Optimization of Culture Conditions for Amoxicillin Degrading Bacteria Screened from Pig Manure

(1) Objective: The objective of this study was to screen amoxicillin (AMX)-degrading bacterial strains in pig manure and optimize the fermentation conditions for these strains to achieve high fermentation rate, which can provide an effective way for the practical application of bacterial strains as antibiotic-degrading bacterial in treating livestock waste for antibiotic residues. (2) Methods: Antibiotic susceptibility tests and high-performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) were employed to screen AMX-degrading bacterial strains in pig manure. The culture conditions were optimized for AMX-degrading bacterial strains using Plackeet–Burman design (PBD), the steepest ascent design, and the response surface methods, coupled with the Box–Behnken design (BBD). The effects of culture time, temperature, rotator (mixing) speed, inoculum level, and initial pH value on the growth of AMX-degrading strains were investigated. Experimental data obtained from BBD were utilized to generate a second-order polynomial regression model for evaluating the effects of the tested variables on the optical density at 600 nm (OD600) of culture solutions as the growth indicator for the screened AMX-degrading strains. (3) Results: The initial pH, culture time, and the inoculum level had significant effects on the OD600 value (growth) of the screened AMX-degrading strains. The initial pH value was found to be the most critical factor influencing the growth of bacteria. The optimized culture condition for the bacterial growth determined by the response surface methodology was: the initial pH of 6.9, culture time of 52 h, and inoculum level of 2%. The average OD value of 12 different fermentation conditions in the initial fermentation tests in this study was 1.72 and the optimization resulted in an OD value of 3.00. The verification experiment resulted in an OD value of 2.94, which confirmed the adequacy of the optimization model for the determining the optimal culture condition. (4) Conclusions: The growth of the screened strain of AMX-degrading bacteria could be optimized by changing the fermentation conditions. The optimization could be achieved by using the Box–Behnken response surface method and Plackett–Burman experimental design.


Introduction
Amoxicillin (AMX) is a broad-spectrum therapeutic antibiotic that can penetrate the cell walls of bacteria and thus kill pathogenic G + and Gbacteria (including coccus and bacillus) [1]. AMX has been commonly used in intensive livestock production because it is considered to be effective in treating a variety of animal diseases. However, like most antibiotics, AMX cannot be completely efficiency. The antibiotic susceptibility test qualitatively determined if an isolated bacterial strain was capable of degrading AMX. In this test, a piece of filter paper containing 10 µg of AMX was applied to the surface of the Luria-Bertani (LB) agar medium (NaCl (5 g/L), yeast extract (5 g/L), 1.5% (w/v) agar powder and tryptone (10 g/L) at pH 7.0) which had been inoculated with a degradation bacterial strain. As the diffusion distance of AMX in the agar increased, the AMX concentration decreased logarithmically to a certain concentration below which the bacterium would not grow, thus forming a transparent antimicrobial circle on the filter paper. The size of this inhibition zone reflected the sensitivity of the test bacterial to AMX, i.e., the smaller the circle, the stronger resistance of bacteria to AMX. The HPLC-MS/MS method was then used on the bacteria strains that had the small inhibitory zones observed in the susceptibility tests to further evaluate quantitatively their effectiveness in degrading AMX. A Liquid chromatography system-UPLC (I-Class)-MS (XEXO TQD) (Waters, Milford MA, US) coupled with Themo LENGEND MICRO 17R (Thermo Fisher Scientific, USA) were used in the HPLC-MS/MS evaluation. The Waters ACQUITY UPLC BEH-C18 column (1.7 µm, 2.1 × 100 mm; Waters, USA) with column temperature of 35 • C was used. A mobile phase of 0.1% (v/v) formic acid in water (A) and 0.1% (v/v) formic acid in acetonitrile (B) was used, filtered through a 0.22 µm membrane filter at a flow rate of 0.4 mL/min. The gradient dilution was performed as follows: 0 min, 95% A; 1 min, 95% A; 3 min, 2% A; 5 min, 2% A; 5.5 min, 95% A; 8 min, 95% A. The injection volume was 1 µL for both samples and standards. The mass spectrometer was used for the detection of analytes in the positive ion mode. The ionization source used a source voltage of 3.00 KV; a source temperature of 400 • C; a gas flow rate of 700 L/Hr; and cone of 50 L/Hr.
The standard stock solution of AMX was prepared with ultrapure water at a concentration of 1 µg/mL, which was then used to dilute to the following concentrations: 500, 200, 100, 50, 20, 10, 5, 2, and 1 ng/mL. The culture medium samples containing the isolated bacteria strains was diluted 1000 times and centrifuged for 10 min at 13,000 rpm. The supernatant was removed and filtered through a 0.22 µm membrane and aliquots of the supernatant were taken for HPLC-MS/MS analysis. The isolated strains that had higher degradation rates were selected for the optimization of fermentation conditions.

