Mitochondrial translation regulates terminal erythroid differentiation by maintaining iron homeostasis

Abstract

Mitochondrial tRNA taurine modifications mediated by mitochondrial tRNA translation optimization 1 (Mto1) is essential for the mitochondrial protein translation. Mto1 deficiency was shown to induce proteostress in embryonic stem cells. A recent finding that a patient with MTO1 gene mutation showed severe anemia led us to hypothesize that Mto1 dysfunctions may result in defective erythropoiesis. Hematopoietic-specific Mto1 conditional knockout (cKO) mice were embryonic lethal and showed niche-independent defect in erythroblast proliferation and terminal differentiation. Mechanistically, mitochondrial oxidative phosphorylation complexes were severely impaired in the Mto1 cKO fetal liver, and this was followed by cytosolic iron accumulation. Overloaded cytosolic iron promoted heme biosynthesis, which induced an unfolded protein response (UPR) in Mto1 cKO erythroblasts. An iron chelator or UPR inhibitor rescued erythroid terminal differentiation in the Mto1 cKO fetal liver in vitro. This mitochondrial regulation of iron homeostasis revealed the indispensable role of mitochondrial tRNA modification in fetal hematopoiesis.

Mitochondrial tRNA modification regulates fetal erythropoiesis through the maintenance of intracellular iron homeostasis.

INTRODUCTION

The hematopoietic system is one of the most proliferative organs in the body and continuously produces new blood cells to maintain the peripheral circulation throughout a lifetime. Among hematopoietic cell lineages, erythrocytes are the most abundant in terms of cell number, accounting for 84% of all human cells (1). Erythropoiesis is a process that generates mature erythrocytes from hematopoietic stem and progenitor cells (HSPCs) to meet the staggering demand for red blood cells, an oxygen supplier (~2 × 10¹¹ per day) (2). Erythroid differentiation from proerythroblasts to enucleated reticulocytes is strictly controlled (3). During erythroid maturation, a large amount of hemoglobin is produced (4), while mitochondria, nuclei, and other organelles are released outside the cell (3). Protein synthesis in erythropoiesis needs to be tightly regulated because aberrant protein production induces cellular stress. Defect in unfolded protein response (UPR) has been reported to be associated with fetal anemia (5) and β-thalassemia (6).

Transfer RNAs (tRNAs) are small RNAs that decode genetic information in mRNA into proteins. tRNAs contain diverse chemical modifications that are post-transcriptionally introduced by tRNA modification enzymes. To date, more than 70 tRNA modification enzymes have been identified in humans (7). Post-transcriptional modifications have also been detected in mitochondrial tRNA (mt-tRNA) (8). A subset of mt-tRNAs [mt-tRNA^Leu(UUR), mt-tRNA^Lys, mt-tRNA^Glu, mt-tRNA^Gln, and mt-tRNA^Trp] contain taurine-derived modifications (9) at wobble position 34 U, which interacts with the third nucleotide of mRNA codons (10). Taurine modifications in 34 U stabilize codon-anticodon interactions, which controls the efficiency of decoding (11). Furthermore, the constitutive knockout (KO) of mitochondrial tRNA translation optimization 1 (Mto1), the core subunit of taurine modification enzyme complexes, was recently shown to markedly suppress mitochondrial protein translation, causing severe dysfunctions in energy production as well as imbalanced proteostasis in both embryonic stem (ES) cells and mice (12). Moreover, it has been reported that a patient with MTO1 gene mutation presented severe anemia though the unknown mechanism (13). These findings led us to hypothesize that Mto1 dysfunctions may result in defective hematopoiesis, particularly erythropoiesis, during which a large amount of protein is produced. However, the role of taurine modifications in mt-tRNAs in hematopoiesis, particularly in erythropoiesis, currently remains unclear.

Here, we generated hematopoietic-specific Mto1 conditional KO mice to examine the role of mt-tRNA taurine modifications in hematopoiesis. This Mto1 conditional KO mouse model showed defects in erythroblast proliferation and terminal erythroid differentiation in a cell-intrinsic manner. We demonstrated that a Mto1 deficiency resulted in reduced formation of OXPHOS complexes and altered intracellular iron homeostasis, which resulted in the up-regulation of heme biosynthesis and an induction of UPR. The present results will contribute to a more detailed understanding of the role of tRNA modification in hematopoiesis and the pathophysiology of mitochondrial diseases.

RESULTS

Hematopoietic-specific Mto1 KO leads to embryonic lethality and severe anemia

Constitutive global Mto1 KO previously shown to result in embryonic lethal (12), suggesting an indispensable role for Mto1 in fetal development. We first checked Mto1 gene expression in a database, expression Atlas (14) with different developmental stages of the mouse, and found that Mto1 was expressed highest at the embryonic day (E) 16 stage and in the fetal liver, which undertakes active hematopoiesis (Fig. 1A). Another database, Bloodspot (15), also showed that the Mto1 expression was the highest in the erythroid lineage, particularly in proerythroblasts (Fig. 1B). Both databases suggested the important role of Mto1 in fetal hematopoiesis, particularly in erythropoiesis.

Fig. 1. Hematopoietic-specific Mto1 KO leads to embryonic lethality and severe anemia.

(A) Mto1 gene expression patterns in mouse embryo tissues retrieved from the FANTOM5 project (see Materials and Methods). Color represents transcripts per million (TPM). (B) Mto1 mRNA expression in hematopoietic lineages retrieved from Bloodspot (see Materials and Methods). Color represents relative expression. (C) Representative image of an embryo (top and fetal liver (lower) of Mto1^fl/fl (fl/fl, hereafter WT, left) and Mto1^fl/fl;Vav-Cre (fl/fl;Vav, hereafter Mto1 KO, right) at E16.5. Scale bars, 1 mm. (D) Level of taurine modifications in mitochondrial tRNA (tm5U and tm5S2U modifications) in fetal liver Ter119⁺ erythroblasts from E16.5 WT (blue) and Mto1 KO (red) embryos determined by mass spectrometric analysis. Data normalized by the cytosolic tRNA modification (ms2t6A) were depicted (n = 4). (E) Absolute number of total fetal liver cells isolated from E16.5 WT (blue) and Mto1 KO (red) embryos (n = 6 from two independent experiments). (F) Peripheral blood parameters of E16.5 WT (blue) and Mto1 KO (red) embryos. RBC, red blood cell; WBC, white blood cell; PLT, platelet; HGB, hemoglobin; MCV, mean corpuscular volume; MCHC, mean corpuscular hemoglobin concentration (n = 7 from three independent experiments). N.D., not detected; ns, not significant; NK, natural killer; *P < 0.05, **P < 0.01, and ***P < 0.001.

To elucidate the biological function of Mto1 in hematopoiesis, we generated hematopoietic-specific Mto1 conditional KO mice by a crossing with Vav-Cre mice (fig. S1A) (16). Hematopoietic-specific Mto1 KO mice (Mto1^fl/fl;Vav-Cre, hereafter Mto1 KO) were embryonically lethal, while Mto1^fl/+;Vav-Cre heterozygous lived as long as their Vav-Cre–negative littermates [Mto1^fl/fl, hereafter wild-type (WT)] (fig. S1B). The Mto1 KO fetus had a paler appearance and its fetal liver was smaller and paler than the WT control, indicating severe anemia (Fig. 1C). Quantitative polymerase chain reaction (qPCR) confirmed that the KO efficiency of the Mto1 gene in the total fetal liver was more than 90% (fig. S1C). Consistently, the level of taurine modifications in mitochondrial tRNA of fetal liver Ter119⁺ erythroblasts were undetectable or severely impaired in Mto1 KO fetus as determined by mass spectrometric analysis (Fig. 1D). Total fetal liver cellularity was significantly reduced in Mto1 KO to approximately 50% of that in WT (Fig. 1E). A blood analysis of the Mto1 KO fetus revealed severe anemia along with marked abnormalities in erythrocyte-related parameters such as mean corpuscular volume (MCV) and mean corpuscular hemoglobin concentration (MCHC) and a milder reduction in white blood cell number (Fig. 1F). To elucidate the cause of anemia in the Mto1 KO fetus, we analyzed the immature hematopoietic cell population containing hematopoietic stem cell and progenitors (HSPCs) of the Mto1 KO fetal liver. A flow cytometric analysis revealed significant decreases in the lineage-negative (Lin⁻) and Lin⁻c-kit⁺ (LK) populations and slight changes in hematopoietic stem cells (HSCs) and multipotent progenitors (fig. S1, D to G) as well as the myeloid-committed progenitors (fig. S1, H and I) of the Mto1 KO fetal liver compared to those in WT control. In contrast, analysis of mature hematopoietic cell populations showed negligible changes in the Mto1 KO fetal liver, except for B cells (fig. S1, J and K). Collectively, these results suggested that the Mto1 deficiency resulted in severe anemia and mild reduction in HSPCs and B cells.

