OPTIMIZATION OF PROCESS VARIABLES BY RESPONSE SURFACE METHODOLOGY

In the present study optimisation of the growth medium for the production of Cyclodextrin glucanotransferase (CGTase) was carried out using response surface methodology. Four important parameters namely starch, yeast extract, K2HPO4 and MgSO4 concentrations were selected as the independent variables and the enzyme activity (CGTase activity U/mL) was the dependent response variable. Each of these independent variables was studied at five different levels as per central composite design (CCD) in four variables with a total of 28 experimental runs. The optimal calculated values of tested variables for maximal production of CGTase were found to be comprised of: starch, 2.16 %; yeast extract, 0.6 %; K2HPO4, 0.62 %; MgSO4, 0.04 % with a predicted CGTase activity of 150 U/ml. These predicted optimal parameters were tested in the laboratory and the final CGTase activity obtained was very close to the predicted value at 148.2 U/ml.


Introduction
Cyclodextrin glucanotransferase (CGTases; EC 2.4.1.19) is an enzyme which converts starch into the cyclodextrins (CDs). Based on the number of glucose moieties the CDs are classified as α-, β-, and γ-CDs. CDs have the capacity to encapsulate hydrophobic molecules within their hydrophobic cavity, based on this nature it is used in the various industries. Leemhuis et al., (2010), Martin Del Valle (2009), Li et al., (2007) and Biwer et al., (2002) reviewed the numerous applications of the CDs in the pharmaceutical, cosmetics, and food and textile industry. CDs have a hydrophilic outside and hydrophobic inside due to this it is used in the encapsulation of hydrophobic molecules which is particularly advantageous as many drug molecules are poorly soluble in water (Loftsson and Duchene, 2007), or to protect guest molecules from light, heat, or oxidizing conditions (Astray et al., 2009). Cyclodextrins are also used to lower the volatility of odour molecules in perfumes and room refreshers for controlled release of the odour. In the chemical industry, CDs are used in the separation of enantiomers to extract toxic chemicals from waste streams (Martin Del Valle, 2009) and in soil bioremediation (Fava and Ciccotosto, 2002). Various other applications of CDs include the suppression of undesirable (bitter) tastes and the extraction of compounds such as cholesterol from foods (Szente and Szejtli, 2004;Szejtli and Szente, 2005).
The composition and concentration of the medium plays a vital role in the growth and enzymes production by the microorganisms. The optimization of the media components and culture conditions are the primary task in a biological process. The traditional optimization approach used is one-at-a-time optimization. In this method one parameter is optimized by changing it at the same time other factors were maintained at a constant level (Suvarna Laxmi et al., 2008). This method of optimization requires a large number of experiments, it is a tedious process and also consumes a lot of chemicals and resources leads to the process development which is cost ineffective. Apart from this, there is a chance for misconception of results because the interaction effects between different factors are unnoticed (Hymavathi et al., 2009). Response surface methodology (RSM) is a useful tool for studying the effect of several factors influencing the responses by varying them simultaneously and carrying out a limited number of experiments. RSM in concise, is explained as a collection of experimental strategies, mathematical methods and statistical inference for constructing and exploring an approximate functional relationship between a response variable and a set of design variables. Very few authors have reported satisfactory optimization of CGTase production from microbial sources using a statistical approach (Gawande and Patkar, 1999;Rahman et al., 2004;Ibrahim et al., 2005).

Research Article
During a screening program, a CGTase activity producing strain was mutated and identified as Bacillus sp. TPR71HNA6. With the help of Plackett-Burman design (PBD) four significant nutritional parameters which influence the CGTase production were selected. The objective of the present study was to optimize the levels of chosen significant nutritional parameters using central composite design (CCD).
In the preliminary studies and PBD it was observed that the parameters namely starch, yeast extract, K2HPO4, and MgSO4 concentrations were playing a vital role in the CGTase production. These four parameters were further optimized based on the Response surface methodology. RSM has been proved to be a powerful tool for optimization of fermentation parameters by many research groups (Hymavathi et al., 2009). This method has been successfully applied in the optimization of fermentation medium components, conditions for enzymatic production as well as CDs production processes. It allows the calculation of maximum enzyme production based on few sets of experiments in which all the factors are varied within selected range and also to study interactive effects of various process parameters.

