Soy sauce moromi samples that have been fermented for 20–90 days (when citrulline was significantly increased) were collected from Guangdong Meiweixian Flavoring Foods Co., Ltd. L-citrulline, arginine, ornithine hydrochloride, and glucose were purchased from Shanghai Aladdin Reagent Co., Ltd. Yeast extract; peptone; and de Man, Rogosa and Sharpe (MRS) and nutrient broth (NB) medium were purchased from Guangdong Huankai Microbial Technology Co., Ltd. Bacterial genomic DNA extraction kit, RNAsimple total RNA extraction kit, and FastKing-RT SuperMix kit were purchased from Tiangen Biochemical Technology (Beijing) Co., Ltd. Real-Time PCR Super Mix (SYBR Green with anti-Taq) kit was purchased from Beijing Polymer Biotechnology Co., Ltd.

The screening of citrulline accumulation microbes was performed using the methods of Zhang et al. (2015) with some modifications. Firstly, the moromi samples were serially diluted with 0.9% NaCl (w/v), plated on MRS and NB agar supplemented with 80 mg/l cycloheximide and 2% NaCl, and then stationary cultured at 30°C for 2–5 days. Subsequently, single colonies were subcultured on MRS and NB agar supplemented with 4 g/l arginine and 0.005% bromocresol purple (w/v) to test the arginine utilization ability. The strains able to utilize arginine were distinguished by forming a purple halo around the colonies. Finally, the candidates were inoculated in MRS (or NB) broth with 18% NaCl (w/v) and 10 g/l arginine and the modified soy sauce broth with 10 g/l arginine, then cultured at 30°C for 7 days. The arginine and citrulline content were determined, and the A/C rate and A/O rate were calculated by Eq. 1.

where C₁ is the citrulline (or ornithine) content, M₁ is the molar mass of citrulline (or ornithine), C₂ is the arginine consumption, M₂ is the molar mass of arginine, and V is the volume.

Isolates (W. confusa and P. acidilactici) were grown in 50 ml MRS broth at 30°C and statically cultured to the late-log phase, as assessed by OD600; the cells were harvested by centrifugation at 8,000 × g for 4 min at 4°C, then washed twice with 0.9% NaCl. The cell pellets were resuspended in saline with OD600 ≈ 1, then 1 ml of strain solution was cultured in 9 ml of modified MRS broth for static culture. The modified MRS broth were based on a medium [peptone (4 g/l), meat extract (5 g/l), yeast extract (5 g/l), sodium acetate (5 g/l), K₂HPO₄ (2 g/l), triammonium citrate (2 g/l), MgSO₄⋅7H₂O (0.2 g/l), MnSO₄⋅4H₂O (0.05 g/l), and Tween 80 (1 g/l)] with addition of NaCl [6, 9, 12, 15, or 18% (w/v)], arginine (2, 4, or 6 g/l), citrulline (2 or 4 g/l), ornithine (0.5 or 1 g/l), glucose (0, 20, 40, or 80 g/l), urea (5, 10, 20, or 40 g/l), ethanol [0.5, 1, 2, or 4% (w/v)], and glycerin [0.5, 1, 2, or 4% (w/v)]. The pH of the medium was adjusted to 4, 4.8, 5.6, 6.4, or 7.2 by acetic acid/NaOH, and the incubation temperatures were set at 15°C, 30°C, and 37°C, respectively. Except for the groups added with NaCl that were cultured at 30°C for 48 h, other groups were cultured for 24 h. After being cultured, samples were centrifuged at 12,000 × g for 6 min, and the supernatants were taken to measure the citrulline and arginine contents. For gene expression and intercellular and extracellular amino acid analysis, the cells were cultured for 5 h.

The phenolic compounds were dissolved in ethanol at the concentration of 10 mg/ml, then filtered through a membrane filter for sterility. The modified MRS fortified with 10 g/l arginine and 18% NaCl was sterilized and cooled down to room temperature. Then, the phenolic solution was added into the sterilized medium to obtain a final concentration of 100 mg/l. The control sample was added with the same volume of ethanol. The prepared cell pellets were then resuspended in the above culture for static culture at 30°C. Arginine, citrulline, and ornithine contents of supernatants were subsequently analyzed.

