Optimization of growth medium and fermentation conditions for improved antibiotic activity of Xenorhabdus nematophila TB using a statistical approach

A sequential optimization approach based on statistical experimental designs was employed to optimize growth medium and fermentation conditions, in order to improve the antibiotic activity of Xenorhabdus nematophila TB. Tryptone soyptone broth (TSB) was chosen as the original medium for optimization. Glucose and peptone were identified as the optimum carbon and nitrogen sources using single factor method. Peptone, fermentation time, initial pH and inoculum volume were identified as the critical factors which highly influenced the antibiotic activity of X. nematophila TB using Plackett– Burman (PB) design. The critical factors were subsequently optimized to locate their optimum domain using steepest ascent method and were further optimized using central composite design (CCD), involving response surface methodology (RSM). The optimum growth medium and fermentation conditions for antibiotic production by X. nematophila TB consisted of glucose 5.00 g/l, peptone 25.6 g/l, NaCl 5.00 g/l, K2HPO4 2.50 g/l, initial pH 7.59, medium volume (100/250 ml), inoculation age (OD600 nm: 2.00), inoculum volume 9.95%, rotary speed 150 rpm, temperature 25°C and fermentation time 54.1 h. An overall of 73.52% increase in antibiotic activity (418.7 U/ml) was obtained as compared with the unoptimized conditions (241.3 U/ml). This study provides a powerful tool to optimize the bioprocess of the antibiotic production by X. nematophila TB.


