Engineering Characterization of Small‐Scale Bioreactors for Large‐Scale hiPSC Production

ABSTRACT

Human induced pluripotent stem cells (hiPSC) have great potential for cell therapy applications. To meet the global demand for hiPSC‐derived cell therapies, the implementation of scalable technologies, such as stirred‐tank bioreactors (STB), is essential. However, the addition of physical cues, including shear stress, can impact cell viability and proliferation and requires precise tuning. In this work, we used an engineering characterization approach to estimate the impeller power number (0.5) and investigate the mixing and suspension dynamics in the first generation of small‐scale (0.2 L) DASGIP bioreactors (DASGIP‐STB). By keeping constant power input per volume (P/V = 4.6 W/m³) as a scale‐up criteria, we successfully transferred a hiPSC expansion process to a 0.2 L single‐use STB (BioBLU‐STB) and scaled it up to a single‐use 2 L STB (Univessel‐STB) without compromising cell expansion, viability, and metabolism, as well as hiPSC quality attributes, including their pluripotent phenotype and differentiation potential.

Graphical Abstract and Lay Summary

Human induced pluripotent stem cells (hiPSCs) hold significant promise for regenerative medicine, and their large‐scale production requires robust and scalable bioprocesses. In this study, we applied an engineering characterization approach to estimate the impeller power number and analyze the mixing and suspension dynamics in the 0.2 L DASGIP stirred‐tank bioreactor (STB). This knowledge enabled the rational transfer of a hiPSC expansion protocol from the DASGIP system to a 0.2 L single‐use STB and its scale‐up to a 2 L single‐use STB, while supporting efficient hiPSC expansion, metabolism, and pluripotent phenotype.

Abbreviations

DISMT

dual indicator system for mixing time

DO

dissolved oxygen

EB

embryoid body

hiPSC

human induced pluripotent stem cell

H _max

maximum homogeneity index

N _P

impeller power number

P/V

power input per unit volume

Re

Reynolds Number

STB

stirred‐tank bioreactor

SUB

single‐use bioreactor

t _M

mixing time

U _T

impeller tip speed

1. Introduction

Human induced pluripotent stem cells (hiPSC) have the unique ability to proliferate continuously and differentiate into various cell types of the human body [1]. Since they can be derived from a patient's own cells, hiPSC serve as valuable platforms for patient‐specific in vitro disease models, drug discovery, and hold significant promise in cell therapy applications [2, 3]; depending on the target therapy, estimated cell doses can reach up to 10¹⁰ cells per patient [4].

The implementation of technologies for hiPSC culture, such as stirred‐tank bioreactors (STB), that allow the systematic production of cells in a robust and cost‐effective manner is thus essential [5]. These systems are easy to scale‐up, require less space and labor, and are equipped with the appropriate process‐monitoring tools for precise control and monitoring of (critical) process parameters, including agitation, temperature, pH, dissolved oxygen (DO), and medium/supplements feeding rates, all of which are critical to maintain optimal conditions for hiPSC culture [6, 7, 8]. Similarly, single‐use bioreactors (SUBs) have gained importance in cell therapy applications, offering many benefits over traditional glass or stainless‐steel equipment–improved efficiency, flexibility, lower costs, reduced contamination risks, and faster production times [9, 10]. Therefore, using rational approaches to transfer the existing protocols to SUBs has considerable benefits compared to developing these processes de novo.

To ensure effective culture mixing and cell suspension in STB while limiting shear stress exposure, the impeller power number (N _P), power input per unit volume (P/V), and the mixing and suspension characteristics are all important factors to consider [11, 12] in the bioprocessing of hiPSC, where the cells are the product. Previous studies have shown that shear stresses above 0.5 Pa can impact on hiPSC culture, causing cell damage, apoptosis, or spontaneous differentiation [13, 14, 15]. As such, studying the characteristic flow environment of a bioreactor system is critical from a scaling‐up perspective, particularly in the bioprocessing of hiPSC where the cells are the products. Key engineering parameters must be identified and used as scaling criteria, defining the limits within the design space to ensure successful bioprocess performance at larger scales.

In this work, we used an engineering characterization approach to estimate the impeller power number (N _P = 0.5) and investigate the mixing and suspension dynamics in the first generation of small‐scale (0.2 L) DASGIP‐STB. By keeping constant the P/V as a scale‐up criteria, we transferred hiPSC expansion process, previously established by our group in DASGIP‐STB [16], to a single‐use 0.2 L BioBLU‐STB and scaled it up to a single‐use 2 L Univessel‐STB.

2. Materials and Methods

2.1. Bioreactor Set‐Up

Two different STB systems were characterized in this study, equipped with radial (DASGIP‐STB) and axial (BioBLU‐STB) flow impellers. The sections below detail the working volume, geometry, and probe design characteristics of each vessel configuration. For the 2 L Univessel SU (Sartorius), we referenced the characterization studies performed and disclosed by the manufacturer, which included an analysis of power input under different agitation conditions [17].

2.2. DASGIP‐STB

A flat‐bottomed DASGIP parallel bioreactor (BioBlock) system (Eppendorf, Germany) with a height of 15.5 cm and a diameter of 6.2 cm was used. It featured a 6 cm diameter trapezoidal two‐blade paddle impeller positioned 1.4 cm above the bottom. Probes were 3D printed in clear acrylic‐based resin to replicate those used in the hiPSC expansion protocol published by our group [16] (Table 1).

TABLE 1.

Probes geometrical dimensions.