Mto1 KO induces defects in terminal erythroid proliferation and differentiation

Because severe anemia was observed in the Mto1 KO fetal liver, erythrocyte differentiation was further analyzed at the E16.5 stage when Mto1 expression was highest (Fig. 1A). The absolute number in Ter119⁺ erythroid progenitor cells was significantly lower in the Mto1 KO fetal liver than in WT (Fig. 2A). To better characterize which stage of erythroid differentiation was affected, we flow-cytometrically subfractionated fetal liver cells into erythroblast populations as previously reported (17). It was found that absolute cell number of erythroblasts from basophilic to orthochromatic erythroblast stages decreased significantly in Mto1 KO fetal liver with most severe reduction of polychromatic erythroblasts, while the number of proerythroblasts was maintained (Fig. 2, B and C). This was further confirmed by another strategy to differentiate erythroblast stages using Ter119 and CD71 (18) that showed erythroblast differentiation block at the early stage of differentiation in Mto1 KO fetal liver (fig. S2, A and B). To evaluate the terminal stage of erythropoiesis, we also investigated the enucleation status of erythroblasts and found that enucleation was impaired in Mto1 KO erythroblasts (fig. S2, C and D). These results indicated that Mto1 regulated the terminal erythroid differentiation.

Fig. 2. Mto1 KO induces defects in terminal erythroid proliferation and differentiation.

(A) Absolute number of the Ter119⁺ population in the fetal liver of E16.5 WT (blue) and Mto1 KO (red) embryos (n = 6 from two independent experiments). (B and C) Representative fluorescence-activated cell sorting (FACS) plots (B) and absolute number (C) of erythroid subpopulations in the fetal liver from E16.5 WT (blue) and Mto1 KO (red) embryos. ProE, proerythroblast; Baso, basophilic erythroblast; Poly, polychromatic erythroblast; Ortho, orthochromatic erythroblast; ret, Reticulocytes (n = 6 from two independent experiments). (D) mRNA expression of Mto1 in erythroid differentiation stages (proerythroblast to orthochromatic erythroblast) from the fetal liver of E16.5 WT (blue) and Mto1 KO (red) embryos (n = 3 from two independent experiments). **P < 0.01 and ***P < 0.001.

Consistent with the substantial reduction of Mto1 KO polychromatic erythroblast, we found that Mto1 mRNA expression was up-regulated at highest level at the polychromatic erythroblast stage of the WT fetal liver (Fig. 2D). Moreover, a tRNA modification analysis by mass spectrometry (MS) showed the highest taurine modification levels at the polychromatic erythroblast stage (fig. S2E). We further compared the transcriptome profile of the polychromatic erythroblasts between the Mto1 KO and WT fetal liver by RNA sequencing. The principal components analysis showed a clear separation of Mto1 KO from WT (fig. S2F), followed by a Gene Ontology (GO) analysis that showed the negative enrichment of DNA replication- and cell cycle–related GO terms and positive enrichment of cytoplasmic translation–related GO terms in Mto1 KO (fig. S2G), implying impaired cell proliferation. We also confirmed the down-regulated expression of the erythropoiesis master regulator, Gata-1 at both the mRNA and protein levels in Mto1 KO polychromatic erythroblasts (fig. S2, H and I).

Collectively, severely impaired erythroid proliferation and differentiation in Mto1 KO fetal liver coincided with the Mto1 mRNA expression and its mediated tRNA taurine modifications during erythroid differentiation, which was highest at the polychromatic erythroblast stage.

Mto1 KO mice show niche-independent and cell-intrinsic erythroid defects

Interactions with the hepatic niche have important roles in fetal hematopoiesis (19, 20). To test whether Mto1 KO affects hepatic niche cells, we flow-cytometrically analyzed stromal and endothelial cell populations in Mto1 KO fetal liver–derived cells and found no significant changes in both populations (Fig. 3, A and B). Macrophages are also key niche components to support and promote erythropoiesis through the enucleation process in erythroblast islands (21). Absolute number of total macrophages (F4/80⁺), tissue-resident (F4/80⁺Ly6c⁻), and inflammatory (F4/80⁺Ly6c⁺) macrophages was unchanged in Mto1 KO fetal liver compared to control (Fig. 3, C and D, and fig. S3A). These results indicated that erythroid defects in the Mto1 KO fetal liver were independent of hepatic niche.

Fig. 3. Mto1 KO mice show niche-independent and cell-intrinsic erythroid defects.

(A and B) Representative FACS plots (A) and absolute number (B) of niche cell population in the WT (blue) and Mto1 KO (red) fetal liver at E16.5 (n = 3 from two independent experiments). EC, endothelial cell; SC, stromal cells. (C and D) Representative FACS plots (C) and absolute number (D) of macrophage populations in the WT (blue) and Mto1 KO (red) fetal liver at E16.5 (n = 3 from two independent experiments). iMac, inflammatory macrophage; tMac, tissue resident macrophage. (E) Experimental scheme of the erythroblast island culture in vitro. (F) Absolute number of erythroblast islands per fetal liver from E16.5 WT (blue) and Mto1 KO (red) embryos (n = 3 from two independent experiments). (G) Representative FACS plots of the in vitro erythroid differentiation of erythroblast islands from the fetal liver from E16.5 WT (upper) and Mto1 KO (lower) embryos. Left and right panels show 2 and 8 days after the in vitro culture, respectively. (H) Time course of erythroblast differentiation during the in vitro erythroblast island culture from WT (blue) and Mto1 KO (red) fetal livers at E16.5 (n = 3 from two independent experiments). ns, not significant; *P < 0.05, **P < 0.01, and ***P < 0.001.

To further confirm the niche-independent erythroid defect caused by Mto1 deficiency, we isolated erythroblast islands from Mto1 KO and WT fetal livers and cultured them in vitro for 8 days according to previously established methods (Fig. 3E and fig. S3B) (22, 23). The in vitro erythroblast island culture revealed that the number of erythroblast islands was significantly lower in Mto1 KO than in WT (Fig. 3F), and the transition from basophilic to polychromatic erythroblasts was impaired, leading to the accumulation of basophilic erythroblasts on day 8 (Fig. 3, G and H). Similar results were found by erythroid differentiation culture of lineage negative cells (fig. S3, C and D) (24). These indicate defect in both cell differentiation in basophilic stage and cell proliferation in polychromatic stage, as suggested by the transcriptome analysis (fig. S2, G to I). These results suggested that Mto1 played an indispensable role in fetal erythropoiesis in a cell-intrinsic manner.

Mto1 KO induces OXPHOS complex defects and cytosolic iron accumulation

MTO1 modifies position 34 U of mt-tRNA with taurine, which stabilizes codon-anticodon interactions and promotes translation of 13 mt-DNA–encoded genes. The mt-DNA–coded proteins are incorporated into the oxidative phosphorylation (OXPHOS) complexes I, III, IV, and V (Fig. 4A). Metabolic labeling of mt-DNA-coded proteins showed that overall mitochondrial translation was significantly suppressed in Mto1 KO fetal liver erythroid progenitor cells (Ter119⁺) compared to WT (Fig. 4B), raising the possibility that some of OXPHOS complex might be defective. To check the formation of OXPHOS complex, we performed blue native polyacrylamide gel electrophoresis (PAGE) followed by immunoblotting and found significantly less formation of these OXPHOS complexes (around 50% reduction) in Ter119⁺ Mto1 KO cells than in WT control, whereas complex II, all subunits of which are coded by nuclear DNA, remained unaffected (Fig. 4C).

Fig. 4. Mto1 KO induces OXPHOS complex defects and cytosolic iron accumulation.

