Iron is a crucial metal for normal development, being required for the production of heme, which is incorporated into cytochromes and hemoglobin. The zebrafish chianti (cia) mutant manifests a hypochromic, microcytic anemia after the onset of embryonic circulation, indicative of a perturbation in red blood cell hemoglobin production. We show that cia encodes tfr1a, which is specifically expressed in the developing blood and requisite only for iron uptake in erythroid precursors. In the process of isolating zebrafish tfr1, we discovered two tfr1-like genes (tfr1a and tfr1b) and a single tfr2 ortholog. Abrogation of tfr1b function using antisense morpholinos revealed that this paralog was dispensable for hemoglobin production in red cells. tfr1b morphants exhibited growth retardation and brain necrosis, similar to the central nervous system defects observed in the Tfr1 null mouse, indicating that tfr1b is probably used by non-erythroid tissues for iron acquisition. Overexpression of mouse Tfr1, mouse Tfr2, and zebrafish tfr1b partially rescued hypochromia in cia embryos, establishing that each of these transferrin receptors are capable of supporting iron uptake for hemoglobin production in vivo. Taken together, these data show that zebrafish tfr1a and tfr1b share biochemical function but have restricted domains of tissue expression, and establish a genetic model to study the specific function of Tfr1 in erythroid cells.
Iron acquisition by developing erythroid cells is necessary to produce hemoglobin, which allows red blood cells to deliver oxygen to body tissues in exchange for carbon dioxide. The processes of iron uptake and intracellular transport are precisely regulated to protect cells from the toxic effects of free iron (Hentze et al., 2004). In vertebrates, the major pathway by which all cells obtain iron occurs through the interaction of transferrin receptor 1 (Tfr1) with its ligand transferrin (Tf) (Aisen, 2004; Andrews, 2000). Tfr1 is a type II membrane protein that facilitates iron uptake by binding to the iron carrier Tf, a plasma glycoprotein that shuttles iron absorbed from the diet. When bound, the Tf/Tfr1 complex is internalized by clathrin-mediated endocytosis. Acidification of the endosome compartment by proton pumps causes the release of iron from Tf. Iron is subsequently delivered to the cell cytoplasm through the action of the transmembrane protein divalent metal transporter 1 (DMT1), which transports iron out of the endosome by means of a proton-coupled process. In erythroid cells, most iron is carried to the mitochondria, where it is incorporated with protoporphyrin to produce heme. Recycling of the endosome restores the apo-Tf/Tfr1 complex to the cell surface, with the pH of the external milieu causing the release of apo-Tf, such that both Tfr1 and Tf are available for repeated cycles of use.
Tfr1 is highly expressed on differentiating erythrocytes, reflecting their substantial iron requirement to support hemoglobin synthesis (Ponka and Lok, 1999). The murine knockout of Tfr1 established that it was required for erythropoiesis and embryonic development (Levy et al., 1999). Mice homozygous for a null Tfr1 allele died of anemia before embryonic day (E) 12.5, and displayed marked growth retardation, edema and signs of tissue necrosis. Tfr1–/– mice also showed neurologic abnormalities, including kinking of their neural tubes and increased neuronal apoptosis. In addition, Tfr1+/– embryos evinced hypochromic, microcytic erythrocytes, consistent with iron deficiency. Analysis of chimeric mice generated with Tfr1–/– embryonic stem cells illustrated that Tfr1 was required postnatally for adult erythropoiesis and lymphopoiesis, as Tfr1–/– cells did not contribute to the bone marrow, spleen or thymus (Ned et al., 2003). Conversely, Tfr1–/– cells did incorporate into all non-hematopoietic tissues, indicating that other pathways of iron uptake were sufficient to permit their survival (Ned et al., 2003).
In addition to Tfr1, mammals also possess transferrin receptor 2 (Tfr2), a type II membrane protein similar to Tfr1. Tfr2 is highly expressed in liver hepatocytes and erythroid precursor cells, and can facilitate Tf-bound iron entry in vitro, but its function remains poorly understood (Fleming et al., 2000; Fleming et al., 2002; Kawabata et al., 1999; Kawabata et al., 2001; Trinder and Baker, 2003). Furthermore, while alternative mechanisms to Tf/Tfr1-mediated iron acquisition have been characterized in a variety of mammalian cell lines, their in-vivo roles remain to be elucidated. These include direct uptake of non-Tf-bound iron (NTBI) (Baker et al., 1998; Goto et al., 1983; Hodgson et al., 1995; Inman and Wessling-Resnick, 1993; Kaplan et al., 1991; Sturrock et al., 1990), Tfr1-independent uptake of Tf-bound iron (Chan et al., 1992; Thorstensen et al., 1995), and receptor-mediated uptake of ferritin (Gelvan et al., 1996; Konijn et al., 1994; Leimberg et al., 2003; Meyron-Holtz et al., 1999). Recently, the soluble protein 24p3/neutrophil gelatinase-associated lipocalin (Ngal) was found to deliver iron to developing mammalian kidney epithelial cells, with a pattern of cell binding and intracellular trafficking independent from that of the Tf/Tfr1, and may present one avenue for the cellular distribution of NTBI (Yang et al., 2002).