Bioreactor	Probe type	Probe number	Height (cm)	Diameter (cm)	Off‐bottom clearance (cm)
DASGIP‐STB	pH	x1	13.3	1.2	2.0
	O2	x1	12.6	1.2	2.7
	Sampling	x1	13.9	0.4	1.4
	Perfusion	x1	13.9	0.4	1.4
	Temperature	x1	14.0	0.4	1.3
BioBLU‐STB	pH	x1	11.5	1.2	0.8
	O2	x1	11.7	1.2	0.6
	Sampling	x2	11.3	0.4	1.0
	Perfusion	x1	11.3	0.4	1.0
	Temperature	x1	11.3	0.4	1.0

The working volume was set to 0.2 L. To program the impeller speed, the standard magnetically driven impeller was replaced with a top‐driven configuration. The motor was connected to an Ultra 3000 Servo drive and controlled using Ultraware software (Rockwell Automation, WI, USA). The system was mounted in a water‐filled glass trough to minimize light refraction, and images were captured using a NET iCube camera (NET, Germany) attached to an adjustable arm. The camera recorded images at increasing speeds for various agitation strategies. To reduce background noise and improve visualization, a white LED panel was placed behind the reactor.

2.3. BioBLU‐STB

A BioBLU 0.3sc single‐use vessel (Eppendorf) with a height of 12.3 cm and a diameter of 6.6 cm was used. It featured a 3.4 cm 8‐blade impeller with a 60° pitch, positioned 1.0 cm above the bottom. Probes were 3D printed in clear acrylic‐based resin to replicate those used in the BioBLU 0.3sc (Table 1). The working volume was set to 0.2 L. The motor was connected to an Ultra 3000 Servo drive and controlled using Ultraware software (Rockwell Automation, WI, USA). The system was mounted in a water‐filled glass trough to minimize light refraction, and images were captured using a NET iCube camera (NET, Germany) attached to an adjustable arm. The camera recorded images at increasing speeds for various agitation strategies. To reduce background noise and improve visualization, a white LED panel was placed behind the reactor.

2.4. Impeller Power Input Characterization

The power input from the impeller at various stirring rates was measured for DASGIP‐STB and BioBLU‐STB. The bioreactors were mounted on an air bearing system with pressurized air supplied through silicon tubing at 0.2 bar as per the experimental protocol described in [18]. This setup prevented self‐rotation but allowed smooth vessel motion during impeller agitation. The torque, M, generated by the impeller was calculated by multiplying the distance from the impeller axis, l, by the force, F, applied by a rotating rod rigidly mounted on the vessel. This force was measured with a digital force gauge (DFG55‐10, Omega Engineering, Manchester, UK), placed at l = 62 mm for the DASGIP‐STB configuration and l = 73 mm for the BioBLU‐STB. Prior to each measurement, the gauge was tared, and the fluid was allowed to reach a steady state for 60 s to minimize measurement fluctuations. Measurements were taken at a frequency of 10 Hz and averaged over 60 s, with three repetitions (n = 3) per condition. All measurements were conducted at room temperature (22.5°C) using MilliQ water or MilliQ water–glycerol mixtures (0%–100% glycerol, Fisher Scientific, UK) to adjust fluid viscosity and determine the power number at low Reynolds numbers (Re), Equation (1). Power numbers were determined for Re ranging from 6.5 × 10² to 7.9 × 10³, and the temperature of the working fluid was monitored before each experiment.

$$Re=\frac{N{D}^2{\rho }_L}{\mu }$$ (1)

where μ is the dynamic viscosity of the liquid, N the stirring rate, D the impeller diameter, and ρ_L the liquid density. The power input, P, and the power number, N _P, were calculated using Equations (2) and (3). The power input per unit volume, P/V, as a function of impeller speed (rpm) was fitted using a power function f(x) = a x ^b.

$$P=2\pi NM$$ (2)

$$N_{\mathbf{P}}=\frac{P}{N^3{D}^5{\rho }_L}$$ (3)

To calculate impeller N _P at specific Re, piece‐wise linear interpolation was applied to the power curve data.

2.5. Mixing Time

The Dual Indicator System for Mixing Time (DISMT) was used to measure mixing time as detailed in Rodriguez et al. [19]. Indicator stock solutions were prepared in 70% ethanol at concentrations of 1.52 mg/mL (Methyl Red) and 1.38 mg/mL (Thymol Blue) on a magnetic stirrer. These were then diluted in Milli‐Q water to 4.26 and 4.67 mL/L, respectively. Subsequently, NaOH was added in 5 µL increments until a pH of approximately 7.0 was measured using a pH probe (Mettler Toledo, USA). This working solution was added to the vessel and mixed by the impeller until the solution turned bright red with the addition of stoichiometric amounts of HCl (0.75 M). The impeller speed was set, and the mixing experiment started by adding a stoichiometric amount of 0.75 M NaOH at the free surface through an inlet using a pipette. The mixing process was monitored by a camera with a frame rate of 10 Hz. Image acquisition stopped when the solution turned fully yellow with no red or green color traces left over (pH ≈ 7.0) or after a maximum of 3 min. The same amount of 0.75 M HCl was then added to prepare for a new measurement. The working solution was replaced after five measurements to ensure consistent color changes. Mixing was tested at impeller speeds from 30 to 200 rpm in 10 rpm increments, with each condition repeated five times (n = 5). Images were analyzed using MATLAB, which measured the change in red, green, and blue (RGB) pixels in the images over time, defining mixing time, t _M, as the time to achieve 95% mixing of all pixels to a 95%‐degree, further details are available in [19, 20, 21].

