Comparative Optimization of α-Amylase Production in Bacillus velezensis sp. Using Response Surface Methodology and Artificial Neural Networks

Background:

This study aims to optimize the production of α-amylase from Bacillus velezensis sp. through a comparative analysis of advanced experimental methodologies, specifically Definitive Screening Design-Response Surface Methodology (DSD-RSM) and Artificial Neural Networks (ANN). The research emphasizes the use of environmentally sustainable and cost-effective agro-solid substrates, namely moong husk and soybean cake, to enhance enzyme yield, addressing the growing industrial demand for efficient enzyme production.

Materials and Methods:

The optimization process began with the identification of nine critical operational variables influencing α-amylase biosynthesis. Initial experiments utilized the One-Factor-at-a- Time (OFAT) approach, providing preliminary insights into the effects of individual variables. Subsequently, a second-order Definitive Screening Design (DSD) was employed within the RSM framework to systematically evaluate the interactions among the variables. The nine parameters analyzed included pH (5.0-7.0), temperature (30ºC-40ºC), carbon source concentration (moong husk at 2% to 6%), nitrogen source concentration (soybean cake at 1% to 3%), K₂PO₄ (0.1%-0.5%), MgSO₄ (0.05%-0.2%), NaCl (0.1%-0.5%), fructose (0.5%-2%) and NaNO₃ (0.1% to 0.5%). Following the DSD-RSM analysis, ANN modeling was utilized to further refine the optimization process, allowing for the prediction of enzyme yield based on the identified parameters. Confirmation experiments were conducted under the optimal conditions established by both DSD-RSM and ANN to validate the predicted enzyme activity.

Results:

The optimized conditions resulted in a significant α-amylase activity of 1092.49 IU/mL, achieved under the following optimal parameters: pH 5.48, temperature 34.27ºC, carbon source concentration of 4.09%, nitrogen source concentration of 2.02%, K₂PO₄ at 0.34%, MgSO₄ at 0.14%, NaCl at 0.23%, fructose at 1.54% and NaNO₃ at 0.53%. This marked a substantial increase in enzyme activity, representing a 2.6-fold enhancement compared to the initial yield of 418.25 U/mL obtained through the OFAT method. Statistical analysis confirmed the robustness of the optimization models, with R²=0.999, value indicating a strong correlation between predicted and actual enzyme activities

Conclusion:

The findings of this research demonstrate that the integration of DSD-RSM and ANN methodologies significantly enhances the production of α-amylase from Bacillus velezensis sp. compared to traditional optimization techniques. This study not only highlights the effectiveness of modern statistical and computational approaches in bioprocess optimization but also contributes to the development of sustainable enzyme production strategies that can meet the increasing industrial demands.