Optimization of Cultural Conditions for Production of Extracellular Polysaccharide by Halomonas xianhensis SUR308 Using Weighted Response Surface Methodology

Optimization of different physico-chemical and cultural parameters such as temperature, pH, incubation time and inoculum dose along with variation of NaCl, glucose and casein hydrolysate concentrations were carried out in the present investigation for the enhancement of extracellular polysaccharide (EPS) production by Halomonas xianhensis SUR308. Weighted response surface methodology (WRSM) was applied to study the interactive effects of these seven significant variables on EPS production by the isolate. A second order polynomial regression model for weighted response of the growth (O. D. at 540 nm) and production of EPS (g/L) was used to analyze the experimental data following the analysis of variance (ANOVA). Such analysis showed that the model was very significant (p < 0.05). Interaction among glucose, NaCl and pH was found to be most effective for growth and EPS production by H. xianhensis SUR308. The estimated Original Research Article Biswas et al.; JABB, 8(3): 1-11, 2016; Article no.JABB.27839 2 optimum conditions of these variables obtained by using desirability function were 3.49% glucose, 2.5% NaCl and pH 6.8. Under such conditions the isolate produced 7.8 g/L of EPS which justify the predicted value (7.45 g/L) of EPS production in WRSM.


INTRODUCTION
Media formulation and process optimization are key steps in bioprocessing employing a rationally selected microbial strain. The influence of both nutritional components as well as environmental factors on bacterial growth and production of extracellular polysaccharides (EPS) needs to be explored for proper understanding of the fermentation process and also to maximize the yield of product. As early as 1985, Margaritas and Pace [1] have pointed out that successful designing of the fermentation process involves selection and optimization of media components, fermentation conditions, designing of fermenter as well as developing superior strains by mutation. Medium optimization by employing the "one-factor-at-a-time" method involves changing one independent variable while keeping all the others at a particular level [2]. This singledimensional approach is laborious and timeconsuming, especially for a large number of variables, and does not guarantee the determination of optimal conditions [3][4]. Such drawback of the one-factor-at-a-time method could be solved by statistical optimization techniques [5]. Factorial design and response surface methodology (RSM) are important statistical methods for optimization of number of variables by following only a few experimental trials [6].
Response surface methodology is a collection of statistical techniques for designing experiments, building models, evaluating the effects of a number of factors and searching for the optimum conditions for process optimization with higher efficiency and simplicity [7]. It is an ideal method to study and quantify the individual and combined effect of different parameters [8]. This statistical method utilizes quantitative data from appropriate experimentation to determine and simultaneously solve multivariate equations.
Currently, interest in finding the optimal experimental conditions for more than one response by using response surface methodology (RSM) is frequent in studies employing product quality improving techniques.
Often in industries, it is difficult to analyze the results obtained from RSM when several dependent variables (responses) of interest, especially of opposing nature, are involved. This fact makes optimization procedure a challenging issue in industrial processes. The analysis of data from a multi-response experiment requires careful consideration of the multivariate nature of the data. Interrelationships that may exist among the responses can render univariate investigation meaningless.
Several publications have presented approaches addressing multiple quality characteristics but very few published papers have focused primarily on the existence of correlation. If we desire to optimize several correlated responses simultaneously, we would end up in obtaining futile separate individual optima. Optimal condition for one response may be far from real optimum or even physically impractical for the others or may produce unsatisfactory results for the remaining responses. Thus, if correlations among quality characteristics are ignored, the designers may miss finding design variable settings that simultaneously improve the quality of all the responses, which in-turn could lead to an unrealistic solution causing model instability, over-fitting and errors of prediction.
Since the growth and production of EPS are highly correlated to each other, then we can combine these two responses into single response giving appropriate weights, called weighted response and we can use this weighted response as a representative of these responses. Then we analyze the data of weighted response using RSM. This process is called weighted response surface methodology (WRSM). Taking into account all the information mentioned above, it seemed prudent to investigate the performance of the combined approach modeling (WRSM) along with desirability function. The application of this new hybrid method is likely to improve the predictive accuracy of the model over the individual methods.
Response surface methodology has been successfully applied in the optimization of media composition [9-10] and fermentation processes [11][12] of production of extracellular polysaccharides using microorganisms. Production of extracellular polysaccharides by microorganisms from hypersaline sources is mainly limited in halophilic bacteria and archea belonging to the genera Halomonas, Alteromonas, Salipiger, Haloferax and Haloarcula [13][14]. Optimization of conditions for EPS production by these strains was mostly accomplished by one factor at a time. Very recently, statistical experimental designs including factorial design, RSM have been implied in optimizing the cultural conditions for EPS production and found more reliable in predicting the substrate concentration with less accidental errors than the classical experiments [15] and maximum EPS production by bacterial strains through fermentation process [16].
During the survey of diversity and EPS production by halophilic bacteria from multi-pond solar salterns, a potent halophilic bacterial strain identified as Halomonas xianhensis SUR308 (Genbank Accession No. KJ933394) has been isolated [17][18] and the physico-chemical properties of the purified extracellular polysaccharide produced by the strain have been evaluated [19]. Here, we report the optimization of cultural conditions for production of extracellular polysaccharide by H. xianhensis SUR308 under batch culture using weighted surface response methodology.

