Modeling Clonal Progression in SF3B1-Mutant Myelodysplastic Syndrome

SF3B1-mutant myelodysplastic syndrome (MDS) has recently been proposed as a distinct disorder characterized by ring sideroblasts, ineffective erythropoiesis and good prognosis. Selected co-occurring genetic abnormalities were reported associated with significantly worse outcome and suggested as exclusion criteria for the proposed entity. However, it remains unclear how a limited spectrum of co-occurring drivers affects SF3B1-mutant MDS biology to determine evolution from a relatively indolent condition to high risk malignancy.

To gain a better insight into the clonal progression of SF3B1-mutant MDS, we analyzed SF3B1 co-mutations in a cohort of 176 SF3B1-mutated patients diagnosed with a myeloid neoplasm. RUNX1 and STAG2 were the only co-mutated genes found significantly associated with advanced disease phenotype (i.e. MDS with excess blasts and secondary acute myeloid leukemia) (OR=18.36 (2.18-862.91), P=0.001, and OR= Inf (3.57-Inf), P<.001, respectively), and reduced overall survival (P<.001). Based on these data, we hypothesized that acquisition of RUNX1 or STAG2 co-mutations in patients with SF3B1-mutant MDS drives progression to high-risk neoplasms.

RUNX1 and STAG2 control hematopoietic stem and progenitor cell self-renewal and differentiation by regulation of gene expression. To explore the biological impact of RUNX1 or STAG2 loss in the context of SF3B1-mutant MDS, we disrupted RUNX1 or STAG2 using CRISPR/Cas9 in an induced pluripotent stem cell (iPSC) model of MDS-RS that we previously established by reprogramming of bone marrow cells from SF3B1-mutant individuals. This patient-derived system allows hematopoietic progenitor expansion through doxycycline-mediated expression of 5 transcription factors (5F-HPC) and multilineage differentiation upon doxycycline withdrawal. We asked how RUNX1 or STAG2 disruption affected SF3B1-mutant HPCs self-renewal and differentiation in our model of MDS-RS, and sought to define the underlying transcriptional changes.

RUNX1-edited cells maintained significantly higher proportion of CD34 ⁺ HPCs, suggesting that RUNX1 mutation increases self-renewal capacity of SF3B1-mutant progenitors. Consistently, SF3B1/RUNX1 double-mutant 5F-HPCs showed positive enrichment of HSC-specific gene signatures. This was accompanied by broad upregulation of inflammatory programs, recapitulating the activated gene signatures we identified in SF3B1/RUNX1 co-mutated patients in our cohort. RUNX1 disruption promoted myeloid skewing at the expense of erythroid differentiation in SF3B1-mutant cells. Consistent with this, granulocyte-monocyte progenitors (GMP) and myelo-erythroid transcriptional programs were positively and negatively enriched, respectively. By contrast, in SF3B1/STAG2 double-mutant 5F-HPCs, both erythroid and myeloid populations were reduced compared to SF3B1 single mutant controls, suggesting a block in both myeloid and erythroid differentiation, despite the presence of pro-differentiation signals. This was supported by a profound down-regulation of genes involved in the response to external stimulus and suppression of GMP specific transcriptional signature.

In summary, we identified RUNX1 and STAG2 mutations as main drivers of disease progression in SF3B1-mutant patients, and generated extensive cell line panels to interrogate their functional interaction with mutant SF3B1. By applying CRISPR/Cas9 editing to our MDS-RS model, we could overcome limited availability of primary MDS samples to show that RUNX1 or STAG2 co-mutations drive progression through distinct biological mechanisms in SF3B1-mutant HPCs.