In the next years, it is expected that the demand for L-asparaginase will expand substantially. This is due to the fact that, in addition to its applications in the medical field, it can also be used in the food processing industry28. Therefore, finding alternative sources of L-asparaginase could be crucial. Multiple Streptomyces species have been examined in terms of their L-asparaginase production, such as S. paulus CA014, S. olivaceus NEAE-11931, alkaliphilic S. fradiae NEAE-8232. In our previous study, we collected 35 soil samples from many regions of Egypt and recovered one hundred and thirty different actinomycete strains31. The L-asparaginase activity of each of these isolates was evaluated using the plate assay method. As shown in Fig.1A, B, the activity of L-asparaginase was verified through the observation of a color transition from yellow to pink in the medium that surrounds the colony. This change was in comparison with the control plate, which was prepared with medium devoid of dye (Fig.1C).
(A, B) The plate assay method used to assess L-asparaginase production after 2 and 5 days, (C) a dye-free medium prepared for the control plate, (D) Streptomyces sp. strain NEAE-99 growth on SSF medium.
The culture medium’s pH increased and turned pink instead of yellow due to the production of L-asparaginase by the isolate. This enzyme caused the hydrolysis of L-asparagine, resulting in the release of ammonia. The increase in pH caused by the ammonia led to the change in color of the culture medium. Among the isolates that were evaluated, Streptomyces sp. strain NEAE-99 emerged as a promising candidate for further research involving L-asparaginase production under SSF conditions (Fig.1D). L-glutaminase activity was also assessed, and the findings indicated that the L-asparaginase that produced is glutaminase-free.
Cultural and morphological characteristics of Streptomyces sp. strain NEAE-99
The culture characteristics of Streptomyces sp. strain NEAE-99 are displayed in Table1. On all tested media, the strain grew well, displayed abundant, well-developed aerial mycelium that ranged in color from grey, whitish blue to bluish grey (Table1). As shown in Table1; Fig.2, the aerial mycelium appeared grey on the ISP3 and ISP7 media, bluish grey color on the ISP2 and ISP4 media, and a whitish blue color on the ISP5 and ISP6 media depending on the medium components. Diffusible blue pigments are produced when the strain grown on ISP2, ISP4, ISP5 and ISP6 media. In both ISP3 and ISP7 media, violet pigment is produced.
The aerial mycelium color of strain NEAE-99 upon ISP4 medium.
Using a scanning electron microscope (SEM) at magnifications ranging from 2000× to 13,000×, the morphology of Streptomyces sp. strain NEAE-99 cultured on ISP2 medium was examined. A scanning electron micrograph indicates the absence of verticils and mycelium fragmentation. The aerial mycelium differentiated into long spiral-shaped spore chains, which may be closed or opened. More than fifty elongated, smooth-surfaced spores with diameters ranging from 0.68 to 0.86 × 0.89 to 1.30μm can be observed in spore chains (Fig.3). The morphological characteristics of strain NEAE-99 align with those recognized among the Streptomyces genus33.
SEM showing the spore surface and morphological characteristics of the spore-chains of strain NEAE-99 at magnifications ranging from 2000 to 13,000 ×.
Physiological properties of strain NEAE-99
The physiological characteristics of the strain NEAE-99 are displayed in Table2. The strain grew at pH values of 5, 7, and 9 and a temperature ranging from 25 to 40°C. It was determined that the optimal temperature and pH were 30°C and 7, respectively. The formation of melanoid pigments was not observed. The strain exhibited a moderate resistance to NaCl, with a maximum concentration of 4% (w/v). The strain was grown using the following sugars: Trehalose, D (–) fructose, D (+) mannose, D (+) glucose, rhamnose, maltose, cellulose, L-arabinose, ribose, and D (+) galactose. Streptomyces sp. strain NEAE-99 produced α –amylase, L-asparaginase, protease, cellulase, uricase, chitosanase, lecithinase and gelatinase. Streptomyces sp. strain NEAE-99 has the capability to degrade lecithin, casein, and starch. Streptomyces sp. strain NEAE-99 displayed positive coagulation and peptonization of milk, but it showed a negative reduction of nitrate to nitrite (Table2). No antibacterial activity has been seen against any of the tested bacterial or fungal strains.
