Characterization of Cooperating Mutations in Murine Acute Leukemias: Towards Establishing Leukemia Models with Genetic Lesions in Defined Cellular Pathways

Acute leukemia (AL) is caused by somatic genetic alterations including chromosomal aberrations and point mutations. AL is genetically extremely heterogeneous with each case having about 4 to 8 cooperating driver mutations. Even with modern gene-editing techniques, like CRISPR-Cas9, it is difficult to generate animal models that have several driver mutations. We established retroviral transduction murine bone marrow transplantation leukemia models (MBMTLM) using two primary driver oncogenes and identified cooperating mutations using whole exome sequencing (WES). One model used the AML1-ETO9a (AE9a) fusion, an alternatively spliced isoform of AML1-ETO, and the second model used the CALM-AF10 minimal fusion protein (CA-MF), which contains the regions of CALM and AF10 required for leukemogenesis. AML1-ETO is associated with acute myeloid leukaemia (AML), whereas CALM-AF10 is seen in AML, T cell acute lymphoblastic leukemia and malignant lymphoma. Previous studies in animal models have shown that these fusion genes are not sufficient to cause leukemia on their own but require the cooperation of additional mutations. All of the AE9a (n= 13) and CA-MF (n=21) transplanted mice developed AL with a median latency of 115 days and 91 days, respectively. Immunophenotypic analysis of the AE9a leukemias showed a myeloid phenotype. Interestingly, immunophenotypic analysis of the CA-MF leukemias showed two distinct phenotypes: myeloid and biphenotypic leukemia. Both leukemias were transplantable into secondary and tertiary recipient mice, which showed the same leukemia phenotypes as their respective primary leukemias but had a much shorter latency.

We hypothesised that the difference in phenotypes and latencies were due to different cooperating mutations which had arisen spontaneously. To identify these cooperating mutations, we performed WES on leukemia cells and the corresponding germline DNA of the donor mice. So far, we have analysed 32 leukemia exomes with a 97x mean coverage (15 AE9a and 17 CA-MF). On average, we found 9.1 somatic potential driver mutations per exome with significantly more mutations in AE9a leukemias than in CA-MF, 14.4 vs 3.8 per exome, respectively. Any somatic mutation with a high (nonsense) or moderate (missense) predicted impact was considered a potential driver mutation. In both CA-MF and AE9a leukemias, we found mutations in genes known to be mutated in human leukemia. One of the CA-MF leukemias had a mutation in the activation loop of the tyrosine kinase domain of Flt3 , while another carried a mutation in the RING finger domain of Cbl , which encodes an E3 ubiquitin ligase and is a negative regulator of FLT3 signaling. Mutations in the protein tyrosine phosphatase Ptpn11 and the polycomb group tumor suppressor gene L3mbtl1 (histone methyl-lysine binding protein) were found in the AE9a leukemias. The mutations found in Cbl , Flt3 and Ptpn11 correspond to known mutations in human leukemia. Even though we have only analyzed 24 leukemias so far, we found one recurringly mutated gene. Two AE9a leukemias and one CA-MF leukemia had mutations in the Ppig gene. Ppig encodes a peptidyl-prolyl cis trans isomerase. The mutated genes belong to the following functional groups: protein tyrosine phosphatases ( Ptpn11, Ptprt , Ptprg) , tyrosine kinases ( Flt3 , Sphk2 ), ubiquitin ligases ( Cbl and Ubc9 ), transcriptional repressors or epigenetic modifiers ( Dpy30 , Spen , L3mbtl1 ), nuclear exporters ( Xpo1 , Ranbp3 ), guanine exchange factors ( Dennd1b and Mcf2 ), DNA replication and DNA damage repair genes ( Rfc3 , Rad54l2 ) and cohesin complex ( Pds5a ). Members of these gene categories are known to be frequently mutated in human leukemia.

Our results strongly suggest that CALM-AF10 and AML1-ETO require additional cooperating events to cause leukemia. MBMTLMs are a unique resource for studying clonal evolution and cooperating mutations. These genomically well characterised mouse leukemias provide more realistic models for human disease and can also lead to the discovery of potentially important novel pathways in leukemogenesis (eg: peptidylprolyl cis-trans isomerases). Our MBMTLMs will be a unique resource for the in depth study of leukemias with multiple driver mutations in defined cellular pathways (eg: growth factor signaling, nuclear export, cohesins, polycomb complex) and should facilitate development of treatment strategies targeting these pathways and combinations of pathways.