Determination of the Growth Curve of Bacteria Strain
The isolated bacteria strains were inoculated on LB liquid medium at 1% inoculation level and cultured at 30 • C for 24 h on a shaker incubator. A turbidimetric method was used to evaluate the growth curve of the bacteria [18,19], in which the growth of the bacteria was measured by the optical density of the bacterial cultural media at 600 nm (OD 600 ) [20] using 722 ultraviolet visible light spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA) in 4, 6,12,15,25,35,37,40,48,50,55, and 60 h of culture. The bacterial growth curve was plotted as the optical density OD 600 vs. the culture time.

Selection of Significant Variables by Plackett-Burman Design (PBD)
Based on the data from the fermentation experiment, the Plackeet-Burman design (PBD) was used to screen the factors that have the most significant influence among a set of factors by using a minimum number of tests [11]. It should be noted that this design is not focused on the interrelationship between factors, but merely on the significant effects of these factors [12]. Specifically, the effects of five parameters that defined the culture condition on the OD 600 value of bacterial cultures were investigated. The number of tests selected was N = 12, with five parameters X 1 , X 2 , X 3 , X 4 , and X 5 representing the fermentation time, the temperature, the inoculum level, the shaker speed, and pH, respectively. The high (+ 1) and low (-1) levels of these parameters are shown in Table 1, with the high level being 1.25 times the low level. The regression analysis was carried out using the software MINITAB 17.0 (Minitab, LLC, State College PA, USA) to assess the statistical significance. Three most significant influencing variables selected by the PBD were further analyzed by the steepest ascent design (SAD) to design the next experiment (selecting the rage of parameters) to further optimize the fermentation condition. The steepest ascent design uses the gradient direction of the experimental data as the climbing direction, and the change step size is determined according to the effect value of each factor, which can approach the target area quickly and economically [13].

Box-Benhnken Design (BBD)
The three significant factors and the best level of these factors identified after PBD and SAD were further optimized using Box-Behnken response surface design (BBD) [14]. This design coupled with the response surface methodology (RSM) is one of the most commonly used methods for process optimization with minimal experimental requirements [23]. In this design, the optical density (OD 600 ) value (growth) of bacteria was considered as the response value and the three identified significant variables the independent variables. Data analysis and experimental design were performed for each single factor using Design-Expert (Version 8.0.6, Stat-Ease Inc., Minneapolis, MN, USA).
Based on the BBD, a set of 12 tests were performed and a second-order polynomial model was used to fit to the data to generate a prediction model for the optimal fermentation condition.
where Y is the response (bacterial growth), X m , X n and X P are the three most significant parameters identified by the PBS among the five factors that were tested in the first set of fermentation tests (Table 1); α 0 is the offset term, and α i , α ii and α ij are the regression coefficients of the first-order main effect, the second main effect, and interaction effect, respectively.

Verification
To verify the effectiveness of the three-step optimization procedure (PBD-SAD-BBD), three separate tests were carried out according to the optimal fermentation conditions identified through the BBD experiments. The result was compared to the values predicted by the optimization Equation (1).

Screening by HPLC-MS/MS
The AMX degradation rates of strains AMX-1, AMX-3, and AMX-6 measured by using HPLC-MS/MS, along with the blank control, are summarized in Figures 2 and 3 and Table 2.

