pH Auto-Sustain-Based Fermentation Supports Efficient Gamma-Aminobutyric Acid Production by Lactobacillus brevis CD0817

: Gamma-aminobutyric acid (GABA) plays a role in several physiological functions. GABA production by lactic acid bacteria has attracted considerable interest; however, there is need to improve production. This study aimed to develop a pH auto-sustain (PAS)-based GABA fermentation process for Lactobacillus brevis CD0817, with L-glutamic acid (solubility ~6.0 g/L and isoelectric point 3.22) as the substrate. Firstly, we determined the optimum levels of vital factors affecting GABA synthesis using Erlenmeyer ﬂask experiments. The results showed that optimal levels of sugar, yeast extract, Tween-80, manganese ion, and temperature were 5.0 g/L, 35.0 g/L, 1.0 g/L, 16.0 mg/L, and 30.0 ◦ C, respectively. The added L-glutamic acid (650 g per liter of medium) mostly existed in the form of solid powder was slowly released to supply the substrate and acidity essential for GABA production with the progress of fermentation. Based on the optimizations, the PAS-based GABA fermentation was performed using a 10 L fermenter. The PAS-based strategy promoted GABA synthesis by the strain of up to 321.9 ± 6.7 g/L after 48 h, with a productivity of 6.71 g/L/h and a substrate molar conversion rate of 99.6%. The ﬁndings suggest that the PAS-based fermentation is a promising method for GABA production by lactic acid bacteria.


Introduction
Gamma-aminobutyric acid (GABA) is a non-proteinogenic amino acid that is extensively distributed in nature [1,2] and functions as the major inhibitory neurotransmitter in mammals [3,4]. GABA is also associated with the auditory cortex [5] and psychosis states [6]. Furthermore, GABA possesses therapeutic effects on liver injury [7], hepatic encephalopathy [8], and mechanical pain [9] and also possesses antidiabetic and hypotensive properties [10]. Therefore, GABA has been widely used in the food and pharmaceutical industries [11]. Moreover, GABA is a precursor for the synthesis of biodegradable polyamide 4 [12][13][14].
Over the past three decades, GABA production by lactic acid bacteria (LAB) has been extensively studied because of the safe status of LAB [15][16][17]. Generally, a glutamic acid decarboxylase (GAD) system in the LAB cell is responsible for synthesizing GABA [18]. The system consists of glutamate/GABA antiporter (GadC) and glutamate decarboxylase (GadA/GadB). The mechanism and process of GABA synthesis by the GAD system is as follows: (i) glutamate is imported into the cell by GadC; (ii) the imported glutamate is decarboxylated to GABA by intracellular GadA/GadB, simultaneously releasing one L. brevis CD0817, a robust GABA producer, was previously isolated from fecal samples from a healthy adult [35]. The strain was stored in 50% glycerol at −80 • C and was resuscitated by culturing at 32 • C for 2 d following streaking on a plate.
The starting seed medium contained 25.0 g/L of yeast extract, 10.0 mg/L of MnSO 4 ·H 2 O, 28.0 g/L of MSG, 2.0 g/L of Tween-80, and 50.0 g/L of glucose and was maintained at a pH of 5.0. The starting fermentation medium contained yeast extract (25.0 g/L), MnSO 4 ·H 2 O (25.0 mg/L), Tween-80 (2.0 g/L), and glucose (25.0 g/L). Just before inoculation, sterilized L-glutamic acid solid powder was added in the amount of 650 g per liter of fresh fermentation broth. The added L-glutamic acid mostly existed in the form of solid powder due to its low solubility. L-glutamic acid, glucose, and the remaining components were separately sterilized at 121 • C for 20 min and mixed together before use [35].
We successively optimized the starting fermentation medium, fermentation temperature, glucose level of the starting seed medium, and seed age. The preparation of the seed medium varied with the advancing optimization. For optimizing the fermentation medium and temperature, the inocula were prepared at 32 • C using the starting seed medium. For optimizing the glucose level of the seed medium, the inocula were made at 32 • C. For optimizing the seed age, the inocula were prepared at 30 • C using the optimized seed medium. The optimized seed medium was identical to the starting seed medium, with the exception of the glucose being reduced to 10 g/L. The cells of L. brevis CD0817 were cultivated in a ZHWY-2102C rotary shaker (Zhicheng Inc., Shanghai, China) under constant rotation (100 rpm) and could be used as the inoculum when the absorbance at 600 nm (A 600 ) reached 3.0-6.0.

