Ensemble modeling for strain development of L-lysine-producing Escherichia coli

One of the main strategies to improve the production of relevant metabolites has been the manipulation of single or multiple key genes in the metabolic pathways. This kind of strategy requires several rounds of experiments to identify enzymes that impact either yield or productivity. The use of mathematical tools to facilitate this process is desirable. In this work, we apply the Ensemble Modeling (EM) framework, which uses phenotypic data (effects of enzyme overexpression or genetic knockouts on the steady-state production rate) to screen for potential models capable of describe existing data and thus gaining insight to improve strains for l-lysine production. Described herein is a strategy to generate a set of kinetic models that describe a set of enzyme overexpression phenotypes previously determined in an Escherichia coli strain that produces increased levels of l-lysine in an industrial laboratory. This final ensemble of models captures the kinetic characteristics of the cell through screening of phenotypes after sequential overexpression of enzymes. Furthermore, these models demonstrate some predictive capability, as starting from the reference producing strain (overexpressing desensitized dihydrodipicolinate synthetase (dapA*)) this set of models is able to predict that the desensitization of aspartate kinase (lysC*) is the next rate-controlling step in the l-lysine pathway. Moreover, this set of models allows for the generation of further targets for testing, for example, phosphoenolpyruvate (Ppc), aspartate aminotransferase (AspC), and glutamate dehydrogenase (GdhA). This work demonstrates the usefulness, applicability, and scope that the Ensemble Modeling framework offers to build production strains.