Optimizing and developing a scalable, chemically defined, animal component-free lentiviral vector production process in a fixed-bed bioreactor

Abstract

Lentiviral vectors (LVVs) play a critical role in gene delivery for ex vivo gene-modified cell therapies. However, the lack of scalable LVV production methods and the high cost associated with them may limit their use. In this work, we demonstrate the optimization and development of a scalable, chemically defined, animal component-free LVV production process using adherent human embryonic kidney 293T cells in a fixed-bed bioreactor. The initial studies focused on the optimization of the culture process in 2D static cultures. Process changes such as decreasing cell seeding density on day 0 from 2.5 × 10⁴ to 5 × 10³ cells/cm², delaying the transient transfection from 24 to 120 h post-seeding, reducing plasmid DNA to 167 ng/cm², and adding 5 mM sodium butyrate 6 h post-transfection improved functional LVV titers by 26.9-fold. The optimized animal component-free production process was then transferred to the iCELLis Nano bioreactor, a fixed-bed bioreactor, where titers of 1.2 × 10⁶ TU/cm² were achieved when it was operated in perfusion. In this work, comparable functional LVV titers were obtained with FreeStyle 293 Expression medium and the conventional Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum both at small and large scale.

Keywords: animal component-free; chemically defined; fixed-bed bioreactor; gene therapy; iCELLis Nano bioreactor; immunotherapy; lentiviral vectors; scalability; serum-free; virus.

Graphical abstract

Figure 1

Cell growth kinetics, LVV production yields, and metabolite analysis phase I of the optimization (A) Cell growth (live cells/cm²) after seeding 5 × 10³, 7 × 10³, and 2.5 × 10⁴ cells/cm², (B) functional LVV titers (TU/cm²), and (C) cell productivity (TU/cell) after seeding the aforementioned three seeding densities and transfecting cells with a total plasmid DNA concentration of 333 ng/cm². (D–F) Glucose consumption and lactate/ammonia production for the different seeding densities 2.5 × 10⁴ cells/cm² (D), 7 × 10³ cells/cm² (E), and 5 × 10³ cells/cm² (F) during the experiments and measured daily from the cell culture medium. Data of results are shown as mean values ±SD of n = 3 technical replicates. Statistical analysis was performed using ordinary one-way ANOVA with Tukey’s multiple comparisons test and significance is shown when p values were ∗∗p ≤ 0.01, ∗∗∗∗p ≤ 0.0001. Only relevant statistical analysis is shown.

Figure 2

LVV production yields during phase II of the optimization (A) Functional LVV titers (TU/cm²) and (B) cell productivity (TU/cell) after seeding 5 × 10³ cells/cm² using two different total plasmid DNA concentration concentrations. Data of results are shown as mean values ± SD of n = 3 technical replicates. Statistical analysis was performed using ordinary two-way ANOVA with Tukey’s multiple comparisons test and significance is shown when p values were ∗∗∗∗p ≤ 0.0001. Only relevant statistical analysis is shown.

Figure 3

LVV production yields during phase III of the optimization (A) Functional LVV titers (TU/cm²) and (B) cell productivity (TU/cell) after seeding 5 × 10³ cells/cm² when medium was exchanged before addition of DNA/PEI complexes (B) or 6 h after the transfection (A). Additives were added either from the beginning of the culture on day 0 (pre-transfection) or 6 h after transfection (post-transfection). Additives, non-essential amino acids + chemically defined lipids + sodium pyruvate; NaBu, sodium butyrate. Data of results are shown as mean values ± SD of n = 3 technical replicates. Statistical analysis was performed using ordinary one-way ANOVA with Tukey’s multiple comparisons test and significance is shown when p values were ∗p ≤ 0.05, ∗∗p ≤ 0.01, ∗∗∗∗p ≤ 0.0001; ns, not significant. Only relevant statistical analysis is shown.

Figure 4

Functional titers during the final studies of the optimization (A) Functional LVV titers (TU/cm²) and (B) cell productivity (TU/cell) after seeding 5 × 10³ cells/cm². Data of results are shown as mean values ± SD. Statistical analysis was performed using ordinary two-way ANOVA with Tukey’s multiple comparisons test and significance is shown when p values were ∗∗p ≤ 0.01, ∗∗∗p ≤ 0.001.

Figure 5

Cell growth kinetics and metabolite analysis in the iCELLis Nano bioreactor (A) Cell growth (total live cells/cm²) after seeding 2.5 × 10⁴ cells/cm² in run 1 or 5 × 10³ cells/cm² for runs 2 to 8, (B) average cell growth in runs 2 to 8, (C) cells distribution in the fixed-bed of the iCELLis Nano bioreactor by the end of runs 3 to 8, (D) glucose levels, and (E) lactate production for the different bioreactor runs. Transfection at 48 h corresponds to run 1, while transfection at 120 h corresponds to runs 2 to 8. Data represented correspond to eight different iCELLis Nano runs. For each time point in (A and C), three different macrocarriers were sampled. Statistical analysis was performed using ordinary one-way ANOVA with Tukey’s multiple comparisons test (ns, not significant).

Figure 6

LVV production yields in the iCELLis Nano bioreactor (A) Fluorescent microscopy image of a macrocarrier sampled from the fixed-bed after GFP transfection, (B) functional LVV titers (TU/cm²), (C) cell productivity (TU/cell), (D) total physical LVV particles, and (E) VP/TU ratio in the different bioreactor runs. Dashed division lines indicate (left to right) runs using pre-optimization conditions, runs in batch mode, and runs in perfusion mode. TU, transducing units; VP, total physical LVV particles. Scale bar, 1,000 mM in (A).