Reprint of “Multiphase mixing characteristics in a microcarrier-based stirred tank bioreactor suitable for human mesenchymal stem cell expansion”

Process Biochemistry xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Process Biochemistry journal homepage: www.elsevier.com/locate/procbio

Reprint of “Multiphase mixing characteristics in a microcarrier-based stirred tank bioreactor suitable for human mesenchymal stem cell expansion”☆ Tanja A. Greina, Jasmin Lebera, Miriam Blumenstocka, Florian Petrya, Tobias Weidnera,b, ⁎ Denise Salziga, Peter Czermaka,b,c,d, a

University of Applied Sciences Mittelhessen, Institute of Bioprocess Engineering and Pharmaceutical Technology, Giessen, Germany Fraunhofer Institute for Molecular Biology and Applied Ecology (IME), Project group Bioresources, Giessen, Germany c Justus Liebig University, Faculty of Biology and Chemistry, Giessen, Germany d Kansas State University, Department of Chemical Engineering, Manhattan, KS, USA b

A R T I C L E I N F O

A B S T R A C T

Keywords: Mesenchymal stem cell Mixing time Transition flow regime Disposable bioreactor Microcarriers Power input Particle suspension

Large-scale human mesenchymal stem cell expansion calls for a bioreaction system, that provides a sufficient growth surface. An alternative to static cultivations systems like cell factories are disposable stirred tank reactors. Here, microcarriers provide the required growth surface, but these make it difficult to achieve a complete homogenization in the bioreactor, while avoiding shear stress. To gain insight into this process, we investigated the impact of different power inputs (0.02–2.6 W m−3) on the mixing time (tm). Whereas tm was inversely proportional to agitation in a one-phase-system, aeration resulted in a constant mixing time at 30–70 rpm. A high microcarrier concentration (30 g L−1) and low stirrer speed (30 rpm) in the liquid-solid system caused a 50fold increase in tm and the formation of a discrete non-mixed upper zone. The effect of the microcarrier concentration on tm became negligible at higher stirrer speeds. In the three-phase system, microcarrier settling was prevented by aeration and a minimal specific power input of 0.6 W m−3 was sufficient for complete homogenization. We confirmed that a low power input during stem cell expansion leads to inhomogeneity, which has not been investigated in the three-phase system up to date.

1. Introduction Human mesenchymal stem cells (hMSCs) are suitable for several applications in regenerative medicine, particularly in the field of cell therapy [1] due to their ability to support self-renewal and multi-linage differentiation, as well as their anti-inflammatory properties [2]. The limitation of hMSCs to anchorage-depend growth means that cell expansion is a challenging process. The expansion of hMSC is typically carried out in static cultivation systems such as tissue culture flasks or cell factories, but the growth surface is limited and elaborate harvesting processes are necessary [3–5]. Medical treatment with hMSCs requires processes that can expand a small number of isolated hMSCs up to an

industrial scale, but the bioreactor system must provide gentle cultivation and harvest conditions because hMSCs are shear sensitive [6,7]. Rather than static systems, hMSCs can also be cultivated on macrocarriers in fixed-bed systems [8,9] or on suspended microcarriers in a stirred-tank reactor (STR). The advantage of the latter strategy is the high growth surface to volume ratio and greater spatial yield. The expansion of hMSCs in dynamic STRs has been demonstrated with different hMSC types and reactor configurations [10–13]. Promising results were achieved in disposable bioreactors e.g. the Mobius® CellReady 3 L bioreactor in fed-batch cultivation mode [14]. The flow regime in a STR becomes particularly important when the cells are growing on microcarriers, because a balance must be achieved between

DOI of original article: http://dx.doi.org/10.1016/j.procbio.2016.05.010 Abbreviations: cv [−], solid fraction of the carrier; cv,max [−], maximum fraction of solids in the closest spherical packing; C [−], constant; D [m], inner vessel diameter; d [m], stirrer diameter; dmt [−], dimensionless mixing time; dpi [−], dimensionless power input; FG [vvm], fraction of gas; g [m s−2], gravity constant; n [s−1], agitation rate; nc [s−1], critical impeller speed for solid suspension; Ne [−], Newton/power number; Nm [−], mixing number; P [W], power input; Pa [W], power input for aerated broth; Re [−], Reynolds number for stirrer; Su [−], suspension number; V [m3], volume; Va [m3 s−1], volumetric air flow rate; tm [s], mixing time, usually for 95% homogeneity; ηl [kg m−1 s−1], dynamic viscosity of medium/liquid; ηm [kg m−1 s−1], mean dynamic viscosity; ρm [kg m−3], liquid density; ρ [kg m−3], density; ρc [kg m−3], mean carrier density; εT [s3 m−2], mean energy dissipation; υ [m2 N−1 s−1], kinematic viscosity; τ [dyn cm−2], shear stress ☆ This article is a reprint of a previously published article. The article is reprinted here for the reader's convenience and for the continuity of the special issue. For citation purposes, please use the original publication details. Process Biochemistry, 51/9, pp. 1109–1119. ⁎ Corresponding author at: University of Applied Sciences Mittelhessen, Wiesenstrasse 14, 35390 Giessen, Germany. E-mail address: [email protected] (P. Czermak). http://dx.doi.org/10.1016/j.procbio.2017.07.025 Received 30 October 2015; Received in revised form 10 May 2016; Accepted 12 May 2016 1359-5113/ © 2016 Elsevier Ltd. All rights reserved.

Please cite this article as: No Author, , Process Biochemistry (2016), http://dx.doi.org/10.1016/j.procbio.2017.07.025

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

the power input required to keep the solid particles in suspension and the protection of the hMSCs from shear stress. The solid microcarriers must remain suspended in the medium to ensure sufficient homogeneity and mixing in the bioreactor, but the power input is limited by the maximum shear stress that can be tolerated by the hMSCs without loss of viability or differentiation capacity [15]. The expansion of hMSCs involves a spatial and temporal mixing procedure based on the distribution of two or more components in the bioreactor volume. These components differ in at least one property, such as concentration, temperature, density, viscosity, particle size or shape. Because mixing is a key process that determines bioreactor performance, and is in turn influenced by many constructive and operational parameters, a detailed quantitative analysis of these factors and their impact on mixing efficiency and distribution is required for process optimization. One useful criterion for the determination of mixing efficiency is the mixing time (homogenization time, tm), which is defined as the time required to achieve a certain degree of homogeneity among the components. It offers specific information concerning bulk mixing (macromixing), but does not allow the quantitation of meso-mixing and micromixing. It can indicate the optimum hydrodynamic regime or bioreactor setup (e.g. stirrer type, use of baffles) and can predict how mixing efficiency is affected when a process is scaled up. In addition, different types of cultivation systems can be compared by calculating the dimensionless mixing number (Nm) [16] which represents the number of stirrer rotations required to homogenize the liquid [17].

