Modification of Medium Composition for Enhancing the Production of Antifungal Activity from Xenorhabdus stockiae PB09 by Using Response Surface Methodology

Xenorhabdus stockiae PB09 bacterium has been shown to exhibit antifungal activity against several plant pathogens. To improve its efficacy, the optimization of the nutritional components in culture media was performed. The medium components that have significant effects on antifungal activity of X. stockiae PB09 were initially identified using a fractional factorial design. Response surface methodology and central composite design were then used to create a model for optimizing the levels of carbon, nitrogen, and mineral sources that maximize antifungal activity of X. stockiae PB09. After that, the suitable carbon, nitrogen, and mineral sources were selected and adjusted by the second-order polynomial regression model, which predicted that 98.62% of antifungal activity could be obtained when the medium contained sucrose, yeast extract, NaCl, and K2HPO4 at 3.24, 23.71, 5.46, and 2.73 g/L, respectively. Laboratory verification of this recipe resulted in the antifungal activity at 97.95% in the shake flask experiment after 48-hour cultivation, which was significantly 27.22% higher than that obtained by using the TSB medium. In addition, X. stockiae PB09 cultured in the verified recipe by using 5 L fermenter could effectively inhibit the mycelial growth of Phytophthora sp., Rhizoctonia solani, Pythium sp., and Fusarium oxysporum. This study demonstrated that the RSM and CCD were shown to be valuable tools for optimizing the culture medium that maximize the antifungal activity of X. stockiae PB09.


Introduction
Entomopathogenic nematodes of the genera Steinernema are effective biological control agents for a wide range of agricultural pests [1]. When Steinernema nematodes infest the insect hosts, they release their symbiotic bacteria Xenorhabdus spp., into hosts' haemocoels.
Antifungal activity of entomopathogenic bacteria has been found to differ qualitatively depending on the bacterial species and strains as well as their culture conditions. Nutrition in medium culture can play an important role in triggering and increasing the levels of secondary metabolite production. e quantity and type of essential nutrients are found to be the effective means for not only restricting the bacterial growth but also specifically affecting their metabolic and regulatory pathways [17]. Optimization of culture media for enhancing the production of antibiotic activity of Xenorhabdus spp. has been developed by using the response surface methodology (RSM) [18][19][20]. Despite the presence of complex metabolic interactions, RSM can be used to evaluate the relative significance of several factors that affect the bacterial pathways [21]. is method is often employed after the "vital few" controllable factors have been identified, and it then needs to find the factors that set the optimization of the responses. is method has successfully been applied in many areas of biotechnology such as lactic acid, antibiotic, and enzyme productions [22][23][24]. e aim of this study was to optimize the recipe of culture medium for maximizing the antifungal activity of X. stockiae PB09 by using stepwise optimization as follows: (1) screening the most effective original medium; (2) modifying that original medium by replacing its carbon and nitrogen sources with new ingredients that significantly affect the antifungal activity by using the one-factor-at-atime approach; (3) optimizing all the significant ingredients, whereby a fractional factorial design (FFD) was initially employed to screen the most significant factors, the steepest ascent method was then carried out for the suitable concentrations, and, finally, a central composite design (CCD) of RSM was applied to search for the optimal medium compositions; and (4) scaling up the cultivation in batch fermentation. e results obtained from this study could be beneficial for further enhancement of secondary metabolite production by X. stockiae PB09, especially in the large-scale settings.

