Scale‐up of an intensified bioprocess for the expansion of bovine adipose‐derived stem cells (bASCs) in stirred tank bioreactors

Cultivated meat is an emerging field, aiming to establish the production of animal tissue for human consumption in an in vitro environment, eliminating the need to raise and slaughter animals for their meat. To realise this, the expansion of primary cells in a bioreactor is needed to achieve the high cell numbers required. The aim of this study was to develop a scalable, microcarrier based, intensified bioprocess for the expansion of bovine adipose‐derived stem cells as precursors of fat and muscle tissue. The intensified bioprocess development was carried out initially in spinner flasks of different sizes and then translated to fully controlled litre scale benchtop bioreactors. Bioprocess intensification was achieved by utilising the previously demonstrated bead‐to‐bead transfer phenomenon and through the combined addition of microcarrier and medium to double the existing surface area and working volume in the bioreactor. Choosing the optimal time point for the additions was critical in enhancing the cell expansion. A significant fold increase of 114.19 ± 1.07 was obtained at the litre scale in the intensified bioprocess compared to the baseline (**p < .005). The quality of the cells was evaluated pre‐ and post‐expansion and the cells were found to maintain their phenotype and differentiation capacity.

cell numbers are only attainable in bioreactor cultures.
To date, a significant number of published studies have demonstrated the ability to expand human mesenchymal stem cells (hMSCs) at different scales (deSoure et al., 2016;Hewitt et al., 2011;Lawson et al., 2017;Rafiq et al., 2013Rafiq et al., , 2017Rafiq et al., , 2018. More recently, bovine adipose-derived stem cells (bASCs) as precursors for fat and muscle tissue were also expanded successfully in spinner flasks (Hanga, Ali, et al., 2020).
However, that study was carried out in a semi-controlled environment as spinner flasks rely on the temperature and gas control within an incubator. The next step for scalable production is translation to fully controlled bioreactors. This is, to our knowledge, the first report of litre scale expansion of bASCs for cultivated meat production.
Microcarrier culture in bioreactors has a significant advantage over monolayer culture due to their high surface-areato-volume ratio. In addition, this surface area can be expanded in bioprocesses through the provision of additional microcarriers, thus maximising cell expansion, while maintaining cell growth in the exponential phase for longer. This approach has been reported previously for expansion of human MSCs on microcarriers (Hervy et al., 2014;Rafiq et al., 2018), as well as for the expansion of skeletal muscle satellite cells (Bodiou et al., 2020;Verbruggen et al., 2018). It is believed that a key mechanism for the success of this approach is the phenomenon known as a beadto-bead transfer. This refers to the migration of adherent cells from one microcarrier to another in suspension culture (deSoure et al., 2016;Rafiq et al., 2018;Takahashi et al., 2017). Bead-tobead transfer and surface area addition are approaches of particular interest from a manufacturing perspective as they facilitate process intensification, thereby increasing cell yield and reducing process time.
In this study, we investigated the surface area addition approach for process intensification, while translating the bioprocess for expansion of bASCs from spinner flasks to litre scale bench-top bioreactors, thus benefiting from full environmental monitoring and control. The continuous monitoring of process parameters possible in such bioreactors and the offline measurements of important representative metabolites (Moutsatsou et al., 2019) and cell counts performed throughout the culture enabled a better understanding and faster optimisation of the expansion bioprocess of these cells.
Such work has not been reported before for bASCs culture and represents a major step forward for this novel use of stem cell culture.