Table 1 Conditions used to determine the mixing time in water at 37 °C and ambient pressure.

Mixing time depends on several geometrical factors (e.g. dimensions of the mixing system and bioreactor) and technological factors (e.g. fermentation conditions, physical characteristics of the medium, power consumption and dissipated energy). This general correlation describes the dependence of tm on several factors:

d V P t m = f ⎛ , n, η, ρ, a , , εT⎞ V Pa ⎝D ⎠ ⎜

Lower limit

Upper limit

Center Points

Working volume (L) Aeration (L min−1) Microcarrier concentration (g L−1) Agitation rate (rpm)

1.0 0 0 30

2.4 0.1 30 110

– – 15 50 and 70

phenolphthalein pH shift color change method [26] in a Mobius® CellReady 3 L bioreactor (Merck Millipore, USA) with a scoping marine impeller (d = 0.0762 m). The bioreactor was assembled for stem cell cultivation with pH and temperature probes (for measurement) and an EXcell 230 sensor for the inline monitoring of absorbance (Exner Process Equipment, Germany). For mixing time determination, the bioreactor was filled with preheated distilled water and supplemented with approximately 2 mL 0.1% phenolphthalein (Fagron, Germany) before adjusting the pH to ∼10 with 5 M NaOH to ensure a complete color change to pink. Here, the mixing time is defined by the time span from addition of acid to fully decolorization of the solution in the vessel. For decolorization, 5 M HCl in 5% excess was added as a pulse through the remaining probe fitting in the head plate. The mixing time measurement was started immediately after the addition of HCl and stopped when the color changed (95% criterion). According to the experimental design (Table 1), five parameters were varied: volume, aeration, microcarrier concentration, and stirrer rotation/agitation rate. For every setting, the mixing time was determined in triplicates. The influence of aeration on the mixing time was determined by passing air through the micro-sparger at 0.1 L min−1. The effect of the microcarrier concentration was investigated in the presence and absence of two different concentrations (15 g L−1and 30 g L−1) which are typical for stem cell expansion [27,28]. Solohill® collagen‐coated microcarriers (Pall Corp., USA) with a density of 1022–1030 kg m−3 and a diameter between 125–212 μm were used. The mixing time under all the test conditions was determined over a range of agitation rates (30, 50, 70 and 110 rpm). Each experiment was done three times. The data represent the mean value and the respective standard deviation.

(1)

Nm = tm·n

Factor

⎟

(2)

Several equations can be used to calculate the mixing time, taking into account the type of fermentation (e.g. aerobic, anaerobic), the rheological characteristics of the broth, and the fermentation conditions [18–21], so the accuracy of such estimates are limited to the special conditions included in the calculation. Moreover, most models can only predict tm for a Reynolds number (Re) exceeding 10,000. Where Re < 10,000, these models often require correction factors [22]. In a mixing process, the initial heterogeneity of the components has a major impact on the power input, especially where the components differ in viscosity and density. During the expansion of hMSCs, the power input is determined by a combination of stirring and the aeration. Three studies have been published concerning mixing times in the Mobius® CellReady 3 L bioreactor, each involving a one-phase pure liquid system [23–25]. As the actual hMSC expansion process involves multiple phases, we investigated the mixing behavior of the this bioreactor in two‐phase and three-phase-systems. Therefore, the mixing time at increasing agitation rates was determined in aerated and nonaerated processes with different working volumes and microcarrier concentrations. The selected parameters reflect the real hMSC expansion process: aeration (0.1 L min−1), working volume (1 and 2.4 L), microcarrier concentration (15 and 30 g L−1), agitation rate (30–110 rpm) and power input (0.01–6 W m−3). The shear stress experienced by the cells under these process conditions was also evaluated.

2.2. Cell expansion in the bioreactor As a prove of concept, we have carried out a bioreactor cultivation with bone-marrow derived hMSCs (passage 3–7; kindly provided by EMD Millipore (USA)). The cells were cultivated at 37 °C in DMEM supplemented with 10% FCS and 2 mM L-glutamine (all Biochrom, Germany) containing 15 g L−1 microcarriers (Solohill® collagen coated, 360 cm2 g−1). The aeration rate was maintained at 0.1 L min−1. Expansion was carried out in fed-batch mode (initial concentration 3350 cells cm−2) in a Mobius® CellReady 3 L bioreactor with an agitation rate of 50 rpm. Feeding protocol was done according to previous published work [13,14,29]. The bioreactor was initially filled with 1 L of fresh medium, on day 8 the working volume was increased to 2 L with fresh medium and sufficient microcarriers to maintain the concentration at 15 g L−1, and the agitation rate was increased to 80 rpm. To avoid substrate limitations, 0.4 L fresh medium and microcarriers were added on day 12 and the agitation rate was increased to 90 rpm. Cell dissociation prior to the addition of the fresh microcarriers has not been carried out in the bioreactor expansion. Adherent cells on microcarriers were counted by membrane lysis and subsequent staining of cell nuclei with crystal violet. The microcarriers were settled in 1 mL homogeneous carrier suspension, the supernatant (900 μL) was removed and replaced with 900 μL 0.1% crystal violet in 0.1 M citric acid. The suspension was mixed and incubated at 37 °C for 45 min on a rocking shaker. The nuclei were counted with a

2. Materials and methods 2.1. Mixing time determination All mixing time experiments were carried out at 37 °C using the 2

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

Fig. 1. Mixing characteristics in a Mobius® CellReady 3 L bioreactor at different working volumes (▴: 2.4 L, ■: 1 L). The mixing time was determined at 37 °C in preheated water without microcarriers or aeration. Data are presented as mean of three independent measurements with standard deviation of the mean (lines are shown for guidance).

