The experimental materials comprised various instruments and equipment. These included a 5-L STR (SRT-5, Bioengineering AG, Wald, Switzerland), as well as 5-L, 50-L, and 500-L IFSB vessels (IFSB-5, IFSB-50, IFSB-500, Zhejiang JYSS Bio-Engineering Co., Ltd., Zhejiang, China). Other essential apparatus included a carbon dioxide oscillation incubator (Infors HT, Bottmingen, Switzerland), Vi-Cell Viability Analyzer (Beckman Coulter, California, United States), Microplate reader (Bio-Rad Laboratories, California, United States), Autodesk Prime protein purification system (Zhejiang JYSS), Protein A affinity HPLC (GALAK Wuxi, China), IEX-HPLC (Thermo Fisher Scientific, MA, US), and SEC-UPLC (Waters, MA, US). Reagents for cell culture, including the growth medium, were sourced from Shanghai Sinomab Biotechnology Co., Ltd. Additionally, anhydrous glucose was procured from Shandong Xiwang Pharmaceutical Co., Ltd., and sodium bicarbonate from Nanchang Baiyun Pharmaceutical Co., Ltd. The kit for measuring nutrients was obtained from the Nanjing Jiancheng Bioengineering Institute.

In this study, we initially assessed the process performance parameters and product quality attributes involved in the preparation of antibody-based biotherapy on a 5-L scale using STR-5 and IFSB-5. The capability of IFSB-5 and its associated consumables in the production of the antibody-based biotherapeutic is assessed, providing key technical data for scaling up cell culture. Rather than opting for a direct scale-up from the 5-L scale to the 500-L scale, we chose a more conservative approach. This involved an intermediate step utilizing a 50-L bioreactor, followed by the eventual transition to the 500-L bioreactor. This stepwise approach serves to minimize the risk of failure by enabling early problem diagnosis on a smaller scale, allowing us to refine our strategy for the subsequent steps. Additionally, it helps to limit material and labor loss in case of any unforeseen challenges. All 5-L–500-L bioreactors were operated in fed-batch mode.

The recombinant CHO K1 cells expressing the antibody-based therapeutic (an anti-TNFα IgG1 antibody) were obtained from Shanghai Zhanjiang Biotechnology Co., Ltd, Shanghai, China. The cell line culture process comprised cell recovery, cell expansion, and a final fed-batch culture. The culture was grown in a serum free medium (CHOM-B03) and equilibrated with a 5% CO₂/air mixture overnight before seeding. It was maintained in a CO₂ incubator with a 5% CO₂/air mixture at 37°C. For the 5 L scale culture, primary seeds were shake flask amplified and after expansion to 0.9 L, seeds were transferred to a 5 L bioreactor at a 1:4.4 inoculation ratio to initiate the culture. For the 50 L scale culture, primary seeds were first expanded in shake flasks, followed by a secondary expansion in 5 L bioreactors. Once the culture volume reached 5 L, it was transferred to a 50 L bioreactor at a 1:4 inoculation ratio. The culture was maintained until day 2, at which point a second feed expanded the volume to 40 L, continuing thereafter with fed-batch cultivation. For the 500 L scale culture, the seeding process mirrored that of the 50 L scale for the initial two stages, with the third stage of seed expansion occurring in a 50 L bioreactor. After the culture volume increased to 50 L, it was transferred to a 500 L bioreactor, also at a 1:4 inoculation ratio. On the second day, the culture was fed twice, reaching a total volume of 400 L, and was then maintained using the fed-batch cultivation method. The batch inoculation density should be approximately 1.0-1.2×10⁶ cells/mL, and the cell viability should not be less than 90%. Cell density, viability, cell status, and glucose levels were measured daily during the culture period. Product expression was measured on days 8, 10, 12, 14, and 16, and product quality attributes were measured after purification by affinity chromatography.

The incubation period lasted for 16 days during which the temperature conditions were set at 37°C for day 0 to day 7 and 34°C for day 8 to day 16. The pH was meticulously regulated within the range of 6.7–7.2 using a combination of CO₂ and sodium carbonate. Dissolved oxygen (DO) levels were consistently kept at 40% air saturation across all IFSBs by means of interaction between the gas-exposed smooth surface of the vessel and the culture medium inside. In he STRs, the DO level was controlled by introducing air at a rate of 1.7 mL/min and oxygen at a rate of 1–10 mL/min. This rate was automatically adjusted based on the DO level. For 5-L, 50-L and 500-L IFSBs, the rocking speeds were consistently maintained at 50–52 rpm, 32–33 rpm and 21.5–24.5 rpm respectively. Simultaneously, the stirring speed of the 5-L STR was held within the range of 230–260 rpm.

