This pilot study was the first known study to expand equine cord blood MSCs on microcarriers in stirred suspension bioreactors. The first step in the process development was to find an appropriate microcarrier to facilitate attachment as well as growth of the eCB-MSCs. Five microcarriers that are commonly used for the expansion of human MSCs were tested. Both Synthemax II and Enhanced attachment microcarriers had very low attachment of eCB-MSCs. These are both polystyrene microcarriers with proprietary coatings. Cytodex 1, an uncoated microcarrier with a dextran matrix, allowed for cell attachment but not long term expansion, which may have been due to the lack of coating which prevented expansion and spreading of the cells on the microcarrier surface. Cell attachment as well as expansion was facilitated on Cytodex 3, a gelatin coated microcarrier with a dextran matrix, as well Cultispher S, a gelatinous macroporous microcarrier. However, cells on Cytodex 3 achieved higher attached cell densities, likely due to poor nutrient and oxygen transfer into the pores of the Cultispher S microcarrier causing increased cell death. These results were not unexpected as Cytodex 3 is commonly used to expand various sources of MSCs, obtaining high cells densities over long term culture period [10, 11, 17,18,19, 28,29,30,31]. Based on these results, Cytodex 3 was chosen for use in the remaining process development.
Several different inoculation process parameters were investigated for the attachment and expansion of eCB-MSCs on microcarriers in bioreactors. Initially the cell attachment kinetics were investigated and attachment of cells to microcarriers was compared to attachment to static T-flasks. Compared to static attachment, cell attachment to microcarriers occurred much more rapidly, with nearly 50% attachment occurring within the first 2 h, compared to less than 10% in static attachment. The attachment may have been enhanced due to shear effects on the cells promoting cell attachment, as low levels of shear have been found to affect proliferation as well as cytokine production of MSCs . The static flasks were also not coated, while the microcarriers were coated in gelatin, which could have led to the enhanced attachment in bioreactor culture. Additionally, it was observed that the cells in static culture undergo a lag phase, while in the bioreactor a lag phase was absent.
Other inoculation phase conditions were also found to affect cell growth. One such condition was the confluency of the T-flask prior to inoculation into the bioreactors. It was found that with T-flasks at low confluency, a lag phase was not observed, however a lag phase was observed when a T-flask at high confluency was used for inoculation. The cells in the low confluency T-flask were in the exponential phase of growth, while the cells in the high confluency T-flask were in the stationary phase of growth, likely contributing to the lag phase observed using this T-flask confluency. This was consistent with the findings by Balint et al. (2015), who found that when cells were passaged from T-flask to T-flask at 10–50% confluency they had significantly lower population doubling times and a higher proliferation rate than when cells were passaged at 40–70% confluency . To our knowledge, no study has been performed to analyze the effect of T-flask confluency on subsequent growth in bioreactors.
The last condition investigated in the inoculation phase was the initial cell to microcarrier ratio. Three different ratios were investigated, 2 cells/MC, 4 cells/MC, and 8 cells/MC. The 2 cells/MC ratio had the highest fold-increase in cell number, and the 8 cells/MC ratio had the highest attached cell densities. The actual choice of cell to microcarrier ratio in a bioprocess is dependent on other process constraints. For example, if the cells are very scarce, then the 2 cells/MC density would be chosen as high cell densities are still achieved despite the low inoculation density. However, if the process is time sensitive, or the cost of the medium is a limiting factor, then an 8 cells/MC density would be chosen as the greatest cell densities are achieved, with the same quantity of medium, and is reached one day earlier than using the 2 cells/MC or 4 cells/MC densities.
The 30%FBS-0bFGF medium was compared to the 10%FBS-5bFGF medium and the cell growth kinetics of the eCB-MSCs were similar in both media, therefore the addition of bFGF was an appropriate substitute to lower the amount of FBS in the medium. This is consistent with several studies that have shown that bFGF in culture medium enhances expansion of human MSCs, as bFGF is a cytokine that enhances motility and proliferation of several cell types [33,34,35]. A study by Ibrahim et al., tested different types of basal media, with 10% FBS, with the addition of either 4 or 10 ng/mL bFGF and found greater expansion with 10 ng/mL bFGF, and found that the bFGF was required for growth .
Using the 10% FBS-bFGF medium, a medium replacement regime was developed by analyzing the metabolic activity of the cells in static and bioreactor culture. There were significant differences between the metabolism in the cells in static culture compared to bioreactor culture, with the bioreactor cells having very low metabolic activity. Studies analyzing the metabolism of human MSCs grown in stirred suspension bioreactors have found that the glucose uptake rate varied between 5 and 15 pmol/cell/d [10, 36], comparing to our results of 2.35 pmol/cell/d for the bioreactor culture and 7.89 pol/cell/d for the static culture. Studies have also found that the lactate consumption rate varied between 12 and 25 pmol/cell/d [10, 36], compared to our results of 3. 32 pmol/cell/d for the bioreactor culture and 22.5 pmol/cell/d for static culture. However, no studies could be found for the metabolic activity of equine MSCs, and it has been found that human MSCs have different metabolic activity than certain species of animal MSCs .