Aeration did not appear to influence the mixing behavior at a working volume of 1 L, but the outcome was quite different when the working volume was increased to 2.4 L. In the one-phase-system, we observed a continuous decline in tm as the agitation rate increased from 30 to 110 rpm, with the shortest mixing time at the highest agitation rate. Aeration at 0.1 L min−1 improved the mixing conditions at an agitation rate of 30 rpm (0.05 W m−3) and reduced tm by ∼50% compared to the non-aerated system. Increasing the power input to 0.67 W m−3 (70 rpm) did not change the mixing time, but increasing the power input to 2.6 W m−3 (110 rpm) led to a further reduction in tm.

hemocytometer. Glucose and lactate concentration in the culture medium were determined with Biosen C-Line Analyzer (EKF Diagnostics, Germany). 3. Results 3.1. Mixing in a one-phase system Using the disposable bioreactor in fed-batch mode allows to increase the working volume of the hMSC expansion process from 1 L to 2.4 L. We therefore investigated how the change in working volume affects the fluid dynamics and mixing behavior in the bioreactor. As shown in Fig. 1, the mixing time was influenced by changes in working volume, particularly at agitation rates lower than 70 rpm. Under these conditions, the mixing time was increased up to 6-fold by increasing the working volume from 1 L to 2.4 L. At agitation rates higher than 70 rpm, the mixing behavior was no longer dependent on the working volume and tm became more or less constant. This suggests a turbulent flow regime in the bioreactor at agitation rates greater than 70 rpm, corresponding to a Re number of 8279. Generally, turbulence and thus constant and volume-independent mixing behavior is reached at Re ≥ 104 for propeller stirrers and at Re ≥ 103 for baffled systems [16]. In the absence of aeration, we correlated the agitation rate and the applied power input (Eq.(3)). During hMSC expansion, aeration is required to supply the cells with oxygen. The power input in aerated bioreactors was calculated using Eq. (4). The correlation of the agitation rate with the volume-specific power input is shown in Fig. 2.

P = Ne⋅ρ⋅n3⋅d5

3.2.2. Homogenization of the microcarriers The presence of suspended microcarriers in the medium forms a liquid-solid two-phase system, so it is also necessary to consider the influence of the microcarriers on mixing behavior, especially when the microcarriers are present at high concentrations. As shown above, the working volume of the bioreactor had a significant impact on tm (Fig. 4). At a working volume of 1 L, the microcarriers appeared to have no effect on the mixing time at different power input levels (Fig. 4A). However, when the working volume was increased to 2.4 L, tm became dependent on the microcarrier concentration (Fig. 4B). The effect of 30 g L−1 microcarriers at a working volume of 2.4 L is shown at increasing agitation rates in Fig. 5. A low agitation rate (30 rpm) increased tm up to 50-fold. In particular at the 2.4 L working volume, the dense microcarrier suspension (30 g L−1) could not be homogenized at 30 rpm and a clear upper phase was visible, separated from the lower phase containing the particles. While increasing agitation rates, the impact of the microcarrier concentration on tm became negligible (70 rpm). We calculated the minimum agitation rate required to homogenize the microcarriers by equalizing the equations of Zogg [30] and Einenkel [31] describing the dimensionless suspension number Su (Eqs. (5) and (6)).

(3) −0.2

F −0.25 ⎛ n2⋅d4 ⎞ ⋅⎜ Pa = P⋅0.1⋅⎛ G ⎞ ⎟ 2/3 ⎝ n⋅V ⎠ ⎝ g⋅Wi ⋅V ⎠

(4)

1

c v ·(ρc , −,ρ l )·vsed,eff ·g·c·d−2.54 ·ρm−1.27 ⎞ 3.27 n c = ⎜⎛ ⎟ ηm−0.27 ⎠ ⎝

3.2. Mixing in a two-phase system 3.2.1. Mixing behavior in the presence of aeration As the successful cultivation of hMSCs depends on an adequate oxygen supply, the impact of aeration on the mixing characteristics of the bioreactor has to be considered. Although the aeration rate during hMSC expansion is low (0.1 L min−1) and often discontinuous, it forms a two-phase (liquid-gas) system. The mixing times in an aerated twophase system are shown in Fig. 3.

(5)

when

Su =

ρm⋅d2⋅n3 = c⋅Re−0.27 c v (ρc − ρl)g⋅vsed,eff

(6) −1

to a 2.4 L Applying a microcarrier concentration of 30 g L working volume, we found that agitation rates of 50 and 69 rpm were 3

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

Fig. 2. Volume specific power input of the Mobius® CellReady 3 L bioreactor with/without aeration (0.1 L min−1) and at different filling volumes: 1 L without aeration (■) and with aeration (●); 2.4 L without aeration (▴) and with aeration (♦). Testing was carried out at 37 °C in preheated water without microcarriers. Data are presented as mean of three independent measurements with standard deviation of the mean (lines are shown for guidance).

Fig. 3. Mixing characteristics of the Mobius® CellReady 3 L bioreactor with/without aeration (0.1 L min−1) and at different filling volumes: 1 L without aeration (■) and with aeration (●); 2.4 L without aeration (▴) and with aeration (♦). Testing was carried out at 37 °C in preheated water without microcarriers. Data are presented as mean of three independent measurements with standard deviation of the mean (lines are shown for guidance).

required for complete suspension,1 and 100% suspension2 respectively. This was confirmed visually: at 50 rpm, the reactor spontaneously formed three vertical phases comprising a dense cloud of suspended microcarriers in the lower phase, a thinner middle phase of suspended microcarriers, and a thin, clear upper phase (Fig. 5, middle). Although exceeding nc by using agitation rates of more than 69 rpm, a thin clear upper phase was visible at a microcarrier concentration of 30 g L−1 and 2.4 L working volume. While mass transport occurs solely by diffusion in this phase, an agitation rate greater than 70 rpm achieved sufficient microcarrier homogenization at all tested combinations of microcarrier concentration and working volume. Under the typical conditions for hMSC expansion (15 g L−1

microcarriers in DMEM incl. 10% FCS), we calculated that an agitation rate of at least 40 rpm would be required to achieve efficient suspension. Otherwise, settling causes the inhomogeneous distribution of microcarriers in the reactor and results in cell clumping, which is associated with a poor cell nutrient supply, slow growth and inefficient harvesting.

