Splicing of erythroid transcription factor is associated with therapeutic response in myelodysplastic syndromes

Anemia is the primary clinical manifestation of myelodysplastic syndromes (MDS), but the molecular pathogenesis of ineffective erythropoiesis remains incompletely understood. Luspatercept, an activin receptor 2B (ACVRIIB-Fc) ligand trap, has been approved to treat anemia, however its molecular mechanism of action is unclear. We found that the ACVR2B, its ligand GDF11, and effector, SMAD2, are upregulated in MDS patient samples. GDF11 inhibited human erythropoiesis in vitro and caused anemia in zebrafish, effects that were abrogated by luspatercept. Upon GDF11 stimulation of human erythroid progenitors, SMAD2 binding occurred in the erythroid regulatory regions, including at GATA1 intron. Intronic SMAD2 binding led to skipping of exon 2 of GATA1, resulting in a shorter, hypomorphic isoform (GATA1s). CRISPR deletion of the SMAD2 binding intronic region decreased GATA1s production upon GDF11 stimulation. Expression of gata1s in a mouse model led to anemia,rescued by a murine ActRIIB-Fc (RAP-536). Finally, RNA-seq analysis of samples from the Phase 3 MEDALIST trial revealed that responders to Luspatercept had a higher proportion of GATA1s compared to non-responders. Moreover, the increase RBCs post-treatment was linked to a relative decrease in GATA1s isoform. Our study indicates that GDF11-mediated SMAD2 activation results in an increase in functionally impaired GATA1 isoforms, consequently contributing to anemia and influencing responses to Luspatercept in MDS.