Animal models that faithfully recapitulate clinical features of myelodysplastic syndrome (MDS) have been difficult to establish. MDS are hematologic disorders associated with ineffective hematopoiesis, blood cytopenias, myeloid dysplasia and an increased risk of acute myeloid leukemia (AML). Although transplantation of primary AML cells into immunocompromised mice has been met with much success, primary MDS patient samples exhibit poor engraftment with no evidence of disease in recipient animals. 1–3 The engraftment of primary MDS samples is complicated further by frequent outgrowth of normal HSC clones. 4 These challenges underscore the current limitations and inefficiencies of engrafting primary MDS cells. Recent improvements in engrafting primary MDS cells have been observed when transplanting immunophenotypically defined HSC from the marrow of MDS patients or when co-injecting stromal cells engineered to produce non-cross-reacting human cytokines. 5, 6 Unfortunately, even with the improved engraftment of primary MDS cells, mice do not succumb to features resembling human MDS, precluding the use of these models for pre-clinical testing. To circumvent the current limitations, we developed a model using immunocompromised recipient mice and a human MDS cell line (MDSL) derived from the non-leukemic phase of an MDS patient with refractory anemia-ringed sideroblasts. 7–9 The MDSL line was derived as a subline of MDS92, and maintains factor dependency for cell growth, but has reduced differentiational capacity compared with MDS92. 10 Herein, we report the successful engraftment of MDSL cells into NOD/SCID-IL2Rγ mice (NSG) and NSG-hSCF/hGM-CSF/hIL3 (NSGS) mice, and reproducible development of disease, including cytopenias, clonal expansion and host hematopoietic suppression. In addition, we show that the MDSL xenograft model is a useful tool for evaluating novel and existing therapeutics for MDS. As reported for the original parental MDS92 line, 11 MDSL cells have maintained cytokine dependence in vitro. Upon hIL-3 withdrawal, MDSL cells stop growing within three days and die within five days (Figure 1a). Given the challenges of engrafting primary human MDS samples, we sought to determine the engraftment and diseaseinitiating potential of MDSL. A total of 1Â 106 MDSL cells were injected intravenously into 8–10 week old sub-lethally irradiated NSG or NSGS mice, and mice were monitored for evidence of hematopoietic malignancy. MDSL engraftment in NSGS or NSG resulted in a fatal hematologic disease associated with cytopenia, marrow failure and organ infiltration, with a median survival of

24 and 70 days post injection, respectively (Figure 1b). The shorter latency observed with the NSGS is attributed to the transgenic expression of three non-cross-reacting human cytokines (SCF, GMCSF and IL3), which provide an enhanced milieu for the myeloid cell grafts. 12 At the time of killing, flow-cytometric analysis revealed efficient MDSL engraftment of hCD45þ cells in the bone marrow (BM), spleen (SP) and the peripheral blood (PB) of both NSG and NSGS recipient mice (Figure 1c). Interestingly, conditioning NSG or NSGS mice with sub-lethal radiation was not requisite for MDSL cell engraftment, however non-irradiated recipients did exhibit a delayed disease latency (data not shown). Intrafemoral BM aspirates of NSG at 53 days post transplant revealed an average human MDSL (CD45þ CD33þ) graft of 2% total marrow (Figure 1d). However, at time of death (82 days post transplant), MDSL cells progressively expanded to occupy B60% of the total marrow (Figure 1d). BM cellularity …