3.3. Mixing in a three-phase system A complete understanding of mixing behavior during hMSC expansion requires the investigation of a three-phase (gas-liquid-solid) system. As described in 3.2 for the two-phase liquid-solid system, vertical heterogeneity was observed at 30 rpm featuring a dense cloud of suspended microcarriers in the lower phase of the reactor (Fig. 6). However, aeration dampened the effect by lifting some microcarriers into the upper phase. A particle-free zone was also present, but was considerably smaller compared to that observed in the non-aerated bioreactor. This effect can also be observed in the mixing time characteristics (Fig. 7) resulting in a progression curve that is similar to the two-phase system (Fig. 3).

1 Complete suspension allows for the settling of small amounts of microcarriers at the base of the bioreactor for a maximum duration of 1 s (the ‘one second criterion’) such that almost all microcarriers are in suspension (only valid when 800 < Re < 106). 2 100% suspension requires the absence of visible suspension at the base of the bioreactor and that the microcarriers are suspended to 90% of the overall fluid height, with the upper 10% devoid of microcarriers (“Schichthöhenkriterium”).

4

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

Fig. 4. Mixing characteristic of Mobius® CellReady 3 L bioreactor in a nonaerated, two-phase system at different working volumes. Mixing times are shown at 37 °C with 30 g L−1 (white), 15 g L−1 (gray) or without (black) Solohill microcarriers. A: 1 L working volume; B: 2.4 L working volume. Data are presented as mean of three independent measurements with standard deviation of the mean (lines are shown for guidance).

Fig. 5. Observation of the mixing in a Mobius® CellReady 3 L bioreactor with a working volume of 2.4 L at 30 rpm, 50 rpm and 70 rpm (from left to right) with a microcarrier concentration of 30 g L−1. Testing was carried out at 37 °C in preheated water without aeration.

Fig. 6. Observation of the mixing in a Mobius® CellReady 3 L bioreactor with a working volume of 2.4 L at 30 rpm, 50 rpm and 70 rpm (from left to right) with a microcarrier concentration of 30 g L−1 and aeration at 0.1 L min−1.

microcarrier concentration and aeration all affect mixing behavior. Bringing these effects together, the dimensionless power input (dpi) and dimensionless mixing time (dmt) are recommended as criteria to measure stirrer performance and mixing behavior [32]. In the system with a 1 L working volume (Fig. 8) every calculated parameter combination showed a linear trend towards faster mixing at higher power inputs, indicating that particle concentration and aeration only have a minor impact on mixing. However, the mixing behavior changes at higher working volumes (Fig. 9). In the one-phase system, a linear trend was observed until the dpi exceeds 1013, at which point the slope declines. The addition of microcarriers increased the dmt compared to the one-phase system. Aeration changed the mixing behavior of systems with and without microcarriers. Initially, increasing the power input gave a constant or

Interestingly, for the 1 L working volume and at low power input, we could show a decline in mixing time while increasing the microcarrier concentration. An increase in power input resulted in the same values for the mixing time at both microcarrier concentrations. Using the higher working volume of 2.4 L, we could observe another effect, also shown in the two-phase system. Here, the mixing time also decreased with increasing microcarrier concentration. Like the gas-liquid system, a constant, power-input independent mixing time was observed in the range 0.05–0.67 W m3. A further increase in the power input again reduced the tm and achieved better mixing conditions. 3.4. Mixing behavior under hMSC expansion conditions The experiments described so far confirm that the working volume, 5

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

Fig. 7. Mixing characteristics in a Mobius® CellReady 3 L bioreactor with aeration at 0.1 L min−1. Mixing times at 37 °C are shown with 30 g L−1 (white) or 15 g L−1 (gray) microcarriers in 1 L (○) and 2.4 L (◊) working volumes. Data are presented as mean of three independent measurements with standard deviation of the mean (lines are shown for guidance).

Fig. 8. Efficiency of the Mobius® CellReady 3 L bioreactor (1 L working volume) at different microcarrier concentrations with (●) and without aeration (■) at 0.1 L min−1. Mixing times at 37 °C are shown with 30 g L−1 (white) or 15 g L−1 (gray) or without (black) microcarriers.

the increase in the mixing time itself at the higher working volume. The dip vs dmt profile changed completely under aeration and in the presence of microcarriers. The dmt was constant in the three-phase system until a certain power input threshold was reached (Fig. 9). As for turbulent mixing conditions the mixing time is known to become independent of aeration and particle concentration. We assumed that transition to a turbulent flow regime was achieved in the 1 L working volume [35,36]. However, while adding up to the final volume of 2.4 L in the last stage of the mentioned process, the mixing behavior in the bioreactor undergoes substantial changes. For complete homogenization, the minimal specific power input was 0.6 W m−3 (> 50 rpm in the 1 L system and > 70 rpm in the 2.4 L system).

slightly higher dmt, but further increase caused the dmt to decline. Under aeration, the presence of microcarriers resulted in a lower dmt. Stirrer flooding, only occurring at higher aeration rates, could not explain this observation. The longer mixing time reflects the energy dissipation at the gas-liquid interface, which makes less energy available for liquid-phase mixing [33,34]. Increasing agitation rates in the threephase-system, brief stagnation of the mixing time was observed even under fully turbulent flow regimes reflecting the interaction between the gaseous and solid phases. Finally, the mixing time decreases to a system-specific value that is no longer dependent on the specific power input. At the beginning of hMSC expansion in the Mobius® CellReady 3 L bioreactor (1 L working volume), the mixing time was unaffected by aeration or particle concentration. We assumed that transition to a turbulent flow regime was achieved in the 1 L working volume [35] and under such conditions the mixing time is known to become independent from aeration and particle concentration [36]. The end phase of hMSC expansion requires a 2.4 L working volume, and the mixing behavior in the bioreactor underwent substantial changes. Only the one-phase system (medium only) behaved like the 1 L system in terms of the dip vs dmt profile, although the line tends toward higher dmt values reflecting

3.5. Determination of shear stress during hMSC expansion The mixing time calculations indicated that a minimal agitation rate of > 70 rpm is required to ensure homogeneity in the vessel under production conditions. However, increasing the agitation rate and introducing aeration can expose the hMSCs growing on the microcarrier surface to shear stress. The mean intensity of shear stress in the stirrer outflow area, on the hMSCs in the disposable bioreactor during aerated 6

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

Fig. 9. Efficiency of the Mobius® CellReady 3 L bioreactor (2.4 L working volume) at different microcarrier concentrations with (♦) and without aeration (▴) at 0.1 L min−1. Mixing times at 37 °C are shown with 15 g L−1 (gray) or without (black) microcarriers (lines are shown for guidance).