Optimization Experiments
Single-factor experiments were performed to determine the optimal levels of key factors affecting GABA synthesis one by one. Once a key factor was optimized, its optimized level was thereafter incorporated into the fermentative formula and kept fixed in the following experiments. The starting fermentation medium, fermentation temperature, glucose level of the starting seed medium, and seed age were successively optimized. For the fermentation medium, levels of the glucose, nitrogen source, Mn 2+ , and Tween-80 were sequentially optimized. To achieve this, fermentation experiments were performed using a 250 mL Erlenmeyer flask (Sichuan Shuniu Co., Ltd., Chongzhou, China). The fermentation was initiated by transferring seed culture (inoculum size, 10%, v/v) into the flask containing 100 mL fermentation medium supplemented with 65 g solid powder L-glutamic acid. The fermentation was conducted at 32 • C for 96 or 72 h to optimize level of glucose/nitrogen source or of Mn 2+ /Tween-80, respectively, in the fermentation medium. The fermentation was conducted for 72 h to optimize the fermentation temperature. The fermentation was conducted at 30 • C for 60 h to optimize the seed medium glucose level or seed age. On the premise of assuring experiment quality, the stepwise reduction of fermentation time enhanced research efficiency. Each Erlenmeyer flask experiment was conducted in triplicate. Samples were aseptically withdrawn every 12 h to determine GABA content and A 600 .

Fermentation Trial
The optimized seed medium contained 25.0 g/L yeast extract, 10.0 mg/L MnSO 4 ·H 2 O, 28.0 g/L MSG, 2.0 g/L Tween-80, and 10.0 g/L glucose and was maintained at a pH of 5.0. The optimized fermentation medium contained yeast extract (35.0 g/L), MnSO 4 ·H 2 O (50.0 mg/L), Tween-80 (1.0 g/L), and glucose (5.0 g/L). Just before inoculation, 650 g of sterilized L-glutamic acid solid powder per liter of fermentation medium was added. The seed culture was made as described above. The seed culture (inoculum size, 10%, v/v) was inoculated in a 10 L fermenter containing 3.0 L of the optimized fermentation medium supplemented with 1950.0 g of solid powder L-glutamic acid and incubated at 30.0 • C under constant agitation (50 rpm) for 48 h. The fermenter was equipped with EasyFerm Plus 120 pH electrodes (Hamilton, Bonaduz, Switzerland) and two layers of six-flat-bladed impellers. The fermentation was performed under sealed conditions, and the cover of the fermentation tank was fitted with a silicone pipe to serve as an exhaust for carbon dioxide. The exhaust pipe was submerged in water to prevent air from entering the tank. Samples were aseptically withdrawn every 2 h to determine the GABA and residual sugar content and cell growth. All the samples were stored at −20 • C for further analysis. The experiment was repeated three times.

Analytic Procedures
The GABA concentration of the culture broth was determined by high-performance liquid chromatography. Briefly, after a 5 min treatment in a boiling bath, a sample of the culture broth was centrifuged at 8000× g for 3 min using a Sorvall ST 8R centrifuge (Thermo Fisher Scientific, Waltham, USA). Thereafter, the supernatant (20.0 µL) was subjected to derivatization at 25 • C for 5.0 min by mixing with 20.0 µL of the derivatization reagent and 100.0 µL borate buffer. Approximately 50 µL of the derived sample was injected into an Agilent 1200 system (Agilent Technologies Inc., Santa Clara, CA, USA) equipped with an Agilent Eclipse XDB-C18 column (4.6 × 150 mm, 5 µm). The high-performance liquid chromatography conditions were as follows: mobile phase flow rate 0.8 mL/min, column temperature 30 • C, and detection wavelength 338 nm. All data were expressed as the means ± standard deviations.
The amino acid derivatization reagent was freshly prepared by dissolving 10.0 mg of o-phthalaldehyde and 10 µL of 2-mercaptoethanol in 2.5 mL of acetonitrile. The borate buffer was prepared by adding 4.9 g of boric acid into 100 mL of ultrapure water, pH was adjusted to 10.4 using NaOH, and the solution was further diluted to 200.0 mL. The mobile phase was prepared as follows: 2.7 g of sodium acetate trihydrate and 200.0 µL of triethylamine were dissolved in 0.9 L of ultrapure water, pH was adjusted to 7.3 using acetic acid, the solution was diluted to a final volume of 1 L, and then mixed with 250.0 mL acetonitrile [36].
Cell biomass was expressed as A 600 value. The cells were recovered by centrifugation at 5000× g for 5 min using a Sorvall ST 8R centrifuge and resuspended in an appropriate volume of 0.8% NaCl according to the dilution factors of different samples. Finally, cell growth was monitored by determining the A 600 value on an UV1200B ultraviolet-visible spectrophotometer (Mapada Instruments, Shanghai, China).
Residual glucose was quantitated by 3,5-dinitrosalicylic acid method, according to the manufacturer's instructions. Briefly, 2 mL of 3,5-dinitrosalicylic acid reagent was added into a tube containing 1 mL of 10-fold diluted samples, mixed, heated for 5 min in a boiling water bath, and immediately cooled with cold running water. Thereafter, 9 mL distilled water was added and mixed, and the A 540 value of the mixture was determined using the UV1200B spectrophotometer.