| Microcarrier culture in spinner flasks
Two different-sized spinner flasks with the same type of impeller were used in this study: 100 ml capacity (Bellco) (Figure 1a) and 500 ml capacity (Wheaton Celstir) (Figure 1b). Before use, all spinner flasks were coated with Sigmacote (Sigma-Aldrich) to prevent cell attachment to the glass . SoloHill Plastic microcarriers (Pall) were chosen for this study based on their excellent performance, animal-free characteristics, and commercial availability (Hanga, Ali, et al., 2020;Rafiq et al., 2016). The microcarriers were weighed to achieve the surface area needed and then suspended in D-PBS without calcium and magnesium (BE17-516F; Lonza) for autoclaving. The surface area concentration used was 5 cm 2 /ml of medium. bASCs were seeded at the optimal seeding density of 1500 cells/cm 2 established in our previous study (Hanga, Ali, et al., 2020). The feeding regime previously established to be optimal at 80% medium exchange was carried out on Day 3 and then every other day until the end of culture (Hanga, Ali, et al., 2020).
Briefly, the agitation was stopped to allow microcarriers to settle (~5 min), then the volume was removed and replaced with a fresh medium. The spinner flasks were then transferred back to the incubator and the agitation was restarted. All spinner flask cultures were agitated at the minimum speed required to just suspend the microcarriers (N js ). N JS is usually assessed visually as the speed at which no particles remain static on the base of the vessel at any point for more than 1 to 2 s (Zwietering, 1958). At this speed, all cells attached to microcarriers are able to access nutrients from the medium and pass metabolites into it. Higher speeds do not enhance those processes, but greatly increase the potential for damage to cells from microcarrier-microcarrier or microcarrier-impeller impacts and enhanced turbulent stresses . The first time the use of N JS was proposed as a preferred operating criterion for the culture of MSC was by Hewitt et al. (2011) for spinner flasks and it was later shown to be appropriate for a range of bioreactors by . For both types of spinner flasks used in this study, this speed was visually assessed to be 30 rpm, which was kept constant throughout the culture. The spinner cultures were carried out for 11 days when the full harvest was done by using a previously published protocol . Three different combinations of parameters were investigated for the ability to intensify the bioprocess and these are shown in Table 1   performed on the pH probe which was then sterilised by autoclaving.

| Microcarrier culture in bioreactors
The DO probe was first autoclaved and then calibrated postassembly in the bioreactor using saturation with air supplemented with 5% CO 2 as 100% DO. Aeration was achieved through the headspace using air supplemented with CO 2 which has been previously established to be sufficient for the oxygen demand of human MSCs up to the 5 L scale Rafiq et al., 2013).
SoloHill Plastic microcarriers (Pall) were also used for the bioreactor cultures and were weighed to provide a surface area of 5000 cm 2 /L. The microcarriers were sterilised by autoclaving before their addition to the bioreactor vessel. Cells were inoculated on the microcarriers at the established optimal seeding density of 1500 cells/cm 2 (Hanga, Ali, et al., 2020). The "baseline" bioreactor runs were initially operated at a working volume of 1 L containing a surface area of 5000 cm 2 throughout the culture. The "MC + volume addition" runs started with 5000 cm 2 in 1 L up to Day 5 after which both surface area and volume were doubled to achieve 10,000 cm 2 in 2 L (Table 1). For all bioreactor experiments, the agitation speed was kept constant throughout culture at the N js, which for this vesselimpeller combination was assessed by visual observation and it was found to be 60 rpm. The bioreactor cultures were also maintained for 11 days when the harvest was performed. Medium exchanges in the bioreactor were carried out on the bench using the peristaltic pump (Watson-Marlow 120S), pre-sterilised glass bottles, and a sterile welder (Terumo TSCD-II) for establishing sterile connections ( Figure 1c). The medium exchange % was limited to 50% when the bioreactor was operated at the 1 L volume because of the positioning of the sampling port on the Mobius vessel, which only allows the removal of a maximum volume of 500 ml. For the "MC + volume addition," the 50% medium exchange was maintained for consistency. The first medium exchange was performed at Day 3, then every other day until the end of the culture. Briefly, the agitation was stopped to allow microcarriers to settle (~5 min), then the volume was removed and replaced with fresh medium. The agitation was then restarted. During all processing steps, aeration and temperature were controlled at all times.

