Molecular cloning, sequence analysis, and expression of the yeast alcohol acetyltransferase gene

The ATF1 gene, which encodes alcohol acetyltransferase (AATase), was cloned from Saccharomyces cerevisiae and brewery lager yeast (Saccharomyces uvarum). The nucleotide sequence of the ATF1 gene isolated from S. cerevisiae was determined. The structural gene consists of a 1,575-bp open reading frame that encodes 525 amino acids with a calculated molecular weight of 61,059. Although the yeast AATase is considered a membrane-bound enzyme, the results of a hydrophobicity analysis suggested that this gene product does not have a membrane-spanning region that is significantly hydrophobic. A Southern analysis of the yeast genomes in which the ATF1 gene was used as a probe revealed that S. cerevisiae has one ATF1 gene, while brewery lager yeast has one ATF1 gene and another, homologous gene (Lg-ATF1). Transformants carrying multiple copies of the ATF1 gene or the Lg-ATF1 gene exhibited high AATase activity in static cultures and produced greater concentrations of acetate esters than the control.

Acetate esters, such as isoamyl acetate and ethyl acetate, are recognized as important flavor compounds in beer and other alcoholic beverages. It has been suggested that alcohol acetyltransferase (AATase) (EC 2.3.1.84) is one of the most important enzymes for acetate ester formation. AATase is an SH enzyme which reacts with acetyl coenzyme A and, depending on the degree of affinity, with various kinds of alcohols (22). The activity of this enzyme is strongly repressed under aerobic conditions or by the addition of unsaturated fatty acids to a culture (10,13,22).
As acetate esters affect the flavor quality of alcoholic beverages, many workers have attempted to clone the AATase gene in order to understand the mechanism of acetate ester synthesis and to control ester production (11,13,22).
Recently, we succeeded in purifying this enzyme from Saccharomyces cerevisiae to homogeneity and determined its internal peptide sequences (13). The molecular weight of this enzyme was estimated to be about 60,000. In this paper, we describe the cloning of the AATase-encoding gene. We cloned the gene from two yeast phage libraries, one constructed with S. cerevisiae DNA and the other constructed with DNA from brewery lager yeast (Saccharomyces uvarum). The nucleotide sequence of the ATF gene isolated from S. cerevisiae revealed that the molecular weight of the encoded protein is 61,059. A Southern analysis of yeast genomes revealed that S. cerevisiae has oneA TFl gene, but brewery lager yeast has oneATF1 gene and another homologous gene. We also obtained expression of these genes by using a multicopy plasmid. The resulting transformants exhibited 6-to 15-fold-greater AATase activity than the control. The concentrations of acetate esters present in cultured supernatants obtained from transformant cultures were also greater than the concentrations present in the control.

MATERIALS AND METHODS
Strains and plasmid. S. cerevisiae Kyokai No. 7 (=K7 = IFO 2347) was obtained from the Institute for Fermentation, Osaka, Japan. Brewery lager yeast strain KBY001 (S. uvarum) was obtained from our culture collection. S. cerevisiae TD4 (a his4-519 ura3-52 leu2-3 leu2-112 trp can) was used for transformation and for expression of the ATF multicopy plasmid. Escherichia coli DH5 (endAI gyrA96 hsdR17 recAl relA1 supE44 thi-1) was used as a host for plasmid construction. Plasmid YEp13K, which is described elsewhere (19), was used as a vector for yeast plasmid construction.
Genomic DNA isolation. Yeast genomic DNA for library construction was isolated as described by Rose and Broach (16). Small-scale preparations of yeast DNA were obtained by the method of Rose et al. (17).
Construction and screening of the libraries. Yeast genomic DNA was partially digested with Sau3AI to give fragments with an average size of 15 to 20 kb. These fragments were ligated into the BamHI site of the X-EMBL3 vector (Stratagene) and packaged in vitro (Stratagene). E. coli P2392 {F-galK2 galT22 hsdR514 [rk-Mk+] X-lacYI or A(lacIZY)6 mcrA mcrB+ metBl P2 supE44 supF58 trpR55} was infected with the recombinant phage.
