Biotechnologically and economically crucial priceless prokaryotes are taken into consideration as natural source for clinically active antibiotics. They hold a prominent position due to their ability to produce secondary metabolites with complex structures and potent biological activities.1 Marine resource has not been explored completely and is a hub for isolation of rare actinomycetes with potency to produce novel bioactive compounds.2 The biosynthesis of the natural products involves the optimization studies for their production that can be engineered with systemic design for suitable physiological conditions and supplementary nutrients. Media constituents and the cultural parameters play important role in antimicrobial agent’s production.3
Traditional optimization includes the study of one-factor-at-a-time which is time consuming and tiresome. Further, the interactive effects of the variables are not known which lead to misleading conclusions, hence it is very critical to establish the optimum conditions that yield highest quantities of the bioactive metabolites with functional properties retained. RSM is a statistical protocol to materialize and systematize the process variables for optimizing the bioactive metabolite production.4,5 RSM was implemented to optimize the process parameters in multi factorial design experiments6 for economizing the number of experimental trials, hence it is extensively applied 7 for theoretical process of optimization8 and interpretation of experiments.9
Bioengineers rely on fermentation kinetic information to design and control microbial processes using mathematical models. Of these, structured models consider cell structure, function and composition aspects; while unstructured models consider only total cellular concentration. However, later ones are proven to be accurate in describing many fermentation processes in a wide range of experimental conditions and media.10-15 To describe the system in better way, evaluation of assumed unstructured models with experimental statistics by considering the whole cell mass to explain the biological system and that in turn elucidates the fermentation profiles of bio product microbial processes.16-17
In this study, the main objectives were isolation and identification of the actinobacterial strain, optimization of different variables that influence the maximal production of bioactive metabolites using RSM and determine kinetic parameters of bioactive metabolite production from S.arabica VSM 25.
MATERIALS AND METHODS
The deep sea sediment samples were collected at different depths of the Bay of Bengal of coastal Andhra Pradesh, India, were transported to the laboratory in sterile bags and air dried at room temperature. The samples were subjected to pre-treatment with sodium dodecyl sulphate (0.05%) and yeast extract (5%).18 One gram of pre-treated sample was suspended in 100 ml of sterile distilled water and serial dilutions were prepared down to 10−4 dilution was plated on Marine Agar medium (containing 0.5% peptic digest of animal tissue, 0.1% yeast extract, 0.01% ferric citrate, 1.94% sodium chloride, 0.88% magnesium chloride, 0.32% sodium sulphate, 0.18% calcium chloride, 0.05% potassium chloride, 0.016% sodium bicarbonate, 0.08% potassium bromide, 0.003% strontium chloride, 0.002% boric acid, 0.0004% sodium silicate, 0.0002% sodium fluorate, 0.0001% ammonium nitrate, 0.0008% disodium phosphate and 2% agar (pH 8)) supplemented with nalidixic acid (50μgml−1) and nystatin (50μgml−1). The inoculated plates were incubated at 30°C for 4 weeks.
The actinomycetes strain VSM-25 isolated from the deep sea marine sediment samples was identified as S.arabica VSM-25 by cultural, morphological, physiological and biochemical studies along with 16S ribosomal DNA (rDNA) analysis (GenBank: KU507598). The strain was maintained on ISP-2 agar medium at 4°C.
RSM is a group of mathematical and statistical evaluation that enables the assessment of the alliance existing between the experimental factors and their response. To achieve significant model, prior information of the process and process variables under study is required. RSM was executed to analyse the optimum levels of the five variables (Time, pH, Temperature, concentrations of Galactose and Peptone) for the production of bioactive metabolites by VSM 25 and its effect against responses was measured in terms of anti-mycotic activity (zone of inhibition in mm), by disk diffusion method. The selection of the variable under study followed the one-factor-at-a-time approach. The coded and uncoded individual variables used in the study are listed (Table 1). Range of variables and central points were selected based on the preliminary experimental information (Supplementary Table 1). The experiments were designed with application of 25 full factorial central composite design for five variables that consists 32 factorial factors, 10 axial factors and 8 replicates. The 8 replicates at centre points used for each categorical variable which includes a total of 50 experiments (Supplementary Table 1), were calculated from the subsequent equation (1).19
A second order polynomial equation was used to express the responses as a characteristic of the independent variables as follows:20
Y = Predicted response, β0 = intercept coefficient, βi = linear coefficient, βij = interaction coefficients, βii = quadratic coefficients, Xi and Xj = coded values of the five additive variables. 3D plots were generated by varying two variables with the experimental range with the other variable constant at the central point. Fractional factorial design with centre points are augmented with a group of “star points” that allows estimation of curvature.