The inability of erythrocytes to obtain adequate iron for hemoglobin synthesis, as well as defects in heme or globin production, causes hypochromic, microcytic anemias in humans. While low dietary iron or blood loss is most frequently the underlying cause, inherited mutations in any number of genes required for hemoglobin synthesis have been attributed to such anemias (Andrews, 1999). We have utilized the zebrafish Danio rerio as a genetic model to study hemoglobin production during vertebrate erythropoiesis (Brownlie and Zon, 1999; Wingert and Zon, 2003). Zebrafish hematopoietic screens have resulted in the identification of nine complementation groups that display hypochromic, microcytic anemia: chardonnay, chianti, frascati, gavi, montalcino, sauternes, shiraz, weissherbst and zinfandel (K.D., P.F., R.A.W. and L.I.Z., unpublished) (Haffter et al., 1996; Ransom et al., 1996). chardonnay (cdy) has a mutation in Dmt1 (Donovan et al., 2002). The weissherbst (weh) mutant is unable to obtain maternal iron yolk stores due to a defect in ferroportin 1 (Fpn1), a transmembrane protein required to transport iron from the yolk into embryonic circulation (Donovan et al., 2000). The mutant sauternes (sau) has a defect in the enzyme aminolevulinate synthase-2 (Alas2), which functions at the first step in heme biosynthesis (Brownlie et al., 1998). Lastly, the zinfandel (zin) mutation has been mapped to the major globin locus, suggesting that zin results from disrupted globin function (Brownlie et al., 2003).
We report here the characterization of the chianti (cia) mutant phenotype and the cloning of the cia gene. We show that cia encodes an erythroid-specific isoform of transferrin receptor 1 (tfr1a) that is solely required for iron acquisition by differentiating erythrocytes. We found that zebrafish have undergone and retained a duplication of Tfr1 during teleost evolution, adding tfr1a and tfr1b to the growing list of gene duplicates in teleosts. To determine the function of zebrafish tfr1b, we utilized a morpholino knockdown approach and found that tfr1b is not required for erythropoiesis, but rather necessary for normal development of non-hematopoietic cells. These findings establish that the combined functions of tfr1a and tfr1b in zebrafish embryos recapitulate the role of mammalian Tfr1. Thus the cia mutant provides a useful genetic model to study the role of Tfr1 in erythropoiesis in the absence of other developmental defects.
Materials and methods
Zebrafish strains and maintenance
Zebrafish were maintained (Westerfield, 1993) and staged as described (Kimmel et al., 1995). ciatu25f was generated on the Tübingen (Tü) strain in a large-scale ENU mutagenesis screen (Haffter et al., 1996; Ransom et al., 1996), and crossed to standard wild-type AB for maintenance. ciahp327, ciahs019, and ciaiu089 were obtained by screening for zebrafish with hypochromia in a second large-scale ENU mutagenesis screen (K.D. and L.Z., unpublished); these alleles were generated and maintained on the Tü background. Matings with ciatu25f homozygotes and linkage analysis were used to assemble the cia complementation group.
o-dianisidine staining, in-situ hybridization and histological analysis
Detection of hemoglobin by o-dianisidine was performed as described (Ransom et al., 1996). We performed whole-mount in-situ hybridization with digoxigenin-labeled RNA probes as described (Thompson et al., 1998). Synthesis of βe1 globin probe was performed as described (Brownlie et al., 1998). Antisense and sense tfr1a and tfr1b probes were synthesized from the respective cDNA clones in pGEM-T easy vector. Adult peripheral blood and kidney tissue samples were isolated and Wright-Giemsa stained as described (Ransom et al., 1996; Brownlie et al., 1998).
Genetic mapping strains were created by mating cia AB or Tü heterozygotes to the polymorphic Dar or WIK strains. Embryos were collected from pairwise matings of mapping strain cia heterozygotes, and scored at 72 hours post fertilization (hpf) for hypochromia. Genomic DNA extraction from individual embryos and bulk segregant analysis were performed as described (Zhang et al., 1998) using primers designed to SSLP markers obtained from the Massachusetts General Hospital Zebrafish Server website (http://zebrafish.mgh.harvard.edu) and synthesized by Invitrogen.
