FLT3^N676K drives Myeloid Leukemia in a Xenograft Model of KMT2A-MLLT3 leukemogenesis

Geneticrearrangements of the KMT2A gene (previously MLL) are found in approximately 10% of human acute leukemia. However, in acute lymphoblastic leukemia (ALL) occurring in infants less than 1 year of age, a leukemia subtype with a particularly poor prognosis, close to 80% have a KMT2A -rearrangement (KMT2A -R). The KMT2A -Rs display an unusual lineage promiscuity and even though the specific fusion gene and cell-of-origin likely has an impact, the mechanisms influencing lineage determination in KMT2A -leukemogenesis remains largely unknown. Recently, when defining the landscape of mutations in infant KMT2A -R ALL, we identified a novel FLT3^N676K mutation in the tyrosine kinase domain that caused factor independent growth of the interleukin-3 dependent leukemia cell line Ba/F3 (Andersson et al., 2015). Notably, FLT3^N676K was the most common FLT3 mutation in our cohort which also included non-infant KMT2A -R ALL and acute myeloid leukemia (AML).

Herein, we studied the biological impact of the FLT3^N676K mutationin KMT2A-MLLT3 driven leukemogenesis in human hematopoietic cells. Retroviral co-expression of FLT3^N676K and KMT2A-MLLT3 in human CD34⁺ enriched cord blood cells transplanted to NOD.Cg- Prkdc^scidIl2rgtm^1Wjl /SzJ (NSG) mice primarily caused AML and rarely ALL. By contrast, expression of KMT2A-MLLT3 alone caused ALL, dual-phenotype leukemia (DPL) or bilineal leukemia (BLL), and rarely AML. Mice receiving cells expressing KMT2A-MLLT3 alone(n=23) or KMT2A-MLLT3+FLT3^N676K (n=26) succumbed to disease at a median latency of 119 and 107.5 days respectively, with no significant difference in disease latency between the groups. However, examining all KMT2A-MLLT3 leukemias combined, with and without FLT3^N676K, highlighted that AML developed with a significantly shorter latency as compared ALL and DPL. Analysis of the bone marrow of moribund mice revealed that almost all cells (98.3±1.9%) expressed human CD45 (hCD45⁺). Mice receiving FLT3^N676K alone were sacrificed at day 209 or 223 with no sign of disease.

Gene expression analysis of highly purified leukemia cells revealed that KMT2A-MLLT3 + FLT3^N676K enforced Myc- and Myb transcriptional modules and increased expression of negative feedback regulators of the RAS signaling pathway. Further, KMT2A-MLLT3 DPL and ALL cells had highly similar gene expression profiles with high expression of PAX5, IKZF1/2, and EBF1, indicating that the DPL cells are lymphoid-primed and aberrantly express CD33. Both KMT2A-MLLT3 DPL and ALL cells displayed weak but distinct expression of CEBPA suggesting that KMT2A-MLLT3 DPL cells originate from a progenitor cell that express CEBPA . Consistent with the high similarities of DPL and ALL cells, a fraction of DPL cells became CD19⁺CD33^-ex vivo in lymphoid culture conditions. Further, we identified enrichment of SMAD signaling in KMT2A-MLLT3 lymphoid leukemia cells as well as in primary human KMT2A -R ALL, suggesting that TGFB-regulation is different in lymphoid and myeloid KMT2A -driven leukemia. Consistent with this, human KMT2A -R ALL cell lines were more sensitive towards a specific inhibitor of SMAD3 as compared to KMT2A -R AML cell lines, suggesting that inhibition of SMAD signaling may be a fruitful approach for KMT2A -R ALL.

Taken together, herein we show that co-expression of KMT2A-MLLT3 and FLT3^N676K in human cord blood cells preferentially drive expansion of acute myeloid leukemia cells. The data suggest that a cooperating lesion can influence the immunophenotype of the resultant leukemia, at least in a xenograft setting, thereby adding new insight into the biology of KMT2A -R leukemia and the mechanism by which leukemia-associated genetic lesions cooperate in leukemogenesis.