Two peptides were chosen for construction of two oligonucleotide mixed probes. The oligonucleotide mixed probes were synthesized with an Applied Biosystems model 380B instrument. The oligonucleotides were purified with oligonucleotide purification cartridges (Applied Biosystems) and were labelled with [y-32P]ATP and T4 polynucleotide kinase. For the initial screening, 30,000 recombinant phages from the X-EMBL3 library were plated onto E. coli P2392. Duplicate transfers of the clones were made onto nylon membranes. The filters were prehybridized for 3 h at 60°C and hybridized for 18 h at 30°C in a solution containing 6x SSPE (1 x SSPE is 0.18 M NaCI, 10 mM NaH2PO4, and 1 mM EDTA [pH 7.7]), 5 x Denhardt's solution, 0.5% sodium dodecyl sulfate (SDS), and 10 ,ug of salmon sperm DNA (Sigma) per ml. Approximately 500,000 cpm of 5'-end-labelled oligonucleotide mixture per filter was used. The filters were washed twice with a solution containing 2x SSC (lx SSC is 0.15 M NaCl plus 0.015 M sodium citrate) for 30 min at 30°C.
To clone the ATF1 gene from brewery lager yeast strain KBYOO1, a 0.4-kb ClaI-EcoRI fragment of the S. cerevisiae ATF1 gene was used as a probe after labelling with [ax-32P]dCTP using a multiprime labelling kit. The filters were hybridized at 65°C for 18 h, washed twice with a solution containing 2x SSC and 0.1% SDS, and then washed twice with 0.5 x SSC for 30 min at 65°C.
Transformation. E. coli transformation was carried out as described by Hanahan (5), using frozen competent cells obtained from Toyobo, Osaka, Japan. Transformation of S. cerevisiae strains was carried out by the lithium acetate procedure (8 (21) were used to clone DNA fragments for sequencing. Sequencing was performed by the dideoxy chain termination method (18), using a Bst DNA polymerase sequencing kit obtained from Bio-Rad Laboratories according to the supplier's instructions. Chemically synthesized sequencing primers were used when it was not practical to use the M13 universal primer. The sequences obtained were analyzed with the DNASIS program (Hitachi Software Engineering Co., Ltd., Yokohama, Japan).
AATase assay. To measure the AATase activity of yeast cells, cultures were grown in 100 ml of YM15 medium in 500-ml flasks at 30°C. The cultures were grown either with shaking (120 rpm) or under static conditions. To prepare a yeast cell extract, all procedures were performed at 4°C or on ice. Yeast cells were harvested and washed once with distilled water and once with buffer A (25 mM imidazole-HCl, 0.1 M NaCl, 20% glycerol, 1 mM dithiothreitol, 0.1% Triton X-100, 0.5% isoamyl alcohol; pH 7.5). The yeast pellet was resuspended in 0.25 ml of buffer A. Then 1.5 g of glass beads was added, and the mixture was vortexed. The beads were washed three times with 0.25 ml of buffer A, and then the cell suspensions were centrifuged at 15,000 x g for 20 min to remove any unbroken cells. AATase activity was measured as previously described (12). The protein concentration in the homogenate was determined with a protein assay kit (Bio-Rad Laboratories).
Nucleotide sequence accession number. The nucleotide sequence determined in this study has been deposited in the GSDB, DDBJ, EMBL, and NCBI databases under accession number D26554.

RESULTS
Cloning of the S. cerevisiae ATF gene. The method used to clone the AATase-encoding gene utilized oligonucleotide probes constructed on the basis of the amino acid sequences obtained for purified S. cerevisiae Kyokai No. 7 AATase. The peptide sequences that were determined are shown in Fig. 1. Two synthetic oligonucleotides, probes 2 and 5 ( Fig. 1), were prepared and used for the initial screening.