Microorganism and Culture Conditions
In the present study a mutated Bacillus sp. TPR71H (GenBank Accession No: FN993946) was used. This culture was stored in nutrient agar slants and subcultured periodically once ever week. The production of CGTase experiments were conducted according to the PBD. The liquid samples are withdrawn and centrifuged at 10,000 rpm for 10min to remove the biomass and other insoluble substrates from the culture. After centrifugation the supernatant liquid was collected and estimated for CGTase activity.

Estimation of CGTase Activity
Enzyme activity was measured by decrease of phenolphthalein colour intensity. Enzyme assay was carried out according to the Kaneko et al., (1987) method. The reaction mixture containing 1mL of 40mg of soluble starch in 0.1M potassium phosphate buffer (pH 6.0) and 0.1mL of the crude enzyme from the culture was incubated in water bath at 60°C for 10min. The reaction was stopped with 3.5mL of 30mM NaOH. Finally, 0.5mL of 0.02% (w/v) phenolphthalein in 5mM Na2CO3 was added and mixed well. After leaving the mixture to stand for 15min at room temperature, the reduction in colour intensity was measured at 550 nm. A blank lacking the enzyme is tested simultaneously with each batch of samples. One unit of enzyme activity was defined as the amount of enzyme that forms 1µgm of β-CD from soluble starch in 1min.

Optimization by Response Surface Methodology (RSM)
Response surface methodology using Central composite design was applied for optimization of CGTase production from mutated Bacillus sp. TPR71HNA6. Four important parameters namely starch (X1), yeast extract (X2), K2HPO4 (X3), and MgSO4 (X4) concentrations were selected as the independent variables and the enzyme activity (CGTase activity U/mL) was the dependent response variable. Each of these independent variables was studied at five different levels as per CCD in four variables with a total of 28 experimental runs. CGTase activity (U/mL) corresponding to the combined effects of four variables was studied in their specified ranges as shown in Table 1. The process variables such as temperature, pH and agitation speed were kept constant throughout the experiment. All the flasks were analysed for CGTase activity at the end of the experiment. The plan of CCD in the coded levels of the four independent variables is shown in Table 2.
For statistical calculations the independent variables were coded as Where Xi is the experimental value of variable; X0 is the midpoint of Xi, δXi is the step change in Xi and xi is the coded value for Xi, i = 1-4.
This response surface methodology allows the modelling of a second order equation that describes the process. CGTase production data was analysed and response surface model given by Eq.
(2) was fitted with multiple regressions through the least squares method.
Where Yi is the predicted response, in the present study CGTase production (Yi) taken as a response, xi xj are input quadratic coefficie

Data Analysis and Interpretation of the Results
The results of the experimental design were analysed and interpreted using the STATISTICA version 7.0 (StatSoft, USA) statistical software. Prediction of optimum fermentation parameters and shape of the curves generated by the model was also done by the same software. Table 3 depicts the results of the 28 runs CCD in four selected variables at five levels for optimization of CGTase production. CGTase production varied markedly in the range of 97-148U/mL with the conditions tested. High CGTase activity was observed in experimental runs with the mid values of the parameters. It was observed from various experimental runs that CGTase production was quite high with higher starch concentration.