Calculations
The substrate molar conversion rate (k) was calculated using the formula below: where c GABA is concentration of GABA (g/L), m Glu is mass of consumed L-glutamic acid (g), 147.1 is molar mass of L-glutamic acid (g/mol), 103.1 is molar mass of GABA (g/mol), and V is volume (L). It should be noted that 1 L of fresh medium increased to 1.41 L after L-glutamic acid was depleted.

Statistical Analysis
All statistical analyses were performed using SPSS statistics V22.0 software (International Business Machines Corporation, Armonk, NY, USA), and a post hoc test was performed using Tukey's test at p < 0.05. All charts and graphs were plotted using the Origin 2017 software (OriginLab, Northampton, MA, USA). Data were represented as means ± standard deviations (n =3).

Effect of Glucose on GABA Production
Glucose is an ideal carbon source for GABA-synthesizing LAB [28,37]. Therefore, we determined the optimal glucose concentration for efficient GABA production. The results indicated that glucose concentration had little effects on the trend of GABA synthesis (Figure 1), that is, GABA was rapidly produced from 0 to 36 h across all the concentrations (0-40.0 g/L), plateauing afterward. However, GABA content varied with glucose concentration, with 5.0 g/L of glucose yielding the highest GABA concentration (283.5 ± 23.0 g/L) after 48 h of fermentation. Higher concentrations of glucose (10.0-40.0 g/L) instead slightly reduced GABA production ( Figure 1). Lactobacillus brevis is a microaerophilic, obligately heterofermentative lactic acid bacterium that uses the phosphoketolase pathway to produce a mixture of lactic acid, ethanol, acetic acid, and CO 2 as products of glucose fermentation [38,39]. The decrease in GABA concentration at higher glucose concentration could be attributed to the effects of the metabolites, such as lactic acid, ethanol, and acetic acid. Moreover, a mildly acidic (around 5.0) [40,41] environment is favorable for maximizing GAD activity, whereas extremely low pH (below 4.0) inhibits GAD activity [42,43]. High content of glucose generates massive amounts of organic acids and, thus, may inhibit GAD activity by further acidifying the environment [34,44]. However, the exact mechanism of glucose in GABA production remains poorly understood and requires further studies. 2017 software (OriginLab, Northampton, MA, USA). Data were represented as means ± standard deviations (n =3).

Effect of Glucose on GABA Production
Glucose is an ideal carbon source for GABA-synthesizing LAB [28,37]. Therefore, we determined the optimal glucose concentration for efficient GABA production. The results indicated that glucose concentration had little effects on the trend of GABA synthesis (Figure 1), that is, GABA was rapidly produced from 0 to 36 h across all the concentrations (0-40.0 g/L), plateauing afterward. However, GABA content varied with glucose concentration, with 5.0 g/L of glucose yielding the highest GABA concentration (283.5 ± 23.0 g/L) after 48 h of fermentation. Higher concentrations of glucose (10.0-40.0 g/L) instead slightly reduced GABA production ( Figure 1). Lactobacillus brevis is a microaerophilic, obligately heterofermentative lactic acid bacterium that uses the phosphoketolase pathway to produce a mixture of lactic acid, ethanol, acetic acid, and CO2 as products of glucose fermentation [38,39]. The decrease in GABA concentration at higher glucose concentration could be attributed to the effects of the metabolites, such as lactic acid, ethanol, and acetic acid. Moreover, a mildly acidic (around 5.0) [40,41] environment is favorable for maximizing GAD activity, whereas extremely low pH (below 4.0) inhibits GAD activity [42,43]. High content of glucose generates massive amounts of organic acids and, thus, may inhibit GAD activity by further acidifying the environment [34,44]. However, the exact mechanism of glucose in GABA production remains poorly understood and requires further studies.
Additionally, the LAB strain examined in this study could grow and produce GABA in a glucose-free medium ( Figure 1), indicating that a nitrogen source could also provide some essential carbons for microbial growth. The carbons mainly comprised peptides, amino acids, and nucleotides as the product label indicated. Overall, 5.0 g/L of glucose supported optimal GABA synthesis by L. brevis CD0817, which was lower than that (≥10.0 g/L) required by other strains of L. brevis for efficient GABA production [28,34,37,44].  Additionally, the LAB strain examined in this study could grow and produce GABA in a glucose-free medium (Figure 1), indicating that a nitrogen source could also provide some essential carbons for microbial growth. The carbons mainly comprised peptides, amino acids, and nucleotides as the product label indicated. Overall, 5.0 g/L of glucose supported optimal GABA synthesis by L. brevis CD0817, which was lower than that (≥10.0 g/L) required by other strains of L. brevis for efficient GABA production [28,34,37,44].