The purification process followed standard antibody purification protocols, which involved protein A capture followed by polishing steps. Initially, proteins were purified using affinity chromatography with an AutoPrep column system, which was equilibrated with a solution containing 20 mM Tris, 150 mM NaCl, and a pH range of 6.8–7.2. The flow rate during this step was maintained at 1.0 mL/min. The culture supernatant was then filtered through a 0.22 μM filter membrane, up-sampled, and subsequently eluted using an elution solution consisting of 25 mM Citrate and a pH range of 2.8–3.2. The equilibrium and eluent used for cation exchange chromatography consisted of a 20 mM solution of sodium acetate-acetic acid, 30 mM of NaCl, with a pH range of 4.8–5.2 by 1.2 mL/min. Similarly, for hydrophobic chromatography, the equilibrium and eluent comprised a 20 mM solution of sodium acetate-acetic acid, 300 mM of NaCl, with a pH range of 4.8–5.2 by 1.5 mL/min.

Initially, the process performance parameters and product quality attributes were compared between the IFSB and STR at a 5-L scale, providing essential technical reference data for scaling up cultivation. Operational parameters were then determined for the 50-L and 500-L IFSBs based on the proposed scaled-up principles, and the transition to larger bioreactors was conducted in a stepwise manner. Bioprocess performance and product quality were compared across scales to further assess the outcomes of the scale-up process.

To investigate the flow field and assess the performance of the IFSB, computational fluid dynamics (CFD) simulations were employed to analyze the shear force acting on the IFSB. The detailed simulation methodology is outlined in the Supplementary Material. The IFSB is propelled by an oscillator, which can be conceptualized as a rigid body undergoing horizontal rotation with eccentricity on an oscillating platform. The trajectory of each point on the IFSB remains consistent, ensuring uniform speed and acceleration.

The scale-independent process parameters, such as medium, temperature, pH, DO, feed regimen, seeding density, and culture cycle, remained consistent across different scales. Figures 1A–C illustrates the typical control profiles of the bioreactor temperature, pH, and DO for STR-5, IFSB-5, IFSB-50, and IFSB-500. The temperature profiles were effectively maintained around the set point. For STR-5 and IFSB-5, it was 37°C from day 1 to day 7°C and 34°C from day 8 to day 16. In the case of IFSB-50 and IFSB-500, the temperature was 37°C from day 1 to day 5°C and 34°C from day 6 to day 16. The DO profiles exhibited a declining phase followed by a control phase, marked by band-like spikes around the set point of 40% air saturation due to oxygen supplementation. pH profiles fluctuated within the set range of 6.9–7.2 for all scales.

The scale-dependent process parameters, including agitation speed, batch volume, and gas flow rate, are presented in Table 1. The scale-dependent process parameters for the 5-L STR refer to the previously well-established and stable cultivation process. The stirring speed and surface air flow rate for IFSB-5 were set at 50 rpm and 1.7 mL/min, respectively, based on the characteristics and application experience of IFSBs. The O₂ flow rate and CO₂ flow rates were determined according to the DO and pH levels of the culture medium, ranging from 0 to 10 mL/min, respectively. The scale-dependent process parameters for IFSB-50 and IFSB-500 are also listed in Table 1. The suggested agitation speeds were calculated based on the 50 rpm agitation speed set in the 5-L bioreactor, using Eq. 2. Additionally, the O₂ flow rate and CO₂ flow rates were calculated based on a 5-L bioreactor, using Eq. 1.

The growth, product formation, and performance of STR-5 and IFSB-5 cultures were assessed by comparing their growth profiles, glucose consumption, and antibody production. IFSB-5 exhibited a superior maximal VCD compared to STR-5 (3.3 × 10⁷ vs 2.2 × 10⁷ viable cells/mL, on average). Cell viability remained consistently high in both bioreactors, surpassing 90%, with no significant differences observed. The integral of the viable cell concentration over time (IVCD) for IFSB-5 progressively exceeded that of STR-5. To maintain optimal conditions, glucose concentration was sustained and supplemented at 5 g/L whenever it fell below 4 g/L. Illustrated in Figure 2, IFSB-5 demonstrated a higher daily glucose consumption than STR-5 from day 9 onwards, attributed to its elevated VCD.

In Figure 3, antibody production was measured from day 8 onwards in the culture. The results indicated a gradual increase in antibody concentration, with IFSB-5 showing a higher final antibody concentration (3.70 g/L) compared to STR-5 (2.56 g/L). The specific productivity (Qp) remained comparable for both IFSB-5 and STR-5, suggesting that the enhanced antibody production was attributed to an increase in the cumulative cell count.

