Reducing diacetyl production of wine by overexpressing BDH1 and BDH2 in Saccharomyces uvarum

Abstract

As a byproduct of yeast valine metabolism during fermentation, diacetyl can produce a buttery aroma in wine. However, high diacetyl concentrations generate an aromatic off-flavor and poor quality in wine. 2,3-Butanediol dehydrogenase encoded by BDH1 can catalyze the two reactions of acetoin from diacetyl and 2,3-butanediol from acetoin. BDH2 is a gene adjacent to BDH1, and these genes are regulated reciprocally. In this study, BDH1 and BDH2 were overexpressed in Saccharomyces uvarum to reduce the diacetyl production of wine either individually or in combination. Compared with those in the host strain WY1, the diacetyl concentrations in the recombinant strains WY1-1 with overexpressed BDH1, WY1-2 with overexpressed BDH2 alone, and WY1-12 with co-overexpressed BDH1 and BDH2 were decreased by 39.87, 33.42, and 46.71%, respectively. BDH2 was only responsible for converting diacetyl into acetoin, but not for the metabolic pathway of acetoin to 2,3-butanediol in S. uvarum. This study provided valuable insights into diacetyl reduction in wine.

Electronic supplementary material

The online version of this article (doi:10.1007/s10295-017-1976-2) contains supplementary material, which is available to authorized users.

Introduction

Wine bouquet is often influenced by a buttery or nutty flavor, which has been directly linked to the accumulation of vicinal diketone diacetyl [23]. As an important group in flavor profiles, the characteristic buttery taste of diacetyl has long been a major problem in wine. The sensory threshold for diacetyl was 1–4 mg/l in wine. In general, this threshold varies according to the categories of wine. Diacetyl with an appropriate concentration can produce a buttery or butterscotch flavor in wine and improve flavor quality. However, diacetyl elicits an unwanted buttery flavor when its content is more than 10 mg/l [5, 16, 19].

In wine, diacetyl as an important odoriferous compound is a byproduct synthesized during alcoholic and malolactic acid fermentation (MLF) [19, 24]. Diacetyl formation and concentration in wine are influenced by many factors, including pH, citrate concentration, temperature, sulfur dioxide concentration, and aeration degree during winemaking [2]. In many previous studies, fermentation conditions have been optimized to reduce diacetyl production in wine [1, 28, 38].

Diacetyl is the product resulting from the chemical oxidative decarboxylation of α-acetolactate. Extensive researches have been devoted to this topic, many researchers have extrapolated that the lactic acid bacteria (LAB) was associated with winemaking and was an important role on producing diacetyl [21, 23]. LAB metabolism is associated with a change in the sensory attributes of the wine and the most important change is the reduction in wine acidity due to the decarboxylation of l-malic acid to form l-lactic acid, which is the main basis for the malolactic fermentation and the diacetyl is the important byproduct during the process [2, 8]. Furthermore, diacetyl is the product resulting from the metabolism of citrate. Diacetyl is formed as an intermediate metabolite in the reductive decarboxylation of pyruvic acid to 2,3-butanediol. In LAB, theoretically, 1 mol of citrate produces 1 mol of acetic acid, 2 mol of carbon dioxide and 0.5 mol of (diacetyl + acetoin and 2,3-butanediol) [28, 31]. The metabolism of citric acid is usually sequential to malic acid during MLF, and is not initiated until more than half of the malic acid has been metabolised [3]. Oenococcus oeni is the preferred species for inducing MLF in wines, as it is the best adapted to survive in the harsh wine conditions of low pH and high alcohol content [20].

Diacetyl synthesis is a complicated process requiring yeast, only a low concentration at the completion of fermentation in wine (<1 mg/l) [14]. But based on current knowledge, the taste threshold of diacetyl varies according to the style and type of wine, such as Martineau et al. (1995) reported that the threshold for Chardonnay and Pinot Noir had been determined were 0.2 mg/l for wine and 0.9 mg/l, respectively [20]. It is worthy to note that metabolism of yeast will influence the diacetyl content in wine. Furthermore, wine yeast are capable of synthesising diacetyl as well as degrading it, The presence of yeast lees, especially when stirred, can further reduce the diacetyl content [27]. In this study, treatment of the wine in MLF by addition of sugar promoted the yeast to ferment again to reduce the diacetyl production in wine.

