The ferrireductase paraferritin contains divalent metal transporter as well as mobilferrin

Jay N. Umbreit, Marcel E. Conrad, Lucille N. Hainsworth, Marcia Simovich


Inorganic iron can be transported into cells in the absence of transferrin. Ferric iron enters cells utilizing an integrin-mobilferrin-paraferritin pathway, whereas ferrous iron uptake is facilitated by divalent metal transporter-1 (DMT-1). Immunoprecipitation studies using antimobilferrin antibody precipitated the previously described large-molecular-weight protein complex named paraferritin. It was previously shown that paraferritin functions as an intracellular ferrireductase, reducing ferric iron to ferrous iron utilizing NADPH as the energy source. It functions in the pathway for the cellular uptake of ferric iron. This multipeptide protein contains a number of active peptides, including the ferric iron binding protein mobilferrin and a flavin monooxygenase. The immunoprecipitates and purified preparations of paraferritin also contained DMT-1. This identifies DMT-1 as one of the peptides constituting the paraferritin complex. Since paraferritin functions to reduce newly transported ferric iron to ferrous iron and DMT-1 can transport ferrous iron, these findings suggest a role for DMT-1 in conveyance of iron from paraferritin to ferrochelatase, the enzyme utilizing ferrous iron for the synthesis of heme in the mitochondrion.

  • iron absorption
  • calreticulin
  • flavin monooxygenase

iron is vital for the survival of all cells. Iron is acquired from foodstuffs and transported into the intestinal absorptive cells as either inorganic iron or heme (3). Little is known about the mechanism of transport of heme iron. The transport of nonheme iron has been the object of considerable recent study (6-8, 11, 14). Unlike most nucleated cells in the body, the luminal surface of the enterocyte lacks transferrin receptors (21). This makes the transport of iron by intestinal cells different from other cells. Although nonintestinal cells are capable of transporting iron in the absence of transferrin (1, 6, 9, 15, 19, 20, 23, 24) by the same pathway used in the intestine, plasma iron is mostly transported bound to transferrin.

Recent data (4) indicate that inorganic iron can enter cells in the absence of transferrin via two different pathways. Ferric iron enters cells in association with the integrin mobilferrin pathway (IMP). This pathway appears to be unique for ferric iron and is not shared with other metals. It was elucidated by isolating iron radiolabeled proteins after intraduodenal instillation of59Fe3+ or by incubating tissue culture cells with 59Fe3+ citrate. IMP utilizes a β3-integrin and mobilferrin (a homologue of calreticulin). Once inside the cells, ferric iron is reduced by a large protein complex (520 kDa) named paraferritin (25). Paraferritin has been shown to contain β3-integrin, mobilferrin, flavin monooxygenase, and β2-microglobulin (25, 26). The ferrireductase activity of paraferritin converts the ferric iron to ferrous iron intracellularly utilizing NADPH, so that the ferrous iron is available for incorporation into heme by means of the enzyme ferrochelatase.

In contrast, ferrous iron is transported into the cell via a transmembrane pump protein named divalent metal transporter (DMT-1, also called Nramp-2 or DCT) (11, 14). This pathway was discovered by expression cloning of proteins upregulated in iron deficiency and was found to be defective in the Belgrade rat and mk mouse (11, 14), which phenotypically have iron deficiency anemia. This pathway is shared with certain other divalent metal cations, including manganese, nickel, and cobalt (4, 14). DMT-1 binds radioactive ferrous iron (4). DMT-1 does not appear to be involved in the initial transport of ferric iron across the plasma membrane. Although the major focus of the research on the IMP and the DMT-1 pathways has been to document and explore their role in inorganic iron uptake by cells, there is accumulating data that these pathways have additional roles in intracellular iron metabolism (4, 5).

Although the ferric pathway (IMP) and the ferrous pathway (DMT-1) are clearly distinct and different modes for the entry of inorganic iron into the cell (4), the conversion of ferric iron into ferrous iron by paraferritin in the cytoplasmic compartment would suggest the possibility of some interconnection between the diverse pathways somewhere downstream from the entry mechanism. This study will show that DMT-1 is a component of paraferritin.



