The productivity limit of manufacturing blood cell therapy in scalable stirred bioreactors

Abstract

Manufacture of red blood cells (RBCs) from progenitors has been proposed as a method to reduce reliance on donors. Such a process would need to be extremely efficient for economic viability given a relatively low value product and high (2 × 10¹² ) cell dose. Therefore, the aim of these studies was to define the productivity of an industry standard stirred-tank bioreactor and determine engineering limitations of commercial red blood cells production. Cord blood derived CD34+ cells were cultured under erythroid differentiation conditions in a stirred micro-bioreactor (Ambr™). Enucleated cells of 80% purity could be created under optimal physical conditions: pH 7.5, 50% oxygen, without gas-sparging (which damaged cells) and with mechanical agitation (which directly increased enucleation). O₂ consumption was low (~5 × 10^-8 μg/cell.h) theoretically enabling erythroblast densities in excess of 5 × 10⁸ /ml in commercial bioreactors and sub-10 l/unit production volumes. The bioreactor process achieved a 24% and 42% reduction in media volume and culture time, respectively, relative to unoptimized flask processing. However, media exchange limited productivity to 1 unit of erythroblasts per 500 l of media. Systematic replacement of media constituents, as well as screening for inhibitory levels of ammonia, lactate and key cytokines did not identify a reason for this limitation. We conclude that the properties of erythroblasts are such that the conventional constraints on cell manufacturing efficiency, such as mass transfer and metabolic demand, should not prevent high intensity production; furthermore, this could be achieved in industry standard equipment. However, identification and removal of an inhibitory mediator is required to enable these economies to be realized. Copyright © 2016 The Authors Journal of Tissue Engineering and Regenerative Medicine Published by John Wiley & Sons Ltd.

Keywords: bioreactor; blood; cost; culture; erythrocyte; manufacture; productivity; red cell.

Figure 1

Erythroid cell proliferation and differentiation is affected by bioreactor operational factors that determine system mass transfer. Mechanical agitation, gas sparging and cell protective PF68 were tested for effect on growth and maturation. (A) Growth curves show gas sparging (S) with stirring (300 or 450RPM) greatly reduced cell proliferation; stirring exacerbated the negative effect of sparging but was not detrimental alone (NS). PF68 supplementation to sparged bioreactors (S, P) protected proliferation from mechanical damage and was equivalent to nonsparged bioreactors (NS). (B) Example flow cytometry plot of CD235a vs. the nuclear stain DRAQ5 shows clear identification of the enucleated population (box). (C) In a nonsparged system, higher mechanical agitation supported a higher enucleation rate after 18 days. (D) Mechanical agitation was shown to have a direct effect on enucleation by transfer of cells from static to bioreactor culture after 19 days; the parallel curves indicate this accelerated enucleation is not associated with increased enucleated cell fragility.

Figure 2

Lower dissolved O₂ increases the percentage of enucleated cells. (A) Lower dissolved O₂ increases the percent of the cell population enucleated (p = 0.004). 25% and 50% O₂ form a statistically distinct group from higher O₂ levels (p ≥ 0.05). pH is not a statistically significant factor (pairwise comparison indicates the difference between pH 7.3 and 7.5 close to significance, p = 0.14). (B) Cell morphology clearly shows higher enucleation levels at lower O₂ (Day 19 cells cytocentrifuged, stained with Leishmans dye, observed with a Nikon Eclipse Ti microscope with a 40× objective. (C) The level of enucleation is higher throughout the culture process at low O₂, not just at final harvest. (D) An interaction chart for pH and O₂ suggests a rise in percent enucleation with increased pH may be more significant when O₂ is at an intermediate level.

Figure 3

Erythroid cells cultured in the established bioreactor system had reproducibly improved homogeneity but lower total proliferation relative to cells generated in the control static culture system. (A) Photographs of cells from static, bioreactor and a primary control show mature enucleated cells in both systems (cytocentrifuged onto glass slides, stained with Leishmans dye, visualised using a Nikon Eclipse Ti microscope equipped with a 40× objective). (B) The diameter of cells in static and bioreactor cultures show a reduction with time to the size of the primary control red blood cells. This was slightly accelerated in the bioreactor. (C) The percentage of enucleated erythroid cells (CD235a+/DRAQ5–) in bioreactor cultures at the point of peak proliferation was greater than in the control static culture. (D) Haemoglobin status of cells obtained from static and bioreactor cultures after 21 days was similar (HPLC). The percentage indicates the proportion of β haemoglobin chain expression. (E) Cumulative population doublings of cells in static and bioreactor cultures and associated table shows reduced total proliferation in the stirred bioreactor (n = 5 independent bioreactor runs from separate primary cell isolations, data is mean ± standard error of the mean).

Figure 4

O₂ is not a factor limiting cell expansion in the Ambr bioreactor system. The specific OUR of erythroblasts at serial time points was monitored via an O₂ sensitive fluorescent probe. OUR is low relative to conventional cell lines and declines as cells mature

Figure 5

Volumetric productivity of the system is dependent on media exchange strategy (A) Cells were cultured starting at 3 × 10⁶/ml, 5 × 10⁶/ml, and 5 × 10⁶/ml including a 30% volume exchange after 5 h. Cells initially proliferated at a constant and equivalent rate under all conditions after which growth became inhibited. (B) The initial deviation of the cell numbers from the extrapolated exponential growth is approximately linear (R² > 95%), and can be used to approximate the time point at which growth became inhibited.

Figure 6

Depletion of medium factors or production of common metabolites and cytokines are not responsible for volumetric productivity limits. (A) As previously, the deviation of cell growth from the initial exponential rate can be plotted as exponential model residuals vs. time. Supplementation after 10 h with amino acids, vitamins, or amino acids, vitamins and phosphate do not change the point at which growth becomes inhibited. (B) A wider range of supplementation strategies were tested including combinations of cytokines, serum, glucose and glutamine. The percent reduction in cells at 24 h compared to that predicted by the exponential model for each strategy is shown indicating no support of additional cell growth relative to control for any supplementation strategy. (C) Ammonia and lactate both inhibit cell growth (p ≤ 0.05). An increase in lactate concentration reduces the inhibitory effect of ammonia at high levels of the latter. (D) Lactate accumulates linearly with increased cell.time. However, at the point of growth inhibition (red dashed line) the level is not inhibitory with reference to (C). Further, glucose and lactate specific rates do not show any notable change as growth becomes inhibited. (E) A screen of cytokines present in media after cell growth inhibition indicated TGF‐β1 as a primary candidate for feedback growth inhibition. (F) TGF‐β1 is shown to be slightly inhibitory to erythroblast cell growth with a maximum of 9% reduction in specific growth rate over 40 h of culture and 10 ng/ml TGF‐β1. (G) TGF‐β1 is produced at the same specific rate in static and bioreactor cultures.