(A) A list of mitochondrial DNA-coded proteins. OXPHOS complexes corresponding to each protein are listed in the right column. (B) Metabolic labeling of mitochondrial translation in Ter119⁺ erythroblasts in E16.5 WT and Mto1 KO fetal livers. Representative FACS histogram (left) and relative geometric mean fluorescence intensity (GeoMFI, right) of mitochondrial translation was shown (n = 3 from two independent experiments). (C) Blue native PAGE analysis of OXPHOS complex proteins in the total protein lysate of Ter119⁺ erythroblasts in E16.5 WT and Mto1 KO fetal livers. Right: Densitometry data of Mto1 KO samples normalized to WT samples. Dashed line indicates the density of WT samples (n = 3). (D) Oxygen consumption rate (OCR) in polychromatic erythroblasts sorted from the E16.5 fetal livers of WT and Mto1 KO embryos (n = 3). (E) Representative FACS histogram (left) and relative geometric mean fluorescence intensity (GeoMFI, right) of mitochondrial membrane potential in polychromatic erythroblasts from the E16.5 fetal livers of WT and Mto1 KO embryos. Mitochondrial membrane potential was measured by MitoProbe, JC-1 Red (n = 5 to 6 from two independent experiments). (F) Representative FACS histogram (left) and mean fluorescence intensity (MFI, right) of mitochondrial iron levels in polychromatic erythroblasts from E16.5 WT or Mto1 KO fetal liver. Mitochondrial iron measured by Mito-FerroGreen (n = 3 from two independent experiments). (G) Cytosolic iron levels in Ter119⁺ erythroblasts from E16.5 WT or Mto1 KO fetal liver measured by ICP-MS. Amounts of iron within a sample were normalized by total amount of protein. (n = 3 to 4). (H) Transmission electron microscope analysis of polychromatic erythroblasts from the E16.5 fetal livers of WT and Mto1 KO embryos. Iron-containing siderosomes are shown in the magnified pictures. Scale bars, 500 nm. Siderosome areas per cell were measured between WT and Mto1 KO embryos (n = 30). *P < 0.05, **P < 0.01, and ***P < 0.001.

Consistent with impaired OXPHOS complex formation in Mto1 KO cells, maximum respiration rate, mitochondrial membrane potential, and mitochondrial reactive oxygen species (ROS) levels in Mto1 KO polychromatic erythroblasts were significantly suppressed compared to those in WT cells despite mitochondrial mass was normal in Mto1 KO polychromatic erythroblasts (Fig. 4, D and E, and fig. S4, A and B). Thus, Mto1 deficiency induced impaired OXPHOS complex formation and subsequently suppressed OXPHOS activity in polychromatic erythroblasts.

To further validate the pivotal role of the OXPHOS complex formation in fetal erythropoiesis, another OXPHOS complex I–deficient mouse model, Ndufs4 KO mouse (25) was analyzed. Ndufs4 KO fetal liver–derived Ter119⁺ erythroblasts showed the impaired formation of only OXPHOS complex I but not other complexes (fig. S4C). Like Mto1 KO mouse, Ndufs4 KO mouse also showed significantly impaired terminal erythroid differentiation in fetal liver, but much milder than Mto1 KO mouse (fig. S4, D to F). No significant change in the mitochondrial membrane potential was observed in Ndufs4 KO fetal liver (fig. S4G).

Multiple forms of iron are incorporated in OXPHOS complex for executing electron transfer. Iron-sulfur clusters are held by complexes I, II, and III while different types of heme by complexes II, III, and IV (26). We hypothesized that impaired formation of OXPHOS complexes I, III and IV, caused by Mto1 deficiency, would alter iron localization in mitochondria and eventually affect intracellular iron distribution between mitochondria and cytosol. To test this hypothesis, the iron level in the mitochondria in Ter119⁺ erythroblast cells was quantified by flow cytometry with a specific iron probe, while intracellular iron levels was quantified by inductively coupled plasma MS (ICP-MS) following cytosol isolation from the erythroblast cells. Those analysis revealed a significant reduction of mitochondrial iron, in contrast to a significant elevation of cytosolic iron in Mto1 KO fetal liver erythroblasts (Fig. 4, F and G). Consistently, transmission electron microscope analysis revealed that Mto1 KO fetal liver polychromatic erythroblasts contained significantly larger siderosomes, iron-containing vesicles in the cytosol than WT control, while the percentage of siderosome-containing cells was comparable (Fig. 4H and fig. S4H). The extent of cytosolic iron accumulation was relatively milder in Ndufs4 KO polychromatic erythroblasts, in that siderosome size was comparable to WT control, whereas the percentage of siderosome-containing cells was slightly higher (fig. S4, I and J). These results indicated likelihood that the extent of cytosolic iron accumulation involves in impaired fetal erythropoiesis.

Collectively, these results indicated that Mto1 deficiency induced inefficient translation and formation of OXPHOS complexes in polychromatic erythroblasts and resulted in less mitochondrial and more cytosolic iron, leading to dysregulation of erythroid proliferation and differentiation.

Up-regulated heme biosynthesis correlates with Mto1 KO-derived cytosolic iron overload

Cytosolic iron levels are sensed by the RNA-binding protein, iron regulatory protein (IRP), which interacts with cis-regulatory hairpin structures, known as iron-responsive elements (IREs) in specific target mRNAs, and regulates protein translation (27). To investigate whether cytosolic iron accumulation facilitates the IRE-related protein translation, the proteome analysis of WT and Mto1 KO polychromatic erythroblasts was performed. Consistent with impaired OXPHOS complex formation in Mto1 KO erythroblasts, a Metascape analysis (28) of differentially expressed proteins showed significant changes in oxidoreductase-related proteins (−Log₁₀ P value = 10.96) (fig. S5). The expression of IRE target proteins showed a similar profile to that under iron-replete conditions (27); the significantly higher expression of ferritin protein [ferritin light chain (FTL)] and 5′-aminolevulinate synthase 2 (ALAS2) and significantly lower expression of transferrin receptors (TFRC) than in WT (Fig. 5A). ALAS2 protein up-regulation in Mto1 KO erythroblasts was further confirmed by antibody-based immunoblotting (Fig. 5B). These data indicated that the IRPs sensed the increased cytosolic iron and regulated the iron-related protein translation via IREs. The up-regulated expression of the ALAS2 protein, an initial enzyme of the heme biosynthesis pathway, is expected to lead to active heme biosynthesis. Another MS analysis confirmed that intracellular heme (hemin) content was significantly higher in Mto1 KO polychromatic erythroblasts by 30% (Fig. 5C). Although this analysis is not per se direct measurement of the heme biosynthesis, the data strongly suggested that heme biosynthesis was substantially up-regulated in Mto1 KO polychromatic erythroblasts. Metabolome analysis revealed a significant change in the metabolic pathway of porphyrin (eight highest ranked metabolism, P value < 0.05), which is an intermediate of heme biosynthesis (Fig. 5D). Porphyrin plays an essential role in heme biosynthesis because porphyrin is commonly used as an intermediate in heme biosynthesis pathway across species from bacteria to higher vertebrates (29). These data validated the cytosolic iron overload derived from OXPHOS defect in Mto1 KO cells that, in turn, induced the up-regulation of iron-heme biosynthesis.

Fig. 5. Upregulated heme biosynthesis correlates with Mto1 KO-derived cytosolic iron overload.

(A) Proteome data of heme synthesis- and heme metabolism-related proteins in polychromatic erythroblasts from E16.5 WT (blue) and Mto1 KO (red) fetal livers (n = 3). (B) Western blot analysis of ALAS2 protein in the total protein lysate of Ter119⁺ erythroblasts in E16.5 WT and Mto1 KO fetal livers. (C) Intracellular hemin concentration in polychromatic erythroblasts (1 × 10⁶ cells) from the E16.5 fetal livers of WT (blue) and Mto1 KO (red) embryos (n = 4). (D) Metabolite set enrichment analysis (MSEA) of polychromatic erythroblasts from E16.5 WT and Mto1 KO fetal livers. The top 25 enriched metabolite sets are shown. *P < 0.05 and **P < 0.01.