The fermented soy sauce moromi is a complex system involving a variety of species, and a central composite design (CCD) experiment was applied to verity the effect of main factors and to optimize the phenolic compounds on citrulline accumulation (Supplementary Table 1). The fresh moromi (after 20 days of fermentation) fortified with 4 mg/l arginine was treated as the simulated fermentation sample. The arginine consumption, the formation of citrulline and ornithine, the A/C rate, and the A/O rate were considered as dependent variables. The experimental design and data analysis were performed using the Design Expert software, version 8.0.6.

All experiments were performed in triplicate unless otherwise noted, with the mean value and standard error reported. The data were analyzed by ANOVA using SPSS (version 22, IBM, United States), and significant differences among the mean values were determined by adopting Tukey’s multiple range test, with significance defined at p < 0.05.

A total of 34 strains were isolated from moromi, and all strains exhibited the ability to degrade arginine and accumulated variable concentrations of citrulline besides Leuconostoc garlicum. We found that six strains were detected to possess a strong ability to decompose arginine and accumulate citrulline through screening tests. Among these isolates, two strains, which were identified as P. acidilactici and W. confusa (the sequence of 16S DNA is shown in the Supplementary Material), exhibited an outstanding citrulline accumulation capability (Table 1). The A/C rates P. acidilactici and W. confusa reached up to 40.9% and 30.2%, respectively, and citrulline was accumulated over 900 mg/l. Representative strains of P. acidilactici and W. confusa isolated from moromi were selected for further analysis.

The effects of different medium constituent conditions (urea, glucose, ethanol, glycerol, NaCl, and pH) on citrulline accumulation of the two strains are shown in Figure 1. The A/C rate was approximately 6% in the mediums containing different concentrations of urea, glucose, ethanol, and glycerol cultured with W. confusa and P. acidilactici, which was much lower than that cultured in modified MRS broth (above 30%). Moreover, those constituents had little influence on citrulline accumulation, directly indicating a small effect on citrulline accumulation. The A/C rates were even less than 2% when W. confusa and P. acidilactici were cultured at a medium with pH 4.0, and the growth rate of the strains was inhibited severely; a similar trend was found in the arginine catabolism. These stains were isolated from moromi when the pH was 4.8, which was the optimal pH for citrulline accumulation.

The addition of different contents of salt significantly affected the citrulline accumulation capacity of W. confusa and P. acidilactici. These strains had a strong ability to consume arginine at salinity less than 6%, and citrulline was less than 1,000 mg/l. The arginine utilization capacity rapidly decreased when the NaCl concentration was higher than 9%. When NaCl was at 9%, the highest citrulline concentration was found for both strains, and the A/C rates significantly increased to approximately 60%. When NaCl increased to 18%, both stains exhibited a strong citrulline accumulation capacity in comparison with the control group, and the A/C rate and citrulline content were slightly higher in P. acidilactici than those in W. confusa. Then, we cultured the strains in modified moromi culture, and the results showed that the A/C rates of W. confusa and P. acidilactici were 51.3% and 44.0%, respectively (Supplementary Table 3). These evidences confirm that the NaCl concentration plays a key role in citrulline accumulation in soy sauce.

The ADI pathway gene expression of W. confusa and P. acidilactici was determined under different NaCl concentrations (0, 9, and 18%). As shown in Figure 2, the expression of arcA and arcC genes was significantly repressed in P. acidilactici by NaCl, decreasing by approximately 20-fold and 50-fold at 9 and 18% NaCl, respectively. The expression of the gene arcB showed a higher reduction at the same conditions, which were, respectively, 53-fold and 86-fold for the two strains. These results mean that the expression of arcB was relatively lower than that of arcA and arcC in the presence of NaCl, especially at 9% NaCl. Likewise, the enzyme activity of the ADI pathway was significantly decreased in the presence of NaCl, and OTC activity declined even more (Supplementary Figure 1). Therefore, the unequal proportional inhibition of ADI and OTC activity directly led to the reduction of arginine utilization and the increase of citrulline accumulation. However, the reduction of ADI and OTC by NaCl was lower than that of relative gene expression.

