E. coli DH5α was used for plasmid propagation and construction. B. subtilis 168 was used for production of γ-PGA.

The media included Luria-Bertani (LB) medium (5 g/L yeast extract, 10 g/L tryptone and 10 g/L NaCl) and fermentation medium (30 g/L glucose, 2.5 g/L yeast extract, 20 g/L L-glutamate, 1 g/L K₂HPO₄, 1 g/L MgSO₄, pH 7.0).

All E. coli strains were cultured in LB medium at 37 °C. For γ-PGA production, an individual colony of B. subtilis strain was inoculated into 5 mL of LB medium at 37 °C overnight for seed culture preparations. Then, 1 mL of seed culture was inoculated into 50 mL of fermentation medium, which was incubated at 37 °C with shaking at 160 rpm for 48 h. For engineering Bacillus strains harboring the promoter P_xylA, 4 g/L xylose was added to the cultures at 8 h, followed by continued incubation for an additional 40 h. Biomass was monitored by measuring the optical density at 660 nm (OD₆₆₀). The shake flask experiments were performed in triplicate.

For colony morphology analysis, fermentation medium agar plates were supplemented with 4 g/L xylose and 2% agar. B. subtilis strains were streaked on the plates and incubated at 37 °C for 24 h.

In the carbon source optimization experiments, glucose was individually replaced with 30 g/L of sucrose, fructose, lactose, mannitol, or glycerol, respectively, while all other components of the fermentation medium remained unchanged. Xylose (4 g/L) was added to the cultures at 8 h, followed by continued incubation for an additional 40 h. In xylose concentration optimization experiments, xylose was added to the cultures at concentrations of 1, 2, 4, 8, and 16 g/L at 8 h, followed by continued incubation for an additional 40 h. All experiments were conducted in triplicate.

In the IPTG concentration optimization experiments, IPTG was added at concentrations of 0.1, 0.2, 0.4, 0.6, 0.8, and 1.0 mM to the cultures at 8 h, followed by an additional 8 h of incubation. To determine the optimal IPTG addition time point, 0.6 mM IPTG was added to the cultures at 0, 4, 8, 16, 20, and 24 h, respectively, followed by an additional 8 h of incubation. The activity of PgdS in the culture supernatant was measured to evaluate the optimal IPTG concentration and addition time point. All experiments were conducted in triplicate.

For scale-up, the cell grew in 5-L fermenter with 3.5 L of media (30 g/L mannitol, 2.5 g/L yeast extract, 20 g/L L-glutamate, 1 g/L K₂HPO₄, 1 g/L MgSO₄, 0.04% (v/v) antifoam (THIX-298), pH 7.0) for 72 h. The inoculation amount and temperature were 2% (v/v) and 37 °C, respectively. The aeration rate was 1.0 vvm, and the agitation speed was 400 rpm. The biomass, γ-PGA titer, concentrations of mannitol and L-glutamate, Mw of γ-PGA were analyzed.

To further improve the γ-PGA production in 5-L fermenter, the effects of inoculation amount (1% (v/v), 2, 5 and 8%), temperature (32 °C, 37 °C, and 42 °C), aeration rate (0.5 vvm, 1.0 vvm, and 1.5 vvm), agitation speed (200 rpm, 300 rpm, 400 rpm, and 500 rpm) and pH (6.0, 6.5, 7.0, and pH-uncontrolled) were investigated.

The primers used for plasmid construction are shown in Supplementary Table S1. The sequences of promoter P_aprE, P₄₃, and P_xylA were amplified from the genomic DNA of B. subtilis 168. Two methods were used to construct the plasmid. One was to digest DNA fragments and plasmids with restriction enzymes, respectively, followed by ligation using T4 DNA ligase (Takara, Dalian). Another method was to insert DNA fragments into plasmids using an ClonExpress Ultra One Step Cloning Kit (Vazyme, China).

For expression of the pgsBCAE gene cluster, the upstream (1.5 kb) and downstream (1.5 kb) regions of pgsB were PCR-amplified from the genomic DNA of B. subtilis 168. These fragments were individually assembled with three different promoter sequences via overlapping PCR. The resulting constructs were then cloned into the plasmid pKSV7, yielding plasmids pKSV-P₄₃, pKSV-P_aprE, and pKSV-P_xylA, respectively.

