5-Azacytidine Reduces Leukemic Burden in a Xenograft Model of Juvenile Myelomonocytic Leukemia

Abstract

Juvenile myelomonocytic leukemia (JMML) is an aggressive myeloproliferative disorder of early childhood with often fatal outcome. Despite many attempts to develop alternative treatment options allogeneic hematopoietic stem cell transplantation (HSCT) remains the only curative modality. In the past our group has linked the prognosis of JMML to differential DNA methylation patterns (Olk-Batz, Blood 2011;117:4871-80 and Poetsch, Epigenetics 2014;9:1252-60), suggesting a key role of epigenetic modifications in JMML pathophysiology. To overcome the lack of suitable preclinical JMML research models we have developed an ex vivo JMML xenotransplantation system using neonatal Rag2-/- gamma-c-/- mice. Transplantation of 1x106 primary JMML cells resulted in stable xenologous engraftment and reproduced a characteristic JMML phenotype including myelomonocytic expansion; infiltration of spleen, liver and, notably, lung; splenomegaly; and reduced survival (median 26 weeks). Persistent human engraftment and leukemic organ infiltration was confirmed by both flow cytometry and immunohistology. Ras pathway mutations present in xenotransplanted patient samples were invariably confirmed in engrafted tissues. In addition, the model sustained serial transplantations and can therefore be used to amplify scarce patient material. We first tested if DNA methylation patterns in JMML cells were stable even after xenologous engraftment because such stability would be a prerequisite if the model were to be used for preclinical investigation of DNA methyltransferase inhibitors. JMML cells before xenotransplantation and those retrieved from the bone marrow of engrafted mice were profiled for global CpG methylation using Illumina 450K arrays. DNA methylation patterns in JMML were patient-specific and surprisingly robust in functional regions over several months of engraftment time (on average, 0.29% of 30877 promoters and 0.25 % of 30725 intragenic regions were called as "differentially methylated" between source and xenograft; 0.2 β-value change cutoff). These findings confirm the suitability of the xenograft model to investigate JMML epigenetics and, more importantly, indicate that patient-specific epigenetic profiles originate in leukemia-initiating stem cells, reinforcing a fundamental role of these alterations in JMML biology. Our group recently published a retrospective case series demonstrating unprecedented clinical efficacy of the DNA methyltransferase inhibitor 5-azacytidine (5AC) to induce partial or complete remissions in JMML before allogeneic HSCT (Cseh, Blood 2015;125:2311-3). To further investigate the drug on the preclinical level we administered 5AC to Rag2-/- gamma-c-/- mice xenografted with primary JMML cells. After a leukemia establishment phase the mice were divided into treatment or mock groups and treated with 5AC (3mg/kg body weight i.p., N=6) or saline (N=6) for 2 cycles (1 dose daily for 5 days; 9 days of recovery). This regimen was tolerated well by the animals. We found that 5AC reduced JMML infiltration in all organs analyzed, with most pronounced effects in spleen (human CD45+ fraction of all CD45+ cells, 0.24% +/- 0.04% vs 39.78% +/- 10.72%; p<0.01) and lung (0.41% +/-0.18% vs 42.88% +/-8.42%; p<0.01). The proportion of early progenitor cells (CD34+) within the human leukemia population in murine bone marrow was dramatically reduced after 5AC treatment (7.89% +/-0.74% vs 32.65% +/-3.76%; p<0.01) while the amount of granulocytes increased simultaneously (44.90% +/-1.74% vs 9.35% +/-1.95%; p<0.01). These findings suggest a loss of JMML cells induced by forced differentiation of more immature cells into mature myelomonocytic cells with reduced proliferation potential. Bisulfite pyrosequencing of the human BMP4 promoter CpG island, a locus frequently hypermethylated in JMML, showed significantly reduced DNA methylation in JMML cells retrieved from 5AC-treated mice (31.32% +/-2.66% vs 52.46% +/-1.39%; p<0.001). In summary we created an ex vivo JMML xenograft model in immunodeficient mice that reflects many important aspects of this disorder and proved its usefulness for preclinical research of DNA methyltransferase inhibition because of extraordinary stability of leukemic DNA methylation patterns. 5AC showed clear preclinical efficacy in this model, supporting its further development in clinical treatment strategies for JMML. Disclosures No relevant conflicts of interest to declare.