Effect of Nitrogen Source on GABA Production
Several studies have shown that the nitrogen level has a considerable influence on GABA production by LAB [29,34,37]. For instance, tryptone, yeast extract, and whey powder are the best nitrogen sources for L. brevis HYE1 [45], L. plantarum EJ2014 [46], and L. brevis A3 [47], respectively; and a mixture of peptone, meat extract, and yeast extract enhanced GABA production by L. brevis CRL 2013 [22]. Therefore, the identification of the right nitrogen source and its optimal level is essential in enhancing GABA production [37,46,47]. In the present study, we examined the effects of 37 nitrogen sources on GABA production. The results indicated that the nitrogen source remarkably modulated GABA synthesis, as a GABA titer difference of up to 26.0-fold was observed among the nitrogen sources examined ( Figure S1). Among which, yeast extract FM408 yielded the highest GABA output (239.4 ± 6.5 g/L), whereas soy peptones 02-19 yielded the lowest output (9.2 ± 0.5 g/L). Therefore, yeast extract FM408 was employed for further study.
Further analysis indicated an increase in GABA production with increase in yeast extract FM408 concentration from 0 to 30.0 g/L ( Figure 2). Although higher concentrations (40.0-50.0 g/L) of yeast extract FM408 increased the cell biomass, it did not increase GABA production further (Figure 2). To guarantee efficient production of GABA, the dosage of yeast extract FM408 was suggested to be 35.0 g/L, slightly higher than 30.0 g/L. The glutamate (1.85 ± 0.05 g/L) introduced by 35.0 g/L of yeast extract FM408 had a minimal effect on the bioprocess.

Effect of Nitrogen Source on GABA Production
Several studies have shown that the nitrogen level has a considerable influence on GABA production by LAB [29,34,37]. For instance, tryptone, yeast extract, and whey powder are the best nitrogen sources for L. brevis HYE1 [45], L. plantarum EJ2014 [46], and L. brevis A3 [47], respectively; and a mixture of peptone, meat extract, and yeast extract enhanced GABA production by L. brevis CRL 2013 [22]. Therefore, the identification of the right nitrogen source and its optimal level is essential in enhancing GABA production [37,46,47]. In the present study, we examined the effects of 37 nitrogen sources on GABA production. The results indicated that the nitrogen source remarkably modulated GABA synthesis, as a GABA titer difference of up to 26.0-fold was observed among the nitrogen sources examined ( Figure S1). Among which, yeast extract FM408 yielded the highest GABA output (239.4 ± 6.5 g/L), whereas soy peptones 02-19 yielded the lowest output (9.2 ± 0.5 g/L). Therefore, yeast extract FM408 was employed for further study.
Further analysis indicated an increase in GABA production with increase in yeast extract FM408 concentration from 0 to 30.0 g/L ( Figure 2). Although higher concentrations (40.0-50.0 g/L) of yeast extract FM408 increased the cell biomass, it did not increase GABA production further (Figure 2). To guarantee efficient production of GABA, the dosage of yeast extract FM408 was suggested to be 35.0 g/L, slightly higher than 30.0 g/L. The glutamate (1.85 ± 0.05 g/L) introduced by 35.0 g/L of yeast extract FM408 had a minimal effect on the bioprocess. The effect of yeast extract FM408 on (a) GABA production and (b) cell growth. This was evaluated using a 250 mL Erlenmeyer fermentation flask at 32.0 °C. Cell growth was expressed as A600. Vertical bars represent standard deviations of means (n = 3). The difference in 48 h GABA concentrations was not significant between 40 g/L and 50 (p = 0.979) or 30 g/L of yeast extract FM408 (p = 0.573), but it was significant between 40 g/L and 20 (p = 0.000) or 0 g/L of yeast extract FM408 (p = 0.000). GABA, gamma-aminobutyric acid; A600, absorbance at 600 nm.