The application of genetic engineering strategies in yeast have been proposed to reduce diacetyl production. For example, ILV3 encoding dihydroxyacid reductase and ILV5 encoding reductoisomerase are overexpressed to minimize diacetyl production in brewing yeast [12, 26, 33]. Acetolactate synthase encoded by ILV2 and ILV6 catalyzes the formation of α-acetolactate, which is the precursor of diacetyl formation. Therefore, deleting ILV2 or ILV6 to reduce diacetyl content has also been proposed [7, 39]. Genes encoding α-acetolactate decarboxylase that catalysed the decarboxylation of α-acetolactate into acetoin have been cloned from different bacteria, such as Enterobacter aerogenes, Klebsiella terrigena, Acetobacter aceti, and Streptococcus lactis and expressed in yeast to construct recombinants [4, 11, 37]. However, previous approaches only focused on preventing the formation of diacetyl, and enhancing the degradation of diacetyl has not yet to be extensively investigated.

Diacetyl is formed through non-enzymatic decarboxylation from α-acetolactate outside a cell and is reabsorbed by the yeast cell. Diacetyl is then converted to acetoin and to 2,3-butanediol by diacetyl reductase or 2,3-butanediol dehydrogenase encoded by BDH1. This conversion may be catalyzed by other ketoreductases that are yet to be fully characterized [21, 22]. In 2009, Ehsani et al. deleted BDH1 in a wine yeast strain model and examined the effect of its disruption during fermentation in an MS medium. They observed that the diacetyl concentration produced by the recombinant strain increases by twofold compared with that produced by a native strain. The amount of produced acetoin increases and the amount of 2,3-butanediol decreases [9]. Acetoin and 2,3-butanediol yield a much higher taste threshold than diacetyl does [3]. BDH2 is a gene adjacent to BDH1, whose encoded protein is 51% identical to Bdh1p, and both genes are regulated reciprocally [10]. In the current study, recombinant Saccharomyces uvarum with BDH1 and/or BDH2 overexpression were constructed to reduce diacetyl production in wine. The recombinant strains were characterized in terms of mRNA levels, diacetyl, acetoin, and 2,3-butanediol content, and other fermentation performance parameters. Our study indicated that BDH1 and BDH2 can effectively degenerate diacetyl in wine.

Materials and methods

Strains and vectors

Saccharomyces uvarum WY1, which was purchased from China Center of Industrial Culture Collection and encoded as CICC1465, was used as the host strain. Escherichia coli DH5α (Φ80 lacZΔM15 ΔlacU169 recA1 endA1 hsdR17 supE44 thi-1 gyrA relA1) that was used for the amplification of plasmids was obtained from the Yeast Collection Center of Tianjin Key Laboratory of Industrial Microbiology, Tianjin University of Science and Technology, China.

The plasmids Yep352 and pUG6 used in this work were purchased from Invitrogen (Carlsbad, CA, USA). Plasmid Puc19-PGK was obtained from the Yeast Collection Center of the Tianjin Key Laboratory of Industrial Microbiology, Tianjin University of Science and Technology, China.

Cultivation conditions

Escherichia coli DH5α strain was grown at 37 °C in Luria–Bertani broth (1% NaCl, 1% tryptone, and 0.5% yeast extract) supplemented with ampicillin (100 mg/l). S. uvarum WY1 was grown at 28 °C in yeast extract peptone dextrose (YEPD) medium (1% yeast extract, 2% peptone, and 2% glucose), which was supplemented with 200 mg/l G418 for selecting Geneticin (G418)-resistant yeast strains after transformation. Then, when the YEPG medium (1% yeast extract, 2% peptone, and 2% galactose) was used for Cre expression in yeast transformants, 500 mg/l Zeocin (Promega, Madison, United States) was added to the YEPD plates for yeast culture to select Zeocin-resistant yeast strains.

Fermentation conditions

A 250 ml flask autoclaved at 121 °C for 15 min before use was filled with 190 ml of unfiltered and formulated Muscat grape juice (20.45 Brix, pH 3.31, SO₂ 80 mg/l). Muscat grape juice was obtained from Hangu, Tianjin, China.