Extraction and Quantification of EPS
The EPS produced by the growing culture was isolated and quantified according to the method as described by Quesada et al. [21]. To extract the soluble EPS, the culture after definite period of growth was centrifuged at 12000×g for 30 min, three volumes of chilled ethanol was added to the supernatant and kept overnight at 4°C for complete precipitation of the EPS. The precipitate was then collected by centrifugation at 12000×g for 30 min. The cell bound EPS was extracted by treating the cell mass with hot normal saline for 10 min under vigorous shaking. Following this the cell mass was separated by centrifugation at 10000×g for 15 min. Subsequently the EPS in the supernatant was recovered as per the method of soluble EPS extraction described above. The soluble and cell bound EPS were pooled, dissolved in known volume of water and quantified for total carbohydrate.
The carbohydrate content was determined by Dubois method [22]. To 1 ml of EPS solution, 0.5 ml of 5% phenol and 3.5 ml of concentrated sulfuric acid was added and incubated at 50°C for 20 min. Absorbance was recorded at 490 nm and the concentration was determined from the calibration curve prepared in the same method using glucose as standard.

Weighted Response Surface Methodology (WRSM)
Weighted response surface methodology (WRSM) was used to determine the optimum concentration of the variables for enhancement of EPS production. Average value and standard deviation were computed for each experimental run. Systematic study of the relative contribution of the seven independent variables, namely NaCl (%), glucose (%), casein hydrolysate (%), inoculum (%), aeration (CVF ratio), temperature (°C), pH (X 1 , X 2 , X 3 , X 4 , X 5 , X 6 and X 7 , respectively), at -2.83, -1, 0, 1, and 2.83 levels (values) ( Table 1) were followed for optimization of EPS production. The growth, O. D. at 540 nm (Y 1 ) and production of EPS, g/L (Y 2 ) was chosen as the dependent variables. As the analysis of data from a multi-response experiment requires substantial inclusion of almost the whole data set or every evaluated response, the response variables should not be investigated individually and independently of one another. Interrelationships that may exist among the responses can render such univariate investigation meaningless. Ribeiro et al. [23] stated that if the Pearson correlation coefficient (r) between two variables in a multiple response system is ≥0.6, it can be considered that the corresponding variables have good degree of linear association between them. Accordingly, a strong linkage was observed between Y 1 and Y 2 (Pearson correlation coefficient, r=0.87128).
Here it was also observed that correlation between (Y 1 +Y 2 )/2 and Y 1 and between (Y 1 +Y 2 )/2 and Y 2 are very high (i.e. 0.97807 and 0.95439 respectively). Thus (Y 1 +Y 2 )/2(=W, say) is taken as a weighted response of Y 1 and Y 2 and we use this weighted response as dependent variable.
In the present investigation, RSM is being selected as it is the most currently used methodology in optimization procedures, especially when the experiment is intended to find improved or optimal process settings. Data were approximated to a second-order polynomial equation and analysis of variance (ANOVA) was generated. The significances of polynomial relations were examined statistically at a probability (p) of 0.05. Statistical analysis was performed in MINITAB (version 16, Minitab Inc., US) software. Weighted response is fitted to a polynomial regression equation with appropriate degree. Desirability function is used for judging the predicted weighted response under optimum condition and for determining the optimum values of the causal factors (independent variables) that maximize desirability (D).
The estimated response surface model for weighted response obtained from the polynomial regression analysis as a function of the uncoded factor levels are as follows:

Response Analysis
Model adequacy was confirmed from the ANOVA table (Table 1), where the p-value for lack-of-fit (>0.05) and adjusted R 2 value (83.27%) show that the model describes the experimental points adequately. Furthermore, it was observed from probability plot (Fig. 1), that the weighted response (W) had the normal distribution. and show that the model describes the experimental points adequately. Furthermore, it was observed from probability plot (Fig. 1), that the ted response (W) had the normal The response surface plot for the weighted response scores as a function of glucose (%, w/v) and pH is shown in Fig. 2a, which indicates that the weighted response (W) are very sensitive to changes in glucose and pH of the medium. The increase in percentag and pH increased the weighted response significantly. Similarly, increase in glucose and NaCl increased the weighted response up to a particular point followed by a decrease (Fig. 2b), while increase in NaCl and pH increased the weighted response up to a particular point followed by a decrease (Fig. 2c).

