Iron Regulatory Protein-1 Protects against Mitoferrin-1-deficient Porphyria*

Background: Heme and [Fe-S] cluster assembly are tightly regulated processes that require mitochondrial iron. Results: Loss of mitochondrial iron activates the [Fe-S]-dependent RNA-binding activity of IRP1 that inhibits protoporphyrin biosynthesis. Conclusion: IRP1 forms a critical feedback mechanism, preventing protoporphyrin accumulation under limiting mitochondrial iron conditions. Significance: This study provides evidence linking heme biogenesis to that of [Fe-S] clusters synthesis. Mitochondrial iron is essential for the biosynthesis of heme and iron-sulfur ([Fe-S]) clusters in mammalian cells. In developing erythrocytes, iron is imported into the mitochondria by MFRN1 (mitoferrin-1, SLC25A37). Although loss of MFRN1 in zebrafish and mice leads to profound anemia, mutant animals showed no overt signs of porphyria, suggesting that mitochondrial iron deficiency does not result in an accumulation of protoporphyrins. Here, we developed a gene trap model to provide in vitro and in vivo evidence that iron regulatory protein-1 (IRP1) inhibits protoporphyrin accumulation. Mfrn1+/gt;Irp1−/− erythroid cells exhibit a significant increase in protoporphyrin levels. IRP1 attenuates protoporphyrin biosynthesis by binding to the 5′-iron response element (IRE) of alas2 mRNA, inhibiting its translation. Ectopic expression of alas2 harboring a mutant IRE, preventing IRP1 binding, in Mfrn1gt/gt cells mimics Irp1 deficiency. Together, our data support a model whereby impaired mitochondrial [Fe-S] cluster biogenesis in Mfrn1gt/gt cells results in elevated IRP1 RNA-binding that attenuates ALAS2 mRNA translation and protoporphyrin accumulation.

Iron is an essential element that is incorporated into prosthetic groups that function in a multitude of biochemical processes. Two of these iron-containing prosthetic groups are heme and iron-sulfur ([Fe-S]) clusters. Heme is a moiety with diverse functions in many cell types and is an integral component of hemoglobin in erythroid cells (1). [Fe-S] clusters are requisite cofactors for many enzymes, including aconitases and succinate dehydrogenase (2,3). Although functionally distinct, assembly of both cofactors require mitochondrial function and mitochondrial iron assimilation (4).
Iron is imported into mitochondria via two transporters: mitoferrin-1 (MFRN1, SLC25A37) and mitoferrin-2 (MFRN2, SLC25A28). MFRN2 is expressed ubiquitously, whereas MFRN1 is specifically induced in differentiating erythroid cells (5)(6)(7). Previous work has shown that loss of Mfrn1 in frascati zebrafish and mice disrupts heme and [Fe-S] cluster synthesis, owing to a severe reduction in mitochondrial iron in erythroid progenitors (7,8). However, neither Mfrn1 Ϫ/Ϫ mice nor zebrafish frascati embryos deficient in Mfrn1 develop porphyria. In fact, conditional deletion of Mfrn1 in hepatocytes only causes porphyria and hepatobiliary stasis when the animals are fed a ␦-aminolevulinic acid (ALA)-rich 12 diet (8). This indicates that reduction in mitochondrial iron stores may concomitantly attenuate early steps of protoporphyrin synthesis.
Recent evidence suggests that cross-talk between [Fe-S] cluster assembly and protoporphyrin synthesis pathways may explain the absence of porphyria in Mfrn1 Ϫ/Ϫ animals (9). Cellular iron homeostasis is largely regulated by iron regulatory proteins-1 (IRP1) and -2 (IRP2) that can directly influence protein expression via two mechanisms (10 -12). IRPs can bind to iron responsive elements (IREs) in the 5Ј-UTRs of mRNAs, blocking translation initiation. Alternatively, mRNAs with IRE(s) in the 3Ј-UTR have increased stability when bound by IRPs. As a result, this latter class of transcripts will have increased accumulation when IRPs are activated by iron deficiency. Although IRP1 and IRP2 exhibit some functional redundancy and can be regulated by common mechanisms (13)(14)(15), IRP1 is unique in that RNA-binding activity is negatively regulated by [Fe-S] cluster biogenesis. Insertion of an [Fe-S] cluster into IRP1 converts IRP1 from an IRE-RNA binding protein to cytosolic aconitase (ACO1). A decline in [Fe-S] cluster synthesis would, therefore, trigger increased IRP1 RNAbinding activity (16 -19).
