Oxygen Uptake Rate Soft-Sensing via Dynamic kLa Computation: Cell Volume and Metabolic Transition Prediction in Mammalian Bioprocesses

Abstract

In aerobic cell cultivation processes, dissolved oxygen is a key process parameter, and an optimal oxygen supply has to be ensured for proper process performance. To achieve optimal growth and/or product formation, the rate of oxygen transfer has to be in right balance with the consumption by cells. In this study, a 15 L mammalian cell culture bioreactor was characterized with respect to k L a under varying process conditions. The resulting dynamic k L a description combined with functions for the calculation of oxygen concentrations under prevailing process conditions led to an easy-to-apply model, that allows real-time calculation of the oxygen uptake rate (OUR) throughout the bioprocess without off-gas analyzers. Subsequently, the established OUR soft-sensor was applied in a series of 13 CHO fed-batch cultivations. The OUR was found to be directly associated with the amount of viable biomass in the system, and deploying of cell volumes instead of cell counts led to higher correlations. A two-segment linear model predicted the viable biomass in the system sufficiently. The segmented model was necessary due to a metabolic transition in which the specific consumption of oxygen changed. The aspartate to glutamate ratio was identified as an indicator of this metabolic shift. The detection of such transitions is enabled by a combination of the presented dynamic OUR method with another state-of-the-art viable biomass soft-sensor. In conclusion, this hyphenated technique is a robust and powerful tool for advanced bioprocess monitoring and control based exclusively on bioreactor characteristics.

PEI-Mediated Transient Transfection of High Five Cells at Bioreactor Scale for HIV-1 VLP Production

“…Glucose, lactate and phosphate concentrations were measured with an ion-exclusion liquid chromatographic method using a sulfonated polystyrene divinyl benzene column (Aminex HPX-87H, Bio-Rad, Hercules, CA, USA) in an Agilent 1200 series HPLC system (Agilent, Santa Clara, CA, USA). A 0.01 N H 2 SO 4 solution was used as the mobile phase with a flow rate of 0.45 mL/min [29]. All measurements were performed with an AZURA UV/VIS detector (Knauer, Berlin, Germany) with a refractive index detector temperature of 35 • C. The standard deviation of the technique was determined as 0.31% for glucose, 0.26% for lactate and 1.01% for phosphate measurement.…”

“…The flow rate was adjusted to 0.64 mL/min and two solvents (solution A and B) were used in the mobile phase. Solution A consisted of 10 mM K 2 HPO 4 and 10 mM K 2 B 4 O 7 and solution B of a 45/45/10% v/v/v mix of acetonitrile, methanol and water, respectively [29]. Amino acids were detected at 266/305 nm for fluorenylmethoxycarbonyl derivates and at 450 nm for o-phthalaldehyde derivates.…”

Bioprocess Control: Current Progress and Future Perspectives

“…Artificial neural network Monitor fermentation process-measurement of glycerol, 1,2 propanediol and biomass [116] Multiple linear regression in combination with mechanistic model Prediction of biomass concentration [117] Empirical models Biomass concentration [118] Empirical modelling of oxygen uptake rate Predicted viable biomass [119] Multivariate adaptive regression spline algorithm in combination with 2D fluorescence spectra and process data…”

Challenges in the Development of Soft Sensors for Bioprocesses: A Critical Review

“…Process knowledge can be implemented into the soft sensor model. Knowledge-based model parts such as first-principle models (Ohadi et al, 2015;Steinwandter et al, 2017;Pappenreiter et al, 2019;Tahir et al, 2019;Krippl et al, 2021) can be employed to develop a more accurate and robust model. Process knowledge in the form of linguistic expressions can be digitalized using approaches based on fuzzy logic, as reviewed by Birle et al (2013).…”