HOM2 Deletion by CRISPR-Cas9 in Saccharomyces cerevisiae for Decreasing Higher Alcohols in Whiskey

Abstract

In typical whiskey, the content of higher alcohols is about 1500–2000 mg/L, leading to a high intoxicating degree (ID). To produce low-ID whiskey, Saccharomyces cerevisiae XF0-h, XF0-H and XF0-LH were successfully constructed by CRISPR-Cas9 gene editing technology to knockout HOM2 (encoding aspartate β-semialdehyde dehydrogenase) in the original strain XF0 and the LEU1 knockout strain XF0-L. The contents of higher alcohols in whiskey fermented by XF0-h, XF0-H, and XF0-LH were 704 ± 8 mg/L, 685 ± 6 mg/L, and 685 ± 19 mg/L, respectively, showing reductions of 23.93%, 25.98%, and 15.81% compared to XF0, XF0, and XF0-L. The fermentation conditions of XF0-LH were optimized through single-factor experiments and the Box–Behnken design. The optimal conditions were a wort concentration of 9.8 °P, hydrolyzed broken rice syrup addition of 78 g/L, and an inoculum size of 2.7 × 10⁶ cells/mL. The low-ID whiskey was brewed with a higher alcohol content of 556 mg/L by 50 L fermenter at the optimal conditions.

1. Introduction

Alcoholic fermentation is a critical process in whiskey production, during which yeast generates both ethanol and higher alcohols [1,2]. The higher alcohols in whiskey mainly include n-propanol, isobutanol, isoamylol, etc., which have an important influence on the flavor and taste of whiskey [3]. A moderate amount of higher alcohols can make the flavor of whiskey fuller and richer. The intoxicating degree (ID) of alcohol is enhanced by excessive higher alcohols, especially when isoamylol is excessive [4,5], because isoamylol causes ethanol and acetaldehyde to remain in the body for a longer time with intoxication symptoms such as headache, nausea and hangover [6,7,8]. The higher alcohols in whiskey mainly originate from the sugar metabolism pathway and amino acid metabolism of yeast [6]. It was reported that higher alcohols can be regulated by key genes of the higher alcohol metabolic pathway in Saccharomyces cerevisiae [9,10,11,12]. Isopropylmalate isomerase encoded by the LEU1 gene is a key enzyme in the pathway of the conversion of pyruvic acid to α-ketobutyric acid. Expression of the LEU1 gene affects the conversion between α-keto isovaleric and α-keto isohexanoic acid, which causes changes in the production of higher alcohols [13,14,15,16].

Aspartate β-semialdehyde dehydrogenase, which is encoded by the HOM2 gene, is a key enzyme in the second step of the pathway that synthesizes threonine and O-acetylhomoserine from aspartate [17]. Disrupted expression of the HOM2 gene affects the amount of threonine and O-acetyl-homoserine synthesized by yeast, which in turn causes changes in the production of higher alcohols. It was reported that aspartate β-semialdehyde dehydrogenase directly influences the Ehrlich pathway [18]. The HOM2 knockout strains show a significant reduction in the production of isoamylol and isobutanol [19]. Knockout of one allele of HOM2 and both of its alleles resulted in a significant reduction in isoamylol content in Chinese rice wine [20]. Fewer studies have been conducted on the effect of the HOM2 gene on the content of higher alcohols in spirits and whiskey.

Conventional knockout methods typically require replacing the target gene with a selectable marker gene. It also necessitates the removal of this marker gene to ensure the yeast is suitable for industrial production [21,22,23]. These methods generally allow for the knockout of only one gene at a time, and when multiple genes need to be modified, the recycling of marker genes is limited. Furthermore, conventional techniques often suffer from low editing efficiency and lengthy transformation cycles. The CRISPR-Cas system, which is a self-defense mechanism used by bacteria and archaea to prevent the invasion of foreign viruses, is now widely used in gene editing research [24,25]. The CRISPR-Cas9 system is an effective tool for simultaneously editing multiple genes in S. cerevisiae. In the CRISPR-Cas9 system, the target sequence is recognized by guide RNA (gRNA). The Cas9 protein then specifically cleaves the target site causing DNA double-strand breaks, which stimulates the cellular repair mechanism and realizes the genetic modification without screening markers [26,27,28].