Isolation of zebrafish Tfr genes, radiation hybrid mapping and mutation analysis
A 329 bp fragment of zebrafish tfr1a was isolated from zebrafish kidney cDNA library using degenerate primers, 5′-TACACMCCWGGMTTCCC-3′ (forward) and 5′-CCTGGRCCCCATGCATCCCTCTG-3′ (reverse). This fragment was used to screen zebrafish kidney cDNA gridded filters, which obtained partial clones of tfr1b. For each tfr1, a combination of 5′ and 3′ rapid amplification of cDNA ends (RACE) was used to determine the entire cDNA sequence; full-length clones were then obtained by RT-PCR from 36 hpf embryos, using 5′-ATGGATCAAGCCAGGACAACC-3′ (forward) and 5′-CTAAAGAGGTGAGCTGAAG-3′ (reverse) primers for tfr1a, and 5′-ATGGCAGGAACAATTGGTCAA-3′ (forward) and 5′-CTAGATTTCGTTGTCCAGGGA-3′ (reverse) for tfr1b. A partial fragment of tfr2 was cloned using online genome sequence data, and the open reading frame determined by 5′ RACE; a full-length clone was obtained by RT-PCR from 36 hpf embryos using 5′-ATGATGGACTCGGTCACAGGA-3′ (forward) and 5′-CTACAGCGGGTTGTCGATGTT-3′ (reverse). Radiation hybrid mapping was done with the following forward and reverse PCR primers: tfr1a, 5′-CAACAACATCCTCGTTCAG-3′ and 5′-CTCTGGACCCCGATCACC-3′; tfr1b, 5′-GCTTCGACATCGACCAGGTGC-3′ and 5′-GCACCTTGAAATGGGAGC-3′; and tfr2, 5′-CCCATCAGCAGATGAACCAACGAA-3′ and 5′-ACATAGGTGTGTTTACCGTTTTCC-3′. Mutation analysis was done by isolating cDNA from each cia allele at 36 hpf. tfr1a was amplified using the primers above, subcloned into pGEM-T Easy vector (Promega), and clones sequenced to determine the mutations.
cDNA overexpression constructs and morpholino designs
Full-length tfr1a and tfr1b cDNAs were subcloned into the pCS2+ vector and mRNA synthesized using SP6 mMessage mMachine (Ambion). mtfr1 and mtfr2 cDNA clones were a gift from Vera Sellers (Children's Hospital, Boston), and were subcloned into pCS2+. For expression in zebrafish ciaiu089 embryos, approximately 500-600 pg of synthetic mRNA encoding tfr1a, tfr1b, mtfr1 or mtfr2 was injected into 1-4-cell stage embryos.
Two morpholino antisense oligonucleotides targeting the tfr1a transcript were obtained from Gene-Tools: tfr1a-MO1 (5′-AGATGGTTGTCCTGGCTTGATCCAT-3′) was designed to the predicted start codon (underlined); tfr1a-MO2 (5′-ACACCTTCGAGTGGACGAAGTAACAC-3′) was designed to the splice donor of exon 13. Embryos were injected with 1 nl of either tfr1a-MO1 at 0.1 mg/ml or with tfr1a-MO2 at 1.5 mg/ml; to rescue tfr1a-MO1, embryos were co-injected with 500 pg of tfr1a cRNA. Morpholinos designed against the tfr1b transcript were as follows: tfr1b-MO1 (5′-CCAATTGTTCCTGCCATGGGATCTG-3′) was designed against the predicted start codon (underlined), tfr1b-MO2 (5′-AACAAAACTTACCATTCTGGAAAC-3′) and tfr1b-MO3 (5′-GCGGCTGTTTACCTATTAACAGAGG-3′) were designed against the respective splice donor and acceptor sites between exons 1 and 2. Embryos were injected with 1 nl of tfr1b-MO1 at 1.25 mg/ml or co-injected with tfr1b-MO1/MO2 at 0.5 mg/ml each; to rescue tfr1b-MO1, embryos were co-injected with 300 pg of tfr1b mRNA.
Iron-dextran microinjection assays
Intravenous iron injection at 48 hpf was performed as previously described (100 mg/ml, Sigma), such that each embryo received approximately 100 ng iron-dextran (Donovan et al., 2000). For 1-cell injections, iron-dextran was diluted to 10 mg/ml, and embryos injected with approximately 10 ng iron-dextran.
GenBank accession numbers
Zebrafish tfr1a, AY649363; zebrafish tfr1b, AY649364; and zebrafish tfr2, AY649365.
