Comparative one-factor-at-a-time, response surface (statistical) and bench-scale bioreactor level optimization of thermoalkaline protease production from a psychrotrophic Pseudomonas putida SKG-1 isolate

Microorganism and protease production

Pseudomonas putida SKG-1 (MTCC 10510) was isolated in our laboratory. This solvent and heavy metal tolerant strain is capable of producing solvent-, psychro-, thermo-, alkali-stable protease [9]. The pure culture was maintained over nutrient agar slants (pH 7.0) at 4°C.

For protease production, 250 ml sterile modified GYE (MGYE) broth [10] of pH 9.0 was taken in 500 ml Erlenmeyer flask, inoculated with 2% (v/v) mother culture of 0.8 OD (A₆₆₀; 1 cm cuvette) containing 2.8 × 10⁸ colony forming units (cfu) ml^-1 and incubated at 25°C in shaker incubator (150 rpm). The fermenting broth (5.0 ml) was periodically drawn aseptically, and growth was assessed by turbidity measurement at 660 nm. Each sample was then centrifuged at 12,000 rpm (4°C) for 10 min, and cell-free supernatant was used for protease assay.

Enzyme assay

The protease activity was assayed by casein digestion method of Shimogaki et al. [11] at 40°C and pH 9.5 (sodium carbonate-bicarbonate buffer). One unit of protease activity is defined as the amount of enzyme required to liberate 1 μg of tyrosine min^-1.

Optimization of nutritional and physical parameters

In the conventional scaling-up approach, various nutritional and physical parameters were optimized by maintaining all factors at a constant level in the basal medium, except the one under study. Each subsequent factor was examined after taking into account the previously optimized factor(s). Among carbon sources fructose, xylose, lactose, sucrose, soluble starch, maltose, glycerol and carboxymethyl cellulose (CMC) were supplemented individually by replacing glucose in the basal medium. All sugars were autoclaved separately at 10 psi for 20 min, and added at 1.0% (w/v) level. For the effect of different nitrogen sources, peptone plus yeast extract in MGYE broth were replaced individually by gelatin, urea, casein, casamino acid, beef extract, yeast extract, tryptone, sodium nitrate, ammonium nitrate and ammonium sulfate at 1.0% (w/v) level. Combination of each nitrogen source with 0.5% (w/v) yeast extract (YE) was also attempted. After optimization of carbon and nitrogen sources along with their concentrations, varied levels of yeast extract (0.1-0.7% w/v) were also studied to optimize its dose. Metal cations (0.01% w/v) studied to enhance protease yield were: Ca²⁺ (CaCl₂), Cu²⁺ (CuSO₄), Mg²⁺ (MgSO_4.), Mn²⁺ (MnSO₄), Fe²⁺ (FeCl₂), Zn²⁺ (ZnSO₄) and combination of Ca²⁺ +Mg²⁺ ions. The dose of best metal ion was optimized by its supplementation at 0.005-0.05% (w/v) concentrations. Initial pH of the medium (7.0-10.0), shaking speed (0-250 rpm), temperature (10-40°C) and time of incubation (0-72 h) were the physical parameters studied for their effect on bacterial growth and protease production.

Statistical optimization of factors affecting protease production by RSM

Where, Y is the predicted response, β_o offset term, β_n liner coefficient, β_nn squared coefficient, β_nm interaction coefficient, X_n n^th independent variable, X_n² squared effect and X_nX_m interaction effects.

Bench-scale bioreactor level optimization

Statistical analysis

Each set of experiment was performed thrice, and all values presented here are average of three independent experiments. The standard deviation for each value is ≤ 5%.

Protease production

The strain SKG-1 exhibited typical sigmoidal growth curve in modified GYE broth (Figure 1). After a steep exponential growth, the onset of stationary phase was at 60^th h onwards, and attained maximum growth and protease production at 66 and 72 h, respectively. The enzyme production initiated at 6^th h of bacterial growth with maximum 65 Uml^-1 during stationary phase at 72 h incubation, which thereafter decreased with time (Figure 1). Maximum enzyme production during stationary growth phase is in accordance with the findings of other researchers [2, 10]. Pseudomonas aeruginosa PseA strain exhibited slow growth up to 12 h, exponential growth up to 48 h, followed by a stationary phase. The protease secretion corresponded with the growth response, and reached maximum during the late exponential/early stationary phase [3].

Optimization of nutritional and physical parameters

Incubation time

The bacterial growth and protease production were in harmony up to 54 h under optimized nutritional and cultural conditions. Bacterial growth was in the exponential phase up to 54 h; thereafter entered the stationary phase. Whereas, enzyme production reached maximum (514 Uml^-1) during early stationary phase at 60 h; thereafter remained nearly constant up to 72 h fermentation (Figure 2). Our findings are in accordance with the results of several other researchers [3, 10]. Zeng et al. [26] reported maximum protease production of only 45 Uml^-1 at 10°C from Pseudomonas sp. strain DY-A during the late stationary phase of growth.

