Results and Discussion

Cell growth of Vero cells in suspension cultures

Screening for commercial media

Suspension adapted Vero cells grew well in shake flasks in IHM03 medium up to a cell density of around 2 × 106 cells/mL and with doubling times of around 48 h as previously reported (Shen et al., 2019). IHM03 is an in-house medium developed and produced in small batches by the NRC, supporting the growth and virus production of Vero suspension cultures. As reported by Shen et al. and Rourou et al., media composition is critical for successfully generating a suspension adapted Vero cell line and no commercial media was able to support Vero cell adaptation so far (Rourou et al., 2019; Shen et al., 2019). Despite these recently reported adaptation successes with in-house media, establishing a process using commercial media could reduce the risk of lot-to-lot variations and would make the platform more amenable to work under standard conditions when media supply is assured.
Therefore, efforts have been dedicated to assessing different commercial media. After 20 passages of gradual adaptation in shake flasks, a Vero cell line was obtained that was able to grow in MDXK medium (Xell AG) supplemented with 4 mM Glutamax and which exhibited cell doubling times of around 48-72 h (data not shown). Other commercial media that were tested, but did not support growth of Vero cells in suspension culture included VP-SFM, OptiPRO, FreeStyle 293 (Thermo Fisher Scientific, USA), HyClone HyCell TransFx-H (GE Healthcare, USA), HEK GM (Xell AG, Germany) and ProVeroTM -1 serum free medium (SFM) (Lonza, Switzerland).
Despite slightly slower growth rates in MDXK than in IHM03 in shake flasks, the cells were able to grow to similar cell densities in batch shake flasks with less formation of cell aggregates. Compared to other mammalian suspension cell lines, like derivatives of HEK293 or CHO with cell doubling times of 24 h, there is still great potential to develop media that can support similar growth rates.

Cell growth in batch bioreactor

The suspension adapted Vero cells showed similar growth in IHM03 medium in 1L batch bioreactors as previously reported, reaching 1.78 × 106 cells/mL after 6 days (Shen et al., 2019). Cell viability was above 99 % during the whole run. The doubling time was around 51 hours for the entire batch process duration between cell seeding and peak in maximum cell density, and around 40 hours for the exponential cell growth phase between cell seeding up until 96 hours (Figure 1 A). The substrates glucose and glutamine were almost depleted at the end of the culture. Ammonia was produced throughout and stayed below concentrations of 4 mM. Lactate production reached a concentration of 24.5 mM after 96 hours, but declined thereafter which can be explained by an uptake of lactate by the cells as previously reported (Quesney et al., 2003).
Cell growths in MDXK medium was slightly slower and only reached 1.45 × 106 cells/mL after 7 days (Figure 1B). However, in contrast to the shake flask experiments, Vero cells in MDXK medium showed a higher degree of aggregate formation in the bioreactor compared to cells in IHM03. This could have led to an underestimation of the cell count, which is also indicated by a higher glucose consumption rate in MDXK. Cell viability again was above 99 % throughout the process. As opposed to the previous run, glucose and glutamine were depleted earlier and required feeding, adjusting glucose to 2 g/L and glutamine to 2 mM once daily starting on day 2 for glutamine and day 6 for glucose. Ammonia production was similar, never exceeding concentrations of 4 mM. Lactate production, however, was significantly higher and surpassed 50 mM at the end of the culture. Lactate was not consumed by the cells in MDXK medium indicating significant differences in the cell metabolism in the two media.

rVSV-ZEBOV production in shake flask

rVSV-ZEBOV production experiments were initially carried out at a smaller scale in shake flasks to test multiple conditions simultaneously. In particular, the effects of varying multiplicities of infection (MOI), which is the ratio of infectious particles to the number of cells at the time of infection (TOI), as well as infections at different cell densities and in different growth media were screened.
During the late stages of the cultivation when cell densities exceeded 1 × 106 cells/mL, cells started to adhere to the surface of the shake flask and cell aggregates were formed, which made it quite difficult to accurately determine the cell count. Therefore, initial experiments to investigate the infection kinetics of rVSV-ZEBOV in suspension adapted Vero cells, were carried out by seeding a single cell culture in fresh medium at a cell density of 1 × 106 cells/mL and cells were infected immediately thereafter.
Previous studies of rVSV-ZEBOV in adherent Vero cells and suspension cultures of HEK293 cells have shown that infection at a reduced temperature of 34 °C led to higher infectious titers compared to 37 °C (Kiesslich et al., 2020)(Gélinas et al., 2019). Based on these studies, the temperature was lowered to 34 °C after infection in all experiments of this work involving rVSV-ZEBOV.