The yeast strains were maintained in YEPD medium at 28 °C, 180 rpm. After incubating for 12 h, the yeast cells were subcultured into 50 ml YEPD medium under similar condition. Bacterial sludge were harvested by centrifugation (5000 rpm, 4 °C, and 5 min) and washed once with 50 ml sterile water. Cell pellets were resuspended in 5 ml sterile water, OD600 measured (0.1 final OD 600), and inoculated to grape juice. The mixture was fermented at 25 °C and weighed every 12 h to monitor fermentation progress until the daily weight loss was less than 0.3 g.

Fermentation broth was removed the scum under aseptic conditions and then transferred to similar bottles. Solid active lactic acid bacteria (0.01 g/l) were placed into the bottles and fermented at 18 °C to activate malolactic acid fermentation. 28 g/l of sugar was added into the wine during the malolactic acid fermentation (the fourth day), which can promote yeast ferment again. In the secondary fermentation, the malic acid content was determined every 3 days until the content tend to a stable value, which suggested that the MLF has finished. All fermentations were performed in triplicate.

Plasmid construction

Polymerase chain reaction (PCR) primers used in this study are listed in Table 1. All oligonucleotide primers were synthesized and purified by DingGuo ChangSheng Biotech (Beijing, China). Plasmid DNA was prepared from E. coli by a plasmid DNA extraction kit (Solarbio, Beijing, China). Genomic DNA was prepared from S. uvarum WY1 by a yeast genomic DNA extraction kit (Solarbio, Beijing, China).

Strains and plasmids used in the current study

Strains or plasmids	Relevant characteristic	Reference or source
Strain
E. coli DH5α	supE44ΔlacU169 (ϕ80lacZΔM15) hsdR17 recAl endAl gyrA96 thi-1 relA	This lab
WY1	Wild-type industrial Saccharomyces uvarum	This lab
WY1-1	BDH1::PGK1p-BDH1-PGK1t-loxP-KanMX-loxP	This study
WY1-2	BDH1::PGK1p-BDH1-PGK1t-loxP-KanMX-loxP	This study
WY1-12		This study
WY1Δ1	BDH1::loxP-KanMX-loxP	This study
WY1Δ2	BDH2::loxP-KanMX-loxP	This study
Plasmid
pUC19-PGK	Apr, cloning vector, containing PGK1p-PGK1t expression cassette	This lab
pUG6	Kanr, containing loxP-KanMX-loxP cassette	This lab
pSH-Zeocin	Zeor, Cre expression vector	This lab
Yep352	Apr, cloning vector	This lab
Yep-KPB1	Apr, Kanr, B1A-PGK1p-BDH1-PGK1t-loxP-KanMX-loxP-B1B	This study
Yep-KPB2	Apr, Kanr, B2A-PGK1p-BDH2-PGK1t-loxP-KanMX-loxP-B2B	This study

Yep352 was used as backbone to construct recombinant plasmids Yep-KPB1 and Yep-KPB2. The PCR-generated PGK1 fragment (1771 bp) amplified from Puc19-PGK plasmid as template with P-F and P-R primers was inserted into the BamHI/SalI of Yep352 plasmid to generate the Yep-P plasmid. A XhoI fragment of BDH1 (1149 bp), which was amplified with primers B1-F and B1-R from WY1 genomic DNA, was ligated to the vector Yep-P by in-fusion [40], Finally, KanMX fragment (1613 bp) digested with BamHI and KpnI as the dominant selection marker during yeast conversion was amplified through PCR using pUG6 as template with Kan-F and Kan-R primers. The fragment was cloned with Yep-PB1 to produce the final plasmid Yep-KPB1, which is shown in Fig. 1a. Plasmid Yep-KPB2, an episomal plasmid with BDH2 under the control of PGK1 promoter, was constructed using the same method.

Fig. 1

a Construction of plasmid Yep-KPB1. b The homologous recombination of B1ZF-loxP-KanMX-loxP-PGK1p-BDH1-PGK1t-B1ZR with the S. uvarum WY1

Yeast transformants and construction of recombinant yeast strain

Selecting single colonies from YEPD plate were maintained in 5 ml YEPD medium at 28 °C, 180 rpm. After incubating for 12 h. Then pipette 0.5 ml from the bacterial solution into 5 ml YEPD medium (180 rpm, 28 °C, and 4 h). S. uvarum transformation was performed using lithium acetate/PEG method [29]. Transformants were screened on YEPD plate containing 200 mg/l G418 and verified through PCR with accurate site integration. Various pairs of primers were designed and listed in Table 2.