Fig. 10. Shear forces during stem cell expansion: 1 L working volume without aeration (■) and with aeration (●); 2.4 L working volume without aeration (▴) and with aeration (♦) (lines are shown for guidance).

increased to ensure proper mixing (maximum 1.2 W m−3 after 8 d and 1.4 W m−3 after 12 d). The growth kinetics of the primary hMSCs was not affected by increasing the working volume or agitation rate, resulting in a steady growth rate of 0.37 h−1 and a glucose consumption rate of 0.002 mmol h−1 per 105 cells during the fed-batch phase.

and non-aerated mixing was calculated as follows [17,37,38]:

P 0.5 ⎞ τ = η ⋅⎛ ⎝ m ⋅ν ⎠

(7)

As shown in Fig. 10, the higher power input caused by aeration significantly increased the shear stress, but at higher working volumes the mean shear stress was lower. Assuming a minimal required agitation rate for complete homogenization of 50 rpm for the 1 L system (70 rpm for 2.4 L) the same intensity of shear stress is applied throughout the hMSC expansion process.

4. Discussion During the hMSC expansion process, the homogenization of three phases – liquid, solid particles and gas – is necessary to avoid gradient formation and thus limiting conditions in the reactor. In this single-use bioreactor, the required power input for homogenization is provided by stirring and aeration. Although three studies have already addressed the mixing time in this bioreactor [23–25], they did not include the influence of aeration, microcarriers or the low power inputs typically used during hMSC expansion. As different experimental setups were used in these studies the acquired results are not directly comparable with the current investigation. It is necessary to include both aeration and microcarriers to

3.6. Expansion of hMSCs using the Mobius® CellReady 3 L bioreactor The expansion of hMSCs in the bioreactor was carried out in fedbatch mode (Fig. 11). Initially, the cells were inoculated in a 1 L working volume and agitated at a maximum power input of 0.58 W m−3. The feeding was done after 8 d, whereupon the additional medium and microcarriers increased the working volume at first to 2 L and finally to 2.4 L after 12 d. At the same time, the power input was 7

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

Fig. 11. Fed-batch stem cell expansion in a Mobius® CellReady 3 L disposable bioreactor with working volumes of 1 L, 2 L and 2.4 L. Cell concentration (●), glucose concentration (◁) (lines are shown for guidance; n = 1).

increases the depth of this clear layer declines. After a certain point, further increases in the agitation cause the clear upper layer to increase in depth again, reflecting the increasing density of the particle suspension. In this case (n < nc), the mixing time is longer than it would be in a system without particles. This effect directly correlates with the concentration of the solid particles: the higher the particle concentration, the longer the mixing time during this phase. Interestingly, we confirmed this behavior in the two-phase system (solid-liquid) but not the three-phase system. In the latter case, with a working volume of 2.4 L and a power input of up to 0.67 W m−3 (70 rpm), the shortest mixing time was achieved in the presence of 30 g L−1 microcarriers. At higher agitation rates (n > nc) the mixing time declined to the values observed in a pure liquid one-phase system. At this point, a homogeneous suspension prevails because all particles follow the flow pattern of the reactor [39]. Once a certain power input threshold is reached, the suspension capacity is high enough to suspend every particle and the influence of the solid content on tm becomes negligible. In the Mobius® CellReady 3 L bioreactor, the mixing behavior in a three-phase system can be divided into three sections [33,43]: complete sedimentation of particles and negligible dispersion of the gas phase (Section 1), partial suspension of particles and adequate dispersion of the gas phase (Section 2), and adequate suspension of both the particles and the gas phase (Section 3). In agreement with this sectional mixing behavior, critical process parameters such as viscosity change due to the mixing process [34]. In a laminar flow regime, the viscosity (ηm) and density (ρm) of the suspension change during the process. The energy dissipation and viscosity changes in a three-phase system reduce the mixing efficiency of the stirrer. This becomes apparent when considering Ne and Re. The two dimensionless numbers differ significantly compared to the one-phase system especially at low agitation rates in a laminar flow regime [41]. The influence of particles on the power number Ne (also known as the Newton number) depends on the particle concentration. At low particle concentrations, Ne is constant in the turbulent flow regime and is equal to that of a one-phase system (e.g. water only). But at high particle concentrations (such as in our experiments), low agitation rates have a strong impact on Ne in two contexts. First, the formation of a particle bed changes the bottom shape of the reactor and redirects the flow pattern in the presence of the particles, thus Ne drops below the value of the one-phase system (liquid only). Second, increasing the agitation rate causes Ne to exceed the value of the one-phase system, thus a higher specific power input is needed to achieve the same mixing efficiency as the particle-free system [44]. Depending on the bioreactor system, the actual composition of the

achieve the expansion of adherent hMSCs in a stirred tank reactor, but our experiments revealed that both these factors have a significant impact on the mixing behavior of this bioreactor. 4.1. Microcarriers and aeration affect mixing behavior Many previous reports have addressed the impact of solid particles [39] and aeration [16,20,33,40,41] on mixing time in different systems. Under laminar flow conditions, mixing time is linearly dependent on agitation/stirrer power input in both one-phase (liquid only) and multiphase systems varying in density or viscosity. With increasing power input, the flow conditions become fully turbulent. Concerning the onephase system, we could also confirm that reaching a specific power input, the mixing time remains constant and cannot be further decreased. We further observed ambivalent mixing behavior in multi-phase systems. Under laminar flow conditions, an increase instead of the expected reduction in mixing time despite the increasing power input was shown. This effect was also observed when using aeration and in the presence or absence of microcarriers. The mixing time is generally longer in three-phase (gas-liquid-solid) systems than two-phase (solid-liquid or liquid-gas) systems [33,42] as we confirmed in our experiments when the agitation rate was greater than 70 rpm. At lower agitation rates, particularly when the working volume was 2.4 L, the mixing time was longer in both two-phase systems compared to the three-phase system. This reflects the smaller volume of liquid due to the presence of solid particles, reducing the diffusion distance between the particles and encouraging faster discoloration due to accelerated mass transfer [16,20,33,35]. The solid particles also interact with gas bubbles which suspends the solid particles in the liquid. Our results support the previously reported stagnation of the mixing time under transitional flow conditions using solid particles [33]. The increased mixing time in the presence of particles at higher agitation rates may reflect energy dissipation at the solid-liquid interface, which is therefore no longer available for the mixing process [16,33]. If the microcarriers are not sufficiently agitated, they form a particle bed below the stirrer. Under conditions where only few particles are lifted, tm does not differ significantly from a system without particles, and the particles in the bed hardly influence the flow pattern of the reactor (n < < nc). Increasing the stirrer speed (n) causes more particles to be suspended. Initially only few particles are lifted and a clear phase remains visible in the upper part of the vessel, but as the agitation rate 8