Multiplicity of infection

Vero cells were infected with rVSV-ZEBOV at different MOIs and samples were taken every 12 hours to determine infection kinetics (Figure 2 ). For the selected range of MOI, peak production of rVSV-ZEBOV occurred between 24 and 36 hours post infection (hpi) and the titers were in the same range with 1.10 × 107 TCID50/mL, 1.05 × 107 TCID50/mL and 1.58 × 107 TCID50/mL at an MOI of 0.001, 0.01 and 0.1, respectively. In all cases, the infectivity declined after the maximum titer had been reached. Similar kinetics have been observed for adherent growing Vero cells, however, the titers were more than eight times higher in adherent cell experiments, for example 8.79 × 107 TCID50/mL at an MOI of 0.01 at 36 hpi (Kiesslich et al., 2020). Further, the cell density at the time of infection was more than three times higher for the suspension cultures than for the adherent cell cultures in the reported study, indicating even lower virus production per cell. It is to mention that adherent Vero cells were cultivated in commercially available media, optimized for cell growth and virus production.
rVSV infections of adherent Vero cells typically lead to a very distinct cytopathic effect, where cells become round shaped and eventually lift off from the surface. Of note, since suspension cells are already round shaped and not attached to a surface, the cytopathic effect induced by rVSV infections was less noticeable in the early stages of infection and only became more apparent when the cell diameter increased due to viral replication and when cells started to die from lysis caused by viral release.
Nevertheless, based on these experiments and in accordance with our previous work, it was decided to continue all subsequent rVSV experiments at an MOI of 0.01.

Cell density

rVSV-ZEBOV replication in Vero cells seeded at different cell densities was investigated to evaluate the effect of varying cell densities and to assess if the production yield was affected by the cell growth phase or metabolite concentrations at the time of infection. Figure 3shows rVSV-ZEBOV infection at 1 × 106 cells/mL, 2 × 106 cells/mL and 4 × 106cells/mL. The viral infection kinetics were slower compared to the previous experiments and infectious titers peaked at 48 hpi for all cases, indicating considerable variation between experiments. Additionally, maximum infectious titers were around three times higher for the run at 1 × 106 cells/mL, reaching 3.28 × 107 TCID50/mL. Throughout the time course of the experiment, titers were even higher at 2 × 106 cells/mL, but the maximum infectious titer was not significantly elevated at 48 hpi. For the run with a seeding cell density of 4 × 106 cells/mL, infectious titers of rVSV-ZEBOV were significantly higher, reaching 1.32 × 108 TCID50/mL, exceeding infectious titers obtained from adherent Vero studies in 6-well plates (Kiesslich et al., 2020). In comparison, for rVSV-GFP produced in suspension cell cultures of Vero cells at different cell densities in a similar experiment, 3.8 times higher virus titers were obtained at 2.5 × 106 cells/mL compared to 0.87 × 106 cells/mL (Shen et al., 2019). But a further increase in cell density from 2.5 × 106 cells/mL to 5 × 106 cells/mL did not result in higher infectious titers.
These results indicate that suspension cultures of Vero cells could be a viable alternative if high cell density processes of suspension cultures can be achieved at larger scale, but further research is necessary to investigate effects of high cell density in more detail. One advantage though is that suspension cultures are not limited by the surface area, as is the case for adherent cell cultures using microcarriers, roller bottles or fixed-bed bioreactors, which is the prevalent mode of virus production in Vero cells. Nevertheless, these results need to be carefully evaluated since it might be challenging to seed bioreactors in fresh medium at high cell density

rVSV-ZEBOV, rVSV-HIV, and rVSVInd-msp -SF-Gtc shake flask production in different media

