The inability to derive functional hematopoietic stem cells (HSCs) in vitro from pluripotent cells prevents widespread utilization of HSCs in the clinic; however, the molecular defects compromising the in vitro generated hematopoietic stem/progenitor cells (HSPCs) are unknown. Using a two-step differentiation method in which human embryonic stem cells (hESCs) were first differentiated into embryo bodies (EBs) and then CD34+ cells from hEBs were co-cultured on OP9M2 bone marrow mesenchymal stem cell (MSC) stroma (hEB-OP9), we were able to derive HSPCs expressing the HSC immunophenotype (CD34+CD38−CD90+CD45+) (hereafter termed CD90+HSPCs). Colony forming and stroma co-culture assays demonstrated that the hEB-OP9 CD90+HSPCs were able to differentiate into myelo-erythroid lineages and T-cells. However, when comparing CD90+HSPCs from hEB-OP9 to those from fetal liver (FL)—an in vivo source of HSCs—the former remained severely functionally limited in their proliferative potential and ability to differentiate into B-cells.
To identify the basis of the proliferative and differentiation defects, we performed microarray analysis to define gene expression differences between CD90+HSPCs derived from hEB-OP9, FL, early 3–5 week placenta (PL) and an earlier stage of hESC differentiation (hEB). This analysis revealed establishment of the general hematopoietic transcription factor network (e.g. SCL, RUNX1, CMYB, ETV6, HOXB4, MYB), demonstrating the successful differentiation and identification of hematopoietic cells using our two-step culturing techniques and immunophenotype criteria. Moreover, evaluation of Spearman coefficients confirmed CD90+HSPCs isolated from hEB-OP9 culture were brought into closer resemblance of the hFL CD90+HSPCs as compared to to the developmentally immature hEB and hPL CD90+HSPCs. Encouragingly, hEB-OP9 CD90+HSPCs displayed downregulation of expression of genes related to hemogenic endothelium development associated with hEB and hPL while genes critical in HSPC function, including DNA repair and chromatin modification, were upregulated to levels comparable to hFL-HSPCs. However, a subgroup of FL HSPC genes could not be induced in hEB-OP9 HSPCs, including the HOXA cluster genes and BCL11A—implicated in HSC self-renewal and B-cell formation, respectively. Interestingly, absence of HOXA genes and BCL11A and poor proliferative potential were also observed in HSPCs from early placenta, suggesting these defects are not in vitro artifacts but instead reflect an inability of hEB-OP9 HSPCs to complete developmental maturation. To validate the necessity of HOXA genes and BCL11A in proliferation potential and multipotency, we next utilized shRNAs to target MLL—the upstream regulator of the HOXA cluster—, individual HOXA genes, or BCL11A in FL-HSPCs to test whether knockdown was sufficient to recapitulate the defects observed in hESC-derived HSPCs. Knockdown of HOXA7 resulted in the loss of CD34+ cells while HOXA9 shRNA-treated cells displayed a loss of more differentiated CD38hi cells. MLL knockdown depleted both CD38+ and CD34+ populations. BCL11A silencing resulted in the loss of B-cells. These studies identify HOXA genes and BCL11A as developmentally regulated genes essential for generating self-renewing, multipotent HSCs from pluripotent cells.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.