Alkaline proteases are robust enzymes with considerable industrial applications in detergent, leather processing, pharmaceutical, food processing, feed and chemical industries, as well as in silver recovery and waste treatment [1,2]. The harsh conditions required for industrial applications and the high cost of alkaline protease production warrants continued search for microbial sources of alkaline proteases. The cost of enzyme is a major obstacle in the successful application of proteases in industry. Protease yield has been improved by screening for hyperproducing strains and/or by optimization of the fermentation medium. Strain improvement by either conventional mutagenesis or recombinant DNA technology is one approach for improving the production of proteases [3]. Approximately, 40% of the production cost of industrially important enzymes is estimated to derive from the cost of the growth medium [4,5]. Hence, optimization of fermentation conditions is necessary for cost-effective enzyme production. Media components exert great influence on extracellular protease production and the effects vary in different organisms. Therefore, medium components and their concentrations have to be optimized accordingly [6]. The ‘one-variable-at-a-time approach’ (OVAAT) of improving fermentation conditions is the most frequently used operation in biotechnology to obtain maximum cell growth and high yields for the desired metabolic product. This approach is not only time consuming but it also ignores the combined interactions among physicochemical parameters [7]. Optimization by statistical approaches such as Plackett-Burman Design (PBD) and Response Surface Methodology (RSM) are generally preferred over OVAAT the approach for achieving significant improvements in yield and reduction in the production cost [8]. RSM is a widely accepted statistical approach for modelling and analysis of problems in which a response e.g. level of extracellular enzyme production is influenced by several variables such as temperature, pH and media components. Moreover, several low cost agro-industrial by-products such as rice bran [9], corn steep liquor [10], soybean meal [11], feather meal [12], potato peel [13] and green gram husk [14] have been used as media components to reduce the cost of enzyme production. The present study aims to improve the protease production by Bacillus sp. KW2, a strain isolated from waste dumping sites, in a low cost medium through optimization of medium components and their concentrations using statistical approaches.

Isolation of alkaliphilic bacteria

Soil samples collected from waste dumping sites at Langol, Manipur, India were serially diluted (10^-2 to 10^-7) in sterile saline solution. The diluted samples were spread plated on Horikoshi medium I [15] and plates were incubated at 30°C for 72 h. Morphologically distinct strains were subcultured and screened for protease production on Horikoshi medium I supplemented with 1% (w/v) skim milk. The pure cultures were preserved as slants at 4°C and as glycerol suspensions (20%, v/v) at -20°C.

Inoculum preparation

A loopful of bacterial culture was inoculated in Horikoshi-I broth and kept incubated under shaking conditions (30°C, 150 rpm, 24 h). The culture broth was centrifuged (10,000 rpm, 30 minutes). The pellet was collected, washed twice with sterile distilled water (SDW) and then centrifuged. The pellet was then dissolved in 10 ml SDW and the optical density (OD) was measured at 600 nm. The OD was then adjusted to 0.5.

Enzyme production

Protease production by Bacillus sp.KW2 was done in submerged fermentation in glucose yeast extract medium (GYM) containing glucose (1% w/v), yeast extract (0.5%, w/v), MgSO₄ (0.05%, w/v), and KH₂PO₄ (0.5 %, w/v). Na₂CO₃ (0.5%, w/v) was autoclaved separately and added to the media. 2% (v/v) inoculum was added in 50 ml production medium and incubated under shaking conditions (30°C, pH 10.5, 150 rpm, 72 h). The culture broth was centrifuged (8,000 rpm, 15 min) and the supernatant was used for protease assay.

