HRI coordinates translation by eIF2αP and mTORC1 to mitigate ineffective erythropoiesis in mice during iron deficiency

Abstract

Iron deficiency (ID) anemia is a prevalent disease, yet molecular mechanisms by which iron and heme regulate erythropoiesis are not completely understood. Heme-regulated eIF2α kinase (HRI) is a key hemoprotein in erythroid precursors that sense intracellular heme concentrations to balance globin synthesis with the amount of heme available for hemoglobin production. HRI is activated by heme deficiency and oxidative stress, and it phosphorylates eIF2α (eIF2αP), which inhibits the translation of globin messenger RNAs (mRNAs) and selectively enhances the translation of activating transcription factor 4 (ATF4) mRNA to induce stress response genes. Here, we generated a novel mouse model (eAA) with the erythroid-specific ablation of eIF2αP and demonstrated that eIF2αP is required for induction of ATF4 protein synthesis in vivo in erythroid cells during ID. We show for the first time that both eIF2αP and ATF4 are necessary to promote erythroid differentiation and to reduce oxidative stress in vivo during ID. Furthermore, the HRI-eIF2αP-ATF4 pathway suppresses mTORC1 signaling specifically in the erythroid lineage. Pharmacologic inhibition of mTORC1 significantly increased red blood cell counts and hemoglobin content in the blood, improved erythroid differentiation, and reduced splenomegaly of iron-deficient Hri^-/- and eAA mice. However, globin inclusions and elevated oxidative stress remained, demonstrating the essential nonredundant role of HRI-eIF2αP in these processes. Dietary iron repletion completely reversed ID anemia and ineffective erythropoiesis of Hri^-/- , eAA, and Atf4^-/- mice by inhibiting both HRI and mTORC1 signaling. Thus, HRI coordinates 2 key translation-regulation pathways, eIF2αP and mTORC1, to circumvent ineffective erythropoiesis, highlighting heme and translation in the regulation of erythropoiesis.

Graphical abstract

Figure 1.

Characterization of erythroid phenotypes of eAA and Atf4^−/−mice in ID. (A) Generation of eAA mice by crossing AATg mice with EpoRCre⁺ mice. (B) Defective eIF2αP in Ter119⁺ cells of eAA mice. Sorted Ter119⁺ erythroblasts (populations I+II) and Ter119^– (supplemental Figure 1A-B) from BM were used. (C) Complete blood cell count (CBC) of Wt, Hri^−/−, eAA, and Atf4^−/− mice under both +Fe and –Fe conditions. (D) Wright-Giemsa stained blood smears. Arrows indicate globin inclusions. Photographs were taken by using a Leitz optical microscope with PixeLINK Capture software. (E) Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) assay of insoluble globin precipitates from blood samples. Precipitated globin protein in 2000g pellets of equal numbers of blood cells (1 × 10⁷) were analyzed. (F) Spleen weights as percentage of body weights. (G) Serum Epo levels. (H) Correlation of serum Epo to hemoglobin levels. P values denote the comparison between Wt and mutant mice under –Fe conditions. *P < .05, **P < .01, ***P < .001. Numbers of mice used in (C and F): Wt+Fe, n = 25; Hri^−/−+Fe, n = 5; eAA+Fe, n = 6; Atf4^−/−+Fe, n = 11; Wt–Fe, n = 22; Hri^−/−–Fe, n = 6; eAA–Fe, n = 11; Atf4^−/−–Fe, n = 10. Numbers of mice used in (G): Wt+Fe, n = 11; Hri^−/−+Fe, n = 8; eAA+Fe, n = 5; Atf4^−/−+Fe, n = 4; Wt–Fe, n = 9; Hri^−/−–Fe, n = 9; eAA–Fe, n = 10; Atf4^−/−–Fe, n = 6. Numbers of mice used in (H): Wt–Fe, n = 6; Hri^−/−–Fe, n = 6; eAA–Fe, n = 10; Atf4^−/−–Fe, n = 6. BW, body weight; MCH, mean corpuscular hemoglobin; MCV, mean corpuscular volume.

Figure 2.

HRI-ISR is activated in ID and is necessary to promote differentiation and mitigate ROS. (A) HRI-eIF2αP signaling and (B) ATF4 expression in erythroid precursors from Wt, Hri^−/−, eAA, and Atf4^−/− mice. (C) Differentiation stages of Ter119⁺ cells from BM and spleen (Spl). +exp and –exp denote with or without splenic erythroid expansion, respectively, of Atf4^−/−–Fe mice (supplemental Figure 2C-E). (D-E) Representative ROS histograms for Ter119⁺ populations from eAA and Atf4^−/− mice. Wt–Fe (green shade), eAA+Fe (blue line), eAA–Fe (red line), and Atf4^−/−–Fe (red line) in BM, spleen, and blood samples. Numbers of mice used in (C): Wt+Fe, n = 34; Hri^−/−+Fe, n = 8; eAA+Fe, n = 6; Atf4^−/−+Fe, n = 4; Wt–Fe, n = 31; Hri^−/−–Fe, n = 9; eAA–Fe, n = 11; Atf4^−/−–Fe +exp, n = 5; Atf4^−/−–Fe, –exp, n = 5. Ret, reticulocyte.

Figure 3.