The arc gene expression of W. confusa was found to be different with P. acidilactici. The expression of arcA and arcB genes in W. confusa slightly decreased (by approximately two-fold) in the presence of NaCl, and the expression rate of arcA/arcB was about 1.0, indicating that citrulline accumulation in W. confusa was not caused by the unequal down-regulation of arc genes. In order to further investigate the citrulline accumulation mechanism of W. confusa, the intercellular and extracellular amino acids were determined at different NaCl concentrations for 5 h. As shown in Figure 3, the sum of citrulline and ornithine content was less than 70% of arginine consumption, especially without NaCl, which indicates that the ADI pathway was not the only way for arginine consumption. Furthermore, arginine metabolism was very different in the presence of NaCl: (1) ornithine was the main product of the ADI pathway when cultivated without NaCl, (2) when cultivated at 9% NaCl, citrulline in the extracellular space was largely increased and ornithine was decreased, and (3) when the NaCl concentration was increased to 18%, the growth of W. confusa was significantly inhibited, so did the arginine and citrulline transportation ability, leading to the accumulation of citrulline inside the cells. These results indicate that the different arginine/citrulline transportation abilities under different salt stresses may be the reason for the citrulline accumulation in W. confusa.

The citrulline content and A/C rate of the two strains were slightly increased at pH 4.8. The effect of pH on citrulline accumulation was then investigated in the presence of 18% NaCl. As shown in Figure 4, P. acidilactici possessed a stronger arginine consumption ability than W. confusa in the medium with NaCl. Citrulline accumulation in P. acidilactici remained stable at three pH values at 18% NaCl, while it exhibited a significantly higher citrulline accumulation capacity at pH 4.8 than that at pH 5.6 and 6.4 in W. confusa. The results indicate that pH can regulate arginine metabolism in the presence of high concentrations of NaCl. There was no significant change in arcA and arcB gene expressions in W. confusa under different pH conditions in 18% NaCl, but a general over-expression (six–seven-fold) of arcC was observed at a low-pH condition (Supplementary Figure 2). The low intercellular arginine and citrulline content at pH 4.8 indicates that large amounts of arginine were converted to citrulline and then secreted into the medium.

Our previous study showed a significant reduction of citrulline when the temperature at the primary fermentation period in soy sauce was set at 15°C (Zhou et al., 2019). In this study, the arginine consumption in P. acidilactici was lesser at 15°C than that at 30 and 37°C without the addition of NaCl, whereas the opposite behavior was observed in W. confusa. Both strains accumulated minimal citrulline at 15°C (Figure 5). As previously mentioned, the A/C rates did not exceed 10% at any culture temperature without NaCl, while they were significantly increased at 18% NaCl, especially when the strains were cultured at 30°C, the A/C rate improved by 10–15-fold. Noticeably, both strains exhibited the lowest A/C rate (lower than 25%) and citrulline content at 15°C, which was much lower than that at 30°C and 37°C, directly explaining the reason of low accumulation of citrulline at low fermentation temperature in soy sauce. Under 18% NaCl, the arc gene expression changes of the two strains were below two-fold at different temperatures (Supplementary Figure 3), suggesting the small effect of temperature on the ADI pathway under high salt stress.

In low-salt solid-state fermented soy sauce, several times of “daochi” during the fermentation process can increase the uniformity and oxygen content of the moromi. While for traditional Chinese-type soy sauces, the moromi is usually fermented in open air for almost 6 months. The effects of oxygen and light on citrulline accumulation of the key strains were then investigated. Small differences of A/C rate and citrulline accumulation were observed in W. confusa when cultured at four environmental combinations of light and oxygen. However, in P. acidilactici, the light improved citrulline accumulation, and the maximum citrulline content (1,459.0 mg/L) and A/C rate (57.1%) were observed in light and anaerobic condition. The opposite results were observed in W. confusa.

Eight phenols were added into the medium to test their effects on citrulline accumulation of the two strains, and both positive and negative effects were observed (Figure 6). In the control group, the A/C rate and citrulline in P. acidilactici at 24 h were 780.38 mg/l and 49.04%, respectively, and increased slightly after 72 h of culture. Noticeably, nearly all of the tested phenols exhibited a strong inhibitory effect against the A/C rate of P. acidilactici at 24 h, but the A/C rate then increased after 72 h of culture. Quercetin, TBHQ, gallic acid, and protocatechuic acid showed an inhibitory effect on citrulline formation by P. acidilactici cultured for 24 h. After 72 h of culture, the level of citrulline in the protocatechuic acid group was increased and had no significant difference compared with the control group, but the citrulline content in the BHA group dropped by nearly half. However, the phenols that showed citrulline inhibitory effects on P. acidilactici exhibited the opposite effect on W. confuse; they rapidly improved the A/C rates, and citrulline accumulation in BHA and TBHQ was significantly increased after 72 h of culture. In summary, among the tested phenols, only quercetin and gallic acid with 100 mg/l could be the potential inhibitors of citrulline formation in soy sauce, and they also showed an inhibitory effect on microbial growth (Supplementary Figure 4).

