Cohesin Loss Dominantly Influences the Accessibility Landscape of Npm1^cA/+ AML

Acute myeloid leukemia (AML) is an aggressive cancer of the bone marrow and blood that affects 20,000 adults annually in the United States (Heimbruch et al. 2021). AML results from the serial accumulation of genetic mutations, which can occur in multiple combinations. While the effects of several individual mutations on disease development and progression are well studied, determining how different mutations interact with one another to influence disease characteristics and outcome is critical to understanding AML complexity. Additionally, the low 5-year survival rate of AML underlines the need for the development of mutation-targeted therapies.

One of the most common recurrent somatic mutations in adults with AML is in the NPM1 gene, which results in an aberrant cytosolic mislocalization of NPM1. This mutation is termed NPM1cA (Falini et al. 2005, Cancer Genome Atlas Research Network 2013). Approximately 50% of patients with an NPM1cA mutation also harbor a mutation in a cohesin complex gene, with cohesin mutation often resulting in haploinsufficiency. The cohesin complex is comprised of STAG2, SMC1A, SMC3, and RAD21 and forms a ring-like structure that regulates promoter-enhancer interactions through DNA looping and participates in three-dimensional genome organization. Recently, we utilized the previously-published Npm1^cAflox/+ and Smc3^flox/+ mouse models to examine the genetic interaction between Npm1^cA/+ and Smc3^-/+ (Vassiliou et al. 2011 and Viny et al. 2015). While mice bearing both mutations do not have decreased disease latency or penetrance, HSPCs from double mutant mice exhibit unique changes in gene expression that are not observed in single mutant HSPCs (Meyer et al. 2022). This suggests that cohesin haploinsufficiency affects the transcriptional environment in a unique way in the presence of the Npm1cA mutation.

Here, we present the mechanistic insights we have uncovered from ATAC sequencing performed on Npm1^cA/+;Smc3^-/+ vs. single mutant HSPCs. We find that Npm1^cA/+;Smc3^-/+ HSPCs have an increased number of open chromatin regions compared to Npm1^cA/+ only HSPCs, suggesting that Smc3 haploinsufficiency affects chromatin structure. We have previously shown that Npm1^cA/+;Smc3^-/+ HSPCs have an increased number of upregulated genes vs. single mutant HSPCs. Interestingly, the top transcription factor binding motifs found in open chromatin promoter-associated regions in both Smc3^-/+ and Npm1^cA/+;Smc3^-/+ HSPCs are for Ets family members (Erg, Fli1, PU.1). In contrast, the open regions in Npm1^cA/+ only HSPCs are enriched for Gata family binding sites, suggesting that the addition of Smc3 haploinsufficiency to Npm1^cA/+ may result in a switch from Gata to Ets-driven gene expression programs. To verify this, we generated a list of genes that were both differentially upregulated and acquired an open-chromatin promoter status in our Npm1^cA/+;Smc3^-/+ vs. Npm1^cA/+ only cells. This gene list was compared to previously-published Ets family member ChIP-seq data from a hematopoietic precursor cell line (HSC7)(Wu et al. 2014). We found that 70% of Fli1-bound genes overlapped with our gene list. Similarly, 71% of Erg-bound genes and 64% of PU.1 and overlapped with our ATAC gene list. In contrast, only 41% overlap was observed between the Gata2 ChIP-seq data and our gene list. This suggests that the addition of cohesin haploinsufficiency to Npm1^cA/+ dominantly causes a switch from Gata-driven to Ets-driven transcription and that inhibition of Ets-driven transcription may be a targetable and effective way to treat Cohesin^mut;NPM1cA leukemias. In support of this, preliminary studies using the Fli1 inhibitor calcimycin significantly reduced the growth of an Npm1^cA/+;Smc3^-/+ leukemic line vs. an Npm1^cA/+ line. In addition, we are analyzing scRNA-seq data to identify if changes are related to lineage skewing, or are alternatively driving a novel gene expression program in small subsets of cells. Our results show that combinatorial genetic models allow for a greater understanding complex genetic interactions and for the identification of novel, mutation-specific targets.

No relevant conflicts of interest to declare.