Model-based process design for surfactin production with Bacillus subtilis

Abstract

Bacillus subtilis is one of the most important production organisms in industrial biotechnology. However, there is still limited knowledge about the kinetics of fed-batch processes in bioreactors, as well as a lack of biological performance indicators, such as production yields, particularly regarding their variation over time. Understanding these kinetics and changes is crucial for optimizing the productivity in fed-batch processes. Fed-batch bioreactor cultures of Bacillus subtilis BMV9 in high cell density processes for surfactin production have been characterized with a kinetic model composed of first-order ordinary differential equations, describing the time course of biomass, substrate, surfactin and acetate. This model contributes to understanding critical restrictions and the knowledge gained was used to design and implement a model-based process. The model integrates biomass growth based on Monod kinetics, substrate consumption, surfactin synthesis and formation of the by-product acetate. After the model was parameterized for B. subtilis BMV9 using 12 different fed-batch bioreactor experiments, the kinetic model was able to accurately describe biomass accumulation, substrate consumption, product formation rates and, to some extent, the overflow metabolism involving acetate. Based on this, the kinetic model was used for a process design, in which the batch was omitted, which led to a product titre of 46.33 g/L and a space–time-yield of 2.11 g/(L*h) was achieved. The kinetic model developed in this study enables the description of the time course of biomass growth, substrate consumption and product formation and thus significantly improves process understanding. The computation of process parameters, which are not analytically accessible at any time, could be realized. A sensitivity analysis identified the maximum specific growth rate, substrate-related maintenance and the maximum acetate formation rate as key parameters influencing model outputs.