Fermented moromi is a complex system. In order to verify the actual inhibitory effect of quercetin and gallic acid on moromi, a CCD experiment was designed with two major factors (pH and temperature). Arginine, citrulline, and ornithine content was determined after 48 h and 7 days of culture, and arginine consumption, citrulline and ornithine formation, A/C rate, and A/O rate were set as the responses. The validity of the models was statistically analyzed by ANOVA as presented in Table 2. The desirable model with p-values < 0.05 and lack of fit >0.05 indicated that the model terms were significant while the pure error was not significant. We found the models of 7 days-Arg, 48 h and 7 days-ornithine, and 7 days-A/C and A/O rate could well fit the real data. The 20-day fermented moromi supplied with 4 mg/ml arginine was used as a simulated fermented model; therefore, citrulline was not only accumulated by P. acidilactici and W. confusa. Among those models, only temperature and quercetin show a significant linear effect (p < 0.05) on all parameters and the 7 days-A/C rate model, while pH and gallic acid were less important. According to these two models, we set the pH value as original value and the temperature at 15°C. The optimum quercetin and gallic acid contents as suggested by the software were 100 and 10 mg/l, respectively, corresponding to the lowest A/C rate of 29.8% and the highest A/O rate of 34.1%. To verify the prediction, cultured moromi in the presence or absence of quercetin and gallic acid was analyzed in triplicate to determine the optimal A/C and A/O rates. The A/C rate and A/O rate obtained at these conditions are 28.4 and 33.9%, respectively, corresponding to citrulline of 1,326.7 mg/l, which was 17.3% lower than that of moromi without phenols (Table 3).

Citrulline has proven to be the major EC precursor in soy sauce, which mainly originates from the catabolism of arginine. Although Saccharomyces cerevisiae can produce citrulline via arginine metabolism, the generation of citrulline is majorly attributed to the metabolism of arginine by malolactic bacteria via the ADI pathway (Fang et al., 2018). In previous work, citrulline content in moromi was found to continuously increase throughout the fermentation period, and it was significantly accumulated when fermented from 20 to 90 days in Cantonese soy sauce (Zhou et al., 2019). In this study, we isolated 34 strains from moromi and determined their arginine utilization ability. The results showed that all strains exhibited the ability to degrade arginine and accumulate variable concentrations of citrulline except for L. garlicum. During koji-making, only six isolates were identified to be able to accumulate citrulline via the ADI pathway (Zhang et al., 2015). We speculated that citrulline did not accumulate in the koji-making period due to the low arginine content, low abundance of strain with ADI pathway, or lack of stimuli (NaCl). P. acidilactici and W. confusa were isolated from moromi as the key strains for citrulline accumulation in soy sauce. Enterococcus faecium and Enterobacter sp. showed strong arginine consumption and weak citrulline accumulation ability when cultured in the 18% NaCl medium, revealing that they can be potential candidates to reduce EC in soy sauce. Noticeably, one strain of P. acidilactici and five isolates of W. confusa were screened out from moromi, while different strains of W. confusa had different citrulline accumulation abilities in the presence or absence of NaCl. The available genome sequencing research revealed an unexpected diversity of both gene arrangement and composition of the ADI gene clusters. Even for closely related microorganisms, the ADI-related genes were organized in one or more clusters (Zúñiga et al., 2002). Further research is required to verify the difference of those closely related microorganisms’ ADI clusters and their effects on arginine metabolism. Beyond that, the ADI pathway was easily induced by substrate, cell energy level, and environmental factors (Araque et al., 2013; Zhou W. et al., 2017; Majsnerowska et al., 2018).