Results and Discussion
CGTase activity (U/mL), the response variable was transferred to natural log values in order to stabilize its variance. ANOVA (analysis of variance) was employed for the determination of significant effects of variables for CGTase production. The experimental results suggest that the variables selected for the fermentation process had strong effect on CGTase production. On the basis of these experimental values statistical testing was carried out using the Fisher's 'F'-test and students'-'t'-test. Analysis of variance for CGTase production shows that fitted second order response surface model is highly significant with Ftest = 18.59 (P < 0.0001).
The coefficients for the linear effect of K2HPO4 and MgSO4 were highly significant while starch and yeast extract concentrations were statistically insignificant (Table  4). In the quadratic terms all variables were significant. The starch and yeast extract concentrations were insignificant at linear terms and significant in quadratic terms which indicate that these two factors are highly influential parameters on the CGTase production. With small variation in the concentration of these variables, a significant change in the production could be observed. The interactive effect between trace elements (K2HPO4 and MgSO4) were not significant, all other remaining interactions are significant. The interaction of the starch and K2HPO4 has the highest magnitude (8.125) when compared to the other interactions (  The coefficient of determination R2 for the above predicted Eq. (3) was 95.94. The correlation coefficient (R2 = 0.9594) was indicating that the statistical model can explain 95.94% of the variability in the response. Therefore this equation can be used for predicting the response at any combination of four variables in and around the experimental range. CGTase activity (U/mL) at specific combination of four variables can be predicted by substituting the corresponding coded values in Eq. (3). Figure 1 depicts the correlation between the observed and predicted values. From this figure it was observed that all of the data points are concentrated near the diagonal line, and no scattered points were observed, it indicates that there is a good correlation between the observed and predicted values. The value of the adjusted determination coefficient is close to the R2 value (Adj R2 = 0.9158) is also very high to advocate for a high significance of the model (Box et al., 1978;Cochran and Cox, 1957). If there are many terms in the model and the sample size is not very large, the adjusted R2 may be noticeably smaller than the R2. Here in this case the adjusted R2 value is 0.9158, which is lesser than the R2 value of 0.9594. The Predicted R2 of 0.7318 is in reasonable agreement with the adjusted R2 of 0.9158. At the same time, a relatively lower value of the coefficient of variation (CV = 3.31%) indicates a better precision and reliability of the experiments carried out (Myers and Montgomery, 1995;Khuri and Cornell, 1987).
(3) were prepared using STATISTICA 7.0 software. The surface plot (Fig 2-7) shows the behavioural change with respect to simultaneous change in two variables. Proper choice of fermentation parameters is desirable for maximum enzyme production and surface plots based on well fitted model provides these choices. Surface and contour plots were prepared for six pairs of variables which were having significant interaction effects in maximizing CGTase production at specific hold values.
The behaviour of CGTase production with respect to change in starch and yeast extract concentrations at specific hold values is shown in Fig 2. From the figure it was observed that the contour plot is slightly inclined towards the starch, indicating that the interaction between these two parameters is significant and starch has a high influence on CGTase production. It was observed that starch at 1.8-2.2% (Fig 2-4) and yeast extract at 0.5-0.65% (Fig 2, 5 & 6) concentrations were effective for enzyme production was noticed. Gawande and Patkar 1999 reported that the nature and concentration of the carbon source plays a vital role in the CGTase production. Ai-Noi et al., 2008 andKhairizal et al., 2004 reported that increasing the sago starch concentration increased enzyme production. Gawande and Patkar 1999 also commented that above certain concentration of carbon source, when other nutrients are kept constant, catabolite repression may occur. It was noticed that the CGTase production with starch concentration above 20-30g/L, resulted in low enzyme production by Bacillus sp (Gawande et al., 1998). Generally, a phosphorus source is considered to be necessary for cells for the synthesis of nucleic acids and phospholipids (Madigan et al., 1997). From the Fig. 3, 5 and 7 it was noticed the interaction behaviour of phosphorus with other variables. It was observed that concentration of phosphorus slightly depends on the starch concentration (Fig 3). A lower concentration of phosphorus is preferable for the effective CGTase production. Swinkels, 1985 reported that starch contains the trace metals, in that case when using the starch, lower concentrations of trace elements addition is preferable. It was observed that K2HPO4 concentration at 0.55-0.65% is optimum for CGTase production by the mutated Bacillus sp TPR71HNA6 (Fig 3, 5 & 7). It was noticed that the concentration of MgSO4 in the range of 0.035 -0.05% (Fig  4, 6 & 7). The low concentrations of these salts were needed to increase the production of CGTase. A similar result was reported by Gawande and Patkar 1999 whereby using experimental design, they found that the concentration of mineral salts (magnesium sulphate in their case) at 0.5g/L can increase the CGTase production by Klebsiella pneumoniae AS-22. Validation of the experimental model A repeat fermentation for CGTase production by mutated Bacillus sp. TPR71HNA6 under optimal conditions was carried out for the validation of optimized parameters. The CGTase production under optimized parameters viz. starch 2.16%, yeast extract 0.6%, K2HPO4 0.62% and MgSO4 0.04% yielded CGTase activity of 150U/mL. The CGTase yield so obtained under optimized parameters was even higher than the predicted value (148.2U/mL) by the model. These validation studies indicate that the proposed model was adequate to predict the optimisation of CGTase production from mutated Bacillus sp.TPR71HNA6. Similarly Rahman et al., 2004 andIbrahim et al., 2005 used the statistical optimization techniques for improvement of the CGTase production.