Integrated multi-dimensional modeling of non-model bacteria identifies engineering targets for acarbose biosynthesis optimization.

Abstract

Metabolic engineering for high-value compounds such as acarbose, a diabetes drug, requires systematic understanding of metabolic regulation. Here, we applied a multi-dimensional systems biology approach in Actinoplanes sp. SE50/110, a non-model acarbose-producing bacterium. We reconstructed an improved genome-scale metabolic model (iASE1267) with expanded metabolic coverage and a MEMOTE score of 80%, enabling more accurate phenotype predictions. Using a dual-objective OptRAM strain design strategy, we identified two sets of static engineering targets, including AcbR overexpression with adenylosuccinate lyase repression, and overexpression of dTDP-glucose 4,6-dehydratase with repression of 4-(cytidine 5'-diphospho)-2-methyl-D-erythritol kinase. Time-course metabolic modeling further revealed dynamic metabolic valves-ASPO1, PC, and PYK-governing flux redistribution. Integrating these targets, we reconstructed a core transcription-metabolism network and identified two pleiotropic negative transcription factors (TFs). Experimental validation of these TFs and metabolic genes increased acarbose titers by 18%-23%. This work establishes a framework integrating static/dynamic metabolic modeling with transcriptional networks for engineering non-model microbes.