Bacterial
Cultures. Xenorhabdus stockiae PB09 isolate was derived from the infective juveniles (IJs) of Steinernema siamkayai Stock, Somsook, and Reid, which was obtained from the Department of Agriculture, Ministry of Agriculture and Cooperatives, ailand, using the method described by Kaya and Stock [25]. X. stockiae PB09 was maintained in phase I and used throughout the study. To ensure phase I, the bacteria were subcultured onto NBTA agar (consisting of peptone, 10 g/L; beef extract, 3 g/L; NaCl, 5 g/L; and agar, 15 g/L, and supplementing with triphenyltetrazolium chloride, 0.040 g/L, and bromothymol blue powder, 0.025 g/L) and incubated at 28°C in the dark for 2-3 days. X. stockiae PB09 in phase I could be distinguished from phase II by its adsorption of bromothymol blue and by formation of blue colonies on NBTA.
A seed culture of X. stockiae PB09 was prepared by inoculating a full loop of phase I colonies into a 250 mL flask containing 100 mL of nutrient broth, which was cultivated at 28°C on a rotary shaker at 200 rpm for 16-24 h in complete darkness to result in the optical density (600 nm) of approximately 2.

Measurement of Bacterial Biomass.
e biomass of bacterial cells was measured by optical density (OD) value of the culture at 600 nm using a spectral photometer, and the dry cell weight (DCW) was calculated as described by Wang et al. [26]. e calibration curve was initially plotted between the dilutions of bacterial cell suspension and the obtained optical density values. A fixed volume of each dilution was centrifuged at 10,000 rpm for 20 min, and its cell pellet was dried at 110°C in a hot air oven for 48 h and finally weighted for converting the optical density value into the dry cell weight (g/L).

Assay of Antifungal Activity.
Cell-free supernatants of X. stockiae PB09 were in vitro tested for their efficacy against Phytophthora sp. mycelia growth by using the poisoned food technique [11]. Phytophthora sp. was provided by the Biocontrol Research Unit, Mahasarakham University, and was cultured on potato dextrose agar (PDA) at 25°C for 5-7 days (L : D � 12 : 12 h). Its young mycelia at the edge of colonies were then punctured by sterile 6 mm cork borer to produce mycelial plugs. In the meantime, cell-free supernatant of X. stockiae PB09 was mixed with sterile, melting PDA that had been cooled down to approximately 50°C at 10% (v/v) in the final volume of 10 mL and then poured into each 9 cm Petri dish. After PDA mixed with cell-free supernatant of X. stockiae PB09 became solidified, each 6 mm mycelial plug was placed upside down onto the center of each plate. e experiment was performed in four replicates. All plates were incubated in an incubator at 30°C for 5 days and the diameter of fungal colony of each plate was measured. e inhibitory activity of cell-free supernatant of X. stockiae PB09 against Phytophthora sp. mycelial growth was calculated using the following formula: I � ((C−T)/C) × 100, where I is the percentage of mycelial inhibition, C is the radial growth in control group, and T is the radial growth in treatment group.

Selection of the Optimal Carbon and Nitrogen Sources.
e one-factor-at-a-time approach was used to determine the optimum carbon and nitrogen sources for enhancing the antifungal activity of X. stockiae PB09. Various sources of carbon (glucose, fructose, sucrose, maltose, and glycerol) and nitrogen (tryptone, yeast extract, peptone, beef extract, and soybean meal) were individually used to replace the corresponding sources in the original TSB medium, while NaCl and K 2 HPO 4 were kept unchanged. e antifungal activity of X. stockiae PB09 bacteria cultured in different media were determined after 48 h of cultivation at 28°C in dark using a rotary shaker at 200 rpm. e aliquots of the resulting culture broths were individually centrifuged and filtered to separate the bacterial cells from the supernatants which were used for bioassays. e supernatants were stored at 4°C until required.

Fractional Factorial Design (FFD).
is study used the FFD to identify the medium's ingredient that had significant effects on the antifungal activity. ree independent variables, carbon (sucrose), nitrogen (yeast extract), and mineral (NaCl and K 2 HPO 4 ) sources, were included in the 2-level fractional factorial design. According to the FFD, each variable was prepared in two levels: −1 for low level and +1 for high level. e design matrix of the tested variables and the levels of each variable are shown in Table 1. e main effect analysis of the influencing factors is shown in Table 2.
e FFD is based on the first-order model: where Y is the predicted response, β 0 is a constant coefficient, β i is the linear coefficient, and X i is the level of independent variables. In this study, three variables were screened by 8 experimental runs in addition with four runs at the center point. Design-Expert software (version 7.1, Stat-Ease, Inc., Minneapolis, USA) was used to design the experiments. Each experiment was conducted in three replicates.