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

input, suggesting the cells are not adversely affected by long-term shear forces. Similarly, previous studies based on surface markers and the differentiation capacity of primary hMSCs found no differences between hMSCs expanded in static T-flasks or on microcarriers in an STR [14].

suspension is often unclear, e.g. how many particles remain in suspension, which makes the precise calculation of mixing times more challenging. 4.2. Heterogeneity occurs during hMSC expansion in the Mobius® CellReady 3 L bioreactor

5. Conclusion Our analysis of mixing times revealed that the mixing behavior changed during the hMSC expansion process due to the fed-batch operational mode. The addition of medium and microcarriers required a higher power input to ensure proper mixing, which was discouraged by the heterogeneity of the components. The gas bubbles and microcarriers are not a uniform size, e.g. the individual microcarrier particles span the size range 125–212 μm. This component-specific heterogeneity hinders the mixing process and generates heterogeneous zones that can reduce the process yield [45]. We found no evidence that such heterogeneous zones had a negative impact on hMSC expansion in this bioreactor, as we have shown in our previous publication [14]. In addition, cells expanded in this bioreactor keep their differentiation ability, a normal karyotype as well as surface marker expression [13,29] defined by the International Society for Cellular Therapy [46]. Heterogeneity can become more challenging in larger-scale processes. For example, local cell lysis has been reported in animal cell cultures due to the addition of concentrated drops of alkaline solution at the top of the reactor [45]. Simulations have shown that insufficient mixing can result in high local pH fluctuations (up to pH 9), even if the overall pH is buffered at 7.2 [47] and significant spatial gradients in dissolved oxygen [41].

It is important to understand the impact of bioengineering characteristics on cell behavior in the exact ranges used in the final process. One important parameter, the mixing time, is strongly affected by multiple components in the bioreactor including the gas provided by aeration, the liquid growth medium, and solid particles such as microcarriers used to provide a growth surface for the cells. Typically, a short mixing time is preferable to avoid inhomogeneity in the bioreactor vessel. In contrast to investigations of the influence of shear stress on stem cells, but in agreement with several stem cell expansion studies in a stirred tank bioreactor, an expansion under moderate mean shear stress will not result in quality loss. The inhomogeneity in the bioreactor is more likely to shorten residence times of the stem cells in the high shear stress zone. Hence, an inhomogeneity in the flow regime with moderate mixing times could be as well favorable in the efficient stem cell expansion process. This shows that the expansion of hMSCs in a disposable bioreactor is a robust process that matches static systems in terms of cell quality while providing all the advantages of a bioreactor, such as process control, online monitoring and process automation. Acknowledgment We would like to thank the Federal Ministry of Education and Research, Germany and the Hessen State Ministry of Higher Education, Research and the Arts for the financial support within the Hessen initiative for scientific and economic excellence. We thank Richard M. Twyman for revising the manuscript.

4.3. Shear stress remains low during hMSC expansion The thorough mixing of all three phases is necessary for hMSCs to receive a sufficient nutrient supply during expansion, but the power input at this stage is a double-edged sword because fluid shear stress can influence cell proliferation and differentiation [6,7,15]. In previous studies, hMSCs have been exposed to a constant high shear stress (3–25 dyn cm−2) for several hours. The results suggest, that the characteristic of the cells will be influenced by applied shear stress depending on the stem cell type. In contrast, previous studies based on surface markers and the differentiation capacity of primary hMSCs found no differences between hMSCs expanded in static T-flasks or fixed bed cultivation or on microcarriers in an STR [8,9,13,14,29]. The process parameters we tested never exceeded a mean shear stress of 0.2 dyn cm−2 which is unlikely to have any negative effects. Kaiser et al. calculated a local maximal shear stress for the Mobius® CellReady 3 L bioreactor of up to 10 times higher (up to 1.2 dyn cm2), located close to the stirrer [48–50]. In this study, a turbulent flow regime in the vessel was assumed. Concerning the three-phase stem cell cultivation, present mixing time investigations could not confirm a fully turbulent flow regime for typically used agitation rates (30–100 rpm). Furthermore, our results suggest that the microcarriers had only a short residence time in this high shear stress zone close to the stirrer. Although greater shear stress is introduced when processes are scaled up, the aforementioned challenges with heterogeneous suspensions are more likely to cause problems with cell growth. Assuming a constant specific power input and geometric similarity, we calculated that the shear forces in a 50 L scale process would – in the worst case – reach 1 dyn cm−2. Despite these facts, we cannot rule out that hMSCs may be affected by low levels of shear stress over the prolonged duration of the stem cell expansion process (generally more than 2 weeks) [14,13]. When animal cells are stressed during expansion, this can be observed by monitoring metabolic kinetics such as the average of glucose consumption and lactate production [51]. We observed a ratio of glucose consumption and lactate production (YGlc/Lac) close to 1 (0.99 ± 0.02) in fed-batch mode, which was independent of the power

