Optimization of carbofuran insecticide degradation by Enterobacter sp. using response surface methodology (RSM)

2.1 Chemicals and reagents

analytical standard Carbofuran PESTANAL@ 99% was purchased from Sigma Aldrich USA. High-performance liquid chromatography (HPLC) grade acetonitrile was purchased from Merch (Germany). All other chemicals and media ingredients used were of highest analytical grade are either from Merch (Germany), Sigma (India) or Fisher Scientific (Singapore).

2.2 Characterization of carbofuran degrading isolate

The bacteria were isolated in soil samples collected from different agricultural farms in peninsular Malaysia and used as initial inoculants for enrichment. Extraction of bacteria from the soil samples was done using serial dilution method as described by Collins and Lyne with little changes. One gram of each soil sample was suspended in 10 mL sterile medium (MSM). The MSM contained in g/L (Na₂HPO₄·12H₂O 6.5, KH₂PO₄ 0.9, FeSO₄·7H₂O 0.01, MgSO₄·7H₂O 0.15 and 2 mL of trace element solution) supplemented with initial 5 mg/L of carbofuran as source of carbon (Collins et al., 1989). Other dilutions were prepared until a dilution of 10⁻¹⁰ was achieved. To ensure maximum extraction of the bacteria, the dilutions were agitated carefully. While for liquid culture experiments, 0.2 mL aliquots of appropriate dilutions of the soil samples were inoculated into MSM flask supplemented with 50 mg/L carbofuran and incubated at 32 °C agitated for 14 days in a shaker incubator at 150 rpm. The experiment was done in aseptic conditions in triplicate. Flask having the same inoculum without addition of carbofuran insecticide serves as control experiment. Using the spread plate as well as pour plates techniques, 0.2 mL aliquot of the incubated culture was pour in agar plates containing 50 mg/l carbofuran insecticide (Chaudhry and Ali, 1988). Using a sterile wire loop the bacterial colonies in the plates were then streaked into nutrient agar plates supplemented with 50 mg/l of carbofuran and incubated at 32 °C for two days. Sub-culturing was done every two weeks on carbofuran containing agar plates until pure colonies were obtained. Isolated single colonies differing in morphological features were then re-suspended in MSM and used for further studies.

2.3 Growth analysis of bacterial culture

Fresh grown culture of Enterobacter sp. was aseptically inoculated into ten ml of nutrient broth (NB) and incubated using shaker incubator at 140 rpm at 37 °C until the growth of 0.8–09 OD_600nm reached. The bacterial cell was then incubated for 24 h at 37 °C using shaker incubator at 150 rpm. The liquid culture was centrifuged at 10,000x g rpm for 10 min. (Chin-Pampillo et al., 2015a). The supernatant was carefully filtered and used for further studies (Ibrahim et al., 2015).

2.4 Optimization using response surface methodology (RSM)

$$\mathrm{Y}={\beta }_0+{\beta }_1{X}_1+{\beta }_2{X}_2+{\beta }_3{X}_3+{\beta }_4{X}_4+{\beta }_{11}{X}_1+{\beta }_{22}{X}_2+{\beta }_{33}{X}_3+{\beta }_{44}{X}_4+{\beta }_{12}{X}_1{X}_2+{\beta }_{13}{X}_1{X}_3+{\beta }_{23}{X}_2{X}_3+{\beta }_{24}{X}_2{X}_4+{\beta }_{14}{X}_1{X}_4+{\beta }_{34}{X}_3{X}_4$$

Response surface methodology (RSM), which comprises regression analysis and factorial design, aids in measuring the significant parameters, in order to study the relations among the factors as well as choosing the optimal conditions of factors or appropriate responses (Ibrahim et al., 2015). In this study, two steps of optimization of carbofuran degrading strain were conducted by response surface methodology RSM (Ramasamy et al., 2017). Initial stage of the optimization was performed using Plackett–Burman Design (PBD) while, significant variables obtain were optimized further using central composite design (CCD) (Li et al., 2011). All tests were conducted in triplicate. (Chen et al., 2011). The percentage of carbofuran degradation was taken as the response (Rajendran et al., 2014). The correlation of the independent variables and that of the response were outline using the second order polynomial model that suits the response factors (El-Naggar Nel and Abdelwahed, 2014). The general formula for the second degree model applied is as follows:

In which the factor Y represent the response predicted while β₀, β₁, β₂, β₃, β₄, β₁₁, β₂₂, β₃₃, β₄₄, β₁₂, β₁₃, β₂₃, β₂₄, β₁₄ and β₃₄ are the regression coefficients of the model and is constant while β₀ represent the intercept, β₁, β₂, β₃, β₄ and β₁₁, β₂₂, β₃₃ and β₄₄ represent the linear as well as squared coefficients. While, X₁, X₂, X₃ and X₄ represent the independent factors. Factors combination (X₁, X₂) display the interface between the variables (Xia et al., 2012).

