Bacillus velezensis ND (CCTCC M 2020983, China Center for Type Culture Collection, Wuhan, China), which was isolated from a soil sample (Zhao et al., 2021), was used in this study. Strain was routinely cultured overnight in Luria-Bertani (LB) agar (15 g/L agar) plates at 37°C. The inocula were scraped and transferred into 100 mL of seed medium (pH 7.0) containing 10 g/L tryptone, 5 g/L yeast extract, and 10 g/L NaCl and cultivated at 28°C, 200 rpm for 18–20 h. Then the resulted seeds (10%, v/v) were inoculated into 50 mL of fermentation medium containing 120 g/L Soybean meal powder, 16 g/L yeast extract, 1 g/L monosodium glutamate, 88.3 g/L glycerol, 0.5 g/L MgSO₄, 7H₂O, 1 g/L KH₂PO₄, 0.15 mg/L FeSO₄, 7H₂O, 5 mg/L MnSO₄·H₂O, and 0.16 mg/L CuSO₄.5H₂O, and adjust the pH to 8.0 using 1 mol/L NaOH solution. The broth was incubated at 28°C in a rotary shaker with 200 rpm speed for 5 days for shake flasks fermentation.

Six different porous carriers were tested, namely P (polyester fiber material), C (porous ceramic material), Q (Ceramic material), k1, k2, and k3 (polyethylene material). The particle size of carriers P, C, Q, k1, k2, and k3 was 4.0, 3.0, 2.6, 1.0, 1.5, 2.0 cm, respectively. The specific surface area was 3,000, 1800, 1,400, 600, 660, 750 m²/m³, respectively. Carrier (approximately 60% culture medium volume) was added into shaking flask fermentation to evaluate its effects on the cell weight and iturin A yield. These 250 mL flasks containing 50 mL of the fermentation medium were incubated at 28 °C with shaking at 200 rpm for 5 days.

A schematic diagram of the biofilm reactor is given in Figure 1. The system consists of two parts: (i) a 3 L stirred-tank fermentor (Baoxing Co., China) and (ii) a glass column reactor (with a diameter of 60mm and a height of 340mm) packed with carriers for biofilm formation. The glass column reactor was connected to the stirred-tank fermentor through a recirculation line. The entire reactor system contained 1 L of the medium and the liquid medium was recirculated continuously between these two parts with a flow rate of 18 mL/min via peristaltic pump. Unless otherwise noted, the stirred-tank fermentor was controlled at 28°C with stirring at 500 rpm and aeration at 3 vvm (air volume/culture volume/min). During the fermentation process, the stirred-tank reactor offers aerobic growth conditions for suspended cells and ensures a mixing of the medium. On the other hand, the broth trickles from the top of the glass column and flow down over the surface of carriers where the biofilm is attached. Then, the broth collected at the bottom of the reactor flows into the stirred-tank reactor. To control foam production, 20 mL polyether defoamer (Xinjiang Fufeng Biotechnologies Co., Ltd. China.) was added to the culture medium before fermentation. When the defoaming electrode detects foam generation in the stirred-tank fermentor during fermentation, defoamer will be automatically fed into fermentor.

In batch fermentation, the reactor containing 1 L of medium was inoculated with seed culture of Bacillus velezensis ND (10%, v/v), and then the fermentation broth was circulated to the connected column reactor (4/5 volume of the reactor is filled with fiber ball carrier) by peristaltic pump for 8 days. Liquid samples were taken from the stirred-tank fermentor at regular intervals for the determination of suspended cell density, residual substrate and iturin A concentration. After each sampling procedure, sterile water was added to the fermentor to restore the fermentation volume to 1 L (High stirring rates and aeration can accelerate water evaporation in the fermentation broth, which will reduce the volume of the broth and cause viscosity). Fed-batch fermentation were carried out by adding steriled glycerol (carbon source) to the fermentor. The glycerol (18 g) was supplemented (one-time addition) at 48 h and 96 h respectively, at which point glycerol was less than 28 g/L.

For comparison, batch and fed-batch fermentations in the suspended cell bioreactor were carried out. Specifically, a 3 L stirred-tank fermentor containing 1 L of medium was inoculated with seed culture of Bacillus velezensis ND (10%, v/v) and fermented at 28°C. The aeration rate was controlled at 3 vvm and the stirring speed was 500 rpm. Fed-batch fermentation were carried out by adding steriled glycerol (carbon source) to the fermentor. The glycerol (18 g) was supplemented (one-time addition) at 96 h and 144 h respectively, at which point glycerol was less than 28 g/L. Biomass, yield and productivity of iturin A, carbon source consumption was compared between suspended cell and biofilm reactor.

The repeated-batch fermentation in biofilm reactor was operated by replacing the fermented broth with 1 L of fresh medium and the culture immobilized on the carrier was used as the inocula for the next batch. Other fermentation conditions were the same as described in batch fermentation in biofilm reactor. Three repeated batch fermentation were carried out to assess the stability of iturin A production by the biofilm reactor.