For genes cwlO, pgdS and ggt deletion, the upstream and downstream fragments regions of each target gene were amplified by PCR and fused by overlapping PCR. The fused fragments were subsequently inserted into plasmid pKSV7, generating plasmids pKSV-cwlO, pKSV-pgdS, and pKSV-ggt, respectively.

For pgdS expression, the coding sequence was amplified by PCR from B. subtilis 168 genomic DNA and inserted into plasmid pHT43, producing plasmid pHT-SP_PgdS-pgdS. The sequences of signal peptide SP_amyQ and SP_aprE were PCR-amplified from pHT43 and pBE-S, respectively, and replaced the SP_PgdS sequence in plasmid pHT-SP_PgdS-pgdS using one-step cloning kit, resulting in plasmids pHT-SP_amyQ-pgdS and pHT-SP_aprE-pgdS.

All recombinant plasmids were verified by DNA sequencing (Sangon Co., Ltd., China). Then, the verified plasmids were transformed into B. subtilis 168 as previously described (Young and Spizizen, 1963). Mutant strains were randomly selected from LB agar plates and verified by PCR. All plasmids and strains used in this work are listed in Supplementary Table S2.

The activity of PgdS was measured as previously described (Sha et al., 2018). Briefly, crude enzyme solution was collected from culture by centrifugation. The 0.5 mL of reaction mixture contained 50 mM phosphate buffer (pH 7.0), 1 g/L γ-PGA and 100 μL crude enzyme solution, and was incubated at 37 °C for 2 h. The reaction was terminated by boiling for 5 min. Then 1 mL of 2 M acetate buffer and 1 mL ninhydrin chromogenic solution were added into the reaction, and was incubated at 100 °C for 15 min. The UV absorption of reaction was measured at 570 nm. The amount of free amino groups was calculated. One unit of PgdS was defined as the amount of enzyme that generated 1 μM of free amino groups per hour.

The culture was centrifuged at 12,000 g for 30 min to remove the bacteria, the supernatant was mixed with 4 times the volume of 95% ethanol, the precipitate was collected by centrifugation at 12,000 g for 5 min and redissolved with deionized water. Then, the solution was centrifuged at 12,000 g for 5 min to remove insoluble impurities, the supernatant was transferred into dialysis bag (aperture: 10 kDa) overnight. Finally, γ-PGA was obtained through lyophilization.

For Mw of γ-PGA analysis, the purified γ-PGA was prepared in 0.3 M Na₂SO₄ as a 1.0 mg/mL aqueous solution. The Mw of γ-PGA was analyzed using gel permeation chromatography (GPC), equipped with a TSKgel G6000 PWXL column (TosohBioscience, Japan). The mobile phase consisted of 0.3 M Na₂SO₄ solution in water, with a flow rate of 0.5 mL/min at 30 °C, and detection was performed with a refractive index detector.

For stereochemical γ-PGA analysis, the purified γ-PGA was dissolved in 6 M HCl and incubated at 105 °C for 12 h. Then, the solution pH was adjusted to 7.0 by 6 M NaOH. Samples was analyzed by HPLC (Shimadzu, Japan), equipped with Chialpak MA(+) (Daicel Corp, Japan). The mobile phase consisted of 2 mM CuSO₄ solution in water, with a flow rate of 0.5 mL/min at 25 °C, and detection at 254 nm.

For NMR analysis, the 10 mg of purified γ-PGA was dissolved in 500 μL of D₂O. Spectra were recorded using NMR spectrometer (Avance III HD600, Bruker).

The γ-PGA titer of the fermentation broth was analyzed as previously described (Zeng et al., 2012). Briefly, the cell was removed by centrifugation (12,000 g), and the supernatants were collected. Then, four volumes of ethanol were added in supernatants. The mixture was centrifuged at 12,000 g to remove supernatants and collect γ-PGA-containing precipitate. Next, the precipitate was dried and redissolved in deionized water. Finally, the UV absorption of γ-PGA-containing aqueous solution was detected at 216 nm. The γ-PGA titer was calculated by the standard curve.

Statistical significance was performed using analysis of ANOVA with a significance level of 0.05 by SPSSAU (www.spssau.com).