INTRODUCTION
Xenorhabdus spp. is Gram-negative bacteria belonging to Enterobacteriaceae.The bacterium is mutually associated with the infective juvenile (IJ) nematodes in the genus Steinernema, family Steinernematidae (Thomas and Poinar, 1979).The IJ of the entomopathogenic nematodes carry bacterial symbionts colonizing a special segment of their gut (Martens et al., 2003).The bacteria play the key role in different stage of the symbiosis, acting as a "butcher'' (killer of the insect) (Forst et al., 1997), a ''guard'' (protecting the colonized cadaver by its antimicrobial compounds) (Webster et al., 2002) as well as a ''cook'' (serving food for the growing nematode population within the insect) (Forst and Clarke, 2002).This nematodebacteria complex has been developed commercially as a biological control agent of insect pests (Ehlers, 1996).
The production of secondary metabolites with antibiotic properties is a common characteristic to Xenorhabdus spp.(Webster et al., 2002).More than 30 bioactive secondary metabolites have been reported from the fermentation broths of Xenorhabdus spp., including hydroxystilbenes and indoles (Paul et al., 1981), xenorhabdins (McInerney et al., 1991a), xenocoumacins (McInerney et al., 1991b), xenorxides (Li et al., 1998), nematophin (Li et al., 1997) and benzylineacetone (Ji et al., 2004).These metabolites not only have diverse chemical structures, but also have a wide range of bioactivities with medicinal and agricultural interest, such as antibiotic, antimycotic, insecticidal, nematicidal and antineoplastic (Webster et al., 2002).X. nematophila has been known to produce xenocoumacins, nematophin and benzylineacetone.Xenocoumacins are highly active against Gram-positive bacteria and also exhibit antimycotic activity and potent activity against stress-induced ulcers (McInerney et al., 1991b).The pharmacological activity of xenocoumacins has been speculated to be similar to that of amicoumacins (McInerney et al., 1991b).Nematophin is active against Gram-positive bacteria, especially on Bacillus subtilis and also shows significant antimycotic activity.In comparison, benzylineacetone is active against Gram-negative bacteria.
Antibiotic production by Xenorhabdus spp.differs qualitatively and quantitatively in the strain types and species.Besides, growth medium and fermentation conditions also play very important roles.Cell growth and the accumulation of metabolic products are strongly influenced by growth medium and fermentation conditions such as carbon sources, nitrogen sources, inorganic salts, pH, temperature, agitation and oxygen availability.Xenorhabdus spp.cultivated in 1% peptone water showed no antibiotic activity, however, other media including yeast extract broth and its modifications (Akhurst, 1982;Sundar and Chang, 1993), Luria-Bertani broth (Sundar and Chang, 1993), sea water medium (Paul et al., 1981) and TSB (Li et al., 1997;Ji et al., 2004) have been used successfully for antibiotic production.Limiting the supply of nutrient not only is an effective means to restrict cell growth but also has specific metabolic and regulatory effects (Doull and Vining, 1990).Therefore, to achieve high product yields, one prerequisite is to design a proper production medium.There is usually a relationship between the growth medium and the biosynthesis of antibiotics (Elibol and Mavituna, 1998;Elibol, 2004).Another prerequisite is to design a proper fermentation conditions in an efficient fermentation process, and there is usually a relationship between the fermentation conditions and the biosynthesis of antibiotics (Chen et al., 1996;Li et al., 1997;Yang et al., 2001Yang et al., , 2006)).A much lower antibiotic activity was observed for the X. nematophilus D1 strain fermentation at 35°C than those at 15 -30°C (Chen et al., 1996).Aeration is essential for antibiotic production by Xenorhabdus spp.(Akhurst, 1982;Chen et al., 1996).Li et al. (1997) showed that the concentration of nematophin differed throughout the duration of fermentation and strain of X. nematophila BC1 produced two to five times more nematophin over the whole period than that of strains D1 and ATCC 19061.
Different statistical approaches can be used to optimize the growth medium and fermentation conditions.The conventional method of single factor optimization by Fang et al. 8069 maintaining other factors involved at an unspecified constant level is not only tedious, but also can lead to misinterpretation of results, especially because the interactions between different factors is overlooked (Lotfy et al., 2007).Compared with it, the advantages of statistical approaches are quite obvious: improved product yields, reduced process variability, closer confirmation of the output response to normal, reduced development time and reduced overall costs (Wang et al., 2008).Statistical approaches can be adopted at various phases of an optimization strategy (Kumar and Satyanarayana, 2007).Plackett-Burman (PB) design (Plackett and Burman, 1946) is a well established and widely used statistical technique for screening critical factors (Sharma and Satyanarayana, 2006).The steepest ascent method is well established to locate the optimum domain of the critical factors.However, central composite design (CCD) using response surface methodology (RSM) provides important information regarding the optimum level of each factors along with its interactions with other factors and their effects on product yield (Elibol, 2004).
To date, the statistical experimental designs have not been used in optimization of growth medium and fermentation conditions together for antibiotic production by X. nematophila.The aim of this work was to use sequential optimization approaches as a tool for fermentation engineering to improve the antibiotic activity of X. nematophila TB, which was isolated from its nematode symbiont Steinernema sp.TB obtained from the soil collected from Taibai Mountain, Qinling, P.R.China (Xu et al., 2006;Fang et al., 2008a).In-vitro and in vivo tests showed that the strain exhibited obvious antibacterial and antimycotic activity against many bacterial and fungal species of medical and agricultural importance (Fang et al., 2008a and b).X. nematophila TB is a very potent producer of new secondary metabolites, and has the potential application for disease control in agroforestry industry.The optimization process was performed sequentially: firstly, screen the original medium; secondly, select the optimum carbon and nitrogen sources by single factor method; thirdly, PB design was applied to identify the critical factors; fourthly, steepest ascent method was used to locate the optimum domain of critical factors; and finally, using RSM optimization of the critical factors by CCD.

MATERIALS AND METHODS
Organism and culture conditions X. nematophila TB was isolated from its nematode symbiont Steinernema sp.TB obtained from the soil of Taibai Mountain, Qinling, P.R.China (Xu et al., 2006).Phase I variant of the bacteria was used throughout the study.The strain was maintained on nutrient agar (NA) slants and subcultured monthly.Due to the instability of the phase I under normal conditions, glycerinated stocks of the bacteria frozen at -70°C were frequently used as a starter materia l for cultivation.NBTA medium, NA s upplemented with triphenyltetrazolium chloride 0.040 g/l and bromothymol blue 0.025 g/l, was used to test the variant of the bacteria.Phase I is distinguished from phase II by its adsorption of bromothymol blue to produce a red core colony overlaid by dark blue and surrounded by a clear zone after 2 -3 days of incubation in darkness at 28°C.