Statistical optimization of factors affecting protease production by RSM

Table 2 shows predicted responses of Box-Behnken design on the basis of above polynomial equation. This regression equation was assessed statistically for analysis of variance (ANOVA), and the results are predicted in Table 3. ANOVA of regression model demonstrated the determination coefficient (R²) 0.9988, which means 99.88% variability in the response could be explained by this model. The R² value is always between 0 and 1.0. The model is stronger and predicts better response when R² value is closer to 1.0 [5]. The value of the adjusted determination coefficient (adjusted R²) is 0.9976. This higher value of adjusted R² indicates greater significance of the model. A very low value of coefficient of variation (C.V., 0.70%) indicates better precision and reliability of the experiments executed. The adequate precision value of 121.595 measures signal to noise ratio, and a ratio > 4.0 is desirable. In this case, higher ratio indicates an adequate signal, and also proves that model can be used to navigate the design space.

The F- value of 820.46 in Table 3 implies that the model is significant. There is only 0.01% chance that a "model F- value" so large could occur due to noise. ANOVA analysis also indicated that the model term linear glucose (P < 0.0001), yeast extract (P < 0.0001), MgSO₄ (P < 0.0001), pH (P < 0.0001), quadratic glucose A² (P < 0.0002), yeast extract B² (P < 0.0001), MgSO₄ C² (P < 0.0001), pH D² (P < 0.0001) and four interaction terms were significant. Smaller the P-value, more significant is the corresponding coefficient. The P < 0.0500 indicates that model terms are significant.

The 3D response surface plots and two dimensional contour plots were used to understand the interaction effects of medium components and optimum concentration of each component required for maximum protease production. Response surface curves for variation in alkaline protease yield were constructed, and are depicted in Figure 3. In each set, two variables varied within their experimental range, while the other two variables remained constant at zero level.

Figure 3a depicts the production of alkaline protease with respect to glucose versus yeast extract. From the interaction response of glucose with yeast extract, protease yields increased with increasing glucose and yeast extract concentration up to 1.25% and 0.5%, respectively. The enzyme activity decreased at > 0.5% yeast extract concentration. However, the response curve did not show curvature, rather it was flattened. This suggested a demand for higher concentration of glucose. Figure 3b represents the interaction effect of glucose and MgSO₄ on production of alkaline protease. With an increase in glucose (0.75-1.25%, w/v) and MgSO₄ (0.0075-0.01%, w/v) concentration, the protease yield increased. Thereafter, an increase in MgSO₄ concentration up to 0.0125% (w/v) resulted in decreased protease production. The optimal values for glucose and MgSO₄ (w/v) were 1.25 and 0.01%, respectively.

Figure 3c reveals that maximum protease was produced at higher level of glucose (1.25%) and slightly lower alkaline pH (8.8) in the design range. This accorded a run number of 10, which is considered as the optimal condition of test variables. Table 2 shows that maximum protease of 617 Uml^-1 were produced at pH 8.8 (-1 in coded unit) and glucose at 1.25% (+1 in coded unit). Figure 3d depicts the interaction of two variables, viz., MgSO₄ and yeast extract on protease production. The protease production increased with increasing concentration of MgSO₄ (0.0075-0.01%, w/v) and yeast extract (0.3-0.5%, w/v). Further increase in their concentrations resulted in decreased alkaline protease production.

Figure 3e illustrates the interaction effect of pH (8.8-9.2) and yeast extract (0.3-0.7%) on alkaline protease production. Maximum enzyme units were produced with 0.5% yeast extract and pH 8.8. Further increase in pH and any deviation in yeast extract concentration from optimal, decreased the enzyme production. The effect of pH and MgSO₄ is shown in Figure 3f. The response curve analysis indicated that protease production decreased with increase in pH from 8.8 to 9.2, and increased with enhanced concentration of MgSO₄ up to 0.01% (w/v). Further increase in MgSO₄ concentration above 0.01% caused decreased enzyme yield.

The above optimized results (Table 2 standard order 10) concerning four variables were finally verified by again performing the batch shake flask experiment. The maximum experimental alkaline protease production of 617 Uml^-1 was very close to 612 Uml^-1 predicted by Box-Behnken design with 1.25% glucose, 0.5% yeast extract, 0.01% MgSO₄ and pH 8.8. Thus, under optimized conditions, the protease yield increased from 514 units in conventional optimization trial to 617 Uml^-1 using RSM at 60 h incubation. This proved that response surface methodology is a slightly better optimization approach as compared to conventional "one-variable-at-a-time" method in terms of improved protease yield in less time, resource and expenditure. Our findings are in agreement with the results of other researchers who have also reported RSM a better approach for enhanced protease production [4, 5, 16]. Reddy et al. [27] reported a 2.3 folds increase in alkaline protease production using Plackett-Burman and Response surface methodology by Bacillus sp. RKY3. Anbu et al. [28] obtained a good correlation coefficient of 0.9996 using Box-Behnken design and alkaline protease production of 112.90 Uml^-1 by Shewanella oneidensis MR-1 strain through response surface optimization.

