(468c) Towards Accurate Prediction of Cell Culture Dynamics in Bioreactor Digital Twins Using Segregated Models.

The industrial production of high-value biotherapeutics from mammalian cells is a complex process that relies on a sequence of unit operations, including bioreactors, centrifuges, membranes, and chromatography. Bioreactors are a core component of this process, and their performance is critical for efficient production. In this regard, bioreactor digital twins can help improve process performance and facilitate complex tasks such as process scale-up and technology transfer. However, to unleash their potential, the model's predictive capabilities need to be adequate and the uncertainty low. One of the steps to improve predictive capabilities and reduce uncertainty is enhancing the underlying mechanistic model, which is the aim of this work and will be pursued by considering a segregated model for proliferating and quiescent cells.

Cell populations are subject to changes in environmental conditions during the cultivation stage, which may lead to alterations of the metabolic pathways normally adopted during the exponential growth phase. These alterations may appear at macroscopic levels through changes in the uptake rate of nutrients, metabolite secretion, and in productivity of a target molecule. For example, it is well established that shifts in lactate metabolism may happen when cells enter the stationary phase, switching from net production to net lactate consumption within the cultivation system [1].

The cell cycle for mammalian cells has been studied extensively, and there is general agreement on the existence of distinguished phases (G0/G1, S, G2M), which constitute the whole series of processes that lead to cell proliferation. The cell cycle distribution is vital when it comes to enhancing productivity in relevant biotherapeutics production systems such as CHO cells culture. One of the main strategies adopted to enhance it relies on interfering with the cell cycle to arrest the cell growth and preferentially adopt the pathways leading to biotherapeutics production [2].

Various strategies for the cell growth arrest for enhanced recombinant protein production have been studied, such as low-temperature exposure, sodium butyrate, or other molecules that can target specific sites that interfere with the normal execution of the cell cycle stages. Some observations suggested that the growth arrest may correspond to the cells leaving the proliferating phase and entering a quiescent state G1/G0, in which the cells shift metabolism, preferring pathways leading to the recombinant protein of interest [3]. Several biological models have been proposed to explain the onset of the quiescent state G0/G1. The main three categories are the 'transition probability (TP)', the 'growth controlled (GC)', and the 'sloppy size control (SSC)' model. Focusing on the transition probability model, it considers a reversible transition between the non-replication (NR-phase, when cells are quiescent or resting) and replication (R-phase, when cells are proliferating or cycling) phases, establishing a probability that one of the two events may happen. Specifically, the model attributes the overall growth rate variation to the proliferating cells moving to the quiescent phase[4]. Fed-batch experiments reported in the recent literature show that cells tend to synchronise into a presumed G0 state as the cultivation progress towards the late stage and stationary phase of the growth. In these experiments, the online measurements using a FUCCI fluorescence with I_Red were adopted to estimate the cell cycle distribution during the cultivation [5].

This work presents a segregated mechanistic model considering a 'transition probability' between proliferating (G1, S, G2/M) and quiescent cells (G0) that allows modulating the cell growth rate based on the transition of the cells from proliferating to quiescent and vice versa. This new segregated model was developed in gPROMS FormulatedProducts® 2022.1 and included in the built-in Upstream Bioprocessing libraries (Siemens, UK). As shown in Fig. (1a), the bioreactor model is connected to a feed model to have the typical representation of a fed-batch system. The model includes the capability to consider the level of cell segregation shown in Fig. (1b). The transition between viable cycling and viable resting cells, and vice versa, are modulated by a transition rate function which establishes a probability for the cells to leave and renter the proliferating phase. The transition rate probability depends on the concentration of a selected pool of nutrients and metabolites, which can limit or inhibit the reversible transition between the two phases.

Fig. (2a) shows the dynamic of the proliferating and quiescent cells, demonstrating as the concentration of quiescent cells increases going towards the late and stationary phase of the growth. At the same time, it can be observed as feed pulses temporarily reverse the net direction of the cell transition, demonstrating the reversibility of the transition. Fig. (2b) shows the experimental dynamic of the VCC, which includes both proliferating and quiescent cells. The model demonstrates that the transition into a quiescent state can modulate the growth rate, allowing experimental results to be reproduced and opening the door for further improvement in the mechanistic modelling of CHO cell bioreactors.

Furthermore, the level of segregation enables us to consider different levels of productivity and nutrient consumption during the two cell cycle phases (Data not shown). The results demonstrate that enhancing the mechanistic description of the system can help reduce the gaps between the mathematical model and experimental results, providing a more accurate tool for the definition of underlying phenomena.

References

[1] F. Zagari, M. Jordan, M. Stettler, H. Broly, F.M. Wurm, Lactate metabolism shift in CHO cell culture: The role of mitochondrial oxidative activity, N. Biotechnol. 30 (2013). https://doi.org/10.1016/j.nbt.2012.05.021.

[2] Z. Du, D. Treiber, J.D. Mccarter, D. Fomina-Yadlin, R.A. Saleem, R.E. Mccoy, Y. Zhang, T. Tharmalingam, M. Leith, B.D. Follstad, B. Dell, B. Grisim, C. Zupke, C. Heath, A.E. Morris, P. Reddy, Use of a small molecule cell cycle inhibitor to control cell growth and improve specific productivity and product quality of recombinant proteins in CHO cell cultures, Biotechnol. Bioeng. 112 (2015). https://doi.org/10.1002/bit.25332.

[3] V. Avello, M. Torres, M. Vergara, J. Berrios, N.A. Valdez-Cruz, C. Acevedo, M.M. Sampayo, A.J. Dickson, C. Altamirano, Enhanced recombinant protein production in CHO cell continuous cultures under growth-inhibiting conditions is associated with an arrested cell cycle in G1/G0 phase, PLoS One. 17 (2022) 1–17. https://doi.org/10.1371/journal.pone.0277620.

[4] G. Yao, Modelling mammalian cellular quiescence, Interface Focus. 4 (2014). https://doi.org/10.1098/rsfs.2013.0074.

[5] J. Möller, K. Bhat, K. Riecken, R. Pörtner, A.P. Zeng, U. Jandt, Process-induced cell cycle oscillations in CHO cultures: Online monitoring and model-based investigation, Biotechnol. Bioeng. 116 (2019) 2931–2943. https://doi.org/10.1002/bit.27124.