UPR is responsible for the Mto1 KO-mediated erythroid defects

Heme has been reported to induce UPR via up-regulation of Xbp1 splicing (30). Because heme expression was up-regulated in Mto1 KO erythroblasts, we speculated that Mto1 deficiency might induce UPR through heme overproduction. Gene expression analyses revealed the up-regulation of mitochondrial UPR related gene (Chop) (31) and cytosolic UPR related genes [Atf4, the spliced form of Xbp1 (Xbp1-s) and Atf6] (32) in Mto1 KO polychromatic erythroblasts (Fig. 6A). Considering posttranslational regulation, up-regulation of ATF4 and XBP1-s proteins in Mto1 KO erythroblasts was confirmed by immunoblotting (Fig. 6B). The ratio of Xbp1-s to total Xbp1 (Xbp1-t) mRNA expression was also up-regulated in Mto1 KO erythroblasts and become more pronounced at the orthochromatic erythroblast stage (Fig. 6C and table S1). UPR induction in Mto1 KO polychromatic erythroblasts is also supported by negative enrichment of cell cycle–related genes (fig. S2G) because UPR was shown to inhibit cell cycle progression (33). As a consequence, apoptosis was significantly induced in Mto1 KO polychromatic erythroblasts (fig. S6, A and B). Although iron accumulation is known to cause ferroptosis, lipid peroxidation, a hallmark of ferroptosis, was similar between WT and Mto1 KO erythroblasts, and terminal erythroid differentiation was not rescued by a selective inhibitor of ferroptosis, ferrostatin-1 (Fer-1), in the erythroblast island culture in Mto1 KO mouse (fig. S6, C and D), suggesting that apoptosis but not ferroptosis partly explains the erythroid differentiation block in Mto1 KO fetal liver besides suppressed cell proliferation.

Fig. 6. UPR is responsible for the Mto1 KO-mediated erythroid defects.

(A) mRNA expression of UPR-related genes in fetal liver polychromatic erythroblasts from E16.5 WT (blue) and Mto1 KO (red) embryos (n = 6 from two independent experiments). (B) Western blot analysis of ATF4 and XBP1-s proteins in the total protein lysate of Ter119⁺ erythroblasts in E16.5 WT (left) and Mto1 KO (right) fetal livers. (C) Ratio of Xbp1-s to Xbp1-t mRNA expression in different erythroid stages during terminal erythroid differentiation in the fetal liver from E16.5 WT (blue) and Mto1 KO (red) embryos. Relative value to WT is shown (n = 5 to 6 from two independent experiments). (D) Representative FACS plots of E16.5 WT (upper) and Mto1 KO (lower) fetal liver-derived erythroblast islands 3 days after the in vitro culture without (Ctrl, left) or with an IRE1α kinase inhibitor (Kira-6, right). (E to G) Percentage of Ter119⁺ cells (E), erythroblast subpopulations (F), and mRNA expression of Gata-1 (G) in the polychromatic erythroblasts of E16.5 WT and Mto1 KO fetal liver–derived erythroblast island 3 days after the in vitro culture without (Ctrl) or with Kira-6 (n = 3 from two independent experiments). *P < 0.05, **P < 0.01, and ***P < 0.001.

To confirm whether the IRE1α-XBP1 signaling pathway induces terminal erythroid differentiation block, WT erythroblast islands were in vitro treated with the UPR inducer thapsigargin, which is known to preferentially induce UPR through the IRE1α-XBP1 signaling pathway (34). The treatment with thapsigargin significantly reduced the Ter119⁺ population and impaired terminal erythroid differentiation (fig. S6, E and F), accompanied by the down-regulated expression of the Gata-1 gene and up-regulated expression of the Xbp1-s gene in treated and isolated WT polychromatic erythroblasts (fig. S6, G and H). UPR can be induced by several cellular stresses such as high level of cytosolic calcium, which is also elevated with thapsigargin treatment (35). To rule out the involvement of calcium-induced UPR in Mto1 KO mouse, we compared intracellular calcium level and found no significant difference between Mto1 KO and WT mice (fig. S6I). Next, we investigated whether the inhibition of the IRE1α-XBP1 signaling pathway rescued the fetal erythroid defect in Mto1 KO mice. When Mto1 KO and WT erythroblast islands were cultured with Kira-6 (36), an IRE1α kinase inhibitor, the down-regulated expression of the Xbp1-s gene was confirmed in Mto1 KO cells (fig. S6J). The down-regulation of ATF protein expression was observed as previously reported (37, 38), whereas ALAS2 protein expression was not affected, indicating that Kira-6 treatment did not alter the cytosolic iron level (fig. S6K). Under these conditions, the number of Ter119⁺ Mto1 KO cells and downstream erythroblast populations increased to normal levels (Fig. 6, D to F). Conversely, Gata-1 gene expression was significantly up-regulated in Mto1 KO erythroblasts treated with Kira-6 (Fig. 6G). Collectively, these results indicated that heme-induced UPR through the IRE1α-XBP1 signaling pathway is responsible for the fetal erythroid defects in Mto1 KO mice, resulting in cell cycle suppression and induction of apoptosis.

Mto1 KO-mediated erythroid defects are rescued by iron chelation

Last, to examine the causal relationship of imbalanced iron localization with fetal erythroblast terminal differentiation, erythroblast islands from Mto1 KO and WT fetal livers were cultured with excess iron. The excess iron treatment significantly decreased the Ter119⁺ population and Gata-1 gene expression in Mto1 KO and WT fetal livers (fig. S7, A to C). Conversely, Xbp1-s gene expression was induced in WT cells by the excess iron treatment and was not observed in Mto1 KO cells (fig. S7D). This iron overloading–induced Xbp1-s expression was confirmed at the protein level using ERAI mice (39), which express the XBP1-s–venus fusion protein upon UPR (Fig. 7A).

Fig. 7. Mto1 KO-mediated erythroid defects are rescued by iron chelation.

(A) Representative FACS histogram and geometric mean fluorescence intensity (GeoMFI) of the spliced form of the XBP1 (XBP1-s)–venus fusion protein in the polychromatic erythroblasts of E16.5 ERAI mouse fetal liver-derived erythroblast islands 3 days after the in vitro culture without (Ctrl, dark blue) or with ferric ammonium citrate (Fe, light blue) (n = 6 from two independent experiments). (B) Representative FACS plots of E16.5 WT (upper) and Mto1 KO (lower) fetal liver-derived erythroblast islands 3 days after the in vitro culture without (Ctrl, left) or with DFO (right). (C) Percentage of Ter119⁺ cells in the polychromatic erythroblasts of E16.5 WT and Mto1 KO fetal liver-derived erythroblast island 3 days after the in vitro culture without (Ctrl) or with DFO (n = 3 from two independent experiments). (D) Percentage of erythroblast subpopulations in the polychromatic erythroblasts of E16.5 WT and Mto1 KO fetal liver–derived erythroblast islands 3 days after the in vitro culture without (Ctrl) or with DFO (n = 3). (E and F) mRNA expression levels of Gata-1 (E) and Xbp1-s (F) in the polychromatic erythroblasts of E16.5 WT and Mto1 KO fetal liver–derived erythroblast island 3 days after the in vitro culture without (Ctrl) or with DFO (n = 3 from two independent experiments). (G) Molecular mechanism underlying erythroblast terminal differentiation defects in Mto1 KO mice. In WT polychromatic erythroblasts, OXPHOS complexes are intact and intracellular iron distribution is balanced, which results in normal heme biosynthesis. In the Mto1-deficient state, the lack of taurine modifications of mitochondrial tRNA impairs the formation of OXPHOS complexes I, III, IV, and V, associated with an attenuated OXPHOS activity and cytosolic iron overload, which results in up-regulation of ALAS2 protein expression. Elevated ALAS2 enhances heme biosynthesis an excess of which induces Xbp-1 splicing pathway-mediated UPR at the polychromatic erythroblast stage in fetal hematopoiesis. *P < 0.05, **P < 0.01, and ***P < 0.001.

Furthermore, erythroblast islands from Mto1 KO and WT fetal livers were cultured with the iron chelator, deferoxamine (DFO). Efficacy of iron chelation by DFO was confirmed by the downregulation of ALAS2 protein expression (fig. S7E). Under the culture condition with DFO, terminal erythroid differentiation in the Mto1 KO fetal liver was partially rescued, whereas only a slight effect was observed in the WT fetal liver (Fig. 7, B to D). In line, Gata-1 gene expression also increased in DFO-treated Mto1 KO fetal liver cells (Fig. 7E), whereas Xbp1-s gene expression (Fig. 7F) was significantly suppressed.