The operating procedures for this method were as follows: fermentation temperature was set at 37°C from 0–24 h, 34°C from 24–36 h, 31°C from 36–48 h, and 28°C after 48 h. This strategy was used during batch fermentations in biofilm reactor. The fermentation conditions except temperature were the same as described in batch fermentation in biofilm reactor.

For comparison, batch fermentations in the suspended cell bioreactor with step-wise temperature control strategy were carried out.

For the determination of cell concentration in broth, 1 mL of fermentation broth mixed with 9 mL of sterile distilled water, and shaken by a vortex. Then, the mixture was serially diluted and spread onto LB-agar plates. After 12 h of incubation at 37°C, cell concentration was determined as colony-forming units per milliliter (CFU/mL).

For the determination of cell dry weight in broth, 10 mL sample was centrifuged at 5,000 rpm for 10 min. The cell were washed twice with 50 mL distilled water and dried to constant weight at 80°C.

For the determination of cell dry weight in biofilm, the carrier was taken out from the reactor to the gauze at the end of fermentation. The biofilm was stripped from the carrier with tweezers, then the biofilm with medium was washed with PBS buffer, and was dried to constant weight at 80°C.

The concentration of glycerol was determined with a Glycerol Assay Kit (APPLYGEN Co., China). The fermentation broth was centrifuged at 5,000 rpm for 10 min, and resulting supernatant was diluted 1,000 times. 5 μl of diluent was mixed with 195 ul detection buffer of the kit. After the mixture had been incubated at 37°C for 15 min, the glycerol concentration was determined by microplate reader at 550 nm.

For the determination of the concentration of iturin A, 0.3 mL of fermentation broth was added to 0.9 mL of methanol and the mixture was vigorously shaken using a vortex mixer for 10 min. The extract was centrifuged at 12,000 rpm for 15 min, and the supernatant was used for iturin A determination. The iturin A concentration was measured on an Agilent RP-HPLC (Agilent Technologies, United States), which equipped with Agilent Lichrospher C18 column (Agilent ZORBAX Eclipse XDB-C18 column, United States). The mobile phase was10 mM ammonium acetate/acetonitrile = (65:35, v/v), and flow rate was 1.0 mL/min. The injection volume was 20 μl and detection wavelength was 210 nm, and concentration of iturin A was calculated by the standard curve made by standard iturin A (purity 95%, Sigma-Aldrich United States).

For the determination of iturin A adsorbed on carrier, the carrier was taken out from the reactor at the end of fermentation and biofilm was rinsed thoroughly with 1,000 mL distilled water. Then, the mixture of biofilm and water was vigorously vortexed for 30 min to facilitate the disintegration of bacterial clumps and was used to extracted iturin A. The iturin A extraction and assay were performed as described above.

All assays were performed at least in triplicate, and data represent the mean of three or more experiments. All observations were repeated and the average values were obtained and demonstrated with standard errors of the repetitions as error bars. Using Origin 2018 ANOVA, any difference with p < 0.05 was considered significant.

Six different types of carrier were tested for their effects on the production of iturin A and cell growth so as to find a suitable carrier. As shown in Figure 2, P, C, k1, k2, and k3 demonstrated a significant promotion effect on cell growth and iturin A production (p < 0.05). This may be attributed to two factors: i) the carrier vibrated in a shake flask, promoting the oxygen and nutrient transfers in broth. ii) Biofilm formed gradually at the surface of these carriers, resulting in a higher amount of biomass. Unlike these carriers, Q appeared to have an inhibitory effect upon cell growth and biofilm was not observed at the surface. The reason of this unexpected observation was still unclear. It was probably due to unknown component of carrier Q which was made of natural raw materials. On the other hand, carrier P yielded the highest iturin A concentration (4.7 ± 0.1 g/L) and biomass (6.5 ± 0.1 g/L). These values were 10.6% and 22.3% greater than the control group (4.3 ± 0.1 and 5.3 ± 0.1 g/L, respectively). The better performance of P could be attributed to its larger specific area resulting from structural differences among the carriers. For these reasons, carrier P was selected as the most suitable packing for cell immobilization in biofilm reactor system and was used in the following experiments.