Phylogenetic analysis
The 16S rRNA gene sequence for strain NEAE-99 has been determined to be 1537 base pairs in length. The DNA sequencing data has been formally submitted to the GenBank database and assigned the accession number KJ676777. A substantial similarity between the strain under study and numerous species of the Streptomyces genus was noted via a GenBank database search using BLAST29. The neighbor-joining algorithm technique was applied in order to create a phylogenetic tree34 and the MEGA-X software30 using the 16S rDNA gene sequences of members of the Streptomyces genus (Fig.4).
A neighbor-joining phylogenetic tree demonstrates the evolutionary relationship between similar Streptomyces species and Streptomyces sp. strain NEAE-99.
Streptomyces sp. strain NEAE-99 exhibits a close phylogenetic relationship to certain related Streptomyces species, as illustrated by the phylogenetic tree. According to the phylogenetic tree, Streptomyces sp. strain NEAE-99 is a member of the same clade with S. violaceoruber strain SC02 (Accession No. MW075679.1), with an identity percentage of 99.93%. Streptomyces sp. strain NEAE-99 is closely related to the type strain of S. violaceoruber owing to its physiological, morphological, and culture characteristics. Consequently, Streptomyces sp. strain NEAE-99 has been identified as S. violaceoruber strain NEAE-99.
Optimization of L-asparaginase production by S. violaceoruber strain NEAE-99 using central composite design
The process of SSF is being optimized with the goal of enhancing L-asparaginase production through the use of ingredients that are less expensive in order to minimize the costs of production. Selecting an appropriate solid substrate is one of the most critical steps in SSF. The appropriateness of some agro-industrial wastes, like soybean and wheat bran (either separately or in combination), was assessed in our earlier study by El-Naggar et al.35 as carbon sources and as supporting material that has been loaded with all the nutrients required for S. violaceoruber growth and L-asparaginase production. The results demonstrate that two substrates significantly increased S. violaceoruber’s production of L-asparaginase.: wheat bran (18.088 U/gds) and soybean (27.985 U/gds). Nevertheless, the highest L-asparaginase production by S. violaceoruber during SSF was achieved by using a mixture of soybean and wheat bran in a 1:1 ratio by weight, resulting in a yield of 41.864 U/gds (Fig.1D). As the mixture of soybean and wheat bran, in a 1:1 ratio by weight, yielded the highest L-asparaginase production, so it was chosen for more fermentation studies to produce L-asparaginase under SSF. Because soybeans contain significant amounts of minerals, carbohydrates, lipids, and proteins, they are a suitable substrate for L-asparaginase production35. According to a study conducted by Isaac and Abu-Tahon36, Fusarium solani AUMC 8615 achieved the highest activity of L-asparaginase (187.9 U/mL) on wheat bran, the most favorable natural substrate out of the seven substrates studied.
The CCD design matrix comprises 30 experimental trials with 6 replicates at the midpoints (5, 6, 7, 10, 17, and 24) used in order to achieve the highest possible L-asparaginase production and to investigate the linear, quadratic, and mutual interactions between four independent variables including: soybean and wheat bran (1:1, w/w) (X1), concentration of dextrose (X2), concentration of L-asparagine (X3), and concentration of KNO3 (X4) (Table3).
In addition, Table3 presents the experimental and theoretical findings of L-asparaginase production for a variety of combinations of the four independent factors. The results indicated that the production of L-asparaginase was significantly influenced by the concentrations of four independent factors. The greatest yield of L-asparaginase, 191.33 U/gds, was achieved in central run number 10 using the optimal experimental conditions: 15g/250 mL Erlenmeyer flask of a mixture consisting of equal weight ratios of soybean and wheat bran, supplemented with freshly prepared ADS broth containing (g/L): L-asparagine 10, dextrose 2, and KNO3. (1) Additionally, run number 23 yielded the minimal production of L-asparaginase, measuring 22.34 U/gds. This occurred when the fermentation medium consisted of 15g/250 mL Erlenmeyer flask of a mixture consisting of equal weight ratios of soybean and wheat bran, supplemented with freshly prepared ADS broth containing (g/L): of dextrose 2, L-asparagine 10, and of KNO3. (2) The decrease in L-asparaginase production during run no. 23 may have been caused by an elevated concentration of KNO3.