Protease assay

Protease activity was determined by using Hammerstein casein as substrate. 1 g casein was dissolved in 100 ml Glycine-NaOH buffer (50 mM, pH 10). 1 ml substrate was added to 1 ml appropriately diluted crude enzyme and incubated at 50°C in a water bath for 10 min. The reaction was stopped by adding 2 ml TCA (10%, w/v) and the reaction mixture was centrifuged (8,000 rpm, 15 min). A control sample (blank) was also included in which TCA was added before the addition of enzyme. Absorbance was measured at 280 nm against the Protease activity was determined by using Hammerstein casein as substrate. 1 g casein was dissolved in 100 ml Glycine-NaOH buffer (50 mM, pH 10). 1 ml substrate was added to 1 ml appropriately diluted crude enzyme and incubated at 500C in a water bath for 10 min. The reaction was stopped by adding 2 ml TCA (10%, w/v) and the reaction mixture was centrifuged (8,000 rpm, 15 min). A control sample (blank) was also included in which TCA was added before the addition of enzyme. Absorbance was measured at 280 nm against the blank. One unit of enzyme activity was defined as the amount of enzyme that led to an increase in absorbance of 0.01 at 280 nm per min.

Experimental design for optimization of medium components

Medium optimization was carried out by OVAAT and statistical approaches. Various physiological and nutritional parameters of the production medium were first standardized in GYM medium by OVAAT approach. Following this step, signal parameters affecting protease production were identified by PBD and interactions among signal parameters were studied by RSM.

One-variable-at-a-time approach (OVAAT)

To evaluate the influence of temperature on protease production, fermentation was carried out at various temperatures (25°C, 28°C, 30°C, 32°C, 35°C, 37°C, 40°C and 45°C).

Effects of sodium carbonate on protease production were studied by varying sodium carbonate concentration in the GYM medium (0.05, 0.1, 0.2, 0.5, 1 and 2 %, w/v).

To assess the effects of carbon (C) sources on enzyme production, the C source present in GYM medium (glucose, 1%, w/v) was replaced with various C sources (maltose, starch, lactose, fructose, tri-sodium citrate and rice bran; 1%, w/v each). To study the effects of nitrogen (N) sources, yeast extract (0.5%, w/v) was replaced with various organic and inorganic nitrogenous compounds (soybean meal, peptone, tryptone, beef extract, KNO₃ and NH₄Cl, 0.5% w/v each).

Screening of factors affecting protease production by pb design

PBD was used to screen signal factors among seven variables viz. rice bran, soybean meal, KH₂PO₄, MgSO₄, CaCl₂, inoculum size and inoculum age. The maximum and minimum ranges of the factors selected for protease production are given in Table 1. Design Expert 6 software (Stat-Ease) was used to generate a set of 12 experiments (Table 2). Protease production was set up in triplicates in 50 ml of the respective production medium and fermentation was carried out for 72 h (30°C, pH 10.7, 150 rpm). For each experiment, protease production was calculated in terms of protease units/ml (U/ml). The effect of each parameter on protease production was determined by calculating Student’s t test and p-values. The effect of each factor was coded as positive or negative and significance was analyzed according to ‘t’ and ‘p’ values [16].

Table 1: Maximum and minimum ranges of parameters selected for PBD.

Table 2: Experimental design used in PBD.

Optimization of protease production by RSM

After PBD, RSM was used in the final round of medium optimization. Four parameters viz. rice bran, soybean meal, KH₂PO₄ and CaCl₂ were studied at five levels (-α, -1, 0, +1, +α) using the Central Composite Design (CCD). The experimental levels of the four parameters used in RSM in terms of actual and coded forms are listed in Table 3. A set of 30 experimental runs including six centre points were designed using Design Expert 6. Protease production was carried out in triplicates by dispensing 50 ml each of production media in 250 ml Erlenmeyer flasks. The media were inoculated with 5% (v/v, 0.5 OD₆₀₀) inocula and kept incubated in the shaker for 72 h (30° C, pH 10.7, 150 rpm). A multiple regression analysis of the data was carried out to obtain an empirical model that defines the response (protease production) in terms of the independent variables. The model equation is represented as:

Table 3: Experimental ranges of four different variables used in CCD.

Where, Y is the predicted response, β₀ is the intercept, βi is the linear coefficient and β_ij is the interaction coefficient. The model was analyzed using ANOVA, 3D curves, contour and one-factor plots in order to study the interactions among various factors and to determine the optimum concentration of each for maximum protease production. Accuracy of the model was determined by validating the model.

