A subset of myeloid progenitors drives leukemogenesis in familial CEBPA-mutated acute myeloid leukaemia

Abstract CEBPA is a transcription factor mutated in 10-15% of Acute Myeloid Leukemia (AML), including in families with germline mutations, predisposing them to AML. The CEBPA gene is transcribed from a single exon and produces 2 isoforms, p30 and p42, with characteristic mutations, affecting either the N-terminal portion (CEBPANterm) leading to production of only the p30 isoform, or in-frame in the bZIP domain in the C-terminal portion (CEBPACterm), disrupting DNA binding. In germline cases, the founder mutation is usually CEBPANterm. These cases have a near 100% penetrance to AML following the acquisition of further mutations (commonly CEBPACterm), and recurrence is also common, confirming this mutation as a driver of leukemogenesis in these patients. We have developed zebrafish models with cebpaNterm and cebpaCterm mutations. All homozygous and compound heterozygous cebpa fish have a loss of mature myeloid cells and die by 8 weeks of age with features of AML. This includes expansion of the haematopoietic stem and progenitor cells (HSPC) and morphological blasts, faithfully recapitulating the human disease. Whilst fish with multiple mutations show development of disease, fish carrying heterozygous mutations appear healthy. Similarly, patients with CEBPA germline mutations do not suffer with any clinical symptoms or haematological problems prior to second hit acquisition and AML. We hypothesise that a subset of heterozygous HSPCs are primed for leukemogenesis by CEBPANterm/+ and these provide a reservoir of leukaemia initiating cells following second hit mutations. We initially undertook bulk RNA-seq of GFP-expressing HSPC in Tg(cd41:eGFP;cebpa) mutants. We found very few differentially expressed genes between cebpaNterm/+and wild type HSPC. We also found no differences in cd41-expressing HSPC numbers or c-myb RNA expression by in situ. However, using the Tg(cmyb:eGFP) transgenic line shows a significant reduction in GFP in cebpaNterm/+zebrafish at 4dpf. This suggests a subset of proliferating myeloid-primed HSPC are altered in heterozygous cebpaNterm/+ fish. We therefore further investigated the effects of cebpaNterm/+ in adult zebrafish. We did not observe significant differences in survival between cebpaNterm and WT siblings (up until 18 month), however, analysis of kidney marrow (bone marrow equivalent) revealed an expansion of cd41-expressing HSPCs at 6 months, followed by an expansion of lyz-expressing myeloid cells and a reduction in the HSPC compartment at 16 months suggesting exaggerated age-dependent myeloid bias and HSPC loss. scRNA-seq of the whole kidney marrow at 16 months reveals a cluster of myeloid progenitors that express the marker genes lyz, mpx, pu.1 and csf3r, that have significant differences in gene expression between heterozygous cebpaNterm/+ and wildtype fish. Pathway analysis of the differentially expressed upregulated genes within this cluster shows upregulation of arginine and lysine transporter activity in cebpaNterm/+ fish. AML blasts have been shown to be highly dependent on arginine for survival and proliferation, and regulation of amino acid transport is vital for the progression of AML (Mussai et al., Blood 2015; 125 (15): 2386–2396). Further we observed significant downregulation of ribosomal proteins genes, as well as NF-κB signalling pathway in cebpaNterm/+ fish compared to siblings in this same myeloid progenitor cluster. Inhibition of NF-κB in AML cells leads to aberrant expression of key genes and pathways in leukemogenesis, including CEBPA expression, translation, and interleukin signalling suggesting a key role for this pathway in leukaemia development (Reikvam, Cells 2020; 9 (7):1677). In summary we demonstrate dynamic haematopoietic dysfunction associated with CEBPANterm/+ mutation and associated putative mechanisms driving leukemogenesis in familial AML with germline CEBPANterm/+ mutations that predispose them to cancer.