A Human IPS Model Implicates Embryonic B-Myeloid Fate Restriction As a Developmental Susceptibility to ETV6-RUNX1

Childhood acute lymphoblastic leukemia (cALL) is clinically distinct from that in adults with higher incidence, better prognosis, a distinct mutational spectrum and that the pre-leukemic initiating events frequently occur in utero . One hypothesis for this difference is that childhood cancers arise in transient cell compartments unique to early human development. We have explored this in ETV6-RUNX1-associated cALL, where evidence from twin studies and neonatal heel prick testing has shown that this translocation occurs in utero and is an initiating event . Consistent with this, epidemiological evidence and changes in placental structure suggest that leukemic initiation occurs in the first trimester when both extra-embryonically and intra-embryonically-derived waves of hematopoiesis coexist. We therefore characterized B-lymphoid development in first trimester human embryos aiming to identify a candidate compartment vulnerable to the pre-leukemic initiating effect of ETV6-RUNX1. Using CD19 as a marker associated with definitive B lineage restriction in adult hematopoiesis, we established that the first CD19⁺ B cell progenitors emerged in the human fetal liver (FL) at Carnegie Stage (CS) 17 and were distinct from their cord blood or adult bone marrow counterparts in that the majority expressed surface IL7 receptor. Analysis of IL7R expression also uncovered a CD19^-IL7R⁺ candidate B progenitor compartment. CD19^-IL7R⁺ cells produced CD19⁺ B cells in vitro, possessed IGH DJ recombination, but also exhibited myeloid (but not megakaryocyte-erythroid) potential. Single cell analysis revealed co-expression of lymphoid and myeloid lineage programs in CD19^-IL7R⁺ progenitors at CS20, whereas at CS17 their transcriptional programs appeared markedly more myeloid, indicating that lymphoid potential is first acquired in CD19^-IL7R⁺ progenitors in this developmental window. Myeloid cells derived from CS20 IL7R progenitors harbored IGH DJ rearrangements, confirming their origins from a RAG-expressing lymphoid progenitor. We tested whether these distinct aspects of embryonic lymphoid development could be modeled using human pluripotent stem cells (hPSCs). In vitro B cell differentiation of hPSCs produced IL7R expressing CD19⁺ B cells as well as a CD19^-IL7R⁺ progenitor that also "switched" from myeloid to B-myeloid programming during culture. At the global transcriptional level the hPSC-derived lymphoid hierarchy maps closely with that from the embryo, with both separating from the adult, suggesting that hPSCs provide a developmentally relevant model of early fetal B lymphopoiesis. We next used CRIPSR-directed homologous recombination to knock in the cDNA of RUNX1 into intron V of ETV6 resulting in expression of ETV6-RUNX1 under the endogenous ETV6 promoter. ETV6-RUNX1 hPSCs displayed a partial block in B lymphoid differentiation at the CD19^-IL7R⁺ progenitor to proB cell transition and competitive co-culture demonstrated a selective advantage to the ETV6-RUNX1⁺ CD19^-IL7R⁺ progenitors over their wild type counterparts. Single cell transcriptional analysis showed that the ETV6-RUNX1⁺ CD19⁺ B cells that do emerge express an abnormal B-myeloid gene expression signature akin to that seen in the CD19^-IL7R⁺ progenitor. Consistent with this programming, when grown in myeloid cytokines, ETV6-RUNX1⁺ CD19⁺ B cells exhibited both a survival advantage and an aberrant ability to form IGH DJ rearranged macrophage colonies. Both the transcriptional and functional phenotypes were dependent on ETV6-RUNX1, as demonstrated by their reversion upon cre-mediated excision of the knock-in cassette. Our data support a model whereby ETV6-RUNX1 inhibits effective resolution of a lympho-myeloid program present in early fetal CD19^-IL7R⁺ progenitors, impeding B lineage differentiation and resulting in the production of B cells with aberrant myeloid gene expression and potential. Such programming may explain the relatively high levels of myeloid antigen expression and lineage promiscuity seen in ETV6-RUNX1 and other cALLs. This hPSC model of ETV6-RUNX1 pre-leukemia will afford systematic evaluation of the contribution of second hit mutations seen in cALL and offers a tractable platform for drug screening. In conclusion we propose that a novel IL7R⁺ lympho-myeloid progenitor in the human embryonic fetal liver is a candidate target cell for in utero pre-leukemic initiation in childhood B-ALL.