Phenotypic analysis of the cia mutant
Four autosomal recessive alleles of the cia mutant, ciahp327, ciahs019, ciaiu089 and ciatu25f, were obtained in large-scale ENU mutagenesis screens for zebrafish with defects in embryogenesis (K.D. and L.Z., unpublished) (Haffter et al., 1996; Ransom et al., 1996). The cia alleles displayed a phenotypic range with regard to the onset and severity of hypochromic anemia. Hypochromia was assessed by staining for hemoglobin with o-dianisidine, and red cell number was evaluated by in-situ hybridization for βe1 globin expression. Hypochromia was first evident at 36 hpf, when ciatu25f embryos exhibited a marked decrease in the number of cells stained with o-dianisidine, while ciahp327, ciahs019 and ciaiu089 had no detectable o-dianisidine-positive cells (Fig. 1A-C). At this time, all cia alleles exhibited wild-type numbers of circulating red cells (Fig. 1D-F). As development proceeded, ciatu25f embryos continued to show less severe hypochromia than the other alleles (Fig. 1G-I). Embryos of each cia allele exhibited a decrease in total red cell number compared with wild-type by 72 hpf; ciatu25f had the least severe anemia (50% of wild-type cell numbers); both ciahp327 and ciaiu089 had a more severe anemia (30% of wild type), and ciahs019 had the most severe anemia (<10% of wild type) (Fig. 1J-L). ciatu25f and ciaiu089 were both homozygous viable, while ciahp327 and ciahs019 were lethal between 7 and 10 days post fertilization (dpf). Hematopoietic commitment and early differentiation were found to progress normally in cia, as the gene expression of scl, gata1 and various embryonic globins were indistinguishable from wild type until 36 hpf; however, at 72 and 96 hpf, gata1 expression persisted in cia circulating cells, suggesting a block or delay in their differentiation (data not shown). By contrast, examination of lymphoid and myeloid cell gene expressions found no differences in cia mutants (data not shown). These analyses suggested that cia had a specific defect in late-stage erythroid differentiation.
The viable ciatu25f and ciaiu089 alleles survived to adulthood without symptoms of prolonged anemia, such as growth retardation, pallor or cardiomegaly. However, examination of peripheral blood from ciatu25f and ciaiu089 adults revealed that definitive erythrocytes in these animals were hypochromic and microcytic. Additionally, an increased number of undifferentiated red cells was present in peripheral circulation (Fig. 2A,B). Kidney smears from ciatu25f and ciaiu089 adults showed an increase in erythroblasts present in the kidney marrow, as well as hypercellularity when compared with wild-type adult zebrafish (Fig. 2C,D). These data demonstrate that the cia genetic lesion also perturbs adult erythropoiesis.
Zebrafish tfr1a is the gene defective in cia
To gain further insight into this phenotype, we isolated the cia gene by a candidate cloning strategy. Using bulk segregant analysis, the cia gene was mapped to linkage group (LG) 2. Linkage analysis of 996 meioses placed the cia locus between SSLP markers z4300 and z7634, approximately 2.7 cM north of the closest genetic marker, z7634 (Fig. 3). Examination of expressed sequence tags (ESTs) in this region revealed synteny to human chromosome 3, which suggested that Tfr1 was a candidate for cia (Fig. 3A). Degenerate PCR was used to isolate a 329 bp tfr1-like zebrafish clone. Using this fragment to screen a zebrafish kidney cDNA library, 29 partial clones were isolated. Sequencing of the library clones revealed that they encoded a different tfr1-like gene, which we designated as tfr1b. Radiation hybrid mapping of both zebrafish tfr1 genes placed tfr1a (the 329 bp clone) within the prospective cia locus on LG2, while tfr1b was mapped to a location on LG24 (Fig. 3A). Interestingly, examination of ESTs on the LG24 radiation hybrid panel in the vicinity of tfr1b also detected synteny to human chromosome 3 (Fig. 3A). This led us to speculate that teleosts had undergone a duplication of the tfr1 locus during the course of evolution. Comparative genomic analysis among vertebrates has strongly supported the hypothesis that the teleost lineage underwent a genome duplication event after the divergence of teleosts and tetrapods (Amores et al., 1998; Gates et al., 1999; Postlethwait et al., 1998; Woods et al., 2000). It has been estimated that at least 20% of the duplicated genes were maintained during subsequent teleost evolution (Postlethwait et al., 2000). With this in mind, in combination with the synteny we observed between the human Tfr1, zebrafish tfr1a and zebrafish tfr1b loci, we hypothesized that tfr1a and tfr1b represented the outcome of an ancestral teleost Tfr1 duplication.
We isolated the full-length cDNAs of zebrafish tfr1a and tfr1b by a combination of 5′ and 3′ rapid amplification of cDNA ends (RACE) PCR. Sequence comparison of the full-length clones revealed that the zebrafish Tfr1a and Tfr1b predicted peptides are 43.5% identical to each other, and respectively 39% and 35% identical to human Tfr1; comparison of the helical domain that is responsible for Tfr dimerization and Tf binding (Aisen, 2004) showed that they shared 44% and 49% identity with human Tfr1, respectively (Fig. 3B and data not shown). Blast search of the pufferfish genome similarly showed two Tfr1-like predicted proteins, further supporting an ancestral teleost tfr1 duplication event. Phylogenetic comparison among known vertebrate Tfr genes revealed that zebrafish and pufferfish tfr1-like genes were in fact most closely related, and clustered separately from the tfr1 of higher vertebrates (Fig. 3C). During these analyses, we discovered that zebrafish also possessed an ortholog to mammalian tfr2. We isolated the full-length cDNA of zebrafish tfr2, and found it clustered with greatest similarity to known human and mouse tfr2 (Fig. 3B,C).