Human duodenal homogenates were freshly obtained during Whipple surgery for pancreatic adenocarcinoma. The mucosa was scraped free of the serosa and incubated with radiolabeled ferric citrate (59Fe3+) for 10 min before homogenization of the cells. Paraferritin was isolated from the supernatant as described previously (25). In brief, the homogenate was centrifuged at 47,000 g for 30 min and the supernatant was collected. Ammonium sulfate was added to 60%, and the pellet was collected, dialyzed, and chromatographed on an AcA 34 gel filtration column. The appropriate fractions were then subjected to DEAE chromatography with a salt gradient. The fractions collected constituted the purified protein. Duodena from rats were obtained following incubation with radiolabeled ferric citrate in isolated duodenal loops in vivo for 10 min (8). The isolated segments were excised and opened, and the mucosa was washed free of luminal contents. Gut mucosal scrapings were homogenized, and paraferritin was isolated as previously described (8, 25).

Tissue culture.

K562 erythroleukemia cells were obtained from the American Type Culture Collection (Rockville, MD). A human kidney cell line (293H; Life Technologies, Gaithersburg, MD) was used for certain experiments. The cells were grown and prepared for study as described previously (6). K562 cells were grown in suspension in RPMI 1640 and 10% heat-inactivated fetal bovine serum (Life Technologies). Cells were incubated in a 5% CO2 incubator (ShelLab, Cornelius, OR) and grown to a density of 5 × 106 cells/ml. 293H cells were grown in SFMII (Life Technologies).


Two polyclonal antibodies were used in these studies raised against a polypeptide sequence of human DMT-1 [amino acids 325–339 (KTNEQVVEVCTNTSS) or amino acids 32–45 (SNSSLPHSTGDSEEPFTT)] (4). Each rabbit was immunized with the polypeptide construct in Ribi adjuvant at 2-wk intervals for three multiple subcutaneous inoculations. Development of antibody against the polypeptide construct was monitored in rabbit serum by an ELISA method. Western blots of whole cell lysates of K562 cells were obtained and electrophoresed on 7.5% SDS-polyacrylamide gels. Western blots were prepared, and a single band (58 kDa) was detected using the antibody to DMT-1, developed with secondary antibody with horseradish peroxidase and chemiluminescence (Amersham, Arlington Heights, IL) as described previously (4). Inundation of the blot with the immunizing peptide (2 mg/ml) before development with the anti-DMT-1 ablated the band (with either of the two antibodies and corresponding peptide), but other nonspecific peptides had no effect. Polyclonal chicken anticalreticulin antibody (PA 1–903) was obtained from Affinity BioReagents (Golden, CO).