2.6. Suspension Time

Two commercially available microcarriers, commonly used in cell culture, were used to mimic the heterogeneity in aggregate size in hiPSC cell culture: Cytodex 3 and Hillex (Table 2). Briefly, microcarriers were stained with 0.4% Trypan Blue (Sigma‐Aldrich, USA) and used at concentrations of 1.5 g/L (Cytodex 3)  and 5 g/L (Hillex) for improved visualization. Suspension speed was assessed by recording side view images of the reactor at 10 Hz for 3–5 min per condition. Impeller speeds from 10 to 300 rpm were tested, with images processed using a purposely written MATLAB code. Suspension speed (NH) and suspension number (NtH) were determined based on the time required to achieve 90% homogeneity. Suspension number was calculated by fitting homogeneity over time to a sigmoidal function with a 95% threshold [21]. A piecewise cubic hermite interpolating polynomial (PCHIP) function was applied to the data to calculate the impeller speeds required to achieve target homogenization levels (defined as 10% and 90% suspension of microcarriers). The temporal homogeneity profiles were modeled using a sigmoidal function, and a 95% homogeneity threshold was selected to determine the suspension number [18].

TABLE 2.

Microcarriers characteristics.

Characteristics	Cytodex 3	Hillex
Specific gravity (g/cm3)	1.04	1.11
D50 (wet, µm)	175	180
Nominal surface area per gram (cm2/g)	2700	500
Dry weight used (g/L)	1.5	5
Diameter (µm)	120–180	150–210
Manufacturer	Cytiva	Sartorius

2.7. hiPSC Expansion

The hiPSC line used in this work, CBiPS1sv‐4F‐40 (RRID: CVCL_V189), was derived from umbilical cord blood and registered at the Spanish National Stem Cell Bank. Cells were routinely propagated in T‐flasks using the Cellartis DEF‐CS Xeno‐Free 500 Culture System (Cat. No. Y30045, Takara BioEurope AB, Göteborg, Sweden) at 37°C in a humidified atmosphere with 5% (v/v) CO₂ and 95% (v/v) air. Cells were typically split every 3–4 days at 80%–90% confluency and seeded at 3–4 × 10⁴ cells/cm² according to manufacturer's manual. The culture medium was changed daily following the manufacturer's instructions. Cells were dislodged by rinsing with Dulbecco phosphate‐buffered saline (DPBS, Thermo Fisher Scientific, Waltham, MA, USA) and incubating with Versene (Thermo Fisher Scientific, Waltham, MA, USA) for 20 min at 37°C. Cells were then sedimented by centrifugation at 300 × g, and the pellet was resuspended in Cellartis DEF‐CS Xeno‐Free 500 culture medium.

2.8. hiPSC Expansion in STB

hiPSC expansion was performed in three distinct STB (DASGIP‐STB, BioBLU‐STB, and Univessel‐STB) operated in perfusion mode. hiPSCs were inoculated at a density of 3 × 10⁵ cell/mL as single cells and cultured in Cellartis DEF‐CS Xeno‐Free 3D Spheroid Culture Medium (Cat. No. Y30047, Takara BioEurope AB, Göteborg, Sweden) as described elsewhere [16]. During expansion, cells were cultured at 37°C with a surface aeration rate via overlay of 0.1 vvm, 5% CO₂, and dissolved oxygen at 4% O₂ (corresponding to 18% air saturation). Perfusion started 48 h post‐inoculation and was operated by automated gravimetric control, as described before [22], at a constant dilution rate (1.3 day⁻¹) for 2 days. The stirring rates used in each STB are summarized in Table 3.

TABLE 3.

Overview of impeller types, number, working volumes, stirring rates, N _P, and P/V across the different STB configurations.

STB type	Impeller type	Impeller number	Working volume (L)	N (rpm)	N P	P/V (W/m3)
DASGIP‐STB	Trapezoidal two‐blade paddle impeller	×1	0.2	80	0.5	4.6
BioBLU‐STB	8‐blade impeller (60° pitch)	×1	0.2	110	3.13	4.6
Univessel‐STB	3‐blade segment impeller (30° pitch)	×2	2	149	1.3[17]	4.6

2.9. DASGIP‐STB Configuration

Cells were expanded in DASGIP parallel bioreactor systems (BioBlock system, Eppendorf, Germany) in a working volume of 0.2 L. A flat‐bottom bioreactor vessels equipped with trapezoid‐shaped paddle impellers with long arms was used, as previously reported [23]. The stirring rate was set to 80 rpm (Table 3). A sintered glass sparger (16–40 µm pore size, Eppendorf) was integrated into the outlet perfusion line to act as a cell retention device, preventing the loss of cell aggregates.