Primers used in the present study

Primers	Sequence (5′ → 3′)	Restriction site
For plasmid construction
B1-F	CCGCTCGAGATGAGAGCTTTGGCATATT	XhoI
B1-R	CCGCTCGAGTTACTTCATTTCACCGTG	XhoI
B2-F	CCGCTCGAGATGAGAGCCTTAGCGTAT	XhoI
B2-R	CCGCTCGAGTCATGTGTGACGCAGTT	XhoI
Kan-F	CGGGGTACCCAGCTGAAGCTTCGTACGCT	KpnI
Kan-R	CGCGGATCCGCATAGGCCACTAGTGGATC	BamHI
P-F	CGCGGATCCTCTAACTGATCTATCCAAAACTG	BamHI
P-R	ACGCGTCGACTAACGAACGCAGAATTTTCGAG	SalI
B1Z-F	ACGTTCTCGTGTTAATCCCGCGGTCTTCTTGTTTTACTAACTTTTCTTTCTCTCATAGCATTCTCTTGACAGCTGAAGCTTCGTACGCT	–
B2Z-F	ATTTTCTTTTTGTTCGTAACTATCTGTGTATGTAGTAGTGTAATCTACTTTTAATTTACTATGCACAGCTGAAGCTTCGTACGCT	–
B2Z-R	TATCATATCAAGAGAAACAGGCTAGGACCCCGTAAGGAGGAAAGAATAGGCAAGGATAGGAAAACATAACGAACGCAGAATTTTCGAG	–
For PCR verification
B1U-F	TAAACGAGAAGGAGTCTACAATCAA	–
B1U-R	TAGCATACACTTACGACCAGCGA	–
B1D-F	CGCTCC TCTTTTAATG CCTTTA	–
B1D-R	TTGTCTTCCAGAGCCTTTTTATTTA	–
B2U-F	GCCGAGCGGGTCGATCAAGAACTAA	–
B2U-R	CGTCAAGACTGTCAAGGAGGGTATT	–
B2D-F	CGCTCC TCTTTTAATG CCTTTA	–
B2D-R	TGGAGTGTGGCCAGCAATGCAGGTC	–
Zeocin-U	CCCACACACCATAGCTTCA	–
Zeocin-D	AGCTTGCAAATTAAAGCCTT	–
For real-time qPCR
B1RT-F	TTCAATCCCTCCAAGCACG	–
B1RT-R	ATGGCGATGTCTCCGTTGT	–
B2RT-F	GAGCCTTAGCGTATTTCGG	–
B2RT-R	CAACTACCTTGTCTCCCACTT	–

The DNA fragment of B1ZF-loxP-KanMX-loxP-PGK1p-BDH1-PGK1t-B1ZR was amplified by PCR using Yep-KPB1 as template with the primers B1Z-F and B1Z-R. Then, the fragment was transformed into the WY1 host strain. The fragment was integrated into the chromosome of WY1 at BDH1 locus through homologous recombination, constructing BDH1 overexpression strain WY1-1, as shown in Fig. 1b. WY1-2 strain with BDH2 overexpression at BDH2 locus was constructed using the same method. The DNA fragments of B1ZF-KanMX-B1ZR and B2ZF-KanMX-B2ZR amplified from Yep-KPB1 and Yep-KPB2 plasmids were transformed into the yeast WY1, respectively. The transformation produced the recombinant strains WY1Δ1 and WY1Δ2. BDH1 and BDH2 overexpression in one strain, simultaneously named WY1-12, was constructed on the basis of WY1-2, that is, the strain WY1-2 was transformed into plasmid Yep-KPB1 after excising KanMX.

Real-time PCR

Yeast strains were cultured at 28 °C in a YEPD medium for 18 h, the mRNA was extracted and transcribed in reverse to produce cDNA using a cDNA synthesis kit (TIANScript RT Kit, TIANGEN, China). The abundance of mRNAs coding for BDH genes was measured by amplifying the genes using corresponding cDNAs as PCR templates. The mRNA level on the gene expression was quantified by real-time quantitative PCR (RT-qPCR), using an Ultra SYBR Two-Step RT-qPCR kit with ROX (reference dye for real-time PCR; CWBIO, China). Actin gene ACT1 and UBC6 were used as reference gene, respectively; expression level of the target genes was normalized with respect to the expression level of ACT1 and UBC6 [6, 32]. The primers used are listed in Table 2 and experiments were conducted thrice.