Next, the kinetics of three variants, namely rVSV-ZEBOV, rVSV-HIV, and rVSVInd-msp -SF-Gtcwere compared and the effect of different media on rVSV production were studied. Based on temperature study results of rVSV-ZEBOV infections in Vero cells, rVSV-HIV infections were carried out at 34 °C in a recent study and this condition was adopted for this work as well (Mangion et al., 2020). The rVSVInd-msp -SF-Gtcconstruct, however, is temperature sensitive and therefore all infections were carried out at 31 °C.
The IHM03 and MDXK adapted cell lines were infected with rVSV to compare the virus production capacity of both media. Despite higher titers obtained at higher cell densities (Figure 3 ), these experiments were carried out at a seeding cell density of 1 × 106cells/mL to avoid cell aggregation. Figure 4 shows rVSV-ZEBOV (A), rVSV-HIV (B) and rVSVInd-msp -SF-Gtc (C) replication in the MDXK adapted cell line at an MOI of 0.01 in comparison to the corresponding experiment conducted in IHM03.
rVSV-ZEBOV replication in MDXK medium reached slightly higher titers than in IHM03 medium (Figure 4 A). Where the infectious titer reached a maximum at 24 hpi in IHM03, the titer in MDXK was with 2.63 × 107 TCID50/mL around 2.5 times higher. Otherwise, almost identical infection kinetics were observed.
rVSV-HIV replicated better in IHM03 with a higher maximum titer of 9.59 × 106 TCID50/mL reached in a shorter period of time (24 hpi) compared to 2.08 × 106 TCID50/mL in MDXK after 36 hpi (Figure 4 B). Titers of rVSV-HIV were lower in both media than rVSV-ZEBOV and the infectivity declined faster than for rVSV-ZEBOV. Besides, rVSV-HIV produced in adherent Vero cells reached up to 3.91 × 107 TCID50/mL at MOI of 0.01. However, the peak was reached significantly later at 96 hpi (Mangion et al., 2020).
In contrast to these two strains, rVSVInd-msp -SF-Gtc reached significantly higher titers (Figure 4 C). In MDXK, 5.19 × 108 TCID50/mL were reached at 36 hpi. In IHM03, almost two-fold higher titers with 1.17 × 109 TCID50/mL were reached. However, it took additional 24 hours to reach this titer, which aligns with the beginning of the replication phase being delayed. In addition to higher infectious titers compared to rVSV-ZEBOV and rVSV-HIV, the infectivity of viral particles did not decline significantly over the following sample time points.
Higher infectious titers of rVSVInd-msp -SF-Gtc in Vero cells could be linked to the use of different transmembrane domains. Whereas rVSV-ZEBOV and rVSV-HIV use Ebola GP or SIV transmembrane domains (Mangion et al., 2020), rVSVInd-msp -SF-Gtc is expressing the native VSV-G protein transmembrane domain gene. It has been shown that the stem region of the VSV-G glycoprotein was important for efficient virus assembly, and viruses with shortened sequences were replicated up 20-fold less (Robison and Whitt, 2000). More research investigating the use of different transmembrane domains with the same extracellular domain could reveal interesting aspects on virus replication rates and identify new targets to improve the rVSV platform. Nevertheless, it might be more appropriate to compare replication of rVSVInd-msp -SF-Gtc to rVSV-GFP production, which also uses the native VSV-G protein transmembrane domain, and where titers of up to 8.93 × 109 TCID50/mL have been obtained in shake flask experiments.
Besides, the lower process temperature of 31 °C is likely affecting the infectivity. For example, the infectivity of this strain was not declining over the following 36 h after the peak titer had been reached thus potentially stabilizing infectivity. Though for rVSV-ZEBOV, an optimal production temperature of 34 °C was determined in adherent Vero cells (Kiesslich et al., 2020), rVSVInd-msp -SF-Gtc is based on VSVInd(GML) which was adapted to replicate well at the lower temperature of 31 °C (Figure 4 C).