Iron overloading directly induced UPR in ex vivo erythroblast islands, whereas Mto1-mediated erythroid defects was partially rescued by iron chelation. This finding suggested that impaired iron homeostasis causes IRE1α-XBP1 signaling pathway-mediated UPR in Mto1 KO erythroblasts and regulates cell proliferation and differentiation.

DISCUSSION

The primary target proteins affected by an mt-tRNA modification enzyme deficiency are the components of OXPHOS complexes, namely, mt-DNA–coded proteins (40, 41). Consistent with this finding, Mto1 KO erythroblasts showed reduced translation of OXPHOS complexes I, III, IV, and V. Because of their iron-rich structures (26), OXPHOS complex defects were followed by the marked imbalanced intracellular distribution of iron, i.e., decreased mitochondrial and increased cytosolic iron, which resulted in the up-regulation of heme biosynthesis in Mto1 KO mice. As it was shown that heme induced cytosolic UPR via ROS generation (30, 42), elevated heme in turn induced cytosolic UPR (Fig. 7G).

Our findings suggested the OXPHOS complex defects as the primary causes of the inefficient erythroid production in Mto1 KO mouse. To support this hypothesis, we have analyzed the OXPHOS complex I–deficient mouse model, Ndufs4 KO mouse (25), which lacks Ndufs4 gene encoding an 18-kDa protein, one of the 45 subunits composing OXPHOS complex I, and thereby show a Leigh-like phenotype in humans (25). Ndufs4 KO mouse partially phenocopied erythroid deficiency found in Mto1 KO mouse, albeit to less extend (fig. S4, C to F). However, it is of note that no significant change in the mitochondrial membrane potential was observed at all in Ndufs4 KO fetal liver (fig. S4G), suggesting that Mto1 deficiency–caused erythroid defects are mediated by other factors rather than attenuated OXPHOS activity. The genotype-phenotype comparison suggested the degree of impaired fetal erythropoiesis correlates with the extent of defect in OXPHOS complexes, but not with mitochondrial activity; four of five OXPHOS complexes were affected in Mto1 KO cells, whereas only one complex was affected in Ndufs4 KO cells (Fig. 4C and fig. S4C). Because multiple forms of iron are incorporated in OXPHOS complexes (26), Mto1 KO-induced OXPHOS complex defects lead to the alteration of the intracellular iron distribution (Fig. 4, F to H), which is more notable than that in Ndufs4 KO cells (fig. S4, I and J). These results indicated that erythroid-deficient phenotype is attributed to the overloaded cytosolic iron. The reduction of mitochondrial iron and the elevation of cytosolic iron in Mto1 KO cells (Fig. 4, F and G) suggest the intracellular iron transmigration from mitochondria to cytosol. Yet, the possibility of the up-regulated iron uptake in Mto1 KO cells cannot be excluded because the extent of cytosolic iron increase is higher than that of mitochondrial iron decrease. Our observation is in line with previous findings showing that Ndufs4 KO mouse displayed signs of iron overload such as increased expression of hepcidin and iron-replete pattern of IRE target protein expression in the liver (43).

Moreover, Mto1 played an indispensable role in fetal erythropoiesis in hepatic niche-independent and cell-intrinsic manner (Fig. 3). Our results also propose the pivotal role of iron homeostasis in this mechanism, given erythrocytes use abundant iron for their differentiation through heme-hemoglobin biosynthesis (4). Although we propose the UPR induced by excess heme as a cause of anemia in this present study, we cannot exclude the possibility that the oxidative stress induced by redox imbalances of increased cytosolic iron is a potential cause of UPR and anemia. Moreover, it is also still possible that the proteostatic stress caused by imbalance of mitochondrial and cytoplasmic translation contributes to them, as suggested in the previous reports (12, 44). The erythroid defect was not fully rescued by DFO treatment in Mto1 KO mouse (Fig. 7, B to F).

Historically, mitochondria have been extensively examined due to their role in energy metabolism. Because mature erythrocytes function as oxygen suppliers and lose their mitochondria during their maturation process (3), the role of mitochondria in erythropoiesis has not been investigated in great detail. Emerging research has identified multifaceted roles of mitochondria, not only as the “bioenergetic powerhouse” but also as the “biosynthetic hub” of cells (45). For example, mitochondrial citrate supplies carbon for the synthesis of fatty acids, cholesterol, and ketone bodies (46), which, in turn, affect epigenetics. In this line, ketone bodies synthesized from mitochondrial citrate epigenetically regulated erythroid genes through their histone deacetylase activity in Tfam KO mouse (47). Because TFAM is an essential transcription factor for transcription and replication of mitochondrial genome, Mto1 KO mouse might share some molecular mechanisms in erythroid defect with Tfam KO mouse. The metabolome analysis of Mto1 KO cells showed the difference in amino acid or nucleotide metabolism (Fig. 5C), which are also found in Tfam KO mouse (47). However, here, we propose a molecular mechanism that links primary mitochondrial protein deficiency and erythroid defects via spatial changes in intracellular iron homeostasis. This mechanism is not necessarily specific to the Mto1 KO but can be generalized to other mitochondrial respiratory chain dysfunction models, such as Tfam KO (47) and Uqcrfs1 KO (48) mouse. It might be interesting to study the role of iron in other models in future. This unique molecular mechanism may provide a better understanding of the pathophysiology of mitochondrial diseases and contribute to the development of novel therapy by providing a previously unexplored point of view in this research field.

In human cases, most of the patients with MTO1 gene mutation suffered from metabolic and cardiac disorders such as lactic acidosis and hypertrophic cardiomyopathy (13, 49, 50). Consistently, we previously reported that heart-specific Mto1 KO mouse died shortly after birth (12) and another group also reported cardiomyopathy caused by Mto1 deficiency (51). On the other hand, hematopoietic symptoms have been reported in only one case so far (13). This is likely that it would be difficult to diagnose anemia in those patients with MTO1 gene mutation because they die very early in their life due to severe organ failures such as cardiovascular and metabolic diseases. Although the direct relevance of the present study for the understanding of the pathophysiology in human cases is limited for now, our study sheds the light on the important role of Mto1 in hematopoiesis, more cases of which would be found in future.

In conclusion, we here identified an indispensable role of mt-tRNA taurine modifications in erythropoiesis in a cell-intrinsic manner. Mechanistically, OXPHOS complex defects caused by an Mto1 deficiency led to alterations in the iron-heme biosynthesis axis, which triggers UPR and subsequently, impair erythroid proliferation and differentiation. This molecular mechanism will contribute to the understanding of the pathophysiology of mitochondrial diseases.

MATERIALS AND METHODS

Database search

Mto1 gene expression data at different developmental stages and in hematopoietic lineages were retrieved from Expression Atlas, FANTOM5 project, Riken (www.ebi.ac.uk/gxa/home) and Bloodspot (www.fobinf.com/?dataset=nl_mouse_data).

Mice

C57BL/6 (CD45.2⁺) and B6.SJL (CD45.1⁺) mice, hereafter referred to as WT, were purchased from Japan SLC (Hamamatsu, Japan) and the Jackson Laboratories (Bar Harbor, ME), respectively. Mto1^fl/fl mice in which exons 1 and 2 of the Mto1 gene were floxed by the LoxP sequence (12) were crossed with Vav-Cre^+/− mice (16) (purchased from the Jackson Laboratories). Ndufs4^+/− mice (25) were purchased from the Jackson Laboratories. ERAI mice (RBRC01099) (39) were provided by RIKEN BRC through the National BioResource Project of MEXT/AMED, Japan. All mice were maintained at the Center for Animal Resources and Development at Kumamoto University. All experiments were approved by the Animal Care and Use Committee of Kumamoto University (study approval number: A2023-025).

Quantitative RT-PCR

Total RNA was extracted and purified with the RNeasy Kit (QIAGEN, Hilden, Germany). Total RNA was then reverse transcribed to cDNA using the PrimeScript RT Master Mix (Takara Bio Inc., Kusatsu, Japan) following the manufacturer’s instructions. A quantitative real-time PCR (qRT-PCR) was performed using SYBR Green Master Mix (Thunderbird qPCR Mix, Toyobo Life Sciences, Osaka, Japan). qPCR was run on the LightCycler-96 real time PCR instrument (Roche Life Science, Penzberg, Germany). Actb gene was used as an internal control. Primer sequences are shown in table S2.

