DISCUSSION
In Figure 1, we observe that plasmid DNA availability plays an essential role in the maintenance of VG titer when cell density at the time of transfection is increased from 1 to 2 million cells per mL. In low-cell-density transient transfection, plasmid DNA is usually delivered on a volumetric basis [45, 47-50]. Like our study, Grieger, Soltys and Samulski [48] also observed a titer increase during production when cell density and plasmid concentration were doubled. The need for a higher plasmid DNA concentration was also observed for productions at 4 × 106 cells/mL (medium cell density, MCD). A 4-fold increase in the number of cells at the time of transfection resulted in an almost linear increase in viral titer (VG/mL) only when plasmid DNA was delivered per million cells (Figure 2B). Moreover, the cell-specific yield was maintained during MCD production, with plasmid DNA concentration maintained on a per-cell basis (Figure 2B). Yields between 1,500 and 3,000 VG/cell were obtained, similar to those reported by Chahal, Schulze, Tran, Montes and Kamen [47]. The improvement in titer was also observed for functional particles (ETU). The functional titer of the MCD production with plasmid DNA delivered on a cell basis was maintained constant (Figure 2D), resulting in a difference of up to 3.9 times compared to the LCD control at 96 hpt. At LCD, the slight decrease in transducing units, but not in genome-containing particles, could be explained by the loss in functionality of the viral vectors. AAV vectors are thermally stable, but their transduction efficiency decreases when maintained at 37°C and as the pH decreases [60]. Extending the culture to 96 hpt can decrease intracellular and extracellular pH due to the accumulation of metabolites, such as ammonia and lactate [61, 62].
The VG/ETU ratio measures viral functionality, showing the proportion of genome-containing viruses competent to transduce a target cell. At 48 hpt, the MCD production had a mean VG/ETU ratio of 67.1 (Figure 2E). Although this value is higher than reported in the literature [47, 48], there was no statistically significant difference from our LCD control. As shown in Figure 7, the production of AAV6 at medium cell density was successfully performed at a 3-L bioreactor. The cells were inoculated into the bioreactor in fresh medium at the desired cell density of 4 × 106 cells/mL to prevent detrimental effects from the spent medium. At larger scales, this could be achieved, for example, by medium exchange using Tangential Flow Depth Filtration (TFDF) as a cell retention device [63, 64]. This bioreactor production resulted in 30- and 7.5-fold increase in VG/mL and VG/cell titers, respectively, compared to previously reported production at a similar scale [47]. Surprisingly, the viral vector yield on the bioreactor was around 10-fold higher than the small-scale satellite culture. Similarly, the LCD bioreactor production also showed a higher yield than the small-scale experiments. This enhanced production in the bioreactors could be associated with better-regulated culture conditions, such as dissolved oxygen and pH. Again, no loss in functional titer was observed, with a maximum value of 7.8 × 109 ETU/mL achieved 48 hours post-transfection and a VG/ETU ratio of 4.6 (Figure 7D and E). These results show an improvement from previous reports [47, 48].
When productions were conducted at cell densities other than 1 × 106 cell/mL, the transfection efficiency, measured by the expression of the transgene 24 hours post-transfection, decreased to about 30% independently of the cell density (Figure 1B, Figure 2C, and Figure 3A). However, the percentage of cells expressing the transgene did not correlate directly with production efficiency. Others observed this same phenomenon during the production of viral-like particles via transient transfection of HEK293SF suspension cells [65]. Hildinger, Baldi, Stettler and Wurm [66] concluded that the transfection efficiency reduction resulted from DNA being provided on a volumetric basis. However, increasing the plasmid DNA availability in our study did not significantly improve transduction efficiency (Figure 1B, Figure 2C, and Figure 3A), as measured by the detection of the product of one of the three plasmids, which corroborates previous findings [47]. Increased energetic demand is reported to occur at higher cell densities and to allow recovery from transfection, an event known to be cytotoxic [67]. Transfection efficiency did not improve during production in a perfusion-like mode, even when the medium was supplemented (Figure 5B). The transfection efficiency was even lower at a higher glucose concentration. Lavado-García, Jorge, Cervera, Vázquez and Gòdia [67] partially explained the low transfection efficiency as a result of downregulated pathways involved in lipid biosynthesis and nuclear transportation of intracellular proteins. The transfection step, per se, is a known bottleneck in the production of AAV vectors. A recent study from our group showed that the proportion of cells that produce viral particles is as low as 7%, despite high transduction efficiency [68].