The difference in metabolic activity between static and bioreactor culture could be due to the mechanism by which the MSCs convert glucose to energy. There are two main mechanisms in which MSCs convert glucose to energy (ATP): glycolysis and oxidative phosphorylation. In oxidative phosphorylation, glucose is metabolised to generate ATP with the consumption of oxygen. This is a very efficient method of energy production, with 1 mol glucose generating ~ 36 mol ATP. In glycolysis, glucose is converted to ATP inefficiently, with 1 mol glucose generating ~ 2–4 mol ATP [37,38,39]. The yield of lactate to glucose was 2 .9g/g in static culture, and 1. 42 g/g in bioreactor culture. Glycolysis may have been occurring in the cells grown in static culture causing the increased glucose consumption, while oxidative phosphorylation may have been occurring in the cells grown in the bioreactor, allowing for a lower glucose consumption while still generating a large amount of energy [37,38,39,40]. An increased oxygen concentration due to the agitation occurring in the bioreactors could have caused the cells in the bioreactor to undergo oxidative phosphorylation rather than glycolysis. The diffusion of nutrients through the bioreactor due to the mixing could also alter the metabolic activity of the cells.
Based on the analysis of glucose and lactate in the medium, a medium replacement regime of a basal medium change of 50% on Day 4, and an addition of bFGF every two days was proposed. Common replacement regimes used in a bioreactor process include, 25% daily or every 2 days, 50% either daily, every 2 days or every 3 days, full medium change every 2 days or 3 days or a perfusion (continuous replacement) regime. However, typically no specific analysis is performed to quantify which specific nutrients are limiting, or if any toxic by-products have built up.
The proposed medium change was used to expand the eCB-MSCs in static and bioreactor culture. Differences were again observed between the cells expanded in static and bioreactor culture. The cells in the bioreactor culture were greatly influenced by the bFGF addition, while the cells in the static culture were greatly influenced by the 50% medium change. This could be related back to the glucose consumption rate, which was observed to be much higher in the static expanded cells, therefore required a higher glucose concentration in the media. This demonstrates differences between bioreactor and static expanded cells, and the need for a custom medium replacement regime for the different modes of expansion.
When cells are expanded using microcarrier-based processes, the agitation must be high enough to maintain cells in suspension. However, studies have also shown that higher agitation rates can achieve greater cell expansion, due to improved nutrient and oxygen transfer, as well as shear stresses can trigger cellular responses through mechanotransduction that can enhance proliferation of cells [41, 42]. Three different agitation rates, 40 rpm, 60 rpm and 80 rpm, were compared for cell proliferation in the 125 mL bioreactor. The average shear stress in 125 mL bioreactors have previously been calculated in our lab to be 0.004 Pa, 0.006 Pa, and 0.008 Pa for bioreactors run at 40 rpm, 60 rpm, and 80 rpm. These values are considerably lower than shear stresses that have been found to damage cells (1.5–3 Pa ), or to alter cell behavior (0.1–1 Pa [44, 45]). However, the maximum shear stresses, occurring at the tip of the impeller, have been found to nearly 40 times greater than the average shear stress, which is within range to alter cell behavior, and could have contributed to the lower final attached cell densities in the 80 rpm bioreactor.
The harvesting stage of a microcarrier process is very important to detach the cells from the microcarriers and filter to achieve a pure, highly viable cell suspension. Enzymatic removal is the most common method of removing cells from microcarriers, however, the type of enzyme to use is process and cell specific. Therefore, this study investigated five different types of enzymes for detachment efficiency and found similar detachment efficiencies using 0.25% Trypsin, Accutase, TrypLE, TrypZean, and lower efficiency with 0.05% Trypsin, which has a much lower activity level than the other four enzymes. Goh et al.  (2013) compared the kinetics of cell detachment using 0.25% Trypsin, TrypLE Express and Type I collagenase and showed that 0.25% Trypsin resulted in the highest cell detachment, as well as higher osteogenic potential compared to TrypLE Express and Type I collagenase. A similar study by Weber et al. (2007) investigated harvesting of human MSCs using 0.25% Trypsin, Accutase, collagenase or a Trypsin-Accutase mixture . Trypsin and Trypsin-Accutase mixtures achieved the highest cell yields and viabilities.
As 0.25% Trypsin was the standard enzyme used to detach eCB-MSCs from static culture, and was successful in removing the cells from the microcarriers, this enzyme was chosen for use in the bioreactor process. The ideal exposure time in the range of 3–15 min was investigated, and it was found that after 9 min the detachment plateaued, therefore this time was chosen for all other experiments. Throughout the harvesting experiments, generally low harvesting efficiencies were observed, despite images showing that the majority of cells had detached. Upon further investigation, it was found that many cells had been trapped in the sieve used for filtering. As the surface area of the sieve was small, compared to the number of microcarriers being filtered, a microcarrier cake built up on top of the sieve, preventing cells from passing through. A sieve with a larger filter area would be advantageous to achieve higher harvesting efficiencies.
Using the developed process, cells from two new donors were compared to the original cell donor for expansion over two passages in the 125 mL bioreactors. The cell densities of Donor 1201 increased slightly between passages, while those from Donor 1409 and Donor 1412 decreased between passages. It is possible that the growth of the eCB-MSCs using our process could have selected for a certain sub-population of cells in Donor 1201, therefore when the cells were passaged, the cells reached greater maximum attached cell densities during the second passage. All the cells were grown at a high passage, specifically donors 1409 and 1412 which were at passage 10 during the first passage in the bioreactor, and passage 11 in the second passage. Some stem cells have been found to reach senescence at high passages. A study by Bonab et al. (2016), found that population doubling times of human BM-MSCs increased substantially during the 10th passage of cells . This could have been attributed to the decrease in cell growth between the two passages. Variability in proliferation potential between donors has previously been observed in both human [48, 49] and equine MSCs [50, 51]. Heathman et al (2016), and Phinney et al. (1999), compared human BM-MSCs donors for proliferation potential in static and found up to a 12-fold difference between donors. Donor to donor variability has also been shown in equine MSCs, with a study by Carter-Arnold et al. (2012) showing high variability in proliferation between 6 different equine BM-MSC donors.
There was variability in not only the expansion of the eCB-MSCs between donors, but also in the harvesting. Donor 1409 cells, which the harvesting protocol was developed for, achieved the highest harvesting efficiencies, followed by Donor 1201 and 1412. It was shown in the kinetic growth data of the cells, that there were differences in the cells from different donors, therefore this could have resulted in differences in efficiency of the enzymatic harvesting procedure. The low harvesting of all donors can be attributed to the filter as discussed earlier.
The donors used in this study were from cells from two different breeds of horses: Quarter horses and Thoroughbreds, as well as both male and female. To decrease donor to donor variability, the process may need to be altered to account for different breeds and/or sexes. However, if an allogenic treatment is utilized, several prospective donors can be screened for proliferation potential, or for other desirable properties such as chondrogenic potential, and only certain donors can be chosen to be used for treatment.
Maximum attached cell densities of 75,000 cells/cm2 were achieved when expanding eCB-MSCs in stirred suspension bioreactors. No other published papers were found that expanded eCB-MSCs in stirred suspension bioreactors, while only one study was found for human cord blood MSCs expanded in stirred suspension, in which cell densities of 45,000 cells/cm2 were reached . Other studies expanding various sources of MSCs on Cytodex 3 achieved attached cell densities ranging from 40,000–70,000 cells/cm2, comparable with our results [28, 29].
The required number of cells to treat a patient (approximately 109 Ref ), could be achieved with a 2 .5L bioreactor. However, it is expected that if these cells were grown in computer-controlled bioreactors, controlling the dissolved oxygen and pH, even greater attached cell densities could be achieved, decreasing the required volume. Comparably, to achieve 109 MSCs in static culture, a 40-layer CellSTACK® would be required, which uses twice the medium volume as a 2 .5L bioreactor, greatly increasing the cost. As well, it would not be possible to control the dissolved oxygen and pH in the CellSTACK® system, thus oxygen and nutrient gradients could occur, affecting cell growth as well as producing a less homogenous product.
The surface marker expression and trilineage differentiation capacity of the eCB-MSCs did not differ between static and bioreactor culture, consistent with a previous report comparing these two expansion methods in human MSCs . The surface markers assessed have been extensively used to characterize equine MSCs, as they appear to be mostly constitutively expressed/not expressed among MSCs from various sources and at different passage numbers [53,54,55,56]. Reports on CD105 and MHC I expression are variable, however we did not observe a difference in expression between culture systems. While there was variability in chondrogenic pellet Toluidine Blue staining and Alizarin Red staining for osteogenesis, both donors evaluated showed capacity for trilineage differentiation at later passages (passage 11). This is likely as a result of the addition of bFGF to the culture media . More characterization is needed to ensure that the immunomodulatory potency and in vivo function of the eCB-MSCs remains unchanged between culture systems.