Single Cell RNA-Sequencing to Uncover Cell Type-Specific Gene Expression Changes during Inv(16) Leukemia Initiation

The transcription factor RUNX1 and its binding partner CBFβ form a heterodimer whose activity is essential for normal blood cell development. A subset of acute myeloid leukemias (AMLs) are marked by chromosomal abnormalities such as inversions and translocations affecting RUNX1 or CBFβ, leading to malignant hematopoiesis. For example, an inversion of chromosome 16 is seen in patients with the M4Eo subtype of AML and leads to the expression of the fusion protein CBFβ-SMMHC. This fusion protein acts as a driver of leukemogenesis but there are still many unanswered mechanistic questions about the leukemogenic process as well as the molecular activity of the fusion protein. The conditional knock-in mouse model (Kuo et. al, 2006; PMID: 16413472) allows us to study the earliest molecular events in leukemogenesis, since it takes 3-4 months to develop leukemia after the expression of CBFβ-SMMHC is induced in the adult mice. Previous work indicated that CBFβ-SMMHC expression leads to the formation of pre-leukemic hematopoietic cell populations in the bone marrow. In this study, we have utilized single cell RNA-sequencing to profile dynamic changes of hematopoietic cell populations based on gene expression analysis in individual cells, in order to understand the function of the CBFβ-SMMHC fusion protein in hematopoietic cells and leukemogenesis.

Flow cytometry was used to enrich for Cd117+ (c-Kit+) stem and progenitor cells from the total bone marrow of a) wild-type mice (n=2) b) pre-leukemic CBFβ-SMMHC expressing mice (n=2), and c) leukemic CBFβ-SMMHC expressing mice (n=2). The 10X Genomics Chromium platform was utilized to capture individual cells in droplets, with each cell and its transcripts uniquely tagged at the time of isolation. Prepared cDNA libraries were sequenced on an Illumina NextSeq 500. After post-sequence alignment and QC, expression matrices were imported into Seurat (Butler et. al, 2018; PMID: 29608179) for further analysis. In total, cDNAs from more than 14,000 cells were sequenced, each expressing between 1500-4000 genes. This high dimensional data was reduced using t-distributed stochastic neighbor embedding (t-SNE), a process that "clusters" the cells that are the most related to one another based on their transcriptional profile (See figure). Cluster identities can then be assigned based on canonical markers expressed within each cluster.

In wild-type mice, tSNE clustering revealed at least 14 unique cell identities among the c-Kit+ cells comprising stem cells, multipotent progenitors, and cells of myeloid, erythroid, and lymphoid lineages. Similar heterogeneity was observed for the c-Kit+ cells from CBFβ-SMMHC expressing mice that were pre-leukemic. The c-Kit+ cells from leukemic CBFβ-SMMHC expressing mice were more homogeneous, indicating a dominant leukemia clone. In the two pre-leukemic mice examined, a small cluster of c-Kit+ cells express some markers of megakaryocyte-erythroid progenitors (MEP), but express numerous other genes that classify them as "abnormal", a leukemogenic cell type that has been reported previously as abnormal myeloid progenitors (AMPs, Kuo et. al, 2006; PMID: 16413472). We have identified novel markers of this population, which expands as leukemia progresses. Interestingly, the top canonical pathway identified from markers in this AMP population is heme biosynthesis. In humans, upregulated heme biosynthesis has recently been reported in aggressive pediatric AMLs which overexpress MYCN (Fukuda et. al, 2017; PMID: 28768907). In the sequencing data from our mouse model, however, we see no alteration in Mycn, potentially suggesting an alternative driver of heme biosynthesis in inv[16] leukemogenesis.

Lastly, we noted numerous other alterations when comparing the data from wild-type to pre-leukemic c-Kit+ cells, both in the number of cells occupying a cluster (cell type) as well as gene expression differences within a cell type. These data truly underscore the large impact CBFβ-SMMHC has on the hematopoietic landscape and, because these data were obtained on the single cell level, we will be able to examine the gene expression alterations the fusion protein drives in specific cell subtypes.

View largeDownload slide

View largeDownload slide

Close modal