Corynebacterium glutamicum Cg1-P30 (CMCC NO. 22680), preserved at China General Microbiological Culture Collection Center, was initially grown in brain heart infusion (BHI) agar medium for activation. The activated strain was maintained on BHI broth at 30°C in a rotary shaker at 200 rpm overnight. Subsequently, 2% of the seed culture was transferred into new flasks containing 50 mL of production medium consisting of 10 g/L glucose, 5 g/L corn steep liquor, 5 g/L urea, 0.25 g/L MgSO₄·7 H₂O, 1.0 g/L K₂HPO₄ and 1.0 g/L KH₂PO₄ and incubated at 30°C and 200 rpm for 48 h. The concentrations of glucose, corn steep liquor, and urea were further optimized as described in the section of optimization of medium for bioflocculant production.

Based on the result of one-factor-at-a-time experiment, response surface methodology through Box-Behnken design (Design Expert software, Version 8.0.6) was applied to evaluate the optimum level and interactive effects of carbon and nitrogen sources on bioflocculant production. The concentration of glucose, corn steep liquor, and urea were selected as independent variables at three different levels, while the response variable was the flocculating activity of the fermentation supernatant after culturing for 48 h. The result of Box-Behnken design was shown in Table 1. A quadratic polynomial equation was fitted to evaluate the correlate relationship between the independent variables and the response as following: Y = β 0 + β 1 X 1 + β 2 X 2 + β 3 X 3 + β 12 X 1 X 2 + β 13 X 1 X 3 + β 23   X 2 X 3 + β 11   X 1 2 + β 22   X 2 2 + β 33   X 3 2 Where Y (%) is the flocculating activity of the fermentation supernatant, β₀ is the intercept coefficient; β₁, β₂, and β₃ are the linear coefficients; β₁₁, β₂₂, and β₃₃ are the quadratic coefficients; and β₁₂, β₁₃, and β₂₃ are the interaction coefficients; X₁, X₂, X₃ are the concentration of glucose, corn steep liquor, and urea, respectively. Three additional experiments were conducted to verify the validity of the statistical experimental strategies.

The flocculating activity was determined using the kaolin clay suspension method (Sun et al., 2012) with some modification. 1 mL bioflocculant sample (the fermentation supernatant was used in the optimization experiment) and 50 μL FeCl₃ (1 M) solution were mixed with 49 mL of kaolin solution (4 g/L) and then incubated for 5 min at room temperature. The absorbance of the upper phase of the mixture was measured by at 550 nm. A control in which the bioflocculant was replaced with deionized water was also conducted under the same conditions. The flocculating activity was calculated according to the following equation: F l o c c u l a t i n g   a c t i v i t y = B − A / B × 100 % Where A and B are the absorbance at 550 nm of the sample and control, respectively.

The cell-free supernatant obtained from the culture broth by centrifugation at 6,000 rpm and 4°C for 30 min was mixed with three volumes of ethanol and kept at 4°C overnight for precipitation. The resulting precipitate by centrifugation was dissolved in distilled water to repeat the precipitation procedure for three times. The final collected precipitate was dissolved in deionized water, dialyzed (molecular weight cutoff of 10,000 Da) to remove low molecular weight compounds and lyophilized at −50°C to obtain the purified bioflocculant.

The total content of polysaccharide, uronic acid, amino sugar and protein in the bioflocculant was measured by the phenol-sulfuric acid method, carbazole-sulfate method, Elson-Morgan method and Bradford method, with glucose, glucuronic acid, glucosamine and bovine albumin as the standard, respectively. For monosaccharide composition analysis, after hydrolysis of the bioflocculant with trifluoroacetic acid at 110°C for 6 h, the hydrolyzed sample was analyzed by using a high-pressure ion chromatography system (Dionex ICS 3000, Thermo Fisher Scientific Inc., United States), equipped with an anion exchange column (Dionex CarboPac PA20, Thermo Fisher Scientific Inc., United States) and a pulsed amperometric detector (reference electrode Ag-AgCl, measuring electrode Au). For neutral monosaccharide analysis, the mobile phase consisting of A (100 mM NaOH) and B (10 mM NaOH), was programmed as follows: 0–5 min, 100% A; 5–25 min, 10% A. For uronic acid monosaccharide analysis, the mobile phase consisting of A (10 mM NaOH) and B (100 mM NaOH containing 1 M sodium acetate), was programmed as follows: 0–20 min, 95% A; 20–30 min, 70% A; 30–35 min, 100% A. The identification of each monosaccharide peak was determined on the basis of the elution time by comparison with standard solutions of different monosaccharides (fucose, arabinose, rhamnose, galactose, glcosamine, glcose, xylose, galacturonic acid and glucuronic acid). The FT-IR spectra of the bioflocculant was recorded after grinded with KBr powder and pressed into a pellet, by a Bruker Vertex 70 spectrometer (Bruker Optics, United States) at the wavelength range of 400–4,000 cm⁻¹. The elemental composition of the bioflocculant was analyzed using a Thermo Scientific K-Alpha XPS system (Thermo Fisher Scientific Inc., United States). The surface morphology of the bioflocculant was observed by SEM (Hitachi SU8010, Japan).

