Stag2 Loss Drives Abnormal Chromatin Remodeling with Fli1 Hyperactivity in Npm1c Leukemia

STAG2 is a member of cohesin complex that is recurrently mutated in >10 cancers and is essential in maintaining the integrity of 3D genome partitioning structures known as topologically associated domains (TADs). Our previous work has demonstrated that depletion of various cohesin factors, including Stag2, leads to increased hematopoietic stem and progenitor population (HSPC) self-renewal and myeloid-biased differentiation. Loss of Stag2 leads to impaired sub-TADs and affects key hematopoietic transcription factors, such as PU.1, to access and engage their target genes. In addition to defective lineage priming, Stag2 loss also results in failure to silence stem-cell programs such as Fli1-targets including Mecom and Hoxa9, which is the direct binding target of NPM1c. Yet, the extent and cooperative transformational role of STAG2 mutation with NPM1c, remains unexplored.

To determine the impact of STAG2 mutation in NPM1c mutant background, we generated STAG2 KO (refer as KO) OCI-AML3 cells. Comparing to isogeneic non-targeting (refer as NT) control cells, KO cells have slower proliferation rate despite normal cell cycle profile. There is an increased cell surface expression of stem cell associated markers, such as CD34 and ESAM. To obtain molecular signature, we performed RNAseq and ATACseq. In RNAseq, we found that the KO cells have increased signature of DNA damage response as previously reported, whereas ATACseq showed an increased chromatin accessibility in the KO compared to NT, especially at the HOXA locus. Among the more accessible regions, motif analysis identified FLI1 as the predicted top transcription factor (TF) to bind. As NPM1c can directly bind chromatin, we analyzed published NPM1c Cut&Run data and identified significant overlap of the NPM1c binding within the more accessible regions of KO cells (33%), suggesting synergistic chromatin remodeling between STAG2 and NPM1c mutations. Transcriptomically, GSEA analysis showed increased expression of FLI1 targeted pathways, suggesting FLI1 hyperactivity as the result of abnormal chromatin remodeling in the KO cells. We are currently performing genetic silencing of FLI1 via siRNA as well as pharmacologically inhibition to determine whether FLI1 would be a therapeutic target in STAG2 KO leukemia.

Next, we generated dual Stag2^ΔNpm1^c/+ murine models with tamoxifen inducible UbcCreER^T2 to establish the functional effects on hematopoiesis. After 4 weeks, LSK cells (Lin-Sca1+Kit+) are increased in Stag2^ΔNpm1^c/+ double mutant mice. Within LSK cells, Stag2^ΔNpm1^c/+ but not Npm1^c/+ has marked expansion of the myeloid biased MPP2 (LSK+Flk2-Cd150+Cd48+) and MPP3 (LSK+Flk2-Cd150-Cd48+) compartment. Transplantation of Stag2^ΔNpm1^c/+ LSK cells showed impaired reconstitution capacity and myeloid biased output in primary and secondary transplantation. Molecularly, we performed bulk ATACseq on MPP3 and found Stag2^ΔNpm1^c/+ cells have increased accessibility compared to Npm1^c/+ cells. Motif analysis identified Fli1 as the top TF in Stag2^ΔNpm1^c/+, which confirms the observation in STAG2 KO OCI-AML3 cell lines. When aged, mice carrying Stag2^ΔNpm1^c/+ developed highly penetrant acute leukemia with macrocytic anemia and thrombocytopenia. Morphological examination of the bone marrow identified hypolobated megakaryocytes and dyserythropoiesis, which suggests myelodysplastic related changes. The leukemia is transplantable, and the phenotype of the recipients recapitulates the primary leukemia.

Overall, we have established both in vitro and in vivo system of cohesin mutant leukemia model. We determined that the abnormal chromatin remodeling driven by STAG2 and hyperactivity of FLI1 is the key dysfunction in Stag2/Npm1c co-mutant. Further studies include determining the phenotypic and chromatin response of FLI1 inhibition in STAG2 mutant models, which will pave the way to highlight new therapeutic potentials of cohesin mutant leukemia.