Temporally Distinct Developmental Waves of Erythropoiesis from Human Pluripotent Stem Cells

Mammalian erythropoiesis during embryogenesis occurs in several distinct stages or "waves" that vary according to timing, site of production, gene expression and physiology. The ontogeny of mammalian erythropoiesis is most thoroughly studied in mice where the earliest circulating erythroblasts released from the yolk sac are termed primitive. Later, the first definitive erythroid lineage is established by erythro-myeloid progenitors (EMPs) that originate in the yolk sac and migrate to the fetal liver for terminal differentiation. A second wave of definitive erythropoiesis is established from hematopoietic stem/progenitor cells that originate in the dorsal aorta and migrate to later stage fetal liver for terminal differentiation. Finally around birth, definitive erythropoiesis shifts to the bone marrow. The ontogeny of erythropoiesis overlaps in mice and humans, although less is known about the latter, as hematopoietic tissues from precisely staged early human embryos are difficult to obtain. We hypothesized that the initial steps of human erythroid ontogeny could be recapitulated by induced pluripotent stem cells (iPSCs) induced to undergo hematopoietic differentiation. We used a serum- and feeder-free protocol to differentiate iPSCs into embryoid bodies (EBs) that produced two sequential waves of distinctly different erythroid precursors. At day 8 of differentiation, EBs began to release hematopoietic precursors. Thereafter, erythroid precursors were released from the EBs in the presence of stem cell factor (SCF), erythropoietin (EPO) and insulin-like growth factor 1 (IGF-1). Erythroid precursors produced during wave 1 (days 12-23 of differentiation) were relatively large and expressed embryonic-type globins (zeta and epsilon), resembling those produced during primitive erythropoiesis. In contrast, wave 2 erythroblasts (day 27 and later) were smaller and expressed mainly gamma and alpha globins with some beta globin, suggestive of fetal-type definitive erythropoiesis. To investigate further the similarity of wave 1 and wave 2 erythroblasts to cells at the primitive and definitive stages of ontogeny, respectively, we used Affymetrix Genechips to analyze the global transcriptomes of stage-matched (CD235⁺ CD71^high) cells. As primary human primitive erythroblasts were not available for comparison, we compared the transcriptomes from the iPSC-derived erythroblasts with those of primary murine definitive and primitive erythroblasts that were flow cytometry-purified from embryonic day 15.5 (E15.5) fetal liver and E10.5 bloodstream, respectively. The comparisons showed that wave 1 erythroblasts from human pluripotent cells resembled more closely the erythroid primitive lineage from mice, while wave 2 erythroblasts from the human cells resembled the erythroid definitive lineage of mice (P-value < 0.05 by a modified Kolmogorov-Smirnov test). For example, SOX6 and BCL11A, preferentially expressed during definitive erythropoiesis, were expressed at relatively high levels in wave 2 erythroblasts. In addition, gene set enrichment analysis (GSEA) demonstrated that wave 2 human iPSC-derived erythroblasts and primary murine definitive erythroblasts expressed numerous genes related to immune/inflammatory pathways that were shown previously to be important for the formation of definitive hematopoietic stem and progenitor cells in zebrafish and mouse embryos. Our findings demonstrate that human iPSC-derived embryoid bodies recapitulate early stages of erythroid ontogeny with respect to the timing of emerging lineages and their gene expression. Additionally, gene expression studies of human iPSC-derived primitive and definitive erythroblasts indicate inflammatory signaling as a potential regulator of the later stage of erythroid development.