Although the genome of B. subtilis 168 contains the γ-PGA biosynthesis gene cluster pgsBCAE, it is usually not transcribed (Urushibata et al., 2002). Therefore, the constitutive promoters P_aprE and P₄₃ and inducible promoter P_xylA were integrated chromosome of the B. subtilis 168 to produce strains BS1, BS2, and BS3 (Figure 1A), respectively. Next, the engineered strains were used for γ-PGA production. The results showed that titers of γ-PGA in BS1, BS2, and BS3 were 3.06 g/L, 2.78 g/L, and 1.51 g/L (Figure 1B), respectively, suggesting that promoter P_aprE may have stronger transcription ability than P₄₃ and P_xylA.

However, the promoter P_xylA can be inhibited by glucose in fermentation (Kim et al., 1996). Therefore, we selected a variety of carbon sources, including sucrose, fructose, lactose, mannitol and glycerol, to replace glucose in fermentation for elimination of glucose inhibition. All the other components of the media remained the same. As shown in Figure 1C, when medium contained sucrose, the biomass and γ-PGA titer were 7.82 and 2.1 g/L, respectively. Compared with glucose, sucrose did not improve the γ-PGA production. The reason was that sucrose was hydrolyzed to produce glucose and fructose in fermentation, producing glucose inhibited the P_xylA. When fructose and mannitol were carbon sources, the corresponding γ-PGA titers were 4.8 g/L and 4.82 g/L, respectively, which both were higher than those of strains BS1 and BS2. Notably, the OD₆₆₀ was highest at 9.31 in 48 h by using mannitol, suggesting mannitol was more favorable to the cell growth. In addition, both biomass and γ-PGA titer decreased by using glycerol, indicating that glycerol was not suitable to γ-PGA production for strain BS3. Considering that there was no significant difference between fructose and mannitol as carbon sources in γ-PGA production, fructose and mannitol were selected as carbon sources for further research.

Since the P_xylA promoter was induced by xylose, the effect of xylose concentration in fermentation was studied when carbon sources were fructose and mannitol, respectively. As shown in Figure 1D, with fructose as carbon source and concentration of xylose was 4 g/L, the γ-PGA titer was maximum at 4.8 g/L. However, with mannitol as carbon source and concentration of xylose was 2 g/L, the γ-PGA titer was 6.08 g/L (Figure 1E). Compared with fructose, mannitol increased γ-PGA production by 26.67% while requiring less xylose. Therefore, mannitol had good performance on γ-PGA production in strain BS3.

Furthermore, the colony morphology of B. subtilis 168 and BS3 were observed. The results showed that B. subtilis 168 formed big and flat colonies (Supplementary Figure S1A). In contrast, the colonies of BS3 appeared globoid and viscous, indicating that it has the ability to synthesize γ-PGA. Next, the purified γ-PGA was analyzed by NMR (Supplementary Figure S1B). These results determined that purified product in this study was γ-PGA. In addition, stereochemical analysis of γ-PGA showed that D/L-glutamic acid ratio was 59: 41 (Supplementary Figure S1C).

The B. subtilis contains hydrolases CwlO, PgdS and GGT. These enzymes can degrade γ-PGA to affect the Mw and concentration of γ-PGA (Mitsui et al., 2011; Scoffone et al., 2013). In order to increase the γ-PGA Mw and titer, the effects of deletion of genes cwlO, pgdS and ggt on γ-PGA production were investigated. Recombinant strains BS4 (BS3ΔcwlO), BS5 (BS3ΔpgdS), and BS6 (BS3Δggt) were constructed and used to produce γ-PGA under optimal culture conditions (fermentation medium containing 30 g/L mannitol, 2.5 g/L yeast extract, 20 g/L L-glutamate, 1 g/L K₂HPO₄, 1 g/L MgSO₄, pH 7.0; 2 g/L xylose added to the cultures at 8 h). The results showed that biomass of BS4 was higher than BS3 and other mutant strains (Figure 2A), while BS5 and BS6 showed no significant difference from BS3. The γ-PGA titer in BS4 was highest at 6.62 g/L at 56 h, and its Mw was 9.53 × 10⁶ Da. In addition, γ-PGA Mw of BS5 were highest at 2.27 × 10⁷ Da at 48 h, which increased by 200.26% compared to BS3. These results indicated that deletion of cwlO improved the γ-PGA titer, while PgdS had a significant impact on the Mw of γ-PGA. In contrast, deletion of ggt did not substantially enhance either the titer or the Mw.