To determine whether mutations in tfr1a were present in the various cia alleles, RT-PCR was used to obtain full-length cDNA clones of tfr1a from embryos of each allele. Each cia allele was found to harbor a mutation in either the tfr1a open reading frame or a conserved splice site (Fig. 3B). Sequence analysis of tfr1a in ciahp327 identified a T-to-A transversion at nucleotide +1889 that results in an I→N missense mutation at codon 630. ciahs019 were found to have a G-to-A transition at nucleotide +1970 that causes a G→D missense mutation at codon 657. The residues mutated in ciahp327 and ciahs019 are both located in the Tfr1a regions of the helical domain involved in Tf binding (Buchegger et al., 1996; Cheng et al., 2004; Lawrence et al., 1999). Mutagenesis studies have further localized the Tf/Tfr binding interface to include a conserved RGD sequence, and mutation of the glycine in particular severely abrogates Tf binding (Dubljevic et al., 1999; Giannetti et al., 2003; Liu et al., 2003; West et al., 2001). As the ciahs019 mutation occurs at this particular glycine, we predict it may directly eliminate Tf binding, although this tripeptide in Tfr1a is replaced by QGS residues. The residue mutated in ciahp327 is located in α helix 1 of the helical domain, adjacent to a lysine residue critical for Tf binding, and may similarly disrupt interaction with the ligand (Giannetti et al., 2003; Liu et al., 2003). We found ciatu25f possessed a G-to-A nucleotide transition at the exon 13 splice donor site that results in inclusion of a 90 bp intron with a premature stop codon, which eliminates the entire Tfr1 helical domain as well as 94 residues (approximately 30%) of the protease domain. RT-PCR of tfr1a from ciatu25f detected the presence of both the wild-type tfr1a cDNA and the mis-spliced variant (data not shown). This was consistent with the ciatu25f hypomorphic phenotype, as it suggests that to some extent, ciatu25f are capable of synthesizing normal Tfr1a. Lastly, we found ciaiu089 had a T-to-C transition at nucleotide +946, with a resulting F→L mis-sense mutation at codon 316. The F316 residue is located on an apical domain loop that is altered in conformation upon Tfr ligand binding, and although a mutation in this loop may cause its local destabilization, it is not clear what the precise consequence might be for overall Tfr function.
We next ascertained the expression pattern of each zebrafish tfr1 during embryogenesis. Whole-mount in-situ hybridization of zebrafish embryos showed tfr1a to be highly expressed in the developing blood, as marked by βe1 globin expression (Fig. 4A-H). tfr1a expression was first detected at 12 hpf in the ventral mesoderm, which converges to form the hematopoietic intermediate cell mass, the zebrafish intraembryonic blood island (data not shown). Blood-specific expression of tfr1a was maintained until 36 hpf in circulating primitive erythrocytes. By contrast, tfr1b was found to be expressed ubiquitously throughout embryogenesis (Fig. 4C,F,I). Notably, expression was not elevated above the level of ubiquitous expression in either developing or circulating primitive erythrocytes.
To provide further evidence that a defect in tfr1a was responsible for the cia phenotype, we used morpholino (MO) antisense oligonucleotides to inhibit either tfr1a mRNA translation or splicing and effect a transient genetic knockdown (Nasevicius and Ekker, 2000). MOs were designed to the tfr1a start site (MO1) and to the tfr1a exon 13 splice donor (MO2). When injected into wild-type zebrafish embryos, both tfr1a MO1 and MO2 prevented erythrocyte hemoglobin synthesis, resulting in embryos that displayed hypochromic erythrocytes that were o-dianisidine negative (Fig. 5A-C and data not shown). Co-injection of tfr1a mRNA was able to rescue the hemoglobin defect in 18% (30/167) of the animals injected with tfr1a MO1, demonstrating that the phenotype in tfr1a morphant embryos was specific (Fig. 5D).
Lastly, we confirmed the ability of tfr1a to rescue hemoglobin production in cia mutants by overexpression. Wild-type tfr1a mRNA was injected into ciaiu089 homozygous embryos, resulting in a partial rescue of o-dianisidine positive cells at the 36-40 hpf stage (Fig. 5E; Table 1). From these experiments we concluded that mutations in tfr1a were causal for the cia phenotype.
