Enhanced production of intracellular dextran dextrinase from Gluconobacter oxydans using statistical experimental methods

Optimization of the fermentation medium for DDase production by Gluconaobacter oxydans M5 was carried out in the shake flasks using two kinds of statistical methods. Four variables, namely glucose, tryptone, yeast extract and sodium chloride, were found to influence DDase production significantly by the Plackett-Burman screening. A four-factor five-level central composite design (CCD) was chosen to explain the combined effects of the four medium constituents. The optimum medium consisted of glucose (17.670 g/L), maltobiose (30 g/L), tryptone (12.198 g/L), yeast extract (13.528 g/L), ammonium nitrate (15 g/L), copper sulfate (0.01 g/L), zinc sulfate (0.01 g/L), and sodium chloride (0.009 g/L); the initial pH 6.0 was set prior to sterilization. The DDase yield obtained from optimized medium increased by 17-fold (0.238 U/mL) or so. Under these optimal conditions, the experimental values agreed with the predicted values, indicating that the chosen method of optimization of medium composition was efficient, relatively simple, time reducing and material saving.


Abbreviations
The difference in structure between Gluconobacter and Leuconostoc dextran had been further studied and G. dextran could be used as a dietary fibre (Yamamoto et al., 1993b).Furthermore, G. dextran was used not only as a fat replacer in acceptable low-fat foods but also for taste improvement of the bitter-sweet compound stevioside (Sims et al., 2001;Yamamoto et al., 1994).What's more, G. dextran displayed shear-thinning flow behavior (Naessens, 2003).According to the former research, the novel polysaccharide displayed lower viscosity than L. mesenteroides dextran of similar molecular weight as a consequence of its higher degree of branching, and it might be suitable for certain food applications not asso-ciated with thickening functionality, such as a source of dietary fibre or a low-calorie bulking agent for sweeteners (Naessens et al., 2005).Above all, G. oxydans DDase could promise alternatives to L. mesenteroides DSase as biocatalysts for the synthesis of dextran and oligodextrans.But the enzyme system had been studied far less extensively than the glucosyltransferase of L. mesenteroides.
In our group, G. oxydans have been studied for about 20 years in the applications of incomplete oxidation, such as synthesizing L-ribulose, D-tagatose, miglitol and chiral aldehydes (Gao et al., 2006;Gao and Wei, 2006;Yang et al., 2008).In the process of these studies, we have found the strain also has the potential ability of interconverting between maltodextrins and dextran.It is known to all that changing environmental factors, such as the medium composition and the concentration of nutrient, can lead to an increase of DDase production.The development of an economically productive medium requires selecting carbon, nitrogen, ion and trace element sources.And there were two ways by which the problem of medium component limitations may be addressed: classical and statistical (Mao et al., 2007).As for the classical experimental design, there are several limitations towards complete optimization.Moreover, medium optimization by single dimensional search needs a great deal of experiments to determine optimum levels, which are unreliable.Furthermore, it was hard to evaluate the relative significance and the presence of complex interactions of several affecting factors (Kammoun et al., 2008;Mao et al., 2007;Tan et al., 2010).Compared with the classical method, the statistical experimental design in the fermentation medium optimization exhibits its advantages, including more advanced results with less process variability, closer confirmation, less development time and less overall costs.
In this study, medium optimization for production of intracellular DDase by G. oxydans was reported to make it clear that fermentation factors influenced the DDase yield under statistical experimental design.

Microorganism
The strain G. oxydans M5 was stored as 1 mL aliquots in 20% glycerol at -72°C.The frozen cultures were plated periodically to control their viability.

Medium and growth conditions
The G. oxydans grew in a medium containing 80 g of D-sorbitol, 20 g of yeast extract, 2 g of KH2PO4 and 0.5 g of MgSO4•7H2O in 1 L of deionized water.The initial pH was set at 6.0 prior to sterilization.
The organism was incubated at 30°C at 200 rpm in a rotary shaker for 24 h.After 24 h cultivation, 5 mL of the seed culture was used to inoculate 50 mL of production medium in a 250 mL glass flask and this culture was grown under the same conditions as the Mao et al. seed culture.In the optimization process, the amounts of every component were changing in different experimental processes.All the experiments were carried out in three parallel tests.

DDase assay
The extraction method of G. oxydans DDase was as described in the former research by Yamamoto et al. (1992).The G. oxydans DDase activity was determined in accordance with the method described by Naessens et al. with some modifications (Naessens et al., 2005).It was based on the release of nitrophenol from pnitrophenyl-α-D-glucopyranoside (NPG), mimicking G2 utilization by the transglucosidase, which was developed.The DDase specificity of the test was proven via an NPG zymogram.All results representted an average of three separate determinations.One unit of DDase activity was defined as the amount of enzyme which liberates 1 µmol of nitrophenol from NPG per min under the standard assay conditions.

