Effect of inoculum density on human‐induced pluripotent stem cell expansion in 3D bioreactors

Abstract Objective For optimized expansion of human‐induced pluripotent stem cells (hiPSCs) with regards to clinical applications, we investigated the influence of the inoculum density on the expansion procedure in 3D hollow‐fibre bioreactors. Materials and Methods Analytical‐scale bioreactors with a cell compartment volume of 3 mL or a large‐scale bioreactor with a cell compartment volume of 17 mL were used and inoculated with either 10 × 106 or 50 × 106 hiPSCs. Cells were cultured in bioreactors over 15 days; daily measurements of biochemical parameters were performed. At the end of the experiment, the CellTiter‐Blue® Assay was used for culture activity evaluation and cell quantification. Also, cell compartment sections were removed for gene expression and immunohistochemistry analysis. Results The results revealed significantly higher values for cell metabolism, cell activity and cell yields when using the higher inoculation number, but also a more distinct differentiation. As large inoculation numbers require cost and time‐extensive pre‐expansion, low inoculation numbers may be used preferably for long‐term expansion of hiPSCs. Expansion of hiPSCs in the large‐scale bioreactor led to a successful production of 5.4 × 109 hiPSCs, thereby achieving sufficient cell amounts for clinical applications. Conclusions In conclusion, the results show a significant effect of the inoculum density on cell expansion, differentiation and production of hiPSCs, emphasizing the importance of the inoculum density for downstream applications of hiPSCs. Furthermore, the bioreactor technology was successfully applied for controlled and scalable production of hiPSCs for clinical use.


| INTRODUC TI ON
The application of human-induced pluripotent stem cells (hiP-SCs) has shown high potential in the field of clinical therapies 1 and pharmaceutical drug development, 2 as this cell type is suitable for generating disease-specific models and patient-specific therapies. [3][4][5][6] However, the utilization of hiPSC models in drug discovery requires high cell quantities of hiPSCs and their derivatives at a constant quality. 7,8 This can hardly be achieved by using conventional 2D cell cultures due to insufficient cell production yields, lack in scalability and difficulty of controlling cell culture parameters. 9,10 In contrast, the use of 3D culture models offers the opportunity of large-scale expansion of hiPSCs under controlled conditions. 9,11 For production of large cell quantities fulfilling the required quality standards, it is important to consider those factors that potentially influence hiPSC expansion and differentiation in 3D culture systems. Such factors include feeding strategies, coating materials, culture media and the cell inoculum density. 9 In the present study, the effect of the inoculum density on cell expansion and differentiation of hiPSCs cultured in perfused hollowfibre-based 3D bioreactors was investigated. For this purpose, 10 × 10 6 hiPSCs resp. 3.3 × 10 6 cells/mL, or 50 × 10 6 hiPSCs resp. 16.6 × 10 6  2.9 × 10 6 cells/mL.

| Bioreactor system/technology
The 3D four-compartment hollow-fibre bioreactor used in this study is based on three independent, interwoven hollow-fibre capillary bundles, two for supplying nutrient media by countercurrent perfusion and one for gas exchange. The space between these capillary bundles (extracapillary space) serves as cell compartment. The capillary system is integrated into a polyurethane housing. The cells, grown in the cell compartment, were constantly supplied with nutrients and oxygen. The bioreactor types used in this study had a cell compartment volume of 3 mL (analytical-scale, AS) or 17 mL (largescale, LS); specific data regarding compartment measurements as well as perfusion conditions are displayed in Table 1. Both bioreactor types, the AS and LS bioreactor, are constructed identically in respect of their capillary configuration ( Figure 1); they only differ in length and number of capillaries. A detailed description of the bioreactor technology can be found elsewhere. 12,13 The bioreactors were connected to a perfusion device consisting of pumps for medium feed and medium recirculation.

| Expansion of hiPSCs in 3D bioreactors
Following pre-expansion, either 10 × 10 6 (AS 10) or 50 × 10 6 (AS 50, LS 50) hiPSCs were inoculated as single-cell suspension into precoated bioreactors (8.68 µg/cm 2 Matrigel, Corning) and cultured over 15 days. The initial cell numbers used in this study are based on previous studies on the hepatic differentiation of hiPSCs in the AS bioreactor, where 100 × 10 6 cells were inoculated. 15 Thus, an initial cell number of 10 × 10 6 cells, resp. 3.3 × 10 6 cells/mL in AS 10 provides the spatial conditions for at least a 10-fold cell expansion, while an initial cell number of 50 × 10 6 resp. a cell density of 16.6 × 10 6 cells/mL in AS 50 should enable at least a 2-fold expansion. The latter was chosen to investigate the influence of a high initial cell density on the expansion procedure. For the feasibility testing of an up-scale of the hiPSC expansion, the LS bioreactor was inoculated with a cell number of 50 × 10 6 resp. a cell density of 2.9 × 10 6 cells/mL, as this equals the conditions of AS 10. To ensure single-cell survival at the beginning of the experiment, 1 mmol/L ROCK inhibitor (Y-27632; Abcam) was included into the culture medium as bolus injection and was rinsed out within the first 24 hours of bioreactor cultures.
The bioreactors were placed into a heating chamber constantly kept at 37°C. The medium recirculation rate was set to 10 mL/min (AS) resp. 20 mL/min (LS), whereas the medium feed was initially set to 1 mL/h (AS) resp. 2 mL/h (LS) and adapted daily to up to 12 mL/h (AS) or 40 mL/h (LS), depending on the glucose consumption rates.
Thereby, glucose levels were kept above 4.4 mmol/L throughout the culture period. The gas perfusion rate was constantly maintained at 20 mL/min (AS) resp. 40 mL/min (LS); CO 2 was added at a percentage of up to 5% for pH regulation to approximately 7.2 (Table 1).

