Modeling the Multi-Step Pathogenesis of Acute Myeloid Leukemia of Down Syndrome

The multistep pathogenesis of Down Syndrome (DS)-associated pre-leukemia and subsequent progression to acute leukemia is one of the better characterized of all human blood malignancies. Children with DS have a 150 fold increased risk of developing acute megakaryoblastic leukemia (AMKL) and greater than 30 fold increased risk of developing B cell acute lymphoblastic leukemia (B-ALL). DS-AMKL is often preceded in late fetal development or soon after birth by a pre-leukemic syndrome termed transient myeloproliferative disorder (TMD), which is characterized by high numbers of abnormal megakaryocytes and megakaryoblasts in the circulation, spleen and liver. Previous work has demonstrated that constitutional trisomy 21 results in expansion of megakaryocyte-erythroid progenitors (MEP) in fetal liver (FL) with a concomitant reduction in fetal pre-pro-B cells. The expanded MEP population subsequently acquires an N-terminal truncating mutation in the transcription factor GATA1 (termed GATA1s), leading to selective expansion of a pre-leukemic erythromegakaryocytic blast population. While the majority of DS-TMD cases spontaneously resolve within 3 months, up to 15% of DS-TMD neonates can develop lethal progressive liver fibrosis. Progression to AMKL following spontaneous resolution of TMD is associated with acquisition of at least one additional germline mutation. While murine models implicate a role for trisomy 21 and GATA1s in the leukemogenic process, they do not faithfully recapitulate the pathology of the human disease. Previous attempts to model DS-associated TMD through xenotransplantation of DS-FL and DS-TMD cells have proven technically challenging. Therefore, there remains a need for a human model to investigate the genetic steps required for initiation of DS-TMD and progression to DS-AMKL.

We previously identified a leukemia stem cell (LSC)-associated miRNA signature by sorting 13 adult AML patient samples into 4 sub-populations based on CD34/CD38 expression, followed by supervised analysis guided by the in vivo leukemia initiating capacity of each sub-population in an optimized xenotransplant model. Interestingly, the top three LSC-associated miRNA candidates are all located on chromosome 21. To determine the role of these miRNA in human leukemogenesis, we engineered a tri-cistronic lentivector for enforced expression. Compared to control vector-transduced cells, tri-cistronic vector-transduced Lin‒CD34+CD38‒ cord blood (CB) cells generated a myeloproliferative syndrome in xenotransplanted mice, with splenomegaly, enhanced CD45+ human bone marrow cellularity and blocked B cell development at the pro B cell stage. Human grafts were enriched for CD45⁺CD33⁺CD117⁺CD123⁺CD41^lo/CD42^lo cells in bone marrow, peripheral blood, spleen and liver. In the CD45‒ compartment, a distinct lineage switch was observed, with CD41+ megakaryocytic output supplanting normal CD235+ erythroid output. High numbers of CD41⁺CD42b⁺CD61⁺CD34^lo human platelets were detected in peripheral blood and spleen. Blood films revealed large dysplastic platelets and megakaryoblast-like cells. Histology showed hCD45+ packed bone marrow cavities, with loss of normal architecture. Bone marrow, spleen and liver all showed extensive reticulin deposition. In the lineage negative (Lin-) fraction of BM, we observed an expansion in the proportion of human MEP and multi-lymphoid progenitors (MLP). To further model leukemic progression, we expressed GATA1s in combination with our tri-cistronic miRNA vector. Mice transplanted with double transduced cells showed intermediate levels of splenomegaly and bone marrow cellularity compared to mice transplanted with cells transduced with tri-cistronic vector alone. The addition of GATA1s induced a complete loss of B cell development while restoring erythroid development. In human Lin‒ cells isolated from the BM, addition of mutant GATA1s further augmented the proportion and total numbers of MEP while restoring the MLP compartment to normal levels.

These data demonstrate that we have generated a human xenograft model of DS-TMD through enforced expression in normal CB cells of a tri-cistron comprising 3 LSC-associated miRNA in combination with mutant GATA1s. With this model in place, we plan to further interrogate the genetic lesions involved in progression from DS-TMD to DS-AMKL.