The bacterial strain (R. erythropolis 4.1491) used in this study was obtained from China General Microbiological Culture Collection Center (CGMCC), R. erythropolis was cultivated in flasks containing ATYP medium for 58.2 h at 15.3 °C with shaking at 180 rpm. The concentration of the viable organisms was estimated to be 10⁸ colony forming units (CFU)/mL by plate count method.

The ingredients of the ATYP medium for propagating R. erythropolis consist of (g/L): KH₂PO₄, 1; CaCl₂, 0.1; NaHCO₃, 3; CH₃COONa, 1; MgCl₂, 0.5; NH₄Cl, 1; NaCl, 1; C₄H₄Na₂O₄, 0.5; yeast extract, 0.5; peptone, 0.5. One milliliter trace element solution and 1 mL vitamin solution were supplemented to each liter of the ATYP medium to maintain the necessary nutrient requirement of the cells. The trace element solution consists of (g/L): FeCl₂·4H₂O, 1.8; CoCl₂·6H₂O, 0.25; NiCl₂·6H₂O, 0.01; CuCl₂·2H₂O, 0.01; MnCl₂·4H₂O, 0.7; ZnCl₂, 0.1; H₃BO₃, 0.5; Na₂MoO₄·2H₂O, 0.03; Na₂SeO₃·5H₂O, 0.01. The ingredients of the vitamin solution consist of (g/L): V_H, 0.1; nicotinic acid, 0.35; V_B1, 0.3; p-aminobenzoic acid, 0.2; C₈H₁₂N₂O₂·2HCl, 0.1; calcium pantothenate, 0.1; V_B12, 0.05. The pH of the medium was adjusted to 5.56.

The aflatoxigenic strain, A. flavus 2810, was kindly provided by Prof. Weijian Zhuang, Fujian Agriculture and Forestry University, China and was maintained on potato dextrose agar (PDA) medium (containing the extract from 200 g boiled potato, 20 g glucose and 20 g agar in 1 liter distilled water) at 4 °C. Spore suspensions were prepared by harvesting seven-day-old sporulating A. flavus cultures with sterile distilled water. Spores were counted with a hemocytometer and then were diluted with sterile distilled water to the desired concentration.

R. erythropolis (10⁸ CFU/mL) and 0.5 mL aflatoxin B₁ (10 mg/L, Sigma, Saint Louis, Missouri, USA) were added to autoclaved ATYP medium in 100 mL flasks. ATYP medium supplemented with AFB₁ without R. erythropolis was used as control. After incubation, the cells were removed by centrifugation at 12000 rpm for 10 min, and AFB₁ was extracted from the supernatants by methyl alcohol and quantified by aflatoxin B₁ ELISA test kit (Beacon Analytical Systems Inc., Portland, Maine, USA).

Peanuts that reach commercial level of maturity were harvested and used immediately or stored at 4 °C for use within 48 h. For preparation, the peanut kernels were washed with tap water, then surface-disinfected with 0.1% sodium hypochlorite for 1 min, cleaned with tap water and air dried [17]. Twenty grams of peanuts were placed in each of the four autoclaved 100 mL flasks (three treatments plus one control). The three treatments were: (A) Two milliliters of R. erythropolis cell suspension (10⁸ CFU/mL); (B) 2 mL B. megaterium cell suspension (10⁸ CFU/mL); (C) 1 mL each of R. erythropolis and B. megaterium cell suspension (10⁸ CFU/mL); and (D) sterile distilled water (control). After incubation at room temperature for three hours, 200 μL of A. flavus spores at a concentration of 10⁶ spores/mL were inoculated into each flask. The flasks were placed in growth chamber (QHX-300BS-III, Shanghai Xinmiao) with controlled humidity at 85% and temperature at 30 °C for incubation. The concentration of aflatoxin B₁ was examined after seven days’ inoculation using high performance liquid chromatography (HPLC) [24].

The design matrix selected for the screening of significant variables for the degradation of AFB₁ and the corresponding responses are shown in Table 2. The adequacy of the model was calculated, and the variables evidencing statistically significant effects were screened via Student’s t-test for ANOVA (Table 3). It is indicated that temperature, pH, liquid volume and incubation time were the most important factors. The polynomial model describing the correlation between the formulation and the variables (x₁ − x₆) and the degradation rate of AFB₁ (y) can be expressed by the following equation: (3)

The results of the t-test for variance between the average of observation of the 2-level experiment and the center point showed that the difference was significant (P < 0.05). This indicated that the optimum point was in the domain of our experiment. The next step in the optimization of the degradation efficiencies was to determine the optimum levels of significant variables. For this purpose, the RSM, using a central composite design, was adopted for optimization of the degradation efficiencies.