Inoculum preparation
A loopful of the phase I of X. nematophila TB growing on an NBTA plate was inoculated into a 250 ml flask containing 100 ml fresh NB (NA without agar) medium, which was adjusted to a final pH of 7.20, and then cultivated in darkness at 28°C on an Eberbach rotar shaker at 150 rpm for 16 -24 h, during which timed the optical density (600 nm) was approximately between 1.50 and 2.00.

Assay of antibiotic activity
Antibiotic activity was measured by agar diffusion plate assay with B. subtilis 08C293 (CCTCC) (Wang et al., 2008;Maxwell et al., 1994).First, 1 ml containing 10 7 -10 8 colonies of B. subtilis was applied to NA plate.After 2 h incubation at 28°C, 100 µl supernatants of the fermentation microfiltrated using a 0.22 µm syringe microfilter were placed on 6 mm disk filters (Whatman 3 mm paper) and air dried.The dried disks were put onto the NA plate and incubated at 28°C for 48 h to determine the relationship between the size of the zones of inhibited bacterial growth and the concentration of the antibiotic.Zones of inhibition were measured from the edge of antibiotic disk to the margin of the zone of inhibition.Antibiotic activity was expressed as units of activity per milliliter the supernatants of fermentation, where 1 U was defined as a 1.0 mm annular clearing around the antibiotic disk.Maxwell et al. (1994) confirmed the assumption that the changes in the size of the zones of inhibition (expressed as units of activity per gram of insect tissue) represented changes in antibiotic concentration (xenocoumacins and nematophin).The antibiotics were extracted from insect larvae killed by X. nematophila by homogenizing the insects in distilled water.The assumption has been used successfully by Wang et al. (2008) to measure the antibiotic activity of X. nematophila.Therefore, the size of the zones of inhibition served as a measure of antibiotic titer of X. nematophila TB.

Measure of cell growth
Cell growth was measured by optical density of the fermentation at 600 nm and biomass concentrations (DCW: g/l) were determined using a calibration curve.The calibration curve was calculated using dilutions of a biomass suspension with known optical density.A fixed volume of the dilutions was centrifuged at 10,000 rpm for 20 min (Himac CR 22G, Japan) and dried the cell pellets at 50°C for 48 h.All the cell pellets were weighed before centrifugation and after drying.Thus, a relationship between biomass concentration (g/l) and optical density can be determined.

Selection of the optimum carbon and nitrogen sources
Single factor method was used to investigate the optimum carbon and nitrogen sources for the antibiotic activity.Based on TSB medium, different carbon sources (glucose, fructose, maltose, sucrose, lactose, starch, glycerol, bran and corn syrup) and nitrogen sources (peptone, soyptone, tryptone, beef extract, cotton cake powder, yeast extract, bean cake powder, urea and fish powder) were used instead of the corresponding carbon and nitrogen source in TSB medium, respectively, while other components were kept constant at their original concentration.All experiments were conducted in triplicate.

Plackett-Burman design
The purpose of this optimization step was to identify which factors of the fermentation medium and conditions have high impact on antibiotic activity of X. nematophila TB.The PB design is very useful for screening the critical factors (Plackett and Burman, 1946).The total number of experiments to be carried out is K + 1, where K is the number of factors.Each factor is represented at two levels, high and low, denoted by (1) and (-1), respectively.All experiments were performed in triplicate and the average values were given.The 'STATISTICA 6.0' software (StatSoft, Inc., Tulsa, USA) was used to analyze the experimental data.

Path of steepest accent (descent)
Based on the results obtained from PB design, the fitted first-order model is (1) Y is the predicted response, β0, βi are the constant coefficients and xi is the coded independent factors.The direction of steepest ascent (descent) is the direction in which Y increases (decreases) most rapidly.One usually takes as the path of steepest ascent (descent) the line through the center of the region of interest and normal to the fitted surface.Thus, the steps along the path are proportional to the regression coefficients βi and started from the center of the PB design.To move away from the first design center along the path of steepest ascent (descent), we moved 3.6, -3.9, 0.2 and 0.8 in directions of peptone, fermentation time, initial pH and inoculation volume, respectively.These new units were determined according to the range of unity level from PB design and estimated coefficient ratio from the first-order model.