| Cell harvest
Cell harvesting from microcarriers in spinner flasks was carried out using the protocol previously developed by . For cell harvest from microcarriers in the bioreactor, the same protocol was adapted to the litre scale with the difference that all steps were carried out directly in the bioreactor on the bench using the peristaltic pump and sterile welder rather than transferring to the biological safety cabinet. Briefly, the agitation was stopped and the microcarriers were allowed to settle for 5 min before removing 50% of the spent medium and replacing it with DPBS for washing under agitation at the established Njs. These steps were repeated three times, followed by removal of 50% of the volume and replacing with 0.25% trypsin-EDTA (25200072; Gibco, Thermo Fisher Scientific). The incubation with the enzyme was carried out for up to 20 min at 37°C, while stirring at 150 rpm. For "Baseline" and "MC addition only" conditions, the exposure time to the proteolytic enzyme was 15 min, while for the "MC + volume addition" conditions where large cell densities were obtained, 15 min was not deemed satisfactory and thus, it was extended to 20 min. A sample was then taken for microscopic evaluation of complete cell dissociation from microcarriers. Once this was confirmed, the enzyme was then inactivated with growth medium to double the volume.
For spinner flasks, the total volume of 100 ml was then taken post-dissociation, while for the bioreactor culture, only a sample of 200 ml was taken post-dissociation for the filtration step to remove the microcarriers. This was achieved using Steriflip filtering devices (Millipore; 90 µm pores). The cell suspension was then centrifuged at 250g for 5 min to obtain a pellet.

| Process analytics
wSamples were taken at the same time points when feedings were performed and were used for cell imaging, cell counting and glucose, and lactate measurements before and after medium exchanges. Live/ dead staining kit (L3224; Thermo Fisher Scientific) comprising calcein-AM for live cells and ethidium homodimer for dead cells was used following the manufacturer's instructions to assess cell viability on microcarriers using a fluorescent microscope (Evos FL, Thermo Fisher Scientific). Spent medium samples were collected before and after medium exchanges and were analysed for glucose and lactate concentrations on the AccuTrend Plus meter (Roche). Fresh growth medium was used as the baseline control. Cell counts were performed directly onto microcarriers using the reagent A100 and reagent B protocol on the Nucleocounter NC-3000 (Chemometec). Briefly, the cell-microcarrier suspension was diluted to a 1:3 ratio with reagent A100 (lysing agent) and reagent B (stabilising agent). The resulting suspension was then loaded onto a Via-1 Nucleocassette pre-loaded with acridine orange and 4′,6-diamidino-2-phenylindole (DAPI).
Based on cell counts, the following parameters were calcu- are cell numbers at the end and the start of the culture.

| Cell product characterisation
All chemicals were purchased from Sigma-Aldrich unless stated otherwise. Cells from both spinner flasks and bioreactor cultures were harvested at the end of culture as described in Section 2.4 and used for characterisation. 2.6.1 | Tri-lineage differentiation potential The StemPro differentiation kits (Thermo Fisher Scientific) were used to drive differentiation of bASCs to adipogenic (A1007001), osteogenic (A1007201), and chondrogenic (A1007101) lineages.
The chondrogenic differentiation medium was then carefully added dropwise. All cultures were kept in their respective media

| Statistical analysis
All spinner flask experiments were performed in four biological repeats, while bioreactor experiments in triplicates. Cell counts were obtained from two independent samples from each repeat.
Data were expressed as mean ± SD. Statistical analysis was carried out using Graph Pad Prism 9. For comparison of multiple data sets, significance was determined by one-way analysis of variance. Statistical significance was considered when p < .05.

| RESULTS AND DISCUSSION
The aim of this study was dual: (1) to investigate the potential for process intensification through surface area addition by utilising the bead-to-bead transfer phenomenon and (2) to translate the improved bioprocess to litre scale in controlled bioreactors. To the best of our knowledge, no such study has been done using bovine stem cells. The assumption was that the bead-to-bead transfer phenomenon would be applicable to the bovine-derived MSCs and would lead to higher cell yields.