The factors that shows significant in Plackett Burman were further optimized by applying the CDD model (Yadav et al., 2015). The response design and the analysis of data was carried out with Design Expert software version 6.0.8 (See Table 1).

2.5 Statistical analysis

All the experiments were conducted in triplicate. Degradation studies were conducted using HPLC. All the data were statistically analyzed using Graph Pad v3.5. One-way ANOVA analysis of variance using the Tukey’s test was carried out and p < 0.05 was considered statistically significant.

3.1 Plackett–Burman experiment

Enterobacter strain BRC05 is being optimized by applying response surface model to examine the best state for its capacity to degrade carbofuran. A total of 12 experimental designs were prepared by applying Placket-Burman design (PBD), comprising five independent parameters namely (A) carbofuran concentration, (B) pH, (C) temperature, (D) carbon source (glucose) and (E) nitrogen source (NH₄CI). The higher and lower values of independent factors were selected according to previous literature. As shown in Table 2.

The PBD use the minimum number of runs to rapidly detect the variables with a significant effect on the response. Plackett-Burman design (PBD) deliver fast and efficient approach in identifying significant variables among greater amount of factors, thus, saves time and sustain extensive facts on each parameter. In Placket-Burman experiment, the main effects contain a complicated interaction with two factors associations (Yang et al., 2016). Therefore, the PBD is applied when studying major effects if the two way interactions are insignificant (Mirizadeh et al., 2014). Practically, PBD and fractional factorial designs were frequently applied to select significant variables that affect the process output measures or product quality (Sharma et al., 2009; Ibrahim et al., 2015; Ravanipour et al., 2015).

Experimental variables that affect carbofuran degradation are the temperature, pH, carbofuran concentration, carbon and nitrogen sources. The minimum and maximum values of the independent factors were chosen based on previous research (Khazaei et al., 2016; Kundu et al., 2016).

Placket-Burman experiment confirms the result of the significant factors runs, which shows that among the variables used, carbon source concentration was insignificant factor for carbofuran degradation using isolate BRC05 an Enterobacter specie with p-value remained greater than 0.05. This might be due to the fact that the bacteria already using carbofuran as source of carbon instead of the additional source (glucose). While the remaining four factors, i.e. Nitrogen source (Ammonium sulphate), pH, initial carbofuran concentration and Temperature were found to be significant with p-values < 0.05. The model coefficient of determination (R²) was 0.999 which may perhaps explain up to 99.9 percent of the response randomness. The Adjusted and the Predicted R-squared values were found to be 0.9987 and 0.9651 respectively, showing the existence of higher correlation between the actual and predicted values (See Tables 3 and 4).

3.2 Analysis using Central Composite design (CCD)

The optimum levels of the significant variables and the relations of these factors on the degradation of carbofuran were examined. Four significant variables achieved using Plackett–Burman design, were all experimented by central composite design. The levels selected for the variables were fixed according to previous studies (Ibrahim et al., 2015). The matrix for the Central Composite Design showing actual and predicted values results is displayed in Table 5. CCD response were utilized to generate response surfaces as well as 2D and 3D contours so as to show the effects of these variables on cabofuran degradation. Response Surface Methodology-CCD assists in understanding how individual factors interact with each other (Roriz et al., 2009).

The outcomes of the analysis acquired from the CCD analysis were fitted in second order polynomial equation so as to define the carbofuran degradation dependence. The effects and interaction of independent variables were presented and also their response were predicted. 30 run experiments performed base on response surface methodology while the response (%) were shown as predicted and actual values. The lowest and highest carbofuran degradation percent were (95.00%) and (19%) respectively Table 5.