The wort used in the fermentation of whiskey has a complex nutrient profile. Yeast obtains the nutrients for growth, reproduction and the production of by-products such as higher alcohols. The ratio of carbon to assimilable nitrogen (C/N) in wort is an important factor affecting the content of higher alcohols, which differed in wort fermented with different wort concentrations [29]. During alcohol fermentation, yeast primarily produces ethanol and higher alcohols. The yield of higher alcohols varies significantly with different inoculum sizes [30].

S. cerevisiae XF0-h and XF0-H were constructed by knocking out HOM2. S. cerevisiae XF0-LH was constructed by knocking out HOM2 and LEU1. At the same time, we optimized the fermentation process to reduce higher alcohols in whiskey to decrease the intoxicating degree and to improve the taste of whiskey. The brewing technology will be obtained to produce high-quality whiskey.

2. Materials and Methods

2.1. Materials and Reagents

2.1.1. Strains and Plasmids

The strains and plasmids are listed in Table S1 (Supporting Information). The Escherichia coli DH5α with p414-Cas9-BleoR or p426-gRNA-HOM2-kanMX was cultured in Luria–Bertani (LB) broth with 100 μg/mL ampicillin (Aladdin, Shanghai, China) at 37 °C. The S. cerevisiae strains were cultured at 30 °C in the YPD medium, and 200 µg/mL of Zeocin™ (Thermo Fisher, Shanghai, China) and 100 µg/mL of G418 (Aladdin, Shanghai, China) were used to screen positive transformants with Zeocin™ and KanMX resistance, respectively [31].

2.1.2. Design and Synthesis of Primers

The primers are listed in Table S2 (Supporting Information). The HOM2 gene sequence (Genomic Sequence: NC_001136) of S. cerevisiae (S288c) was retrieved from the National Center for Biotechnology Information (NCBI). Primers were designed by SnapGene and synthesized by Shanghai Sangon Bioengineering Co., Ltd. (Shanghai, China).

2.1.3. Wort Preparation

The crushed malt was added with distilled water at the ratio of 1:4 (kg/L) and stirred well for saccharification. Saccharification procedure: The mixture was heated at 53–55 °C for 60 min, then at 63–65 °C for an additional 60 min, and finally at 72 °C for 20 min. Following this, the wort was filtered through 4 layers of sterile gauze.

2.1.4. Preparation of Hydrolyzed Broken Rice Syrup [32,33]

The broken rice was crushed and collected through an 80-mesh sieve. It was mixed with water at 1:5 (g/mL) and enzymolysised with 30 U/g of thermostable α-amylase at 95 °C for 30 min to achieve liquefaction. The liquefied solution was enzymolysised with 300 U/g of glucoamylase at 55 °C for 48 h, followed by heating and concentration to produce hydrolyzed broken rice syrup (dextrose equivalent > 95%).

2.2. Construction of CRISPR-Cas9 System

The CRISPR-Cas9 system comprises the Cas9 protein and guide RNA (gRNA). In constructing the Cas9 expression plasmid, the p414-Cas9 plasmid and the pPICZ (alpha) plasmid as templates were amplificated by PCR (Cas9 gene: 3 min at 98 °C, and then 25 cycles of 98 °C for 10 s, 55 °C for 15 s, 72 °C for 9 min, followed by a final step at 72 °C for 10 min. BleoR gene: 3 min at 98 °C, and then 28 cycles of 98 °C for 10 s, 55 °C for 15 s, 72 °C for 2 min, followed by a final step at 72 °C for 4 min) to obtain fragments containing the Cas9 gene and the BleoR gene, respectively. The primers were designed with 20 bp homologous sequences at ends of the BleoR gene fragment and the Cas9 gene fragment. The two purified fragments were then ligated in vitro by the ClonExpress^® II One Step Cloning Kit (Vazyme, Nanjing, China). The ligation product was introduced into competent E. coli, which was plated on LB. Positive colonies were selected to extract the Cas9-expressing plasmid p414-Cas9-BleoR by the plasmid extraction kit (Magen, Shanghai, China) [14].