Path of Steepest
Ascent. e method of steepest ascent was used to move rapidly to the vicinity of the optimum response. e center point of the FFD was considered as the origin of the path. e direction of steepest ascent was parallel to the normal of the contour line of the response curve of the model (1). e experimental design and the corresponding response of steepest ascent are shown in Table 3.

Central Composite Design (CCD).
Continued from the above studies, the central composite experimental design (CCD) was used to optimize the concentrations of three effective nutrients. e three independent variables (sucrose, yeast extract, and mineral) were used as main variables and designated as X 1 , X 2 , and X 3 , respectively. e minimum and maximum ranges of variables were used as shown in Table 4. For developing the regression equation, the tested factors were coded according to the equation below: where x i is the coded value of an independent variable, X i is the actual value of independent variable, X 0 is the actual value of the X i at the center point, and ΔX i is the step change value.  e response (percentage of mycelial inhibition) was added to a second-order model in order to estimate its correlation with the independent variables. An empirical second-order polynomial model is shown in the following equation: where Y is the predicted variable, X i and X j are input variables which influence the response variable Y, β 0 is the intercept term, and β i , β ii , and β ij are measures of the effects of variables X i , X 2 i , and X i X j , respectively. According to the central composite design, a 2 3 fullfactorial CCD for three independent variables (each at five levels with four axial points and six replicates at the center points) was employed to fit the second-order polynomial model, which clearly showed that optimization of medium constituents would require at least 20 experiments.
For regression and graphical analyses of the obtained data, the Design-Expert software (version 7.1, Stat-Ease, Inc., Minneapolis, USA) was used. By optimizing the second-order polynomial equation (3) based on the desirability criterion of maximum percentage of mycelial inhibition, the suitable combinations of independent variables could be achieved. Student's t-test for the estimated coefficients was also applied for statistical analyses. Moreover, statistical analysis of the model was performed by analysis of variance (ANOVA). Fisher's F-test (overall model significance) was used for determination of the second-order polynomial model equation, and the quality of the fit of regression model equation was given by the coefficient of determination, R 2 . e cultivation temperature was 28°C with agitation speed of 200 rpm and aeration rate of 2.5 L/min. e seed culture of X. stockiae PB09 was transferred to 3 L of sterile medium (the optimized medium) in the fermenter at the ratio of 10% (v/v). e pH profile was adjusted by using 2 mol/L NaOH and 2 mol/L HCl. e batch fermentation process was performed for 72 h. e sample (5 mL) was taken out every 3 h, and then, 5 mL of the culture broth was spun (10,000 rpm, 20 min, 4°C) using centrifuge MPW-380R and filtered using 0.22 µm-syringe filters to obtain cell-free supernatant, which was stored at 4°C until required. Four plant pathogenic fungi, namely, Phytophthora sp., Pythium sp., Rhizoctonia solani Kuhn, and Fusarium oxysporum Schlechtendahl, were used to determine the reduction of mycelial growth using the optimum medium obtained above. e experiment was repeated in triplicates.

Effects of Different Media on Biomass and Antifungal
Activity of X. stockiae PB09. e effects of three different media, TSB, LB, and YSG, on biomass and antifungal activity of X. stockiae PB09 cells after cultivation by using shake flasks at different periods of time are shown in Table 5. e maximum dry cell weight was also found when the bacteria were cultured for 48 h on TSB (11.90 g/L). e cell-free supernatant of X. stockiae PB09 cultivated by using TSB medium for 48 h exhibited the highest inhibitory activity against the mycelial growth of Phytophthora sp. (70.73%). erefore, TSB medium was shown to be the optimum medium for maximizing biomass and antifungal activity of X. stockiae PB09.