Analytical methods of diacetyl content and volatile flavor compounds

The concentration of diacetyl was determined after MLF. Diacetyl content was measured using a diacetyl distiller and was a modification of a method for beer analysis. Different concentrations of diacetyl solution (100 ml) were prepared at 4 °C. A distillate (25 ml) was collected in a 50 ml graduated cylinder from the diacetyl distiller, and pure water was prepared as blank control. The solution was injected into 4.5 ml pure water and to a 0.5 ml of 1% (w/v) o-phenylendiamine solution that was confected with 4 mol l⁻¹ HCl solution. Pipette 10 ml from above 30 ml mixture into a colorimetric tube and added into 2 ml 4 mol l⁻¹ HCl solution, subsequently. Then, the tubes were placed in the dark for 20–30 min after shaking. Finally, diacetyl content was detected with a 1 cm cuvette at an optical density of 335 nm (OD335). A standard curve based on the absorbance of different diacetyl concentrations at 335 nm was established [35].

In this work, 100 ml wine broth post-fermentation was added into diacetyl distiller. A total of 25 ml fermentation broth distillate was collected using 50 ml graduated cylinder. The absorbance of diacetyl at 335 nm was detected according to the above method. Then, diacetyl content was calculated in accordance with the standard curve and the analysis was performed in triplicate.

Other volatile flavor compounds were detected as previously described methods [30]. The analyses were performed in triplicate.

Result

Diacetyl content assay

The diacetyl production levels of the host strain (WY1), WY1Δ1, and WY1Δ2 were measured after fermentation, which was performed in triplicate for each strain. The results showed that diacetyl production by WY1Δ1 and WY1Δ2 were 10.41 and 9.50 mg/l, thus, increased by 1.37- and 1.25-fold, respectively, compared to that of WY1. These results indicated that BDH1 and BDH2 affected the formation of diacetyl in S. uvarum. Therefore, we investigated the diacetyl production of the overexpressed strains. Our findings showed that the diacetyl content of wine brewed with WY1, WY1-1, WY1-2, and WY1-12 was 7.60, 4.57, 5.06, and 4.05 mg/l, respectively. As shown in Table 3, the diacetyl production of engineered strains was always lower than that of the host strain WY1. The concentration of diacetyl was decreased by 39.81, 33.42, and 46.71%, compared with the control strain. This finding indicated that BDH1 and/or BDH2 overexpression decreased diacetyl production in wine effectively.

Determination of diacetyl and other volatile flavor compounds in the wine fermented with host strain and engineered strains^a

	WY1	WY1-1	WY1-2	WY1-12	WY1Δ1	WY1Δ2
Diacetyl (mg/l)	7.60 ± 0.04	4.57 ± 0.02	5.06 ± 0.02	4.05 ± 0.02	10.41 ± 0.05	9.50 ± 0.04
Acetoin (mg/l)	25.150 ± 0.20	22.250 ± 0.31	31.380 ± 0.21	20.331 ± 0.25	–	–
2,3-Butanediol (g/l)	0.950 ± 0.010	0.998 ± 0.015	0.960 ± 0.016	1.120 ± 0.010	–	–
Propyl alcohol (mg/l)	38.9025 ± 0.45	38.3915 ± 0.54	34.5136 ± 0.47	41.36 ± 0.63	–	–
Isobutyl alcohol (mg/l)	35.0677 ± 0.65	34.8714 ± 0.55	30.5938 ± 0.49	36.6924 ± 0.58	–	–
Isoamyl alcohol (mg/l)	131.4775 ± 2.50	146.5142 ± 3.00	129.8257 ± 2.80	133.5871 ± 3.50	–	–

^a Data are the average of three independent experiments ± the standard deviation (P < 0.05)

Effects of BDH1 and/or BDH2 overexpression on the production of acetoin, 2,3-butanediol volatile, and other flavor compounds

To investigate the effect of the genes BDH1 and/or BDH2 overexpression on diacetyl metabolism, the contents of acetoin and 2,3-butanediol fermented by the recombinant strains and the host strain was measured. As shown in Table 3, acetoin production by the engineered strains WY1-1 and WY1-12 were reduced by 11.53 and 19.16%, respectively, compared with that of WY1. Then, 2,3-butanediol content was increased significantly. However, acetoin content of WY1-2 increased from 25.150 ± 0.20 to 31.380 ± 0.21 mg/l, compared with the parental strain WY1. The production of 2,3-butanediol of WY1-2 was almost similar to that of WY1. This result suggested that overexpression of BDH1 can affect acetoin and 2,3-butanediol content significantly. However, overexpression of BDH2 has little influence on 2,3-butanediol formation.

