Erythroid Ferritin Chains In Vivo Are Dispensable for Hemoglobinization but Play a Critical Non-Canonical Role in Erythropoietin Signaling

During erythroid differentiation, a surge occurs in the flux of extracellular iron to mitochondria to accommodate hemoglobinization. Using radiolabeling of human erythroblasts, Vaisman et al. tracked iron movement from extracellular transferrin to intracellular ferritin to heme, identifying a lysosomal requirement for iron to shift from ferritin to hemoglobin (Blood, 1997). Subsequently, the Philpott and Mancias labs mapped molecular components of this pathway, implicating the chaperone Pcbp1 in iron delivery to apoferritin and the cargo receptor Ncoa4 in holoferritin delivery to lysosomes; mice with deletions of either gene have defects in red cell (RBC) hemoglobinization, supporting an in vivo role for ferritin in iron trafficking to heme (Ryu et al., JCI, 2017 & Santana-Codina et al., Haematologica, 2019). However, the mild phenotypes in both models suggest in vivo contributions of ferritin-independent pathways. Recent studies have also highlighted non-canonical ferritin functions with relevance to erythropoiesis. One such function concerns ferroxidase activity of ferritin heavy chain (Fth1) which may protect these iron-laden cells from ferroptosis. Another non-canonical role of ferritin, recently described by our laboratory, consists of microtubule stabilization, potentially linking iron availability with trafficking of erythropoietin receptor (EpoR) vesicles (Nat. Comm., 2021). Nevertheless, in vivo roles for ferritin heavy and light (Ftl) chains in erythropoiesis remain ill-defined 3 decades after the initial studies.

In the current studies, deletion of Fth1 or Ftl in erythroid cells and erythroid-associated macrophages with EpoR-Cre revealed distinct developmental requirements. Fth1^fl/fl;EpoR-Cre mice showed partial lethality (P < 0.001, chi squared), while Ftl^fl/fl;EpoR-Cre had a normal Mendelian yield. Phenotypic characterization of the animals used the corresponding flox/wt-EpoR-Cre mice as controls. The rare Fth1^fl/fl;EpoR-Cre adult survivors surprisingly showed no abnormalities in any complete blood count (CBC) parameters, including reticulocyte count (retic), but had marked upregulation of RBC Ftl on immunoblot (> 10-fold). Within Fth1-deficient RBC, mass spectrometry identified ~2-fold increases in antioxidant factors (Txnrd1, Nqo1, Pirin, P < 0.01), suggesting compensation to mitigate ferroptosis. Ftl^fl/fl;EpoR-Cre mice had mild anemia (12% Hb decrease, P < 0.01), a slight increase in retic (1.5-fold, P < 0.01), and ~2-fold increase in RBC anti-proteotoxic factors (Bag6, Pfdn2, Hectd3, P < 0.01). Fth1-deficient fetal liver (d13.5) showed erythroid deficits in viability (P < 0.01), cell number (P < 0.05), differentiation (P < 0.05), proliferation (P < 0.05), along with increased apoptosis (~10-fold increase in cells with cleaved caspase 3, P < 0.05). We previously reported that the mutant EpoR-H allele, encoding a truncated EpoR deficient in endocytosis (“surface trapped”), renders mice resistant to anemia in the face of iron deficiency (J. Exp. Med., 2018). Strikingly, introduction of EpoR-H eliminated the phenotypic abnormalities associated with Fth1 deficiency: Fth1^fl/fl;EpoR-Cre;EpoR-H mice displayed normal Mendelian yield and normal fetal liver erythropoiesis. Effective loss of erythroid Fth1 expression was confirmed by immunoblot. For Fth1 + Ftl loss, Fth1^fl/wt;Ftl^fl/fl;EpoR-Cre were crossed with Fth1^fl/fl;Ftl1^fl/fl mice, revealing complete lethality in adult (P < 0.0001) and d13.5 embryo (P < 0.05) genotypes. Unexpectedly, the EpoR-H allele rescued full development of double-deleted mice; erythroid loss of Fth1 + Ftl in these animals was confirmed by immunostaining. The Fth1^fl/fl;Ftl^fl/fl;EpoR-Cre;EpoR-H mice had all normal CBC parameters including Hb, MCHC, and retic; Cre+ and Cre- controls were included (Fth1^fl/fl;Ftl^fl/wt;EpoR-Cre;EpoR-H and Fth1^fl/fl;Ftl^fl/fl;EpoR-H;EpoR-H). To probe the relationship between ferritin and EpoR, we conducted immunoprecipitations on erythroid HUDEP-2 cells and found co-precipitation of Fth1 and β-Tubulin with EpoR in 3 independent experiments. Thus, genetic and biochemical studies both support an interaction between ferritin and EpoR. Our in vivo findings show that at steady state erythroid ferritin is fully dispensable for hemoglobinization and ferroptosis prevention in RBC. By contrast, erythroid ferritin plays a novel, indispensable role in Epo signaling.