Effect of Various Carbon and Nitrogen Sources on
Antifungal Activity of X. stockiae PB09. Xenorhabdus bacteria are potential producers of secondary metabolites with antifungal activity. e carbon and nitrogen are the important nutritional components in the medium for bacterial cultivation, and Xenorhabdus spp. bacteria have been shown to require both of these components for their metabolite production. Different species of Xenorhabdus bacteria seem to require rather different types of carbon and nitrogen  International Journal of Microbiology sources, for example, glucose and peptone were the best carbon and nitrogen sources for antibiotic production by Xenorhabdus nematophila TB [19]; however, Xenorhabdus bovienii was found to be greatly influenced by glycerol and soytone [20]. In addition, tryptone and dextrose were shown to significantly increase the protease production by Xenorhabdus indica KB-3 [27]. TSB medium was used as a base medium for selecting the alternative carbon and nitrogen sources, whereby both of which were maintained at the concentrations of 2.5 and 20 g/L, respectively ( Figure 1). e effect of different carbon and nitrogen sources on X. stockiae PB09 antifungal activity is shown in Figure 1. e results showed that glucose could maximally enhance the antifungal activity, followed by sucrose, but both of which were not significantly different. Similar results were obtained in the study of Zhang et al. [28] that found the highest production of active antifungal substances by Streptomyces hygroscopicus BS-112 when glucose was used as a carbon source. In addition, Song et al. [29] also reported that glucose was the optimal carbon source for production of antifungal substances by Brevibacillus laterosporus.
For nitrogen sources, the highest antifungal activity could be achieved when using yeast extract ( Figure 1). Furthermore, yeast extract was found to be the optimal nitrogen source for production of antifungal substances by Mycena leptocephala [30]. In addition, M. leptocephala was found to have increased biomass and antifungal activity when malt extract plus glucose was used as the carbon source and yeast extract was used as the nitrogen source. However, Elibol [21] reported that sucrose, glucose, yeast extract, and peptone had the most profound effects on actinorhodin production by Streptomyces coelicolor A3 (2) in complex medium. erefore, in this study, sucrose and yeast extract were selected for further maximizing the antifungal activity of X. stockiae PB09.

Significant Factors.
For the first optimization step, the FFD was used to screen the relatively significant variables for the production of antifungal activity. e effects of the medium components, including sucrose (X 1 ), yeast extract (X 2 ), and mineral (X 3 ), on antifungal activity were determined. e experimental design and the FFD results are shown in Table 1. e values of the regression coefficients were calculated, and an equation of the first-order model could be written from the coefficients: where Y was the percentage of mycelial growth inhibition, and X 1 , X 2 , and X 3 are coded values of sucrose, yeast extract, and mineral, respectively. Regression analysis of the FFD in Table 2 shows that sucrose, yeast extract, and mineral in the concentration range test had significant effects on antifungal activity (P < 0.05). erefore, these three variables were used for further optimization experiment by using RSM.

e Steepest Ascent Path Test Results of Significant Factors.
e path of the steepest ascent was used to attain the proper direction of changing variables, that is, increasing of sucrose, yeast extract, and mineral (NaCl and K 2 HPO 4 )  International Journal of Microbiology concentrations, to improve the antifungal activity. e direction of the steepest ascent path was determined by using the first-order model (4). e experimental design and corresponding results of the steepest ascent were listed in Table 3. When considering the results from the path of steepest ascent, it was clearly seen that the percentage of mycelial inhibition showed the maximum of 90.50% at run number 2 (Table 3). us, the combination of run number 2 was selected as the center point of CCD.