All the above experiments were repeated in triplicates and the final values have been presented as means ± S.D.

Isolation of bacteria

Nine (9) morphologically distinct bacterial strains were recovered. Of these six (6) strains showed proteolytic activity. Depending on the zone of clearance on skim milk agar medium, a strain (KW2) was selected for further studies. Strain KW2 forms round, smooth and light cream colonies with entire margins and convex elevations. The organism was gram positive, endospore forming, rod shaped and motile. It showed positive results for catalase, oxidase, gelatin liquefaction, and starch hydrolysis tests but negative for Tween 20/80 hydrolysis. KW2 could grow at pH 8-12 with optimal growth at pH 10. It could tolerate up to 5% NaCl (w/v) and grew well at the temperature range of 25-45°C with optimum at 30°C. The phenotypic features of KW2 corresponded with those outlined in description of the genus Bacillus in the Bergey’s Manual of Determinative Bacteriology [17].

Optimization by OVAAT approach

Protease production by Bacillus sp. KW2 increased progressively with increase in incubation temperature and peaked at 30°C (246 ± 16 U/ml). Protease production declined beyond 30°C; at 40°C, enzyme production was reduced to 50 ± 6 U/ml (Figure 1).

Figure 1: Effects of incubation temperature on protease production by strain KW2.

Optimum production of protease at 30°C was also reported for Bacillus sp. B21-2 [18], Bacillus sp. MIG [19] and Bacillus pantothenticus [20]. Sodium carbonate was found to significantly affect medium pH and protease production by Bacillus sp. KW2. Maximum protease production was observed at 1% (w/v) Na₂CO₃ concentration. Presence of Na₂CO₃ (1% w/v) maintains pH of the medium at around 10.7 (Figure 2). Similarly, Horikoshi also used Na₂CO₃ (1% w/v) to adjust pH of the medium to alkaline levels for growth and alkaline condition protease production by Bacillus No 221 [21]. However, Bacillus horikoshii [22] and Bacillus circulans MTCC 7942 [23] were reported to produce alkaline protease optimally in presence of 0.6% (w/v) Na₂CO₃. During protease production by alkaliphiles, Na₂CO₃ incorporation as a source of alkalinity to their growth media is routinely practiced so as to simulate their growth [23]. The role of Na₂CO₃ is not only to maintain alkalinity of the medium but also to provide Na⁺ ions required for growth of the alkaliphiles.

Figure 2: Effects of Na₂CO₃ concentration on pH of the medium and protease production.

Effects of the carbon sources on protease production by Bacillus sp. KW2 were studied by using various C sources instead of glucose in GYM medium. Enhancement of protease production was observed in presence of rice bran (496 ± 4 U/ml), fructose (365 ± 5 U/ml), and starch (364 ± 8 U/ml). However, lactose and tri-sodium citrate inhibited protease production with resultant titer values of 112 ± 5 U/ml and 100 ± 4 U/ml respectively (Figure 3).

Figure 3: Effects of carbon sources on protease production. Statistical significance was determined by student’s t test at p ≤ 0.05. Values with same alphabets are not statistically significant.

Enhancement of protease production in presence of rice bran is an interesting feature because rice bran is a by-product of the rice milling industry that can be used as cheap and readily available substrate for enzyme production. Organic nitrogen sources such as soybean meal, yeast extract and peptone are well known inducers of protease production. In this study, soybean meal was found to be the best nitrogen source for enhanced protease production by the strain KW2 (Figure 4). Similar results were also reported by Joo et al. [24] indicating enhanced protease production by soybean meal in Bacillus clausii. However, inorganic nitrogen sources such as KNO₃ and NH₄Cl inhibited protease production by KW2. The inhibitory effects of inorganic nitrogen sources on protease production have also been reported for Bacillus sp. JB-99 [25], Bacillus horikoshii [4], Bacillus clausii [26] and Streptomyces sp. [27]. After optimization by OVAAT approach, the GYM medium was modified. The modified medium was composed of rice bran (1% w/v), soybean meal (0.5%, w/v), MgSO₄ (0.5%, w/v), KH₂PO₄ (0.5 %, w/v) and Na₂CO₃ (1.0%, w/v).

