2.1 Microorganisms

Four bacterial strains [nitrate-reducing Pseudomonas sp. 61-02 and three LAB strains (LM-1, LP-2, and LS-4)] were used in this study. They were previously isolated from Yamahai-shubo described before (Fujiwara et al., 2013; Wakai et al., 1990). The yeast strain, KZ-06 U, used in this study was a strain of S. cerevisiae that was developed for industrial-scale sake brewing at Kizakura Co., Ltd.

2.2 Yamahai-shubo preparation

The initial mixture of Yamahai-shubo was prepared as follows. The mixture included 333 g of total rice [250 g of raw rice (102 g of koji and 231 g of steamed rice)] and 255 mL of water with the addition of KNO₃ (60 mg/L). koji is a type of rice that is made from koji mold (A. oryzae FK-5, Konnomoyashi Co., Ltd., Hyogo, Japan) with polished rice made by the industrial koji manufacturing system (Churitsu Industry Co., Ltd., Tokyo, Japan). Fermentation with koji followed the same method described before (Aya, 1995). A polishing ratio of 70% was employed for the raw rice and applied to koji and steamed rice.

Four bacterial strains [nitrate-reducing Pseudomonas sp. 61-02 and three LAB strains (LM-1, LP-2, and LS-4)] were prepared as follows. Strain 61-02 was independently cultured in PN medium [1% (W/V) polypeptone, 1% (W/V) KNO₃] for 3 days at 15°C (Wakai et al., 1990) and was adjusted to reach a cell concentration of 10⁴ cells/mL. Three LAB strains were separately cultured in koji extract (sake meter −60) for 3 days at 20°C and were added to the initial mixture of Yamahai-shubo at a cell concentration of 10³ cells/mL. Yamahai-shubo preparation was totally completed in 25 days with the initial mixture of Yamahai-shubo after adding any of the nitrate-reducing bacterial strain 61-02 and/or three LAB bacterial strains (a total of 16 variations depending on the bacterial strain(s) added; see Table 1).

The temperature during Yamahai-shubo preparation was kept at 8°C until day 4. After day 5, the preparation was warmed up by contact with hot water at 65°C to increase the temperature by one degree per day until it reached 21°C on day 16. After holding the temperature of the mixture at 21°C for another 5 days, the temperature was decreased to 15°C on day 21, to 13°C on day 22, to 12°C on day 23, and was then maintained at 12°C until day 25. The yeast strain KZ-06 U was separately cultured in koji extract (sake meter −60) for 2 days at 30°C and added between days 10 and 15 of the preparation to reach a cell concentration of 10⁶ cells/mL when the nitrite concentration of the Yamahai-shubo preparation was less than 1.0 mg/L. The timing of yeast addition in the absence of strain 61-02 was the same as the timing of yeast addition in the presence of strain 61-02. The final preparation reached approximately 500 mL in volume and was called “Yamahai-shubo.” For use at the industrial scale, the process was considered complete at day 25, even if the disappearance of nitrite was delayed and the yeast growth period was shortened.

The cell numbers of strain 61-02 and three LAB strains (LM-1, LP-2, and LS-4) were determined by the method described previously (Fujiwara et al., 2013; Wakai et al., 1990).

2.3 Yamahai-shikomi sake brewing

The polishing ratio of raw rice was set at 70%, just as in the case for Yamahai-shubo preparation. Brewage ingredients (Yamahai-shubo, koji, steamed rice, and water) were mixed at three different times (soe, naka, and tome feeding procedures) to smoothly complete the fermentation (Anzawa et al., 2014; Furukawa et al., 2000). The soe procedure (23 mL of Yamahai-shubo, 9.6 g of koji, 26.6 g of steamed rice, and 33 mL of water were added) was carried out at 15°C from day 1 to day 2. Both the naka (day 3, 12.0 g of koji, 54.6 g of steamed rice, and 64 mL of water) and the tome (day 4, 14.4 g of koji, 99.4 g of steamed rice, and 112 mL of water) feeding procedures were carried out by adding the components at 10°C. After the tome procedure (170 g of raw rice, total volume 450 mL), sake brewing was monitored by measuring the weight reduction of the fermentation mixture, corresponding to CO₂ evolution for 25 days. The temperature of the fermentation mixture was increased by about 0.5°C for another 6 days, kept at 13°C for an additional 7 days, and then promptly lowered to 10°C while the temperature was maintained for an additional 12 days. After sake brewing (25 days), the samples were separated by centrifugation (6800 × g, 20 min, 4°C), and the supernatants were called “Yamahai-shikomi” sake samples (a total of 16 types of Yamahai-shikomi sake samples depending on 16 variations of Yamahai-shubo; see Table 2). The volume of sake and the weight of sake lee were measured. Sake lees are the final sediments obtained in the sake brewing process and are alternatively called “sake-kasu.”

