Report

A Heme Export Protein Is Required for Red Blood Cell Differentiation and Iron Homeostasis

See allHide authors and affiliations

Science  08 Feb 2008:
Vol. 319, Issue 5864, pp. 825-828
DOI: 10.1126/science.1151133

This article has a correction. Please see:

Abstract

Hemoproteins are critical for the function and integrity of aerobic cells. However, free heme is toxic. Therefore, cells must balance heme synthesis with its use. We previously demonstrated that the feline leukemia virus, subgroup C, receptor (FLVCR) exports cytoplasmic heme. Here, we show that FLVCR-null mice lack definitive erythropoiesis, have craniofacial and limb deformities resembling those of patients with Diamond-Blackfan anemia, and die in midgestation. Mice with FLVCR that is deleted neonatally develop a severe macrocytic anemia with proerythroblast maturation arrest, which suggests that erythroid precursors export excess heme to ensure survival. We further demonstrate that FLVCR mediates heme export from macrophages that ingest senescent red cells and regulates hepatic iron. Thus, the trafficking of heme, and not just elemental iron, facilitates erythropoiesis and systemic iron balance.

Aerobic cells require heme, a cyclic tetrapyrole containing a centrally chelated iron. It serves as the prosthetic group for hemoglobin, cytochromes, and other hemoproteins. Heme also initiates globin transcription through inhibiting the DNA binding of the repressor, Bach1 (1), and globin translation through inhibiting substrate phosphorylation by the repressor, erythroid-specific eukaryotic initiation factor 2α kinase (2). However, the trafficking of heme and its role in iron homeostasis are poorly understood.

The feline leukemia virus, subgroup C (FeLV-C), receptor, FLVCR, is a heme export protein (3). Cats viremic with FeLV-C develop pure red cell aplasia (PRCA), characterized by a block in erythroid differentiation at the CFU-E (colony-forming unit–erythroid)–proerythroblast stage, reticulocytopenia, and severe anemia (4, 5). Studies with chimeric retroviruses suggest that the surface unit of the FeLV-C envelope protein induces this phenotype by blocking FLVCR function (6, 7). Although all bone marrow cells are infected (8), white cell and platelet production remain normal, which suggests that FLVCR is uniquely important for CFU-E–proerythroblast survival or differentiation.

To prove that FLVCR is required for erythropoiesis, we generated constitutive (Flvcr+/–) and inducible (Flvcr+/flox;Mx-cre) Flvcr mutant mice (9) (fig. S1). Interbred Flvcr+/– animals yielded no null offspring (Flvcr–/) among 109 progeny (table S1). Intrauterine deaths occurred at one of two embryonic times: at or before embryonic day 7.5 (E7.5) and between E14.5 and E16.5.

Developmental expression of Flvcr is high in the yolk sac at E7.5, the ectoplacental cone at E8.5, and the placenta after E9.5 (Fig. 1A); all are sites of nutritional transport from mother to conceptus. These are also sites of high heme oxygenase–1 expression (10). As heme catabolism helps to support normal fetal development (10), FLVCR might complement this function at or before E7.5.

Fig. 1.

Embryonic FLVCR analyses. (A) Wild-type mouse Flvcr expression (colored red) by in situ hybridization. Ectoplacental cone (ec), yolk sac (ys), embryo proper (ep), liver (li), neural tissue (n), placenta (pl) and intestine (in). Additional information is in SOM text. (B) E14.5 FLVCR-null embryo and a littermate control. The skeletal abnormalities are less apparent in embryos derived from interbreeding Flvcr+/ parental mice backcrossed to C57BL/6 mice for five to seven generations (SOM text). (C) Representative flow cytometric analyses of E14.5 liver cells from control and FLVCR-null embryos immunostained with antibodies to CD71 and Ter119. The relative percentages of the nucleated cells in each of the populations I to V are indicated.

