Chinese Hamster Ovary (CHO) cells are an invaluable cell line for robust and reliable production of therapeutic proteins in the biopharmaceutical industry. There is significant pressure on the industry to accelerate process development to increase productivity and economics, but conventional empirical strategies, which are time-consuming and labor-intensive, are not well-suited to meet the growing demand for biologics and requirements to lower the cost of goods sold. Recent advances in -omics technologies, such as genomics, transcriptomics, proteomics, and metabolomics, that effectively link the genotype to the phenotype of an organism has greatly increased our understanding of CHO cells at a cellular level. Genome-scale models of CHO metabolism can be used for guiding effective process optimization, CHO cell engineering efforts, and bioreactor monitoring and control.
Published in Biotechnology & Bioengineering, “A genome-scale nutrient minimization forecast algorithm for controlling essential amino acid levels in CHO cell cultures,” authors describe a newly developed genome-scale model that can forecast key CHO cell process parameters including cell growth and essential amino acid consumption to guide cell culture media optimization. Knowing how nutrients are utilized by the cells can ultimately improve bioprocess control through the optimization of media, feeding strategies, and other process parameters to maintain optimal nutrient profiles to achieve higher process performance while controlling toxic metabolite accumulation in the spent media.
Recently, Michael Betenbaugh, Ph.D, and Yiqun Chen, Ph.D, from Johns Hopkins University (JHU) joined Ji Young Anderson and Milla Neffling, Ph.D, from 908 Devices for a dynamic discussion to learn more about the genome-scale model to predict amino acid levels and the critical role of new analytical tools like the REBEL analyzer (908 Devices, Inc.) in model generation.
Watch the Dynamic Discussion
Listen to this 20-minute roundtable discussion hosted by 908 Devices on the highlights of a new peer-reviewed article exploring model-based approached to cell culture optimization
The authors investigated how a genome-scale model can help to control a process and to design feeding strategies that lower production costs by utilizing a nutrient minimization strategy. First, the model was shown to accurately predict key nutrient levels by the simple measurements of viable cell density. This model was tested and validated by daily measurements and was used to limit the concentration of three key nutrients: the essential amino acids Leucine, Valine and Lysine.
The experimental design goal was to achieve the same growth and titer in the control, as in the nutrient minimization experiment. The growth patterns were nearly identical in the two experiments, whereas the titer was slightly lower in the nutrient minimization experiment. The authors discuss how the model can be further refined by always including the amino acid measurements when refining the model to specific cell lines and bioprocesses therefore highlighting the need for rapid at line amino acid measurements.
Nutrients, such as amino acids, must be supplied through the culture media to maintain growth and productivity. However, optimizing essential amino acid levels for a given cell line can be a challenge—too much or too little can have a negative impact on cell growth, and ultimately process productivity. Maintaining amino acid levels at their optimum by preventing over- or under-feeding of the culture contributes to better amino acid utilization for maximum productivity and can help reduce operational costs.
The development of models is a data-intensive endeavor, and the industry has struggled with access to high-quality datasets, which has been a stumbling block in the past for model development. New analytical tools like the REBEL analyzer solves the data acquisition issue by bringing amino acid analysis to the point-of-need in the lab allowing for faster and more frequent measurements. Streamlining the amino acid data analysis allowed the authors access to the necessary data required to support the larger initiative of developing these predictive models.
Overall, the authors found that the forecast model accurately predicted the concentration of essential amino acids based on viable cell density (VCD) measurements. This finding holds great potential to enhance process control by guiding media composition or feed control strategies for CHO cell-based bioprocesses. This model supports a nutrient minimization approach to bioprocess optimization that can lower nutrient cultivation costs and limit accumulation of undesirable metabolic by-products that negatively impact performance.
While no model is perfect, as evidenced by minor deviations between observed and model-predicted concentrations, the authors emphasize that the expansion of at-line measurement with tools like the REBEL allow for continual updating, refining, and adapting of the model. Continuous improvement of the model fit helps increase the accuracy and forecast reliability.
“Instruments like the REBEL and genome-scale models form a critical partnership that will become increasingly important in this data-driven generation. They work together to provide better understanding of systems like CHO cells that can also be applied to other systems like cell and gene therapies across a wide spectrum of technologies.” said Mike Betenbaugh, AMBIC Director and JHU Professor of Chemical and Biomolecular Engineering.
Currently, the model can be accessed through a graphical user interface developed with the Advanced Mammalian Biomanufacturing Innovation Center (AMBIC). Users can use the model to predict the consumption levels of up to twelve essential amino acids in CHO cell-based bioprocesses.