2.4 Analysis of fermentation factors in Yamahai-shubo preparation and sake samples

Basic fermentation factors for sake meter (SM), ethanol (Alc), total acidity (TA), amino acidity (AA), and Ultraviolet absorption (UV) were determined by the standard method established by National Research Institute of Brewing (National Research Institute of Brewing, 2017), while glucose (Glc) was measured by the method described previously (Fujiwara et al., 2013) using a specific instrument (Arkray ADAMS Glucose GA-1151, Kyoto, Japan). Fermentation factors for volatile organic compounds [ethyl acetate (EtAc), isobutyl acetate (iBuAc), isobutyl alcohol (iBuOH), isoamyl acetate (iAAc), isoamyl alcohol (iAOH), ethyl caproate (EtCp), and isoamyl acetate/isoamyl alcohol (iAAc/iAOH)] were determined by following the same standard methods (National Research Institute of Brewing, 2017). Other fermentation factors, including concentrations of other organic acids (pyruvic acid, citric acid, succinic acid, malic acid, and lactic acid), were determined by the method described previously (Kohsaka et al., 2016): 25 μL of sample was taken in a 10 mL screw-top test tube, and 1 mL of 0.3 M HCl/1-propanol solution was added. It was heated at 100°C for 1 h. After cooling, 1 mL of methyl tert-butyl ether containing 4 μg of methyl laurate as an internal standard and 1 mL of distilled water were added and mixed. The ether layer was analyzed by a gas chromatography GC-2014 (Shimazu Corporation, Kyoto, Japan). Diacetyl was derivatized by Shinwa DS-DA (Shinwa Chemical Industries, Ltd., Kyoto, Japan, https://shinwa-cpc.co.jp/products/ds/) and analyzed using the same gas chromatography system. All data are the averages of two or three independent experiments.

The qualities of Yamahai-shikomi sake samples were graded by sensory scores evaluated by four or six well-trained panelists. The sensory scores of sake samples held at 20°C consisted of qualitative evaluations of aromas and tastes according to five different grades (1, very good; 2, good; 3, normal; 4, bad; 5, very bad).

2.5 Principal component analysis (PCA)

Principal component analysis (PCA) was performed with Excel Statistical analysis 2008 software for Windows (Social Survey Research Information Co., Ltd., https://www.bellcurve.jp/), employing fermentation factors (SM, Alc, TA, AA, Glc, UV, EtAc, iBuAc, iBuOH, iAAc, iAOH, EtCp, and iAAc/iAOH) and sensory scores at the final point (25 days) of Yamahai-shubo (Table 1) and Yamahai-shikomi sake samples (Table 2). In addition, the data for sake volume and sake lee, sake lee/sake volume, were used for the analysis.

3.1 Effect of four bacterial strains (nitrate-reducing pseudomonas sp. 61-02, Le. Mesenteroides LM-1, Lp. Plantarum LP-2, and Ll. Sakei LS-4) on Yamahai-shubo preparation samples