Provision of iron rescues cia hypochromia
To demonstrate the requirement of tfr1a for iron uptake in cia erythroid precursors, we attempted to rescue embryonic hypochromia with two methods of iron delivery. First, we tested whether injection of ciaiu089 embryos with iron-dextran at the 1-cell stage could remedy their hypochromia. We hypothesized that by overloading the embryo cytoplasm with iron-dextran at the 1-cell stage, we would deposit a supply of usable iron to be distributed during subsequent embryo cleavage, such that the cytoplasm of all cells would directly receive iron and circumvent the need for tfr1a function. We found that injection of 1-cell staged ciaiu089 with iron-dextran robustly rescued hypochromia at 40 hpf (138/297; 46%) (Fig. 6A). By facilitating the delivery of cytoplasmic iron to erythrocytes, we have demonstrated that the biochemical function of Tfr1a is to facilitate iron uptake across the zebrafish red cell membrane.
Second, previous work had found that intravenous injection of iron-dextran at 48 hpf rescued hemoglobin production over subsequent days in zebrafish mutants with inadequate iron present in circulation if the primitive erythrocytes were competent to uptake and utilize iron (Donovan et al., 2000). Thus we used intravenous injection of iron-dextran to assess whether (excess) available iron would ameliorate the cia hemoglobin defect. We anticipated that intravenous iron would not rescue defective hemoglobination in cia, due to the loss of Tfr1a function. Injection of iron-dextran into the circulation of ciaiu089 or ciatu25f at 48 hpf was in fact unable to rescue hypochromia when assessed over subsequent days (0/87; 0%) (Fig. 6B). The failure of this late-stage iron injection to rescue cia hemoglobin synthesis shows the inability of the cia erythrocytes to acquire iron when the circulation had been saturated with a usable iron source. In addition, this result highlights that erythrocytes only acquire appreciable cellular iron levels through the function of Tfr1a.
Zebrafish tfr1b is used for iron uptake by non-hematopoietic tissues
Based on the in-situ hybridization, as well as RT-PCR analyses from embryos between 24 hpf and 6 dpf, it was evident that tfr1b was expressed throughout embryogenesis (Fig. 4 and data not shown). To investigate the role of tfr1b during development, we used antisense MOs to evaluate loss of tfr1b function. In particular, we were interested in whether abrogation of Tfr1b would affect hemoglobinization during erythroid cell differentiation. A MO designed against the tfr1b translational start site (MO1) was injected into wild-type zebrafish embryos at the 1-cell stage and the embryos were examined throughout development. At 18-24 hpf, tfr1b MO1-injected animals exhibited brain necrosis and growth delay. By 36-48 hpf, embryos were still developmentally delayed, and could be grouped into three classes according to their overall growth progression and the severity of their nervous system necrosis (Table 2, Fig. 7). Class I injected animals (7.8%) exhibited a moderate growth delay in comparison with uninjected wild-type embryos (Fig. 7A-D). Class II injected embryos (87.5%) displayed severe growth retardation, being much smaller with a markedly curved trunk and tail (Fig. 7E,F). Class III injected embryos (4.7%) included the morphants with extreme growth retardation (Fig. 7G,H). All Class II and III affected morphants died before 5 dpf. Despite this, all tfr1b morphant groups underwent normal hemoglobinization, with visibly red blood in circulation and o-dianisidine concentrations indistinguishable from uninjected wild-type siblings. Similar results were observed with co-injection of MOs targeted to the splice donor and acceptor sites at the junction between exons 1 and 2 (data not shown).
To exclude the possibility that the tfr1b MO phenotype was due to MO toxicity, we attempted to rescue the generalized growth defects and nervous tissue necrosis with concomitant tfr1b mRNA overexpression. Co-injection of tfr1b MO1 and tfr1b cRNA into 1-cell stage wild-type embryos resulted in a greater number of Class I animals (30.6%) (Fig. 7I,J; Table 2). The ability of tfr1b overexpression to ameliorate the combination of growth and necrosis effects of tfr1b MO1 demonstrated that the observed embryo phenotype was reflective of tfr1b loss of function. From these experiments, we conclude that zebrafish tfr1b is not necessary for erythroid iron uptake. The tfr1b MO phenotype and the ubiquitous tfr1b expression pattern suggest that Tfr1b is most likely used by all non-erythroid cells for iron assimilation.
Multiple Tfr genes are sufficient to rescue hypochromia in cia
We next determined if tfr1b could compensate for the loss of tfr1a function in cia erythrocytes. We overexpressed tfr1b in ciaiu089 mutants and found that hemoglobin production was partially rescued in 27% of animals injected (Fig. 5F; Table 1). The number of ciaiu089 rescued and the degree of rescue were similar to that observed when tfr1a was overexpressed. These results confirm that Tfr1b functions to deliver cellular iron. We also examined expression of tfr1b in cia mutants to assess if its expression might be altered, and hence partly accountable for the homozygous viability of ciaiu089 and ciatu25f. However, we detected no changes in tfr1b expression in any cia allele by whole-mount in-situ hybridization (data not shown). Thus, taken together with the morpholino data, this shows that while tfr1b is capable of iron delivery into erythrocytes, it is not normally utilized by developing erythroid cells.