K562 cells and 293H cells were provided with fresh media on the day before study. Then 10 ml of the cellular suspension was centrifuged at 1,600 rpm (470 g) for 5 min at 4°C (BT6000D; Sorvall, Wilmington, DE). The cells were resuspended in 10 ml of PBS with three washes. The cells were resuspended in buffer (10 mM HEPES, 10 mM NaCl, pH 7.4) with 1% Triton X-100. The suspension was placed on a rotator for 1 h at 4°C. The mixture was centrifuged at 2,000 rpm (840g) for 5 min at 4°C, and the supernatant was diluted 1:1 with buffer. It may be noted that under these conditions membrane-bound DMT-1 is not solubilized (4). For immunoprecipitates utilizing antimobilferrin, anticalreticulin antibody (5 μl) obtained from chicken (utilizing the COOH-terminal sequence #PAI-903 from Affinity BioReagents) or for the anti-DMT immunoprecipitates (using the antibody to amino acids 32–45), 5 μl of the previously described antibody was added to the solution and incubated on a rotator for 1 h (4°C). Then 50 μl of protein G-Sepharose (diluted 1:1 with buffer) (Amersham Pharmacia Biotech, Piscataway, NJ) was added and incubated overnight on a rotator (4°C). The slurry was centrifuged at 840 g for 5 min, and the precipitate was resuspended and washed in five changes of a buffer containing 5 ml of Nonidet P-40, 0.5 g SDS, 1.3 g HEPES, and 4.5 g NaCl in 500 ml of water. The final pellet was washed in 10 mM HEPES, 10 mM NaCl (pH 7.4) and centrifuged at 840 g for 5 min. The precipitate was heated at 100°C for 5 min in Laemmli sample buffer (1 ml), and an aliquot was loaded on a 10% acrylamide gel (Criterion; Bio-Rad, Hercules, CA). After electrophoresis, the gel was transferred to Immobilon-P (Millipore, Burlington, MA). For detection of DMT-1, the Western blot was developed with the primary antibody (1:1,000) added in blocking buffer (5% instant skimmed milk with 0.02% Tween 20 and PBS, pH 7.4) for 1 h on a shaker. The blot was washed with PBS with three changes of the buffer. The secondary antibody (donkey anti-rabbit, horseradish peroxidase; Amersham) in blocking buffer (1:1,000) was incubated for 1 h in a shaker. This was washed with PBS for three changes. The blot was developed using chemiluminescence (ECL Plus; Amersham) according to the manufacturer's instructions (rapid immunodetection method). For detection of mobilferrin in the DMT-1 immunoprecipitates, the primary antibody was the mixed chicken antibody to the COOH terminal (PAI-903) and NH2 terminal (PAI-902) at 1:800. The secondary antibody the was horseradish peroxidase-donkey anti-chicken from Jackson ImmunoResearch Laboratories used at 1:1,000. Otherwise the protocol was identical.

For detection of the transferrin receptor, whole cell lysate from 1,000,000 cells was used. The immunoprecipitate was obtained utilizing the chicken anticalreticulin (COOH terminal). The precipitates were collected in 1 ml of sample buffer. From this, 30 μl was loaded to the gel. The Western blot was developed in a primary antibody of 1:200 antitransferrin receptor polyclonal antibody (CD71, #SC9099 from Santa Cruz Biotechnology, Santa Cruz, CA, and identical results were obtained with the monoclonal CD 71 antibody from Novocastra Laboratories), and the secondary antibody was horseradish peroxidase rabbit anti-mouse used at 1:1,000 (Amersham Pharmacia Biotech).

Biotinylation of K562 and 293H cells.

Cells were grown in tissue culture to a concentration of 1 × 106 cells/ml. Fifteen milliliters of cells were centrifuged at 1,600 rpm (470 g) for 5 min at 4°C. Cells were resuspended and centrifuged in PBS and brought to 0.5 ml with PBS. Fifty microliters of biotin reagent (NWS-LC-Biotin, 1.8 mg/ml distilled water; Pierce, Rockford, IL) was added to the suspension and incubated at 4°C for 2 h on a rotator. The mixture was centrifuged at 11,000 rpm (12,000 g) for 15 min at 4°C. PBS (1.5 ml) was added to 0.5 ml of the supernatant. This was dialyzed with 4 l of PBS using a Slide-A-Lyzer (Pierce) overnight with one change of buffer. The dialysate was recovered and incubated with 5 μl chicken anticalreticulin (direct to the COOH terminus; Affinity BioReagents) on a rotator for 2 h (4°C). Protein G-Sepharose was added, and the procedure was completed as described above.

Ferrireductase activity.

The ferrireductase activity was measured in mobilferrin and DMT immunoprecipitates. The precipitates were collected on beads, but instead of suspending in Laemmli sample buffer they were resuspended in 1 ml of 10 mM HEPES, pH 7.4, and 10 mM NaCl. The assay was carried out as previously described (25). In brief, in a final volume of 1 ml, 60 μl of 10 mM ferric citrate, 100 μl of 0.5 M HEPES, pH 7.4, and 10 μl of 45 mM NADPH were added. The immunoprecipitates on the Sepharose beads were added, and the change in optical adsorption at 340 nm was measured in a response recording spectrophotometer (Gilford Instruments, Oberlin, OH). Only linear initial rates with a standard deviation of <0.0030 were used. Controls without iron added were subtracted to obtain the iron-dependent rate.


