Genetic Investigation of the Role of GDF11 in the Treatment of β-Thalassemia and MDS

The current treatment of β-thalassemia only partially mitigates the phenotype of the disease, making the need for novel therapeutic agents imperative. The investigational drug Luspatercept (ACE-536) is a ligand trap that contains the modified extracellular domain of activin receptor IIB (ACVR2B) and induces red blood cell production in an erythropoietin independent fashion. ACE-536 binds with high affinity to members of the transforming growth factor (TGF) β superfamily and therefore alters activin/GDF signaling through the intracellular SMAD complex. In search of the specific ligands, recent studies in a mouse model of β thalassemia intermedia identified growth differentiation factor 11 (GDF11) as a possible target of the drug. It has been proposed that GDF11 is overexpressed in thalassemic erythroblasts and inhibits terminal erythroid maturation via SMAD complex phosphorylation. A negative role of GDF11 in erythropoiesis has been postulated also in myelodysplastic syndrome (MDS).

We recapitulate, by a genetic approach, the phenotype of thalassemic and MDS mice treated with RAP-536, the murine counterpart of ACE-536. We generated and analyzed animals with GDF11 deletion in erythroid cells (Hbb^th3/+ Gdf11^fl/flEpoR-Cre and NUP98-HOXD13 Gdf11^fl/flEpoR-Cre) and in all hematopoietic tissues (Hbb^th3/+Gdf11^fl/flVav-Cre and NUP98-HOXD13 Gdf11^fl/flVav-Cre). We did not detect any changes in red blood cell number, reticulocyte count, hemoglobin or hematocrit levels compared to thalassemic or MDS mice in absence of the floxed gene. Focusing on thalassemic mice, administration of RAP-536 significantly improved the anemia and other hematopoietic parameters in the peripheral blood, decreased spleen size and ameliorated ineffective erythropoiesis as indicated by an increased ratio of mature to immature splenic erythroblasts analyzed by flow cytometry. Similar endpoints were seen comparing floxed and non-floxed animals treated with RAP-536. Therefore, the lack of GDF11 in erythroid or bone marrow derived cells did not prevent a response to the drug. To assess the effect of a pancellular absence of GDF11, we are currently in process of generating a model of β-thalassemia with total Gdf11 deletion.

To investigate the possible effects of RAP-536, we treated erythroid cells derived from normal or thalassemic patients with the drug. Erythroid cell viability, number, differentiation and cell cycle remained unvaried. Of note, we did not detect significant expression of GDF11/Gdf11 in human and mouse erythroid cells. To investigate the effect of an exogenous source of GDF11 production on erythroid cells, we treated murine erythroleukemia (MEL) cells with recombinant GDF11. Upon treatment we observed phosphorylation of the SMAD2/3 complex by western blot. This effect was hindered by co-treatment of GDF11 with RAP-536.

We further assessed the effects of RAP-536 on the mouse erythroid transcriptome, using RNA seq analysis in splenic erythroid populations. After administration of a single dose of RAP-536, thalassemic mice were euthanized. We used flow cytometry to identify possible alterations on differentiating erythroid populations in the spleen. Notably, between 60 and 72h we observed reduced numbers of basophilic and increased numbers of polychromatophilic erythroblasts. Analysis at 60h revealed that signal transducer and activator of transcription 5a (Stat5a), cyclin-dependent kinase 6 (Cdk6) and other cell cycle-related and metabolic genes were increased in the basophilic erythroid progenitors treated with RAP-536. This effect suggests that RAP-536 promotes proliferation and/or differentiation of erythroblasts.

Thus, our genetic analyses suggest that lack of GDF11 may be required but not sufficient to improve erythropoiesis. Furthermore, erythroid cells do not produce but can respond to exogenous GDF11, likely synthesized by non-erythroid cells and under conditions of ineffective erythropoiesis. Even though we detected in vitro effects, these may not mimic physiological effects, as experimental conditions may not correlate with GDF11 concentrations in vivo. As additional ligands have been proposed (such as GDF8 and Activin B), our future studies will focus on the potential role of these molecules. Altogether, these results reveal a potential alternative target of action for ACE-536 and may lead to the discovery of new therapeutic molecules.