2.10. BioBLU‐STB Configuration

BioBLU 0.3sc Single‐Use Vessels (Eppendorf) with an 8‐blade impeller (60° pitch) and a working volume of 0.2 L were used. To ensure consistency in hydrodynamic conditions, P/V was kept constant between the DASGIP‐STB and BioBLU‐STB systems. This was achieved by maintaining the energy transferred from the impeller to the fluid. The corresponding agitation speed for the BioBLU‐STB was estimated using Equation (4), based on the impeller power number (N _p), liquid density (ρ), impeller diameter (D_i ), agitation rate (N), and working volume (V):

$$\begin{array}{ccc}& & P/V_{\mathbf{DASGIP}-\mathbf{STB}}\hfill \\ & & =P/V_{\mathbf{BioBLU}-\mathbf{STB}}\iff {\left(\frac{N^3\rho D{\mathbf{i}}^5{N}_{\mathbf{p}}}{V}\right)}_{\mathbf{DASGIP}-\mathbf{STB}}\hfill \\ & & ={\left(\frac{N^3\rho D{\mathbf{i}}^5{N}_{\mathbf{p}}}{V}\right)}_{\mathbf{BioBLU}-\mathbf{STB}}\iff N_{\mathbf{B}\mathbf{i}\mathbf{o}\mathbf{B}\mathbf{L}\mathbf{U}-\mathbf{S}\mathbf{T}\mathbf{B}}\hfill \\ & & =\sqrt[3]{{\left(\frac{N^3\rho D{\mathbf{i}}^5{N}_{\mathbf{p}}}{V}\right)}_{\mathbf{DASGIP}-\mathbf{STB}}{\left(\frac{V}{\rho D{\mathbf{i}}^5{N}_{\mathbf{p}}}\right)}_{\mathbf{BioBLU}-\mathbf{STB}}}\hfill \end{array}$$ (4)

Based on this calculation, the stirring rate for the BioBLU‐STB was set to 110 rpm (Table 3), and a sintered glass sparger (16–40 µm pore size, Eppendorf) was integrated into the outlet perfusion line to act as a cell retention device, preventing the loss of cell aggregates.

2.11. Average Shear Stress Estimation

Empirical correlations derived from turbulent flow theory were used to estimate the average energy dissipation rate (ε) and corresponding shear stress (τ_mean) experienced by cells in suspension culture. The calculations were based on P/V = 4.6  W/m³, which was maintained constant across all bioreactor conditions as part of the scale‐up strategy.

$$\epsilon =\left(P/V\right)/\rho$$ (5)

$$\tau \_mean=C\times \rho \times \surd \left(\epsilon \times \nu \right)$$ (6)

The following values were used to estimate ε and τ_mean: fluid density (ρ = 1000 kg/m³), kinematic viscosity (ν = 0.691 × 10⁻⁶  m²/s at 37°C), and an empirical constant (C = 5.33).

2.12. Univessel‐STB Configuration

A 2 L Univessel SU STB (Sartorius), equipped with two 3‐blade segment impellers angled at 30° and operated at a working volume of 2 L, was used for the scale‐up. To preserve equivalent hydrodynamic conditions across scales, the P/V was matched to that of the DASGIP‐STB system. The agitation speed required for the Univessel‐STB was determined based on Equation (7), incorporating the impeller power number (N _p), fluid density (ρ), impeller diameter (D _i), stirring rate (N), and bioreactor working volume (V):

$$\begin{array}{ccc}& & P/V_{\mathbf{DASGIP}-\mathbf{STB}}\hfill \\ & & =P/V_{\mathbf{Univessel}-\mathbf{STB}}\iff {\left(\frac{N^3\rho D{\mathbf{i}}^5{N}_{\mathbf{p}}}{V}\right)}_{\mathbf{DASGI}\mathbf{P}-\mathbf{STB}}\hfill \\ & & ={\left(\frac{N^3\rho D{\mathbf{i}}^5{N}_{\mathbf{p}}}{V}\right)}_{\mathbf{Univessel}-\mathbf{STB}}\iff N_{\mathbf{Univessel}-\mathbf{STB}}\hfill \\ & & =\sqrt[3]{{\left(\frac{N^3\rho D{\mathbf{i}}^5{N}_{\mathbf{p}}}{V}\right)}_{\mathbf{DASGIP}-\mathbf{STB}}{\left(\frac{V}{\rho D{\mathbf{i}}^5{N}_{\mathbf{p}}}\right)}_{\mathbf{Univessel}-\mathbf{STB}}}\hfill \end{array}$$ (7)

Using this approach, the stirring rate for the Univessel‐STB was set to 149 rpm (Table 3), and a metal sintered sparger (D12*ID9.1*H90.5 mm, of 30–40 µm pore size, Hengko) was integrated into the outlet perfusion line to act as a cell retention device, preventing the loss of cell aggregates.

2.13. Assessment of hiPSC Concentration and Viability

Viable cell concentration was determined using the automated cell counter NucleoCounter NC‐200 (ChemoMetec), in accordance with the manufacturer's guidelines. Specifically, the “Viability and Cell Count Assay” protocol was applied for hiPSCs cultured in 2D static systems, while the “Viability and Cell Count—A100 and B Assay” protocol was used for 3D aggregates cultured in suspension. For the analysis of cell viability, hiPSC were incubated with 20 µg/mL of fluorescein diacetate (FDA, Sigma‐Aldrich) and 10 µg/mL with a DNA‐binding dye, propidium iodide (PI, Sigma‐Aldrich). Imaging was performed using an inverted fluorescence microscope (DMI6000 B, Leica). Metabolically active cells, that hydrolyzed FDA, exhibited green fluorescence and were classified as viable, whereas PI‐positive cells exhibited red fluorescence and were identified as non‐viable.

2.14. Aggregates Diameter Analysis

To monitor the formation and size of hiPSC aggregates, samples were collected from bioreactor cultures and distributed into 96‐well plates at a volume of 100 µL per well. Images were acquired using an inverted microscope (DMI6000 B, Leica) and subsequently analyzed using the open‐source Image J software (National Institutes of Health, Bethesda, MD, USA; ImageJ). To estimate aggregate diameter, the average Feret diameter was calculated from multiple image frames obtained during live imaging. A minimum of 100 aggregates per condition were assessed to ensure statistical robustness.