Bioreactor processes of suspension adapted Vero cells

Bioreactor production of rVSV-ZEBOV

Two bioreactors of Vero cells were infected at a cell density of 1.37 × 106 cells/mL and 1.02 × 106 cells/mL after cells grew for 4 days in IHM03 and MDXK medium, respectively (Figure 5 ). The cell growth phase corresponded well to the data shown in Figure 1 . Despite a lower cell density at the TOI, maximum infectious titers were with 3.87 × 107 TCID50/mL more than one log higher than in IHM03, were only 3.55 × 106TCID50/mL were obtained. In addition, replication was faster in MDXK, were peak production occurred at 24 hpi compared to 36 hpi, respectively. The MDXK bioreactor also exhibited an almost 15-times higher cell specific productivity with 37.9 TCID50/cell compared to 2.6 TCID50/cell. Moreover, the ratio of total viral particles to infectious particles was lower in MDXK (282 VG/TCID50) than in IHM03 (817 VG/TCID50), further indicating a better quality of the final product if harvested at the time of peak infectious titer.
Compared to the shake flask experiments, rVSV-ZEBOV replication in MDXK medium reached a slightly higher maximum infectious titer, indicating a successful scale-up. For IHM03, titers were three times lower than in shake flask and the peak was reached 12 hours later.
Further, in comparison to adherent Vero bioreactor productions of rVSV-ZEBOV, the production using suspension adapted Vero cells in MDXK appears superior. The infectious titer and the cell specific productivities were slightly higher compared to the production in a microcarrier bioreactor (1.42 × 107TCID50/mL, 7.6 TCID50/cell) and a fixed-bed bioreactor (2.59 × 107TCID50/mL, 11.2 TCID50/cell). The ratio of total viral particles to infectious particles was four times lower than in the microcarrier but nine times higher compared to the fixed-bed process (Kiesslich et al., 2020). Overall, the suspension Vero system is a viable alternative to the current Vero manufacturing system carried out in roller bottles for this Ebola virus disease vaccine (Monath et al., 2019).
When set side by side to a suspension bioreactor production of rVSV-ZEBOV in HEK293-SF, where a maximum of 1.19 × 108TCID50/mL was reached, production in Vero cells in MDXK was 3 times lower. However, it can be expected that future media development, bioprocess and cell line engineering of suspension Vero can lead to significantly higher titers comparable to HEK293-SF (Gélinas et al., 2019).

Bioreactor production of rVSV-HIV

Two bioreactors were prepared as before, and Vero cell growth phase in IHM03 and MDXK medium was consistent with the data from Figure 1 and Figure 5 . The two cultures were infected with rVSV-HIV after 4 days (Figure 6 ). In contrast to the previous experiment, virus production was favoured in IHM03 over MDXK. In IHM03, rVSV-HIV reached a maximum titer of 2.12 × 107TCID50/mL, which was 25-times higher than in MDXK. However, production kinetics of viral genomes were almost identical. This is supported by a lower ratio of 143 VG/TCID50(IHM03) compared to 7041 VG/TCID50 (MDXK). Additionally, the fact that the cell specific productivity in MDXK was 0.9 TCID50/cell implies that the rVSV-HIV/MDXK system failed to scale-up and is not an adequate production system.
Besides, the shake flask experiment (Figure 4 B) already indicated the superiority of IHM03 for rVSV-HIV replication. Moreover, bioreactor production of rVSV-HIV in IHM03 exceeded titers from the smaller scale, whereas bioreactor titers in MDXK subsided the small scale.

Bioreactor production of rVSVInd-msp -SF-Gtc

Finally, rVSVInd-msp -SF-Gtcproduction in bioreactors of Vero suspension cell cultures was studied (Figure 7 ). Again, the cell growth phase was comparable to the previous runs, and cells were infected after 4 days. Infectious titers of rVSVInd-msp -SF-Gtcpeaked in both media at 48 hpi. The beginning of the replication phase was delayed in IHM03, as already seen in the shake flask experiments. Nevertheless, titers were similar with 2.38 × 109TCID50/mL in IHM03 and 3.59 × 109TCID50/mL in MDXK. Thus, results from the shake flask experiment were exceeded by two-fold and seven-fold, respectively. On the one hand, the quality in terms of total viral particles to infectious particles was with 3.0 VG/TCID50 better in IHM03 than in MDXK, where this value was 6.0 VG/TCID50. On the other hand, the cell specific productivity was doubled in MDXK compared to IHM03, with 3670 TCID50/cell and 1803 TCID50/cell, respectively.
In comparison, rVSV-ZEBOV and rVSV-HIV productions in the bioreactor peaked earlier. However, the number of viral genomes continued to increase in those experiments even when the infectious titer declined. Therefore, the rate of viral degradation is higher than the viral production rate after the corresponding peak was reached in the case of these two strains, potentially due to the higher process temperature of 34 °C versus 31 °C and its impact on viral stability.
Overall, the scale-up of rVSVInd-msp -SF-Gtcproduction to the bioreactor was successful, exceeding small scale results. In addition, this strain appears to replicate to much higher titers, with a superior cell specific productivity and an improved ratio of VG/TCID50 as compared to rVSV-ZEBOV and rVSV-HIV.