Recently, the improvement of AAV8 production at high-cell density with a medium exchange strategy was reported [64]. Conversely, the cell density effect was confirmed during our high-cell-density productions (Figures 3B and 5). CDE refers to the diminished production at high cell densities resulting from decreased cell-specific productivity [55, 69]. This effect has been previously documented for the generation of AAV using other production systems, such as insect cells [53, 59, 70-72]. During production, 25 to 30% of the AVV6 vectors are released in the supernatant [47]. Because of the high cell density and the elevated shaking speed to properly oxygenate the bioreactors used, it is possible that the cells were under increased sheer stress and could have released more vectors into the supernatant, which were removed during the medium exchange. This could explain the sudden drop in viral titer at 96 hpt in the HCD production (Figure 3B), which is not expected since AAV vectors are considered very stable at a wide range of temperatures [60, 73]. The same drop, however, was not observed when the medium was supplemented (Figure 5C).
The cell density effect is thought to occur mainly due to metabolic limitations caused by low concentrations of nutrients or the accumulation of inhibitory metabolites [53, 55, 74]. During the production at HCD, a perfusion-like mode was employed to provide enough nutrients to support high cell density and remove inhibitory metabolites; however, the conditions selected were insufficient to prevent the CDE (Figure 3). The glucose level dropped significantly at HCD, even though the medium was fully exchanged daily. For that reason, the basal medium, HyCell TransFx-H, which contains around 6 g/L of glucose, was supplemented with either glucose or Cells Boost 5. The amount of supplement added was based on the cell-specific consumption rate of glucose so that a minimum of 2 g/L of glucose was available at any given time [75], mimicking the observed glucose concentrations in the LCD productions. Medium supplementation with glucose alone did not restore the cell-specific productivity of AAV6 (Figure 5C). The CDE was alleviated, and the cell-specific yield (VG/cell) was restored only when a medium supplemented with 15% Cell Boost 5 was used (Figure 5C), indicating that other critical nutrients were limiting viral vector yield. Besides glucose, other components of the Cell Boost 5 supplement are thought to contribute to alleviating the CDE. While producing adenoviral vectors using the HEK293SF cells, Shen, Voyer, Tom and Kamen [76] emphasized that the composition of the medium used is crucial to support high yields at HCD. However, the complexity of culture media and supplements, such as Cell Boost 5, which can contain hundreds of different components at variable concentrations, complicates the understanding of individual nutrients’ effect on viral vector productions [55]. An optimized cell culture medium to produce AAV vectors could improve titers and vector quality, as seen and suggested for other viral vectors [61, 67, 75, 76].
As discussed above, supplementing the medium with Cell Boost 5 alleviated the cell density effect in terms of genome-containing particles; however, the same was not observed for functional particles. Whereas there was an increase in functional yield when the medium was supplemented, the cell-specific functional titer (ETU/cell) was not fully maintained, resulting in higher VG/ETU ratios compared to the LCD control (Figure 6). One possible explanation is the misassembly of the viral capsid, which contains a stoichiometry of 1:1:10 (VP1:VP2:VP3). If this proportion is altered, the efficacy of the AAV to deliver the transgene decreases [77]. Due to nuclear localization signals, the VP1 and VP2 proteins play a crucial role in transduction, specifically in nuclear transportation [78]. The N-terminus of VP1 also contains a phospholipase A2 domain responsible for viral escape from the endosome [79-82]. Our results highlight the importance of assessing viral transduction efficacy by measuring biologically active vectors [83] as part of optimizing AAV vector production. This measurement is sometimes overlooked by authors [49, 51], even though the quality attributes of a viral vector, including functional titer, are an essential part of the potency tests recommended by the FDA [84].
In summary, we demonstrated that it is possible to produce AAV6 via triple transient transfection at medium cell density without losing cell-specific productivity or functional titer. For that, the plasmid DNA must be provided on a cell basis. Our medium-cell-density productions achieved titers at the order of 1010 VG/mL of crude lysate. The yields obtained by this method represent a significant increase compared to a well-established production protocol [47]. The medium-cell-density method for AAV vector production described in this manuscript was shown to be efficient at a 3-L bioreactor scale. It could be used as the foundation for large-scale production processes, potentially contributing to solving the current vector shortage in AAV manufacturing. The cell density effect could be alleviated at a higher cell density, resulting in similar cell-specific productivity (VG/cell) when a perfusion-like operation was conducted with a supplemented medium. However, the cell-specific functional productivity (ETU/cell) was reduced, highlighting the importance of evaluating both viral genomes and transducing units during bioprocess optimization. Further research is necessary to fully understand how the cell density effect results in reduced functional titer and how to alleviate it fully.