Central Composite Experimental Results.
is experiment was designed to optimize the concentrations of three selected variables in order to maximize the antifungal activity of X. stockiae PB09 by using the CCD of RSM. e three significant independent variables, that is, sucrose, yeast extract, and mineral (NaCl and K 2 HPO 4 ) from the FFD were selected for designing the matrix. e design matrix of the tested variables is shown in Table 6.
By applying multiple regression analysis on the experimental data, the following second-order polynomial equation was used to explain the antifungal activity: where Y is the percentage of mycelial inhibition (%), and X 1 , X 2 , and X 3 are the code values of sucrose, yeast extract, and mineral, respectively. e results of response surface quadratic model in the form of analysis of variance (ANOVA) are shown in Table 7.
is model was shown to be very significant by using Fisher's F-test (F model , mean square regression/mean square residual � 16.93) with a very low probability value ((P model > F) < 0.0001). By determining the coefficient (R 2 ), the goodness of fit of the model was investigated. In this case, the value of the R 2 (0.9384) for (5) suggested that the sample variation of 93.84% for antifungal activity was the result of variables, and only 6.16% of the total variations was not explained by the model. High level of adjusted determination coefficient (adj. R 2 � 0.8830) also indicated the strong significance of the model. In addition, the good predictability of the model could be suggested by its statistically insignificant lack of fit ((P model > F) � 0.3149). Moreover, the low value of coefficient of variation (CV � 3.14%) suggested that these experiments were precise and reliable. e normal probability plot of residuals was the most significant diagnostic method for the model and therefore set as default (Figure 2). To indicate whether there were no signs of problems in the data, the normality in the error term could be verified by linear patterns [23,31]. All in all, for  prediction within the range of selected variables, this model was considered to be suitable. e results obtained from experimental mycelial inhibition were compared with that of the statistical model (5) and shown in Figure 3, which indicates that the empirical model gave predicted data of the response in the same direction with the experimentally obtained data. e 2D contour plot and the 3D response surface curve ( Figure 3) were the graphical representations of the regression equation, which presents the effect of two variables on the antifungal activity of X. stockiae PB09 when maintaining the other two variables at their respective zero level. e significant interactions between the independent variables were indicated by the elliptical nature of the contour plots. e 2D contour plot and 3D response surface curve are drawn in Figures 3(a)-3(c), according to the combined effects of sucrose-yeast extract, sucrose-mineral, and yeast extract-mineral, respectively, from which the percentage of mycelial inhibition for different concentrations of the variables could be predicted. Here, the response surface plots helped to understand the interactions between two nutrients and therefore to locate their optimum levels. e 2D contour plot of sucrose and yeast extract suggested their remarkable interaction, whereby the 3D response surface curve clearly indicated the sucrose and yeast extract concentrations (Figure 3(a)). From these results, the significant interactions between sucrose and yeast extract could imply that the ratio between C and N was critical for the production of active substances. e ratio between C and N was found to directly influence the levels of bacterial biomass and accumulation of their metabolites [23]. Interestingly, the previous study showed that the C/N ratio of optimized medium was lower than that of nonoptimized medium, and this ratio might be suitable for accumulation of active antifungal substances [23]. Nonetheless, the abovementioned interactions could be even more complex; therefore, the RSM was applied for further medium optimization. ese plots (Figures 3(a) and  3(b)) represented that the optimal sucrose concentration was about 3.2 g/L. However, the relatively circular nature of the contour curves suggested that sucrose (Figure 3(b)) had no effect on minerals and showed that the optimal concentration of yeast extract was around 24.0 g/L. Also, the optimal value of minerals (Figures 3(b) and 3(c)) was around 8.2 g/L.
3.6. Validation of the Experimental Model. By applying the regression analysis in (3), the prediction of optimum levels of sucrose, yeast extract, and mineral (NaCl and K 2 HPO 4 ) could be achieved. e optimal values of evaluated components in coded units were as follows: X 1 � 0.464, X 2 � 0.647, and X 3 � 0.313 with the corresponding Y � 98.62%, and their actual values were 3.24 (g/L) sucrose, 23.71 (g/L) yeast extract, and 8.19 (g/L) mineral (NaCl :K 2 HPO 4 at 2 : 1). Consequently, the recipe of medium optimized by statistical approach was then evaluated by shake flask experiments. e maximum experimental antifungal activity was 97.95 ± 2.63%, which was similar to that of the predicted value ( Figure 4). erefore, this developed model was considered to be precise and reliable for predicting the production of antifungal activity by X. stockiae PB09. As the result of optimization, the antifungal activity could be enhanced by 27.22%, when compared to the original TSB medium (70.73 ± 1.48%) (Figure 4).
By applying this optimal medium for cultivation of X. stockiae PB09, the significance of linear and quadratic effects of sucrose, yeast extract, and mineral (NaCl and K 2 HPO 4 ) was found. is may suggest that sucrose, yeast extract, and mineral (NaCl and K 2 HPO 4 ) had direct influence on antifungal activity. e previous study has suggested that the production of microbial secondary metabolites could be significantly affected by even the small manipulations in the culture medium composition [32]. e RSM has been applied as the statistical tool for the studies of the influence of medium constituents, which results in the increase of antibiotic production in several antibiotic discovery programs. For instance, Wang and Liu [23] applied the RSM approach for optimizing the medium for Paenibacillus sp. to produce antifungal active substance and reported the 3.05-fold increase in antifungal activity when compared with the basal medium. Moreover, Zhang et al. [28] reported that a 2.8-fold increase of antifungal production by Streptomyces hygroscopicus BS-112z could be accomplished by using the RSM approach.