Screening for the important factors
For screening, the Plackett-Burman design was applied with which 12 runs would enable us to estimate all main effects for up to factors.Table 1 showed the Plackett-Burman design used in this study (Plackett-Burman, 1946).Since there were only 9 factors in this case, less than the maximum allowable factors, the remained factors became "dummy factors" which would be used to estimate errors.The signs -1 and +1 represented the lower and higher levels of the corresponding components under investigation.In this experiment, the higher levels of the components were chosen to equal 1.25 times of their lower levels (Table 2).

Central composite design (CCD) method
Sodium chloride, tryptone, glucose and yeast extract were four effective nutrients in the Plackett-Burman design.These variables were selected to find the optimum condition for higher DDase production using central composite design (CCD).In this case, a 2 full factorial central composite design for four independent variables each at five levels with eight star points and six replicates at the centre points was employed to fit a second order polynomial model, in which 30 experiments were required in this procedure (Box et al., 1978, Box, 1951).Each variable was designed at five levels (−2, −1, 0, 1, 2) and the lowest and the highest concentration were: sodium chloride, 0 and 0.02 g/L; tryptone, 2 and 18 g/L; glucose, and 55 g/L; yeast extract, 2 and 18 g/L , respectively.Table showed the CCD design for the values and the observed response for the DDase activity.The central values (zero level) chosen for experimental design were same as the lower level (designated -1) in the Plackett-Burman design.
For statistical calculation, the relationship between the coded and actual values is described as the following equation: where Xi is the coded value of the ith variable, Ui is the actual value of the ith variable, U 0 i is the actual value of the ith variable at the center point and ∆Ui is the step change of variable.
The response variable (DDase activity) suitable to a quadratic

S/N
where Y is the measured response, Xi, Xj are input variables which influence the response variable Y; b0 is a constant; bi is the ith linear coefficient; bii is the squared coefficient and bij is the ijth cross-product coefficient.
The 'Design Expert' software (version 7.1.6,Stat-Ease, Inc., Minneapolis, USA) was utilized to analyze the significance of experimental results.The central composite design (CCD) employed to fit a second order polynomial model, had four variables, of which everyone was at five levels with eight star points and six replicates at the centre points and the result indicated that 30 experiments were required for this procedure (Cochran, 1957).

Screening for the important factors
The effects of different carbon sources on the growth of G. oxydans and DDase activity were studied in the media containing yeast extract (2%) as the nitrogen source.The results showed that glucose and maltobiose were better than other carbon sources for the DDase production; the effects of nitrogen sources on the DDase production by G. oxydans were investigated using glucose (2%) as carbon source and it exhibited that tryptone was the best and yeast extract and ammonium nitrate was inferior (data not shown).Then the DDase yield was determined for fermentation containing glucose (2%) and tryptone (1%) as carbon and nitrogen sources and one of the following ions (0.001, 0.01 and 0.1%) were used as additives: KCl, NaCl, High yield of DDase was observed when CuSO 4 or ZnSO 4 or NaCl was used as trace mineral "cocktails" for fermentation (data not shown).According to above study, which was under the premise of keeping the other factors constant, we optimized the composition of the production medium as follows: glucose 30 g/L, maltobiose 30 g/L, tryptone 10 g/ L, yeast extract 10 g/L, ammonium nitrate 15 g/L, copper sulfate 0.01 g/L, zinc sulfate 0.01 g/L, sodium chloride 0.01 g/L and the initial pH 6.0 was set prior to sterilization.The culture was incubated at 30 °C at 200 rpm in a rotary shaker for 62 h.After above optimization, the Plackett-Burman technique was used to select the most important factors which would influence the DDase yield.In this section, twelve experiments were minimum runs to screen the importance of nine complications, namely glucose, maltobiose, tryptone, yeast extract, ammonium nitrate, copper sulfate, zinc sulfate, sodium chloride and pH.The Model F-value of 20.84 implied the model was significant.There was only a 4.66% chance that a "Model F-Value" this large could occur due to noise.Values of "Prob > F" less than 0.0500 indicated model terms were significant.The experimental date analysis showed that four variables, namely glucose, tryptone, yeast extract and sodium chloride out of nine variables influenced the DDase yield significantly.