| Biochemical parameters
The metabolic activity of cultured cells was analysed by daily meas- GmbH. AFP levels were analysed every 5 days, or more frequently when concentrations were above the detection limit.  The schematic image on top shows a section of the capillary structure inside the bioreactor, consisting of the following four compartments: medium capillaries I (red) and II (blue) for countercurrent medium perfusion, gas capillaries (yellow) and the space surrounding the capillaries, which serves as the cell compartment (white). The scale bars correspond to 2 cm

| Gene expression analysis
Gene expression analysis was performed as described previously 15,17 using human-specific primers and probes as listed in Table 2.
Expression values of measured genes were normalized to expression values of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and fold changes of expression levels were calculated using the ΔΔC t method. 18

| Immunhistochemistry analysis
Upon termination of bioreactor cultures, sections of the capillary bed containing cell material were removed and prepared for immunofluorescence staining as described previously. 19 Nuclei were counterstained with Dapi (blue). The antibodies used for immunohistochemistry are displayed in Table 3. Viability Assay as well as cell quantification data, population doublings and doubling times were compared using the unpaired, twotailed Student's t test.  Figure 2D). In contrast, there was no AFP detectable in perfusates of AS 10 during the entire culture period ( Figure 2D). For LS 50 ( Figure 2H), a slight increase was measured from day 14 onwards, but maximum values were almost five times lower than those observed in AS 50.

| Metabolic activity of hiPSCs during bioreactor expansion
In conclusion, higher cell densities led to a significantly higher overall cell activity in AS bioreactors; however, higher cell densities also led to beginning differentiation as indicated by increasing AFP levels in AS 50. The highest values for energy metabolism were achieved in LS 50, being more than three times as high compared with maximum values obtained in AS 50.

| Gene expression profiles of hiPSC cultures
For the characterization of hiPSCs after expansion in 3D bioreactors, the gene expression of pluripotency as well as differentiation markers relative to the undifferentiated state were analysed.
The expression data of the two pluripotency markers POU5F1 and NANOG ( Figure 3A

| Cell activity of hiPSC cultures
The CellTiter-Blue ® Cell Viability assay (CTB) was performed in all bioreactors as well as in 2D cultures, which were cultured in paral- In conclusion, the highest cell activity was detectable in LS 50, followed by AS 50 and AS 10. The increase in cell activity of AS 50 was significantly larger than that of AS 10.

| Immunohistochemical characteristics of hiPSC cultures
Staining with the pluripotency marker POU5F1 (

| D ISCUSS I ON
Since the application of hiPSCs in the medical field requires large cell quantities at high-quality standards, it is of great interest to evaluate factors that influence hiPSC expansion in 3D culture systems. Therefore, the effect of the inoculum density on the hiPSC expansion procedure, cell differentiation and the cell yield was investigated in this study.  the compartmentalized bioreactor provides a countercurrent "arteriovenous" media flow and decentralized gas perfusion via capillaries, thereby enhancing mass exchange for an optimized nutrient and oxygen supply for the cultured cells. 39 Also, cells are not affected by shear stress, which occurs for example if cells are cultured in stirred tank bioreactors, 40 and can cause cell damage. 41 The total amount of cells produced in LS 50 (5.4 × 10 9 cells) would suffice for single-patient treatments in heart and liver therapies as well as treatment of diabetes. 11 Cell numbers of this relevance have to date been only achieved by Kwok et al, 42  Several research groups observed that higher cell densities support differentiation processes of pluripotent stem cells. 31,36,[43][44][45][46][47] However, the majority of studies were performed in 2D culture models, where medium is usually exchanged discontinuously, and cells are limited to growing horizontally, instead of three-dimensionally. In contrast, perfused 3D cultures enable a continuous supply with nutrients and oxygen, while maintaining cell pluripotency. 48 In particular, the four-compartment hollow-fibre bioreactors used in this study aim to mimic the in vivo situation in the tissue, thereby enabling increased cell densities at physiological levels. 17 Therefore, the results gained in 2D cultures may not be directly comparable to the 3D cultures used in the present study.
Furthermore, the results presented in 2D studies regarding critical cell densities varied, depending on the initial cell type and the desired differentiation outcome. For example, Selekman et al 49 found that a human pluripotent stem cell density of 6500 cells/cm 2 is optimal for an epithelial differentiation considering the balance between purity and yield of cells. In contrast, initial seeding densities of dental and oral stem cells for neural induction in 2D cultures laid between 3000 cells/cm 2 and 20 000 cells/cm 2 . 50 Overall, the expansion of hiPSCs in 3D hollow-fibre bioreactors was successful for different cell inoculation conditions and bioreactor sizes. The use of a larger bioreactor (17 mL) resulted in clinically relevant cell yields. The findings also show that the inoculum density has significant influence on the growth behaviour and the differentiation state of the cells in 3D bioreactors. A high cell inoculation number led to a faster expansion with higher maximum values for the glucose uptake and growth, but also to cells more prone to differentiation. In contrast, lower initial cell numbers led to slower expansion, but showed less differentiation and required less time and effort for pre-expansion of the inoculum to the 3D bioreactor. The latter is especially of relevance for efficient hiPSC expansion in large-scale bioreactors. Based on the described results, we conclude that 3D perfusion bioreactors should be inoculated with low cell numbers for achieving a successful long-term hiPSC expansion for clinical purposes. In order to avoid differentiation, additional repeated cell harvesting or cell aggregate dissociation may be included into the expansion procedure.

ACK N OWLED G M ENTS
The work for this study was supported by the German Ministry

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare.