Discovery of the First Pathogenic Human EPO Mutation Provides Mechanistic Insight into Cytokine Signaling

Congenital hypoplastic or Diamond-Blackfan anemia (DBA) is a rare bone marrow failure disorder characterized by a paucity of red blood cells and their precursors in the bone marrow. The majority of cases of DBA are due to haploinsufficient mutations in ribosomal protein genes and in rare cases result from GATA1 mutations. However, nearly half of the DBA cases do not have an identified genetic etiology. While analyzing whole exome sequencing data from a cohort of over 450 patients with a clinical diagnosis of DBA, we encountered the case of a male child of a first cousin consanguineous union who was diagnosed with DBA as an infant and remained transfusion dependent. The patient responded to corticosteroid therapy for a year as a toddler, but this treatment was discontinued due to side effects. The patient subsequently remained transfusion dependent and at 6 years of age an allogeneic bone marrow transplant from a matched maternal aunt was performed. Surprisingly, despite achievement of robust donor chimerism, the patient remained transfusion dependent. Unfortunately the patient developed severe graft-versus-host disease and died of resultant complications. Both the potential recessive nature of the mutation, given parental consanguinity, and the lack of anemia correction following transplant made this case extremely unusual. Thus we evaluated this patient's whole exome sequencing data. We identified a homozygous recessive mutation in the erythropoietin gene (EPO), which resulted in an R150Q substitution in the mature EPO protein. This mutation was absent from a cohort of 60,706 individuals depleted for Mendelian disease and fit the model of complete penetrance in the family. The R150Q mutation was expected to disrupt the high-affinity binding site to the EPO receptor (EPOR). However, we found by producing recombinant proteins that the EPO R150Q mutation only reduced the EPOR binding affinity by 3-fold. Surprisingly, the patient had an over 100-fold elevation in their serum EPO levels, suggesting that this mutation did not cause disease through altered affinity. Rather we observed altered EPOR binding kinetics by this mutant ligand. There was a slightly increased on-rate with a much faster dissociation rate (t_1/2 of 10 seconds for the mutant vs. 6 minutes for the wild type). Using human erythroid cells and primary hematopoietic stem and progenitor cells, we could show that this mutant ligand never reached the same efficacy as the wild type (WT) EPO in promoting erythroid differentiation and proliferation. To better characterize this abnormal activity, we examined downstream signaling responses. We found identical phosphorylation of STAT5 at maximally potent concentrations of the WT (1 nM) and R150Q mutant (100 nM) EPO. By surveying a broad array of >120 phosphorylation events using intracellular flow cytometry, we demonstrated that maximal levels of STAT3 and STAT1 phosphorylation were reduced by 30% and 25%, respectively, with the R150Q (100 nM) compared to WT (1 nM) EPO. To determine the mechanistic basis for variation in downstream effector activation by the R150Q mutant ligand, we used inhibitors of both the JAK2 kinase and the SHP1/2 phosphatases that are respectively up- and downstream of STAT phosphorylation. While SHP1/2 inhibition did not alter STAT phosphorylation, JAK2 inhibition by ruxolitinib more potently inhibited STAT1/3 phosphorylation compared to STAT5. Interestingly, treatment with a low dose of ruxolitinib (40 nM) reduced erythroid proliferation to the same extent at maximally potent concentrations of the WT or R150Q EPO, demonstrating that the impairment in signaling by the R150Q EPO was primarily due to reduced JAK2 activity. Finally, we utilized single molecule fluorescent imaging of EPOR dimerization at the intact cell surface to directly show that the kinetically-biased R150Q EPO has a reduced ability to promote productive dimerization as compared to the WT EPO, even at maximally potent concentrations. Collectively, our results demonstrate how the R150Q mutant EPO - the first pathogenic mutation in EPO identified in humans - results in biased agonism of EPOR signaling through reduced receptor dimerization and consequently impaired JAK2 activation. More broadly our findings reveal how variation of cytokine-receptor binding kinetics can be used to tune downstream responses, which has broad implications for modulating the activity of numerous hematopoietic cytokines.