Clonal Trajectories and Therapeutic Targeting of High-Risk SF3B1-Mutant Myelodysplastic Syndromes

Abstract

Myeloid neoplasms are driven by acquisition of clonal genetic lesions, yet it remains unclear how these lesions cooperate to transform hematopoietic stem and progenitor cells (HSPCs) programs and modify drug responses. The integration of genomic profiling with clinical and cytogenetic data in the recent IPSS-M has significantly improved risk stratification of patients with myelodysplastic syndromes (MDS). Spliceosome mutations in SF3B1 are common initiating events in MDS and confer favorable prognosis, while secondary mutations in RUNX1 and STAG2 transform this low-risk condition into high-risk MDS or secondary acute myeloid leukemia. To understand how these events cooperate with SF3B1 mutations, we modeled these clonal trajectories in primary human HSPCs. SF3B1 K700E mutation was introduced in CD34+ cells isolated from cord blood or peripheral blood of healthy donors, using CRISPR/Cas9 editing and AAV6-mediated homology directed repair. By introducing an intronic fluorescent reporter, this system allows for the tracking and isolation of SF3B1-mutant cells for molecular and functional analysis. Edited cells were heterozygous for the K700E mutation and recapitulated alternative 3' splice site selection of genes previously identified in SF3B1-mutant patients. To model disease progression, we introduced secondary RUNX1 (S-R), STAG2 (S-S), or control AAVS1 (S-A) mutations into SF3B1 K700E knock-in CD34+ HSPCs, leading to a reduction of RUNX1 or STAG2 expression in SF3B1-mutant cells (50% and 90%, respectively). The S-R and S-S trajectories induced divergent alterations in lineage specification of SF3B1-mutant HSPCs. S-R promoted myeloid skewing at the expense of the erythroid lineage, while S-S induced a block in differentiation, impairing both myeloid and erythroid differentiation. Consistently, introduction of secondary RUNX1 or STAG2 mutations in an induced pluripotent stem cell (iPSC-HPCs) model of SF3B1-mutant MDS recapitulated S-R myeloid expansion and S-S maturation arrest identified in primary HSPCs. S-R and S-S genotypes resulted in divergent transcriptional programs including inflammation, immune response, and myeloid differentiation. However, both high-risk clonal trajectories expanded the immature CD34+CD38- HSCs/multipotent progenitors (MPPs). S-R selectively expanded CD34+CD38-CD133+ HSCs, whereas S-S expanded the more mature CD34+CD38-CD133- population. These data suggest that, despite divergent molecular pathways, high-risk mutations converge on expansion of stem cell potential of SF3B1-mutant HSPCs. To understand how genetic (co-mutation patterns) and epigenetic (HSC state) heterogeneity can influence drug responses, we profiled a panel of spliceosome inhibitors and 166 FDA approved and investigational compounds. Both S-R and S-S genotypes conferred decreased response to single and combinatorial agents conventionally used for the treatment of myeloid neoplasms. High-risk genotypes maintained elevated sensitivity to SF3B inhibition, but conferred differential response to novel classes of spliceosome modulators, with STAG2 but not RUNX1 loss selectively promoting response to type I PRMTs inhibitor MS023. By contrast, CHK1 inhibitor Prexasertib was highly selective for SF3B1-mutant cells irrespective of co-mutations, inhibiting growth and cell cycle progression. Taken together these data indicate that the presence of SF3B1 mutation offers therapeutic vulnerabilities that can be leveraged to target high-risk genotypes, and different SF3B1 co-mutated genes can further modulate drug response of SF3B1-mutant cells. In conclusion, progression from low-risk SF3B1-mutant MDS to high-risk disease is mediated by molecularly distinct trajectories driven by RUNX1 and STAG2 mutations that converge on expansion of the HSC compartment. Moreover, clonal progression is associated with genotype-specific drug responses and increased resistance to standard agents, and ongoing studies are elucidating how genetic and epigenetic states affect therapeutic responses.