro10p and S. kudriavzevii/S. uvarum acetyltransferases Atf1p and Atf2p contains significant amino acid changes which produced differences in their activity, substrate affinity and impacted the wine’s aroma profile when they were expressed in a wine S. cerevisiae strain (Stribny et al., 2016a,2016b). The binuclear cluster protein Aro80p showed the highest divergent scores in both S. uvarum and S. kudriavzevii (Stribny, 2016).

Since Aro80p regulates the expression of both Ehrlich pathway genes ARO9 and ARO10 (Iraqui et al., 1999; Godard et al., 2007), we studied the effect of the S. uvarum and S. kudriavzevii ARO80 alleles (ARO80‐Su and ARO80‐Sk) on the PE and PEA production from the phenylalanine precursor. When these alleles were expressed in the wine S. cerevisiae strain T73, the PE and PEA production was increased compared to the S. cerevisiae wild‐type allele (ARO80‐Sc). Considering the regulatory function of Aro80p, we analysed the expression pattern of the Aro80p‐regulated genes ARO9 and ARO10 at different times. It has been reported that overexpression of both genes increases the PE titer (Kim et al., 2014). Indeed, the ARO80‐Su and ARO80‐Sk alleles induced higher expression levels of those genes in short times than the ARO80‐Sc allele, which correlates with increased PE and PEA productions. Because only ARO80 ORFs were swapped, their ability to induce higher expression levels might be explained by the identified amino acid changes. Aro80p belongs to the zinc cluster proteins which are found exclusively in fungi and exhibit characteristic functional domains shared by the members of this family (Iraqui et al., 1999; MacPherson et al., 2006). However, since functional studies in ARO80 have not been conducted yet, we came out with some suggestions based on published data. The substitution D45A45/56 shared by both non‐cerevisiae species is located at the loop that separates the two cysteine‐rich substructures of the DNA‐binding domain (DBD) metal‐binding portion of the protein. Although Aro80p is constitutively bound to the UASARO elements independently of the nitrogen source (Lee and Hahn, 2013) and the DBD is related only to the ability of the protein to bind to its cis element, it has been reported that similar changes in the regulatory protein Leu3p could improve not just the DNA‐binding activity but also the transcriptional activity and therefore an increased expression in vivo of the regulated genes in the presence of the inducer (Bai and Kohlhaw, 1991). We observed many changes located in the middle region, between the DBD and activation domain (AD) in the proteins Aro80p‐Su (Y352C367, N502C517 and G690C705) and Aro80p‐Sk (S482I486, R488G492, D496A500 and N502Y506). In most of the zinc cluster proteins, the region between these two domains contains a motif referred to as the middle homology region (MHR) and it is believed that it regulates the transcriptional activity of the protein (Schjerling and Holmberg, 1996; MacPherson et al., 2006). Although Iraqui et al. (1999) reported a recognizable part of this motif in Aro80p between the positions 370 and 412, it seems to be absent and remains unclear which are the regulatory regions.

Some of the zinc cluster proteins that regulate the genes involved in the catabolism of specific amino acids act also as nutrient sensors (Sellick and Reece, 2005; MacPherson et al., 2006). A well‐reported case is the protein Put3p, which upregulates the expression of the proline catabolite genes PUT1 and PUT2 in the presence of the inducer proline as the sole nitrogen source. The proposed mechanism (Sellick and Reece, 2005) involves the interaction between Put3p and the proline to induce conformational changes which, in turn, unmask the AD that recruits the proteins to initiate the transcription. Indeed, it has been demonstrated that Put3p binds directly to proline through its pyrrolidine ring and then it induces the transcription of genes containing Put3p‐binding sites (Sellick and Reece, 2003). A similar model has been hypothesised for the regulator Leu3p (Kohlhaw, 2003). Because the aromatic amino acids are the inducers of Aro80p, the same activation mechanism might be conserved, but a direct interaction has not been demonstrated yet (Lee and Hahn, 2013). The substitution G532R in Put3p impairs the activation by proline without affecting the activity of other domains but an additional R764T substitution not just recovers but increases the activity of Put3p, which becomes proline insensitive (Ann des Etages et al., 1996). Other substitutions in the middle region have an impact on the AD masking which could produce either a permanent or loose unmasking that increases the transcription activity (Kohlhaw, 2003). Since the substitutions found in the Aro80p middle region are located at similar positions relative to the reported proteins, this suggests that such changes could affect the transcriptional activity by either changing a potential interaction between Aro80p and the phenylalanine or diminishing the AD masking. The substitution N939I954 is located at the C‐terminal end of the AD in the Aro80p‐Su and it probably exerts a minimal effect since the acidic/hydrophobic core (Schjerling and Holmberg, 1996) between positions 899 and 925 is conserved, and deletion experiments in Gal4p and Leu3p have demonstrated that the last twelve and nine residues, respectively, are dispensable for their transcriptional activities (Leuther et al., 1993; Wang et al., 1997). These results and suggested observations indicate that the different Aro80p proteins of this study could be used as a model to identify the functional domains of the protein and promote future studies addressing how the different identified amino acid changes affect the Aro80p transcriptional activity.