Multiple regression analysis and ANOVA
Tables4, and 5 present the results of the multiple regression analysis and ANOVA performed to investigate the extracellular L-asparaginase production using S. violaceoruber as influenced by the effects of the independent process factors namely: soybean and wheat bran (1:1, w/w) (X1), concentration of dextrose (X2), concentration of L-asparagine (X3), and concentration of KNO3 (X4). An assessment of the model’s validity was conducted by analyzing the data provided in Table4, including the coefficient estimates and the values of determination coefficient (R2), predicted R2, adjusted R2, F-value (Fisher value), P- (probability value). Furthermore, an assessment of lack of fit was conducted to verify the precision of the model. The model used exhibited an R2 value of 0.9891 (Table4), indicates that the model accurately predicts 98.91% of the variation in L-asparaginase production, with only 1.09% of the whole variation remaining unexplained. R2 values, ranging from 0 to 1, indicate the degree to which variability in measured response values may be attributed to the variables used in the experiment and their interactions. A greater R2 value (approximate 1) signifies a more robust model and a higher capacity for predicting the response2. The high R2 value indicates a robust correlation between the predicted and measured response values, revealing the variability of the response values around their mean.
The adjusted-R2 value of the regression model used for studying the production of L-asparaginase by S. violaceoruber was determined to be 0.9789. A higher adjusted R2 value implies a high level of model accuracy. The adjusted coefficient of determination (adjusted-R2) provides an explanation for the variation in the response as influenced by the independent factors. The current model’s predicted-R2 value of 0.9546 indicates its significance for predicting L-asparaginase production in future experiments. The theoretical and experimental values of L-asparaginase production are significantly correlated, as evidenced by the high level of agreement among the predicted-R2 value of 0.9546 and the adjusted-R2 value of 0.9789. When evaluating the predictive power of the model to predict the response values at various levels of the assessed process variables in the future experiments, the predicted R2 is applied37.
Interactions between two variables can be categorized into two categories: Antagonism is demonstrated by negative coefficient values, which suggest an antagonistic relationship among the variables, and synergism, indicated by a positive coefficient that means a synergistic interaction between the variables. Table4 clearly indicates that the coefficients values of the four process factors are positive. These variables included soybean and wheat bran (1:1, w/w) (X1), concentration of dextrose (X2), L-asparagine concentration (X3), and KNO3 concentration (X4). These positive coefficients values indicate that the process factors increase L-asparaginase production using S. violaceoruber within the ranges that were evaluated. Furthermore, the mutual interaction effects between X1 and X2; X2 and X3; X3 and X4 had a positive influence on L-asparaginase production by S. violaceoruber. The data indicated that the mutual interaction effects between X3 and X4, as well as X2 and X3, have significant positive coefficients with higher coefficient estimates (32.09 and 7.05, respectively) and lower P-values (< 0.0001 and 0.0031, respectively). This meaning that they act as limiting factors, and their mutual interaction effects will enhance L-asparaginase production by S. violaceoruber. Conversely, the coefficient estimates of 0.36 indicates that the mutual interaction effects between X1 and X2 are negligible positive coefficients with the lowest effect. On the other hand, the quadratic impacts of all process variables and the mutual interaction impacts between X1 and X3; X1 and X4; X2 and X4 have negative coefficient values (Table4), which indicates that they have a negative impact on L-asparaginase production by S. violaceoruber. It can be concluded that the variable significantly influences the response when the value of the calculated coefficients is large, irrespective of their sign. If the coefficient value is close to zero, it is hypothesized that the variable has minimal or no impact on the production. If the coefficient is positive, it suggests that production increases as the value of the tested variable increases. In contrast, a negative sign signifies an increase in production when the variable reaches its minimum value. The variable exerts a negligible or nonexistent influence on the ultimate outcome, as indicated by a coefficient value approaching zero. A positive coefficient associates with an increase in the value of the tested variable, signifying that such an increase corresponds to a rise in production. Conversely, when the variable reaches its minimum value, a negative sign signifies an increase in production.