Screening by HPLC-MS/MS
The AMX degradation rates of strains AMX-1, AMX-3, and AMX-6 measured by using HPLC-MS/MS, along with the blank control, are summarized in Figures 2 and 3 and Table 2.

Screening by HPLC-MS/MS
The AMX degradation rates of strains AMX-1, AMX-3, and AMX-6 measured by using HPLC-MS/MS, along with the blank control, are summarized in Figures 2 and 3 and Table 2.    The peak time of non-inoculated AMX standard solution detected by HPLC-MS/MS was 1.86 min ( Figure 2E). Ninety (90) min after inoculation, AMX was still detected in the solutions of strains AMX-3 and AMX-6, but not in AMX-1 (a new unknown substance with an absorption peak at 2.29 was detected, Figure 2A). In other words, AMX was degraded and new degradation products were produced under the action of AMX-1 strain [24]. Regression analysis yielded a linear relationship between the concentration of AMX and the corresponding peak: Y=39.31X−1.71, R 2 =0.999. Using this relationship, the degradation rate by AMX-1 was calculated be to 98.5%, which was significantly higher than the other two strain (Figure 3). Therefore, AMX-1 bacterial strain was selected for the optimization of fermentation conditions.

Growth Curve of AMX-1
The growth of AMX-1 strain roughly followed a typical "S" curve ( Figure 4). A lag period occurred from 0 to 25 h, followed by a logarithmic growth period from 25 to 50 h, where the number of live bacteria increased rapidly and the maximum growth rate was reached. The growth reached a stable state after 55 h, which indicated the total number of living bacteria reached the maximum. The peak time of non-inoculated AMX standard solution detected by HPLC-MS/MS was 1.86 min ( Figure 2E). Ninety (90) min after inoculation, AMX was still detected in the solutions of strains AMX-3 and AMX-6, but not in AMX-1 (a new unknown substance with an absorption peak at 2.29 was detected, Figure 2A). In other words, AMX was degraded and new degradation products were produced under the action of AMX-1 strain [24]. Regression analysis yielded a linear relationship between the concentration of AMX and the corresponding peak: Y = 39.31X − 1.71, R 2 = 0.999. Using this relationship, the degradation rate by AMX-1 was calculated be to 98.5%, which was significantly higher than the other two strain (Figure 3). Therefore, AMX-1 bacterial strain was selected for the optimization of fermentation conditions.

Growth Curve of AMX-1
The growth of AMX-1 strain roughly followed a typical "S" curve ( Figure 4). A lag period occurred from 0 to 25 h, followed by a logarithmic growth period from 25 to 50 h, where the number of live bacteria increased rapidly and the maximum growth rate was reached. The growth reached a stable state after 55 h, which indicated the total number of living bacteria reached the maximum.

Effects of Individual Factors
The measured optical density (OD600), representing the bacterial growth rate, increased first with temperature until about 35 ℃ and then decreased, with the quickest increase between 25 and 30 ℃ (Figure 5a). It was interesting to note that when the temperature was 40 ℃, the OD600 value was even lower than that at 30 ℃. It was clear that the optimal temperature range for AMX-1 growth was 30-35 ℃, while other condition parameters were at the 0-level as described in Table 1.