The influence of pH was tested in the kaolin clay suspension system with a pH of 2.0–8.0 adjusted by 2 M HCl or 2 M NaOH. The cations including NaCl, KCl, MgCl₂, CaCl₂, CuCl₂, FeCl₂, FeCl₃, and AlCl₃ were used to test the effect of metal ions on the flocculating activity at the final concentration of 1.00 mM. The dose effects of the biopolymer (5–30 mg/L) and Fe³⁺ (0.50–1.50 mM) were studied at pH 2.0. The thermal stability was test at the temperature of 20, 40, 60, 80, and 100°C. The zeta potential of kaolin suspension before and after flocculation was analyzed by a Zetasizer (Zetasizer Pro, Malvern, United Kingdom).

The acid mine wastewater was sampled from iron ore processing plant of Xiangtan City in Hunan Province, China. In order to analyze the application potential of the bioflocculant in acid mine waste water treatment, the biofluccolant was added into the acid mine wastewater at the final concentration of 25 mg/L without pH adjustment. The suspension was shaken at 200 rpm and 30°C for 30 min. Then the flocculating activity was calculated according to the change of turbidity based on the absorbance at 550 nm and the pH change was record by a pH meter ((FiveEasy Plus pH, Mettler, United States). The supernatant was collected for detecting the concentration of Fe, Al, Mn, Cu, Pb, Zn, Ni and Cd by inductively coupled plasma atomic emission spectroscopy (ICP-AES, Hitachi Limited., Japan). The metal removal rate was calculated as follows: R e m o v a l   r a t e   % = C 0 − C / C 0 × 100 where C₀ and C were the initial and final concentrations of the metal, respectively.

Data are expressed as mean values with standard deviation (±SD) from three independent experiments. Statistical comparisons were performed using analysis of variance (ANOVA), and differences at p < 0.05 were considered statistically significant.

A three-level, three-factor Box-Behnken design was used to optimize the medium composition containing corn steep liquor in shake-flask fermentation. The obtained quadratic regression model representing the flocculating activity as a function of the carbon and nitrogen source was written as follows: Y = − 255.46 + 31.16 X 1 + 6.54 X 2 + 35.07 X 3 − 0.17 X 1 X 2 + 0.35 X 1 X 3 − 1.03   X 2 X 3 − 1.31   X 1 2 − 6.54   X 2 2 − 6.12   X 3 2 where Y was the flocculating activity of the fermentation supernatant, X₁, X₂, X₃ were the concentration of glucose, corn steep liquor, and urea, respectively. The statistical significance of the regression model was assessed by F-test and p-value. The analysis of variance (ANOVA) for the quadratic model was summarized in Table 2. Similarly to the study of Agunbiade et al. (2022), the model F-value of 99.09 indicated that the model was highly statistically significant at p < 0.0001. The lack of fit F-value and p-value was found to be 4.4 and 0.093, indicating that the suitability of the model to predict the variations. The p-values showed that X₁, X₂, X₃, X₁ ², X₂ ², and X₃ ² were statistically significant at the 95% confidence level. In contrast, X₁X₂, X₁X₃, and X₂X₃ had no significant influence on the flocculating activity. In addition, the value of the determination coefficient (R ²), the adjusted determination coefficient (R² _adj) and the predicted determination coefficient determination coefficient (R² _pred) were 0.9922, 0.9811, and 0.9015, respectively, indicating a good correlation between experimental and theoretical results. At the same time, a low value (1.36%) of the coefficient of the variation (C.V.) demonstrated the experimental values were precise and reliable.

From the regression model analysis, a maximum flocculating activity of the fermentation broth was estimated as 88.93% under the optimal medium composition with glucose at 12.20 g/L, corn steep liquor at 4.00 g/L and urea at 3.60 g/L. Under this optimal condition, the actual flocculating activity was 88.79% and was very close to the predicted value. By combination of repeated precipitation and dialysis, 0.52 g of purified bioflocculant was finally obtained from 1 L of fermentation broth.

Chemical analysis revealed that the purified bioflocculant contained neutral sugar (37.50%), uronic acid (10.03%), aminosugar (6.32%), protein (16.51%) and nucleic acid (3.72%). The detected rhamnose, arabinose, galactose, glucose, mannose, fucose, galacturonic acid and glucuronic acid in monosaccharide composition analysis were at a molar ratio of 1.11:2.29:1.84:3.67:4.67:2.93:5.67:4.85 (Figure 1).

The FTIR spectrum of the purified bioflocculant was displayed in Figure 2. The broad band around 3,420 cm⁻¹ was representative of the stretching vibration of O-H group. The weak band at 2,930 cm⁻¹ was attributed to the stretching vibration of C-H aliphatic bands (Huang et al., 2019). The band at 1,640 cm⁻¹ was resulted from C=O stretching vibration (Poorgholy et al., 2017). The bands at 1,557 and 1,407 cm⁻¹ were generated by N-H and C-H bending vibration, respectively (Salaberria et al., 2014; Yang et al., 2022) The band around 1,059 cm⁻¹ was from the stretching vibration of the C-O group (Fang et al., 2021).