Effect of the Optimized Medium in Batch Fermentation.
To further determine the possibility of the regression models, the predicted optimal medium composition was evaluated by 5 L scaled fermenter. Figure 5 showed the batch profile of dry cell weight (DCW) and antifungal activity of X. stockiae PB09 cultivated by using the optimal medium in 5 L fermenter. e antifungal activity increased rapidly during the first 42 h and remained stable until 72 h. After 72 h of incubation, X. stockiae PB09 was found to   Figure 5). is was corresponding to the previous studies that Xenorhabdus spp. was capable of producing the antifungal substances which inhibit several fungal plant pathogens, including, Phytophthora nicotianae, Erwinia amylovora [33], Phytophthora capsici [11], Phytophthora cactorum, Fusicladium effusum [12], R. solani, Pythium sp., and F. oxysporum [1].
In addition, the maximum dry cell weight of X. stockiae PB09 at 12.19 ± 0.24 g/L was obtained after 42 h of incubation. erefore, this indicated that the optimized medium could improve the production of antifungal substances by X. stockiae PB09 in the large scale. Remarkably, high levels of antifungal activity and biomass of X. stockiae PB09 could be    International Journal of Microbiology achieved in 5 L fermenter faster than that in the shake flask scale. is may be due to the addition of different neutralizing agents into various voluminal bioreactive systems. Similar results were obtained in the study of Wang and Zhang [34] and Wang et al. [26] that reported the strong influence of aeration and agitation on biomass and antibiotic production by X. nematophila YL001 in fermenter scale.

Conclusions
In this study, the response surface method (RSM) and central composite design (CCD) with the second-order polynomial model were shown to be valuable tools for optimizing the culture medium that maximize the antifungal activity of X. stockiae PB09. e optimal culture could enhance the antifungal activity from 70.73% to 97.95% in the shake flask experiments, which was very close to the predicted level at 98.62%. An overall 27.22% increase in antifungal activity was obtained as compared with TSB medium. Furthermore, X. stockiae PB09 cultured in the 5 L fermenter could effectively inhibit the mycelial growth of Phytophthora sp., R. solani, Pythium sp., and F. oxysporum.
Data Availability e data sets analyzed during the current study are available from the corresponding author on reasonable request.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.