Next, the double-gene deletion of γ-PGA hydrolases was conducted. The strains BS7 (BS3ΔcwlOΔpgdS), BS8 (BS3ΔcwlOΔggt) and BS9 (BS3ΔggtΔpgdS) were constructed and used for γ-PGA production. The results showed that there was no significant difference in the biomass between strains BS7, BS8 and BS3 (Figure 2B). The biomass of strain BS9 was slightly lower than that of BS3 and other double-gene deletion strains after 48 h. The maximum titers of γ-PGA in BS7, BS8 and BS9 were 6.83 g/L, 7.01 g/L and 7.27 g/L at 56 h, respectively, which were 19.41, 22.56 and 27.1% higher than that of BS3. The γ-PGA Mw in BS7 and BS9 obviously higher than BS8 and BS3. BS9 achieved the highest Mw at 2.31 × 10⁷ Da at 56 h. Although it was higher than BS7 (2.11 × 10⁷ Da) and BS8 (9.17 × 10⁶ Da), it was only 1.8% higher than single-gene deletion strain BS5 (2.27 × 10⁷ Da). These results suggested that all double-gene deletion strains increased the titer and Mw of the generated γ-PGA compared to BS3.

Furthermore, the three genes cwlO, pgdS and ggt were knocked out to produce strain BS10 (BS3ΔcwlOΔggtΔpgdS). Production of γ-PGA by strain BS10 was assayed. As shown in Figure 2C, the biomass of BS10 was slightly lower than BS3. But γ-PGA titer and Mw were 8.44 g/L and 2.42 × 10⁷ Da at 56 h, respectively, which increased by 47.81 and 220.1%. These results determined that deletion of cwlO, pgdS, and ggt can further increase the γ-PGA titer and Mw.

In our present study, it has been determined that PgdS had a significant effect on the γ-PGA Mw. So we attempted to regulate expression level of PgdS for control of the Mw of generated γ-PGA. The inducible promoter P_grac by IPTG was selected to control PgdS expression level. The plasmid pHT-SP_PgdS-pgdS was constructed and transformed into BS10 to produce strain BS11. In addition, PgdS contains a signal peptide SP_PgdS (amino acids codes: 1–32) that transport PgdS into the culture to hydrolyze γ-PGA. Considering that the efficiency of PgdS secretion affects its activity in the culture, we sought to enhance PgdS secretion to produce γ-PGA with lower Mw, thereby expanding the range of γ-PGA Mw. Therefore, the signal peptides SP_amyQ and SP_aprE replaced SP_PgdS to construct strains BS12 and BS13, respectively. The specific activity of PgdS in supernatant of culture was measured. The results showed that the specific activities of PgdS in strains BS11, BS12, and BS13 were 5.16 U/mL, 4.91 U/mL, and 3.23 U/mL (Figure 3A), respectively, suggesting that SP_PgdS was more favorable for PgdS secretion. Thus strain BS11 was selected for further study.

Next, the effect of IPTG concentrations on PgdS activity were studied. As shown in Figure 3B, when concentrations of IPTG were 0–0.6 mM, the activity of PgdS increased with increasing IPTG concentrations. However, when IPTG concentrations exceeded 0.6 mM, the activity of PgdS decreased. Therefore, 0.6 mM IPTG was used as an optimal concentration. In optimal IPTG concentration, effect of IPTG addition time on PgdS activity was assayed. The results showed that PgdS activity in supernatant was highest at 20.51 U/mL when IPTG was added at 16 h (Figure 3C).

Finally, the effect of IPTG addition time on γ-PGA production was studied. The results showed that the Mw of generated γ-PGA was related to IPTG addition time (Figure 3D). When IPTG was not added, the γ-PGA titer and Mw in BS11 were 7.49 g/L and 2.15 × 10⁷ Da, respectively. Compared with control, the γ-PGA Mw decreased to 9.55 × 10⁴ Da, 9.95 × 10⁴ Da, and 6.5 × 10⁵ Da, respectively, when IPTG was added at 16 h, 24 h, and 36 h. However, the γ-PGA titer increased with IPTG addition time increase (16–24 h). The γ-PGA titer was lowest at 6.28 g/L when IPTG was added at 16 h. When IPTG was added after 32 h, γ-PGA titer was stable. These results suggested that change of IPTG addition time can produce γ-PGA with specific Mw in an engineered strain. In summary, the γ-PGA with Mw ranging from 9.55 × 10⁴ Da to 2.15 × 10⁷ Da was produced by an engineered strain, with the titer of 6.28–8.57 g/L.