Central composite design (CCD) and response surface methodology (RSM)
After the factors were identified by PB design and the optimum vicinity was detected by path of steepest accent (descent), the CCD with five coded levels was conducted in the optimum vicinity to locate the optimum levels of the critical factors and describe the nature of the response surface.According to this design, the total number of treatment combinations is 2 k + 2k + n0 where 'k' is the number of independent factors and 'n0' is the number of repetitions at the center point.For statistical calculation, the factors Xi have been coded as xi according to the following transformation: (2) Where xi is dimensionless coded value of the factors Xi, X0 the value of the Xi at the center point, and δX is the step change.For the four factors, the trial was essentially a full 2 4 factorial design has eight star points and seven replications of center points, resulting in a total number of 31 experiments formulated using statistical software.The behavior of the system was explained by the following quadratic equation: (3) Fang et al. 8071 Where Y is the predicted response, β0 is the intercept term, βi is the linear effect, βii is the squared effect and βij is the interaction effect.The statistical analysis of the model was performed in the form of analysis of variance (ANOVA) including the Fisher's F-test, associated probability P(F), determination coefficient R 2 and correlation coefficient R that measure the goodness of fit of regression model.

Effect of different medium on cell growth and antibiotic activity
The effect of different medium on cell growth and antibiotic activity of X. nematophila TB are shown in Figure 1.TSB medium gave the maximum antibiotic activity (246.7 U/ml), followed by WYH (201.3U/ml), NB+NaCl (196.7 U/ml), LB (183.3U/ml), YSG (131.3U/ml), YS (83.3 U/ml) and NB (71.3 U/ml).In addition, TSB medium also showed the maximum DCW as compared with the others.Among the media tested, TSB medium was found to be the optimum medium for biomass and antibiotic activity of X. nematophila TB, which confirmed the previous reports that TSB medium was able to support relatively high productivity of antibiotics (Maxwell et al., 1994;Ji et al., 2004).Hence, it was employed as the original medium for the following optimizations.

Effect of different carbon and nitrogen sources on antibiotic activity
Carbon and nitrogen sources are the important nutritional components of the medium which highly influence antibiotic activity of X. nematophila (Yang et al., 2001;Yang et al., 2006;Wang et al., 2008).X. nematophila TB produced the maximum antibiotic activity in glucose (256.7 U/ml).
The antibiotic activity in fructose, maltose, sucrose, lactose, starch and glycerol ranged from 210.0 U/ml to 238.3 U/ml.Antibiotic activity with the addition of crude carbon sources (bran) was 185.0 U/ml, and the minimum (83.3U/ ml) in corn syrup (Figure 2a).As for the various nitrogen sources, the maximum antibiotic activity was obtained in peptone (263.3U/ml), followed by 246.7 U/ml in soyptone, 241.3 U/ml in tryptone and 226.7 U/ml in beef extract.Lower antibiotic activities were observed with yeast extract, urea and crude nitrogen sources (cotton cake powder, bean cake powder and fish powder) (Figure 2b).Therefore, glucose and peptone were chosen as the optimum carbon and nitrogen sources for antibiotic activity of X. nematophila TB and for further experiments.Similar results were obtained by Yang et al. (2001) and Wang et al. (2008) in determining the effect of various carbon and nitrogen sources on the antibiotic activity of X. nematophila YL001 and X. nematophila BJ, respectively.However, maltose and glycerol had the strongest effect on the antibiotic activity of Xenorhabdus sp.D43 (Yang et al., 2006).