Conclusions

In this study, we demonstrated that alleles ARO80‐Su and ARO80‐Sk increased both PEA and PE production compared to the strain carrying ARO80‐Sc by enhancing the expression of the ARO9 and ARO10 genes in the presence of the inducer phenylalanine. We also identified candidate amino changes between Aro80p proteins that might explain differences in the expression of their target genes ARO9 and ARO10. Finally, we propose that changes in the ARO80 gene can be interesting to characterize the Aro80p regulatory domains, and with that, the developing of novel strains showing higher expression in ARO9 and ARO10 genes and, hence, increasing the PE and PEA production in wine.

Conflict of interest

None declared.

Fig. S1. Amplified KanMX cassette from plasmid pUG6 used to disrupt the ARO80 gene in AQ2775 background. The PCR was carried out using the primers aro80Δ‐Fw and aro80Δ‐Rv. Ld: GeneRulerTM 1kb DNA ladder (Thermo Fisher Scientific, Waltham, MA, USA)

Click here for additional data file.

Fig. S2. Confirmation PCR of ARO80 gene disruption. (Left): PCR performed using the diagnosis primers ARO80‐Sc‐test‐Fw and K2. (Right): PCR carried out using the diagnosis primers ARO80‐Sc‐test‐Fw and ARO80‐Sc‐test‐Rv. C1‐5: aro80∆ colonies, WT: Parental strain AQ2775, B: Negative control (water), Ld: GeneRulerTM 1kb DNA ladder (Thermo Fisher Scientific, Waltham, MA, USA).

Click here for additional data file.

Fig. S3. pRCC‐N PCR product containing gRNA for KanMX. PCR was carried out using the protospacer incorporation primers gRNA‐KanMX‐Fw and gRNA‐KanMX‐Rv. Plasmid pRCC‐N was used as a template. Ld: GeneRulerTM 1kb DNA ladder (Thermo Fisher Scientific, Waltham, MA, USA).

Click here for additional data file.

Fig. S4. ARO80 alleles PCR products containing homologous sequences to S. cerevisiae ARO80 locus. The PCR was carried out using the pair primers ARO80‐Sc‐Fw/ARO80‐Sc‐Fw, ARO80‐Sk‐Fw/ARO80‐Sk‐Fw and ARO80‐Su‐Fw/ARO80‐Su‐Fw to amplify the ARO80 alleles from total DNA of the strains AQ2775, AQ4013, and AQ2901 respectively. Ld: GeneRulerTM 1kb DNA ladder (Thermo Fisher Scientific, Waltham, MA, USA).

Click here for additional data file.

Fig. S5. Confirmation PCR of ARO80 allele swapping. The PCR confirmed the presence of ARO80 alleles at the S. cerevisiae ARO80 locus. The reverse primers are specific for each allele and a PCR for the KanMX cassette at the ARO80 locus was done as control. C1‐3: ARO80(Sc), C4‐5: ARO80(Su), C6‐9: ARO80(Sk), C10: aro80[ORF]::KanMX, C11: AQ2775, Ld: InvitrogenTM 100 bp DNA ladder (Thermo Fisher Scientific, Waltham, MA, USA).

Click here for additional data file.

Fig. S6. Integrated ARO80 alleles amplification for Sanger sequencing. The PCR was carried out using the pair primers ARO80‐Sc‐test‐Fw/ARO80‐Sc‐Rv2. The strains C2 (ST44‐Sc), C4 (ST44‐Su) and C6 (ST44‐Sk) were confirmed as positive transformants and used for further experiments. C1‐3: ARO80(Sc), C4‐5: ARO80(Su), C6‐9: ARO80(Sk), C10: aro80[ORF]::KanMX, C11: AQ2775, Ld: GeneRulerTM 1kb DNA ladder (Thermo Fisher Scientific, Waltham, MA, USA).

Click here for additional data file.

Table S1. Aro80p functional analysis results.

Click here for additional data file.

Table S2. Phenylethanol and phenylethyl acetate quantification data.

Click here for additional data file.

Table S3. ARO9 and ARO10 relative quantification data.

Click here for additional data file.