The significance of individual variables is evaluated, and interactions among the factors under consideration is determined using P-values and F-values. Moreover, it was established that process factors are assumed to significantly affect the response if their P-values are < 0.0537. The regression model’s significance is demonstrated by the F-value of 97.07 and the P-value of less than 0.0001. L-asparaginase production by S. violaceoruber is significantly impacted by the linear effects of X1, X2, and X3, as indicated by P-values less than 0.05 and F-values of 40.79, 70.46 and 12.96; respectively (Table4). On the basis of the F-value of 0.01 and the probability value of 0.9303, it is evident that the linear effect of KNO3 (X4) has no significant effect on the production of L-asparaginase by S. violaceoruber (Table4). Conversely, P-value < 0.0001 and a negative coefficient estimate value of -37.48 support the importance of the quadratic impact of KNO3, indicates that it greatly decreases the production of L-asparaginase. The mutual interaction effects between X1 X3 (P-values of 0.0404); X2 X3(P-values of 0.0031) and X3X4 (P-values of < 0.0001) have significant effects on the production of L-asparaginase by S. violaceoruber. Conversely, the P-values of 0.8593, 0.6791, and 0.5638 indicate that the mutual interactions effects between X1 X2; X1 X4 and X2X4; respectively, do not exert statistically significant impacts on production of L-asparaginase by S. violaceoruber. The significance of the quadratic impacts of X1, X2, X3, and X4 is supported by P-values below 0.0001, which indicate that these impacts have a substantial influence on L-asparaginase production. The C.V. %, mean, Std. Dev. and PRESS values are 8.61, 93.2, 8.02 and 4019.9; respectively. When evaluating the signal-to-noise ratio, the value of adequate precision is crucial, the adequate precision value in the current study is 28.22. The precision of the model can be determined by a signal-to-noise ratio that is more than 4 which is considered to be desirable38.
The following equation can be used to determine the maximal predicted production of L-asparaginase (Y):
$$\begin{aligned} {\text{Y}} & \,=\,{\text{175}}.{\text{6}}0\,+\,{\text{1}}0.{\text{46}}{{\text{X}}_{\text{1}}}\, - \,{\text{13}}.{\text{75}}{{\text{X}}_{\text{2}}}\,+\,{\text{5}}.{\text{9}}0{{\text{X}}_{\text{3}}}\,+\,0.{\text{15}}{{\text{X}}_{\text{4}}}\, \\ & \;\;+\,0.{\text{36}}{{\text{X}}_{\text{1}}}{{\text{X}}_{\text{2}}}\, - \,{\text{4}}.{\text{5}}0{{\text{X}}_{\text{1}}}{{\text{X}}_{\text{3}}}\, - \,0.{\text{85}}{{\text{X}}_{\text{1}}}{{\text{X}}_{\text{4}}}\,+\,{\text{7}}0.0{\text{5}}{{\text{X}}_{\text{2}}}{{\text{X}}_{\text{3}}}\, - \,{\text{1}}.{\text{18}}{{\text{X}}_{\text{2}}}{{\text{X}}_{\text{4}}}\, \\ & \;\;+\,{\text{32}}0.0{\text{9}}{{\text{X}}_{\text{3}}}{{\text{X}}_{\text{4}}}\, - \,{\text{21}}.{\text{18}}{{\text{X}}^{\text{2}}}_{{\text{1}}}\, - \,{\text{13}}.{\text{27}}{{\text{X}}^{\text{2}}}_{{\text{2}}}\, - \,{\text{31}}0.0{\text{7}}{{\text{X}}^{\text{2}}}_{{\text{3}}}\, - \,{\text{37}}.{\text{48}}{{\text{X}}^{\text{2}}}_{{\text{4}}} \\ \end{aligned}$$
(2)
X1, X2, X3, and X4 are the values of the independent factors including soybean and wheat bran (1:1, w/w) (X1), concentration of dextrose (X2), concentration of L-asparagine (X3), and concentration of KNO3 (X4).
