Maximum entropy and population heterogeneity in continuous cell cultures

Mammalian cell cultures in continuous bioreactors exhibit complex behaviors, including nonlinear dynamics like multi-stability and hysteresis. Prior research has shown that mathematical models can propose control strategies for these systems. However, most models assume population homogeneity, despite known variability in clonal populations. This gap motivated the development of a new approach to capture phenotypic diversity. No prior work had resolved how to model this heterogeneity while accounting for metabolic coupling and toxic byproducts. Existing models lack the ability to describe sharp transitions between metabolic states. The assumption of homogeneity may overlook critical system behaviors. This uncertainty drove the use of a new modeling framework. The need to incorporate cell-to-cell variability into predictive models has remained unmet. Understanding how heterogeneity affects system dynamics could improve bioprocess control.

The aim of this study is to model phenotypic heterogeneity in continuous mammalian cell cultures using the maximum entropy principle. The specific problem is the assumption of population homogeneity in current models, which may not reflect real-world variability. The motivation is to better understand how heterogeneity affects system dynamics and control strategies. The study addresses how toxic byproduct accumulation impacts cell viability and system behavior. The researchers propose to use a statistical framework to describe the distribution of cellular phenotypes. This approach allows for the inclusion of extracellular variables in the model. The goal is to provide a formal solution for the stationary state of the chemostat. The study also seeks to demonstrate the model's application in two distinct metabolic scenarios.

The study applies the maximum entropy principle to describe phenotypic distributions in a chemostat. The model couples cell metabolism with extracellular variables representing the bioreactor state. The researchers account for the impact of toxic byproduct accumulation on cell viability. They derive a formal solution for the stationary state of the system. The approach is tested using a simplified metabolic model with a tractable solution. A genome-scale metabolic network of CHO cells is also used as a second example. The model incorporates nonlinear dynamics such as multi-stability and hysteresis. The study evaluates how heterogeneity influences system behavior and byproduct concentrations.

The model demonstrates that heterogeneity can cause qualitative changes in the system's dynamical landscape. Byproduct concentrations increase in heterogeneous populations compared to homogeneous ones. The stationary state solution reveals how dilution rate affects phenotypic distribution. The simplified model shows how the maximum entropy framework captures metabolic transitions. The genome-scale CHO model confirms the framework's applicability to complex networks. Heterogeneity leads to larger population sizes under certain conditions. The model predicts that toxic byproducts may accumulate in heterogeneous cultures. These findings suggest that heterogeneity significantly impacts system behavior.

The authors propose that the maximum entropy principle provides a useful framework for modeling population heterogeneity. They suggest that heterogeneity can alter the system's dynamical landscape and byproduct accumulation. The model may help explain sharp transitions between metabolic states in continuous cultures. The study shows that heterogeneity can increase population sizes in certain conditions. The researchers propose that this approach improves the accuracy of bioprocess models. The findings may inform strategies for controlling bioreactor dynamics. The model's application to both simplified and genome-scale networks supports its utility. The study highlights the importance of accounting for phenotypic diversity in continuous cell cultures.