The flavor component profiles of wild type and recombinant strains were measured via GC analysis after fermentation. As shown in Table 3, the contents of main higher alcohols had no significant difference and were within the range of suggested wine quality. Propyl alcohol, isobutyl alcohol, and isoamyl alcohol were formed as a result of the metabolism of pyruvic acid and respective amino acids aspartic acid, valine, and leucine, each metabolic pathway are controlled by a series of corresponding genes. For example, pyruvic acid and valine were catalyzed in the presence of different catalysts to produce α-ketoisovarelate. Then, isobutyl alcohol is formed in the presence of PDC1, PDC5, and PDC6 genes. The results of isoamyl alcohol formed α-ketoisocaproate under the catalysis of branched-chain-2-oxoacid decarboxylase (encoded by gene THI3) and α-ketoisocaproate was generated from α-ketoisovarelate and leucine in the presence of different catalysts [18]. These results reflect that BDH1 and/or BDH2 overexpression did not observably affect the aromatic compound content of wine.

BDH1 and BDH2 expression levels

The mRNA levels of the expressed genes were quantified through RT-qPCR to confirm BDH1 and BDH2 overexpression. The results showed that the mRNA levels of BDH1 and BDH2 in the recombinant strains were higher than the parental strain WY1. The expression levels of the target genes were evidently increased. As shown in Fig. 2a, BDH1 overexpression results showed that the cDNA of the recombinant strainWY1-1 and WY1-12 increased to 2.88- and 2.55-fold than that of the host strain WY1, respectively. BDH2 overexpression results (Fig. 2b) showed that the cDNA of recombinant strain WY1-2 and WY1-12 increased to 3.58- and 2.32-fold than that of the native strain WY1, respectively. To get accurate results, the genes expression levels were quantified with gene UBC6 as reference. Compared with those of WY1, the mRNA expression levels of BDH1 in WY1-1 and WY1-12 increased to 3.12- and 2.74-fold, respectively. Meanwhile, BDH2 overexpression results showed that the cDNA of WY1-2 and WY1-12 increased to 3.85- and 2.55-fold, respectively. The result was essentially consistent with the conclusion that kept the gene ACT1 as reference and also suggested that the mRNA expression levels of BDH1 and BDH2 were correspondingly enhanced by the promoter PGK1 control.

Fig. 2

a, b Determination of BDH1 and BDH2 gene expression levels in the recombinant strainWY1,WY1-1,WY1-2 and WY1-12. Data are the average of three independent experiments. Error bars represent ±SD

Fermentation characteristics of engineered strains

Stable performance is crucially important in industrial fermentation. The physiological characteristics (biomass, ethanol, and residual sugar) of the engineered strains WY1-1, WY1-2, and WY1-12 were investigated to assess their fermentation ability compared with that of the native strain WY1. The wild type and recombinant strains had similar fermentation characteristics. The effects of these mutants and the host strain on the final biomass yield were not significant (Fig. 3); they exhibited the same growth rate, which reached the stationary phases of fermentation after 10 h. Furthermore, the alcohol production of the recombinant strains was similar to those of the host strain WY1. The residual sugar content of WY1-12 was slightly higher than that of WY1, whereas those of WY1-1 were slightly lower compared with WY1, as shown in Fig. 4. In general, the content were in the range of index in wine, no significant difference in the recombinant strains from the host strain WY1 was confirmed by repeated measures ANOVA (P > 0.05), indicating that the basic indices of the recombinant yeast strains were consistent with those of the host strain WY1.