We screened a total of 30,000 recombinants from the X-EMBL3 library, and we obtained 14 positive clones. The DNA inserts were analyzed by restriction enzyme digestion peptide 1 peptide 2a peptide 2b peptide 3 peptide 4 peptide 5 peptide 6 peptide 7 peptide 9 peptide 1 0 peptide 1 1 Nucleotide sequence of the ATFI gene. The 6.6-kb XbaI fragment shown in Fig. 2a was partially sequenced. The resulting sequence is shown in Fig. 3. A computer analysis revealed that the largest open reading frame of the sequenced fragment extended from nucleotide 234 to nucleotide 1808. This open reading frame encoded a 525-amino-acid protein with a molecular weight of 61,059. All 10 peptide sequences which were determined by sequencing the purified AATase ( Fig. 1) were present in this predicted protein (Fig. 3). In addition, the molecular weight of the protein was consistent with the value (60,000) estimated by SDS-polyacrylamide gel electrophoresis of the purified enzyme.
From the codon usage data for the ATF1 gene, a codon bias index of 0.07 was calculated by the method of Bennetzen and Hall (2). This low codon bias index value for ATF1 suggested that the level of expression of the ATF1 gene might be very low.
The Atfl protein contains 14 cysteine residues out of a total of 525 amino acids. This number of cysteine residues is greater than the numbers of cysteine residues in common acetyltransferase and acyltransferase. It is well known that AATase is highly labile, and it is possible that the high number of cysteine residues is partially responsible for the lability of this enzyme. Figure 4 shows the hydrophobicity profile of the Atfl protein. The AATase has been recognized as a membranebound enzyme, and solubilization with Triton X-100 was necessary to purify the Atfl protein. However, interestingly, the results of our hydrophobicity analysis of the Atfl protein indicated that this protein does not have a significantly hydrophobic domain (>30 amino acids); its mean index was calculated to be -0.38 by the method of Kyte and Doolittle (9).
Cloning of the ATF1 and Lg-ATFI genes from brewery lager yeast. Figure 2d shows the results of a Southern blot analysis of yeast genomic DNA digested with either ClaI orXbal when the 0.4-kb ClaI-EcoRI fragment shown in Fig. 2a was used as the probe. In S. cerevisiae Kyokai No. 7 only one band was detected under strict hybridization conditions (Fig. 2d, lanes 1 and 3). However, in brewery lager yeast strain KBY001, two bands were detected under the same hybridization conditions (Fig.  2d, lanes 2 and 4). One band was similar in size to the band obtained for S. cerevisiae, and it exhibited strong hybridization with the probe. The other band was a different size and exhibited weak hybridization with the probe. These results suggested that S. cerevisiae has a unique ATF] gene and that brewery lager yeast has one ATF1 gene and another, homologous gene.  We then cloned the AATase-encoding genes from the X-EMBL3 library constructed from KBY001 DNA by the 0.4-kb ClaI-EcoRI fragment from the S. cerevisiae Kyokai No. 7 ATF1 gene as the probe. We screened a total of 30,000 recombinants from the X-EMBL3 library, and we obtained 17 clones which exhibited strong hybridization with the probe and 11 clones which exhibited weak hybridization with the probe.
Figures 2b and c show restriction maps of the DNA inserts which were cloned in the strongly hybridized clones and the weakly hybridized clones, respectively. It is clear that the strongly hybridized DNA fragment encodes the ATF gene of brewery lager yeast, because its structure is quite similar to the structure of the ATF1 gene of S. cerevisiae.
The weakly hybridized DNA fragment has a different structure. However, because it has been suggested that brewery lager yeast is an allopolyploid and has two sets of genes, which are structurally different but have similar functions (4, 6, 7, 14), we speculated that this homologous gene might be a derivative of the ATF1 gene. This homologous gene appears to be specific to brewery lager yeast, and we designated this gene the Lg-ATFI gene.