| Bioprocess intensification for bASCs expansion in spinner flasks
The first objective was to investigate if, through fresh microcarrier addition and utilising the bead-to-bead transfer phenomenon, the freshly added microcarriers would be populated with attached cells, thereby maintaining cell growth and enhancing total cell numbers over the same culture period. Enhanced total cell numbers in the same timeline is an indication of process intensification, which is highly advantageous leading to an increased cell number while reducing process time. Bead-to-bead cell transfer is a phenomenon that can be influenced by multiple factors, including medium composition, attachment substrate (i.e., type of microcarrier), feeding regime, time The "baseline" bioprocess for the bASC expansion on microcarriers in the Bellco spinner flasks was developed in our previously published work (Hanga, Ali, et al., 2020) and it was used here as a starting point for the intensification of the expansion bioprocess by keeping cell seeding density, initial surface area, and the feeding regime the same. The first critical parameter for process intensification is the selection of the optimal time point for microcarrier addition. In this study, two time points were investigated: In the "MC only addition" runs, different levels of cell growth were found when the microcarriers were added at either Day 5 or 7. At Day 7, a fold increase of 30.78 ± 4.60 was obtained which was lower than the "baseline" bioprocess that yielded a fold of 37.10 ± 6.73. On the other hand, when microcarriers were added at the earlier time point in culture (Day 5), it resulted in a slightly higher fold increase of 43.13 ± 9.17 (Figure 3a). It was postulated that when HANGA ET AL.  (Ferrari et al., 2012;Wu et al., 2014). The existence of aggregation at Day 7 was assumed to lower the efficiency of the beadto-bead transfer, thus explaining the diminished growth registered.
However, although the early addition of extra microcarriers gave a better measurable performance, the differences compared to the "baseline" were not statistically significant (ns; p > .05). In these runs, there is a possibility that, as known in early work on hMSCs (Hewitt et al., 2011), above a certain concentration of microcarriers per ml of medium, the culture performance deteriorates. More recent work has emphasised quantitatively how the microcarrier collisions with rotating impellers and particularly between themselves increases with concentration . As the cells are situated on the surface of the microcarriers, these collisions can cause damage and a reduction in culture performance and thus possibly resulting in cell death. This mechanism might also explain why the difference in cell growth between the "baseline" and the two different addition time points, were not statistically significant.
Another possible cause of this finding could be that nutrients provided in the constant volume were either depleted faster or were not sufficient for the amount of cells now proliferating on the doubled surface area. Alternatively, metabolite production (e.g., lactate) could also have an inhibitory effect on cell growth (Qie et al., 2012;Schop et al., 2009). However, neither of these latter reasons are relevant here as glucose depletion or lactate accumulation didn't occur in either the "baseline" or the "MC only addition" runs ( Figure 3d).
Other nutrients such as glutamine or metabolites such as ammonia could have been depleted or respectively produced in high concentrations, thus influencing cell growth negatively (Schop et al., 2009). However, these were not measured here. The next step was to combine microcarrier addition at the determined better time point of Day 5 with volume doubling at the same time ("MC + volume addition"). This new approach yielded approximately 1.8 × 10 8 cells at Day 11 (Figure 2c) equivalent of a fold increase of 116.65 ± 9.73. This increase was 3.14 times higher than the "baseline" yield and 2.7 times higher than the "MC only addition" yield (****p < .0001) (Figure 3a). The "MC + volume addition" resulted in a significantly higher specific growth rate of 0.0180 ± 0.00 h −1 (****p < .0001) ( Figure 3b) Figure 4. The initial cell seeding density, feeding regimes, and concentration of microcarriers per volume of medium (5 cm 2 /ml) were kept the same. It was not possible to translate the 80% medium exchange from spinner flasks to the Mobius bioreactor when operated at 1 L because of the vessel design and positioning of the medium removal port, which creates a dead volume of approximately 500 ml. Nevertheless, earlier work Rafiq et al., 2013) suggested that similar results could be obtained at 50% medium exchange and with increasing scale, a smaller exchange volume would represent a significant cost saving.
bASCs cultured on Plastic microcarriers in the Mobius bioreactor using the "baseline" operating parameters grew successfully, achieving an average yield of 3.02 × 10 8 cells in 11 days of culture, which was the equivalent of a fold increase of 40.31 ± 3.53 ( Figure 5a). Moreover, the cells remained viable on the microcarriers as shown by the live/dead staining (Figure 5b). Even with the increased levels of aggregation which occurred in the last days of culture, cell viability was maintained high with only a small number of dead cells (red) observed ( Figure 5b). As anticipated, following the smaller scale studies in the spinner flasks, the "MC + volume addition" bioprocess also yielded a significantly higher number of cells (1.01×10 9 cells) compared to the "baseline" (****p < .0001) ( Figure 6a). The equivalent fold increase of 134.75 ± 36.94 was 3.34 times higher than the "baseline" bioprocess (*p < .05) (Figure 6b). The specific growth rate (h −1 ) was calculated as 0.0184 ± 0.00004 h −1 in the "MC + volume addition" bioprocess compared to the "baseline" which was only 0.0140 ± 0.0003 h −1 (**p < .005) (Figure 6c); and the doubling time was significantly lower in the same bioprocess at 37.56 ± 1.93 h compared to 49.55 ± 1.13 h (**p < .005) in the "baseline" (Figure 6d).
