A novel dynamic flux balance analysis for modeling CHO cell fed-batch cultures with pH and temperature shifts

While Dynamic Flux Balance Analysis provides a powerful framework for simulating metabolic behavior, incorporating operating conditions such as pH and temperature, which profoundly impact monoclonal antibodies production, remains challenging. This study presents an advanced dFBA model that integrates kinetic constraints formulated as functions of pH and temperature to predict CHO cell metabolism under varying operational conditions. The model was validated against data from 20 fed-batch experiments conducted in Ambr®250 bioreactors. To mitigate overparameterization, a bi-level optimization approach utilizing the Bayesian Information Criterion was employed to systematically identify the most effective kinetic constraints. This optimization reduced the number of parameters (from 253 to 205) while improving predictive accuracy by up to 8.3% for training and 2.68% for validation datasets. The results highlight the model’s ability to predict cell growth, titer, and also capture metabolic shifts, including glucose, lactate, and ammonia metabolism and amino acid utilization, across different temperature and pH conditions with high predictive precision (average $R^2\ge 0.97$ for cell growth and titer and average $R^2\ge 0.85$ for other metabolites). This optimized dFBA framework offers a robust tool for studying model-based optimization for CHO cell metabolism, identifying optimal operating conditions to balance growth and productivity.