(191de) Metabolic Pathway Engineering in Mammalian Cells through Kinetic Model Optimization

Fast growing cells, including stem cells, cancer cells, and many immortal cell lines, have a metabolic behavior known as aerobic glycolysis, characterized by a high rate of glycolysis and lactate production. This is in contrast to the metabolic behavior of many quiescent cells, which consists of a lower glucose consumption, low lactate production, and a larger reliance on oxidative phosphorylation for energy production. Altering the metabolic phenotype of aerobic glycolysis has been an active area of research in developing cancer therapies. In industrial cell culture, reducing lactate production resulting from rapid growth has been shown to enhance productivity. A novel strategy to alter cellular metabolic behavior may advance cancer therapy and cell culture bioprocessing.

The metabolic reaction network is highly nonlinear due to complex allosteric feedback and feedforward controls and different combinations of isoenzymes. A detailed kinetic metabolic model thus has a large parameter space with many possible combinations of enzyme kinetics and expression. This necessitates a method to intelligently search the parameter space for key changes which can be made to manipulate metabolism, as well as other biological reaction networks to specific ends.

A promising approach for deciding which enzymes and genes to experimentally manipulate is to embed a kinetic model into an optimization problem in order to target desirable modifications to metabolic behavior. We employed a kinetic model of metabolism which encompasses glycolysis, the pentose phosphate pathway, and the TCA cycle with the known allosteric regulations.

The model was solved using a local optimizer embedded in General Algebraic Modeling System (GAMS) to find the combinations of parameters that enable the target metabolic behavior with stated constraints. A multi objective optimization problem is also constructed to demonstrate the metabolic tradeoff between rates of glucose consumption and lactate production. Additionally, a penalty is imposed to changes from the original state incurred in the optimization process; the penalty increases with increasing magnitude of genetic alteration. This penalty also provides insight into the variables which have the most impact on the objective function. Varying the weight of this penalty allows for improvements in target metabolism to be balanced by the number of desired changes to the system, limiting the complexity of resulting experiments. The optimization schema was first applied to reduce lactate production while maintaining the energetic requirements for growth by allowing the expression level of multiple enzymes to vary. Then, the effect of knockdown or amplification of genes in the pathway were evaluated, and combinations of genetic alterations that can steer the metabolic behavior from the aerobic glycolysis type to the target type were identified.

Here we have demonstrated a method for rationally guiding cellular engineering through optimization. By identifying combinations of parameter changes which yield desirable outcomes, optimization of kinetic models can greatly reduce the effort required to engineer cell metabolism and provide deep insight into the reaction networks and their behavior.