Central composite design (CCD) method
With the central composite design method, the experiments with different combination of glucose, tryptone, yeast extract and sodium chloride were performed.A four-factor five-level central composite design was used based on the "Design Expert" software (version 7.1.6)and experimental data is presented in Table 3.
By statistical analysis, a mathematical model was obtained to describe the relationship between the DDase production (Y) and the test variables in coded factors (concentration of glucose, tryptone, yeast extract and sodium chloride, respectively): The coefficient values of the model and factors were calculated by Design Expert Software and their values are shown in Tables 4 and 5.The results of the second order response surface model fitting in the form of ANOVA were given in Table 4.
In Table 5, the Model F-value of 13.82 implied that the model was significant.There was only a 0.01% chance that a "Model F-Value" this large could occur due to noise.The "Lack of Fit F-value" of 48.97 implied the Lack of Fit was significant.The well fit of the model was checked by determination coefficient (R 2 ).In this case, the value of the determination coefficient (R 2 = 0.9280) indicated that it was reasonable to use the regression model to analyze the tendency (Akhnazarove, 1982;Khuri and Cornell, 1987).Table 5 shows the significance of each coefficient determined by F-value and P-value.Values of "Prob > F" less than 0.0500 indicate model terms were significant.In this case X 2 , X 3 , X 4 , X 3 X 4 , X 1 2 , X 2 2 , X 3 2 , X 4 2 were significant model terms.Figures 1 -6 shows the response surface for the variation in the production of DDase, from which the values of DDase activity for different concentrations of the variables could be predicted.Each contour curve represented an infinite number of combinations of two test variables with another maintained at their zero level.Figures 1 -3 show that increasing the sodium chloride concentration from 0.003 to 0.01 g/L increased DDase production.The maximum production was obtained with the concentration of sodium chloride at about 0.01 g/L.The two nitrogen sources, tryptone and yeast extract, were important to increase the DDase yield.From the response surface contour plot, the concentrations of tryptone and yeast extract were around 12 -14 g/L, respectively, to result in the maximum DDase yield.In this study, glucose was also a significance parameter.Figures 2, 4 and 6 have shown that the maximum DDase activity appeared at the concentration of 17.5 -18 g/L.
According to the optimized mathematical model, the optimal values of the test variables in coded unit were as follows: X 1 = -0.156,X 2 = 0.549, X 3 = -0.986,X 4 = 0.882     with the corresponding maximum DDase production Y = 0.209 U/mL (a possible variation is between 0.17 and 0.25 mg/L) in the confidence range of 95%.Putting the respective values of X i in Equation ( 1), the concentration of sodium chloride, tryptone, glucose and yeast extract were at 0.009, 12.198, 17.670 and 13.528 g/L, respectively.It was clear that the optimal values from response surface plots were almost consistent with those from optimized mathematical equation.
In order to verify the predicted results, the experiment was performed with the optimized medium and the maximum DDase production was found to be 0.238 U/mL, which was obvious in close agreement with the model prediction.After optimization, compared with the minimum medium, the production of DDase was enhanced about 17 fold, experimentally (Table 6).To our best knowledge, the DDase production in this research is higher than any other reported results (Naessens et al., 2005).

Conclusions
Plackett-Burman screening and response surface methodology (RSM) for medium optimization were employed for the DDase production from G. oxydance.The method used was effective in the screening for nutritional requirements in a limited number of experiments.It enabled us to screen a large number of experimental factors and was useful to determine the optimum levels of medium components concentration that significantly influenced the DDase production in a perfect way.The chosen method of optimization of medium composition was efficient, relatively simple, time reducing and material saving.The final composition of definite medium after the optimization was as follows: glucose 17.670 g/L, maltobiose 30 g/L, tryptone 12.198 g/L, yeast extract 13.528 g/L, ammonium nitrate 15 g/L, copper sulfate 0.01 g/L, zinc sulfate 0.01 g/L, sodium chloride 0.009 g/L, respectively, and the initial pH was set at 6.0.The optimization of the medium resulted in an about 17-fold (0.238 U/mL) increase in DDase yield.This work would provide a stable foundation for the industrialization of DDase.

Figure 1 .
Figure 1.Effect of sodium chloride and trypotone concentration on the DDase production by G. oxydans.Other variables are held at zero level.

Figure 2 .
Figure 2. Effect of sodium chloride and glucose concentration on the DDase production by G. oxydans.Other variables are held at zero level.

Figure 3 .
Figure 3.Effect of sodium chloride and yeast extract concentration on the DDase production by G. oxydans.Other variables are held at zero level.

Figure 4 .
Figure 4. Effect of trypotone and glucose concentration on the DDase production by G. oxydans.Other variables are held at zero level.

Figure 5 .
Figure 5.Effect of trypotone and yeast extract concentration on the DDase production by G. oxydans.Other variables are held at zero level.

Figure 6 .
Figure 6.Effect of glucose and yeast extract concentration on the DDase production by G. oxydans.Other variables are held at zero level.

Table 2 .
Assigned concentrations of variables at different levels in Plackett-Burman design.

Table 3 .
The CCD design for the values in coded and the observed values in response.

Table 4 .
ANOVA for response surface quadratic model.

Table 5 .
The least-squares fit and coefficient estimate.

Table 6 .
Experimental verification of combined effect of optimized medium on the response of DDase production.