Session 10 - Theme: Bioreactors and Bioprocesses (2)

Transient gene expression (TGE) has evolved from a laboratory-scale tool for candidate screening to a rapid platform for preclinical manufacture of recombinant proteins and viral vectors. In TGE, plasmid DNA is complexed with a cationic transfection reagent to form nanoscale DNA-polymer aggregates (i.e., polyplexes). Polyplex formation is governed by a temporal incubation window during which size, charge, and colloidal stability are established prior to introduction into the bioreactor, where cellular uptake and expression occur. While reported titres at laboratory scale have increased from mg/L to g/L over the past decade via biological optimizations (e.g., host cell lines, vector and promoter design, and DNA-to-reagent ratios) (1), industry reports document losses in titre and specific productivity when scaling up from 7 L stirred-tank reactors (STRs) and 22 L rocking bioreactors to 50 L and 200 L STRs (2). This study addresses the hypothesis that transient transfection performance is governed by mixing-controlled exposure histories during and immediately following DNA-polymer complex addition and the finite stability window of polyplexes. Therefore, an engineering characterization was performed in two representative scale-down systems: an ambr® 250 STR and a 2 L rocking bioreactor (CultiBag®). Volumetric power input (P/V), power number, Reynolds number (Re), flow regime, mixing time, and mixing number were quantified across a wide range of operating conditions relevant to industrial TGE. In the STR, mixing time measurements were extended beyond conventional dual-indicator system methodologies by employing transfection-representative addition volumes (> 1 % compared with traditional tracer injections of 0.05 % in < 1 L systems). In parallel, rocking bioreactor hydrodynamics were characterized to resolve free-surface-driven mixing, volume-dependent flow transitions intrinsic to wave-induced systems, and to develop a novel Re formulation appropriate for motion-driven flow. The results demonstrate that matching global metrics such (e.g., P/V) is insufficient for TGE scale-up. Instead, mechanistic scaling criteria must explicitly account for addition-driven mixing dynamics, bioreactor geometries, and transient exposure histories.

Cultivation of mammalian cells is typically performed in stirred and aerated bioreactors, where cells are exposed to complex hydrodynamic forces generated by mixing and aeration. While adequate mixing is essential for efficient heat and mass transfer, maximum hydrodynamic stress (τmax) can impact cell viability and productivity. In this study, shear-stress-sensitive aggregates, were employed as a low-cost and universal probes for τmax characterization in stirred bioreactors. Due to small volume of the tested ambr®250 mL bioreactor system we utilized optical microscopy image analysis for aggregate size determination. An optical flow-through cell was designed to enable continuous sampling and was tested using the high-throughput ambr®250 mL bioreactor system equipped with multiple impeller configurations. Obtained results of the hydrodynamic stress will be compared with the computational fluid dynamics (CFD) simulations utilizing lattice-Boltzmann LES method covering both single and multiphase flow conditions.

Whilst the fundamentals of conventional stirred tank mixing processes have received great attention over the last 70 years, there have been very few studies devoted to vibromixers – which mix with an oscillating plate – despite the fact that they have been applied in industry for over 30 years. Indeed, there is almost no fundamental knowledge of flow and mixing performance of vibromixers (particularly high-frequency/low-amplitude devices) documented in the literature. This work experimentally characterizes the hydrodynamics and mixing performance of a vibromixer, operating at 100 Hz and amplitudes of a few millimeters. Experiments have been performed with a Fundamix vibromixer with a perforated oscillating plate in a cylindrical dish-bottomed glass tank (T=0.15m, H=T). Several diameters of vibromixer plates (D/T=0.3–0.43) were used and at different off-bottom clearances (C/T=0.33–0.5). PIV was used to measure velocity fields and to estimate turbulence kinetic energy and shear rate. Power consumption was determined using a piezoelectric force sensor. Mixing time was evaluated using the coloration-decolorization technique. PIV results show the vibromixer creates one circulation loop similar to a rotating axial flow impeller in turbulent flow whilst shear rates are low (< 0.05 s-1). The power number follows a similar trend to conventional stirred tanks whereby it is constant in turbulent flow and inversely proportional to Reynolds number in laminar flow. However, specific power input is of the same order of magnitude or less than a stirred tank. Mixing times are short compared with a turbulent impeller stirred tank.