Homogenates of the tissue culture cell line K562 (a human erythroleukemia cell line known to synthesize hemoglobin) were labeled with biotin and treated with antibody to mobilferrin (calreticulin). After adsorption onto protein G beads and extensive washing, the resulting immunoprecipitate was subjected to SDS electrophoresis. The gel was blotted to Immobilon-P, and the protein bands were developed with streptavidin-horseradish peroxidase using a chemiluminescent method. The banding pattern exemplifies paraferritin (Fig.1). The diffuse set of bands between 60 and 80 kDa contained the peptides for mobilferrin and flavin monooxygenase (FMO) as shown previously (25). However, other proteins were clearly present. The upper two bands correspond to the peptides of the integrin component (95 and 150 kDa). The paraferritin complex immunoprecipitated because it contained mobilferrin, and the binding between the peptides of paraferritin was sufficiently strong that the complex was recovered substantially intact.

Fig. 1.

Immunoprecipitation of paraferritin. K562 cells were lysed by freeze thawing and uniformly labeled with biotin. Homogenates of K562 cells were made with 1% Triton X-100, centrifuged to remove insoluble debris, diluted to 0.5% detergent, and incubated with anticalreticulin (mobilferrin). The immune complex was collected on protein G beads, washed in detergent containing buffer, then eluted with SDS gel sample buffer at 100°C for 5 min. The supernatant was subjected to SDS electrophoresis, blotted to Immobilon-P, and developed with streptavidin-horseradish peroxidase chemiluminescence as described in materials and methods. The banding pattern seen was characteristic of paraferritin and showed the expected multiple bands (25, 26). The confluent set of bands at ∼60–70 kDa represent flavin monoxygenase, divalent metal transporter-1 (DMT-1), and mobilferrin, as previously shown (25).

Identical immunoprecipitations utilizing antibody to mobilferrin were performed with homogenates from K562 and 293H. After SDS gel electrophoresis and blotting onto Immobilon-P, the gels were developed with primary antibodies to DMT-1. This antibody has previously been described. It is specific for DMT-1, and it detects the ∼60 kDa form of the protein. Both cells lines demonstrated that DMT-1 was present in the paraferritin fractions isolated by immunoprecipitation, strongly suggesting that there was an interaction between the DMT-1 and the paraferritin complex (Fig. 2). The ∼60 kDa band was typical for the DMT-1 identified in these cells and is usually accompanied by the smaller-molecular-weight fragment seen in the figure, whose significance is not known. This specific antibody developed none of the other peptides known to be present in these immunoprecipitates (Fig. 1). The higher-molecular-weight form of DMT-1, thought to be due to glycosylation (11), was not detected on these gels.

Fig. 2.

Western blot of immunoprecipitates of paraferritin demonstrate DMT-1. K562 cells (right) and 293H cells (left) were treated as described in Fig. 1, except the biotinylation step was omitted. After SDS electrophoresis and transblot to Immobilon-P, the blots were developed with anti-DMT-1 antibody as described inmaterials and methods using the antibody to amino acids 325–339 of DMT-1. Equivalent results were obtained using a second antibody specific for the amino acid sequence 32–45. A band was seen at the expected molecular mass for DMT-1, and an accompanying smaller-molecular-weight fragment often seen with DMT-1 was also apparent. No other bands were developed.

Immunoprecipitates of K562 homogenates utilizing the antibody to DMT-1 were also analyzed by Western blots, developing the blot with the antibody to mobilferrin (Fig. 3). A band with the molecular weight expected for mobilferrin was visualized, and no other bands were seen. DMT-1 antibody also appeared to precipitate paraferritin with the associated mobilferrin component.

Fig. 3.

Western blots of K562 homogenates immunoprecipitated with anti-DMT-1. The paraferritin was precipitated as described in Fig. 2except that antibody to DMT-1 (amino acids 32–45) was used. This previously described antibody is specific for DMT-1 and was produced against a synthetic peptide. The precipitates were electrophoresed in SDS, transblotted, and developed with the antimobilferrin antibody derived from chicken. The lanes represent 1, 5, and 10 μl of the sample obtained from the beads. A band at the expected molecular mass for mobilferrin was detected.