Effects of Individual Factors
The measured optical density (OD 600 ), representing the bacterial growth rate, increased first with temperature until about 35 • C and then decreased, with the quickest increase between 25 and 30 • C (Figure 5a). It was interesting to note that when the temperature was 40 • C, the OD 600 value was even lower than that at 30 • C. It was clear that the optimal temperature range for AMX-1 growth was 30-35 • C, while other condition parameters were at the 0-level as described in Table 1.
The bacterial growth increased with the inoculum level rapidly between 1.0-1.5% (v/v), reaching the maximum at about 1.5% (v/v) (Figure 5b). Further increase in inoculum level resulted in a slower growth because oxygen and nutrients available to the bacteria in the medium became limiting and accompanied by bacterial proliferation, affecting the production of primary and secondary metabolites [25].
The shaker speed was related to the dissolved oxygen content in the culture solution [26,27]. Through the vibration of the shaker, the oxygen in the air was continuously dissolved into the culture solution for the bacteria. The bacterial growth increased with the rotation speed of the shaker linearly, indicating that the bacterial growth was closely related to the aeration level (Figure 5c), and that the bacterial strain was aerobic.
The bacterial growth increased with time and reached a maximum value at 50 h ( Figure 5d). The growth stayed almost constant when the time was extended to 60 h, indicating that the bacterial growth had entered a stable state after 50 h. The possible reasons were: (i) depletion of nutrients in the medium; (ii) imbalance of nutrients, such as the inappropriate C/N ratio; (iii) accumulation of harmful metabolites such as acid, alcohol, toxins, or hydrogen peroxide (H 2 O 2 ); and (iv) physical and chemical conditions such as pH and redox potential becoming less suitable for the bacteria [28,29]. The bacterial growth increased with the inoculum level rapidly between 1.0-1.5% (v/v), reaching the maximum at about 1.5% (v/v) (Figure 5b). Further increase in inoculum level resulted in a slower growth because oxygen and nutrients available to the bacteria in the medium became limiting and The bacterial strain grew well only in a narrow range of pH, with the optimal pH around 7 (Figure 5e). In the acidic or alkaline environment with a pH of 5 or 10, the OD 600 value of this strain decreased significantly, which indicated that the initial pH value could significantly affect the growth of this strain.

Plackett-Burman Design (PBD) Screening
The purpose of PBD was to identify which conditions of fermentation had a significant effect on the growth of the AMX-1 strain. The optical density OD 600 ranged from 1.589 (run 7) to 1.899 (run 9), as measured in the fermentation experiment ( Table 3). The results of statistical analysis on the main effect are presented in Table 4. The significance of each factor was assessed using the P-value at the significance level of 0.05. The significance of the five variables in terms of their effects on the optical density (OD 600 ) was ranked as: X 1 (Time) > X 3 (Inoculum level) > X 5 (pH) > X 2 (Temperature) > X 4 (Rotation speed). Among them, the effects of the fermentation time, inoculum level and pH were statistically significant (P < 0.05). Therefore, these three variables were considered as the significant factors in the next experiment.

Steepest Ascent Design
The Coefficient Estimate of X 1 , X 3 and X 5 presented in Table 4 indicated that the fermentation time and the initial pH value exhibited positive effects and the inoculum level exhibited a negative effect. Based on these observations, the directions of the gradients for X 1 , X 2 , and X 4 in the steepest ascent design were determined (Table 5). Specifically, the time and pH were gradually increased, while the inoculum level was decreased. The corresponding optical density (OD 600 ) value first increased and then decreased (Table 5), with the maximum OD 600 at the fermentation time 50 h, pH 6.5, and inoculum level 1.5% (v/v). Table 5. The design and results of steepest ascent test.

Test
Step Size X 1 X 3 X 5 OD 600

Optimization: Box-Behnken Design
Based on the data in Table 5 (the maximum OD 600 at fermentation time X 1 = 50 h, the inoculum level X 3 = 1.5% (v/v), and pH X 5 = 6.5, a BBD design was proposed to further refine the optimal fermentation condition ( Table 6). The corresponding experimental results based on this BBD are shown in Table 7. Table 6. The factors levels in BBD.
The details of regression analysis are summarized in Table 8. Note: The P-value less than 0.001 for the difference is extremely significant; the P-value less than 0.01 is highly significant; the P-value less than 0.05 is significant.
The P-value for the overall model was 0.0044, which demonstrated that the quadratic equation model was highly significant [30]. Based on the P-values associated with the three factors, their influences on the optical density (OD 600 ) of bacterial growth were ranked as: X 5 (pH) > X 3 (Inoculum level) > X 1 (Time). The P-value for misfitting was 0.5733, indicating that the misfitting term was insignificant relative to the absolute error, and the misclassification is not significant. At the same time, the adjusted determination coefficient (Adj R-squared, Table 9) (0.82 > 0.80) and coefficient of variance (CV%) was 9.19, which further indicated that the multivariate regression relationship between the dependent variable and the all independent variables was significant [31]. In other words, the regression equation was adequate in predicting the optical density (OD 600 ) under different fermentation conditions [32] (Figure 6). level) > X1 (Time). The P-value for misfitting was 0.5733, indicating that the misfitting term was insignificant relative to the absolute error, and the misclassification is not significant. At the same time, the adjusted determination coefficient (Adj R-squared, Table 9) (0.82 > 0.80) and coefficient of variance (CV%) was 9.19, which further indicated that the multivariate regression relationship between the dependent variable and the all independent variables was significant [31]. In other words, the regression equation was adequate in predicting the optical density (OD600) under different fermentation conditions [32] (Figure 6).