Analysis such as F-value, R-squared value and lack of fit were calculated so as to estimate the model. Analysis of variance (ANOVA) of the model shows that the model is highly significant, as it is evident from the F test with quite low probability value (<0.0001). Non significance of the lack of fit with a probability value greater than 0.1 is essential. The lack of fit test assesses the model failure to represent data in the experimental domain at points, which are not included in the regression. Therefore, the test is expected to be non-significant to signify the model (Ahmad et al., 2018). In this study, the non-significant value of lack of fit (F value 0.35) shows that the quadratic model is statistically significant for the response, and hence, it can be used for further analysis. (Mohajeri et al., 2010) reported that when the lack of fit is insignificant then the model is a good fit. Also, (Ibrahim et al., 2015) stated that if lack of fit is significant in a model, the model is weak to be used to predict the response. The lack of fit test indicates whether the chosen statistical model is adequate to define the observed data. The outcome of the response surface model indicates that A, B, C, A², B², C², BD were found to be the significant model terms (See Table 6).

The determination coefficients, R-squared and the adjusted R-squared value of the response model were 0.9719 and 0.9457 respectively, which signify that the correlation that exist between the predicted and actual values are very high. Also, the predicted R-squared was 0.9292 which is in agreement with the actual adjusted R-squared. Thus, this shows that the regression model offers an outstanding justification of the bond between the independent variables as well as the response. Adeq Precision calculate the signal to noise ratio in which a ratio of greater than 4 is desirable. In this study the ratio of 27.706 indicates an adequate signal. Indicating that the model can be used to navigate the design space. The coefficient of variation (CV) at minimum value is 5.03% showing that the model was accurate and reliable. The model’s coefficient and their significances assessed using multiples linear regression for carbofuran degradation period. Analysis using multiple regression analysis by ignoring the values that are non-significant, shows the equation below as the basic quadratic model.$$\text{Y}=87.09-3.73\text{A}+1.90\text{B}+3.06\text{C}--2.88{\text{A}}^2--2.31{\text{B}}^2--14.31{\text{C}}^2+2.98\text{BD}$$

in which Y represent response value (%), while (A, B C, and D) represent the coded levels of carbofuran concentration, nitrogen source, pH and temperature respectively (See Fig. 3).

Close

Fig. 3

Agarose gel electrophoresis of PCR product of 16S rRNA gene of strain BRC05.

Export to PPT

The 2D and 3D contour plot signify graphical illustration of regression equation. It is aimed to know the interaction of parameters and trace the best level of each factor for maximum response (Li et al., 2011). Each plotted contour graph for carbofuran degradation indicates different collections of two test variables at a particular time, and retaining the other parameters at constant level (Ibrahim et al., 2015). The curved response surfaces indicate that there are distinct optimal parameters. If the surfaces are rather flat and symmetrical near the optimum, the optimized values may not vary widely from single parameter conditions (Kundu et al., 2016).

3.3 Interaction among factors

The model contour plots were presented to provide information on the response surface variables according to the second-order model through keeping two parameters to remain constant at zero and changing the last two parameters over the experimental range (Fig. 4). The interfaces between any of the two factors revealed that the values near the center point were best interacted. The shape of the contour plots revealed the interactive effects among the factors (See Figures 2A and 2B).

Close

Fig. 4

Phylogenetic relationship using Neigbour-joining method cladogram showing relation between Enterobacter cloacae strain BRC05 and other related strains based on the 16S rRNA gene sequence analysis.

Export to PPT

Circular pattern specified an insignificant interaction while an elongated response surface plot designated that the relationship between the subsequent variables were significant (Arasu et al., 2019; Xia et al., 2012). The nature of a contour plot specified weather the common interaction among the independent variables were significant or not (Ahsan et al., 2017). The response surface for the degradation of carbofuran by enterobacter sp. showed a clear peak, meaning the optimal point was inside the design boundary level as Illustrated in Figs. 2A and 2B.

Interactive effect of temperature and initial carbofuran concentration on the degradation of carbofuran by Enterobacter sp. was shown in Fig. 2A. It was observed that carbofuran removal ability increased at optimum conditions of temperature or initial carbofuran concentration and degradation rate reduced with a further increase with the temperature or initial carbofuran concentration. Also, the interaction between temperature and pH value was presented in Fig. 2B. The elliptical plot showed that the degradation ability of carbofuran was greatly affected by the change in pH or temperature. The decline in carbofuran was increased with increasing temperature. After the optimum temperature, the biodegradation ability decrease as the temperature was further increased (See Fig. 5).

Close

Fig. 5

Some Chromatographic profiles for biodegradation of carbofuran showing the carbofuran peak and its metabolites at various days of incubation.

Export to PPT