Fig. 3

Growth curve of mutants and starting strain. Data are the average of three independent experiments. Error bars represent ±SD

 

Fig. 4

Residual sugar and ethanol production of the engineered strains WY1-1, WY1-2 and WY1-12 and the parental strain WY1. Data are the average of three independent experiments. Error bars represent ±SD

Discussion

Release of flavor substances during consumption is a key quality parameter of beverages [15]. In the brewing industry, diacetyl, as one of an important flavor compounds, must be controlled within a specified range in wine, due to its unpleasant butter [34]. Therefore, studies have mainly aimed to control the production of diacetyl during wine fermentation [8, 25, 36]. In accordance with the method used for the introduction of a plasmid carrying a multi-copy of regulatory or structural genes to test gene regulation models at molecular levels, a similar process was performed in the present work to investigate the differences between the effects of BDH1 and BDH2 overexpression on diacetyl metabolism. This study demonstrated that BDH1 and BDH2 genes are critical factors on diacetyl degradation in wine.

In the present study, WY1-1, WY1-2, and WY1-12 were constructed, and BDH1 or/and BDH2 were overexpressed under the control of PGK1 promoter. After fermentation occurred in grape juice, the strains elicited positive effects on the reduction of diacetyl production. These results demonstrated that the degradation of diacetyl was significantly correlated with the two genes. Compared with that in the parental strain, the overexpression of both BDH1 and BDH2 resulted in different effects, including changes in residual sugar and alcohol production (Fig. 4). Previous studies showed that protein overproduction does not always increase the corresponding product yields [13, 17]. On the contrary, this overproduction may be detrimental to cell growth and influence residual sugar and alcohol production in wine. Regardless of diacetyl production, residual sugar, or alcohol production, BDH1 overexpression is more effective than BDH2 overexpression. The diacetyl production of BDH1 disruption is more than that of BDH2 disruption. This observation is possibly attributed to the more dominant role of BDH1 than that of BDH2. Consequently, the former exhibits a more significant effect on the degradation of diacetyl than the latter does. Ehsani et al. reported that BDH1 participates in the reduction of diacetyl, and the expression level of BDH1 is a limiting factor of the 2,3-butanediol pathway, and this finding is consistent with our observation [9].

Our findings on BDH2 were not completely consistent with the conclusion described by González et al., who expounded that BDH2 in S. cerevisiae cannot affect the metabolism of diacetyl to 2,3-butanediol in wine [10]. In the present study, the effects of WY1-2 on 2,3-butanediol were not significant compared with those of the parental strain, but this strain affected the production of diacetyl and acetioin. As a result, the diacetyl production decreased and the acetoin content increased. Compared with those of WY1, the mRNA expression levels of BDH2 were correspondingly enhanced in WY1-2. Thus, the overexpression of BDH2 alone could trigger the conversion of more diacetyl into acetoin. These results suggested that the amount of BDH2 is another rate-limiting factor in the conversion of diacetyl into acetoin. BDH2 may be the gene that encodes diacetyl reductase or isoenzyme of diacetyl reductase in S. uvarum because BDH2 plays the same function as diacetyl reductase. Further studies should be performed to elucidate this phenomenon. The diacetyl production by WY1-12 was, respectively, decreased by 17.62 and 25.37% compared with those obtained by WY1-1 and WY1-2. This finding indicated that diacetyl production was significantly reduced. Acetoin content decreased and 2,3-butanediol content significantly increased in WY1-12 probably because the general expression levels of BDH1 and BDH2 were higher than those of the gene overexpression level alone. These results were also verified by the identification of the mRNA levels of BDH1 and BDH2 (Fig. 2).

In conclusion, brewing with recombinant industrial yeast strains overexpressing BDH1 and/or BDH2 can significantly decrease diacetyl production in wine. Our investigations showed that the overexpression of BDH1 alone was more effective than the overexpression of BDH2 alone. Based on WY1-1, overexpressing BDH2 can elicit an enhanced effect on the reduction of diacetyl production in wine. The decrease in diacetyl production achieved using the engineered yeast strains indicated that the new strains are useful for the development of novel industrial strains for winemaking. Our work also confirmed that BDH2 participates in the metabolic pathway of diacetyl to acetoin but not in the metabolic pathway of acetoin to 2,3-butanediol. Our work also provided a basis for further research on the reduction of diacetyl production in wine.

Acknowledgements

The current study was financially supported by the National High Technology Research and Development Program of China (863 Program) (2012AA022108), the National Natural Science Foundation of China (31271916), and the Key Technologies R & D Program of Tianjin (Grant no. 15ZCZDNC00110).

Compliance with ethical standards

Ethical statement

This manuscript is in compliance with Ethical Standards. This manuscript does not contain any studies with human participants or animals performed by any of the authors.

Conflict of interest

The authors declare that they have no competing interests.