One mRNA harboring a 5Ј-IRE is erythroid-specific aminolevulinate synthase 2 (Alas2), which encodes the enzyme that catalyzes the first step in the heme biosynthetic pathway (20,21). Wingert and colleagues (9) previously demonstrated that inhibition of glutaredoxin 5 (Glrx5), a gene required for mitochondrial [Fe-S] cluster assembly, resulted in anemia in zebrafish via increased IRP1-mediated inhibition of ALAS2 translation. Thus, a reduction in [Fe-S] cluster assembly in response to iron deficiency can feedback to attenuate the very proximal step of heme assembly (9,22). Here, we examined whether loss of Mfrn1 can also trigger IRP1 activity in erythroid cells. We generated a model to show that in the absence of MFRN1, IRP1 blocks ALAS2 translation, preventing the accumulation of protoporphyrins in vitro and in vivo. Loss of IRP1 or the 5Ј-IRE in Alas2 mRNA resulted in porphyria. Our work indicates that IRP1 functions as a critical link coupling heme and [Fe-S] cluster biosynthetic pathways and that in the absence of MFRN1-mediated mitochondrial iron import, IRP1 protects erythroid cells against porphyria.

EXPERIMENTAL PROCEDURES
Cell Culture-Friend murine erythroleukemia cells were cultured and differentiated as described previously (23). Mouse embryonic stem (mES) cells containing a gene trap insertion in Mfrn1 (Mfrn1 ϩ/gt cell line, XB454, derived from strain 129P2) were obtained from the University of California, San Francisco BayGenomics (24). Insertion of the gene trap vector (containing the coding sequence for ␤-geo) into Mfrn1 was verified by direct sequencing of cDNA obtained by 5Ј-rapid amplification of cDNA ends (25). mES cells were maintained on gelatincoated 100-mm dishes in mES cell medium (7). Null Mfrn1 gt/gt mES clones were derived using G418 selection as described previously (7).
Mouse Blastocyst Injections and Knock-out Mouse Creation-ES cells were injected into C57BL/6J blastocysts and transferred to pseudopregnant C57BL/6J hosts using standard techniques (26). Male chimeras identified by agouti coat color were mated to C57BL/6J females to generate heterozygotes for interbreeding to produce homozygous knock-out progeny. Progeny were genotyped as described below. All mice were maintained at The Jackson Laboratory in a climate-controlled room with a 12-h light cycle and provided acidified water and NIH 5K52 chow ad libitum. The Jackson Laboratory Animal Care and Use Committee approved all protocols. In accordance with the International Committee on Standardized Genetic Nomenclature for Mice, the official genetic designation for this Mfrn1 targets strain is B6.129P2-Slc25a37ϽGt(XB454)BygϾ/Llp.
Mixed Chimera Analysis-Mfrn1 gt/gt ES cell clones were injected into C57BL/6J mouse blastocysts and transferred to C57BL/6J recipients. The degree of mosaicism in offspring was estimated by coat color. The contributions of donor (129P2) versus recipient (C57BL/6) ES cells to the red cell and leukocyte compartments were determined by cellulose acetate electrophoresis to detect strain-specific hemoglobin and glucose-6isomerase alleles as described previously (28) in three chimeric mice (one female with moderate chimerism, two males with moderate to high chimerism) at 10 weeks of age.
o-Dianisidine Staining, CFU-E, and Cytospin Assays-Wildtype E14 and Mfrn1 null ES clones were generally split 1 day after thawing with 10 6 mES cells plated onto gelatinized 100-mm plates in mES media as described previously (7). The following day, mES cells were maintained in "switch media" (7). Two days after the split, the cells were collected by trypsinization and replated on untreated 100-mm dishes at a density of 3 ϫ 10 3 cells/ml (total 6 ϫ 10 4 cells) in embryoid body (EB) media with murine VEGF (10 ng/ml) on days 1 and 5. ALA (0.25 mM) was added to EB cultures for the last 24 h of day 6 primitive differentiation (7). Hemoglobinization of the primitive differentiated cells was assessed by o-dianisidine staining as described (29).