The construction of the gRNA expression plasmid required the replacement of the target site with the HOM2 target sequence. Using the HOM2 gene sequence, the protospacer adjacent motif (5′-NGG-3′ or 5′-NAG-3′) was identified by Snap Gene, and the 20 nt HOM2 target sequence was selected at the upstream of the protospacer adjacent motif. The 60 bp fragment containing the 20 nt HOM2 target sequence was obtained through overlap extension PCR (3 min at 98 °C, and then 35 cycles of 98 °C for 10 s, 55 °C for 15 s, 72 °C for 1 min, followed by a final step at 72 °C for 2 min). The gRNA plasmid p426-gRNA-kanMX and the primer HOM2r-F/HOM2r-R was amplified by PCR (3 min at 98 °C, and then 25 cycles of 98 °C for 10 s, 55 °C for 15 s, 72 °C for 7 min, followed by a final step at 72 °C for 9 min) to obtain the vector fragment (6395 bp) [14]. The primers were designed with 20 bp homologous sequences at ends of the vector fragment (6395 bp) and the 60 bp fragment. The two fragments were then ligated in vitro by the homologous recombination kit. The ligation product was introduced into competent E. coli DH5α, resulting in the gRNA expression plasmid p426-gRNA-HOM2-kanMX, capable of recognizing HOM2 targets.

Design of HOM2 donor DNA (Figure 1): The genome of S. cerevisiae (S288c) was used as a template, and two homology arms upstream and downstream of the HOM2 target sequence were amplified by PCR (3 min at 98 °C, and then 35 cycles of 98 °C for 10 s, 55 °C for 15 s, 72 °C for 5 s, followed by a final step at 72 °C for 1 min) with primers HT-F1/HT-R1 and HT-F2/HT-R2, respectively (20 bp of homology was set at the 5′ end of primers HT-R1 and HT-F2). The two segments were then ligated by overlap extension PCR (3 min at 98 °C, and then 28 cycles of 98 °C for 10 s, 55 °C for 15 s, 72 °C for 2 min, followed by a final step at 72 °C for 4 min) to obtain donor DNA.

2.3. Yeast Transformation

The purified p414-Cas9-BleoR plasmids were transformed into XF0, XF0-h and XF0-L by electrotransformation (Electroporation apparatus, 1.5 kV, 5 ms). The transformed yeasts were cultured in YPD medium (200 µg/mL Zeocin™) at 30 °C for 48 h. Then, the positive transformants were selected by colony PCR. The yeast with the p414-Cas9-BleoR plasmid was used to prepare the competent yeast, mixed with the purified p426-gRNA-HOM2-kanMX plasmid (100 ng/µL) and donor DNA (100 ng/µL) at a 1:10 (v/v) ratio. The mixture was subjected to electrotransformation, and the transformed yeasts were cultured in YPD medium (200 μg/mL G418) at 30 °C for 48 h. Then, the positive transformants were selected by colony PCR.

2.4. Screening and Verification of Gene Knockout Strains

The positive yeast transformants were selected by colony PCR (3 min at 98 °C, and then 30 cycles of 98 °C for 10 s, 55 °C for 15 s, 72 °C for 2 min, followed by a final step at 72 °C for 5 min) with the primer pair HOM2-A/HOM2-D. The HOM2 knockout strains were successfully screened and the expression levels of the HOM2 gene were assessed by Real-Time Quantitative PCR (30 s at 95 °C, and then 40 cycles of 95 °C for 5 s, 60 °C for 30 s) using a kit [31]. Total RNA of strains was extracted by a total RNA extraction kit (Sangon, Shanghai, China).

2.5. Discard Plasmids

The gene knockout strains were cultured in 25 mL of YPD broth at 30 °C for 24 h. Then, they were subcultured for over 12 generations. Recombinant yeasts, XF0-h, XF0-H, and XF0-LH, which had lost the plasmid and exhibited no resistance, were selected in YPD medium with 200 μg/mL G418.

2.6. Yeast Inoculation and Fermentation

We dissolved the wort and adjusted the concentration to 8 °P, and then we added hydrolyzed broken rice syrup at a concentration of 60 g/L. We transferred the wort into 100 mL sterile measuring cylinders, ensuring volume of 100 mL. We centrifuged a specific volume of the seed culture (12,000 rpm, 2 min), collected the cells, washed them twice with sterile water, and then transferred the washed cells into the fermentation wort. The original and recombinant strains were inoculated with 2 × 10⁶ cells/mL in the measuring cylinders and then incubated anaerobically at 27 °C, with the weight loss of CO₂ recorded every 24 h. After fermentation, we measured the higher alcohol content and other physical and chemical parameters.