Effect of Ions on GABA Production
The effects of several compounds including MnSO4, MgSO4, K2SO4, NaCl, CuSO4, ZnSO4, FeSO4, Ba(OH)2, CoSO4, and NiSO4 on the GABA production were individually evaluated. The results demonstrated that only MnSO4 had a positive effect on GABA synthesis (data not shown). Therefore, the effect of MnSO4 concentration on GABA production by L. brevis CD0817 was assessed. As shown in Figure 3a, Mn 2+ was an indispensable factor for efficient GABA production. Overall, 11-55 mg/L Mn 2+ supported peak GABA production; however, maximal cell growth occurred at 2.7-5.5 mg/L Mn 2+ (Figure 3b). We speculated that the high GABA content resulting from 11-55 mg/L Mn 2+ may have suppressed cell growth. Compared with the Mn 2+ -free control, 11-55 mg/L Mn 2+ increased GABA production by 2.3-fold. To assure the need of the strain for Mn 2+ , 16 mg/L (slightly higher than 11 mg/L) of Mn 2+ (equal to 50.0 mg/L MnSO4·H2O) was adopted. . The difference in 48 h GABA concentrations was not significant between 40 g/L and 50 (p = 0.979) or 30 g/L of yeast extract FM408 (p = 0.573), but it was significant between 40 g/L and 20 (p = 0.000) or 0 g/L of yeast extract FM408 (p = 0.000). GABA, gamma-aminobutyric acid; A 600 , absorbance at 600 nm.

Effect of Tween-80 on GABA Production
Tween-80 has been widely used as an enhancer in the manufacture of lactic acid bacterial biologics, such as GABA [16,37], bacteriocins [52], and riboflavin [53]. Therefore, the effects of different concentrations (0-5.0 g/L) of Tween-80 on GABA production were evaluated in this study. The results showed that GABA production increased with increase in Tween-80 concentration from 0 to 1.0 g/L (Figure 4a), which was similar to the previous findings [22,34,54]. However, higher concentrations of Tween-80 did not significantly affect GABA synthesis. Therefore, 1.0 g/L of Tween-80 was selected for further experiments.  Ions can affect GABA production in a strain-specific manner [41]. For instance, Mn 2+ , Cl − , and Ca 2+ plus Na + have been shown to significantly enhance GABA production by L. brevis NCL912 [34], Lactococcus lactis NCDO 2118 [48], and L. brevis K203 [30,37], respectively. Additionally, GADs of Streptococcus salivarius ssp. thermophilus Y2, L. zymae, and L. sakei A156 were activated in vitro by Ba 2+ [49], Mg 2+ [50], and Co 2+ plus Zn 2+ [40], respectively. Generally, some ions can improve GAD activity by acting as cofactors [51]. Hence, it is necessary to screen and optimize ions for a specific GABA-producing LAB strain.

Effect of Tween-80 on GABA Production
Tween-80 has been widely used as an enhancer in the manufacture of lactic acid bacterial biologics, such as GABA [16,37], bacteriocins [52], and riboflavin [53]. Therefore, the effects of different concentrations (0-5.0 g/L) of Tween-80 on GABA production were evaluated in this study. The results showed that GABA production increased with increase in Tween-80 concentration from 0 to 1.0 g/L (Figure 4a), which was similar to the previous findings [22,34,54]. However, higher concentrations of Tween-80 did not significantly affect GABA synthesis. Therefore, 1.0 g/L of Tween-80 was selected for further experiments.
As an oleate, Tween-80 is a growth-stimulating factor for LAB strains [52]. Accordingly, the cell biomass of L. brevis CD0817 increased with increase in Tween-80 concentration in this study (Figure 4b). Increased cell biomass may contribute to GABA production. Moreover, as a non-ionic surfactant, Tween-80 may support oleic acid uptake by cells, which is then converted to cyclopropane fatty acid (a membrane fluidity enhancer) [52]. These findings indicated that Tween-80 may elevate GABA production by improving cell membrane permeability [34]. However, further studies should investigate whether Tween-80 can help maintain the acidic pH by promoting glucose metabolism.
Tween-80 has been widely used as an enhancer in the manufacture of lactic acid bacterial biologics, such as GABA [16,37], bacteriocins [52], and riboflavin [53]. Therefore, the effects of different concentrations (0-5.0 g/L) of Tween-80 on GABA production were evaluated in this study. The results showed that GABA production increased with increase in Tween-80 concentration from 0 to 1.0 g/L (Figure 4a), which was similar to the previous findings [22,34,54]. However, higher concentrations of Tween-80 did not significantly affect GABA synthesis. Therefore, 1.0 g/L of Tween-80 was selected for further experiments.

