With scalable process prospective, MSC cultures on suspended microcarriers are consid-ered as best choice for larger cell production. However, particle properties and interactions with the hydrodynamics must be intended to improve cell adhesion and expansion, to limit potential cell damages (cell death, detachment or diﬀerentiation) and to allow an eﬃcient cell detachment. Cultures are generally performed at the just-suspended state, ensur-ing all particles suspension, to limit the hydromechanical stress generated. This section will first present microcarriers specific properties and cell adhesion mechanisms in order to introduce the agitation impact on cell growth. An outlook of the microcarrier-based MSC cultures and the current eﬀorts will be then presented. Finally, just-suspended state relevancy and determination will be here developed in the case of stirred tank bioreactors.
WJ-MSC microcarrier-based cultures using hPL
In the case of WJ-MSC, only two studies have been conducted on microcarriers, in a culture medium supplemented with hPL, adapted for clinical use. Petry et al. com-pared five diﬀerent microcarriers (Hillex II, ProNectin F-coated, Plastic, Plastic-Plus and Glass-coated) in static mode and validated the dynamic culture in Spinner flasks with the one giving the best growth and attachment condition in static mode . In paral-lel, De Soure et al. validated cells isolation and also dynamic cultures in spinner flasks with Plastic microcarriers precoated with a culture medium supplemented with 50 % hPL . However, such conditions have only been validated at the laboratory scale (until 100 mL), and comparison of microcarriers have only been performed in static mode for WJ-MSC.
Particles just-suspended state
Microcarrier suspension in stirred tank bioreactors is fundamental for MSC expansion. The agitation must be suﬃcient to allow culture medium homogenization and avoid O2, pH and nutrient gradient limitations, to enhance mass transfer, and to maximize the available liquid-solid interfacial area. Hence, suspension of all particles is required to avoid microcarrier settling in the bottom and aggregation . However, a too vigorous agitation may generate damages to the cells through cell lysis, cell detachment or by inducing their diﬀerentiation [153, 154]. One reasonable strategy to get a suﬃcient mixing performance and minimization of hydromechanical stress is to operate at the impeller just-suspended agitation rate Njs at which complete beads suspension is ensured (Fig. 1.23). This means that bioreactor hydrodynamics must therefore be intensively studied to guarantee cells viability and allow a process scale-up, leading to the need of reliable models.
The critical agitation rate Njs is function of the diﬀerent parameters, presented in Fig. 1.22. In the case of microcarrier suspension in small scale bioreactor, no sparging is necessary. Table 1.7 presents the impacts of the diﬀerent parameters, for a given vessel shape. In addition, the impeller design significantly impact Njs value, according to the fluid macroscopic flow pattern in the vessel. As presented in Fig. 1.24, axial flows are composed of large circulation loops, generally over the liquid height (axial downward pumping impellers), radial flows involve four circulation loops and often lead to fluid compartments, and mixed-flows involve a large circulation loop in the opposite direction of the one generated by axial flows and a potential small circulation loop above (axial inward pumping impellers). In the first case, in down-pumping mode.
Hydromechanical stress and turbulence description
As previously presented, agitation is required to fully suspend microcarriers in the bio-reactor. It is nevertheless source of hydromechanical stress leading to potential damages to cells adhered the particle surfaces. MSC cultures are mostly performed in hypoxia and O2 consumption is suﬃciently slow to ensure suﬃcient dissolved oxygen in the reactor only by surface aeration for small scales. Lavrentieva et al. reported a O2 consumption of 0.024 ± 0.002 and 0.095 ± 0.005 pmol h−1 cell−1, respectively in hypoxia (1.5 % O2) and norxia (21 % O2) . Although dissolved oxygen in large-scale cultures should be con-trolled to avoid O2 limitations, it has been shown by Heathman et al. that direct sparging at the just-suspended state was detrimental to BM-hMSC growth . So, unlike other mammalian cell cultures in STR, damages related to sparging will not be taken into ac-count in the present state of art, which will focus on damages due to the agitation only. Before describing cell physiological and performance responses to hydromechanical stress encountered in the bioreactor, impact of turbulence on suspended particles will be pre-sented. The diﬀerent length scales related to the turbulent phenomena are indeed crucial for understanding the link between turbulence and hydromechanical stress encountered by particles, related to the potential cell damages. Hence, the present section aims at describing the phenomena related to the turbulence and at drawing a parallel with their impact for MSC cultivated on suspended microcarriers at the diﬀerent turbulence length scales. Considering microcarrier and cell sizes, these length scales will help to determine the critical damaging scales. Figure 1.25 summarize the relationship between the turbu-lence and the impact on microcarriers and cells, which will be developed in more details in the following sections.
Hydrodynamics characterization in STR
As described by Henzler et al., particle stress is a consequence of a velocity diﬀerence between the particles and the fluid . In laminar flow, the stress τ is calculated using the Newton’s law in Eq. 1.9. However, in the case of turbulent flow, the relative velocity that determines this stress comes from the turbulent fluctuations. The Reynolds decomposition allow to define the velocity field of a flow u as the sum of the mean velocity u¯ and a fluctuating term u0. The Reynolds stress tensor τt is then calculated, using the turbulent fluctuation velocity √u02, as presented in Eq. 1.10. Stress determination is dependent on the turbulent energy dissipation rate ε and the Kolmogorov length scale λK , which allow to define the turbulent range. Relationships are presented in Fig. 1.28 for the diﬀerent ranges. Regarding these results, and independently of the flow regime, the stress applied on particles τ is only related to the turbulent energy dissipation rate ε.
dx τ = µ du (1.9).
u02 τt = ρ√ (1.10).