The four bacterial strains previously found in Yamahai-shubo were examined to evaluate their effects along the time course of Yamahai-shubo preparation (a total of 25 days). Changes in three fermentation factors are depicted in Figure 1, and others are shown in Table 1. In Figure 1, three factors—nitrite concentration (a-1 and a-2), acidity (b-1 and b-2), and amino acid concentration (c-1 and c-2)—were described in the absence (a-1, b-1, and c-1) and presence (a-2, b-2, and c-2) of nitrate-reducing Pseudomonas sp. 61-02, following the method as described in Materials and Methods. Nitrite is known to be essential in Yamahai-shubo preparation process; it is converted from nitrate, which is derived from PN medium for the preparation of strain 61-02 (see Materials and Methods) and inhibits the growth of wild yeasts that are concomitant in the initial mixture for Yamahai-shubo. As shown in Figure 1a-1,a-2, strain 61-02 was responsible for nitrite production; its concentration increased in the first half of Yamahai-shubo preparation and drastically decreased before the second half. However, the coculture of strain LM-1 only with strain 61-02 was found to retard the decrease in the concentration of nitrite (Figure 1a-2). It is noted that the delay is a characteristic of strain LM-1 when it is cocultured with strain 61-02. The acidity increased along the time course of Yamahai-shubo preparation in the absence and presence of strain 61-02, which is necessary for the growth of yeast in sake brewing of strain KZ-06 U; on the other hand, there was a slower rise in acidity for the coculture of strain LM-1 only with strain 61-02 and the simple culture of strain 61-02 without any LAB strains (Figure 1b-1,b-2). Furthermore, there seemed to be no remarkable difference in amino acid concentration in the presence and absence of strain 61-02 (Figure 1c-1,c-2). In addition, with the exceptions of Yamahai-shubo preparations contained strains 61-02 and LM-1 (No. 10 in Table 1) and only strain LM-1 (No. 2 in Table 1), the viable cell numbers of four bacterial strains (61-02, LM-1, LP-2, and LS-4) in Yamahai-shubo became extremely decreased after the preparation for 25 days (Figure 1d-1,d-2 and Table 1). In the first half of the preparation of Yamahai-shubo, there seemed to be no significant difference in cell number of LABs (strains LM-1, LP-2, and LS-4) in the presence and absence of strain 61-02 (Figure 1d-1,d-2). However, Yamahai-shubo preparations contained only strain 61-02 (No. 9 in Table 1) and no addition of any strains (No. 1 in Table 1) had no LABs on day 1, but they increased LABs on day 5. Although the growth of LABs is confirmed in the industrial scale Yamahai-shubo preparation without the addition of microorganisms (Ashizawa & Saito, 1966a; Koyanagi et al., 2016), it is known that the process is not stable at the industrial scale. On the other hand, we have previously confirmed that Yamahai-shubo can be stably incubated at an industrial scale by adding cocci LAB and bacilli LAB (Fujiwara et al., 2013) without being affected by environmental microorganisms, and therefore, as a negative control, we used Yamahai-shubo preparations contained only strain 61-02 and no addition of any strains. After yeast strain KZ-06 U was added in Yamahai-shubo preparation between day 11 and day 19, alcohol concentration changes were almost the same at the 16 patterns (data not shown).

Comparing the components at the final point (25 days) of Yamahai-shubo preparation in Table 1, the addition of strain LP-2 (Nos. 5 and 11) contributed to the highest level of the alcohol concentration (Alc) and total acidity (TA), simultaneously resulting in essential requirements of strains 61-02 and/or LM-1 for Yamahai-shubo utilizing the advantages of strain LP-2 without strain LS-4.

We further investigated fermentation factors for Yamahai-shikomi sake similarly as Yamahai-shubo (Table 2) and applied them to principal component analysis (PCA).

3.2 Principal component analysis (PCA) for Yamahai-shubo and Yamahai-shikomi sake

To investigate the suitability of the bacterial combination of the nitrate-reducing Pseudomonas sp. 61-02 and three LAB strains (LM-1, LP-2, and LS-4), we performed principal component analysis (PCA) by using fermentation factors for 16 samples of Yamahai-shubo and Yamahai-shikomi sake, as shown in Tables 1 and 2. As a result of PCA score plotting of fermentation factors, the contribution ratios of principal component 1 (PC1) and principal component 2 (PC2) were 50.0% and 14.8%, respectively (Figure 2). In addition, the eigen values of principal component 1 (PC1) and principal component 2 (PC2) were 15.5 and 4.6, respectively.

In the score plot (Figure 2a), when Yamahai-shubo preparations and Yamahai-shikomi sake samples contained strains 61-02 and LM-1 (No. 10 in Tables 1 and 2) or only strain LM-1 (No. 2), their plots were located in the first and fourth quadrants, respectively, apart from other combinations of strains that occurred in the second and third quadrants. In the second and third quadrants, two major groups were formed. Plots for the group containing strain LP-2 (LP-2 group; Nos. 3, 5, 11, 13, and 15) occurred in the third quadrants, forming one cluster. In contrast, plots for the group including strain LS-4 (LS-4 group; Nos. 4, 6, 7, 12, and 14) occurred in the second quadrant. The sample only contained strain 61-02 (No. 9) was located in LS-4 group. Interestingly, combinations of three LAB strains (LM-1, LP-2, and LS-4) with (No. 16) and without strain 61-02 (No. 8) were included in this LS-4 group. These results showed that the characteristics of strain LS-4 in the LS-4 group were predominant in the second quadrant when strain LP-2 was present with strain LM-1.