Iron, Heme, and Protoporphyrin Assays-Intracellular iron measurement was performed as described previously (30). Cells were washed in nitric acid, and sample iron content was determined using a Perkin-Elmer inductively coupled plasma optical emission spectrometer. Results were normalized by cell numbers. Mitochondrial iron was similarly quantitated by inductively coupled plasma spectroscopy in mitochondria that were isolated from wild-type Mfrn1 ϩ/ϩ or Mfrn1 gt/gt EBs using the mitochondrial isolation kit according to the manufacturer's instructions (Promega, Madison, WI) and normalized by mitochondrial protein content. For heme and protoporphyrin assays, EBs or embryos were sonicated and heme and protoporphyrin were extracted in 19:1 acetone:concentrated HCl as described (31). The resulting supernatant was injected immediately into a computer-controlled Waters reverse phase high-pressure liquid chromatography system that consisted of a 996 Photodiode Array Detector, a 474 scanning fluorescence detector, an AllianceHT 2795 separations module, and a Bondapak C18 3.9 ϫ 300 mm column with a guard column attached. Bovine Type 1 crystalline heme (Sigma, H-2250) or Protoporphyrin IX (Sigma, P-8293) was used as the standard. The absorbance values were normalized by cell number or protein concentration. EBs were supplemented with 0.25 mM ALA (Sigma) for 24 h prior to cell harvest and biochemical analysis.
Western Analysis-Immunoblotting was performed as described previously (32). Custom, affinity-purified anti-mouse MFRN1 rabbit polyclonal antibody was generated using the following peptide: C-HESHVQEVSHKTSPT (Genemed Synthesis, San Antonio, TX). Goat polyclonal anti-HSP60 (K-19) was purchased from Santa Cruz Biotechnology. Anti-ALAS2 was a generous gift of Dr. Makoto Murakami (Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan). Quantitative densitometry was performed using ImageJ software (NIH, Bethesda, MD) and normalized to HSP60 expression levels.
Lentivirus Infection-Lentivirus production and infection was performed as described previously (35). Briefly, 293T cells were plated at 6 ϫ 10 6 cells per 100 mm dish and incubated overnight. Cells were transfected with retroviral vectors together with psPax2 and pMD2G. Forty-eight hours after transfection, the supernatants were collected and ultra-centrifuged at 28,000 rpm for 2 h. Supernatants from three 10-cm dishes were concentrated into 100 l of PBS. Murine ES cells in 24-well plates were transduced using various amounts of concentrated virus supplemented with 8 g/ml polybrene. After 24 h, the cells were changed back to standard medium. Three days after transduction, the cells were dissociated by 0.05% trypsin and sorted by FACS to obtain the GFP-positive cells.

