An Optimized Human Erythroblast Differentiation System Reveals Cholesterol-Dependency of Robust Production of Cultured Red Blood Cells Ex Vivo

The generation of cultured red blood cells (cRBCs) ex vivo represents a potentially unlimited source for RBC transfusion and other cell therapies. However, efficient differentiation and terminal maturation (symbolized by enucleation) into cRBCs ex vivo highly depend on replenishing human plasma. In addition to cost, this requirement results in variable potency across donors or batches and complicates consistent cRBC production required for clinical translations. Thus, exploration of key factor(s) in plasma and development of a compositionally defined medium would be in need.

To investigate the role of human plasma in erythroblast terminal maturation, we first optimized a widely-used differentiation system that we and other have used in the past decade. We found that culture of erythroblasts in a newly developed medium called human plasma like medium (HPLM) greatly improves the yield and consistence of erythrocyte maturation. However, human serum (1-2%) was indispensable for efficient terminal maturation in the absence of plasma. The serum-derived factor or factors (likely a protein or complex) is essential for erythroblast growth assessed at day 3 after induction of terminal differentiation. We used size exclusion chromatography (SEC) and other protein biochemistry approaches to fractionate human serum and assessed the biological effect of each fraction. We found that the activity is fractions containing proteins or complexes of >100K Da. Proteomics revealed that the fractions with biological activities are enriched for low density lipoproteins (LDL)-containing proteins such as apolipoprotein B-100. In addition, we performed RNA-sequencing (RNA-seq) analysis on cultured erythroblasts with or without serum. Transcriptomic data revealed that the genes associated with de novo biosynthesis of steroids and cholesterol are upregulated in the absence of serum, although production of enzymes encoded by these genes alone is insufficient to provide cholesterol required for erythroblast differentiation and maturation. Hence, we hypothesize that cholesterol-carrying LDL that are naturally present in blood is sufficient to replace human serum for erythroblast survival and terminal differentiation ex vivo.

We next examined directly whether LDL purified from human blood is sufficient for erythroid commitment and enucleation. Through LDL binding and uptake assays, we confirmed that LDL binds to erythroblasts that expressed the LDL receptor and is transported intracellularly. Furthermore, LDL purified from human plasma can essentially replace 2% unfractionated serum for efficient generation of erythrocytes in culture. In addition to using human erythroblasts derived from peripheral blood mononuclear cells (PBMCs), we also extended this optimized protocol to human erythroblasts from various sources, including cord blood-derived mononuclear cells, CD34+ HSPCs, and extensively expanded erythroblasts after BMI1 gene transduction. After induction of terminal differentiation for 8 days, erythroblasts from all these sources effectively enucleated (>50%) and stably produced reticulocytes with no significant difference compared to the serum-containing system. We are also in progress to substitute LDL by chemical synthetic cholesterol.

In sum, we developed a compositionally defined culture medium (LDL-containing Optimized Maturation Medium, LOAM) enabling robust generation of cRBCs from multiple sources of erythroblasts with high enucleation efficiency and higher reticulocyte yield compared to culture in traditional media supplemented with human plasma or serum. Using this advanced human erythroblast differentiation model, we further revealed that cholesterol carried by LDL is critical to erythroblast terminal maturation.