The metabolic analysis showed many differences between the "baseline" and the "MC + volume addition" bioprocesses, particularly towards the end of the culture (Figure 7). These differences were observed in both glucose consumption and lactate production. In the baseline, glucose depletion was not reached at any time point in culture with a minimum of 1.36 mmol/L reached (Figure 7a1), while lactate concentration reached a maximum of 3.73 mmol/L as an average (Figure 7a2). On the other hand, the "MC + volume addition" bioprocess resulted in almost two times higher lactate concentration at Day 11 with a steep increase from Day 5 onwards despite the doubling of volume (Figure 7b2). The glucose concentration also showed a steep decline from Day 5 onwards, reaching low levels (<1 mmol/L) at the final day of culture (Figure 7b1).
From Day 9 onwards, increasing the glucose concentration in the culture medium proved difficult with the 50% exchange and feeding regime at the frequency adopted, which could explain the small drop in total cell numbers at Day 11. As the lactate concentration reached levels of over 7 mmol/L, to avoid inhibitory concentrations, medium refreshment might be needed. The metabolite inhibitory levels are cell line and type dependent (Schop et al., 2009) and to the best of our knowledge, no published study has investigated those levels for bovine ASCs. Overall, the glucose and lactate trends correlated well with the cell growth obtained in both bioprocesses. pH and dissolved oxygen (DO, %) were continuously monitored throughout the bioreactor cultures. In the "baseline" bioprocess, the pH was recorded and maintained within the | 3183 range of 7.3 to 7.4. However, for the "MC + volume addition" bioprocess, the pH range recorded was wider, from 6.88 to 7.3 ( Figure 8a) and it was outside the physiological pH range, which also may have impacted the cell number towards the end of the culture. The DO levels showed a steep decrease over time in both the baseline and the intensified bioprocesses. In the "baseline" bioprocess, the lowest DO level was 34.2% ( Figure 8b). Such a level would not be expected to cause a problem with the cells; indeed there is some evidence to suggest that even lower values are beneficial (Heathman et al., 2019). However, that study was done using human MSCs and not bovine. The "MC + volume addition" bioprocess reached low levels of DO (4.63%) meaning there was a higher oxygen demand which correlates with higher cell numbers. The "MC + volume addition" conditions in both spinner flasks and the Mobius bioreactor resulted in higher cell concentrations of 4.37 × 10 5 and 5.05 × 10 5 cells/ml, respectively, compared to maximum 3 × 10 5 cells/ml for baseline and "MC addition only" conditions (Table 1, column 10).
Moreover, the volumes of media required to produce 1 × 10 6 cells were the lowest for the "MC + volume addition" conditions with only 4.8 ml needed in spinner flasks and 4.9 ml in the Mobius bioreactor compared to 15.1 ml (spinners) and 9.9 ml (Mobius) for baseline (Table 1, column 13). This suggests that the intensified bioprocesses would be more cost-efficient particularly given that the media is the main contributor to the overall manufacturing cost. Overall, the volume of media required to produce 1 × 10 6 cells was lower in the bioreactor at the litre F I G U R E 8 Process parameters monitored during the bASCs expansion on microcarriers in the Mobius bioreactors. (a) pH and (b) DO (%). Data expressed as mean ± SD, n = 3 F I G U R E 9 Cell quality assessment of bASCs pre-and post-expansion in the Mobius bioreactor. Immunocytochemistry staining for CD73 and CD90 markers expression (green), phalloidin (red) for cytoskeleton, and DAPI (blue) for nuclei for (a) pre-bioprocessing. Scale bar = 100 µm. Additionally, the final cell densities (cells/cm 2 ) followed a similar pattern with the highest obtained at the litre scale and in the "MC + volume addition" conditions (Table 1, column 11) suggesting that in the intensified bioprocesses, better utilisation of surface area was achieved.
3.3 | Potential ways to further improve the intensified culture Several issues were identified in these experiments when scaling up the "MC + volume addition" bioprocess in the Mobius bioreactor even though this process still gave the best performance. From Day 9 onwards, the frequency of feeding with only a 50% medium exchange could not prevent the decrease in glucose concentration to almost depletion.
However, this issue could be easily resolved by either increasing the % of medium exchange from 50% to~75% which with a 2 L operating volume would be possible in the Mobius bioreactor (and probably in other bioreactor designs). Alternatively, supplementation with glucose and/or other depleted nutrients could also be used, with the latter being more cost-effective. The pH was also found to drop below the physiological range in the intensified bioprocess particularly towards the end of the culture. However, this difficulty is again easy to address by controlling the pH through the addition of a base or acid, rather than just relying on the buffer system present in the medium. The very low DO levels reached towards the end of the intensified bioprocess could also be an issue. Aeration in these experiments was done via the headspace with no further supplementation with oxygen. This approach was found sufficient until Day 9 of culture, but not beyond when in the final stages of the culture, oxygen supplementation is necessary. However, it is also possible that the lower DO was one of the reasons for the improvement as reported in the growth of hMSCs and is certainly worthy of further investigation.