Mass spectrometric analysis of tRNA modifications

Total RNA was isolated from sorted erythroblast (Ter119⁺; ProE; Baso; Poly; Ortho) cells using TRIzol (Thermo Fisher Scientific, Waltham, MA) after homogenization with a TissueRuptor II (QIAGEN) equipped with a disposable probe (QIAGEN). Extracted RNA was enzymatically digested as previously described (52), with slight modifications. Five micrograms of RNA was digested with 0.5 U of Nuclease P1 (Fujifilm Wako Pure Chemical Corp., Osaka, Japan) and 0.05 U of alkaline phosphatase (Takara Bio Inc., 2120A) in 5 mM ammonium acetate (pH 5.3) and 20 mM Hepes-KOH (pH 7.0) at 37°C for 24 hours. Samples were subjected to a mass spectrometric analysis, as previously described (53, 54).

Hematological parameter analysis

Peripheral blood (PB) was obtained by left ventricle puncture from adult mice and by decapitation from fetuses. Hematological parameters were assessed using the hematology analyzer, Celltac α MEK-6358 (Nihon Kohden, Tokyo, Japan).

FACS analysis and cell sorting

All antibodies used in the present study were purchased from BioLegend (San Diego, CA) unless otherwise stated. In the mature cell analysis, cells were pre-incubated with a purified anti-CD16/32 antibody (93) to block FcγR, followed by staining with antibodies against B220 (RA3-6B2), CD3ε (145-2C11), F4/80 (BM8), Ly6G (1A8), Ly6C (HK1.4), and CD11b (M1/70). In the early hematopoietic cell analysis, cells were incubated with the following biotinylated antibodies against Lin markers: NK1.1 (PK136), CD11b (M1/70), Ter119 (Ter119), Gr-1 (RB6-8C5), CD4 (GK1.5), CD8ε (53–6.7), CD3ε (145-2C11), B220 (RA3-6B2), and IL-7Rα (SB/199); and the following fluorescence-conjugated antibodies: c-Kit (2B8), Sca-1 (D7), CD34 (HM34), Flt3 (A2F10), CD150 (TC15-12F12.2), CD48 (HM48-1), and CD16/32 (93). In common lymphoid progenitor (CLP) staining, IL-7Rα (A7R34) was excluded from Lin markers and stained separately to define CLP. In the erythroid population analysis and sorting, the primary antibodies used were as follows: Ter119 (Ter119, BD Biosciences, Franklin Lakes, NJ), CD71 (C2, BD Biosciences), and CD44 (IM7, BD Biosciences). Stained cells were analyzed and sorted using FACSCanto II and FACSAria III flow cytometers, respectively (BD Biosciences). All data were analyzed using FlowJo (BD Biosciences).

RNA sequencing

One hundred polychromatic erythroblasts were sorted with flow cytometry and were subjected to the RamDA method (55). Briefly, the first strand of cDNA was synthesized with the PrimeScript RT Reagent Kit (Takara Bio Inc.) and NSR (not-so random) primers. The second strand was synthesized with Klenow Fragments (3′-5′ exo-; New England Biolabs, Ipswich, MA) and complement chains of NSR primers. The double-stranded cDNA was purified with AMPure XP beads (Beckman Coulter, Brea, CA) and subjected to library preparation using the Nextera XT DNA sample Prep kit (Illumina, San Diego, CA). The quantity and quality of the isolated cDNA library were determined with Agilent 4150 TapeStation system (Agilent Technologies, Santa Clara, CA), and the library was sequenced on the NovaSeq X system (Illumina). Quality check and trimming of single-end sequences were completed with the trim_galore (version 0.4.3) package. Quality and length parameters were quality 30 and length 30. Filtered sequences were processed to remove mitochondrial mRNA with the SortMeRNA and were aligned to mouse reference sequences (GRCm38, M20 version) from (www.gencodegenes.org/mouse/release_M20.html) with ultrafast RNA-seq aligner STAR (version 2.7.0) (56). All the aligned bam files were used with mouse GFF annotation files (GRCm38, M20 version) as input into the featureCounts program from the Subreads program (version 2.0.1) to count the raw reads for each gene and sample and to create a gene count matrixDES. To calculate differentially expressed genes, the DESeq2 (version 1.30.1) package was used in R. For genes to be considered significantly and differentially expressed, α = 0.05 was used. GO analysis of differentially expressed genes (log₂ fold change > 1, P < 0.01) was performed using DAVID (https://davidbioinformatics.nih.gov/) (57, 58) to generate a network enrichment via GO Processes.

Western blot analysis

In total, 2 × 10⁴ polychromatic erythroblasts were sorted with flow cytometry or 1 × 10⁶ Ter119⁺ erythroblasts were sorted by magnetic beads selection using biotinylated antibodies against Ter119 (BioLegend) and streptavidin microbeads (Miltenyi Biotec) and lysed in 10 μl of Laemmli sample buffer (Bio-Rad laboratories, Hercules, CA). Proteins were separated on 4 to 20% gradient TGX precast gels (Bio-Rad laboratories) and transferred to polyvinylidene difluoride (PVDF) membranes. Transferred membranes were probed with the primary antibody at 4°C overnight, followed by an incubation with the secondary antibody at room temperature for 1 hour. The protein signal was detected with Pierce ECL Western Blotting substrate (Thermo Fisher Scientific). The following antibodies were used: a rabbit monoclonal antibody to GATA-1 (#3535, Cell Signaling Technology, Danvers, MA), a rabbit monoclonal antibody to β-actin (#4970, Cell Signaling Technology), a rabbit monoclonal antibody to ALAS2 (#ab184964, Abcam, Cambridge, UK), a rabbit monoclonal antibody to ATF4 (#11815, Cell Signaling Technology), a rabbit monoclonal antibody to XBP-1 s (#40435, Cell Signaling Technology), and a secondary anti-rabbit immunoglobulin G (IgG), horseradish peroxidase (HRP)–linked antibody (#7074, Cell Signaling Technology). Total loaded protein on the membrane was stained with Pierce Reversible Protein Stain Kit for PVDF membranes (Thermo Fisher Scientific). Intensities of the bands were quantified using ImageJ software (https://imagej.net/ij/).

In vitro erythroid differentiation

Erythroblast islands were isolated from the fetal liver at E16.5. Briefly, the fetal liver was minced in liver digest medium (Thermo Fisher Scientific), the suspension was slowly loaded onto 10 mL RPMI 1640 medium with 30% fetal calf serum (FCS), and erythroblast islands were allowed to settle by gravity at room temperature for 60 min. Settled erythroblast islands were seeded on a six-well plate with erythroid differentiation medium [RPMI 1640 medium (Sigma-Aldrich) supplemented with 15% FCS, erythropoietin (EPO) (0.2 U/ml) and holo-transferrin (4 μg/ml; Sigma-Aldrich)] and incubated for the indicated number of days. The following reagents were added to the culture:10 μM Fer-1 (Selleck Chemicals, Houston, TX), 4 μM Kira-6 (Selleck Chemicals), 2 μM thapsigargin (Selleck Chemicals), 45 μM DFO (Abcam), and 1 mM ferric ammonium citrate (Sigma-Aldrich, St. Louis, MO). At the end of differentiation, cells were collected and analyzed or sorted by flow cytometry or magnetic beads selection for subsequent analyses.

Lineage-negative cells derived from the fetal liver at E16.5 or adult bone marrow (BM) were cultured in vitro for erythroid differentiation following a protocol described in detail previously (24) with slight modifications. Briefly, lineage-negative cells were isolated by magnetic depletion of lineage-positive cells using above-mentioned antibodies against lineage markers and streptavidin microbeads (Miltenyi Biotec, Bergisch Gladbach, Germany). Isolated lineage-negative cells were cultured in erythroid differentiation medium [Iscove modified Dulbecco’s medium (Sigma-Aldrich) supplemented with 15% FCS, 1% bovine serum albumin (Sigma-Aldrich), holo-transferrin (500 μg/ml; Sigma-Aldrich), EPO (0.5 U/ml), recombinant human insulin (10 μg/ml; Sigma-Aldrich), and 2 mM GlutaMAX supplement (Thermo Fisher Scientific)] at 37°C for 2 days. At the end of differentiation, cells were collected and analyzed by flow cytometry.