2.15. Flow Cytometry

To assess the surface marker expression profile of hiPSCs, flow cytometry analysis was carried out. Single‐cell suspensions were obtained after dissociation of (i) cell monolayers with Versene (Gibco, Thermo Fisher Scientific) for 20 min, or (ii) cell aggregates with Versene (Gibco, Thermo Fisher Scientific) for up to 40 min, at room temperature (RT, 18–22°C). For each condition, a minimum of 0.5 × 10⁶ cells were used. Cells were stained with primary antibodies against TRA‐1‐60, SSEA‐4, SSEA‐1, or with the appropriate isotype control (Table S1), followed by a 1 h incubation at 4°C. For unconjugated primary antibodies, cells were incubated with a secondary antibody for an additional 30 min, at RT protected from light (Table S1). Data acquisition was performed on a BD FACSCelesta Cell Analyzer (BD Biosciences), and data analysis was conducted using FlowJo software (FlowJo LLC, http://www.flowjo.com/). A minimum of 20,000 events were recorded for the target population in each sample.

2.16. In Vitro Pluripotency Assay

The assessment of cell pluripotency was conducted in vitro through the formation of embryoid bodies (EBs) and their subsequent spontaneous differentiation, as described elsewhere [16, 24]. Cell aggregates from Day 4 of the BioBLU‐STB and UniVessel‐STB were dissociated using Versene, and the resulting single‐cell suspension was cultured in suspension culture on ultra‐low attachment plates for 7 days in 20% (v/v) KO‐Serum, 1% (v/v) MEM‐NEAA, 0.1 mM 2‐mercaptoethanol, and 2 mM Glutamax (all reagents from ThermoFisher Scientific). Following this period, EBs were harvested and transferred to 0.1% (w/v) gelatin‐coated plates (Sigma‐Aldrich) and cultured for 2 weeks, with medium changed every 2 days. Cells were then fixed in 4% (w/v) paraformaldehyde with 4% (w/v) sucrose in DPBS for 20 min at RT. Fixed cells were washed twice with DPBS and permeabilized and blocked for 30 min with 0.1% (v/v) Triton X‐100 (Sigma‐Aldrich) and 0.2% (v/v) gelatin from cold water fish skin (Sigma‐Aldrich). Afterward, cells were incubated with primary antibodies (Table S2) at 4°C overnight in a humidified chamber. The next day, cells were washed three times with DPBS, and incubated with the secondary antibody (Table S2) for 1 h at RT. Nuclei were counterstained with DAPI reagent (ThermoFisher Scientific) in a 1:2000 dilution in DPBS.

2.17. Gene Expression Analysis

Following centrifugation at 300 × g for 5 min, cells were rinsed with DPBS, snap‐frozen in liquid nitrogen, and stored at −80°C until further processing. Total RNA was isolated using the High Pure RNA Isolation Kit (Roche), following the protocol provided by the manufacturer. RNA concentrations were determined with a NanoDrop 2000c spectrophotometer (Thermo Scientific). For complementary DNA (cDNA) synthesis, 100 ng of total RNA per sample was used with the Transcriptor High Fidelity cDNA Synthesis Kit (Roche). Quantitative real‐time PCR (RT‐qPCR) was carried out on a LightCycler 480 Instrument II (Roche) configured with a 384‐well plate format. The thermal cycling protocol included an initial denaturation at 95°C for 10 min, followed by 45 amplification cycles consisting of denaturation at 95°C for 15 s, annealing at 60°C for 1 min, and a final extension at 72°C for 5 min. All reactions were performed in triplicate, and gene expression levels were calculated using the 2^−ΔΔCt method [25], normalized against housekeeping genes RPLP0 and GAPDH. Primer sequences used for target genes are listed in Table S3.

2.18. Metabolic Profiling

To evaluate cellular metabolic activity, cells were centrifuged at 300 × g for 5 min, after which the supernatants were collected for biochemical analysis. Glucose and lactate concentrations were quantified using the Cedex Bio Analyzer (Roche). Specific rates of glucose consumption and lactate production were determined based on the general mass balance approach, as described in Equation (8):

$$q=\frac{\frac{\mathrm{\Delta }C}{\mathrm{\Delta }t}-D\times \left(C_{\mathbf{in}}-C_{\mathbf{out}}\right)}{\overline{{\mathbf{X}}_{\mathbf{V}}}}$$ (8)

where q represents the specific metabolic rate, ΔC/Δt corresponds to the rate of concentration change in the supernatant, D is the dilution rate, C _in and C _out denote the concentrations at the inlet and outlet, and X refers to the average viable cell concentration over the time interval Δt.

2.19. Statistical Analysis

All statistical evaluations were carried out using GraphPad Prism version 9.0.1. Results are presented either as mean ± standard deviation (SD) for technical replicates, or as mean ± standard error of the mean (SEM) when referring to independent biological experiments (n). The differences between groups were analyzed by using two‐way ANOVA. To assess the degree of similarity between the expansion profiles obtained in DASGIP‐STB, BioBLU‐STB, and Univessel‐STB systems, Pearson's correlation coefficient (r) was calculated, where a value of r = 1 indicates a perfect linear relationship between datasets.