The industrial-scale production of adeno-associated virus (AAV) by transient transfection of HEK293 cells remains a key challenge for gene therapy manufacturing. Scaling from shaken culture systems to stirred tank bioreactors introduces changes in hydrodynamic conditions, where biological and engineering parameters influence process robustness and productivity. Few studies investigate these engineering factors, making detailed understanding essential for successful scale-up. This work presents the characterisation of the Xcellerex™ Xplatform reactor (Cytiva) equipped with the 6B-R50 impeller. A 1 L geometrically scaled-down model of the 50 L system is evaluated across a range of working volumes, agitation speeds, vessel configurations and impeller rotation directions. Emphasis is given to the effective baffle volume immersed in the liquid, as this parameter influences vortex formation. Experiments demonstrate that the 6B-R50 impeller exhibits power numbers of 1.7 and 2.4 under anticlockwise and clockwise rotation, respectively. These values align with those reported for three-pitched blade impellers, supporting its applicability for mammalian cell culture. Reduced baffle submergence leads to earlier vortex formation and increased free-surface deformation, conditions that may negatively impact mammalian cell cultures due to hydrodynamic stress. Power consumption is determined using the air-bearing technique, while mixing and flow dynamics are assessed by the Dual Indicator System for Mixing Time and particle image velocimetry, PIV, respectively. This study presents the first detailed characterisation of the Xplatform bioreactor. The findings support rational process design and scale-up strategies, providing a solid foundation for viral vector production in stirred tank bioreactors.

Addressing mixing issues in an industrial bioreactor requires describing the variability within cellular populations as well as the two-way coupling between cells behavior and concentration gradients developing in large-scale aerated, stirred reactors. A multivariate population balance model is used to discriminate individuals according to their properties and, coupled with mass balances. It provides a framework for describing the dynamics of cell populations driven by internal noise and external forcing. We developed a particle-based Monte Carlo code to address these challenges in a general way [1]. It combines a CFD-based compartment model for scalars, a stochastic cell transport model and particle-based methods for biological source term calculations. Cell division and the partitioning of properties at division are treated explicitly and act as a source of diversification. Mixing, gas–liquid mass transfer, population dynamics over time and space, uptake adaptation, metabolically governed cell division, and cell elongation processes are included [2]. We validated the code through a comparison with available analytical solutions [3] considering length, age, instantaneous elongation rate, and metabolic reaction rates as cell properties. With this tool, we predict the emergence of sub-population with distinct metabolic behavior as a consequence of the exposure to heterogeneous concentration fields. [1] J. Morchain, C. Mayorga, P. Villedieu, A. Liné, A dynamic compartment model for spatially heterogeneous reactors: Scalar and Monte-Carlo particle mixing, https://doi.org/10.1016/j.cherd.2024.04.014. [2] K. Biswas, N. Brenner, Universality of phenotypic distributions in bacteria, https://doi.org/10.1103/PhysRevResearch.6.L022043. [3] J. Cullum, M. Vicente, Cell growth and length distribution in Escherichia coli, https://doi.org/10.1128/jb.134.1.330-337.1978.

Predicting heterogeneities in bioreactors is crucial for scaling up fermentations, as spatial gradients can strongly influence microorganisms and fermentation performance. Although Computational Fluid Dynamics (CFD) coupled with biokinetic models can capture these effects, such simulations are computationally intensive and expensive [1-3], limiting their use for rapid assessment of dynamically operated systems such as fed-batch fermentations. Compartment models (CM) provide an attractive alternative, offering a computationally efficient yet sufficiently detailed representation of hydrodynamics and mass transport in stirred tank bioreactors [4-7]. When combined with machine-learning techniques, these models can further capture the impact of varying operating conditions without relying on CFD-derived flow fields [8]. In this presentation, we discuss our approach to integrating compartment models with machine learning to construct spatially resolved surrogate models for exploring bioreactor design and operation. Depending on the application, supervised or unsupervised methods can be used – either to approximate the flow field and decouple hydrodynamics from biokinetics [8], or to directly infer substrate gradients and metabolic regimes resulting from their interaction [9-10]. Furthermore, deep learning enables the incorporation of categorical variables, expanding the range of design and operational scenarios that can be explored. Our surrogate models achieve a runtime reduction of up to three orders of magnitude compared to a fully coupled CFD model, while maintaining resolution suitable for industrial decision-making. We also highlight key considerations for deployment, including enforcing linear constraints to ensure mass conservation and mitigating spurious predictions outside expected operating ranges.