Mitochondrial translation

The labeling of mitochondrial protein synthesis was examined using Click-iT Metabolic Labeling Reagents (Thermo Fisher Scientific) following the manufacturer’s instructions. Ter119⁺ erythroblasts were sorted by magnetic beads selection using anti-Ter-119 microbeads (Miltenyi Biotec) and seeded in methionine-free RPMI medium (Thermo Fisher Scientific) with 10% dialyzed FCS (Thermo Fisher Scientific). To inhibit cytosolic translation, cycloheximide (Merck, Darmstadt, Germany) was added to the medium at 100 μg/ml for 10 min. Subsequently, 50 μM Click-iT AHA (Thermo Fisher Scientific) was added to the cells for 2 hours. After incubation, cells were collected, fixed by 4% paraformaldehyde (Fujifilm Wako Pure Chemical Corp.) and permeabilized by 0.25% Triton X-100 (Merck). L-Azidohomoalanine (AHA)-labeled proteins were further labeled with Alexa Fluor 488 Alkyne (Thermo Fisher Scientific) by click chemistry using Click-iT Cell Reaction Buffer Kit (Thermo Fisher Scientific) following the manufacturer’s instructions. Stained cells were analyzed using FACSCanto II flow cytometer (BD Biosciences), and data were analyzed using FlowJo (BD Biosciences).

Blue native PAGE

Ter119⁺ erythroblasts were sorted by magnetic beads selection using anti–Ter-119 microbeads (Miltenyi Biotec) and lysed in EzProteoLysis Native (ATTO, Tokyo, Japan) lysis buffer. The protein concentration was measured with Pierce Rapid Gold BCA Protein Assay Kit (Thermo Fisher Scientific). Sixty micrograms of proteins was subjected to blue native PAGE using 4 to 20% gradient TGX precast gels (Bio-Rad laboratories) and EzRun BlueNative (ATTO) buffer system. Separated proteins were transferred to PVDF membranes, which were destained with methanol followed by blocking with 5% skim milk. Transferred membranes were probed with the primary antibody at 4°C overnight, followed by an incubation with the secondary antibody at room temperature for 1 hour. The protein signal was detected with Pierce ECL Western Blotting substrate (Thermo Fisher Scientific). The following antibodies were used: a rabbit polyclonal antibody to ND2 (C I, #19704-1-AP, Proteintech, Rosemont, IL), a mouse monoclonal antibody to SDHA (C II, #ab14715, Abcam), a rabbit polyclonal antibody to CYTB (C III, #55090-1-AP, Proteintech), a mouse monoclonal antibody to COX IV (C IV, ab14744, Abcam), a mouse monoclonal antibody to ATP5A (C V, ab14748, Abcam), and secondary anti-mouse or anti-rabbit IgG, HRP-linked antibodies (#7076 and #7074, respectively, Cell Signaling Technology). For the densitometry analysis, the density of each band was quantified by ImageJ software (https://imagej.nih.gov/ij/). Data were normalized to the density of OXPHOS complex II in which mt-DNA–coded proteins were not incorporated.

Mitochondrial functions

For oxygen consumption rate (OCR) analysis, 3 × 10⁵ polychromatic erythroblasts were sorted with flow cytometry and plated into Seahorse XF HS PDL miniplate (Agilent technologies) in Seahorse XF DMEM containing 10 mM glucose, 2 mM l-glutamine, and 1 mM pyruvate (pH 7.4) (Agilent technologies). The miniplate was centrifuged at 200g for 1 min and incubated at 37°C in a non-CO₂ incubator for 40 min before OCR measurement on a Seahorse XF HS mini (Agilent technologies). Oligomycin (1 μg/ml), 1 μM carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone, 100 nM rotenone, and 10 μM antimycin were used during the measurement.

For mitochondrial ROS production, mitochondrial membrane potential, and mitochondrial mass analysis, fetal liver cells were stained with 5 μM MitoSOX Red (Thermo Fisher Scientific) at 37°C for 10 min or stained with 2 μM MitoProbe JC-1 (Thermo Fisher Scientific) or 50 nM MitoTracker Deep Red FM (Thermo Fisher Scientific) at 37°C for 30 min, followed by surface marker staining with antibodies (Ter119 and CD44) on ice for 30 min. Stained cells were analyzed by flow cytometry.

Intracellular iron quantification

For mitochondrial iron quantification, fetal liver cells were stained with 5 μM Mito-FerroGreen (Dojindo, Kumamoto, Japan) at 37°C for 30 min, followed by surface marker staining with antibodies (Ter119 and CD44) on ice for 30 min. Stained cells were analyzed with flow cytometry.

For determination of iron levels in isolated cytosolic fractions by ICP-MS, 1 × 10⁷ Ter119⁺ erythroblasts sorted by Moflo XDP (Beckman Coulter) were suspended in buffer A [10 mM tris-HCl (pH 7.4), 1 mM EDTA, and 320 mM sucrose], followed by homogenization with a Dounce homogenizer (40 strokes). The homogenate was centrifuged twice at 600g for 10 min at 4°C to remove debris and nuclei. The supernatant was further centrifuged at 8000g for 10 min at 4°C, followed by centrifugation at 12,000g for 10 min at 4°C to remove organelles. The resulting supernatant was saved as the cytosolic fraction. The cytosolic fractions were mixed with an equal volume of concentrated nitric acid (FUJIFILM Wako Pure Chemical, #143-09741) and incubated at 80°C, followed by centrifugation at 12,000g for 2 min. The resulting supernatant was diluted with 25 volumes of demineralized water, and iron levels were measured by 7900 ICP-MS (Agilent Technologies). The ICP-MS instrument was operated in helium (He) collision cell gas mode. Amounts of iron within a sample were normalized by total amount of protein, as determined by Bradford assay.

Transmission electron microscopy analysis

Fetal liver cells were stained with fluorescent antibodies, washed in phosphate-buffered saline (PBS), and centrifuged at 1500 rpm for 5 min. A pellet of stained fetal liver cells was fixed using double volume of 4% paraformaldehyde in 0.1 M PB for 20 min at room temperature. A 1 × 10⁶ polychromatic erythroblasts were sorted using flow cytometry, washed, and centrifuged at 1500 rpm for 5 min, and then the pellet was fixed for electron microscopy (EM) using double volume of 2.5% glutaraldehyde in 0.1 M PB for 30 min. Cells were applied to a poly-l-lysine–coated coverslip using cytospin centrifugation 800 rpm for 5 min. PB was applied quickly onto the coverslip to avoid drying of the specimen. Post-fixation was performed using 1% reduced OsO₄ that was prepared by a mixture of equal volumes of 2% aqueous OsO₄ and 3% potassium ferrocyanide in 0.2 M PB. Specimens were dehydrated using graded series of ethanol and embedded in epoxy resin by the inverted method for 48 hours at 60°C. After polymerization of the resin, coverslips were removed from the resin cylinder. Ultrathin sections were cut from the top surface of the resin cylinder where cells were embedded, stained with uranium acetate and lead citrate, and examined in EM (HT7700, Hitachi). Fifty polychromatic erythroblasts were randomly picked and analyzed to calculate the percent of siderosome-containing cells and quantify area of iron-containing siderosomes. Siderosome area was measured using ImageJ software.