References [1] D. Freimark, P. Pino-Grace, S. Pohl, C. Weber, C. Wallrapp, P. Geigle, R. Pörtner, P. Czermak, Use of encapsulated stem cells to overcome the bottleneck of cell availability for cell therapy approaches, Transfus. Med. Hemother. 37 (2010) 66–73. [2] N. Escacena, E. Quesada-Hernandez, V. Capilla-Gonzalez, B. Soria, A. Hmadcha, Bottlenecks in the efficient use of advanced therapy medicinal products based on mesenchymal stromal cells, Stem Cells Int. (2015), http://dx.doi.org/10.1155/ 2015/895714. [3] C.L. Elseberg, D. Salzig, P. Czermak, Bioreactor expansion of human mesenchymal stem cells according to GMP requirements, Methods Mol. Biol. 1283 (2015) 199–218. [4] C. Weber, S. Pohl, R. Pörtner, P. Pino-Grace, D. Freimark, C. Wallrapp, P. Geigle, P. Czermak, Production process for stem cell based therapeutic implants: expansion of the production cell line and cultivation of encapsulated cells, in: C. Kasper, M. Griensven, R. Pörtner (Eds.), Bioreactor Systems for Tissue Engineering II, vol. 123, Springer Berlin Heidelberg, 2010, pp. 143–162. [5] D. Salzig, A. Schmiermund, P. Grace, C. Elseberg, C. Weber, P. Czermak, Enzymatic detachment of therapeutic mesenchymal stromal cells grown on glass carriers in a bioreactor, Open Biomed. Eng. J. 7 (2013) 147–158. [6] W. Yi, Y. Sun, X. Wei, C. Gu, X. Dong, X. Kang, S. Guo, K. Dou, Proteomic profiling of human bone marrow mesenchymal stem cells under shear stress, Mol. Cell. Biochem. 341 (2010) 9–16. [7] G. Yourek, S.M. McCormick, J.J. Mao, G.C. Reilly, Shear stress induces osteogenic differentiation of human mesenchymal stem cells, Regen. Med. 5 (2010) 713–724. [8] C. Weber, S. Pohl, R. Pörtner, P. Pino Grace, D. Freimark, C. Wallrapp, P. Geigle, P. Czermak, Expansion of human mesenchymal stem cells in a fixed-bed bioreactor system – part b: modeling and scale up, Int. J. Artif. Organs 33 (2010) 782–795. [9] C. Weber, S. Pohl, P. Pino Grace, D. Freimark, C. Wallrapp, P. Geigle, P. Czermak, Expansion of human mesenchymal stem cells in a fixed-bed bioreactor system – part a: inoculation, cultivation, cell harvest procedures, Int. J. Artif. Organs 33 (2010) 512–525. [10] C.L. Elseberg, J. Leber, D. Salzig, C. Wallrapp, M. Kassem, M. Kraume, P. Czermak, Microcarrier-based expansion process for hMSCs with high vitality and undifferentiated characteristics, Int. J. Artif. Organs 35 (2012) 93–107. [11] C. Justice, J. Leber, D. Freimark, P. Pino Grace, M. Kraume, P. Czermak, Online- and offline- monitoring of stem cell expansion on microcarrier, Cytotechnology 63 (2011) 325–335.

9

Process Biochemistry xxx (xxxx) xxx–xxx

T.A. Grein et al.

97–125, http://dx.doi.org/10.1002/9783527612703.ch03. [33] V.B. Rewatkar, K.S.M.S.R. Rao, J.B. Joshi, Critical impeller speed for solid suspension in mechanically agitated three-phase reactors. 1. Experimental part, Ind. Eng. Chem. Res. 30 (1991) 1770–1784. [34] P.M. Weinspach, Stoffübergang im aufsteigenden Flüssigkeitsstrom, Chem. Ing. Tech. 37 (1965) 1215–1220. [35] M. Kraume Mischen und Rühren, Grundlagen und Moderne Verfahren, Wiley; 2006. http://dx.doi.org/10.1002/3527603360. [36] P. Bonvillani, M.P. Ferrari, E.M. Ducrós, J.A. Orejas, Theoretical and experimental study of the effects of scale-up on mixing time for a stirred-tank bioreactor, Braz. J. Chem. Eng. 23 (2006) 1–7. [37] P.E. Cruz, A. Cunha, C.C. Peixoto, J. Clemente, J.L. Moreira, M.J.T. Carrondo, Optimization of the production of virus-like particles in insect cells, Biotechnol. Bioeng. 60 (1998) 408–418. [38] G.A. Hughmark, Power requirements and interfacial area in gas-liquid turbine agitated systems, Ind. Eng. Chem. Process Design Dev. 19 (1980) 638–641. [39] W. Bujalski, K. Takenaka, S. Paoleni, M. Jahoda, A. Paglianti, K. Takahashi, A.W. Nienow, A.W. Etchells, Suspension and liquid homogenization in high solids concentration stirred chemical reactors, Chem. Eng. Res. Des. 77 (1999) 241–247. [40] N. Kuzmanić, B. Ljubičić, Suspension of floating solids with up-pumping pitched blade impellers; mixing time and power characteristics, Chem. Eng. J. 84 (2001) 325–333. [41] A.W. Nienow, C. Langheinrich, N.C. Stevenson, A.N. Emery, T.M. Clayton, N.K. Slater, Homogenisation and oxygen transfer rates in large agitated and sparged animal cell bioreactors: some implications for growth and production, Cytotechnology 22 (1996) 87–94. [42] A.B. Pandit, J.B. Joshi, Mixing in mechanically agitated gas-liquid contactors, bubble columns and modified bubble columns, Chem. Eng. Sci. 38 (1983) 1189–1215. [43] J. Medek, I. Fort, Mixing in vessel with eccentrical mixer, Proc. 5th Europ Conf on Mixing BHRA Press Cranfield (1985) 263–271. [44] M. Micheletti, L. Nikiforaki, K.C. Lee, M. Yianneskis, Particle concentration and mixing characteristics of moderate-to-dense solid-liquid suspensions, Ind. Eng. Chem. Res. 42 (2003) 6236–6249. [45] S. Ozturk, Engineering challenges in high density cell culture systems, Cytotechnology 22 (1996) 3–16. [46] M. Dominici, K. Le Blanc, I. Mueller, I. Slaper-Cortenbach, F. Marini, D. Krause, R. Deans, A. Keating, D. Prockop, E. Horwitz, Minimal criteria for defining multipotent mesenchymal stromal cells: the International Society for Cellular Therapy position statement, Cytotherapy 8 (2006) 315–317. [47] M. Reuss, S. Schmalzriedt, M. Jenne, Structured modelling of bioreactors, in: E. Galindo, O. Ramírez (Eds.), Advances in Bioprocess Engineering, Springer, Netherlands, 1994, pp. 207–215. [48] S.C. Kaiser. Characterization and optimization of single-use bioreactors and biopharmaceutical production processes using computational fluid dynamicsTechnische Universität Berlin, Doctoral Thesis, 2014. http://dx.doi.org/ 10.14279/depositonce-4313. [49] V. Jossen, R. Eibl, S.C. Kaiser, M. Kraume, R. Pörtner, D. Eibl, Mass production of mesenchymal stem cells: impact of bioreactor design and flow conditions on proliferation and differentiation. 2014. http://dx.doi.org/10.5772/23496. [50] S.C. Kaiser, C. Löffelholz, D. Eibl, S. Werner, CFD for Characterizing Standard and Single-use Stirred Cell Culture Bioreactors, INTECH Open Access Publisher, 2011. [51] S. Quesney, A. Marc, C. Gerdil, C. Gimenez, J. Marvel, Y. Richard, B. Meignier, Kinetics and metabolic specificities of Vero cells in bioreactor cultures with serumfree medium, Cytotechnology 42 (2003) 1–11.