3. Results and Discussion

3.1. Hydrodynamic Characterization of DASGIP and BioBLU Bioreactors

To ensure successful process scale‐up, it is crucial to be familiar with engineering parameters such as the impeller power number (N _P), homogeneity index (H), and mixing time (t _M). These parameters are commonly used in or as engineering correlations to ensure consistent hydrodynamic conditions (e.g., mixing efficiency, heat and mass transfer, cell suspension, shear stress) across different‐sized bioreactors [26, 27, 28]. In our study, we estimated the values of N_P , t _M, and H for two 0.2 L STB, one glass vessel STB (named “DASGIP‐STB” from now on), and one single‐use vessel (named “BioBLU‐STB” from now on), commonly used for hiPSC expansion.

Using the setup shown in Figure 1A, we were able to estimate the N _P at different flow regimes for both DASGIP‐STB (Figure 1B) and BioBLU‐STB (Figure S1A). At low Reynolds numbers (Re < 500), N _P decreased as Re increased; at Re > 500, N _P plateaued, indicating the transition to turbulent flow. This plateau reflects the natural baffling provided by internal probes [29]. Without baffles, power numbers usually decrease continuously with increasing Re due to solid body rotation in unbaffled tanks, leading to poor mixing [26, 30]. Since STB for suspension cell cultures are typically operated under a turbulent regime, the estimated power numbers were determined. For the DASGIP‐STB, N _P was 0.5. For the BioBLU‐STB, N _P values were direction‐dependent: 3.1 for clockwise impeller rotation and 3.3 for counterclockwise rotation. The H was assessed in DASGIP‐STB (Figure 1C) and BioBLU‐STB (Figure S1B) using microcarriers of different sizes and densities (i.e., Cytodex 3 and Hillex) (Table 2) to simulate the aggregate diameter heterogeneity of hiPSC during expansion process [16]. Cytodex 3 are smaller and less dense than Hillex, thus, the stirring rate needed to achieve a fully homogeneous distribution of microcarriers across the bioreactor volume is lower for Cytodex 3 than for Hillex, regardless of the vessel used. In the DASGIP‐STB configuration, 40 rpm was sufficient to fully suspend Cytodex 3 (i.e., maximum H), consistent with previous reports for the same setup without internal probes [20]; for Hillex microcarriers, maximum H was achieved at 80 rpm (Figure 1C). In our published hiPSC expansion protocol using the DASGIP‐STB setup [16], a stirring rate of 80 rpm ensured thorough mixing and suspension of a heterogeneous mixture of cell aggregates, thus supporting the data obtained herein. In the BioBLU‐STB configuration, higher stirring rates were needed, i.e., 60 rpm for Cytodex 3, and 110 rpm for Hillex (Figure S1B).

FIGURE 1.

Hydrodynamic characterization of DASGIP‐STB. (A) Experimental setup for the power number (N _p) measurements. (B) Power number (N _p) as a function of Reynolds from the laminar to turbulent regimes. (C) Homogeneity index (H) as a function of stirring rate (rpm), for Hillex and Cytodex 3 microcarriers, the dotted line represents the 95% homogeneity threshold. (D) Mixing time (t _M) as a function of stirring rate (rpm). Error bars represent SEM; n = 3 independent experiments.

The third engineering parameter examined was the t _M. Estimates of t _M in DASGIP‐STB (Figure 1D) and BioBLU‐STB (Figure S1C) were made for stirring rates ranging from 30 to 200 rpm. As anticipated, t _M decreased with increasing stirring rate; importantly, the values achieved in DASGIP‐STB were lower than those in BioBLU‐STB, suggesting that fewer revolutions are needed for complete mixing. For a stirring rate of 80 rpm, reported as optimal for hiPSC expansion in our previous study [16] and N _P of 0.5 (herein determined), the estimated P/V is 4.6 W/m³. Under this setup, the average shear stress estimated using established empirical correlations for turbulent flow (Equations 5 and 6) was 0.3 Pa, which is below the levels considered in other studies to cause cell damage, apoptosis, or spontaneous differentiation (i.e., 0.5–1.5 Pa), thereby impacting stem cell proliferation/differentiation capacity, phenotype, and purity [13, 14, 15]. Additionally, at such P/V value, 14 s (Figure S1D) and 18 impeller revolutions (Figure S1E) were sufficient for complete mixing. In contrast, 29 s and 54 revolutions were required to achieve full mixing at the same P/V in the BioBLU‐STB setup. These differences highlight the importance of thorough hydrodynamic characterization, as mixing time directly influences mass transfer efficiency and microenvironmental homogeneity. Longer t _M​ values can lead to transient nutrient and oxygen concentration gradients and affect cell aggregation, which may affect cell proliferation, viability, and differentiation by creating heterogeneous culture conditions [31].

The values of N _P, t _M, and H herein determined are within the expected for the type and volume of the bioreactors studied [18, 20, 21, 29]. In general, for similar impeller geometries (i.e., diameter, blade width, and blade length), axial impellers (as in BioBLU‐STB) induce lower power numbers than radial impellers (as in DASGIP‐STB) due to reduced drag and power consumption [32]. However, the number of blades of the two impellers used in our study were significantly different (2‐blades in DASGIP‐STB and 8‐blades in BioBLU‐STB). With an increasing number of blades, form dissipation increases and turbulence in the impeller region increases. In addition, the interaction between the flow streams from different blades increases with an increase in the number of blades, which can justify the lower N _P achieved in the DASGIP‐STB [33].

Maximum homogeneity index (H _max) varies with biological system, with larger size and/or higher density cell entities requiring higher stirring rates to achieve a fully homogeneous distribution across the bioreactor volume [34]. Finally, while small‐scale cell culture bioreactors are characterized by having mixing times in the order of seconds, large‐scale vessels tend to have mixing times that extend to minutes [35, 36].