Proteomic analysis

A label-free proteomic analysis was performed with tandem mass tag (TMT) quantitation. In the label-free analysis, polychromatic erythroblasts (3 × 10⁵ cells for each sample) were sorted from the fetal liver at E16.5 by flow cytometry and centrifuged to form cell pellets. Cell pellets were heated with 30 μl of NuPAGE 1× LDS sample buffer (Thermo Fisher Scientific) at 80°C for 10 min. Samples were loaded and run on a 10% NuPage NOVEX Bis-Tris gel (Thermo Fisher Scientific) for 10 min at 180 V in 1 × MES buffer (Thermo Fisher Scientific). Gels were then stained and fixed with Coomassie Brilliant Blue G250 (Sigma-Aldrich) for 15 min. The initial destaining of gels was performed overnight with water and followed by further destaining with a 50% ethanol and 50 mM ammonium bicarbonate (pH 8.0) solution. Proteins were reduced in 10 mM dithiothreitol at 56°C for 1 hour and then alkylated with 50 mM iodoacetamide at room temperature for 45 min in the dark. The in-gel digestion of proteins was performed with trypsin (Sigma-Aldrich) at 37°C overnight. Acetonitrile (30%) in a 50 mM ammonium bicarbonate (pH 8.0) solution twice and a 100% acetonitrile three times was used to extract peptides from the gel which was then evaporated in a concentrator (Eppendorf, Hamburg, Germany) and loaded onto the activated C18 material StageTips (CDS Analytical LLC, Oxford, PA) (59). A liquid chromatography tandem MS (LC-MS/MS) analysis was performed on an Easy-nLC 1200 (Thermo Fisher Scientific) and Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific). Peptides were separated on a 50-cm self-packed column (New Objective, Littleton, MA) with an inner diameter of 75 μm filled with ReproSil-Pur 120 C18-AQ (Dr. Maisch GmbH, Ammerbuch-Entringen, Germany). We used a 103-min gradient from 3 to 40% acetonitrile with 0.1% formic acid at a flow rate of 250 nl/min. The mass spectrometer was operated with a top 20 MS/MS data-dependent acquisition scheme per MS full scan. To identify protein groups, MS raw files were searched with MaxQuant version 1.5.2.8. Database searches were performed with MaxQuant standard settings with additional protein quantification using the label-free quantification (LFQ) algorithm and the match between runs option activated. Contaminants, reverse database hits, protein groups only identified by site, and protein groups with less than two peptides (at least one of them classified as unique) were removed. Missing values were imputed by shifting the compressed beta distribution obtained from the LFQ intensity values to the limit of quantitation.

Regarding TMT quantitation, polychromatic erythroblasts (1 × 10⁶ cells for each) were sorted from the fetal liver at E16.5 by flow cytometry and centrifuged to form a cell pellet. Proteins were extracted with 100 mM triethyl ammonium bicarbonate containing 12 mM sodium deoxycholate (SDC) and 12 mM sodium lauroyl sarcosinate (SLS) and were reduced using 10 mM dithiothreitol for 30 min followed by alkylation with 50 mM iodoacetamide for 30 min. Tryptic digested peptides were labeled with Tandem Mass Tag reagents and combined samples. After the removal of SDC and SLS by a phase transfer method (60, 61), peptides were separated using a High pH Reversed-Phase Peptide Fractionation Kit (Thermo Fisher Scientific). NanoLC-MS/MS were conducted using an Orbitrap Fusion Tribrid mass spectrometer (Thermo Fisher Scientific) and Easy nLC-1000 UHPLC (Thermo Fisher Scientific) equipped with a nanoHPLC capillary column (Nikkyo Technos, Tokyo, Japan). MS data were subjected to a search with MaxQuant version 1.6.17.0 (62). A list of differentially expressed proteins between Mto1^fl/fl and Mto1^fl/fl;Vav-Cre (log₂ ratio > 1 or ≤ 1) was analyzed by Metascape (28).

Intracellular heme quantification

The total heme (hemin) content was measured by MS according to a previously described method with modification (63). Briefly, 1 × 10⁶ polychromatic erythroblasts were sorted using flow cytometry and suspended in 0.1 ml of hemin extraction buffer [acetonitrile: 2 N HCl (8:2)]. After a brief sonication and vortexing, cell lysate was centrifugated at 10,000g for 5 min. The supernatant containing total hemin was subject to MS analysis (Shimazu LCMS8050). Multiple reaction monitoring mode was used to detect hemin [precursor ion: mass/charge ratio (m/z) 616, product ion: m/z 557]. The peak corresponding to hemin was validated using an authentic standard purchased from TCI (Tokyo Chemical Industry, Japan, catalog number H0008). The concentration of hemin was calculated using the standard hemin and normalized to the cell number.

Metabolite extraction

Polychromatic erythroblasts (3 × 10⁶ cells for each) were sorted from the fetal liver at E16.5 by flow cytometry and centrifuged to pellet cells. After washing with 5% mannitol solution, cells were treated with 800 μl of methanol and 550 μl of Milli-Q water containing internal standards [H3304-1002, Human Metabolome Technologies Inc. (HMT), Tsuruoka, Yamagata, Japan] was added to the cell extract. The extract was then centrifuged, and 700 μl of the supernatant was centrifugally filtered through a Millipore 5-kDa cutoff filter (Ultrafree MC-PLHCC, HMT) at 9100g at 4°C for 120 min to remove macromolecules. The filtrate was then evaporated to dryness under a vacuum and reconstituted in 50 μl of Milli-Q water for a metabolome analysis at HMT.

Metabolome analysis

A metabolome analysis was conducted according to HMT’s ω Scan package, using capillary electrophoresis Fourier transform MS (CE-FTMS) based on previously described methods (64). Briefly, a CE-FTMS analysis was performed using an Agilent 7100 CE capillary electrophoresis system equipped with Q Exactive Plus (Thermo Fisher Scientific), an Agilent 1260 isocratic HPLC pump, Agilent G1603A CE-MS adapter kit, and Agilent G1607A CE-ESI-MS sprayer kit (Agilent Technologies). The systems were controlled by Agilent MassHunter workstation software LC/MS data acquisition for 6200 series TOF/6500 series quadrupole orthogonal acceleration–time-of-flight version B.08.00 (Agilent Technologies) and Xcalibur (Thermo Fisher Scientific) and connected by a fused silica capillary (50 μm inner diameter × 80 cm in total length) with commercial electrophoresis buffer (H3301-1001 and I3302-1023 for cation and anion analyses, respectively, HMT) as the electrolyte. The spectrometer was scanned from m/z 50 to 1000 in the positive mode and from m/z 70 to 1050 in the negative mode (64). Peaks were extracted using MasterHands, automatic integration software (Keio University, Tsuruoka, Yamagata, Japan), to obtain peak information including m/z, peak areas, and migration times (MTs) (65). Signal peaks corresponding to isotopomers, adduct ions, and other product ions of known metabolites were excluded, and the remaining peaks were annotated according to the HMT’s metabolite database based on their m/z values and MTs. The areas of the annotated peaks were then normalized to internal standards and sample volumes to obtain relative levels for each metabolite. A metabolite sets enrichment analysis (66) was performed using the web-based metabolomics data analysis platform, MetaboAnalyst (67).

Apoptosis assay

Total fetal liver cells were stained with Pacific Blue–conjugated annexin V (BioLegend, San Diego, CA) in binding buffer for annexin V (Thermo Fisher Scientific) at room temperature for 15 min, followed by surface marker staining with antibodies (Ter119 and CD44) on ice for 30 min. Stained cells were washed once with binding buffer and resuspended in binding buffer containing 7-aminoactinomycin D (7-AAD) (Thermo Fisher Scientific) before being analyzed on a FACSCanto II flow cytometer (BD Biosciences, Franklin Lakes, NJ). All data were analyzed using FlowJo (BD Biosciences).

Ferroptosis assay

Total fetal liver cells were stained with 5 μM C11-BODIPY 581/591 (Thermo Fisher Scientific) in PBS at 37°C for 30 min, followed by surface marker staining with antibodies (Ter119 and CD44) on ice for 30 min. Stained cells were washed once and analyzed on a FACSCanto II flow cytometer (BD Biosciences, Franklin Lakes, NJ). All data were analyzed using FlowJo (BD Biosciences).

Intracellular calcium quantification

Total fetal liver cells were stained with 1 μM Fluo-4, AM (Thermo Fisher Scientific) at 37°C for 30 min, followed by surface marker staining with antibodies (Ter119 and CD44) on ice for 30 min and flow cytometric analysis.

Quantification and statistical analysis

All data were analyzed using GraphPad Prism 6 software. Statistical analyses were performed with an unpaired Student’s t test to compare two groups and multigroup comparisons were performed by a one-way analysis of variance (ANOVA) followed by the Tukey’s multiple comparisons test unless specified otherwise. Statistical analysis for the blue native PAGE data was performed with BootstRatio (68). ns, not significant; *P < 0.05, **P < 0.01, and ***P < 0.001. Data are shown as the means ± SEM unless specified otherwise.

Supplementary Materials

The PDF file includes:

Figs. S1 to S7

Legend for table S1

Table S2

Other Supplementary Material for this manuscript includes the following:

Table S1