Determination of Optimal Fermentation Condition
The response surface contour maps for the factors were generated based on the regression Equation (2) to search for the optimal fermentation condition ( Figure 7). The profiles of the response surfaces between fermentation time and inoculum level, fermentation time and pH, inoculum level and pH were all convex with an open downward direction, indicating a parabolic relationship between the OD600 value and the three factors of fermentation time, inoculum level and pH [33].The surface plots showed the interactive effects of the three significant fermentation factors on the bacterial growth. Figure 7c,b clearly demonstrates that the OD600 increased with pH, with the optimal pH value of 7.00 (Figure 7a) [29]. Comparing Figure 7a,c, and b indicates that the pH also had the most significant effect on the bacterial growth among the three The profiles of the response surfaces between fermentation time and inoculum level, fermentation time and pH, inoculum level and pH were all convex with an open downward direction, indicating a parabolic relationship between the OD 600 value and the three factors of fermentation time, inoculum level and pH [33].The surface plots showed the interactive effects of the three significant fermentation factors on the bacterial growth. Figure 7b,c clearly demonstrates that the OD 600 increased with pH, with the optimal pH value of 7.00 (Figure 7a) [29]. Comparing Figure 7a-c indicates that the pH also had the most significant effect on the bacterial growth among the three variables as the response surface curvatures were smaller (Figure 7a). Figure 7b,d, and m shows that the contour lines were circular, indicating the interaction of these three factors was not obvious [29,31]. Overall, the optimum fermentation condition for the screened bacterial strain was predicted to be: the fermentation time of 52.1 h, the inoculum level of 2%, and pH of 6.855, and this condition resulted in a predicted OD 600 value of 3.00.

Experimental Verification
To verify the prediction model (Equation (2)), the predicted optimal fermentation condition (fermentation time X 1 = 52 h, the inoculum level X 3 = 2%, and pH X 5 = 6.9) was used to conduct a set of tests to measure OD 600 . The average OD600 of five replicates was 2.94 with a standard deviation of 0.045. The difference between the predicted and measured values was about 2%, indicating that the model was adequate in predicting the optimal fermentation condition.

Conclusions
An effective Amoxicillin (AMX) degrading bacteria strain was successfully isolated and screened from pig manure through the antibiotic susceptibility testing and HPLC-MS/MS. A three-step procedure was developed to optimize the fermentation conditions for this screened bacteria strain. In the developed optimization procedure, the Plackett-Burman design (PBD) was first used to select most significant parameters; the steepest ascent design (SAD) was then used to define the gradient direction for the Box-Benhnken design (BBD); and finally the BBD with the response surface method (RSM) was used to refine the optimal fermentation condition, and a predictive model of quadratic polynomial was developed. It was found that among the five factors that defined the fermentation conditions, namely the inoculum level, the fermentation time, temperature, pH and shaker speed, three had significant effect on the growth of screened bacterial strain in the following order: pH, the fermentation time, and the inoculum level. The determined optimal fermentation condition was: pH 6.9, the fermentation time 52.1 h, and the inoculum level 2%. The proposed optimization process increased the bacterial growth rate by 67.6%. The proposed quadratic polynomial model adequately (within 2%) predicted the optimal fermentation condition.