Effect of Temperature on GABA Production
In the present study, we examined the effect of temperature (25-40 • C) on GABA synthesis by L. brevis CD0817. Generally, temperature significantly regulated GABA generation ( Figure 5), with peak GABA production during the first 48 h achieved at 30 • C. However, 25 • C was slightly more suitable than 30 • C for GABA production by L. brevis CD0817 after 48 h, which could be because GAD activity is generally more stable under lower temperatures [49]. Additionally, temperatures above 30 • C significantly inhibited GABA production ( Figure 5). Overall, 30 • C was the optimal temperature for GABA synthesis by L. brevis CD0817, which was similar to findings in L. brevis strains CRL 1942 [55] and RK03 [37]. Therefore, 30 • C was selected for further experiments. As an oleate, Tween-80 is a growth-stimulating factor for LAB strains [52]. Accordingly, the cell biomass of L. brevis CD0817 increased with increase in Tween-80 concentration in this study (Figure 4b). Increased cell biomass may contribute to GABA production. Moreover, as a non-ionic surfactant, Tween-80 may support oleic acid uptake by cells, which is then converted to cyclopropane fatty acid (a membrane fluidity enhancer) [52]. These findings indicated that Tween-80 may elevate GABA production by improving cell membrane permeability [34]. However, further studies should investigate whether Tween-80 can help maintain the acidic pH by promoting glucose metabolism.

Effect of Temperature on GABA Production
In the present study, we examined the effect of temperature (25-40 °C) on GABA synthesis by L. brevis CD0817. Generally, temperature significantly regulated GABA generation ( Figure 5), with peak GABA production during the first 48 h achieved at 30 °C. However, 25 °C was slightly more suitable than 30 °C for GABA production by L. brevis CD0817 after 48 h, which could be because GAD activity is generally more stable under lower temperatures [49]. Additionally, temperatures above 30 °C significantly inhibited GABA production ( Figure 5). Overall, 30 °C was the optimal temperature for GABA synthesis by L. brevis CD0817, which was similar to findings in L. brevis strains CRL 1942 [55] and RK03 [37]. Therefore, 30 °C was selected for further experiments.

Effect of Glucose in Seed Medium on GABA Production
The glucose level of the starting seed medium for L. brevis CD0817 was up to 50.0 g/L [35]. The findings of this study have shown that 5.0 g/L of glucose in the fermentation medium was the optimal level for GABA production. Therefore, it was necessary to reduce the glucose level of the seed medium for an efficient fermentation process. To

Effect of Glucose in Seed Medium on GABA Production
The glucose level of the starting seed medium for L. brevis CD0817 was up to 50.0 g/L [35]. The findings of this study have shown that 5.0 g/L of glucose in the fermentation medium was the optimal level for GABA production. Therefore, it was necessary to reduce the glucose level of the seed medium for an efficient fermentation process. To achieve this, we examined the effect of the glucose level of the seed media on GABA fermentation efficiency. The results indicated that inoculum prepared using 10.0 g/L glucose resulted in an almost identical GABA titer to those prepared using higher levels of glucose (Figure 6), indicating that 10.0 g/L glucose was appropriate for seed medium preparation.

Effect of Seed Age on GABA Production
Next, we examined the effect of seed age on GABA production. GABA fermentation was initiated using seeds with A600 values ranging from 3.0 to 6.0 (4.8-9.7 × 10 8 CFU/mL). The data showed that seed age exhibited marginal effects on either GABA synthesis or strain growth (Figure 7), reflecting flexibility of inoculation. Figure 7. The effect of seed age on GABA production and cell growth. This was evaluated using a 250 mL Erlenmeyer fermentation flask at 30.0 °C. Fermentations were performed using inocula with different A600 values, which were prepared from the seed media containing 10 g/L glucose. Cell growth was expressed as A600. Vertical bars represent standard deviations of means (n = 3). There was no significant difference in 48 h GABA concentration between the different treatments (inocula different A600 value; p = 0.979). GABA, gamma-aminobutyric acid; A600, absorbance at 600 nm. Figure 6. The effect of seed medium glucose content on GABA production and cell growth. Inocula prepared using seed media with different levels of glucose were used for the fermentation experiment, which was performed in a 250 mL Erlenmeyer fermentation flask at 30.0 • C. Cell growth was expressed as A 600 . Vertical bars represent standard deviations of means (n = 3). There was no significant difference in 48 h GABA concentration between the different glucose levels examined (p = 0.267). GABA, gamma-aminobutyric acid; A 600 , absorbance at 600 nm.