Batch fermentation was first performed to investigate the effectiveness of iturin A production by using the biofilm reactor. For comparison, batch and fed-batch fermentations in the suspended cell bioreactor were also carried out. Figure 3 shows the time courses of the biomass, residual glycerol and the concentration of iturin A in biofilm reactor and suspended cell reactor. As showing in Figure 3A, cell started growing rapidly both in suspended cell reactor and biofilm reactor during culture (0–72 h), and the maximum growth rate is almost the same. At 96 h of culture, suspended cell reactor achieved a highest biomass (184.0 ± 7.9 × 10⁸ CFU/mL) which is higher than biofilm reactor (137.3 ± 2.9 × 10⁸ CFU/mL). However, it should be noted that the biomass only represented growth in the broth and not in the biofilm. Actually, biofilm formed gradually in biofilm reactor and the total amount of biomass in broth and biofilm is higher than that in suspended cell reactor (Table 1). In both reactors, there was a rapid decrease in cell density after the culture reached its maximum concentration value, however, the iturin A production continued. This result indicated that iturin A is a non-growth-associated product. In biofilm reactor, faster glycerol consumption was observed and iturin A concentration reached maximum levels at the same time as the depletion of glycerol (Figure 3B). Figure 3C shows that iturin A concentration reached its maximum at 5.4 ± 0.1 g/L after 144 h in biofilm reactor and 4.6 ± 0.1 g/L after 192 h in suspended cell reactor. This implies that biofilm fermentation increases the yield by 17.3% and the productivity by 56.4% compared with suspended cell fermentation. In addition, 355.1 ± 31.7 mg iturin A was obtained from the biofilm on the carrier so that the total amount of iturin A in biofilm reactor was 5.7 ± 0.1 g/L which increased by 25.2% compared with the suspended cell fermentation.

The evidence of substrate limitation occurred in batch fermentation and substrate inhibition present at high glycerol level (data not shown) indicated the appropriateness of fed-batch fermentation. Figure 4 shows the time course of fed-batch fermentation in biofilm reactor and suspended cell reactor. The glycerol was supplemented in biofilm reactor at 48 h and 96 h respectively, at which point glycerol was less than 28 g/L. For suspended cell reactor, the glycerol comsuption is slower and glycerol was supplemented at 96 h and 144 h respectively. As showing in Figure 4A, the feeding of glycerol prolonged the cell growth phase and achieved highest biomass at 144 h in both reactor. Accordingly, a higher cell yield than batch process in both biofilm and suspended cell reactor (166.7 ± 4.7×10⁸ and 237.3 ± 5.8 × 10⁸ CFU/mL, respectively) was achieved. As shown in Figure 4C, iturin A reached it maximum at 6.8 ± 0.1 g/L after 144 h in biofilm reactor while 5.5 ± 0.1 g/L after 192 h in suspended cell reactor. Compared with batch process, the yield of iturin A is further improved.

Three cycles of repeated-batch fermentation were carried out in order to validate the stability of biofilm and long-term performance of the biofilm reactor. As shown in Figure 5, the profile of each fermentation cycle is nearly the same. Specific data of three cycles of repeated-batch fermentation are shown in Table 2. The yield of iturin A in three batches was 5.4 g/L, 5.3 g/L and 5.3 g/L, respectively, and the cells dry weight at the end of each batch increased gradually from 6.3 g/L to 6.9 g/L. Also, the trend of cell growth, substrate consumption and iturin A production in repeated-batch fermentation were nearly the same as single batch fermentation in biofilm reactor. These results indicated that the biofilm on the carrier could maintain their long-term stability, providing the seeds for the next batch. It should be noted that the repeated-batch process omitted inoculation requirements between fermentation cycles and saved costs for seed culture, but it did not showed a shorter period of fermentation and a higher productivity than single batch fermentation. However, long-term stability of biofilm in carrier implied a possibility of continuous fermentation process and the yield of iturin A could be further improved through continuous fermentation.

According to our previous study of this strain, the optimal temperature for good cell growth is 37 °C while the optimal temperature for iturin A production is 28°C (data not shown). To address the inconsistent temperature in cell growth and iturin A production, a step-wise temperature control strategy was introduced in this study. During the first 24 h of cultivation, the temperature was set at 37°C for good cell growth. Subsequently, the temperature was decreased step-wise to 28°C over the next 24 h and maintained until the end of fermentation, favoring the production of iturin A. The result of batch fermentation was shown in Figure 6. The cell density in broth increased rapidly during the first 24 h and slow down as temperature decreased (24–48 h), reaching its highest value at 48 h (249.3 ± 8.5 × 10⁸ CFU/mL) (Figure 6A). This value is 81.8% higher than that of batch process without step-wise temperature control strategy. Meanwhile, glycerol was also consumed at a rapid rate and nearly depleted at 72 h (Figure 6B). The highest level of iturin A was achieved (6.6 ± 0.1 g/L) at 120 h (Figure 6C). These results clearly indicated that the fermentation period was shortened and the productivity (55.1 ± 0.3 mg/L/h) was greatly improved by the step-wise temperature control strategy.

Interestingly, the step-wise temperature control strategy did not work in the suspended cell fermentation. During the culture (0–48 h), suspended cell reactor showed a faster cell growth than process without step-wise temperature control but the maximum biomass showed only a small increase (198.0 ± 8.1 × 10⁸ CFU/mL vs. 184.0 ± 7.9 × 10⁸ CFU/mL). At the same time, the consumption rate of glycerol was also increased, accompanied by the acceleration of cell growth. Glycerol was also depleted at 72 h and the cells entered decline phase prematurely in the suspended cell reactor. Although the same glycerol consumption rate was observed, the highest iturin A concentration in the suspended cell fermentation was only 4.0 ± 0.1 g/L after 196 h which is lower than process without step-wise temperature control.