A combined mathematical and experimental investigation of lactate metabolism in industrial CHO cell cultures

The shift in lactate metabolism of industrial mammalian cell lines is a widely studied phenomenon that has yet to be fully understood. Many mammalian cell lines, including Chinese Hamster Ovary (CHO) cells, which are mainly used to manufacture therapeutic glycoproteins, exhibit an unusual shift in lactate metabolism when grown in batch or fed-batch suspension cultures. Understanding the underlying mechanism of this metabolic switch and whether it is beneficial in terms of productivity, might open up large areas of currently infeasible process space for exploitation. Herein, a combined experimental and computational approach utilizing Flux Balance Analysis (FBA) coupled with Multivariate Data Analysis (MVDA) was developed. The latest available CHO genome-scale metabolic model (iCHO1766) was studied in detail, in order to gain further insights into the metabolic effects that cause this shift in lactate metabolism and its impact on culture performance. Experimental data pre- and post-lactate shift were collected from batch and fed-batch cultures across multiple cell lines and were used to constrain the model. A novel carbon-based constraining algorithm was developed and employed in order to rationally and systematically refine the accuracy of the model’s predictions. Using principal components analysis (PCA) and partial least squares discriminant analysis (PLS-DA) well defined and clearly separated clusters in the solution space corresponding to the lactate producing (LP) and lactate consuming (LC) metabolic states were identified. Furthermore, through analysis of the PC loadings and variable importance in projection (VIP) scores key reaction groups whose activities differed significantly between the two metabolic states were identified. These key metabolic differences were used to formulate hypotheses regarding potential mechanisms and actuators of the switch in lactate metabolism. Based on the computational analysis, experiments exploring enzymatic inhibitors, additional media components, and the introduction of a non-native enzyme into the CHO genome were performed in an effort to test the initial hypotheses.