Serially Transplantable Patient-Derived Xenograft Model of Chronic Myelomonocytic Leukemia (CMML)

Introduction: The establishment of preclinical mouse models for MDS or MDS/MPN overlap syndromes from patient cells often fails or results in low engraftment. If successful, engrafting cells in primary mouse recipients usually do not engraft secondary or later recipients, thus precluding genetic manipulation of the cells for functional studies. We hypothesized that oncogenic complementation of patient cells would a) allow engraftment and serial transplantation, b) enable genetic modification of the cells for functional evaluation and c) preserve the disease-associated biology of the cell. To study these hypotheses we chose the MN1 oncogene for complementation of cells from a patient with chronic myelomonocytic leukemia (CMML). Importantly, healthy cord blood cells, when transduced with MN1, induce myeloproliferation in vivo, but cannot be serially transplanted and do not induce acute myeloid leukemia (Imren et al. Blood 2014).

Aim: To establish a serially transplantable CMML patient-derived xenograft (PDX) model and perform a targeted in vivo shRNA screening in this model.

Patient and Methods: Bone marrow (BM) cells were harvested from a CMML patient with 15% blasts in BM (CMML-2). BM cells were depleted for T cells and transduced with a lentiviral vector encoding MN1 and GFP. One million MN1-transduced or untransduced cells were injected intravenously in sublethally irradiated NOD/SCID IL-2Rγ^-/- mice transgenic for SCF, IL3 and GM-CSF (NSGS). BM and spleen cells from engrafted mice were immunophenotyped and retransplanted in NSGS recipients up to 6 times. Mutations were evaluated by whole exome sequencing and confirmed by Sanger sequencing. A targeted shRNA library was generated and lentivirally transduced in serially transplantable CMML cells. Representation of transduced cells was monitored by next-generation sequencing of individual shRNAs.

Results: MN1 transduction of human CMML cells resulted in mean engraftment of human CD45+ cells of 9% in peripheral blood (PB) and 41% in BM at week 12 after first transplantation. In contrast, untransduced CMML cells were hardly detected in PB and BM even 22 weeks after transplantation (mean engraftment 1.1% and 1.7%, respectively). Transplantation of BM cells from primary into secondary recipients led to stable engraftment of MN1-transduced CMML cells in PB (11%), spleen (4%) and BM (26%). In contrast, transplantation from primary to secondary recipients of untransduced CMML cells showed no engraftment in PB over a period of 40 weeks. Importantly, MN1⁺CMML cells have been serially transplanted up to the sixth recipient with good engraftment levels in PB, BM and spleen and could be re-derived from frozen cells. Overall, MN1⁺CMML cells could be expanded in vivo to allow engraftment in 33 mice so far.

MN1⁺CMML cells in spleen and BM showed a myeloid phenotype expressing CD33, CD38 and CD14, but lacked expression of lymphoid markers CD3 and CD19. The stem and progenitor phenotype CD34+CD38- was found in 2.5% of these cells. By morphologic evaluation, monocytic cells prevailed in BM up to the third transplantation, while blast cells increased in later transplantations. Molecular analysis of primary patient cells identified mutations in BCOR, DNMT3A, U2AF1, NOTCH1 and NRAS . All mutations were still present in human cells isolated from recipient mice after the 4^th and 6^th transplantation, confirming the genetic stability of the model.

A shRNA pool directed against the above mentioned mutated genes, MN1, and luciferase was lentivirally transduced into cells from the fifth recipient and transplanted into the sixth recipient. After 8 weeks, shRNAs against luciferase and BCOR were well preserved in vivo in MN1⁺CMML cells, while shRNAs against U2AF1 and MN1 were depleted. Thus, MN1⁺CMML cells remained dependent on at least one additional gene besides MN1, thus validating our hypothesis that the disease biology is preserved in our induced PDX model.

Conclusion: We present a novel approach to establish serially transplantable MDS/MPN-PDX models by complementation with the MN1 oncogene. Our CMML model preserves the genetic and morphologic characteristics of the disease and allows dissection of the functional architecture of CMML and preclinical drug studies in this underserved disease.