Genome Editing: the Fetal Origins of Acute Myeloid Leukemia in Down Syndrome

For patients with childhood malignancies, leukemia-initiating genetic mutations frequently occur antenatally, followed by the acquisition of additional mutations in the first few years of life.¹  A classic example of this phenomenon occurs in children with trisomy 21 (T21) who have a 150-fold increased risk of developing acute myeloid leukemia (AML). This is recognized as a unique form of pediatric AML: myeloid leukemia associated with Down syndrome (ML-DS).^2–5  Real-world observational studies in children born with T21 have identified somatic GATA1 mutations as the first molecular event leading to the development of a preleukemic clone, seeding the subsequent development of ML-DS.⁶ GATA1 mutations are found in approximately 30 percent of children born with T21, and have been detected as early as 21 weeks gestation,⁷  suggesting that the first mutations required for the development of ML-DS occur in utero. GATA1 mutations result in the production of a truncated isoform of GATA1 (GATA1-short or GATA1s),⁸  which functionally impacts normal hematopoiesis leading to thrombocytopenia, excess blasts in the peripheral blood film, and a pre-leukemic state known as “transient abnormal myelopoiesis” (TAM).^6,9–11  TAM is only described in the context of T21 and GATA1 mutations; it has never been associated with leukemia in the absence of T21.¹²  In most cases, TAM spontaneously resolves in the neonatal period with only 10 to 20 percent of this patient group subsequently developing ML-DS, typically within the first four years of life.^13–15  ML-DS cases always carry GATA1 mutations, demonstrating that this is an absolute requirement for the development of ML-DS. However, additional molecular defects, predominantly in cohesin complex genes such as STAG2, are detected in patient samples following transformation to ML-DS.^16–18 

These observations raise several questions regarding the evolution of ML-DS: What is the cell of origin of ML-DS during fetal life? Are these ML-DS–initiating cells unique to fetal hematopoiesis? What is the molecular mechanism by which an extra copy of chromosome 21 predisposes to a pre-leukemic TAM phenotype when a GATA1s mutation is acquired? Does the cell of origin (or underlying molecular mechanisms) present possible therapeutic vulnerabilities that could be used to treat or even prevent leukemia development in high-risk children? To experimentally address these important questions, sophisticated model systems are required; however, modelling and characterization of aggressive childhood leukemias is extremely challenging due to early initiating events occurring during antenatal life in developmentally restricted hematopoietic stem and progenitor cell (HSPC) populations. Advances in genome editing technologies have created new opportunities to manipulate HSPCs at different stages of ontogeny to address these challenging problems.

Dr. Elvin Wagenblast and colleagues used CRISPR/Cas9 genome editing of HSPC from human fetal liver to induce TAM- and ML-DS–related mutations. The authors characterized the hematopoietic phenotype of purified HSPC derived from normal disomic fetal liver (N-FL) and T21 fetal liver (T21-FL). T21-FL HSPCs showed reduced proliferation in vitro, increased megakaryocyte-erythroid progenitors (MEP) as previously described,¹⁹  and a lower level of engraftment in xenotranplantation assays. A preleukemic TAM-like phenotype was initiated in T21-FL but not in N-FL through the introduction of GATA1 mutations into long-term hematopoietic stem cells (LT-HSC). This in turn induced a proliferation advantage in T21-FL LT-HSC, accompanied by an increased production of CD41⁺ megakaryocytes, higher level of engraftment in xenotransplant assays, and an increase in blasts, together indicating that T21-FL LT-HSC is likely the cell of origin for ML-DS. Importantly, serial transplantation studies demonstrated that this T21-FL–induced preleukemia underwent spontaneous resolution, recapitulating the phenotype of TAM. The authors explored the molecular mechanism of the T21-associated FL development, and through analysis of the distinct pattern of DNA-binding occupancy of GATA1s, which, combined with gene expression analysis and lentiviral transduction, identified that simultaneous overexpression of three chromosome 21 microRNAs (MiR-99a, miR-125b-2, and miR-155) could recapitulate a T21 phenotype. Furthermore, knockout of these micro-RNAs inhibited the development of GATA1s-induced preleukemia in T21-FL.

Introduction of GATA1s in combination with STAG2 deletion or other cohesin complex genes led to the development of a leukemic phenotype with reduced survival of engrafted mice and markedly increased blast cells. Intriguingly, this phenotype was largely independent of T21. Unlike the GATA1s T21-FL pre-leukemia, multiple different progenitor cell populations were shown to be capable of propagating the leukemia. Moreover, the leukemia was serially transplantable but intriguingly could only be induced using fetal and early postnatal, but not adult, HPSCs, mimicking the developmental restriction of ML-DS. The leukemia was characterized by aberrant CD117 expression in the leukemic HSPCs, which the authors demonstrated could be used as a marker for disease-propagating cells. This was used as a potential therapeutic target through demonstration of a promising effect of KIT inhibitors on the GATA1s-induced preleukemic state and primary TAM samples, making it worthy of future clinical evaluation as a potential therapy to prevent leukemic progression of TAM.