A central composite design (CCD) under RSM was used to analyze the interactive effect of temperature, pH, liquid volume and incubation time to reach an optimum level. The design matrix and the corresponding results of the RSM experiments to determine the effects of four independent variables are shown in Table 4.

Experimental data were used in the response surface regression (RSREG) procedure of SAS to find the coefficients of the response function. The coefficients of the response function for the degradation efficiencies are listed in Table 5. In this case, temperature, pH, liquid volume and incubation time had a significant effect on degradation of AFB₁ (P < 0.05), as well as the quadratic terms of temperature. The fitness of the model was examined by the coefficient of determination R², which was found to be 0.8096, indicating that the sample variation of 80.96% was attributed to the variables. The model can be shown as follows: (4)

Table 6 shows the analysis of variance of the regression parameters of the predicted response surface quadratic models for the degradation of AFB₁. As can be seen, both linear and quadratic parameters were very significant. However, the statistical analysis showed that the interaction among parameters was insignificant. This means that all factors considered in this study had a specific effect on the degradation. However, there were no effects of multi-variable interaction present. The effect of individual variable factors is more significant than that of the multi-factor interaction [25].

Figure 1, Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6 portray the 3D response surface plots constructed using the quadratic regression model generated by regression analysis. Figure 1 shows the effects of temperature and pH on the AFB₁degradation_. The degradation of AFB₁ increased with temperatures up to 25 °C. In temperature ranges over 25 °C, a negative effect of the temperature on AFB₁ degradation was observed. AFB₁ degradation increased with an increase in pH. The effects of temperature and liquid volume are shown in Figure 2. An increase in temperature above 25 °C resulted in less degradation of AFB₁. As the liquid volume increased, the degradation of AFB₁ increased at all temperatures probably due to the fact that a high liquid volume is conducive to the growth of bacteria. Figure 3 shows the effects of the interaction of temperature and incubation time on the degradation of AFB₁. As the length of the incubation time increased to 84 h, the degradation of AFB₁ increased at all temperatures. However, further increase in incubation time resulted in little increase in the degradation of AFB₁. Figure 4 is a response surface plot indicating the effects of pH and liquid volume on the degradation of AFB₁. pH and liquid volume had a positive linear effect on the degradation of AFB₁. The plot also reveals that properly increasing pH is conducive to increasing degradation. Figure 5 shows the effects of the interaction of pH and incubation time on the degradation of AFB₁. At 96 h, a linear increase in degradation was observed when the pH was increased. In alkaline conditions, increase in incubation time resulted in increased degradation. A similar profile is observed in Figure 6 with liquid volume and incubation time.

According to the canonical analysis, the results predicted by the model showed that the maximum degradation could be achieved when the temperature, initial pH, liquid volume and incubation time were set at 23.2 °C, 7.17, 24.6 mL/100 mL and 81.9 h, respectively. The predicted optimum rate of AFB₁ degradation was 96.7%. In order to confirm the optimization results, a further degradation test was carried out under the optimal conditions based on the results from the model; the experimentally observed optimum rate of AFB₁ degradation was 95.8% which was quite in agreement with the predicted value.

The strain of marine B. megaterium was isolated from the Yellow Sea of East China, and showed reduced postharvest infection of peanut kernels caused by A. flavus [17]. The results showed both R. erythropolis and B. megaterium could significantly reduce the biosynthesis of aflatoxins in peanuts (p < 0.01) (Table 7). The effect of R. erythropolis plus B. megaterium was better than that of a single bacterial strain.

Tejada-Castañeda et al., reported that a strain of Nocardia corynebacteroides was safe to detoxify aflatoxin-contaminated feed for chickens, because it could transform AFB₁ to other less toxic compounds [26]. Can R. erythropolis degrade AFB₁ completely or can it only transform AFB₁ to other less toxic compounds? To answer this question, the metabolites of AFB₁ incubated with R. erythropolis should be studied further. Furthermore, the work of Taylor et al., describes the enzymes in Mycobacterium smegmatis that degrade aflatoxin, which is subsequently corroborated by the work of Lapalikar et al., who show that enzymes from R. erythropolis also degrade aflatoxin [17,27]. These enzymes are dependent upon the cofactor F₄₂₀, whose production has been shown to be maximal at around 96 hours [28,29]. In a next step, the enzyme activity of F₄₂₀H₂-dependent reductases in the strain of R. erythropolis used in this study and the expression of aflatoxin pathway genes in A. flavus corresponding to the effect of R. erythropolis should be studied in greater detail, because the results could help us elucidate the mechanism of degradation of aflatoxin by R. erythropolis.