We hypothesize that the later death results from deficient red cell production, because definitive fetal erythropoiesis in the mouse begins in the liver at ∼E12 (11), hepatic FLVCR expression is high from E12.5 onward (Fig. 1A), and FLVCR-null embryos have pale livers (Fig. 1B). Flow cytometric analyses of E14.5 fetal liver cells double-stained for Ter119 (erythroid-specific antigen) and CD71 (transferrin receptor) allow quantitative assessment of the maturational stages of differentiating erythroblasts (12) and confirm this concept. Normally, differentiation proceeds clockwise from population I to IV (control in Fig. 1C). In contrast, the null embryos lack Ter119high cells, consistent with a block at the proerythroblast stage, before hemoglobinization (population II). Circulating yolk sac–derived erythroblasts do not express Flvcr by in situ hybridization and have normal morphology (fig. S3), which indicates that embryonic (primitive) erythropoiesis does not require FLVCR.

Although the null embryos appear normal at E8.5, E10.5, and E12.5, defective growth is evident at E14.5. Mutants have abnormal limb, hand, and digit maturation; flattened faces; and hypertelorism (Fig. 1B)—abnormalities that resemble human congenital PRCA, termed Diamond-Blackfan anemia (13, 14). Gross and microscopic examination of the cardiac, pulmonary, and genitourinary systems shows that they are normal. Although it is theoretically possible that the observed phenotype is developmentally appropriate for a growth-retarded embryo, these specific abnormalities are not reported in other mouse models lacking definitive erythropoiesis (11, 15). Thus, FLVCR may serve roles during embryogenesis distinct from its critical erythropoietic function.

Although null animals die in utero, Flvcr+/– mice are clinically indistinguishable from controls (table S2); they have low mRNA expression, as anticipated, but compensate with normal FLVCR protein expression (fig. S4).

We next evaluated postnatal mice lacking FLVCR [Flvcr flox/flox;Mx-cre (fig. S1 and Fig. 2, A to F)]. Within 4 weeks of Flvcr deletion, the mice are runted with pale paws. Necropsy reveals cardiomegaly and splenomegaly [Flvcr-deleted spleen: 326.7 mg ± 22.9 (n = 7) versus control spleen 72.9 ± 5.5 (n = 7); means±SEM, two-tailed Student's t test, P <10–4], likely responses to their severe anemia.

Fig. 2.

Conditional deletion of Flvcr causes PRCA. Unless noted, data are from representative 6-week-old mice, 5 weeks post deletion, (left) controls (n = 13), (right) Flvcr-deleted (n = 11). (A) Hematologic parameters (means ± SEM, one-tailed Student's t test), hemoglobin (HGB), mean cell corpuscular volume (MCV). Duplicate spun hematocrit tubes from two control and two Flvcr-deleted mice. (B) Flow cytometric analyses of marrow from a control and Flvcr-deleted mouse immunostained with antibodies to CD71 and Ter119. Gating methods in Fig 1C. Ratio of the percent of cells in population IV to I and II: Flvcr-deleted: 49.2% ± 11.6% (n = 9) versus control: 77.1% ± 11.0% (n = 9); means ± SD, two-tailed Student's t test, P <10–4. The severity of block is variable between deleted animals and does not appear to correlate with the degree of anemia. (C) Hematoxylin-and-eosin–stained spleen sections from a control and Flvcr-deleted mouse. (D to F) Representative Prussian blue–stained liver sections (D) from a 6-week-old control and a Flvcr-deleted mouse, and duodenum (E) and spleen (F) sections from a 11-week-old (10 weeks post deletion) control and a Flvcr-deleted mouse. Blue staining indicates iron. Scale bars in microns.