Suppression of elevated mTORC1 activity by rapamycin attenuates IE of iron-deficient Hri^−/−, eAA, and Atf4^−/−mice. (A) CBC analysis of blood samples after 12 days of vehicle or rapamycin (Rapa) treatments. (B) Spleen weights as percentage of body weight. (C) Serum Epo levels. P values denote the comparison between vehicle and rapamycin treatment of each genotype. *P < .05, **P < .01, ***P < .001. (D) Representative blood smears before (day 0) and after 12 days (day 12) of rapamycin treatments. Arrows indicate globin inclusions. (E) Erythroid differentiation of BM and spleen. (F) mTORC1 activity in total BM and spleen cells. Numbers of mice used in (A-C and E), +Fe conditions: Wt, n = 3; Wt+rapa, n = 3; –Fe conditions: Wt, n = 5; Wt+rapa, n = 4; Hri^−/−, n = 4; Hri^−/−+rapa, n = 5; eAA, n = 4; eAA+rapa, n = 5.

Figure 4.

INK128 reduces IE of Hri^−/−–Fe mice. (A) Reticulocyte percentage in blood samples and (B) spleen weights of Wt–Fe and Hri^−/−–Fe mice treated with vehicle or INK128 for 3 days. (C-D) Percentages of Ter119⁺ cells in spleens and BM. P values denote the difference between vehicle and INK128 treatment of each genotype. *P < .05, **P < .01. (E) Representative Wright-Giemsa stained blood smears before (day 0) and after 3 days (day 3) of INK128 treatment. (F) Representative cell pellets of BM and spleen samples. (G) Erythroid differentiation of BM and spleen. (H) Representative cell morphology of BM and spleen samples stained with May-Grunwald-Giemsa. Three mice were used for each condition and genotype.

Figure 5.

Iron repletion restores anemia and erythroid differentiation by attenuating HRI and mTORC1 signaling in iron-deficient mice. (A) RBC numbers and Hb levels in blood samples of iron-deficient (–FeR) and iron-replete (+FeR) mice. Mice were in ID for 16 to 20 weeks before FeR. (B) Representative Wright-Giemsa stained blood smears on day 0 and day 10 +FeR. Arrows indicate globin inclusions. (C) Representative histograms of ROS levels in reticulocytes of blood samples. (D) Spleen weights as percentage of body weight (BW). (E) Serum Epo levels. (F) Erythroid differentiation of BM and spleen samples. (G) Inactivation of HRI upon FeR in total spleen cells. (H) Downregulation of mTORC1 signaling upon FeR in spleen and blood cells. P values denote the difference between with and without FeR of each genotype. *P < .05, **P < .01, ***P < .001. Numbers of mice used: Wt–FeR, n = 3; Wt+FeR, n = 3; Hri^−/−–FeR, n = 4; Hri^−/−+FeR, n = 5; eAA–FeR, n = 5; eAA+FeR, n = 4; Atf4^−/−–FeR, n = 5; Atf4^−/−+FeR, n = 2.

Figure 6.

Inhibition of mTORC1 activity and protein synthesis in vivo by INK128. (A) mTORC1 activities measured by pS6 and p4EBP1 levels and (B) in vivo protein synthesis in the erythroid and non-erythroid cells in BM, spleen, and blood samples. Both Wt–Fe and Hri^−/−–Fe mice were treated with vehicle or INK128 for 6 hours. Equal numbers of nucleated cells from BM and spleen were loaded, and the exposure time for developing the western blot was the same for BM and spleen. For blood samples, equal volumes of packed cells were loaded. For measurement of in vivo protein synthesis, the entire nitrocellulose membrane was incubated with anti-puromycin antibody. Puromycin signals from the entire lane of western blots demonstrate protein synthesis activity in this particular sample, which were quantified by ImageJ software and indicated at the bottom of each lane (details available in supplemental Methods). The protein synthesis in Wt–Fe BM samples is defined as 1 for samples from BM and spleens; for Wt–Fe blood samples, it is defined as 1 for comparison of protein synthesis in blood (right panel).

Figure 7.

Proposed model of heme regulation of erythropoiesis during ID in vivo by HRI through coordinated translational control at eIF2αP and mTORC1 signaling. Left: Steady-state erythropoiesis in the BM under iron sufficiency. In iron and thus heme abundance, HRI homodimer is inactive because of full occupancies of heme onto the 4 HRI heme binding domains unable to phosphorylate eIF2α, thus permitting global protein synthesis, mainly globin proteins in erythroid cells. Sufficient hemoglobin production maintains oxygen-delivering capacity in blood without hypoxia. Middle: Activation of HRI-ISR under ID mitigates IE. Under iron/heme deficiency, HRI in BM erythroid precursors is activated by the dissociation of heme. HRI then induces ISR, phosphorylating eIF2α, which inhibits globin protein synthesis and results in a decrease of hemoglobin content and consequently induction of tissue hypoxia stress. In addition, eIF2αP selectively enhances the translation of ATF4 mRNA to alleviate ROS levels. Most important and novel here is that HRI-ISR inhibits mTORC1 signaling to mitigate IE in the spleen. Right: Elevated mTORC1 signaling and development of IE in mutant mice defective in HRI-ISR signaling in ID. Hypoxia induced by ID stimulates Epo production in the kidney and increases Epo in blood circulation. In the spleen, binding of Epo to its receptors in erythroid precursors induces AKT/mTORC1 signaling, thus phosphorylating 4EBP1 and S6K/S6 to increase protein synthesis, promote proliferation, and inhibit erythroid differentiation, which are the characteristics of IE. HRI-ISR serves as feedback to inhibit mTORC1 signaling activity inhibiting the development of IE in ID.