The species of lactic acid bacteria, such as Lactobacillus, isolated from the winemaking process was mainly involved in malolactic fermentation. The ADI pathway genes and arginine degradation were deeply studied (Vrancken et al., 2009a; Araque et al., 2011, 2013; Rimaux et al., 2012). The effect of environmental pH and ethanol on arginine metabolism and citrulline release was determined by regulating the expression of ADI pathway genes. In case of P. acidilactici and W. confusa, the arginine metabolism was slightly affected by pH, ethanol, glycerin, glucose, and urea, while NaCl had a noticeable effect on citrulline accumulation. The A/C rate significantly increased in both strains when cultured under salt stress. Vrancken et al. (2009b) reported that the A/C rate of L. fermentum was not affected by temperature, but it increased with increasing NaCl concentrations, and salt directly facilitated the direct ornithine production. Zhang et al. (2015) pointed out that the presence of a high level of NaCl was a critical environmental factor for citrulline accumulation for P. acidilactici isolated from koji, and they advised that a better control of the production process to keep a low NaCl concentration may be a good strategy to reduce the risk of the product. However, in this study, P. acidilactici and W. confusa were found to accumulate more citrulline in the presence of 9–15% NaCl, and the A/C rate of P. acidilactici was higher at 9 and 15% NaCl than that at 18%. Therefore, fermentation at a low NaCl concentration may not only fail to reduce citrulline but also improve the citrulline accumulation. To the best of our knowledge, research on the regulation of the ADI pathway by salt stress is rare. We found that the expression of the three arc genes was suppressed in the presence of 18% NaCl, and the inconsistent reduction rate of arcA/arcB expressed in the ADI pathway was the reason causing the accumulation of citrulline in P. acidilactici, while the interference of the transport system of three key amino acids in the presence of NaCl was the reason for citrulline accumulation in W. confusa.

Mitigation of EC by lowering the temperature was proven to be an efficient method via reducing the reaction of EC precursors. Although the adaptability of L. fermentum through the ADI pathway under salt stress was much better than that under constant temperatures (Vrancken et al., 2009b), temperature also regulated the ADI pathway even in the presence of a high concentration of NaCl (Figure 5). For P. acidilactici and W. confusa, the A/C rate reduced to approximately 20% at 15°C, and citrulline content at 15°C was less than half of that at 30°C. In the absence of NaCl, the extracellular citrulline concentration in L. fermentum decreased with increasing temperature, while the ornithine concentration rose even higher (Vrancken et al., 2009b). However, a drastic reduction of the growth rate of P. acidilactici and W. confusa was observed at a high concentration of NaCl; actually, they stopped growing when the NaCl content in the culture medium was greater than 9%. The cells started to metabolize arginine mainly into citrulline and minor amounts of ornithine. Therefore, an alternative to reduce citrulline content is to lower the fermentation temperature.

Phenols have been proven to be effective to adjust the metabolism of arginine via the ADI pathway (Alberto et al., 2012). Gallic and protocatechuic acids can regulate EC catabolism and reduce the EC content by up to 91.9% in the course of yellow rice wine leavening by reducing the arginine consumption in Lactobacillus hilgardii, but the remaining citrulline concentration increased (Zhou W. et al., 2017). Similar effects were found for bamboo leaves extract (Zhou et al., 2020). Therefore, the accurate screening of target phenols is essential for the usage of phenolic compounds to decrease EC content in fermented food within a given environment. In this study, both positive and negative effects on citrulline accumulation in key strains were observed for different phenols, and quercetin and gallic acid were observed to be more efficient to inhibit citrulline accumulation. Furthermore, the simulation experiment results indicated that quercetin had a significant effect on the A/C rate in the fermentation process of moromi, and the A/C rate in fermented liquor with optimum phenols was 17.7% lower than that in the control group. In addition, gallic acid did not significantly affect EC formation, but quercetin significantly reduced EC content by 48.38% by preventing the transformation from citrulline to EC (Zhou et al., 2021). The addition of quercetin and gallic acid improved the A/O rate by up to 33.9%, resulting in high ornithine residues, which was also beneficial for the reduction of EC (Fang et al., 2013; Zhou et al., 2021). Quercetin and gallic acid were the common phenolic compounds in fermented food with strong antioxidant and anticancer activity (Sachithanandam et al., 2020). Therefore, future study may focus on investigating the exact inhibiting effect and mechanism of quercetin and gallic acid on citrulline accumulation and EC formation.

This study provides a better understanding on the citrulline accumulation mechanism by the key strains in soy sauce and also proposes a potential method for effectively controlling citrulline by lowering fermentation temperature and addition of polyphenols.