The elemental composition of the biofocculant was analyzed by XPS. The mass fraction of C, O, N, S, and P was detected as 63.48%, 30.58%, 4.65%, 1.00%, and 0.29%, respectively. Moreover, the core level peaks of C 1s, O 1s and N 1s were deconvoluted to vertify the corresponding functional groups. As Figure 3 shown, the peaks located at 284.7, 286.2, and 287.6 eV in the C 1s high-resolution spectrum were assigned to C–C, C–OH/C–N and C=O groups, respectively (Li et al., 2009). The peaks located at 531.5, 532.3, and 533.1 eV in the O 1s high-resolution spectrum were attributed to C=O, O–C–O/O-C=O and C-OH, respectively (Liang et al., 2020). The peak at 400.28 eV in the N 1s high-resolution spectrum was assigned to the protonated nitrogen in aminosugar, while the peak at 399.5 eV was originated from the non-protonated nitrogen in NH or NH₂ group (Zhang et al., 2013; Zhong et al., 2018).

Effects of metal ions, pH, dosage, and temperature on flocculating activity of the bioflocculant were shown in Figure 4. The bioflocculant alone had negligible flocculating activity, but the situation changed by addition of metal ions. At pH 2.0, the bioflocculant at 10 mg/L demonstrated a good flocculating activity of 88.79% and 82.72% toward 4 g/L kaolin suspension, in the presence of 1.0 mM FeCl₃ and AlCl₃, respectively. If the metal ion changed or the pH increased, the flocculating activity dramatically dropped. The trivalent cations were more favorable for promoting flocculation than divalent and monovalent cations.

Considering that though the zeta potential of kaolin solution at pH 2.0 was changed from −12.8 to 9.55 mV by addition of the bioflocculant, no obvious fluccolation was observed, however, in the presence of both Fe³⁺ and the bioflocculant, the zeta potential of kaolin solution changed to 35.45 mV and the flocculation efficiency was significantly improved, it was speculated that trivalent cations with higher charge density could more effectively neutralize the negative charge of kaolin particles to form bridges (Salehizadeh et al., 2000). Variations in pH also cause differences in the electrostatic charge of the suspended particles, thus affecting the bridging efficiency for kaolin clay particles. The absolute values of zeta potential of kaolin suspension in the study were increased from −12.8 to −39.9 mV with the increase of pH 2.0 to 6.0.

The synergistic effect of bioflocculant and Fe³⁺ on the flocculating activity was tested at different dosages of bioflocculant and FeCl₃. The maximum flocculating activity achieved 91.92% at the dosage of 25 mg/L biolfocculant by combination of 1.0 mM Fe³⁺. When the bioflocculant concentration varied from 10 mg/mL to 30 mg/mL, its flocculating activity kept increasing with the increase of Fe³⁺ concentration from 0.5 to 1.0 mM, and at the presence of 1.0 mM Fe³⁺ the flocculating activities were above 88% with not much change during this bioflocculant concentration range, but overdosage of Fe³⁺ would reverse the surface charge of kaolin particles from negative to positive and reduced flocculation obviously by electrostatic repulsion forces (Wang et al., 2015; Nguyen et al., 2021; Hyrycz et al., 2022). With respect to the influence of temperature, the bioflocculant retained high flocculating activity above 85% when temperature varied from 20°C to 100°C, indicated it was stable under high temperature.

A real acid mine wastewater sample without pH adjustment was directly used to investigate the application possibility of the bioflocculant in treatment of acid mine wastewater. As shown in Figure 5, after treatment with the bioflocculant at a dosage of 25 mg/L for 30 min, the acid mine wastewater sample changed from yellow-brown to colorless and clarified, meanwhile, an amount of yellow-brown sediment appeared at the bottom, and the flocculating activity was tested as 96.39%. The yellow-brown color of the acid mine wastewater was caused by large quantities of ferric ions resulted from the oxidation of ferrous ions, and ferric ions could start to precipitate at a pH as low as 3.0 (Potgieter-Vermaak et al., 2006), it seemed that the addition of the bioflocculant accelerated the precipitation process of ferric ions by flocculation. ICP analysis further confirmed that the bioflocculant was able to remove metal ions from the acid mine wastewater sample, with a removal rate higher than 70% to Fe, Al, Zn, Pb, and Cd. Under the action of the bioflocculant, the final concentration of Pb²⁺ was lower than 1 mg/L, meanwhile, though the final concentration of Mn²⁺ and Cu²⁺ did not meet this standard, their concentration obviously changed from 49.7 to 28.6 mg/L and 5.72 to 2.45 mg/mL, respectively. Moreover, the pH of the acid mine wastewater was effectively increased from 3.04 to 6.86. The zeta potential of the acid mine wastewater was changed from −19.12 to 2.89 mV by addition of the bioflocculant. SEM images showed that the bioflocculant had an irregular stacked layer structure and aggregated with acid mine wastewater to form large clumps with net-like structure (Figure 6).