Regression and graphical analysis of the data obtained was executed using Design-Expert 7. By optimizing the second-order polynomial equation and analysing the 3D plots,21 the optimum combinations of individual variables were determined. Statistical model analysis was performed by ANOVA which include Fisher’s F-test, its probability P(F), correlation coefficient R and determination coefficient R2 which measures the goodness of the model fit. Adequate precision have been taken into consideration. The quadratic model for each variable was represented as 3D plots.
Unstructured Kinetic Model
The proliferation of VSM 25 with restricting carbon substrates impacts the antifungal metabolite production. Many researchers have used mathematical and unstructured kinetic models to define the substrate utilization and growth-related production kinetics in a batch system.22
An unstructured kinetic model was developed to study the performance of Nocardiopsis litoralis VSM-8 23 and the similar model is used, here, to predict the kinetics of S.arabica VSM 25 fermentation for its growth, substrate utilization and antifungal metabolite production. The kinetic parameters viz., specific cell growth rate (μmax), d−1, unique bioactive metabolite production rate, d−1, maximum specific cell growth rate, d−1, etc., were estimated.24
Central Composite Design
50 experiments were executed in randomized run order. Supplementary Table 1 explains the experimental design matrix which determines the impact of the five variables on the production of the bioactive metabolite by VSM 25 and its effect on the five responses (Zone of inhibition in mm). The suggested sequential model analyse the sum of squares and lack of fit tests for the best outcome quadratic model, for all the five responses (Table 2).
Response surface analysis of bioactive metabolite production
The highest antimycotic activity of the bioactive metabolite produced by VSM 25 was recorded when the strain was grown in a medium containing 2% galactose, 1% peptone with pH 8 incubated at 30°C for 11 days. The maximum zone of inhibition against all the five responses was found to be 20.8mm (Aspergillus niger), 18.9 mm (Aspergillus flavus), 20 mm (Fusarium oxysporum), 18mm (Fusarium solani) and 19 mm (Candida albicans).
RSM regression analysis was performed to fit the experimental data of each response through second-order polynomial equation. The model is found to be significant with the p-value (<0.0001) for all the five responses. The lower the p value the more significant is the model. Using the experimental data for the bioactive metabolite production, the polynomial model was regressed by only considering significant terms (Table 3).
|−1 (Low)||0 (Middle)||+1 (High)|
|A||Time of Incubation (days)||10||11||12|
|D||Concentration of Galactose (% w/v)||1.0||2.0||3.0|
|E||Concentration of Peptone (% w/v)||0.5||1.0||1.5|
Model adequacy checking
Results of ANOVA used to test the model adequacy at confidence level of 99% were given in Table 4. Coefficient of determination (R2) indicates the proportion of the total variability of the model explained and suggests that model to be a good fit. The R2 value of the model was found to be > 0.9 and reflected to be a very good fit between the observed and the predicted response values (Table 4). The values of the adjusted R2 (<0.9) for model is high which indicate high significance of the model. Signal to noise ratio is measured by adequate precession and values greater than four are desirable. Coefficient variation is the standard deviation expressed in terms of percentage of the mean and should be less than 10%.25 The model showed statistical insignificant Lack-of-fit (Table 4) which indicate the model to be good fit. The experimental values are in close agreement with the predicted values. Which imply that model for each response fits the experimental data adequately.
Variables and interaction effects
3D plots of RSM illustrate the interactive effects of the variables on the five responses with two varying parameters at a fixed value of the third operating parameter (Figure 1, 2, 3, 4, 5). The evaluation of the plots show that all the five parameters recorded a positive effect at the beginning and attained a maximum production and further increase in the range of values and concentration showed a negative effect on the production of the bioactive metabolites by VSM-25.
Optimum operating conditions and corresponding results
Optimum incubation time, pH, temperature and concentration of galactose and peptone was found to be 11 days, 300 C, pH 8 and concentration of galactose 2% and peptone 1% (Figure 6). The recorded zones of inhibition (mm) against the five responses was found to be 20.8 mm for Aspergillus niger, 18.9 mm for Aspergillus flavus, 20 mm for Fusarium oxysporum, 18 mm for Fusarium solani and 19 mm for Candida albicans. All the controlling parameters were found to be within the range, whereas the production of the bioactive metabolite by VSM-25 was found to be high.
