BMI-1 Transduced Human Erythroid Cells: A Representative System for Modeling Transcriptional Programs during Terminal Erythroid Differentiation

Background: The innovative development of cultured red blood cells (cRBCs) not only offers a solution to the perennial issue of donor blood shortages but also holds the potential to reduce the risks of alloimmunization and other transfusion-related complications. We recently reported that BMI-1 transduction and subsequent overexpression in human erythroblasts confers self-renewal capabilities and produces extensively expanded erythroblasts (E3 cells). E3 cells not only proliferate for over 60 days in culture, but also retain the ability to undergo terminal differentiation in response to growth factor signals. In vitro differentiation of E3 cells produces cRBCs^BMI-1, which traverse the same stages of erythroid maturation as primary bone marrow derived erythroid cells based on similar immunophenotypic and morphologic characteristics. Furthermore, an RNA-sequencing comparison of differentiating erythroid cells indicates that cRBCs^BMI-1 likewise closely recapitulate the sequential transcriptional programs of primary human erythroblasts during terminal maturation. These results suggest that cRBCs^BMI-1 can be valuable tools for studying molecular signaling during erythropoiesis.

Methods: Fluorescence-activated cell sorting (FACS) using antibodies against the surface markers CD235a, CD71, and CD49d was applied to primary bone marrow cells as well as 2 E3 cell lines (from different adult donors) that were stimulated to undergo erythroid maturation. We isolated 5 distinct subpopulations from each culture condition: proerythroblasts (ProE), basophilic erythroblasts (Baso E), polychromatic erythroblasts (Poly E), orthochromatic erythroblasts (Ortho E), and reticulocytes (Retics). To confirm the stage of differentiation and the purity of each FACS sorted subpopulation, cytospins were evaluated by a trained hematopathologist blinded to culture conditions. We performed RNA sequencing on each isolated subpopulation to analyze the dynamic transcriptomic changes occurring during human erythropoiesis in BMI-1 stimulated E3 cells and in native bone marrow.

Results: The FACS sorting protocol resulted in isolated erythroid subpopulations that were more than 95% pure based on morphologic evaluation. Using RNA sequencing, we first quantified the number of annotated genes expressed at each differentiation stage. The number genes decreased during erythroid maturation from an average of 22,071 expressed human genes in ProE to 8,838 in Retics. The most dramatic changes in gene expression occurred during the transition from OrthoE to Retics (586 upregulated vs. 2388 downregulated transcripts) and from ProE to BasoE (96 upregulated, 201 downregulated). Furthermore, principal component analysis revealed clear differences in gene expression between distinct stages of erythroid differentiation, regardless of the cell source. In contrast, at each stage of differentiation, gene expression patterns were very similar between erythroid cells from BMI-1 lines or primary bone marrow cultures.

Conclusions: Under appropriate culture conditions, BMI-1 transduction of human peripheral blood mononuclear cells yields a highly proliferative population of erythroblasts (E3 cells) which can divide in culture for over 60 days. In this study, we demonstrated that during terminal maturation E3 cells progress through several stages of differentiation indistinguishable from erythroid cells derived from primary bone marrow. At each of 5 stages (ProE through Retics), differentiating erythroid cells demonstrated similar immunophenotypes, morphology, and gene expression signatures regardless of the source. These results not only validate E3 cells as attractive sources for production of mature RBCs for therapeutic transfusion, but also indicate that BMI-1 transduced human erythroblasts can serve as a model to dissect the differentiation programs that occur during terminal erythroid maturation.