The quality of the various products was assessed by examining critical attributes of antibody-based biotherapeutics, including aggregation, structural integrity, and charge variants.

Antibody aggregates, formed through intermolecular forces during production, pose potential harm, triggering immune responses in humans, leading to hypersensitivity reactions (Lundahl et al., 2021). Such aggregates, if present during intravenous administration, may even lead to capillary blockage and reduced microcirculation. Therefore, controlling aggregates is crucial for the development and production of protein-based drugs. The HPLC-SEC results indicated that the IFSB-5 product had a purity of 99.4% with 0.6% aggregates, while the STR-5 product had a purity of 99.3% with 0.7% aggregates (Table 2). The presence of aggregates in both bioreactors was minimal.

Charge variants, categorized as acidic and basic, can impact various aspects of antibody-based biotherapeutics. These include binding activity, bioactivity, pharmacokinetics, and structural stability (Beck et al., 2022). The effect of bioreactors on product quality was further assessed by analyzing charge variants using IEX HPLC. IEX HPLC results showed that the basic variants of IFSB-5 and STR-5 were comparable (19.4% vs 18.4%, respectively), while the acidic variants of IFSB-5 showed improvement (21.5% vs 27.0%, respectively). The main peak of IFSB-5 was higher than that of STR-5 (59.1% vs 54.6%) (Table 2).

Fc glycosylation profiles significantly impact the clinical efficacy and safety of antibody-based biotherapeutics. Mannose-type N-glycans may affect efficacy, pharmacokinetics, immunogenicity, and stability. Marketed antibody-based biotherapeutics exhibit significantly higher levels of mannose glycosylation compared to endogenous biotherapeutics (Mastrangeli et al., 2020). Examination of high mannose glycosylation levels revealed that IFSB-5 and STR-5 had similar levels of high mannose-type N-glycans (2.14% vs 2.10%, respectively), both displaying low levels of high mannose glycosylation (Table 2).

The bioprocess performance and product quality of the IFSB at a 5-L scale were comparable to or even better than those of a STR. Therefore, the scalability of the IFSB process was further assessed at 50-L and 500-L scales. The growth and product formation at each scale were qualitatively evaluated by comparing the profiles of growth, glucose consumption, and antibody production. The results indicated that the maximum VCD of IFSB was similar at all scales, with average values of approximately 3.29 × 10⁷, 3.32 × 10⁷, and 2.85 × 10⁷ viable cells/mL for IFSB-5, IFSB-50, and IFSB-500 respectively (IFSB-500 exhibited a slightly lower VCD). There was no significant change in cell viability or IVCD between the scales, with cell viability consistently remaining above 90%. Glucose concentration was effectively controlled though daily supplementation when it fell below 4 g/L. The culture cycle for IFSB-500 was extended to 18 days due to its excellent cell status on day 16 (Figure 4).

As illustrated in Figure 5, the assessment of antibody production revealed a consistent final product titer between IFSB-5 and IFSB-50 (3.70 g/L for both). However, IFSB-500 exhibited a slightly lower titer on day 16 (2.95 g/L vs 3.70 g/L), possibly attributed to the lower VCD. Nevertheless, due to the robust cell condition, the culture cycle was extended to 18 days, resulting in IFSB-500 achieving a final titer of 3.51 g/L comparable to IFSB-5 on day 16.

The comparability of the scaled-up process with the IFSB process was further assessed through analytical testing of the products generated at each scale. The results indicated that the purities of IFSB-5, IFSB-50, and IFSB-500 were comparable, exhibiting similar acidic variants (21.5%, 20.6%, and 22.7%), basic variants (19.4%, 18.8%, and 22.4%), and high mannose-type N-glycans (2.14%, 2.26%, and 1.85%, respectively). Minimal aggregation levels were observed, and these decreased further at the IFSB-500 scale. Extending the culture period to 18 days resulted in an increase in acidic variants (from 22.7% to 25.3%), while basic variants remained stable (Table 3).

Given an IFSB with good process performance and scalability, this study simulated the shear of an IFSB using CFD fluid simulation. The Supplementary Material provides a detailed description of the simulation methodology. The results indicate that the volume-averaged shear rate of the IFSB tends to decrease with increasing reactor volume. Furthermore, the volume-averaged shear rate for all IFSB sizes was found to be less than 20 s^-1. The shear rate generated by the IFSB is significantly lower than the maximum shear rate that mammalian cells can withstand. As a result, the damage caused by IFSBs to CHO cells can be neglected (Born et al., 1992; Chalmers and Ma, 2015). Therefore, IFSBs meet the mechanical requirements of mammalian cell cultures.