Effect of Seed Age on GABA Production
Next, we examined the effect of seed age on GABA production. GABA fermentation was initiated using seeds with A 600 values ranging from 3.0 to 6.0 (4.8-9.7 × 10 8 CFU/mL). The data showed that seed age exhibited marginal effects on either GABA synthesis or strain growth (Figure 7), reflecting flexibility of inoculation.
Certain factors influence GABA production by LAB in a strain-specific manner [16,37]. The results of the present study showed that glucose, yeast extract, Tween-80, Mn 2+ , and temperature are key factors influencing GABA production by L. brevis CD0817, with optimal levels of 5.0 g/L, 35

Effect of Seed Age on GABA Production
Next, we examined the effect of seed age on GABA production. GABA fermentation was initiated using seeds with A600 values ranging from 3.0 to 6.0 (4.8-9.7 × 10 8 CFU/mL). The data showed that seed age exhibited marginal effects on either GABA synthesis or strain growth (Figure 7), reflecting flexibility of inoculation. Figure 7. The effect of seed age on GABA production and cell growth. This was evaluated using a 250 mL Erlenmeyer fermentation flask at 30.0 °C. Fermentations were performed using inocula with different A600 values, which were prepared from the seed media containing 10 g/L glucose. Cell growth was expressed as A600. Vertical bars represent standard deviations of means (n = 3). There was no significant difference in 48 h GABA concentration between the different treatments (inocula different A600 value; p = 0.979). GABA, gamma-aminobutyric acid; A600, absorbance at 600 nm.
Certain factors influence GABA production by LAB in a strain-specific manner [16,37]. The results of the present study showed that glucose, yeast extract, Tween-80, Mn 2+ , and temperature are key factors influencing GABA production by L. brevis CD0817, with optimal levels of 5.0 g/L, 35.0 g/L, 1.0 g/L, 16.0 mg/L, and 30 °C, respectively. Therefore, the seed medium was reconstituted to contain 10.0 g/L of glucose, 28.0 g/L of MSG, 35.0 g/L of yeast extract, 1.0 g/L of Tween-80, and 50.0 mg/L of MnSO4·H2O, and pH was Figure 7. The effect of seed age on GABA production and cell growth. This was evaluated using a 250 mL Erlenmeyer fermentation flask at 30.0 • C. Fermentations were performed using inocula with different A 600 values, which were prepared from the seed media containing 10 g/L glucose. Cell growth was expressed as A 600 . Vertical bars represent standard deviations of means (n = 3). There was no significant difference in 48 h GABA concentration between the different treatments (inocula different A 600 value; p = 0.979). GABA, gamma-aminobutyric acid; A 600 , absorbance at 600 nm.