3.2. Process Transfer From DASGIP to BioBLU Bioreactor

The hiPSC expansion protocol established in house based on the DASGIP‐STB [16] was adapted to the single‐use BioBLU‐STB, using P/V as the critical engineering parameter to keep constant across bioreactor types. To achieve a P/V = 4.6 W/m³ (identified as optimal for hiPSC expansion in the DASGIP‐STB) in the BioBLU‐STB, a stirring rate of 110 rpm would be needed; this value was estimated using the N _P reported in this study. Expansion of hiPSC cells in the BioBLU‐STB was assessed using the aforementioned stirring rate, and cell growth performance and phenotype were compared to DASGIP‐STB (Figures 2 and S2). Online monitoring of DO and pH confirmed that in both the DASGIP‐STB and BioBLU‐STB systems, stable culture conditions were maintained, with DO consistently controlled at 18% and pH close to 7.0 with minimal fluctuations (Figure S2A,B).

FIGURE 2.

hiPSC expansion in 0.2 L DASGIP‐STB and BioBLU‐STB. (A) Representative fluorescence images of hiPSC aggregates at Day 0, 2, and 4, stained with fluorescein diacetate (FDA, viable cells, green) and Propidium iodide (PI, dead cells, red). Scale bar = 200 µm. (B) Concentration of hiPSC during expansion, with the solid line representing the best linear fit to the data (r ², Pearson's correlation), and the shaded area represents the 95% confidence interval. (C) Diameter of hiPSC aggregates during expansion, with the solid line representing the best linear fit to the data (r ², Pearson's correlation), and the shaded area represents the 95% confidence interval. (D) Flow cytometry analysis of pluripotency‐specific markers (TRA‐1‐60 and SSEA4) and early differentiation markers (SSEA1), performed at last day of hiPSC expansion (Day 4). (E) Relative expression of pluripotency‐specific markers (Nanog and POU5F1) and early differentiation markers (MESP1 and SOX17), performed at last day of hiPSC expansion (Day 4). (F) Immunofluorescence images of in vitro spontaneous differentiation of hiPSC harvested at Day 4, cells were labeled for α‐SMA (mesoderm, in red), FOXA2 (endoderm, in green), and β‐III tubulin (ectoderm, in green). Nuclei were stained with DAPI (blue). Scale bars: 200 µm. Error bars represent SEM; ns, not significant; n _DASGIP‐STB = 3, n _BioBLU‐STB = 3 independent experiments.

Aggregation of hiPSC in the BioBLU‐STB was achieved, with high cell viabilities and increased aggregate size being observed over time (Figure 2A). Cell growth kinetics (including cell expansion factor, cell growth rate, and population doubling level, Figures 2B and S2C–F); aggregate size distribution (Figures 2C and S2G), and glucose consumption and lactate production rates (Table 4) were similar in both STB evaluated. Likewise, hiPSC phenotype (indicated by high percentages of TRA‐1‐60 and SSEA‐4 positive cells, and low percentage of SSEA‐1) was maintained across vessels (Figure 2D). In addition, the expression levels of genes considered relevant for early differentiation (MESP1, SOX17) and pluripotency (NANOG, POU5F1) were comparable in both platforms. hiPSC expanded in the 0.2 L BioBLU‐STB were able to form EBs and spontaneously differentiate into cells from ectodermal, mesodermal, and endodermal lineages (Figure 2F). These findings demonstrate that the hiPSC expansion protocol was successfully transferred from the glass‐vessel DASGIP‐STB to the single‐use BioBLU‐STB by maintaining comparable hydrodynamic conditions. This underscores the critical role of engineering parameters and correlations in achieving rational process transfer. While alternative conditions may influence process outcomes, the use of P/V provides a systematic framework to achieve reproducible results.

TABLE 4.

Cell specific glucose consumption (q _GLC) and lactate production (q _LAC) rates, and apparent lactate from glucose yield (Y _LAC/_GLC).

	DASGIP‐STB	BioBLU‐STB	Univessel‐STB
q GLC (µmol/(106 cell hour))	0.47 ± 0.02	0.47 ± 0.01	0.52 ± 0.03
q LAC (µmol/(106 cell hour))	0.74 ± 0.02	0.82 ± 0.03	0.85 ± 0.05
Y LAC/GLC	1.58	1.73	1.64

3.3. Process Scale‐Up From 0.2 L DASGIP to 2 L Univessel Bioreactor

The hiPSC expansion protocol established in the 0.2 L DASGIP‐STB was scaled‐up to the 2 L single‐use Univessel STB (named “Univessel‐STB” from now on), and cell growth performance and phenotype were compared across scales (Figures 3 and S3). The stirring rate to apply in the Univessel‐STB was estimated so that the resulting P/V value could match that identified as optimal for hiPSC expansion in the DASGIP‐STB, i.e., 4.6 W/m³; the stirring rate estimated was 148 rpm, based on the N _P value reported in the literature for the 2 L Univessel‐STB [17]. Online monitoring of DO and pH confirmed that in the Univessel‐STB system, stable culture conditions were also maintained, with DO consistently controlled at 18% and pH close to 7.0 with minimal fluctuations (Figure S3A,B).

