Improvement of the constraint-based modeling tool for the metabolic optimization of Bacillus and the production of lipopeptides from agro-resources

Introduction and Objectives Lipopeptides produced by Bacillus subtilis have emerged as sustainable alternatives to their synthetic counterparts, offering superior biodegradability and reduced environmental impact [1]. However, the high cost of substrates continues to pose a major challenge to their commercial viability [2]. To address this issue, genome optimization of the production strain is essential for developing efficient microbial chassis able to grow on cheap agro-resources. Nevertheless, the vast design space can hinder the timely identification of optimal genetic modifications. This study aims to enhance existing biocomputational tools for the systematic genome editing of B. subtilis, with the goal of engineering a lipopeptide overproducing strain capable of utilizing cost-effective xylose-rich renewable agro-industrial feedstocks such as miscanthus or corn straws. Materials and Methods The relevant metabolic pathways were modeled using XML-based code and Petri Net graphical computing frameworks. Computational analysis to identify viable solutions compatible with both surfactin overproduction and xylose utilization was conducted in accordance with constraint-based method with graphical abstraction of a metabolic model [3]. Genome editing was performed to obtain knock-out (KO) and knock-up (KU) modifications. The resulting mutant strains were cultivated in shake flasks to evaluate the accuracy of in-silico predictions. Lipopeptide production was subsequently quantified using UPLC-MS. Results, Discussion and Conclusions The introduction of gene knock-up led to a redesign of the prediction tool, integrating new Boolean constraints. Moreover, the models have been further adapted for target computation under steady state conditions. The tool enabled the identification of more than 61 non-redundant candidate genes for genetic modifications (KO or KU) in the seven newly modeled metabolic networks, all compatible with the conversion of xylose into lipopeptides or their precursors. The proposed solutions are supported by consistent metabolic logic and have been experimentally validated, for instance, through the KO of genetic targets such as spxA, alsR, or the KU of other targets such as the operon Ilv-Leu. However, the tool shows limitations when dealing with global metabolic regulators involved in multiple pathways. A revision of the prediction assumptions could improve the results. These findings validate the efficacy of the partial kinetic based prediction method to determine suitable genetic targets based on formal models.