Characterization of an Acute Myeloid Leukemia Murine Model Driven By MLL/AF9: Effect of Retroviral Insertion Sites and Somatic Mutations on Gene Expression

The MLL/AF9 fusion is found in approximately 30% of MLL-rearranged leukemias and has an intermediate prognosis. Genomically well-characterized murine leukemia models enable us to understand leukemogenesis.

We generated a retroviral transduction murine bone marrow transplantation leukemia model (MBMTLM) using the MLL/AF9 fusion gene. Fifteen of 20 mice transplanted with syngeneic bone marrow transduced with a MLL/AF9 carrying retrovirus developed leukemia after a median latency of 149 days. Half a million leukemic bone marrow (LBM) cells from two of these primary leukemias, MA03-P and MA86-P, were transplanted into irradiated recipient mice to establish secondary leukemias, MA03-S (n=3) and MA86-S (n=4). Half a million LBM cells from these secondary leukemias were further transplanted into irradiated recipient mice to generate tertiary leukemias, MA03-T (n=3) and MA86-T (n=4). The latency of the leukemias shortened from 141 days in MA03-P to 18 and 22 days in MA03-S and MA03-T, respectively. Similarly, MA86-P had a latency of 98 days, and the latency was reduced to about 28 days in MA986-S and MA986-T.

We used retroviral insertion sites (RISs) to track leukemia clones during serial transplantation. We identified 5 RISs in MA03-P. One RIS, RIS#1-03 at chromosome 7:4602500-4609499 accounted for 52.5% of the total RIS-related reads in MA03-P, while the other four RISs were each represented by fewer than 5% of the reads. Only RIS#1-03 was detected in all of the MA03 secondary and tertiary leukemias , indicating that the cells with RIS#1-03 were the dominant clone in MA03 leukemias. Two RISs were detected in MA86-P: RIS#1-86 at chromosome 19:41338500-41341999 and RIS#2-86 at chromosome 10:127106000-127109499 at 46.7% and 2.5%, respectively . RIS#1-986 was contained in the dominant clone as only this RIS was subsequently detected in the secondary and tertiary MA86 leukemias.

The relatively long latency to leukemia development in our MLL/AF9 model was most likely due to the requirement of cooperating somatic mutations. We performed whole exome sequencing on DNA from LBM (n=15) and DNA from their corresponding germline (n=2). An average of 4.5 of single nucleotide variants (SNVs) and 11.4 indels affecting protein coding sequences were found in the MA03 family of leukemias (n=7) which, among others, mutated genes involved in tyrosine kinase pathways such as Epha5 and Pik3r1. We identified an average of 14.8 (SNVs) and 0.5 indels per exome in the MA86 leukemias (n=8). Transcription regulator (Brd1) and tumor suppressor genes (Stk11 and Trp53) were affected by somatic changes in the MA86 family.

RNA sequencing was performed on LBM (n=15) and healthy bone marrow (HBM) (n=8). Principal component analysis (PCA) on the expression profiles showed that LBM samples clustered together. Differential gene expression analysis identified genes such as Six1, Eya1 and Bcor which had been reported in previous studies to be essential for leukemogenesis in MLL/AF9 murine model. We also observed downregulation of genes such as Gata2, Btg1, Ifitm1, which had been implicated in other types of leukemias.

We next investigated the effect of the RISs and somatic mutations on gene expression. RIS#1-903 was in intron 1 of Ppp6r1. A reduction in fragments per kilobase of transcript per Million mapped reads (FPKM) of Ppp6r1 was observed in MA03 family leukemias compared to leukemias of the MA86 family which did not have RIS#1-03 and showed no difference to HBM samples (MA03: 87.71±1.5; MA86: 132.1±5.1; HBM: 77.56±1.7, p< 0.001). We then determined the expression of Tm9sf3 as it is located 600bp away from RIS#1-986. The FPKM of Tm9sf3 was significantly higher in LBM (both of MA903 and MA986 leukemias) than in HBM (LBM: 146.0±12.7; HBM: 64.66±2.8, p<0.001). In MA86 leukemias which all have RIS#1-86, the FPKM of Tm9sf was two fold higher than in MA03 leukemias without RIS#1-86 (MA86: 189.3±4.4; MA03:97.59±1.7, p < 0.001). In contrast, none of the somatic mutations had a significant effect on the expression of any of the mutated genes.

In conclusion, we have established a MBMTLM driven by the MLL/AF9 fusion gene. This well-characterized model provides insights to further understand leukemia development and drug testing. Moreover, we demonstrated that RISs can have an impact on gene expression. Future work on whether Ppp6r1 and Tm9sf3 identified by our RIS analysis are drivers in MLL/AF9 leukemias is warranted.