Peripheral blood and bone marrow findings are diagnostic of PRCA. Flvcr-deleted mice develop a severe hyperchromic macrocytic anemia (Fig. 2A and table S3) and reticulocytopenia. Flow cytometric analyses of their bone marrow show a block in erythroid maturation at the proerythroblast stage (Fig. 2B), as do liver cells from E14.5 FLVCR-null embryos. These results are mirrored in the spleen and account for the large spleens with expanded interfollicular regions (Fig. 2C). Erythroid colony assays confirm the flow cytometry findings; CFUs-E are absent and BFUs-E (burst-forming units–erythroid) expand suboptimally [supporting online material (SOM) text], similar to results in cats viremic with FeLV-C (5). In addition, mice transplanted with Flvcr flox/flox;Mx-cre bone marrow and then treated with polyinosinic-polycytidylic acid [poly(I):poly(C)] to delete Flvcr specifically in engrafted cells also develop PRCA (table S4). This confirms that a lack of FLVCR in hematopoietic cells (and not the microenvironment) accounts for the disease.

We then evaluated the effect of FLVCR overexpression. Pep3b (CD45.1) bone marrow was transduced with retroviral vectors, MFIG or MXIG, encoding green fluorescent protein with or without human FLVCR, respectively, and transplanted into C57BL/6 (CD45.2) mouse recipients. Twelve weeks after transplantation, the MFIG mice displayed mild hypochromic, microcytic anemia [supporting online material (SOM) text]. Because hypochromasia and microcytosis only result from heme or hemoglobin deficiency, FLVCR must export heme from differentiating erythroid cells in vivo. Because the anemia is mild, FLVCR does not outcompete globin for heme.

These observations lead us to hypothesize that FLVCR is required during definitive red cell differentiation to maintain intracellular free heme balance. In the absence of FLVCR, free heme, which is toxic, accumulates in proerythroblasts, the stage when heme synthesis intensifies (16), and triggers molecular pathways that result in cell apoptosis or senescence. Although this may seem counterintuitive because red cells have high heme requirements for hemoglobin, we suspect that FLVCR functions as a safety valve to protect proerythroblasts from heme toxicity when globin expression [which is transcriptionally and translationally regulated by heme (1, 2)] is insufficient. In human tissues, FLVCR is highly expressed at sites of high heme flux, including placenta, uterus, duodenum, liver, and cultured macrophages (Fig. 3), which suggests that FLVCR prevents heme toxicity or facilitates heme iron trafficking in non-erythroid cells as well.

Fig. 3.

FLVCR protein levels differ in human tissues. (A) Western blot analyses of human tissues, bone marrow mononuclear cells (BM MNC), and CD34+ stem/progenitor cells. (B) Densities of the 60-kB FLVCR band [shown in part (A)]. We also assayed FLVCR expression in macrophages isolated from human peripheral blood by plastic adherence for 2 hours, then cultured for 4 days with cytokines (intensity = 5214 ± 260). Quantitative RT-PCR confirmed that FLVCR expression is regulated posttranscriptionally (SOM text) (3, 24).

When senescent red cells are phagocytosed and digested by macrophages, hemoglobin is degraded to heme and, subsequently, to iron, biliverdin, and carbon monoxide. Ferroportin exports iron to plasma transferrin for delivery to the marrow or liver (17). Hepcidin regulates this pathway by inducing the internalization and degradation of ferroportin, thereby blocking intestinal iron absorption and iron release from cellular stores and macrophages (18). To delineate the role of FLVCR in macrophage heme iron recycling, we exposed marrow-derived macrophages from Flvcr-deleted and control mice to ferric ammonium citrate (FAC) or opsonized red blood cells, in the presence or absence of hepcidin, and measured ferritin (Fig. 4A). Deleted and control macrophages exposed to FAC accumulate equivalent amounts of ferritin, which increase equivalently with hepcidin treatment. However, Flvcr-deleted macrophages exposed to opsonized red cells accumulate more ferritin than controls both with and without hepcidin treatment. These data support the model of macrophage heme iron recycling diagrammed in Fig. 4B; under normal physiologic conditions, heme can be exported via FLVCR or can be metabolized to iron, which is subsequently exported through ferroportin or stored as ferritin. When FLVCR is absent, the amount of iron that is generated exceeds ferroportin's export capacity, resulting in an increase in ferritin, which increases further if hepcidin is present and both heme iron and inorganic iron export is blocked. Our data confirm that not all heme in macrophages is broken down (19), but rather some traverses the cell intact via FLVCR. We further verified this export function by 55Fe-heme and zinc mesoporphyrin studies (fig. S6).