In order to evaluate the fermentation factors that contribute to the principal component scores, we confirmed the factor loadings by their plots (Figure 2b). The factor loadings are the correlation coefficients between the principal components and the variable fermentation factors. The closer the values of factor loading are to 1 or −1, the more the factors contribute to the principal components. On the PC1 axis, the factor loading of TA for Yamahai-shikomi sake was 0.91. On the other hand, the factor loading of the sensory score for taste showed a negative value of −0.79. On the PC2 axis, the factor loadings of iBuAc for Yamahai-shubo and of EtAc for Yamahai-shikomi sake were 0.88 and 0.30, respectively. Since EtAc and iBuAc generally provide aromas undesirable for sake products (Yoshizawa, 1999), their positive values are disadvantageous in the quality of sake. In contrast, the factor loading of the sensory score for aroma of Yamahai-shikomi sake showed a negative value of −0.31 (Figure 2b). The factor loading plots showed that negative values of sensory evaluation contributed to the principal components of the PC1 and PC2 axes (Figure 2b). In the score plot, since the LP-2 group showed negative values on the PC1 and PC2 axes (third quadrant, Figure 2a), strain LP-2 was confirmed to be responsible for a group with a higher sensory score. On the other hand, the sample containing strain 61-02 and LM-1 (No. 10) showed positive values on the PC1 and PC2 axes (first quadrant, Figure 2a). EtAc and iAAC/iAOH for Yamahai-shubo were both located in the first quadrant (Figure 2b). Although iAAC/iAOH is the indicator of the good quality of sake, EtAc for Yamahai-shubo may decrease sake quality while iAAC/iAOH for Yamahai-shubo may be less effective in improving sake quality.

Judging from these results, it was found that strains LP-2 and LS-4 were important in the Yamahai-shikomi sake in the presence of strains 61-02 and LM-1.

3.3 Effect of Lp. Plantarum LP-2 and Ll. Sakei LS-4 in Yamahai-shubo preparation on the concentration of organic acids in Yamahai-shikomi sake

The sample coexisting strains LP-2 and LS-4 was located in the LS-4 group (Figure 2a). So, to evaluate the effect of coexisting bacilli-LABs on sake quality, Yamahai-shubo samples were prepared by adding different proportions of strains LP-2 and LS-4 to its initial mixture in the presence of conventional amounts of strains 61-02 and LM-1. After sake brewing with those Yamahai-shubo samples for 25 days, the Yamahai-shikomi sake samples were withdrawn and prepared by following the method as described in the Materials and Methods section, and then concentrations of organic acids (pyruvic acid, citric acid, succinic acid, malic acid, and lactic acid) in Yamahai-shikomi sake and Yamahai-shubo samples were determined for each sample (Figure 3). Although the sake samples were produced in the same way as those Yamahai-shubo samples including bacterial strains with different proportions of strains LP-2 and LS-4, there seemed to be no significant difference in the concentration of five organic acids (Figure 3a). However, taking into account the change in each acid relative to the proportion of strains LP-2 to LS-4 strains in Yamahai-shubo preparation, only lactic acid was found to decrease with an adequate correlation to the decrease in the ratio of LS-4 strain (R² = 0.9074; Figure 3b). Although the proportion of strains LP-2 and LS-4 was unequal, there seems to be a tendency for lactic acid concentration to increase as the mixing ratio of strain LS-4 decreases.

3.4 Effect of Lp. Plantarum LP-2 and Ll. Sakei LS-4 in Yamahai-shubo preparation on the concentration of diacetyl in Yamahai-shikomi sake

We investigated their effect on the concentration of diacetyl. The sake brewing was carried out in the same way as shown in Figure 3, and the concentrations of diacetyl in Yamahai-shikomi sake samples were determined. As shown in Figure 4, the sample prepared in the absence of strain LS-4 significantly exhibited the lowest concentration of diacetyl. Lower concentrations of diacetyl are so favorable for improving sake quality that the use of strain LS-4 in Yamahai-shubo preparation had an undesirable effect on the final sake samples. Although diacetyl concentration of LP2 + LS4 (10²:1) was higher than that of LP2 + LS4 (2:1), there was no significant difference between them (Figure 4). This result was supported by the statistical analysis for the sensory scores for aroma of each Yamahai-shikomi sake sample (Figure 5). The sensory scores for aroma showed that Yamahai-shikomi sake prepared with strain LP-2 had significantly lower values than those with strain LS-4, with the exception of the case with a proportion of LP-2:LS-4 = 10²:1.

Through this investigation, in order to reduce the diacetyl concentration in Yamahai-shikomi sake, it was found that the addition of strain LS-4 should be avoided in Yamahai-shubo preparation, and it is likely that strain LP-2 plays a more significant role in reducing the concentration of diacetyl with strains LM-1 and 61-02 in Yamahai-shubo preparation and Yamahai-shikomi sake brewing.