Bench-scale bioreactor optimization

Effect of aeration

In this set of experiment, the bacterial growth and protease production were studied at a fixed agitation speed of 200 rpm and variable aeration rates of 0-1.5 vvm (Figure 4). The enzyme production was drastically low at aeration rates of 0 vvm (102 Uml^-1 at 84 h) and 0.2 vvm (326 Uml^-1 at 72 h). However, at 0.4 and 0.6 vvm, the protease production was 630 U and 798 Uml^-1, respectively at 60 h incubation. Further increase in aeration rate to 0.8 vvm produced maximum protease of 846 Uml^-1 at just 48 h fermentation. Still higher aeration rates of 1.0 and 1.5 vvm were detrimental for protease production, and resulted in reduced 741 Uml^-1 in 48 h and 436 Uml^-1 in 60 h, respectively (Figure 4). Bacterial growth pattern (Figure 5) was similar to protease production (Figure 4), which was very slow at lower aeration rates, and increased with increasing rate of aeration approaching maximum at 0.8 vvm. The DOT was initially 100% at 0 h, which reduced rapidly with increasing bacterial growth. The DOT was inversely related with the bacterial growth, i.e., at maximum exponential bacterial growth, the DOT was minimum and started increasing after the onset of stationary growth phase (Figure 5).

Under RSM optimized conditions, protease production was maximum at 60 h incubation, while it was highest in bioreactor trial at 48 h, thereby led to significant time saving of 12 h. Optimization of aeration rate at fixed agitation speed of 200 rpm indicated that oxygen supply to bacterial cell mass is a critical parameter for enzyme production under aerobic fermentation process. This could be achieved by optimizing the agitation speed and maintaining proper aeration; however, an appropriate balance of agitation and aeration is imperative to avoid any mechanical damage to bacterial cells which can reduce the enzyme yield. From the results of aeration optimization, it can be inferred that a direct correlation existed between bacterial growth and protease production. The reduction in DOT followed a similar pattern as experienced in previous experiment. It reduced rapidly during exponential growth phase, and rose again on commencement of stationary phase onwards (Figure 5). Rao et al. [29] reported highest protease production of 238.77 Uml^-1 by a Beauveria bassiana isolate on 6^th day of fermentation in a 5 litre stirred tank bioreactor at 150 rpm and 0.6 vvm aeration. Maximum protease of 340 Uml^-1 by Bacillus licheniformis NCIM-2042 was reported at aeration and agitation rates of 3 vvm and 200 rpm, respectively [7].

Effect of agitation at optimized aeration

After optimization of suitable aeration rate (0.8 vvm), we attempted to optimize agitation speed (100-250 rpm) for studying its effect on protease yield and the results are depicted in Figure 6. The enzyme production enhanced with increase in agitation rate, which was maximum at 150 rpm within 48 h of incubation. The order of protease units produced at 250, 200 and 150 rpm were (Uml^-1): 817 < 846 < 882 during 48 h fermentation. However, further decrease in agitation speed to 100 rpm reduced the enzyme yield only to 761 Uml^-1 at extended 60 h incubation (Figure 6). Throughout the study, the DOT started declining concomitantly with increase in bacterial growth, and reached minimum when bacterial growth was maximum. After commencement of stationary phase, the DOT again increased slowly (Figure 7). At optimized fixed aeration rate of 0.8 vvm and 150 rpm agitation, the DOT initially declined to 92% at 6 h incubation. It further declined sharply with increasing bacterial biomass, and approached minimum (18%) at 42 h, followed by slow increase to 49% at 72 h incubation (Figure 8). Thus, the protease production enhanced by ~43% (882 Uml^-1) at bench-scale bioreactor level during 48 h incubation compared to the conventional and RSM optimization at flask level (617 Uml^-1) during 60 h batch fermentation.

It is evident from the findings that an optimum agitation of fermenting broth is necessary for proper oxygen supply to bacterial cells. However, above the optimum speed it may damage the cells or change the cell morphology, which leads to reduced protease yield. Maximum protease production at 300 rpm has also been reported by Potumarthi et al. [7] in a stirred tank bioreactor. Any deviation in agitation speed to 200 or 400 rpm reduced the protease yield. They concluded that mixing is important for maximal protease production by optimizing both agitation and aeration for better oxygen mass transfer rate. This results in better product formation through avoiding/minimizing any mechanical damage to the bacterial cells. Many other researchers also reported the optimum agitation speed range of 150-300 rpm for protease production from different isolates [30–32]. It may, therefore, be inferred that a proper ratio of agitation and aeration is mandatory for appropriate oxygen transfer to bacterial cell mass, and also to minimize the shearing effect for maximum growth and protease production.