Identification of the critical factors
A total of 11 independent factors, which may influence the antibiotic activity of X. nematophila TB, including four nutritional factors (glucose, peptone, NaCl and K 2 HPO 4 ) of growth medium and seven factors (initial pH, inoculation age, inoculation volume, medium volume, rotary speed, temperature and fermentation time) of fermentation conditions, were screened in 12 combinations organized according to PB design.The experimental design matrix and corresponding responses are shown in Table 1.Each row of the table represents an experiment involving all the 11 independent factors.The estimates of the factors on the responses, the associated T-values and significant levels are shown in Table 2.The Student's T-test and P-values were used as a tool to check the significance of factors.The larger the magnitude of the T-value and the smaller the P-value, the more significant is the corresponding factors (Elibol, 2004).Out of 11 factors studied, four factors: peptone, fermentation time, initial pH and inoculum volume showed the highest T-Value of 28.7228, -25.7525, 14.8614 and 13.1980, respectively, with the confidence level all above 99%.Therefore, the four most significant factors (P < 0.01) were considered to be the critical factors for the antibiotic activity of X. nematophila TB.A wide variation in antibiotic activity (130.0 to 261.7U/ml) was observed which highlighted the importance of further optimization.The critical factors identified by this design were subsequently optimized by path of steepest accent (descent), and the other factors were at the levels of glucose 5.00 g/l, NaCl 5.00 g/l, K 2 HPO 4 2.50 g/l, inoculation age (OD 600 nm : 2.00), medium volume (100/250 ml), rotary speed 150 rpm and temperature 25°C according to the results of PB design.

Path of steepest accent (descent)
This design started from the center of the PB design and moved along the path in which the peptone concentration, initial pH and inoculum volume increased, and fermentation time decreased.The design and corresponding results are shown in Table 3. Antibiotic activity of Experiment No.4 was the highest (386.7 U/ml), and then decreased following the value changing.The results mean that this point was near the region of the response of the maximum antibiotic activity, and was chosen as the center point of CCD.

Central composite design (CCD) and response surface methodology (RSM)
Based on the results of PB design and path of steepest accent (descent), critical factors and their central point chosen for CCD were: peptone 25.8 g/l, fermentation time 54.3 h, initial pH 7.60 and inoculum volume 9.9%.Each factor was studied at five different levels with all factors taken at a central coded value of zero (Table 4).The CCD design and the corresponding experimental data are shown in Table 5.

Run Factors Antibiotic activity (U/ml)
a Experimental values are the average of triplicates within ±5% standard error.where Y is the predicted response (antibiotic activity), x 1x 4 are the coded values of the independent factors, viz., peptone, fermentation time, initial pH and inoculation volume, respectively.The statistical significance of Equation ( 4) was checked by F-test, and the analysis of variance (ANOVA) for response surface quadratic model is summarized in Table 6.The Fisher variance ratio, Fvalue, is a statistically valid measure of how well the factors describe the variation in the data.The greater the F-value is from unity, the more certain it is that the factors explain adequately the variation in the data, and the estimated factor effects are real (Imandi et al., 2008).The analysis of variance (ANOVA) of the regression model demonstrates that the model is highly significant, as is evident from the F-value (F model = 66.474) and a very low probability value (P model >F = 0.0001).Moreover, the computed F-value (F 0.01 (14, 16) = 66.474) is greater than the tabular F-value (F 0.01 (14, 16)Tabular s = 3.62) at the upper 1% level, indicating that the treatment differences are highly significant.This proves that the model equation expressed in Eq. ( 4) provides a suitable model to describe the response of the experiment pertaining to antibiotic activity.The model was found to be adequate for prediction within the range of factors employed.The goodness of the fit of the model can be confirmed by the determination coefficient R 2 , which provides a measure of how much variability in the observed response values can be explained by the experimental factors and their interactions.The R 2 value is always between 0 and 1.00.The closer the R 2 value is to 1.00, the stronger the model is and the better it predicts the response (Kaushik et al., 2006).In this case, the value of R 2 (0.9831) implies that the total variation of 98.31% for antibiotic activity is attributed to the given independent factors and only 1.69% of the total variations cannot be explained by the model.The value of the adjusted R 2 (0.9683) also improves a high significance of the model.The value of correlation coefficient R (0.9915) indicates a good agreement  between the experimental and predicted values of the antibiotic activity.These measures indicate that the accuracy and general ability of the polynomial model are good and that analysis of the response trends using the model is reasonable.Analyses of the residual diagnostics of the quadratic model of the observed versus predicted yields showed that no significant violations of the model were found in the analysis, with a good correlation of the model with the experimental data obtained.Among the factors tested, peptone, fermentation time and inoculum volume had very significant effect on antibiotic activity (P < 0.01), but initial pH had insignificant effect (P > 0.05).These factors also showed very significant negative quadratic effect, indicating that antibiotic activity increased as the level of these factors increased and decreased as the level of these factors increased above certain values.The interactions between peptonefermentation time, peptone-initial pH and fermentation time-initial pH were very significant (P < 0.01) for interactive terms, and peptone-inoculum volume was significant (P < 0.05).But the interactions between fermentation time-inoculum volumes, and initial pH-inoculum volume were found to be insignificant (P > 0.05).These suggested that peptone, fermentation time, initial pH and inoculum volume had a direct relationship with the antibiotic activity.Since the concentration of peptone has the highest linear coefficient (-3.6042) and quadratic coefficient (-37.8097) with the lowest P values (< 0.01), meaning it is the most important nutrient of the medium and controls the biosynthesis of antibiotic.
The predicted optimum values of peptone, fermentation time, initial pH and inoculum volume were obtained by applying the regression analysis to the Equation (4) and they are peptone 25.60 g/l, fermentation time 54.1 h, initial pH 7.59 and inoculum volume 9.95%, with the corresponding antibiotic activity 401.3 U/ml.

