Leukemogenic Second Hits in CEBPA-Mutated AML Result in Diverse Effects on Haematopoiesis

Acute myeloid leukaemia (AML) is thought to develop through the stepwise acquisition of genetic aberrations within a haematopoietic stem or progenitor cell (HSPC) resulting in clonal expansion. N terminal mutations of the myeloid transcription factor CEBPA are not only a frequent event in sporadic AML, but they are also inherited in the germ line, conveying familial AML predisposition and thus can be assured to be an initiating event in leukaemogenesis. CEBPA-mutated AML also shows a remarkably conserved landscape of secondary mutations. Around a third of cases harbour mutations in the epigenetic regulator tet methylcytosine dioxygenase 2 (TET2), a gene frequently implicated in the pathogenesis of myeloid malignancy. A further third of cases are found to have mutations in the transcription factor, GATA2, moreover this latter association appears to be almost exclusive to CEBPA-mutated cases in AML.

We have used a zebrafish model to interrogate the effects of these signature mutation patterns on haematopoiesis in juvenile and adult fish. A CEBPA N terminal mutant line (Cebpa^Nterm) was generated using TALEN pairs, with a 5bp deletion leading to a frameshift between the first and second reading frame of this single exon gene, emulating that seen in both familial and sporadic AML cases. In homozygosity cebpa^NTerm leads to complete absence of mature myeloid cells until 4 weeks of age. However, at 6 weeks of age all surviving cebpa^Nterm/Nterm fish had developed granulated myeloid cells on assessment of whole kidney marrow (WKM), with a dosage dependent increase in numbers with loss of wild-type (WT) Cebpa. By contrast in the peripheral blood (PB), a dose dependent reduction in both thrombocyte and myeloid cell numbers was observed with loss of WT Cebpa, but no evidence of anaemia. No homozygotes have survived past 2 months, due to susceptibility to infection and development of leukaemia with primitive features.

To assess the effects of Tet2 loss in combination with cebpa^Nterm/+ we utilised a Tet2 null mutant line (tet2^zdf20) crossed to cebpa^Nterm/+. Tet2^zdf20/+ showed on haematopoietic lineages at 3 and 5 months of age as previously described {Gjini:2015bs}. However, when combined with heterozygosity for cebpa^Nterm, loss of Tet2 showed a cumulative deficit in both HSPC and myeloid cell numbers. Further evidence of cooperation between the two mutations was seen by assessment of thrombocyte number in PB, with platelet numbers steadily trending to their nadir at the maximum allelic loss i.e. cebpa^WT/Nterm ; tet2^zdf20/zdf20 noting that cebpa^Nterm/Nterm do not survive to this age. Whilst no anaemia was observed at 3 months in any genotype, by 5 months cebpa^WT/Nterm ; tet2^zdf20/+ fish showed a significant reduction in erythroid progenitors, suggesting an anticipatory effect of cebpa mutation on the tet2^zdf20/+ myelodysplastic phenotype not usually observed until 8 months of age.

Combination of cebpa^Nterm/+ with a gata2a (one of 2 zebrafish GATA2 orthologues) mutant carrying a truncating mutation prior to the zinc finger domains (gata2a^um27) revealed a different picture. On assessment at 5 months gata2a^um27/+ showed a reduction in both HSPC and myeloid cell numbers in the WKM, while cebpa^Nterm/WT showed an increase. In keeping with each individual mutation's influence on cell numbers, cebpa^Nterm/WT; gata2a^um27/+ compound heterozygotes appeared showed near WT levels in both lineages. No abnormalities in erythroid progenitors or peripheral anaemia was evident in any genotype.

In summary, we have described diverse haematopoietic effects of two known secondary mutations in cebpa^Nterm mutated AML. Mutations in Tet2 cooperate with cebpa^Nterm resulting in reduced cellularity across lineages. In the case of Gata2a, the addition of heterozygous cebpa^Nterm mutation appears to rescue a myeloid progenitor defect present in gata2a^um27/+. Finally, in contrast to cebpa^Nterm/Nterm, we did not observe any cases of frank leukaemia in cebpa^Nterm combined with Gata2a or Tet2 loss. We speculate this is because the most common second hit in cebpa^Nterm, is a cebpa mutation in the carboxy terminal of the second cebpa allele, and this may be requisite for the development of AML. We have now generated a zebrafish with C terminal Cebpa mutation in order to test this hypothesis.