Estimation of kinetic parameters
The profiles of S. arabica VSM 25 growth, restraining substrate utilization results obtained from shake flask experiments and model kinetics have been compared in Figure 7. The assessment of experimental versus model predicted inhibition zones of produced bioactive metabolite on media, inoculated with A. niger, A. flavus, F. oxysporum, F. solani and C. albicans strains over the time are shown in Figure 8. A very good fit observed between model predicted and experimental results. In this study, Microsoft Excel Solver 2010 was used to fit the experimental data with unstructured Logistic models, nonlinear regression using least-square method and estimated biokinetic parameters are tabulated in Table 5. It also indicates determination coefficient (R2) values obtained through fitting Logistic (L), Logistic Incorporated Leudeking-Piret (LILP) and Logistic Incorporated Modified Leudeking-Piret (LIMLP) models to the experimental data were found to be high, thus revealing correct precision of the models.
The L model parameters μmax, X0, Xmax and LILP model parameters ɑ and β, were estimated from shake-flask data and a higher ɑ values than β showed that antifungal metabolite production by VSM 25 is more growth related than non-growth associated in shake flask. The simulated LIMLP model parameters, γ and η, are also in desirable agreement with the experimental data, confirms the chosen model is more appropriate to predict the substrate consumption pattern in bioactive metabolite production by VSM 25.26 Further, Table 6 shows similarity between experimental and model predicted zones of inhibition from agar diffusion tests. Hence, the unstructured models provided a better approximation of kinetic profiles of bioactive metabolite production by S.arabica VSM 25 in submerged shake flask fermentations.
RSM based experimental approach with central point was used to determine the significant parameters and their interactions that influence the optimization process for the antifungal metabolite production by S.arabica VSM 25. All the 50 experiments were accomplished in randomized run order to determine the interactive effects of process variables, their effect against the responses (Inhibition of the growth of the pathogenic fungi was represented in mm). The optimization of bioactive metabolite production parameters was generated and the predicted values of the yield were well consistent with experimental values. The statistical significance of the model is determined by the p-value and the p value of all the five response was found to be <0.0001.27 The model is said to be highly significant if the value of p is < 0.05 28. Regression analysis of ANOVA showed a good fit of the experimental data of each response through second order polynomial equation. Coefficient determination (R2) indicates the proportion of the total variability of the model explained and suggested that the model is a good fit. The suggested R2 value for a good fit model should be close to 1 or at least 0.80.29 The value (R2) of the 5 responses was > 0.9 (Table 4) which indicate that model is a good fit. But R2 increases with factors added which are not significant.30 Hence adjusted R2 is used for model adequacy, since adjusted R2 adjusts for the models’ size.25 The insignificant factors added to the model are decreased by the adjusted R2.30 The adjusted (R2) value of all the responses was > 0.9 which indicate the model to be a very good fit. Signal to noise ratio is measured by adequate precession and values greater than four are desirable for the model to be significant. Coefficient variation (CV) is the standard deviation expressed in terms of percentage of the mean and should be less than 10%.25 Adequate precession and Coefficient variation of the five responses in the present study was found to be greater than four and less than 10%.
3D plots obtained showed the interactive effects of the process variables on the bioactive metabolite production and its effect against the five responses. All the five variables studied had significant impact on the bioactive metabolite production by VSM 25.25 The optimum values of time of incubation, pH, temperature and concentration of galactose and peptone was found to be 11 days, 30°C, pH 8 and concentration of galactose 2% and peptone 1%. The recorded zones of inhibition (mm) against the five responses included 20.8 mm for Aspergillus niger, 18.9 mm for Aspergillus flavus, 20 mm for Fusarium oxysporum, 18 mm for Fusarium solani and 19 mm for Candida albicans by the bioactive metabolite produced by VSM-25.
The estimated kinetic parameters for the S.arabica VSM 25 growth limiting substrate utilization and bioactive metabolite production (in terms of inhibition zones measured for 5 fungal pathogens) showed good regression squares with good fit. Therefore, the unstructured models furnished a better approximation of kinetic profiles of bioactive metabolite production by S.arabica VSM 25 in submerged shake flask fermentations. Subsequently, over all findings carry the insights of the crucial information for large scale production of the antimycotic natural metabolites from novel VSM-25 strain that may be used to find new compounds for the clinical utility.
A novel actinomycete VSM-25 was able to grow and synthesize the bioactive metabolite under a wide range of experimental conditions. RSM based experimental study determine the significant parameters and their interactions affecting the optimization process. Optimization of bioactive metabolite production parameters were designed by CCD of RSM, and the predicted values are in agreement with experimental values. High values of R2, adjusted R2, predicted R2 and adequate precession along with low values of the coefficient of variation concluded that the models for responses fit the experimental data adequately. The unstructured models furnished a better approximation of kinetic profiles of bioactive metabolite production by S. arabica VSM 25 in submerged shake flask fermentations.