An appropriate and highly significant polynomial model is chosen from linear, 2FI, and quadratic models that suit L-asparaginase production by S. violaceoruber, depending on the results of the fit summary in Table5. The results indicate that the quadratic model has an insignificant lack of fit (P-value = 0.6141, F-value = 0.85) and a very small P-value (< 0.0001). Consequently, the model is considered appropriate model and highly significant for L-asparaginase production by S. violaceoruber (Table5). In addition, the quadratic model exhibits greater efficiency compared to other models, as shown by its higher predicted R2 value (0.9546), adjusted R2 value (0.9789), and R2 value (0.9891). In addition, the quadratic model’s summary statistics indicated a lower PRESS value of 4019.9 and a lower standard deviation of 8.02 which demonstrate the validity of the model and its capacity to represent the data accurately.
Three-dimensional (3D) surface plots
To elucidate the impact of the variables being investigated, their interactions, and the optimal concentrations needed to achieve maximum L-asparaginase production, 3D surface plots have been created. Figure5A–C illustrates L-asparaginase production on the Z-axis as a function of a pair of variables, while keeping the values of the other two variables at their zero levels.
Figure5A demonstrates the effects of soybean and wheat bran mixture (g/250 mL Erlenmeyer flask) (X1), dextrose concentration (X2) on the L-asparaginase production using S. violaceoruber, while maintaining L-asparagine (X3) and KNO3 (X4) at their zero levels. The experimental results demonstrate that at low as well as high levels of soybean and wheat bran (g/250 mL Erlenmeyer flask) and dextrose, production of L-asparaginase was decreased. The highest yield of L-asparaginase was achieved at moderate levels of both soybean and wheat bran mixture (g/250 mL Erlenmeyer flask) and dextrose concentration. Consequently, the point prediction tool of Design expert software (version 12) was applied to determine the optimal value for each factor in order to achieve the highest predicted production of L-asparaginase. It is possible to achieve the maximum production of 180.47 U/gds L-asparaginase by employing a mixture of 16.4g/250 mL Erlenmeyer flask of soybean and wheat bran in a ratio of 1:1; w/w, dextrose (2.5g/L), when the concentrations of L-asparagine and KNO3 maintained at their central levels (at 10, 1g/L; respectively). The results of this study agree with the findings of El-Naggar et al.35 , who reported that S. brollosae NEAE-115 utilized soybean and wheat bran in a ratio of 1:1; w/w as substrates for L-asparaginase production under SSF conditions. Various investigations on enzyme production under SSF conditions have shown that, the use of numerous substrates increases the production of metabolites and microbial growth more than the use of a single substrate alone39,40 ,41. According to Sharma and Mishra42, It has been shown that combining a variety of substrates can provide a wide range of nutrients that are not available from a single source. By combining soybean, wheat bran, and other inexpensive substrates, a significant number of nutrients can be obtained. Dharmsthiti and Luechai43 discovered that Aspergillus niger AK10 effectively used soybean as a substrate under SSF conditions. Soybean is an ideal for L-asparaginase production due to its high content of minerals, lipids, proteins, and carbohydrates. Dextrose (glucose) has been found to be the most effective carbon source for promoting the production of L-asparaginase by S. olivaceus NEAE-119 during submerged fermentation at a concentration of 3g/L44. The utilization of cost-effective agricultural byproducts, such as rice bran, sesame oil cake, wheat bran, soybean meal, groundnut oil cake, and tea trash, aids in cost reduction and supports environmental sustainability45,46.
Figure5B illustrates how different amounts of soybean and wheat bran (g/250 mL Erlenmeyer flask) (X1), L-asparagine concentration (X3) affect L-asparaginase production, while maintaining concentrations of both dextrose (X2) and KNO3 (X4) at their zero levels. The experimental results demonstrate that at low as well as higher levels of both L-asparagine and soybean and wheat bran (g/250 mL Erlenmeyer flask), L-asparaginase production was decreased. Abdel-Hamid et al.47. reported that a significant decrease in L-asparaginase production was observed when the concentration of L-asparagine was increased. This reduction may be attributed to the negative impact of L-asparagine on L-asparaginase gene expressions or the downregulation of nitrogenous compound availability. On the other hand, the reduction in L-asparaginase production at elevated concentrations of soybean and wheat bran may be attributed to catabolic inhibition effect or enzyme inactivation. It was reported that the production of microbial L-asparaginase faces catabolic inhibition and required less amount of carbon source48. A maximum yield of L-asparaginase was achieved at moderate levels of both soybean and wheat bran mixture (g/250 mL Erlenmeyer flask) and L-asparagine concentration. It is possible to achieve the highest predicted L-asparaginase production of 177.041 U/gds by using 16.4g/250 mL Erlenmeyer flask from soybean and wheat bran mixture and 10.40g/L of L-asparagine, while the concentrations of dextrose and KNO3 were maintained at their respective central levels (at 2, 1g/L; respectively). The results of the study indicate that the production of L-asparaginase is significantly impacted by the nitrogen source. Microorganisms utilize various organic and inorganic nitrogen sources in order to produce vital biological substances, such as industrial enzymes, proteins, nucleic acids, and amino acids, as well as the cell wall49.
Figure5C illustrates the effects of soybean and wheat bran (X1), KNO3 (X4) on L-asparaginase production, while keeping dextrose (X2) and L-asparagine (X3), at their zero levels. In Fig.5C, it was seen that the maximum amount of L-asparaginase was produced at intermediate concentrations of both soybean and wheat bran mixture and KNO3. Increased levels of these variables have been shown to reduce L-asparaginase production. It is possible to achieve the highest predicted L-asparaginase production of 176.88 U/gds by using 16.2g/250 mL Erlenmeyer flask from the mixture of soybean and wheat bran and 1.01g/L of KNO3, while the concentrations of dextrose and L-asparagine were maintained at their respective central levels (at 2, 10g/L; respectively). The increase in L-asparaginase production resulting from the addition of KNO3 suggests that nitrogen is involved in the regulation of its production. For the production of L-asparaginase by B. licheniformis, ammonium sulfate was found to be the most efficient nitrogen source (the enzyme production increased by 35.56%) when compared to other nitrogen sources including asparagine, casein, ammonium chloride, sodium nitrate, yeast extract, urea and potassium nitrate50.
Figure5D illustrates the impacts of dextrose (X2) and L-asparagine (X3) on L-asparaginase production, while maintaining the level of soybean and wheat bran mixture (X1) and KNO3 (X4) at their zero levels. Figure5D demonstrated that L-asparaginase production increase with the increase in the concentration of both dextrose (X2) and L-asparagine (X3). At high concentrations of both dextrose (X2) and L-asparagine (X3), L-asparaginase production decreased. The highest L-asparaginase production is observed when the levels of dextrose (X2) and L-asparagine (X3) are beyond middle levels. L-asparaginase production is enhanced when carbon concentrations are increased, yielding considerable quantities of the enzyme51. It has been observed that the higher concentrations of dextrose (glucose) act as a catabolic repressor in the bacterial production of L-asparaginase by Erwinia aeroideae and Escherichia coli52,53. This may be attributed to the suppression of catabolites of lactate transport components, which promoted the production of L-asparaginase54. The highest predicted L-asparaginase production of 179.87 U/gds can be achieved by applying 2.5g/L dextrose and 10.8g/L of L-asparagine, while the concentrations of soybean and wheat bran mixture (X1) and KNO3 concentration (X4) were maintained at their respective central levels (at 15g/250 mL Erlenmeyer flask, 1g/L; respectively).
3D plots demonstrate L-asparaginase production using S. violaceoruber as influenced by four process factors, their interactions, and their optimal concentrations required for achieving the highest production of L-asparaginase.
Figure5E shows the effects of dextrose (X2) and KNO3 (X4) on production of L-asparaginase, while keeping soybean and wheat bran mixture (X1) and L-asparagine (X3) at their zero levels. Figure5E demonstrated that the production of L-asparaginase was induced by an increase in the concentrations of dextrose (X2) and KNO3 (X4), and then the production of L-asparaginase was subsequently reduced by a subsequent increase in the concentrations may be due to catabolic inhibition effect. The highest predicted L-asparaginase production of 179.13 U/gds can be achieved by applying 2.5g/L dextrose and 1.01g/L of KNO3, while the concentrations of soybean and wheat bran mixture, and L-asparagine were maintained at their central levels (at 15 and 10g/L; respectively).
Figure5F displays the effects of L-asparagine (X3) and KNO3 (X4) on production of L-asparaginase, while maintaining soybean and wheat bran mixture (X1), and dextrose (X2), at their zero levels. A subsequent increase in the concentrations of L-asparagine (X3) and KNO3 (X4) to their moderate levels resulted in an increase in the production of L-asparaginase. In general, the utilization of various nitrogen sources can be utilized to enhance enzyme production. A further increase in their concentration resulted in a decrease in the production of L-asparaginase. The decrease in L-asparaginase production at higher concentrations may be attributed to the nitrogen regulation of L-asparaginase production at excess L-asparagine and KNO3, which may be a result of the catabolic inhibition effect. It is possible to achieve the highest predicted L-asparaginase production of 175.95 U/gds by using 10.5g/L L-asparagine and 1.02g/L of KNO3, while the concentrations of soybean and wheat bran, and dextrose maintained at their central levels (at 15g/L and 2g/250 mL Erlenmeyer flask; respectively).
Model accuracy checking
The statistical analysis was performed to validate the precision of the design. Figure6A shows the normal probability plot (NPP) of the residuals. NPP is a crucial diagnostic tool used to assess if the residuals follow to a normal distribution. The residuals data points should be arranged in close proximity to a diagonal line and relatively uniformly dispersed. The data points for the residuals exhibited a normal distribution, located along the diagonal line for predicted L-asparaginase production using S. violaceoruber, confirming the validation of the model. Figure6B displays a graph comparing the predicted and actual L-asparaginase production. The dots are gathered around the line of best fit on the graph, demonstrating that the experimental and the predicted L-asparaginase production are highly correlated3. Figure6C displays a graph illustrating the residuals versus the predicted L-asparaginase production. Figure6C indicates that the residuals have a consistent variance. The model is valid because the residuals follow a uniform and random distribution, uniformly distributed both up and down the zero line, with no apparent pattern55.
(A) NPP of the residuals, (B) a graphic showing the actual versus predicted, (C) plot of residuals versus predicted and (D) Box- Cox plot for power transforms.
The Box-Cox graph is illustrated in Fig.6D. The blue line illustrates the current transformation (Lambda = 1). The green line illustrates the optimal lambda value (0.84). The red lines display the lowest and highest values of the confidence intervals, which range from 0.53 to 1.17. The model is in the optimal zone because the blue line, which represents the current transformation, is located between the two red lines, which represent the minimum and maximum values of the confidence intervals. Therefore, there is no demand for transformation of the data as the model accurately correlates to the experimental results56.
ANN modeling prediction for L-asparaginase production
Artificial Neural Networks (ANN) are an innovative approach in artificial intelligence that enables the creation of reliable computational models to interpret and analyze data in a way that resembles human brain processes16. The production of L-asparaginase by Aspergillus terreus MTCC 1782 was predicted using an artificial neural network (ANN) strategy57. The ANN model achieved better results compared to the response surface methodology (RSM). Two factors significantly influence the topology or construction of ANN: the numbers of hidden layer neurons or nodes and the overall number of layers. A standard neural network design consists of three layers: input, hidden, and output. These layers are made up of interconnected artificial neurons58.