| Cell quality assessment pre-and postbioprocessing
In the absence of tests proscribed for bovine ASCs and as discussed and justified in our earlier paper, cell quality was assessed using the guidelines proposed by Dominici et al. (2006). Pre-and post-bioprocessing, the bASCs retained their ability to differentiate towards adipogenic, osteogenic, and chondrogenic lineages. Moreover, in all conditions tested, the cells retained their expression of CD90 and CD73 markers (Figure 9).

| CONCLUSIONS
The first step in the production of cultivated meat is the expansion of the sourced cells. This study focused on the translation to the litre scale and intensification of a bioprocess for the expansion of bASCs as a source of fat and muscle. This aim was achieved through a combination of surface area and volume addition during the culture. This approach took advantage of the bead-to-bead transfer phenomenon in which cells already attached to microcarriers have the ability to populate fresh microcarriers when added during culture. However, the time point for the addition is critical as the cells lose this ability when the level of aggregation is too high. Here, we found Day 5 to be the better time point for surface area addition compared to Day 7. Microcarrier addition was much more efficient when combined with volume addition rather than when the volume was kept constant, probably because with the "surface only" addition, the increased concentration of microcarriers per ml resulted in an increased probability of microcarrier collision during agitation with cell damage as a consequence. The "surface + volume addition" approach lead to significantly higher cell numbers in the same timeline in a spinner flask and to an even better performance in the well-instrumented Mobius bioreactor. Overall, intensification using the "surface + volume addition" in the bioreactor lead to the highest number of cells/cm 2 and the highest cells/ml and reduced the volume of media required for the production of 1 × 10 6 cells from about 15 ml in the baseline in the spinner flask to just under 5 ml. This bioprocessing approach could be implemented with other adherent cell types and other species for applications not only in cultivated meat production but also cell manufacturing for therapeutics.

ACKNOWLEDGMENTS
This study was made possible through funding received from The Good Food Institute, US through the 2018 GFI Competitive Grant Program.