Expression of the ATF and Lg-A TF genes in S. cerevisiae. To confirm that the three cloned fragments really encode AATase, the 6.6-kb XbaI fragment from S. cerevisiae Kyokai No. 7 and the 6.6-kb XbaI fragment and 5.7-kb BglII fragment from brewery lager yeast strain KBY001 were subcloned into yeast shuttle vector YEp13K; the resulting plasmids were designated YATK1 1, YATL1, and YATL2, respectively. These plasmids were used to transform S. cerevisiae TD4. Each transformant was grown in a static culture in YM15 medium at 30°C for 24 h, and then AATase activity was measured.
All of the transformants exhibited very high levels of AATase activity, but the levels of activity differed depending on the origin of the gene ( Table 1). The ATF1 gene from brewery lager yeast exhibited the highest level of activity (15 times greater than the control level), and the Lg-ATFI gene exhibited the lowest level of activity (6.5 times greater than the control level).
When these transformants were cultured with vigorous shaking, they exhibited very low levels of AATase activity (Table 1). These results indicated that all of the cloned fragments encoded an AATase gene.
Ellect of the ATF gene on acetate ester production in yeast cells. AATase is recognized as an enzyme that plays a primary role in acetate ester synthesis in many alcoholic beverages. To evaluate the effect of the cloned ATF and Lg-ATF1 genes on ester synthesis during fermentation, the transformants and the parental strain were cultured at 30°C for 24 h in YM15 medium, and then the volatile ester concentrations in the culture supernatants were determined.
Compared with the parental strain, the YATL1 transformants exhibited a 27-fold increase in isoamyl acetate production and a 9-fold increase in ethyl acetate production, and the YATL2 transformants exhibited a 17-fold increase in isoamyl acetate production and a 2-fold increase in ethyl acetate production ( Table 2).
The production of ethanol and other higher alcohols did not change. These results indicate that AATase activity is a limiting factor in the production of acetate esters in fermented media. DISCUSSION Molecular cloning and nucleotide sequence of the ATFE gene. The volatile ester concentration is one of the most important characteristics of alcoholic beverages. The ester concentration depends on oxygen (20), CO2 pressure (15), and other factors (23).
AATase is known to be responsible for acetate ester synthe-     (20,23). It has been suggested that AATase activity dues in other known acetyltransferases and acyltransferases.
iibited by unsaturated fatty acids (22) and that the AATase is known to be highly labile, and it is possible that the ntration of unsaturated fatty acids in the cell membrane high proportion of cysteine residues is responsible in part for s AATase activity (23). Recently, Malcorps et al. (10) the lability of this enzyme. sed that gene repression is the main cause of decreases in The sequence analysis also revealed a unexpected feature of ase activity in the presence of oxygen and unsaturated AATase. Although AATase has been shown previously to be a acids. It was necessary to clone the AATase-encoding membrane-bound enzyme (10,12,13,22,23), a hydrophobicity in order to understand the mechanism of ester formation. analysis of the Atfl protein revealed that this protein is not -ause there is no simple method to measure AATase hydrophobic. Although there are some short hydrophobic ty on plates, enzyme purification was a necessary step in segments which could possibly interact with membranes, the ig the AATase gene. The 6.6-kb XbaI fragment that was sequencing results suggested that the Atfl protein tends to be d from S. cerevisiae was shown to encode the ATF1 gene.
hydrophilic rather than hydrophobic and that potentially hynucleotide sequence of theATFl gene revealed that this drophobic transmembrane segments are absent (Fig. 4). The encodes a protein with a molecular weight of 61,059. This Atfl protein has no processed segment in the N-terminal is consistent with the predicted molecular weight of the region to act as a potential signal sequence for secretion or as ed AATase (60,000). All 10 peptide sequences which an intracellular targeting segment, as do proteins destined for determined by sequencing the purified AATase (Fig. 1) the mitochondria and endoplasmic reticulum. Malcorps and resent in this predicted protein (Fig. 3). When yeast cells Dufour proposed that the enzyme might be loosely bound to transformed with the ATFl gene carried on a multicopy the vacuole (11). Our results suggest that the Atfl protein is kid, the resulting transformants exhibited high levels of not an integral membrane protein but is a membrane-associase activity in static cultures. These results strongly sug-ated protein.
hat the Atfl protein is the same as AATase.