Fig. 4.

FLVCR exports heme iron from macrophages. (A) Bone marrow–derived macrophages from control (striped) and mice in which Flvcr was deleted neonatally (black) were incubated in the absence or presence of FAC (10 μM Fe) for 24 or 48 hours, then washed; ferritin was measured by enzyme-linked immunosorbent assay (ELISA) (i). Cells were incubated with FAC for 24 hours (ii) or with immunoglobulin G–coated red blood cells (RBC) for 90 min (iii), washed, then incubated for an additional 24 hours with or without hepcidin (1 μg/μl) and ferritin assayed. Data represent ferritin values in macrophages derived from two control and two deleted mice ± SEM of triplicate samples per mouse. (B) Model of macrophage heme iron recycling. HO-1, heme oxygenase–1.

To evaluate the role of FLVCR more broadly, we examined other tissues in Flvcr-deleted mice. Within 5 weeks, mice with the deletion develop pronounced iron loading in hepatocytes and subsequently within duodenal enterocytes and splenic macrophages (Fig. 2, D to F). By 7 months, there is swelling of hepatocytes lining bile canaliculi and bile stasis. In contrast, the mice in which Flvcr is deleted only in hematopoietic cells show no iron overload after 5 to 6 weeks (fig. S5). Liver hepcidin expression by reverse transcription polymerase chain reaction (RT-PCR) is comparably increased in mice with the deletion [1.7 ± 0.2 times control; deleted (n = 5), control (n = 5); means±SEM; two-tailed Student's t test, P = 0.04] and mice lacking FLVCR only in hematopoietic cells [2.0 ± 0.3 times control; lacking FLVCR (n = 6), control (n = 3); P = 0.03]. These data demonstrate that hepcidin alone does not account for the iron overload and biliary pathology. One possibility consistent with our data is that FLVCR exports heme from liver into bile, thus allowing iron to exit the body.

The high hepcidin levels in Flvcr-deleted animals contrasts with levels in other iron-loading anemias with ineffective erythropoiesis, such as thalassemia and congenital dyserythropoietic anemia, where hepcidin is low despite high serum iron and systemic iron overload (20). High hepcidin levels are seen in anemic mice prevented from mounting an erythropoietic response by the use of irradiation, chemotherapy, or an antibody to erythropoietin (21, 22), which indicates that erythropoietic activity is the most potent suppressor of hepcidin synthesis. Our results demonstrate that the inhibitory signal must originate from cells more differentiated than proerythroblasts and, thus, are consistent with the recent finding that growth differentiation factor GDF15 inhibits hepcidin expression (23).

Together, our data show that FLVCR exports heme in vivo and is required by definitive erythroid progenitors at the CFU-E–proerythroblast stage to complete terminal differentiation. We propose that heme toxicity causes PRCA in FLVCR mutant mice and FeLV-C–infected cats and may be a common pathophysiology in other models of failed erythropoiesis where heme synthesis and globin expression are dysregulated, which results in a transient excess of intracellular free heme, for example Diamond-Blackfan anemia (SOM text). Our data demonstrate that FLVCR functions in macrophage heme-iron recycling and show that systemic iron balance involves heme-iron trafficking via FLVCR, in addition to the well-described elemental iron pathways.

Supporting Online Material

www.sciencemag.org/cgi/content/full/319/5864/825/DC1

Materials and Methods

SOM Text

Figs. S1 and S6

Tables S1 to S4

References

References and Notes

View Abstract

Navigate This Article