[12] Q. Rafiq, K. Brosnan, K. Coopman, A. Nienow, C. Hewitt, Culture of human mesenchymal stem cells on microcarriers in a 5 L stirred-tank bioreactor, Biotechnol. Lett. 35 (2013) 1233–1245. [13] D. Jing, N. Sunil, S. Punreddy, M. Aysola, D. Kehoe, J. Murrel, M. Rook, K. Niss, Growth kinetics of human mesenchymal stem cells in a 3-L single-use, stirred-tank bioreactor, BioPharm. Int. 26 (2013) 28–38 available online at http://www. biopharminternational.com/growth-kinetics-human-mesenchymal-stem-cells-3-lsingle-use-stirred-tank-bioreactor. [14] K. Cierpka, C.L. Elseberg, K. Niss, M. Kassem, D. Salzig, P. Czermak, hMSC production in disposable bioreactors with regards to GMP and PAT, Chem. Ing. Tech. 85 (2013) 67–75. [15] S. Stolberg, K.E. McCloskey, Can shear stress direct stem cell fate, Biotechnol. Progr. 25 (2009) 10–19. [16] Zlokarnik M. Rührtechnik, Theorie Und Praxis, Springer Verlag, Heidelberg, 1999http://dx.doi.org/10.1007/978-3-642-58635. [17] P.M. Doran, 7 – fluid flow and mixing, in: P.M. Doran (Ed.), Bioprocess Engineering Principles, Academic Press, London, 1995, pp. 129–163. [18] N.W.F. Kossen, N.M.G. Oosterhuis, Modelling and scaling-up of bioreactors. In BH (Ed), Fundamentals of Biochemical Engineering, vol. 2, 1985, VHC; Weinheim. [19] S.M. Kresta, D. Mao, V. Roussinova, Batch blend time in square stirred tanks, Chem. Eng. Sci. 61 (2006) 2823–2825. [20] N. Ma, M. Mollet, J.J. Chalmers, Aeration, mixing and hydrodynamics in bioreactors, Encyclopedia of Cell Technology, page 225-248, John Wiley & Sons, Inc., 2003, http://dx.doi.org/10.1002/0471250570.spi006. [21] A.W. Nienow, On impeller circulation and mixing effectiveness in the turbulent flow regime, Chem. Eng. Sci. 52 (1997) 2557–2565. [22] C. Oniscu, A.-I. Galaction, D. Cascaval, Modelling of mixing in stirred bioreactors 3. Mixing of aerated simulated broth, Rom. Biotech. Lett. 7 (2002) 817–828, http:// dx.doi.org/10.2298/HEMIND0212506C. [23] S.C. Kaiser, R. Eibl, D. Eibl, Engineering characteristics of a single-use stirred bioreactor at bench-scale: the Mobius CellReady 3 L bioreactor as a case study, Eng. Life Sci. 11 (2011) 359–368. [24] G. van Eikenhorst, Y.E. Thomassen, L.A. van der Pol, W.A.M. Bakker, Assessment of mass transfer and mixing in rigid lab-scale disposable bioreactors at low power input levels, Biotechnol. Progr. 30 (6) (2014) 1269–1276, http://dx.doi.org/10. 1002/btpr.1981. [25] J. Dekarski, Scalability of the Mobius CellReady single-use bioreactor systems, BioPharm. Int. 26 (2013). [26] E.A. Fox, V.E. Gex, Single-phase blending of liquids, AIChE J. 2 (1956) 539–544. [27] F. dos Santos, A. Campbell, A. Fernandes-Platzgummer, P.Z. Andrade, J.M. Gimble, Y. Wen, S. Boucher, M.C. Vemuri, C.L. da Silva, J.M.S. Cabral, A xenogeneic-free bioreactor system for the clinical-scale expansion of human mesenchymal stem/ stromal cells, Biotechnol. Bioeng. 111 (2014) 1116–1127. [28] A.C. Schnitzler, A. Verma, D.E. Kehoe, D. Jing, J.R. Murrell, K.A. Der, M. Aysola, P.J. Rapiejko, S. Punreddy, M.S. Rook, Bioprocessing of human mesenchymal stem/ stromal cells for therapeutic use: current technologies and challenges, Biochem. Eng. J. (2016), http://dx.doi.org/10.1016/j.bej.2015.08.014. [29] D. Jing, S. Punreddy, N. Sunil, M. Aysola, J. Murrel, K. Niss, Characterization of human mesenchymal stem cells, BioProcess Int. 11 (2013) 30–36 available online at http://www.bioprocessintl.com/upstream-processing/upstream-contract-services/ characterization-of-human-mesenchymal-stem-cells-340980/. [30] M. Zogg, Einführung in Die Mechanische Verfahrenstechnik: Mit 29 Tabellen Und 32 Berechnungsbeispielen, Teubner, 1993. [31] W.-D. Einenkel, Fluiddynamik des suspendierens, Chem. Ing. Tech. 51 (1979) 697–704. [32] Zlokarnik M. Homogenization, Stirring, Wiley-VCH Verlag GmbH, 2007, pp.

10