Mfrn1 gt/gt ES Cells Are Defective at Erythroid Differentiation-
Mfrn1 has been studied previously in mice using tissue-specific Cre/Lox technology (8). Although global loss of Mfrn1 in this model leads to embryonic lethality due to a profound embryonic anemia, a cell-autonomous role for MFRN1 in adult hematopoiesis has never been described. Thus, to address this and to study the underlying molecular mechanisms, we developed a model where we could manipulate MFRN1 function in vitro and in vivo. Gene trap mutagenesis was used to disrupt the Mfrn1 locus between introns 1 and 2 (Fig. 1A). Proper integration in murine ES cells was confirmed using PCR and Southern blot analysis (Fig. 1, B and C). Wild-type (Mfrn1 ϩ/ϩ ), heterozygote (Mfrn1 ϩ/gt ), and null (Mfrn1 gt/gt ) ES cells were then cultured in vitro under conditions that facilitate primitive or definitive erythropoiesis. Mfrn1 gt/gt cells, lacking MFRN1 protein expression, showed defective hemoglobinization, failed to form both primitive and definitive CFU-E colonies and were morphologically distinct from wild-type differentiated erythroid cells (Fig. 2, A-E). Semi-quantitative RT-PCR analysis also revealed that expression of erythroid markers, ␣-globin, ␤H1globin, ␤major-globin, and Alas2, were diminished in differentiating Mfrn1 gt/gt cells compared with wild-type (Fig. 2F), indicating that erythroid maturation in mutant cells was blocked. In contrast, the early erythroid marker, Gata-1, and ubiquitously expressed genes such as Mfrn2, Hprt, and GAPDH were normal compared with wild-type and Mfrn1 ϩ/gt cells. Total intracellular iron, mitochondrial iron, and intracellular heme content were all decreased without protoporphyrin accumulation (Fig.  2, G-I, and data not shown). We also examined both endogenous aconitase 1 activity and IRP1 RNA-binding (data not shown). However, we were unable to detect significant changes using either assays with this embryoid culture system. We believe that this is due to a heterogeneous population of cells in our differentiation cultures. Under optimal conditions, only 30% of ES cells undergo proper erythroid maturation (Fig. 2C). RNA-binding and aconitase assays may lack sufficient sensitivity to detect moderate changes.
Mfrn1 Is Required for Primitive and Definitive Erythropoiesis in Vivo-We next examined whether the primitive erythropoietic defects in ES cells can be recapitulated in vivo by injecting wild-type or Mfrn1 gt/gt ES cells into early-stage embryos and generating mixed-chimeric mice. Mfrn1 ϩ/gt and Mfrn1 gt/gt offspring were obtained following germ line transmission of the mutant allele. Surrogate lacZ-staining for Mfrn1 expression in heterozygote and gene-trapped embryos detected staining in the yolk sac blood islands and fetal liver at E8.5 and E11.5, respectively (Fig. 3, A and B). At E11.5, Mfrn1 gt/gt embryos were severely anemic compared with wild-type littermates (Fig. 3B), and no mutant embryos survived past E11.5 (Table 1). MFRN1 is also required in vivo for definitive erythropoiesis. In mixed chimeras, only the host recipient C57BL/6J-derived hemoglobin isoform is detected by cellulose acetate electrophoresis of adult red cells (Fig. 3C). In contrast, genetic contributions from both recipient and donor (mutant, 129P2) glucose-6-isomerase isoforms were detected in leukocytes, indicating the essential role of MFRN1 in the development of the definitive erythroid lineage and its nonessential role in non-erythroid lineages in adult mice. These data confirm that our gene trap model provides an accurate representation of MFRN1 function in vitro and in vivo.
IRP1 Attenuates Protoporphyrin Production-Because disruption of [Fe-S] cluster synthesis has previously been linked to Alas2 mRNA translational repression by IRP1 (9), we next used our model to ask whether ALAS2 translation had any influence on protoporphyrin production in the absence of MFRN1. Mfrn1 gt/gt EBs had lower expression of ALAS2 protein compared with Mfrn1 ϩ/ϩ EBs (Fig. 4A), suggesting that ALAS2 translation is indeed compromised in the absence of MFRN1. To examine the contribution of the 5Ј-IRE in MFRN1 regulation of ALAS2 function and expression, Mfrn1 gt/gt ES cells were transduced with lentivirus expressing either empty vector (mock) or zebrafish alas2 cDNA constructs harboring a wildtype IRE (WT) or mutant IRE (ϪIRE) in the 5Ј-UTR. An additional mutant alas2 cDNA with a premature translational stop codon (ϪIRE/Stop) was also included as a negative control (Fig.  4, B and C). These cells were then differentiated into EBs. As expected, ectopic expression of WT-alas2 resulted in a modest increase in protoporphyrin levels above mock-transduced control cells (Fig. 4, C and D). However, there was almost a 2-fold increase in protoporphyrin levels in cells expressing the ϪIRE mutant construct due to increased ALAS2 expression (Fig. 4, C  and D). The level of protoprophyrin IX (PPIX) dropped to basal levels in cells expressing the double mutant alas2 construct (ϪIRE/Stop) (Fig. 4, C and D). Furthermore, Mfrn1 gt/gt EBs supplemented with exogenous ALA to by-pass the block in protoporphyrin synthesis resulting from diminished ALAS2 protein expression fully restored PPIX but not heme production to levels comparable with Mfrn1 ϩ/ϩ cells (Fig. 4E). Thus, inhibition of protoporphyrin synthesis through action of the 5Ј-IRE in Alas2 mRNA is substantially increased when mitochondrial  [Fe-S] cluster assembly is compromised. A deficiency of MFRN1 should result in elevated cytosolic IRP1 RNA-binding activity that would attenuate translation of ALAS2 and lower PPIX production and its intracellular accumulation.