The results of the computer analysis showed that there was P results of the ATFI gene and Atfl protein sequence no extensive sequence similarity between the translated amino sis revealed some well-known features of AATase. First, acid sequence of the Atfl protein and any other known Adon usage data suggested that the Atfl protein has a low protein-encoding sequence in the GenBank, EMBL, NBRFbias index value (0.07), as defined by Bennetzen and PIR, and SWISS-PROT databases. However, by comparing the Atfl protein with previously described acetyltransferases and acyltransferases, we found that the Atfl protein has a short gene originated from S. cerevisiae K7.
and Heml proteins also contain numerous cysteine residues gene originated from brewery lager yeast.
Expression of the ATFJ and Lg-ATFJ genes in yeast cells. The results of a Southern analysis of yeast genome DNAs suggested that S. cerevisiae has a unique ATFE gene, while brewery lager yeast has one A TF1 gene and another, homologous gene (the Lg-ATFl gene).
Transformants which had the ATF1 gene or the Lg-ATFl gene exhibited high levels of AATase activity in static cultures. This finding suggested that the Lg-ATFl gene encodes AATase or its activator. We speculated that this homologous gene might be a derivative of the A TF1 gene, because one of the major characteristics of brewery lager yeast is that it is an allopolyploid which has at least two diverged genomes. Functionally similar but structurally different (or homologous) alleles have been reported for many other genes, including ERG10 (4), LEU2 (14), MET2 (6), URA3, CYC7, HIS4, and MAT (7). To confirm that the Lg-A TFE gene is a derivative of the ATF gene, sequence data will be necessary.
The results of a comparison of the ATF] gene and the Lg-A TFl gene suggested that the levels of AATase activity produced are different for the ATFI gene and the Lg-ATFl gene. The ATFI gene transformants exhibited greater AATase activities (Table 1) and produced greater concentrations of esters than the Lg-ATF gene transformants (Table 2). We confirmed by Southern analysis that the differences in the copy numbers of plasmids for three transformants were small (less than 10%) under the culture conditions used (data not shown). It is not clear at present whether the differences in AATase activity result from differences in the levels of gene expression or are due to differences in the specific activities of the gene products. Investigations in which Northern (RNA) analysis is used should clarify this issue.
ATF gene plays a key role in ethyl acetate and isoamyl acetate synthesis. The results of an analysis of ester production by the transformants clearly demonstrated that the ATF gene is useful for the control of ester production. The concentrations of both ethyl acetate and isoamyl acetate in the culture supernatants increased compared with the parent strain, and these increases depended on the levels of AATase activity of the transformants, while the production of ethanol and other higher alcohols did not change.
Ashida et al. isolated a mutant which produces high concentrations of isoamyl alcohol and isoamyl acetate during fermentation (1). These authors suggested that the level of isoamyl alcohol production is a major limiting factor in isoamyl acetate production during fermentation. Our data strongly suggest that AATase activity is also a major limiting factor in isoamyl acetate and ethyl acetate production.
However, it should be noted that the increases in production of ethyl acetate and isoamyl acetate are not identical. It was observed that for either the ATF gene or the Lg-A TFl gene the increase in isoamyl acetate concentration was greater than the increase in ethyl acetate concentration, compared with the control.
This difference in the ratio of increases may be due to AATase substrate specificity, since it has been suggested that AATase has a greater affinity for isoamyl alcohol than for ethanol (13,22).
Recently, one other type of acetyltransferase has been reported to be responsible for the production of ethyl acetate in yeast cells (10). Our data strongly suggest that the Atfl protein plays a key role in both isoamyl acetate synthesis and ethyl acetate synthesis. However, in our preliminary experiment, an atfl::URA3 strain still exhibited AATase activity which was 5 to 10 times lower than the activity of the control. It is possible that yeast cells contain many different types of AATase and that this makes it difficult for brewers to control ester production. In order to evaluate the precise role of the ATFE gene in isoamyl acetate and ethyl acetate synthesis during fermentation, further studies are necessary.