(27bu) Framework for Optimizing Modulations of Enzyme Expression Levels and Kinetic Parameters for Computational Microbial Strain Design

Strain design algorithms have largely relied on linear algebraic models, containing reaction stoichiometry and thermodynamic constraints, to suggest genetic interventions that maximize the production of a specified target. Contrarily, kinetic formalisms of metabolism connect metabolic fluxes to the underlying physical states of enzyme levels and metabolite concentrations and encode allosteric regulatory interactions. Such kinetic models generally require in vivo data for parameterization and recent advances have made this process more practical. Here, we describe a framework that uses parameterized large-scale kinetic models of metabolism to generate microbial strain design strategies for targeted metabolites. Specifically, this method grounds the designs to the physical states of enzymes via modulating enzyme levels or kinetic elementary parameter values. These suggestions can then be interpreted as changes in protein expression or modifying enzyme physical properties including substrate binding affinity, turnover and/or regulatory interactions. The framework uses an interior-point method for nonlinear optimization. Application to case studies in Escherichia coli, Clostridium thermocellum, and Saccharomyces cerevisiae are explored. These case studies compare the suggested designs using the framework with those obtained by other strain design tools, such as using OptForce which operates only on stoichiometric models. The impact of including additional criteria for the solutions, such as restricting the extent of increased expression and the number of allowed interventions, is further examined. For example, for acetate production in E. coli, the framework suggested strategies capable of meeting 90% of maximum theoretical yield with less than 5-fold increases in specific enzyme levels over their wild type values (e.g., acetate kinase and aspartate transaminase) and decreases in expression (e.g., leucine transaminase). In summary, this tractable method enables proposing systems-wide modifications of metabolism satisfying varied metabolic engineering design goals.