To address whether this was the case in vivo and to confirm whether IRP1 was responsible for negatively regulating ALAS2 expression, we examined heme and protoporphyrin concentrations in four cohorts of murine embryos, Mfrn1 ϩ/ϩ ;Irp1 ϩ/ϩ , Mfrn1 ϩ/gt ;Irp1 ϩ/ϩ , Mfrn1 ϩ/ϩ ;Irp1 Ϫ/Ϫ , and Mfrn1 ϩ/gt ;Irp1 Ϫ/Ϫ , at E10.5. Mfrn1 heterozygosity has little effect on heme levels in Irp1 ϩ/ϩ background (Fig. 4F). Protoporphyrin levels decreased, consistent with the notion that a reduction in mitochondrial iron import, while having little effect on heme biosynthesis, activates IRP1 due to a mitochondrial [Fe-S] cluster biogenesis defect (7) to inhibit ALAS2 translation. It should be noted that ferrochelatase (FECH), the terminal and rate-limiting enzyme in heme biosynthesis, is a [Fe-S] cluster protein and would also be influenced by the ability to form [Fe-S] clusters (37). However, the decreased FECH activity is unlikely to be responsible for the decline in protoporphyrin levels because diminished FECH function is predicted to exacerbate protoporphyrin accumulation. In contrast, in the absence of IRP1, loss of one copy of Mfrn1 causes a sharp rise in cellular protoporphyrin content, providing in vivo evidence that IRP1 prevents porphyria when MFRN1 is deficient. Taken together, our data strongly suggest that IRP1-mediated translational inhibition of ALAS2 in erythroid cells constitutes a feedback mechanism to attenuate protoporphyrin synthesis and accumulation when mitochondrial iron import through MFRN1 is compromised (Fig. 5).

DISCUSSION
The role of MFRN1 in primitive erythropoiesis and heme synthesis has been established in zebrafish and murine models (7,8). Although these genetic studies have been highly informative, a cell-autonomous in vivo role for MFRN1 has never been described. It was also unclear as to why porphyria was never observed with MFRN1 deficiency. In this work, we developed a genetic and biochemical model to address these questions. We provide evidence that MFRN1 is essential for primitive and definitive erythropoiesis in vivo. These defects could be recapitulated in ES cell cultures, which allowed us to study the underlying The mutant embryos were also stained for lacZ and have a profound anemia due to a defect in primitive erythropoiesis. C, erythrocytes and leukocytes from recipient, donor, or chimeric mice with medium or light penetrance (agouti coat color) were isolated and lysed, and isoforms were resolved on starch-cellulose gel electrophoresis and Ponceaured stained for hemoglobin or glucose-6-isomerase (GPI). MFRN1-deficient donor-derived cells do not contribute to adult erythrocytes but do contribute to the myeloid lineage, indicating that MFRN1 is dispensable for myelopoiesis.  (Fig. 5). Mfrn1 mutant embryos only developed porphyria in the absence of IRP1 (Fig. 4). Although our studies have focused on IRP1, both IRP1 and IRP2 function in red blood cells to regulate ALAS2 translation (9,17,38). Studies of translational regulation of 5Ј-IRE-containing mRNA in a cell culture model of erythropoiesis found that unlike ferritin mRNA, which was almost totally transla-tionally repressed, Alas2 mRNA was partially repressed with significant amounts of the mRNA present on polyribosomes as well as in the translationally silent ribonucleoprotein pool (39). Thus, an erythropoietic phenotype tied to altered ALAS2 expression is predicted in response to either a reduction or an increase in IRP RNA-binding activity. Irp2 Ϫ/Ϫ mice develop erythropoietic protoporphyria resulting from translational derepression of Alas2 mRNA and subsequent overproduction of protoporphyrin intermediates (16,17). In contrast, impairment