Combinatorial CSF3R and RUNX1 Mutations Involved in the Leukemic Progression of Severe Congenital Neutropenia Confer a G-CSF-Dependent Premalignant State in Mice

Background: Severe congenital neutropenia (SCN) is a genetically heterogeneous disorder characterized by a lack of peripheral neutrophils, leading to life-threatening bacterial infections. The disease is alleviated by sustained G-CSF treatment, which restores neutrophil production in the majority of patients. While on G-CSF therapy, SCN patients frequently develop hematopoietic clones with acquired CSF3R mutations. The appearance of CSF3R mutant clones is seen in multiple genetic subtypes of SCN and is strongly associated with progression towards MDS or AML. At this stage, CSF3R mutant clones acquire secondary mutations, most often (in ~75% of cases) in RUNX1 . The majority of these mutations affect the Runt homology domain (RHD) of RUNX1. This combination of CSF3R and RUNX1 mutations is exclusively seen in SCN/MDS/AML, suggesting a unique pattern of leukemic progression of SCN. However, the molecular and cellular consequences of combined CSF3R and RUNX1 mutations and the effects of sustained G-CSF therapy on leukemia development are unknown.

Aim: To determine the cooperative role of CSF3R and RUNX1 mutations in sequential steps of malignant transformation in vivo and to investigate how sustained G-CSF therapy affects this process.

Methods: We used our previously generated Csf3r -d715 mouse strain, which copies the most frequent G-CSFR truncation mutant in SCN patients. Lethally irradiated wild type mice were transplanted with Csf3r-d715 HSPCs transduced with an IRES-GFP backbone lentiviral vector expressing RHD mutant RUNX1-D171N or empty vector control. Mice were treated 3x per week with G-CSF or PBS (solvent control). Biweekly, peripheral blood (PB) samples were analyzed for GFP-marked engraftment, immunophenotypes and cytology. At regular intervals, in vitro progenitor cell colony assays were performed. Bone marrow (BM) and spleen samples were used for phenotypic analyses and secondary transplantations. RNA-seq experiments were carried out on FACS purified LK populations to identify molecular pathways associated with the observed phenotypes.

Results: Mice transplanted with Csf3r -d715/ RUNX1 -D171N (RHD mutant) BM cells started to accumulate GFP-marked Lin negative, c-Kit positive (LK) progenitor cells in the PB, 6 weeks after initiation of G-CSF treatment. The percentages of GFP+ LK cells in the PB ranged from 0.7% to 27.8% (Mean=5.9%; n=11) and fluctuated over time in individual mice, indicating that they were maintained but not vastly expansive at this stage. This phenotype was fully dependent on the combined presence of the Csf3r and RUNX1 mutations and was only seen in the G-CSF treated mice. The PB-LK cells had a blast-like morphology and formed myeloid colonies in semi-solid cultures. Transcriptome analysis of in vitro expanded LK cells by RNA-seq and gene set enrichment analysis (GSEA) showed that the combination of Csf3r -d715 and RUNX1 -D171N mutations and G-CSF stimulation selectively enhanced c-Myc , NF-κB and E2F1-driven transcriptional programs. Of note, fully differentiated GFP-positive neutrophils were also formed, both in vitro and in vivo and mice did not succumb to this pre-leukemic condition. Csf3r -d715/ RUNX1 -D171N BM cells isolated after 30 weeks could successfully engraft secondary recipients (n=6), again resulting in sustained G-CSF dependent accumulation of immature blast (LK) cells in the PB. In these mice, the blast cells in the BM exceeded 85% at week 7 after initiation of G-CSF treatment. At this stage, the blasts had become entirely defective in neutrophil differentiation, indicating that they had now undergone full transformation to AML. RNA-seq and genomic analyses of these AML blasts are currently underway to assess the molecular steps involved in the transition from the premalignant to the fully transformed state.

Conclusion: The combination of Csf3r -d715 and RUNX1 -D171N (RHD) mutations frequently seen in SCN/MDS/AML perturbs myeloid development in an in vivo transplant model in a G-CSF dependent manner. A sustained pre-leukemic state characterized by the selective accumulation of LK cells is seen in primary recipients, followed by a fully transformed AML phenotype in secondary recipients. This model is attractive for unraveling the molecular steps of malignant transformation driven by CSF3R and RUNX1 mutations and for preclinical testing of new therapeutic options for SCN/AML.