As Tf receptors are conserved in structure throughout vertebrate evolution, we wondered if multiple family members would be able to rescue cia when similarly overexpressed in the embryo. In support of a common biochemical function, we found that the overexpression of mouse tfr1 mRNA partially rescued hemoglobin synthesis in 17% of ciaiu089 embryos (Fig. 5G, Table 1). Again, the number of cia embryos rescued and quantity of cells per embryo in which hemoglobin was detected were similar to the overexpression experiments conducted with tfr1a and tfr1b. Although the function of Tfr2 in body iron metabolism has yet to be elucidated, mammalian Tfr2 is capable of binding and transporting Tf-bound iron in vitro (Kawabata et al., 1999; Kawabata et al., 2001; West et al., 2000). Based on these data, we tested if overexpression of Tfr2 would rescue hemoglobin production in cia. Injection of ciaiu089 mutants with mouse tfr2 mRNA at the 1-cell stage partially rescued hypochromia in 27% of mutants (Fig. 5H; Table 1). Thus mammalian Tfr2 was also able to compensate for loss of tfr1a function in erythroid cells. This series of Tfr overexpression experiments suggests that the presence of any number of known Tf receptors on a differentiating erythrocyte can facilitate iron uptake adequate for the production of hemoglobin.
Our study of the cia mutant illustrates that a defect in tfr1a function specifically causes a hypochromic, microcytic anemia in zebrafish. We have shown that tfr1a is required for normal hemoglobin synthesis during zebrafish erythropoiesis. Our results demonstrate that tfr1a is a transferrin receptor exclusively used by erythrocytes, since tfr1a is highly expressed in the developing blood, each cia allele exhibits a blood-specific phenotype, and transient knockdown of tfr1a using morpholinos resulted in hypochromic blood in otherwise normal embryos. The injection of iron-dextran at the 1 cell-stage rescued the hemoglobin defect in cia embryos, showing that loss of tfr1a prevents erythrocytes from acquiring iron. In all other respects, cia erythrocytes are capable of normal hemoglobin production. Thus the biochemical role of zebrafish tfr1a to mediate cellular iron acquisition is conserved with that of vertebrate Tfr1s. In contrast to tfr1a, zebrafish tfr1b is not required for erythropoiesis. Rather, tfr1b function is indispensable for proper growth and development of non-hematopoietic tissues in the embryo. The sum of zebrafish Tfr1a and Tfr1b functions equate to that of the single mammalian Tfr1. Although other pathways of cellular iron delivery have been characterized, they are not sufficient to compensate for the role of tfr1a in erythropoiesis or the role of tfr1b during ontogeny.
Functional equivalence among vertebrate Tfr genes in vivo
The overexpression of several Tfr genes in cia embryos was shown to attenuate the cia hypochromic phenotype. These results illustrate that when expressed in differentiating erythroid cells, a number of Tf receptors are capable of mediating iron uptake sufficient for hemoglobin biosynthesis. Our data specifically show the functional equivalence between zebrafish tfr1a, zebrafish tfr1b, mouse Tfr1, and mouse Tfr2 in the developing zebrafish. It is particularly noteworthy that cia could be rescued with mouse Tfr2, as this is the first in-vivo illustration of Tfr2 facilitating erythroid iron uptake. This finding emphasizes that functional differences between mammalian Tfr1 and Tfr2 result from differences in the spatiotemporal expression of these respective genes.
Complementary roles of the zebrafish tfr1 duplicate genes during embryogenesis
The phenomenon of an erythroid-specific transferrin receptor in zebrafish is most likely explained by a teleost genome duplication event postulated to have happened following the split between teleost and tetrapod lineages (Amores et al., 1998; Gates et al., 1999; Postlethwait et al., 1998; Woods et al., 2000). Based on the presence of gene duplicates discovered in multiple teleosts, including zebrafish, pufferfish and medaka, the genome duplication is predicted to have occurred at least 100 million years ago in a teleost ancestor that pre-dates the radiation of teleosts (Amores et al., 1998; Aparicio et al., 1997; Christoffels et al., 2004; Gates et al., 1999; Naruse et al., 2000; Santini and Tyler, 1999; Taylor et al., 2003; Wittbrodt et al., 1998). Our finding that the summation of zebrafish Tfr1a and Tfr1b functions equate to that of mammalian Tfr1 is consistent with the subfunctionalization model of gene duplicate preservation, in which gene copies retain part of the original gene's function (Force et al., 1999; Ohno, 1970; Prince and Pickett, 2002; Zhang, 2003). Common with other examples of subfunctionalization, the respective expression patterns of the tfr1a and tfr1b have diverged, such that together they recreate the expression of the single ancestral gene, even though in this case the proteins have interchangeable biochemical functions (Bruce et al., 2001; Chiang et al., 2001; de Martino et al., 2000; Dorsky et al., 2003; Lister et al., 2001; Nornes et al., 1998; Oates et al., 1999; Pfeffer et al., 1998). We speculate that non-overlapping tfr1a and tfr1b expression was made possible by the evolution of different sets of regulatory elements for each duplicate. Future work in defining the promoter elements of tfr1a and tfr1b will better characterize the potential differences in the regulation and developmental expression of these genes.