Turbulence analysis is classically performed with a single phase, although it is known that particle suspensions may aﬀect the hydrodynamics. In the case of the STR, two regions may be distinguished: (i) the impeller region and (ii) the rest of the vessel. Shear stress is much higher around the impeller, with local turbulent dissipation rates ε up to 100 times the mean value in the vessel < ε >, while ε can lead to only 0.25 × < ε > in the rest of the vessel . The energy dissipation rate is mostly used to characterize the hydrodynamics. In contrast to the agitation rate, it allows the comparison of diﬀerent bioreactor geometries.
Fluid flow characterization using CFD
Computational Fluid Dynamics (CFD) may be used as a complementary approach to the experimental data. In the present case, CFD allows to better understand flow structures in bioreactors and their potential heterogeneity, aﬀecting cell culture eﬃciency. It is no-tably used to complete the limited data in the literature for fluid flow and shear stress in stirred bioreactors, particularly at the particle just-suspended state where the system is not homogeneous. Concerning bioreactor designs and scale-up, CFD enables to under-stand and compare diﬀerent design concepts from the bench to the industrial scale. So, for scale-up applications, the number of experiments at the pilot-plant level to valid or adapt a design may be reduced using preliminary numerical predictions.
CFD allows to solve the Navier-Stokes equations, that govern fluid flow, to furnish quantitative analysis and prediction of fluid flow phenomena. Microcarrier suspension in stirred tank bioreactors involves two phases, namely the culture medium as the liquid phase and the microcarriers as the granular solid phase, whose interactions (drag force, turbulent dispersion) have to be modelled. Biological reactions may also be implemented to model cell growth or cell metabolic activity but, to our knowledge, it has never been applied for MSC cultures on microcarriers. Cell time scales (doubling time > 1 day) are indeed much larger than hydrodynamic phenomena time scale, and cell growth modelling is complex, due to a high sensitivity to the environment (biochemical concentrations, available sur-face, hydromechanical stress…) and to diﬀerences according to the cell source. Current works are mainly based on microcarrier dispersion characterization and distribution of the shear stress encountered by the particles. Single-phase velocity fields may be experi-mentally validated using PIV (Particle Image Velocimetry).
Number of studies conducted CFD simulations of solid-liquid suspensions in stirred tank bioreactors at or close to the just-suspended state with particles significantly heavier than microcarriers, like sand [187, 188, 155, 189, 190, 191]. On this basis, recent studies adapted the models for microcarrier suspensions in diﬀerent bioreactor design. They are summarized in table 1.12.
Cell detachment from microcarriers
The last criterion to compare microcarriers adequacy for MSC cultures is the ease of de-tachment. Results of the microscopic analysis performed on microcarriers recovered after filtration used for the cell-particles separation are presented in Fig. 2.22. Cells presented good detachments from Star-Plus, Plastic-Plus and Hillex II, with almost no cells ob-served on microcarriers. On the contrary, many cells remained attached on Cytodex-1.
Cell number and viability were controlled after cell detachment and filtration (Table. 2.10 and 2.11). In static mode, the cell viability was similar for all microcarriers, between 81 and 85 %. However, cells on Cytodex-1 presented lower viabilities of 83 % and 73 % after cultures with respectively both orbital and mechanical agitations, than those on Star-Plus and Plastic-Plus microcarriers. Concerning Hillex II, in stirred cultures, the detachment step damaged cells and led to cell viabilities of 61 % (orbital agitation) and 62 % (me-chanical agitation), explaining the poor number of cells counted on microcarriers of only 0.5 to 0.6 105 cells.
Table of contents :
1 State of the Art
1.2 Mesenchymal stem / stromal cells and their therapeutic benefits
1.3 MSC sources and population heterogeneities
1.4 Expansion process challenges
1.5 Microcarrier-based cultures
1.6 Hydrodynamics potential impact on MSC
1.7 Thesis aims and objectives
2 Improvement of WJ-MSC culture on microcarriers
2.2 Preliminary results
2.3 Development of an automatic cell counting method on microcarriers using image analysis
2.4 Impact of the type of microcarrier in different agitation modes on the expansion performances
2.5 Cell culture on home-made microcarriers
2.6 Chapter conclusions
3 Modelling of microcarriers just-suspended state in a stirred tank bioreactor
3.2 CFD models investigation for simulation of microcarriers suspension in stirred-tank bioreactors
3.3 Dimensional analysis and CFD simulations of microcarrier just-suspended state
3.4 Chapter conclusions
4 Optimization of the impeller design for MSC culture on microcarriers in a stirred tank bioreactor
4.2 Impeller design comparison according to the hydromechanical stress encountered by the microcarriers
4.3 CFD-based strategy to optimize the impeller design adapted to microcarrier suspension in a minibioreactor
4.4 WJ-MSC culture in STR using selected EE impeller designs
4.5 Chapter conclusions
Résumé détaillé de la thèse
Appendix A: Supplementary materials and methods
Appendix B: Solid-liquid suspension of microcarriers in stirred tank bioreactor
– Experimental and numerical analysis
Appendix C: Design of experiments of the impeller design optimization
Abstract / Résumé