The network architecture of ANN modeling comprises the validation of the final ANN model in addition to the training and learning processes59. In order to optimize the ANN’s performance in this investigation, the following parameter adjustments were made: The model NTanH (20), the number of tours (5000), the learning rate (0.1), and the validation technique (holdback, 0.2) were all applied. Initial data for four independent variables including soybean and wheat bran mixture in a ratio of 1:1; w/w (X1), dextrose (X2), L-asparagine (X3), and KNO3 (X4) are collected by the input layer. The hidden layer, which serves as a link between the input and output layers, has twenty neurons (Fig.7A). A series of mathematical operations is performed on the data that is provided via the input layer by hidden layer neurons to generate output at the output layer. The number of neurons in the hidden layer influences the prediction accuracy of ANN60. The ANN analysis and model performance in terms of its ability to predict extracellular L-asparaginase production by S. violaceoruber are illustrated in Table6.
ANN for L-asparaginase production by S. violaceoruber (A), predicted versus actual (B), predicted versus the residuals of training and validation processes for production of L-asparaginase (C).
Evaluation of an ANN model
ANN was employed to predict the production of L-asparaginase using the S. violaceoruber. The predictions were based on the experimental data, which are presented in Table3. Comparison of the experimental results for L-asparaginase production against the predictions of the ANN model is illustrated in Fig.7B. The data points are closely organized around the best prediction line in both the training and validation procedures, suggesting the model’s reliability. Furthermore, the residual data points exhibit a symmetrical normal distribution, wherein an equivalent number of points are situated on each side of the regression line (Fig.7C). This demonstrates the appropriateness of the ANN model.
The comparative predictive capability of ANN and CCD
In order to achieve maximum L-asparaginase production, it was necessary to find the optimum value for each factor through the use of prediction models like CCD or ANN. Compared to the CCD, the ANN-predicted values of L-asparaginase production had smaller residuals and showed better agreement with the experimental data (Table3). JMP Pro14’s model comparison tab was applied to assess the precision of the CCD and ANN predictions and identify the most appropriate model based on their predictions for L-asparaginase production and the corresponding experimental data. The following metrics were used to compare the predictive power of the ANN and CCD models: mean absolute deviation (MAD), R2 value, sum of squared errors (SSE), and root mean square error (RMSE)17,61. In a comparison of the predictive capabilities of ANN and CCD, it is evident that ANN demonstrates a substantially higher level of accuracy (Table6). Table6 shows that ANN has a higher R2 value of 0.9916, whereas RASE and AAE have lesser values of 4.96 and 3.62, respectively. Consequently, ANN is more capable of predicting the optimal value of each factor to maximize the production of L-asparaginase. The improved predictive capability of the ANN that has been observed can be attributable to the function of repeatedly training the neurons for a range of different independent factors16.
L-asparaginase production predictions using the desirability function (DF)
In order to maximize L-asparaginase production, the optimal predicted conditions have been determined using the desirability function (as illustrated in Fig.8). The JMP Pro14 software’s desirability function can be set to any value between 0 (representing undesirability) and 1(representing desirability). Prior to experimental verification of the optimization process, the desirability function’s value is usually estimated mathematically. The maximal predicted L-asparaginase production using S. violaceoruber was determined using the desirability function to be 216.19 U/gds. The optimal predicted conditions resulted in the highest predicted L-asparaginase production were 8.46g/250 mL Erlenmeyer flask of soybean and wheat bran mixture (1:1; w/w), 2.2g/L of dextrose, 18.97g/L of L-asparagine, and 1.34g/L of KNO3. The desirability value for these conditions was 0. 997. The verification revealed that the ANN was highly accurate and predictive, as the experimental values (207.55 U/gds) approximately matched the theoretical values.
The optimization plot illustrates the optimal predicted values of the process factors for maximal predicted L-asparaginase production by S. violaceoruber, along with the desirability function.