2.7. Optimization of Fermentation Conditions

2.7.1. Fermentation with Different Concentrations of Wort

We dissolved the wort and adjusted the concentration to 4 °P, 6 °P, 8 °P, 10 °P, and 12 °P. Then, we dispensed 100 mL of each wort concentration into sterile 100 mL measuring cylinders and added hydrolyzed broken rice syrup at a concentration of 60 g/L. We inoculated 2 × 10⁶ cells/mL of XF0-LH into each cylinder and incubated these anaerobically at 27 °C for 5 days.

2.7.2. Fermentation with Different Inoculum Sizes of Yeast

We dissolved the wort and adjusted the concentration to 8 °P and dispensed 100 mL into 100 mL measuring cylinders. We added hydrolyzed broken rice syrup at a concentration of 60 g/L. The inoculum sizes of XF0-LH were 0.5 × 10⁶, 1 × 10⁶, 2 × 10⁶, 3 × 10⁶, and 4 × 10⁶ cells/mL. These were incubated anaerobically at 27 °C for 5 days.

2.7.3. Response Surface Optimization Experiment

A total of 17 experiments were conducted by the Box–Behnken design for the response surface methodology. As shown in

Table 1. Coded and actual values of factors in Box–Behnken design.

Factors	Levels
	−1	0	1
A: Wort concentration (°P)	6	8	10
B: Hydrolyzed broken rice syrup addition (g/L)	40	60	80
C: Inoculum size of XF0-LH (cells/mL)	1 × 106	2 × 106	3 × 106

, the design included three factors at three levels: wort concentration (A), hydrolyzed broken rice syrup addition (B), and inoculum size of XF0-LH (C). The levels were set at −1, 0, and 1, with the higher alcohol content (Y) as the response variable. This approach was employed to optimize the experimental conditions through response surface optimization.

2.8. Analytical Methods

The supernatant of the fermentation broth was collected by centrifuge (8000 rpm, 5 min) and was filtered through a 0.22 μm membrane. The content of higher alcohols and ethanol in whisky was measured by gas chromatography (GC) with the internal standard method. Butyl acetate was used as the internal standard for the higher alcohols. The test sample consisted of 980 μL of fermentation broth and 20 μL of butyl acetate (17.66 mg/mL). Nitrogen was used as the carrier gas at a flow rate of 1.5 mL min⁻¹. The FID and injector temperatures were 230 °C and 250 °C, respectively. The injection volume was 0.4 μL. The column temperature program was as follows: started at 60 °C without stopping, 60 °C to 120 °C at the rate of 10 °C min⁻¹, and 120 °C to 200 °C at the rate of 30 °C min⁻¹, and then 200 °C was maintained for 3 min. In addition, the ethanol was determined with acetone as the internal standard. The sample consisted of 800 μL of H₂O, 100 μL of fermentation broth and 100 μL of acetone (7.32 mg/mL). The column temperature procedure was 50 °C for 5 min and 50 °C to 200 °C at the rate of 30 °C min⁻¹, and 200 °C was maintained for 3 min [31].

2.9. Statistical Analysis

Due to the varying degrees of fermentation in the experimental samples, the results were uniformly converted and analyzed under a 50% vol alcohol system. Each experiment was conducted in triplicate, and the results are presented as “mean ± standard deviation”. Significance analysis was performed using SPSS 27.0, response surface data were processed with Design Expert 13, and bar charts, growth curves, and response surface plots were generated using Origin 2022.

3. Results

3.1. Construction of Recombinant Yeast Strains

3.1.1. Construction of gRNA Targeting Plasmid of HOM2 Gene

The recombinant plasmid p426-gRNA-HOM2-kanMX was successfully constructed following the method described in Section 2.2, as shown in Figure 2 [14].

3.1.2. Results of HOM2 Gene Knockout

After electrotransformation, XF0-h, XF0-H and XF0-LH were selected by PCR with the HOM2-A/HOM2-D primers and template strains’ DNA, with XF0, XF0-h and XF0-L as negative controls. The results are shown in Figure 3.