Figure 4: Effects of nitrogen sources on protease production: Yeast extract (Y.E), Beef extract (B.E), Soybean meal (S.M). Statistical significance was determined by student’s t test at p ≤ 0.05. Values with same alphabets are not statistically significant.

Optimization by statistical approaches

PBD: A total of seven variables were studied by using PBD. Higher levels of rice bran, soybean meal, inoculum age and inoculum size enhanced protease production whereas higher level of KH₂PO₄ reduced protease production (Table 4). Effect of divalent metal ions (Ca⁺² and Mg⁺²) on protease production were also investigated. In this study, Ca⁺² was found to be stimulatory for protease production; however, high level of Mg⁺² (0.5%, w/v) had inhibitory effect on protease production. Ca⁺² and Mg⁺² were reported to be essential for protease production by Bacillus sp. N-40 [28]. In contrast to the present study, Tiwary and Gupta [16] reported that keratinase production by Bacillus licheniformis was inhibited by Ca⁺² while Mg⁺² did not exert any significant effect. Four parameters (rice bran, soybean meal, KH₂PO₄ and CaCl₂) were selected for further optimization studies. MgSO₄ was kept at lower level (0.02 w/v) while inoculum age and inoculum size were fixed at higher levels (24h and 5%, v/v respectively).

Table 4: Screening of signal parameters for protease production by PBD.

Optimization by RSM

The effects of four parameters (rice bran, soybean meal, KH₂PO₄ and CaCl₂ concentrations) on protease production were studied using RSM. A set of 30 experiments with different combinations of the four selected factors were performed. Both observed and predicted values were found to be similar suggesting the authenticity of the model. The results of the RSM experiments for KW2 are presented in Table 5.

Table 5: Results of RSM studies for protease production by strain KW2.

The protease production (Y) by Bacillus sp. KW2 can be expressed in terms of the following regression equation:

Where A, B, C and D represent rice bran, soybean meal, KH₂PO₄ and CaCl₂ concentrations respectively. The model coeeficients determined by multiple linear regression and analysis of variance (ANOVA) are presented in Table 6. The Model F-value of 33.7 implies that the model is significant. There is only a 0.01% chance that "Model F-Value" this large could occur due to noise.Values of "Prob>F" less than 0.05 indicate that model terms are significant. R2 value (0.959) is high indicating that the model is significant. The predicted R2 of 0.8515 is in reasonable agreement with the adjusted R2 of 0.9312. "Adeq Precision" measures the signal to noise ratio. A ratio greater than 4 is desirable. The ratio of 17.713 obtained in ANOVA table indicates an adequate signal. One factor, contour and 3D response surface curves were plotted to study interactions among the four factors to determine the optimum concentration of each factor for maximum protease production. Based on the predictions drawn from the one factor plot, rice bran and KH₂PO₄ at their maximum and minimum levels were inhibitory for enzyme production. Higher level of soybean meal inhibited protease production whereas CaCl₂ did not have significant effect on protease production. The interactions among the variables are shown in Figures 5A-5D.

Table 6: ANOVA values of the CCD model.

Figure 5: 3D response surface curve showing interaction effects of variables on protease production (A-D) from Bacillus sp. KW2. Out of four variables, interactions between two variables were studied by keeping two variables at their optimum levels as selected from the RSM study. A) Interactions between rice bran and soybean meal at fixed CaCl₂ (0.07%) and KH₂PO₄ (0.5%) concentrations.

Figure 5B: Interactions between rice bran and KH₂PO₄ at fixed soybean meal (1.31%) and CaCl₂ 0.07%) concentrations.

Figure 5C: Interactions between CaCl₂ and rice bran at fixed soybean meal (1.31%) and KH₂PO₄ (0.5%) concentrations.