FIGURE 3.

hiPSC expansion in 0.2 L DASGIP‐STB versus 2 L Univessel‐STB. (A) Representative fluorescence images of hiPSC aggregates at Days 0, 2, and 4, stained with fluorescein diacetate (FDA, viable cells, green) and Propidium iodide (PI, dead cells, red). Scale bar = 200 µm. (B) Concentration of hiPSC during expansion, with the solid line representing the best linear fit to the data (r ², Pearson's correlation) and the shaded areas represent the 95% confidence interval. (C) Diameter of hiPSC aggregates during expansion, with the solid line representing the best linear fit to the data (r ², Pearson's correlation) and the shaded areas represent the 95% confidence interval. (D) Flow cytometry analysis of pluripotency‐specific markers (TRA‐1‐60 and SSEA4) and early differentiation markers (SSEA1), performed at the last day of hiPSC expansion (Day 4). (E) Relative expression of pluripotency‐specific markers (Nanog and POU5F1) and early differentiation markers (MESP1 and SOX17), performed at last day of hiPSC expansion (Day 4). (F) Immunofluorescence images of in vitro spontaneous differentiation of hiPSC harvested at Day 4, cells were labeled for α‐SMA (mesoderm, in red), FOXA2 (endoderm, in green), and β‐III tubulin (ectoderm, in green). Nuclei were stained with DAPI (blue). Scale bars: 200 µm. Error bars represent SEM; ns: not significant; n _{STB_0.2L} = 3, n _{STB_2L} = 2 independent experiments.

Our results show that aggregates size increases over time while maintaining high viability, demonstrating successful hiPSC aggregation in the 2 L Univessel‐STB (Figure 3A). The hiPSC growth kinetics (including cell expansion factor, cell growth rate, and population doubling level, Figures 3B and S3C‐F), aggregates’ size distribution (Figures 3C and S3G), rates of glucose consumption, and lactate production (Table 3) were similar between the two culture systems. Likewise, pluripotent phenotype (Figure 3D) and expression levels of early differentiation (MESP1, SOX17) and pluripotency (NANOG, POU5F1) markers (Figure 3E) were comparable. In fact, hiPSC expanded in the 2 L Univessel‐STB were able to form EBs and spontaneously differentiate into cells from ectodermal, mesodermal, and endodermal lineages (Figure 3F). These findings validate the successful scale‐up of the hiPSC expansion process from 0.2 L (as established in the DASGIP‐STB and BioBLU‐STB) to the 2 L Univessel‐STB. Importantly, the results achieved emphasise the critical role of engineering parameters and correlations in achieving successful process scale‐up. For example, in a recent study, the scale‐up of hiPSC expansion from a 0.25 L Ambr250 STB to the 2 L Univessel STB was successfully achieved using impeller tip speed (U _T) as the scale‐up criteria (i.e., constant U _T across scales) [37], instead of P/V as in our study. U _T is frequently employed in the pharmaceutical industry for scaling up processes [38], either microbial fermentations or animal cell cultivations [39], alongside other engineering parameters and correlations such as P/V [40, 41], oxygen transfer rate [39], and volumetric mass transfer coefficient (kla) [39]. In our systems, U _T ranged from 0.20 to 0.42 m/s; if scale‐up was to be done based on U _T, this would imply lowering specific power input and subsequently mixing and mass transfer [41], potentially compromising cell growth. In contrast, maintaining P/V during scale‐up enhances mixing and mass transfer at larger scales, compared to a U _T scale‐up criterion. Although Reynolds number and U _T increase, the impact on shear rate and stress is negligible. Importantly, the Kolmogorov eddy length scale (λ) [41] remains constant, ensuring similar mixing and mass transfer across scales.

4. Conclusion

The results achieved in this study showcase the importance of engineering parameters and correlations for scaling up hiPSC processes. Using P/V as the scale‐up criterion, we were able to successfully transfer the hiPSC expansion process established at 0.2 L (for DASGIP‐STB and BioBLU‐STB) to the 2 L Univessel‐STB. From a 4‐day culture, we could reach 8 × 10⁹ hiPSC cells in the 2 L Univessel‐STB, a 10‐fold increase in the number of hiPSC cells when compared to the 0.2 L DASGIP‐STB and BioBLU‐STB. Considering the cell numbers required for therapeutic applications, which can reach up to 10¹⁰ cells per patient [4], the process established at 2 L using the Univessel‐STB would be able to closely deliver such numbers.

Author Contributions

Pedro Vicente: investigation, data curation, methodology, formal analysis, writing – original draft, writing – review & editing, visualization, conceptualization. Ana Meliciano: investigation, data curation, formal analysis, writing – review & editing, visualization. Cláudia Diniz: investigation, data curation, formal analysis, writing – review & editing, visualization. Artemis Charalambidou: writing – review & editing, visualization, methodology, supervision. Ana Paula Terrasso: writing – review & editing, visualization, methodology, supervision. Catarina Freitas: writing – review & editing, visualization, methodology, supervision. Andrea Ducci: resources, writing – review & editing, visualization, methodology, conceptualization, supervision. Paula M. Alves: resources, writing – review & editing, visualization, supervision, project administration, funding acquisition. Martina Micheletti: conceptualization, resources, methodology, writing – review & editing, visualization, supervision. António Roldão: conceptualization, resources, methodology, writing – review & editing, visualization, supervision, project administration, funding acquisition. Margarida Serra: conceptualization, resources, methodology, writing – review & editing, visualization, supervision, project administration, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: biot70106‐sup‐0001‐SuppMat.docx.

Vicente P., Meliciano A., Diniz C., et al. “Engineering Characterization of Small‐Scale Bioreactors for Large‐Scale hiPSC Production.” Biotechnology Journal 20, no. 9 (2025): 20, e70106. 10.1002/biot.70106

Data Availability Statement

Data will be made available on reasonable request.