If the HOM2 gene is not knocked out, a 1376 bp negative fragment will be amplified by PCR. If both alleles of HOM2 are knocked out, a 917 bp positive fragment (knockout 459 bp) will be amplified by PCR. If one allele of HOM2 is knocked out, both 917 bp and 1376 bp fragments will be amplified by PCR. As shown in Figure 3, XF0-LH, XF0-h and XF0-H were successfully constructed.

3.1.3. Relative Expression Levels of HOM2

To further validate the expression of the HOM2 gene after knockout, total RNA was extracted from XF0, XF0-h, XF0-H, XF0-L, and XF0-LH, and reverse transcription was performed for cDNA. Then, Real-Time Quantitative PCR was conducted using the cDNA as the template to analyze HOM2 gene expression, as shown in Figure 4. The results indicate that the HOM2 gene expression levels in the XF0-H and XF0-LH are extremely low, confirming the successful knockout of the HOM2 gene. In the case of the XF0-h with one HOM2 allele knocked out, the HOM2 gene expression level is only 6% of the XF0.

3.1.4. Verification of Plasmid Loss

The XF0-h, XF0-H and XF0-LH strains, which had lost resistance plasmids, were screened by the replica plating method. As shown in Figure 5, strains without plasmids normally grew on plates without antibiotics but did not grow or slowly grew on plates with G418. This indicates that the plasmid p426-gRNA-HOM2-kanMX has been successfully eliminated from XF0-h, XF0-H, and XF0-H.

3.2. Effect of HOM2 Gene Knockout on Higher Alcohols

Fermentation was conducted with the original strains XF0 and XF0-L, as well as the recombinant strains XF0-h, XF0-H and XF0-LH. The content of higher alcohols in the fermentation broth was determined, as shown in Figure 6.

Figure 6A illustrates that the higher alcohol content in the fermentation broth of XF0-h (704 ± 8 mg/L) and XF0-H (685 ± 6 mg/L) was reduced by 23.93% and 25.98%, respectively, compared to XF0 (925 ± 9 mg/L). The higher alcohol content in the fermentation broth of XF0-LH (685 ± 19 mg/L) was reduced by 15.81% compared to XF0-L (814 ± 29 mg/L). This indicates that the deletion of the HOM2 gene can decrease the higher alcohol content.

As shown in Figure 6B, the isoamylol content in the fermentation broth of XF0-h (374 ± 6 mg/L) and XF0-H (409 ± 3 mg/L) was reduced by 34.99% and 28.96%, respectively, compared to XF0 (575 ± 12 mg/L). The isoamylol content in the fermentation broth of XF0-LH (273 ± 2 mg/L) was reduced by 49.25% compared to XF0-L (537 ± 15 mg/L). This indicates that the deletion of the HOM2 gene significantly decreases the isoamylol content.

Figure 6C demonstrates that the n-propanol content in the fermentation broth of XF0-h (59 ± 7 mg/L) and XF0-H (41 ± 4 mg/L) was reduced by 52.53% and 67.28%, respectively, compared to XF0 (124 ± 13 mg/L). The n-propanol content in the fermentation broth of XF0-LH (48 ± 1 mg/L) was reduced by 55.96% compared to XF0-L (109 ± 12 mg/L). This indicates that the deletion of the HOM2 gene can significantly decrease the n-propanol content.

3.3. Effect of HOM2 Gene Knockout on Fermentation Rate

Fermentation was conducted with the XF0, XF0-L, XF0-h, XF0-H, and XF0-LH strains. The weight loss was recorded every 24 h to generate cumulative weight loss curves. The results are shown in Figure 7.

As shown in Figure 7, the fermentation rates of XF0 and XF0-L are similar, with cumulative weight loss stabilizing after 96 h, marking the end of fermentation, which is indicative of a rapid fermentation process. In contrast, the HOM2 knockout strains (XF0-h, XF0-H, XF0-LH) display slower fermentation rates, with the cumulative weight loss after 240 h remaining lower than that of the original strains. This indicates that the deletion of the HOM2 gene extends the fermentation cycle.

3.4. Experimental Results of Process Optimization

Based on the combined information from Figure 6 and Figure 7, this indicates that, despite having similar fermentation rates, the higher alcohol production of XF0-LH shows no significant difference compared to XF0-h and XF0-H. However, the relative production of isoamylol is significantly reduced by XF0-LH. Therefore, XF0-LH was selected as the fermentation strain for subsequent process optimization experiments.

3.4.1. Effect of Wort Concentration on Higher Alcohols in Whiskey

As shown in Figure 8, the C/N ratio in the fermentation broth of wort concentrations of 4 °P, 6 °P, 8 °P, 10 °P, and 12 °P, adding 60 g/L of hydrolyzed broken rice syrup, shows a decreasing trend as the wort concentration increases from 4 °P to 12 °P. The content of higher alcohols in the fermentation broth initially decreases and then increases, with the lowest higher alcohol content of 761 ± 65 mg/L at 8 °P wort. In 4 °P and 6 °P wort, the assimilable nitrogen content cannot satisfy the metabolic demand of yeast, and yeast will synthesize amino acids through the sugar metabolism pathway, and a large amount of α-keto acids will be formed. One part of the α-keto acids is converted to the amino acids needed by yeast, and the other part is decarboxylated and dehydrogenated to produce higher alcohols. It produced more higher alcohols. In 10 °P and 12 ° P wort, the osmolality increase leads to a decrease in the rate of viable cells of the yeast, and the content of assimilable nitrogen is higher than the metabolic needs of the yeast, so amino acids will be generated to produce more higher alcohols. An 8 °P wort’ C/N ratio is appropriate to produce the lowest higher alcohols content of 761 ± 65 mg/L because the synthetic α-keto acids are converted to the corresponding amino acids.

3.4.2. Effect of Yeast Inoculum Size on Higher Alcohols in Whiskey

As shown in Figure 9, as the yeast inoculum size increases from 0.5 × 10⁶ to 4.0 × 10⁶ cells/mL, the higher alcohol content in the fermentation broth initially decreases and then increases. The lowest higher alcohol content is observed at the yeast inoculum size of 2.0 × 10⁶ cells/mL. At the yeast inoculum size of 0.5 × 10⁶ cells/mL and 1.0 × 10⁶ cells/mL, the yeast population in the fermentation broth is low, leading to a primary focus on growth and reproduction. During this phase of active metabolism, the yeast produces higher alcohols. Additionally, the insufficient yeast population results in incomplete consumption of fermentable sugars, leaving a high residual sugar concentration, which further drives higher alcohol production through anabolic pathways. At the yeast inoculum size of 4.0 × 10⁶ cells/mL, the yeast population is high, causing nutrient limitation early in the fermentation process, which triggers increased production of higher alcohols.

3.5. Optimization of Fermentation Conditions Using Response Surface Methodology

3.5.1. Response Surface Methodology Model and Statistical Significance Analysis

The Box–Behnken design was employed for the response surface methodology, with investigating factors including wort concentration (A), hydrolyzed broken rice syrup addition (B), and an inoculum size of XF0-LH (C). The response value was the higher alcohol content (Y) in the whiskey. The experimental results are detailed in

Table 2. Response surface test design and results.

RUN	Factors			Higher Alcohols Content (g/L)
	X1	X2	X3
1	−1	−1	0	569
2	1	−1	0	934
3	−1	1	0	643
4	1	1	0	660
5	−1	0	−1	605
6	1	0	−1	1048
7	−1	0	1	612
8	1	0	1	715
9	0	−1	−1	854
10	0	1	−1	913
11	0	−1	1	931
12	0	1	1	595
13	0	0	0	628
14	0	0	0	616
15	0	0	0	649
16	0	0	0	667
17	0	0	0	661

.

The experimental data were analyzed using Design Expert 11 to obtain a multiple quadratic regression equation: $Y=-930.785+315.759A+4.006B+119.893C-2.178AB-42.573AC-4.929BC-2.617A^2+{0.169B}^2+111.434C^2$.

3.5.2. Variance and Confidence Analysis of Higher Alcohols Content in Whiskey

A further analysis of variance was conducted on the results of each experimental group from Table 2, and the findings are presented in

Table 3. Response surface regression model analysis of variance.

Source	Sum of Squares	Df	Mean Square	F-Value	p-Value	Significance
Model	3.50 × 105	9	3.89 × 104	39.90	<0.0001	**
A	1.08 × 105	1	1.08 × 105	110.70	<0.0001	**
B	2.86 × 104	1	2.86 × 104	29.35	0.001	**
C	4.00 × 104	1	4.00 × 104	41.08	0.0004	**
AB	3.04 × 104	1	3.04 × 104	31.19	0.0008	**
AC	2.90 × 104	1	2.90 × 104	29.78	0.0009	**
BC	3.89 × 104	1	3.89 × 104	39.92	0.0004	**
A2	461.30	1	461.30	0.48	0.5134
B2	1.93 × 104	1	1.93 × 104	19.78	0.003	**
C2	5.23 × 104	1	5.23 × 104	53.69	0.0002	**
Lack of fit	4.98 × 103	3	1.66 × 103	3.61	0.1235
Residual	6.82 × 103	7	973.76
Pure error	1.84 × 103	4	459.67
Cor total	3.57 × 105	16

Note. **: highly significant (p < 0.01).

.

As shown in Table 3, a quadratic model was applied to regress the higher alcohol content in whiskey. The model showed a highly significant difference with a p-value < 0.0001, indicating a robust fit and suitability for subsequent optimization analyses. The lack-of-fit test had a p-value > 0.05, suggesting that the quadratic model provided an adequate fit across the regression range, effectively modeling the higher alcohol content in whiskey. An analysis of F-values revealed that the most influential factor was wort concentration (A), followed by an inoculum size of XF0-LH (C) and hydrolyzed broken rice syrup addition (B). In the significance analysis, using a threshold of p < 0.05, it was found that the linear terms A, B, C; the quadratic terms B² and C²; and the interaction terms AB, AC, and BC all had significant effects on the higher alcohol content in whiskey (p < 0.01). The model’s R² value was 0.9809, indicating a strong correlation between the actual and predicted values, with 98.09% of the variance in the higher alcohol content explained by the model. The adjusted R² value was 0.9563, suggesting that 95.63% of the variability was accounted for by the investigated variables, demonstrating that the optimized model effectively captures the process conditions. The predicted R² value was 0.7686, with the difference between the predicted and actual R² values being less than 0.2, further validating the model’s accuracy and low error rate.

3.5.3. Response Surface Analysis and Verification Test of Higher Alcohols Content in Whiskey

Based on the results in Table 2, three-dimensional response surface plots were generated using Design Expert to visualize the interaction effects between the investigating factors AB, AC, and BC, as shown in Figure 10. According to the fundamental principles of the response surface methodology, circular contour lines typically indicate a minimal interaction between two factors, while elliptical or saddle-shaped contour lines suggest a significant interaction. Furthermore, a steep response surface indicates strong interaction between factors, whereas a flatter surface suggests weaker interaction [34,35,36].

As shown in Figure 10, when the hydrolyzed broken rice syrup addition is constant, the higher alcohol content in whiskey increases with the wort concentration. When the wort concentration is constant, the higher alcohol content in whiskey initially decreases and then increases with the amount of the hydrolyzed broken rice syrup. The interaction between wort concentration and the hydrolyzed broken rice syrup addition is significant, with the wort concentration having a more pronounced effect on the higher alcohol content than the hydrolyzed broken rice syrup addition. When the inoculum size of XF0-LH is constant, the higher alcohol content in whiskey increases with the concentration of the wort. When the wort concentration is constant, the higher alcohol content in the whiskey initially decreases and then increases with the inoculum size of XF0-LH. The interaction between the wort concentration and inoculum size of XF0-LH is significant, with the wort concentration having a more pronounced effect on the higher alcohol content than the inoculum size of XF0-LH. When the inoculum size of XF0-LH is constant, the higher alcohol content in whiskey initially decreases and then increases as the hydrolyzed broken rice syrup addition is varied. When the hydrolyzed broken rice syrup addition is constant, the higher alcohol content in whiskey initially decreases and then increases with the inoculum size of XF0-LH. The interaction between the hydrolyzed broken rice syrup addition and the inoculum size of XF0-LH is significant, with the inoculum size of XF0-LH having a more pronounced effect on the higher alcohol content than the hydrolyzed broken rice syrup addition.