PAS-Based GABA Bioprocess
Based on the results of the single-factor experiments, a PAS-based GABA fermentation bioprocess was prepared for L. brevis CD0817, and the efficiency of the bioprocess was evaluated. The findings of the single-factor experiments also showed that rapid GABA production occurred from 0 to 36 h (Figures 1-7). Therefore, the PAS-based bioprocess was operated within a 48 h culture period.
The PAS-based fermentation process was performed using 3.0 L of the optimized fermentation medium, 10.0% (v/v) of inoculum size, and 1950.0 g of L-glutamic acid solid powder at 30 • C and 50 rpm. As shown in Figure 8, rapid cell growth occurred immediately after inoculation and decreased considerably after 14 h. GABA production increased external osmolarity that triggered water loss of bacteria. A negative effect on cell growth occurred when dehydration reached a physiologically unsustainable level at high GABA content. The high osmotic pressure arising from high content of GABA was assumed to cause the decline of cell growth [48,56]. Glucose was depleted within 10 h, but cell growth continued till 14 h, which could be attributed to carbon supply by yeast extract. GABA production during the first 6 h of the fermentation process was slow, followed by rapid production till 38 h, after which the production plateaued. The final GABA titer was 321.9 ± 6.7 g/L, with a productivity of 6.71 g/L/h and a substrate conversion ratio of 99.6%. The substrate was depleted within 48 h. Compared with the unoptimized process [37], the optimized process increased GABA production by 28.0%.
In the unoptimized process, the starting seed medium and the starting fermentation medium, described in the Section 2, were used to make the seed medium and conduct fermentation at 32 • C, respectively. The other parameters of the unoptimized process were identical to those of the optimized process. By analyzing the parameters and the results between the two fermentations, we attribute the variation of GABA contents to the differences in the glucose level and culture temperature, which have been discussed in the Section 3.1. "Effect of glucose on GABA production" and Section 3.5. "Effect of temperature on GABA production", respectively. continued till 14 h, which could be attributed to carbon supply by yeast extract. GABA production during the first 6 h of the fermentation process was slow, followed by rapid production till 38 h, after which the production plateaued. The final GABA titer was 321.9 ± 6.7 g/L, with a productivity of 6.71 g/L/h and a substrate conversion ratio of 99.6%. The substrate was depleted within 48 h. Compared with the unoptimized process [37], the optimized process increased GABA production by 28.0%. Figure 8. pH auto-sustain-based GABA fermentation in a 10 L fermenter. The fermentation experiment was conducted at 30.0 °C and 50 rpm. Each liter of fermentation medium was supplemented with 650 g solid powder L-glutamic acid. Cell growth was expressed as A600. Vertical bars represent standard deviations of means (n = 3). The 48 h GABA concentration was 321.9 ± 6.7 g/L. GABA, gamma-aminobutyric acid; A600, absorbance at 600 nm.
In the unoptimized process, the starting seed medium and the starting fermentation medium, described in the Materials and Methods section, were used to make the seed medium and conduct fermentation at 32 °C, respectively. The other parameters of the un- Figure 8. pH auto-sustain-based GABA fermentation in a 10 L fermenter. The fermentation experiment was conducted at 30.0 • C and 50 rpm. Each liter of fermentation medium was supplemented with 650 g solid powder L-glutamic acid. Cell growth was expressed as A 600 . Vertical bars represent standard deviations of means (n = 3). The 48 h GABA concentration was 321.9 ± 6.7 g/L. GABA, gamma-aminobutyric acid; A 600 , absorbance at 600 nm.
The fresh fermentation medium saturated by L-glutamic acid had an extremely low pH (3.3). The pH was enhanced to 4.2 after inoculating the 150.0 mM MSG-based seed culture. The pH was maintained at 4.2 for the first 6 h of fermentation, followed by a gradual increase to approximate 6.0. The increase in pH followed a similar trend with GABA formation, which could be attributed to the consumption of H + during decarboxylation [16,44,57]. Clearly, it is almost impossible to always maintain the optimal pH (around 5.0) required for decarboxylation during the PAS-based fermentation. However, the broth remained acidic due to the organic acids metabolized from carbons and the PAS activity of the substrate. Even if the substrate was depleted, the organic acids would prevent the final pH from exceeding 7.0. As one control, we conducted a non-pH-controlled MSG-based GABA fermentation (initial pH 5.0), which was identical to the PAS-based method with the exception of substituting 0.75 M MSG for L-glutamic acid. The results indicated that the pH was gradually increased, up to~7.5 and~8.0 at 48 and 72 h, respectively; meanwhile the decarboxylation rate was progressively decreased (data not shown). These data implied that the GAD-induced cell can decarboxylate the substrate even in a weakly alkaline environment, although its decarboxylation ability is decreased with the increase of pH. It, thus, could be concluded that the suboptimal acid environment should guarantee a satisfactory GABA bioconversion during the PAS-based fermentation, which was confirmed by the high GABA titer of 321.9 g/L ( Figure 8). Overall, these findings indicated that L-glutamic acid was necessary for maintaining the low pH of the medium. However, whether L-glutamic acid is involved in other functions deserves further investigation.
Compared with this process, MSG is used as the substrate in the traditional GABA fermentation process. MSG has a neutral isoelectric point because of the substitution of sodium ion (Na + ) on the α-carboxyl group for a hydrogen (H + ). Park et al. [46] found that high levels of MSG can inhibit GABA production by introducing large amounts of Na + . Meanwhile, the alkalization of the environment attributed to the decarboxylation also exerted a negative effect on the GABA formation. In such a scenario, supplying an acid solution is necessary to maintain mild acidity [28]. However, the introduced acid radical can inhibit GABA synthesis and reduce the quality of the fermented product [42,43,58]. The PAS-based fermentation process developed in this study possesses some advantages over the MSG-based process. For instance, the substrate inhibitory effect on GABA production observed in MSG-based process was negligible in the PAS-based one. A brief comparison of the PAS-based GABA fermentation with the MSG-based ones is shown in Table 1. MSG: monosodium L-glutamate; NS: not specified; pH-controlled refers to pH-controlled fermentation in which pH was controlled by the feeding of an acid agent.

Conclusions
In the present study, a PAS-based GABA fermentation process was developed for efficient GABA production by L. brevis CD0817, with L-glutamic acid as the substrate. After 48 h of fermentation, GABA titer reached 321.9 g/L with a productivity of 6.71 g/L/h and a substrate conversion rate of 99.6%. Overall, these findings demonstrate the need for LAB strain-specific fermentation methods to maximize GABA production.
Supplementary Materials: The following supporting information can be downloaded at https: //www.mdpi.com/article/10.3390/fermentation8050208/s1: Figure S1: The effect of nitrogen sources on GABA production.