Validation of the experimental model
The validity of the model was tested by carrying out experiments under the predicted optimum conditions.The actual antibiotic activity was 418.7 ± 5.8 U/ml (average of six repeats), which was obviously in close agreement with the model prediction 401.3 U/ml.Therefore, the model was accurate and reliable for predicting the antibiotic activity of X. nematophila TB.After optimization, the antibiotic activity was improved by 73.52% as compared with the control 241.3 ± 3.1 U/ml.Antibiotic production was sustainable in shake flasks (0.5, 1.0 and 2.0 L) and bioreactors (5.0, 20 and 70 L), respectively.A slight decline in antibiotic activity was observed with increasing volume of shake flasks, which could due to the impropermixing of nutrients and rotary problems at higher volumes (Kumar and Satyanarayana, 2007).A slight increase in antibiotic activity was observed with increasing volume of bioreactor, which could be attributed to the better mixing, mass and oxygen transfer in bioreactor (Banik et al., 2007).The statistical experimental designs applied in our investigation have been successfully applied in many recent biotechnological applications (Lotfy et al., 2007;Imandi et al., 2008).However, to the best of our knowledge, no report was obtained on optimization the fermentation medium and conditions together to improve the antibiotic activity of X. nematophila TB using statistical methods.A 35% increase in antibiotic activity of X. nematophila YL001 by optimization of the fermentation medium was obtained by RSM method (Wang et al., 2008).
In this study, an overall 73.52% increase in antibiotic activity was obtained as compared with the control.The chosen method was proved to be a powerful tool for the optimization of the bioprocess for antibiotics production of X. nematophila TB, which was efficient, relatively simple, time and cost saving.Furthermore, the information obtained is considered fundamental and useful for the

Figure 1 .
Figure 1.The effect of different media on DCW (a) and antibiotic activity (b) of X. nematophila TB.

Figure 2 .
Figure 2. The effect of different carbon (a) and nitrogen (b) sources on antibiotic activity of X. nematophila TB.

Table 1 .
The matrix of Plackett-Burman design and the corresponding responses.

Table 2 .
Regression analysis of the data generated by the Plackett-Burman design.

Table 3 .
Experimental results of the path of steepest ascent (descent).

Table 4 .
Coded and actual values of the factors tested in CCD.

Table 5 .
The matrix of the CCD experiment and the corresponding experimental data.
a Experimental values are the average of triplicates within ± 5% standard error.

Table 6 .
Analysis of variance (ANOVA) for the selected quadratic model of antibiotic activity of X. nematophila TB.
development of X. nematophila TB fermentation process for efficient production of antibiotics on a large scale.