Based on Figure 10, it is indicated that the interactions between wort concentration and the hydrolyzed broken rice syrup addition (AB), wort concentration and the inoculum size of XF0-LH (AC), and the hydrolyzed broken rice syrup addition and the inoculum size of XF0-LH (BC) all significantly affect the higher alcohol content in whiskey. The optimal parameters were a wort concentration of 9.808 °P, hydrolyzed broken rice syrup addition of 77.743 g/L, and yeast inoculum size of 2.651× 10⁶ cells/mL. So, the predicted higher alcohol content in whiskey is 565 mg/L. A 50 L scale-up fermentation experiment was conducted to obtain whiskey with a higher alcohol content of 556 mg/L at a wort concentration of 9.8 °P, hydrolyzed broken rice syrup addition of 78 g/L, and yeast inoculum size of 2.7 × 10⁶ cells/mL. The result is in close agreement with the theoretical optimized value, so the optimal parameters are applicable for practical whiskey production.

4. Discussion

The higher alcohols produced by yeast during alcohol fermentation play an important role in the taste of whiskey. The HOM2 gene plays a key role in the metabolism of higher alcohols in yeast. Therefore, genetic analysis of the HOM2 gene is important for controlling the content of higher alcohols in whiskey. It shows that HOM2 knockout strains can reduce the content of n-propanol, isoamylol and higher alcohols in whiskey, which is consistent with previous reports [37]. This is due to the HOM2 knockout, which reduces the expression of aspartate β-semialdehyde dehydrogenase, which further reduces the generation of α-ketobutyric acid from aspartylphosphate, resulting in the reduced production of n-propanol and isoamylol (Figure 11) [38,39]. Isoamylol in the fermentation broths of HOM2 and LEU1 knockout strains was significantly lower than that of the HOM2 knockout strains. This is because LEU1 knockout resulted in reduced production of α-ketoglutarate to produce α-ketohexanoic acid, which in turn reduced isoamylol production.

Substrata also affect the content of higher alcohols in whiskey. Differences in the composition and C/N ratio of wort and hydrolyzed broken rice syrup affect the physiology and the fermentation performance of the yeast. When the C/N ratio is large, the nitrogen content is low and yeast produces alcohol slowly and produces α-keto acids quickly, resulting in synthesizing higher alcohols quickly [40]. When the C/N ratio is in the optimal range, yeast produces alcohol by normal metabolic pathways, resulting in reducing the production of higher alcohols. When the C/N ratio is small and the nitrogen content is high, yeast converts amino acids to form higher alcohols by the Ehrlich metabolic pathway [40]. Adding hydrolyzed broken rice syrup to the fermentation broth can adjust the concentration of sugars, thereby increasing the alcohol content of the whiskey. A low concentration of sugars leads to insufficient yeast metabolism [41], reducing the alcohol yield and affecting yeast growth, reproduction, and metabolic pathways. Excessive sugars can alter yeast metabolic pathways, resulting in increased higher alcohol production. In high osmotic pressure environments, yeast cells may be damaged, leading to inhibited metabolic activity and reduced fermentation capacity, which can cause incomplete fermentation and increased residual sugars. When the free amino nitrogen level in the fermentation broth is appropriate, moderately increasing the concentration of sugars can maximize the ethanol yield when yeast cells have sufficient nutrients [42].

5. Conclusions

The optimum content of higher alcohols can enrich the flavor of whiskey. But when there is an excessive content of higher alcohols in whiskey, especially isoamylol, the drinker is more likely to experience symptoms of intoxication. It has been shown that HOM2 knockout strains could reduce the content of isoamylol and higher alcohols, which could be used to brew low-ID whiskey. It has also been shown that wort concentration and an inoculum size of XF0-LH affect the content of higher alcohols in whiskey. The optimal fermentation conditions for high-quality whiskey were obtained through single-factor experiments and the Box–Behnken design: a wort concentration of 9.8 °P, hydrolyzed broken rice syrup concentration of 77.7 g/L, and inoculum size of 2.65 × 10⁶ cells/mL. So, through scaled-up experiments, low-ID whiskey is successfully brewed by a 50 L fermenter with the higher alcohols content of 556 mg/L.