Zebrafish as a vertebrate model to study the metabolism of iron and other essential metals
As part of our assessment of tfr1a function in erythropoiesis, we developed a means to distinguish between cell extrinsic and intrinsic defects in iron utilization. We utilized a previously characterized method of zebrafish intravenous iron-dextran injection to determine if low plasma iron was a factor in the failure of cia erythrocytes to hemoglobinize normally (Donovan et al., 2000). In this assay, the ability to rescue hemoglobin synthesis with intravenously-supplied iron demonstrates that the erythrocytes are fully capable of iron uptake, intracellular trafficking and metabolism; in direct contrast, the inability to rescue indicates that the presence of a defect(s) intrinsic to the mechanism of cellular iron uptake or utilization in erythrocytes is present. Our de-novo method of iron-dextran injection at the 1-cell stage then serves to test whether developing erythrocytes can traffic and metabolize intracellular iron subsequent to iron uptake. With this assay, we believe that provision of excess iron to the embryo cytoplasm before the onset of cleavage acts to bypass the later necessity for erythroid iron internalization, because all cells in the embryo have been saturated with an excess of usable iron. This novel combination of iron-dextran injection assays is a valuable tool that can now be employed to categorize hypochromic mutants with unknown gene defects currently being studied in our laboratory.
We found it surprising that injection of iron-dextran at the 1-cell stage was relatively non-toxic to the embryo, and we expect this forecasts broader applicability of similar assays. Single cell injection of any number of conjugated trace metals, such as copper or zinc, could be utilized to investigate their function and metabolism in a developmental setting. As we have done, the method could be applied to track various maternal yolk storage components, and could be applied to characterize the defect(s) in genetic mutants. Zebrafish present a unique opportunity to better understand the transit and utilization of iron and other metals during embryogenesis, and such studies in genetic mutants will enable investigation of the pathophysiology of numerous disease states.
In recent years, elucidation of the defects in several zebrafish with hypochromic anemia, in conjunction with the ongoing development of assays to understand their biology, have made significant contributions to the understanding of vertebrate iron metabolism. In this report we have presented evidence that cia represents a specific defect in erythrocyte iron uptake due to an ancestral duplication of the teleost tfr1 locus. Thus cia provide a model to further define the role of Tfr1 in erythropoiesis without a panorama of complicating tissue defects. Furthermore, the zebrafish system provides the ability to implement targeted genetic and chemical screens that could identify additional pathways with a role in the maintenance of iron homeostasis.
Tübingen 2000 Screen Consortium
F. Bebber van, E. Busch-Nentwich, R. Dahm, H. G. Frohnhöfer, H. Geiger, D. Gilmour, S. Holley, J. Hooge, D. Jülich, H. Knaut, F. Maderspacher, C. Neumann, T. Nicolson, C. Nüsslein-Volhard, H. Roehl, U. Schönberger, C. Seiler, C. Söllner, M. Sonawane, A. Wehner, C. Weiler and B. Schmidt at the Max-Planck-Institut für Entwicklungsbiologie, Spemannstrasse 35, 72076 Tübingen, Germany.
U. Hagner, E. Hennen, C. Kaps, A. Kirchner, T. I. Koblizek, U. Langheinrich, C. Metzger, R. Nordin, M. Pezzuti, K. Schlombs, J. deSantana-Stamm, T. Trowe, G. Vacun, A. Walker and C. Weiler at Artemis Pharmaceuticals/Exelixis Deutchland GmbH, Neurather Ring 1, S51063 Köln, Germany.
We thank members of the Zon laboratory for review of this manuscript, D. Giarla for excellent administrative support, V. Sellers for the gift of mouse Tfr1 and Tfr2 clones, and A. Giannetti and P. Bjorkman for helpful discussion of Tfr1 functional domains. R.A.W. is supported by Hematology Training Grant, T32 HL07623, is a Harvard University Fellow of the Albert J. Ryan Foundation, and thanks G. Wingert and V. Wingert for their tremendous love and support. L.I.Z. is an Investigator of the Howard Hughes Medical Institute. This work was supported by HHMI and NIH grants HL073427, DK53298 and HL32262.
- © 2004.