Figure 5D: Interactions between KH₂PO₄ and soybean meal at fixed rice bran (1.0%) and CaCl₂ (0.07%) concentrations.

On the basis of the overall predictions and the interaction plots, it was observed that maximum protease production is achieved when concentrations of rice bran, KH₂PO₄ and CaCl₂ were kept at their central levels (i.e. 1%, 0.5% and 0.07%, w/v respectively) and soybean meal at 1.3% (w/v). Rice bran contains 49.4% carbohydrates, 21.3% fat and 16.5% protein [29]. In India and other countries, approximately one million tons of rice bran are produced annually and used predominantly for animal feed [30]. Use of rice bran as substrate for microbial enzyme production has been reported earlier especially for solid state fermentation [31-33]. There are meagre reports on use of rice bran in submerged fermentation. And there are no reports so far on use of rice bran in RSM studies. This study will be the first report on use of rice bran in optimization of alkaline protease production by statistical approaches. The accuracy of the values predicted by the model was verified by 10 additional independent experiments. The predicted and experimental values of protease production are presented in Table 7.

Table 7: Validation of the CCD model.

The experimentally determined values were in close agreement with the predicted values showing the authenticity of the model. Based on the response surface curves and validation experiments, the final optimized medium for protease production by Bacillus sp. KW2 was as follows: rice bran (1%, w/v), soybean meal (1.3%, w/v), KH₂PO₄ (0.5%, w/v), CaCl₂ (0.07%, w/v), MgSO₄ (0.02%,w/v) and Na₂CO₃ (1.0%, w/v). The Final Optimized Medium was designated as FOM. FOM is a relatively cheaper medium as compared to GYM. Protease production by strain KW2 was recorded as 246 ± 5 U/ml in the unoptimized medium (GYM) and 1302 ± 56 U/ml in the FOM medium. Overall, a 4.8 fold increase in protease production was achieved after optimization. Similar fold increase in protease production after optimization by statistical methods were reported earlier for Bacillus mojavensis (4.2 fold) [34], Microbacterium sp. (3.6 fold) [12] and Bacillus amyloliquefaciens (3.92 fold) [35]. However, higher fold increase in protease production have also been reported Bacillus sp. BGS (6.36 fold) [8] and Bacillus sp. RG-14 (12.8 fold) [36].

Studies of time course for protease production by Bacillus sp. KW2 indicated that protease production started at the exponential phase and reached maximal levels at the late stationary phase of growth (84 h) (Figure 6). Similar to our findings, production of protease at late stationary phase have also reported for Bacillus cereus at 96 h incubation [37] and a marine actinomycete, Saccharopolyspora sp. at 14 days incubation [38].

Figure 6: Time course of growth and protease production of Bacillus sp. KW2 in FOM medium. Bacterial growth was monitored by measuring the OD at 600 nm.

The present study shows that protease production is greatly influenced by cultural conditions and media constituents. The optimum conditions for protease production by Bacillus sp. KW2 was observed at 30°C, pH 10.7 and 84 h of incubation. This study also established that the strain KW2 could produce alkaline protease in a low cost medium (FOM) containing rice bran and soybean meal as major carbon and nitrogen sources. Protease production was also optimized at two levels i.e. one-variable-at-a-time and statistical approaches using Design Expert 6 software. After optimization, protease production was increased by 4.8 fold. Strain KW2 may be a promising candidate for production of alkaline protease for commercial applications.

PK gratefully acknowledges the award of a JRF/SRF research fellowship by the Council of Scientific and Industrial Research (CSIR), Government of India, which facilitated the completion of this research work. The authors would also like to acknowledge Prof. Rani Gupta, Department Of Microbiology, University of Delhi South Campus for help in analysis of the Design Expert 6 software. The authors gratefully acknowledge financial support through the Department of Biotechnology (DBT) Biotech Hub scheme (BT/04/NE/2009) and DBT Twinning Program (BT/178/NE/TBP/2011) awarded by DBT, Govt. of India.

Authors do not have any competing or conflicting interests in submission of this manuscript.