Next Generation Humanized Mice Support Engraftment of Myelofibrosis CD34+ Cells

Introduction While the key transforming genetic events occur in the developing cancerous cell, this cell is dependent on its environmental context and interaction for competitive outgrowth and subsequent tumor-development. Myelofibrosis (MF) represents a model cancer disease with stepwise development from a chronic state that depends on microenvironmental interactions to a more aggressive disease. Engraftment of primary MF patient cells in murine xenograft models is poor (Wang et al., JCI 2012) and is possibly explained by the lack of supportive microenvironmental factors. Thrombopoietin (TPO) has been implicated in the pathogenesis of MF (Schepers et al., Cell Stem Cell 2013, Dadfarnia et al., Blood 2014, Abdel-Wahab et al., Annu Rev Med 2009). Also, the interaction between human hematopoietic cells and SIRPα expressed on mouse macrophages is critical for human engraftment in xenografts (Takenaka et al., Nature Immunology 2007). We hypothesized that the constitutive expression of human TPO and human SIRPα may promote the development of the human MF clone in mouse xenografts. Methods Purified peripheral blood CD34+ cells were collected from six patients with primary MF or post-PV/ET MF and low to intermediate 2 risk disease according to the dynamic international prognostic scoring system (DIPSS). Four patients carried a JAK2-V617F mutation and two patients carried a calreticulin (CALR) mutation. CD34+ cells were intrahepatically transplanted into sublethally irradiated newborn humanSIRPα-transgenic/humanTPO-knockin Rag2-/- gamma-/- (TPO-SIRPα) mice (Rongvaux et al., Ann Rev. Immunol 2013). NSG mice were used as controls and injected with the same number of CD34+ cells. Two to three mice were injected with ≥1 million CD34+ cells from the same patient sample each. Mice were sacrificed 12-16 weeks after transplantation and human engraftment and hematopoietic cell lineage distribution was assessed by flow cytometry using human specific antibodies. Tissues were collected for immunohistochemistry, assessment of fibrosis and spleen weight. DNA was extracted from whole bone marrow and a qualitative PCR was performed to determine the presence of the JAK2-V617F or CALR-mutations. Results Three out of six samples generated a human graft of ≥20% human CD45+ cells, while the three other samples generated engraftment of 0.1-3%. The human graft was mainly composed of myeloid cells and monocytic differentiation was observed. In 2/2 experiments analysed, a JAK2-V617F and a CALR type 2 mutation were detected in the bone marrow of engrafted mice transplanted with the respective patient sample. Development of fibrosis was not observed three months post-transplantation, presumably due to the short observation time. Spleen weight was significantly increased in mice engrafted with human MF and was the consequence of increased murine extramedullary hematopoiesis. We then aimed to identify factors that could predict human MF engraftment in TPO-SIRPα mice. While neither the DIPSS, nor the presence of myeloid precursors in the peripheral blood (blasts excluded) were predictive of human MF engraftment, the presence of blasts in the peripheral blood significantly correlated with engraftment potential. Importantly, none of the patients developed acute leukemia during follow-up. Finally, preliminary evidence suggests that TPO-SIRPα mice are more supportive of human MF engraftment than NSG mice. Conclusions This is the first xenograft model that supports robust engraftment of human peripheral blood MF cells and further supports a role for TPO in the pathogenesis of MF. In contrast to previous models TPO-SIRPα mice strongly promote myeloid rather than lymphoid engraftment. The tight correlation between the presence of peripheral blood blasts and the human MF engraftment potential suggests that human MF stem cells reside in the blast population. In summary, the xenograft model presented here constitutes a powerful tool to assess heterogeneity regarding MF biology, microenvironmental dependence of the MF clone and likely also therapeutic response of MF in vivo. Disclosures No relevant conflicts of interest to declare.