Page 21 - 2021_09-Haematologica-web

Manufactured red blood cells
Holotransferrin
The majority of current protocols use between 0.33 and 0.7 mg/mL holotransferrin isolated from human plasma,5-7 which represents the highest costing individual culture reagent. Holotransferrin is the natural carrier used to deliver iron to the developing erythroid cell by binding to CD71 and being internalized. It is then recycled by the cell and released back into the medium in its apo-form. A cheaper, plant-derived recombinant holotransferrin, optiferrin, is available but this is still expensive. Theoretically, the trans- ferrin concentration could be reduced in culture media as long as iron is also supplemented to bind apo-transferrin, without causing cell toxicity or increasing the likelihood of bacterial growth. For example, Timmins and colleagues used 0.12 mg/mL holotransferrin in combination with 900 ng/mL ferrous sulfate and 90 ng/mL ferric nitrate.27 Olivier et al. used holotransferrin at a concentration of 0.05 mg/mL and 3 mM FeIII-EDTA in small scale immortalized pluripo- tent stem cell cultures, reportedly without affecting the yield of reticulocytes, but the potential impact on mean cell hemoglobin concentrations and viability during storage of the mRBC produced was not measured.61 Other iron sup- plements that could be tested include reagents that would deliver iron to erythroid cells in a CD71-independent man- ner. These include the small lipophilic molecule hinokitiol that can carry iron across the cell membrane into erythroid cells62,63 or alternatively, ferric carboxymaltose and iron sucrose, both already prescribed to patients suffering from iron deficiency.64
Bioreactors and growing erythroid cells at larger scale
The majority of erythroid cultures described in the liter- ature are small and rely on the use of static tissue plastic flasks. For the reported larger scale cultures, a variety of culture vessels have been utilized. The original 2.5 mL packed mRBC produced under GMP conditions and tested in a single volunteer were cultured in static plastic flasks.5 Spinner flasks (of 1.5 L and 3 L volumes) have since been used successfully from day 7 onwards for constantly batch-fed cultures reaching a volume of ~28 L to produce 10 mL of packed filtered reticulocytes.6 Zhang et al. used rotating wall vessels to grow 2x108 cells from cord blood CD34+.13 Most recently, Heshusius and colleagues used a 1 L gas-permeable, rapid expansion bioreactor (G-Rex; Wilsonwolf) which, as well as facilitating partial media replenishment, allowed 90% of the expansion medium to be removed and replaced by differentiation medium. Although the expansion was 10-fold lower in the G-Rex compared to static dishes, the enucleation rate was similar and by extrapolation the authors predicted that ~4.5 mL mRBC could be produced using this bioreactor.7 The cell numbers from these studies are encouraging but are still a long way from the prediction by Timmins et al.,27 who suggested it may be possible to produce 500 units from a single cord blood donation.
There are still many types of bioreactors to choose from and explore further for erythroid culture including: (i) con- tinuous stirred tank bioreactors or spinner flasks which contain internal impellers; (ii) fluidized bed bioreactors in which cells are kept in suspension by the culture medium moving upwards; (iii) rocking heated platforms (wave-like
bioreactors) onto which large disposable bags are attached; (iv) rotating wall vessel bioreactors also known as roller bottles and finally, (v) multi-layered static flasks. Whatever becomes the bioreactor of choice for erythroid cell culture, it will need to facilitate higher density culture, be scalable and incorporate automation. In the long term, this will make erythroid culture more cost-effective by: (i) reducing labor costs - cells would be cultured in a single container making the cultures easier to feed and less labor- intensive, with the possibility of automated, remote feed- ing of media and/or specific depleted nutrients; (ii) reduc- ing the footprint and space required for each batch pro- duction; (iii) easing scale up; and (iv) minimizing human error and batch-to-batch variation by carefully controlling different parameters (such as pH, agitation and oxygena- tion) for optimal culture conditions. One can eventually then imagine rooms filled with bioreactors manufacturing mRBC continuously at scale for clinical use.
The challenge for producing mRBC in any of the above types of bioreactors type lies in the variety of culture con- ditions required during this 3-week process. All cultures are initiated from a small number of HSPC or approxi- mately 100x106 PBMNC, seeded in a small volume of medium. The erythroid cells then proliferate reaching a fold expansion of >105 and requiring low cell densities for optimum growth (2-8x105 cells/mL) or medium replenish- ment (constant versus repeated batch feeding). Orthochromatic erythroblasts then enucleate and are rela- tively fragile during this process. Moreover, erythroid cul- tures are inherently asynchronous, a proportion of cells start enucleating while others are still proliferating; this is noticeable in the last week of culture. To maximize yield, culture conditions have to support cells at different stages of terminal differentiation and nascent reticulocytes must remain viable until the highest percentage enucleation is reached. A manufacturing process that uses a bioreactor efficiently, minimizing the transitions between different types of culture vessels during the 21 days of culture, pro- ducing a high yield of reticulocytes and automating feed- ing, still needs to be identified.
Once the final product is made (i.e., mRBC have been filtered and stored), new cultures have to be reinitiated using new donor-derived HSPC. This is where immortal- ized pluripotent stem cells or an immortalized cell line that enucleates efficiently would be a game-changer as stocks of the same cell phenotype could be maintained.
Reticulocyte filtration and storage
At the end of the procedure, large-scale erythroid cul- tures need to be volume reduced and filtered to separate the mRBC from the nucleated cells and expelled nuclei (also known as pyrenocytes). Currently, the filters used are dead-end leukoreduction filters routinely used by blood banks. These have been designed to filter whole blood, which typically consists of ~ 5x109 RBC and only ~5x106 nucleated cells per milliliter of blood. In compari- son, the percentage enucleation for mRBC cultures is approximately 80% and the medium contains pyreno- cytes as well as free DNA released by disintegrated nuclei. It is notable that very few studies on mRBC yield test the filterability of their final product or report the yield after filtration. This is a key parameter as the mRBC must be purified before clinical use and this process currently alters
haematologica | 2021; 106(9)
2309