of [Fe-S] cluster biogenesis when ABCB7 (40 -42) or GLRX5 (9, 22) are disrupted specifically activates IRP1 RNA- . Mitochondrial heme and Fe/S cluster biogenesis are co-regulated. A, mitochondria were isolated from undifferentiated (Undiff.) and differentiated (Diff.) Friend murine erythroleukemia (MEL) cells as well as Mfrn1 ϩ/ϩ and Mfrn1 gt/gt embryoid bodies and subjected to Western blot analysis using the indicated antibodies. The asterisk denotes a cross-reacting band, and the arrow represents ALAS2. Densitometry analysis revealed that the intensities of the ALAS2 bands normalized to HSP60 expression levels in wild-type and gt/gt lanes are 100% and 61.5%, respectively. B and C, Mfrn1 gt/gt ES cells infected with lentivirus expressing various alas2 constructs were lysed, subjected to SDS-PAGE, and immunoblotted (IB) with anti-ALAS2 and anti-HSP60 antibodies. The schematic in B highlights the features of the alas2 constructs as well as the anticipated results. D, PPIX levels were analyzed per milligram of total protein in Mfrn1 gt/gt cells transduced with various lentiviruses, shown in B. E, heme and PPIX levels were measured in WT (Mfrn1 ϩ/ϩ ) and gt/gt (Mfrn1 gt/gt ) embryoid bodies with (ϩALA) or without exogenous (ϪALA), showing a dramatic increase in protoporphyrin but not heme in Mfrn1 gt/gt cells because mitochondrial iron is limiting. F, murine embryos of various genotypes were examined for heme and protoporphyrin content at E10.5. Shown are mean Ϯ S.E.; all experiments were performed at least twice. IB, immunoblot.
binding activity leading to repression of Alas2 mRNA translation. The observed increase in IRP1-mediated inhibition of ALAS2 via its 5Ј-IRE suggests that [Fe-S] cluster insertion into ACO1 is also compromised in the absence of MFRN1. Thus, by further supporting the concept that cytosolic [Fe-S] biogenesis is downstream of mitochondrial [Fe-S] formation that is fueled, in part, by the action of MFRN1, our findings also illustrate the unique role of IRP1 in linking mitochondrial iron metabolism with protoporphyrin synthesis in erythroid cells (9,40,43,44). Taken together, these studies highlight differing roles of IRP1 and IRP2 in regulating ALAS2-IRP1 links protoporphyrin synthesis with [Fe-S] cluster synthesis and mitochondrial function, whereas IRP2 may have a more general "housekeeping" role in iron metabolism.
Our work presented here has important clinical implications. They suggest that although Mfrn1 mutations may contribute to human porphyrias, Mfrn1 inactivation alone is not likely sufficient for development of erythropoietic protoporphyria. In accord, expression of aberrant Mfrn1 and a concomitant decrease in Fech expression is associated with human erythropoietic protoporphyria (45). Interestingly, C-terminal ALAS2activating mutations (45)(46)(47) were also found in these erythropoietic protoporphyria patients with abnormal Mfrn1 transcripts. Also, Alas2 has been shown to be a genetic modifier in patients with congenital erythropoietic porphyria (48 -50). This indicates that the IRP1/ALAS2 negative feedback loop also exists in humans and that disruption of multiple regulators in the heme and [Fe-S] cluster synthesis pathways are required for protoporphyrin accumulation (45). In Mfrn1 Ϫ/Ϫ conditional knock-out mice, this porphyria effect can be recapitulated by exogenous chemical supplementation of ALA in the diets (8). Given the cytotoxic effects of protoporphyrin and its biosynthetic intermediates, the IRP1/ALAS2 feedback mechanism may have evolved in mammals to tightly link protoporphyrin synthesis with iron availability to guard against their deleterious overaccumulation.