; Article no.JABB.27839
The response surface plot for the weighted response scores as a function of glucose (%, w/v) and pH is shown in Fig. 2a, which indicates that the weighted response (W) are very sensitive to changes in glucose and pH of the medium. The increase in percentage of glucose and pH increased the weighted response significantly. Similarly, increase in glucose and NaCl increased the weighted response up to a particular point followed by a decrease (Fig. 2b), while increase in NaCl and pH increased the nse up to a particular point

Determination of Optimum Condition by Desirability Function
The desirability function approach is one of the most widely used methods in industry for the optimization of multiple response processes. It is based on the idea that the "quality" of a product or process that has multiple quality characteristics, with one of them outside of some "desired" limits, is completely unacceptable. The method finds operating conditions that provide the "most desirable" response values. This function transforms the estimated response variable (in this case, W ), into a desirability value d. The values of d vary in the interval 0 ≤ ≤ 1 (Eq. (1)), increasing as the desirability of the corresponding response increases.
1; > Thus, in order to optimize W scores, the value of d is calculated [Eq. (1)], that was prioritized by specifying the goals and boundaries (maximum and minimum values) of W score obtained within the experimental region. Based on the proposed upper and lower bounds, the predicted W scores (= 9.25) and its desirability (= 1.000) at optimal condition is presented in Fig. 3. The behaviour of the predicted W was generated from the optimized factor of 2.5% of NaCl, 3.49% of glucose, 0.5% of casein hydrolysate, 7.88% of inoculum, 1:5 of CVF ratio (aeration) at 30°C and pH 6.8 (Fig. 3) (in order to make these parameters feasible in experimental runs, these observed optimum parameters were drawn to the nearest round figures [24]. The observed experimental values (mean of 30 measurements) were compared to the predicted values, for verifying the predicted factor settings and to check whether or not the optimal conditions derived from the experiment actually result in an improvement in product or process quality. The predicted values could realistically be achieved within a 99% confidence interval of experimental values (Table 2 and Fig. 4a, b and c).
Furthermore, it was observed from probability plot (Fig. 5a, b and c), the weighted response (W), response 1 and response 2 obtained under optimal conditions had the normal distribution. The predicted value of response 2 was much lesser than the experimental readings, but was close to the lower limit of the 99% prediction intervals. Nonetheless, in the confirmatory experiment, the experimental values are in reasonable agreement within the said confidence intervals for these optimized conditions. The closeness between the experimental and predicted values of the quality parameters also indicated the suitability of the corresponding models.

DISCUSSION
Seven variables such as temperature, pH, aeration, NaCl, glucose, casein hydrolysate and inoculum density were selected and values of responses (growth and EPS production) were obtained under different experimental sets as in CCD experimental plan. To determine optimal levels of the test variables for the amount of EPS production by H. xianhensis SUR308, the 3D response surface was described by the regression model (Table 1).  The corresponding analysis of variance (ANOVA) is represented in three dimensional graphical response surface plots (Fig. 2, 3 and 4). The surface plots showed a significant P suggested that the data was good fit with the model and was further confirmed by determination of coefficient (R  The corresponding analysis of variance (ANOVA) represented in three dimensional graphical response surface plots (Fig. 2, 3 and 4). The surface plots showed a significant P-values that suggested that the data was good fit with the model and was further confirmed by determination of coefficient (R 2 ) with R 2 of 95.6% for response 1 and 90.9% for response 2.
The response surface plots have provided a visual interpretation of interaction between variables and help in determining the optimal conditions by revealing the sign  Under experimental set up with these cultural variables the predicted value of EPS production by the strain H. xianhensis SUR308 was 7.45 g/L with 3.49% (w/v) glucose and 2.5% NaCl at pH 6.8, while the actual value was found to be 7.8 g/L under these identical conditions. The higher actual value reveals the higher accuracy of the model. A maximum EPS production of 9 g/L by a marine bacterium with glucose was reported by Nahas et al.

CONCLUSION
The present study demonstrated the interactive effect of different cultural conditions on growth as well as EPS production by the isolate H. xianhensis SUR308 following WRSM. Statistical analysis of the experimental findings revealed that the optimum combination of cultural conditions for EPS production by the isolate comprised of 3.49% glucose, 2.5% NaCl, 0.5% casein hydrolysate, 7.8% initial inoculum, 1:5 CVF at 30°C and pH 6.8. Under these established set of combination, the actual production of EPS (7.8 g/L) was higher than the predicted value (7.45 g/L), which further justified the advantage of RSM methodology in production technology.