To explore the production potential of strain BS11, the 5-L fermenter was used for large-scale experiments. The results showed that the γ-PGA Mw and titer were 1.85 × 10⁷ Da and 4.86 g/L at 56 h (Supplementary Figure S2), respectively, which were lower than shake flask. Considering the many factors have an impact on fermentation, such as inoculation amount, temperature, aeration rate, stirring speed and pH. Therefore, the fermentation conditions of γ-PGA should be optimized.

First, the inoculation amounts were set 1% (v/v), 2, 5, and 8% in fermentation. As shown in Figure 4A, when inoculation amounts were 1 and 2%, respectively, their biomass was higher than the 5 and 8% of inoculation amounts. The γ-PGA titer was highest at 4.96 g/L in fermentation with 2% of inoculation amount. The γ-PGA Mw in 2% of inoculation amount was highest at 1.93 × 10⁷ Da. These results suggested that when inoculation amount was 2%, the efficiency of γ-PGA synthesis was highest.

Then, in optimal inoculation amount, the effect of temperature on γ-PGA production was studied. The results showed that when temperature was 32 °C, the biomass was higher than at 37 °C and 42 °C (Figure 4B). However, the γ-PGA titer was lowest at 32 °C. The γ-PGA titer kept stable after 40 h at 37 °C, its maximum titer was 5.6 g/L. In addition, the γ-PGA Mw at 32 °C was higher than at 37 °C and 42 °C during 0–40 h. But the γ-PGA Mw at 37 °C increased with the increase of fermentation time and showed no significant difference from that at 32 °C after 48 h. Considering γ-PGA Mw and titer, the 37 °C temperature was selected for optimal temperature.

In the optimal inoculation amount and temperature, the effect of aeration rate on γ-PGA production was explored. As shown in Figure 4C, the cell growth was faster at aeration rate of 1.0 vvm and 1.5 vvm, corresponding to higher titers than 0.5 vvm. The γ-PGA titer was highest at 6.58 g/L at 1.5 vvm. The γ-PGA Mw at 1.5 vvm significantly higher than that of 0.5 vvm and 1.0 vvm, with a highest Mw of 2.02 × 10⁷ Da. So the 1.5 vvm was used for optimal aeration rate.

In optimal inoculation amount, temperature and aeration rate, the agitation speed in fermentation was optimized. The results showed that the cell growth was better at 300 rpm and γ-PGA titer was maximum at 7.63 g/L compared to other agitation speeds (Figure 4D). The γ-PGA Mw was highest at 2.06 × 10⁷ Da at 300 rpm. These results indicated that 300 rpm was more conducive to γ-PGA production.

Based on above optimal conditions, the effect of pH on γ-PGA production was investigated. As shown in Figure 4E, although biomass at pH 7.0 was higher, the γ-PGA titer was maximum at 7.4 g/L when pH was not controlled. The γ-PGA Mw was highest at 1.87 × 10⁷ Da. Therefore, the production efficiency of γ-PGA was better in the pH-uncontrolled group, which also reduced the cost.

Finally, the optimal culture conditions were further applied to produce γ-PGA in 5-L fermenter. Meanwhile, IPTG was added at 0.6 mM in culture at 24 h. The results showed that the biomass at 40 h was the highest (Figure 5), and then decreased. The maximum titer was 7.76 g/L at 56 h. The γ-PGA Mw increased from 8 to 24 h, with a highest Mw was 1.84 × 10⁷ Da. Then γ-PGA Mw decreased after IPTG addition. The lowest Mw was 8.3 × 10⁴ Da at 72 h. These results suggested that IPTG-induced expression of PgdS to change γ-PGA Mw was feasible in large-scale experiments. Although IPTG was only added at 24 h, we are confident that γ-PGA with specific Mw can be produced by changing the IPTG addition time in large-scale experiments in future study.