It was still possible that some plasma membrane or endosomal fragment had remained intact in the homogenate and that an entire membrane fragment was precipitated. Although the banding pattern seen in the stained precipitates (Fig. 1) showed only paraferritin and not whole membranes, a small contamination was still possible. Immunoprecipitates using antibody to mobilferrin were analyzed on Western blots developed with polyclonal antibody to the transferrin receptor (Fig.4). Whereas whole homogenates showed the presence of a band corresponding to the transferrin receptor, the mobilferrin immunoprecipitates did not reveal any transferrin receptor. Thus contamination of paraferritin isolates by intact membrane fragments appeared unlikely.

Fig. 4.

Western blots of K562 homogenates immunoprecipitated with antimobilferrin and developed with antitransferrin receptor. To demonstrate that the immunoprecipitates did not contain membrane or endosomal fragments, the blots obtained as described in Fig. 2 were developed with polyclonal antibody to the transferrin receptor (TfR). The whole homogenate, solubilized in SDS, which should contain both membranes and endosomes, clearly showed the presence of the transferrin receptor (lane C), whereas in the immunoprecipitates no transferrin receptor could be demonstrated (lane P).

The immunoprecipitates retained ferrireductase activity. The mobilferrin immunoprecipitates had a rate of ferric iron-dependent NADPH oxidation of 20 ± 5.0 nM/min per 103cells and the DMT-1 immunoprecipitates had a rate of NADPH oxidation of 26 ± 5.0 nM/min per 103 cells. Although the rabbit anti-mobilferrin previously described inhibited the ferroxidase activity (25), the chicken antibody used here did not, and whereas the DMT-1 antibody inhibited iron transport (4), it did not inhibit the ferrireductase activity.

Although the association of DMT-1 with paraferritin seemed certain based on these findings, it was conceivable that some weak and physiologically less relevant binding might have occurred or that somehow the precipitates had “trapped” the DMT-1 nonspecifically. However, highly purified paraferritin (25) also contained DMT-1. The protein complex was purified from human (Fig.5, left) and from rat duodenum (Fig. 5, right). The purification scheme previously described involves many chromatography steps, including a gel exclusion column chromatography in high salt. The paraferritin eluted with an apparent molecular mass of 520 kDa and should not be incidentally contaminated with a protein the size of DMT-1 (60 kDa). Furthermore, this protein was isolated from the soluble fraction of the tissue homogenate, whereas the noncomplexed DMT-1 protein is believed to be membrane associated and not soluble in nondissociating buffers. SDS gels of purified paraferritin were transblotted to Immobilon-P and developed with DMT-1 as primary antibody. A band was detected at the expected molecular weight, consistent with the findings from the whole cell homogenate immunoprecipitates (Fig. 2). Again, the DMT-1 antibody, confirming the known specificity of the reaction, bound none of the other peptide bands.

Fig. 5.

Western blots of purified paraferritin from human and rat intestine demonstrate DMT-1. Paraferritin purified to homogeneity from human (left) and rat (right) duodenum were subjected to SDS gel electrophoresis, transblotted, and developed with antibody to DMT-1. A single band at the expected molecular mass for DMT-1 was observed, indicating the presence of this protein in the purified paraferritin isolates. The small-molecular-weight fragment seen in the total cell immunoprecipitates was not seen in the purified duodenal paraferritin, which may indicate a tissue difference.


Inorganic iron crosses the cell membrane of tissue culture cells by two distinct pathways (4). Ferric iron is transported by the IMP pathway, which includes the ferrireductase paraferritin. Ferrous iron crosses the membrane facilitated by the DMT-1 protein, but the intracellular pathway is not known. The entrances to the cell by these two pathways are distinct and independent from each other. Both pathways eventually deliver ferrous iron to ferrochelatase for the production of hemoglobin. Whereas the DMT-1 pathway does not require an additional reduction step, the IMP pathway involves the reduction of the ferric cation after transmembrane transport by an NADPH-dependent ferrireductase called paraferritin.

Paraferritin has been purified from duodenal homogenates as a 520-kDa protein complex that had several unexpected features (25) It was shown to be an intermediary in the uptake of ferric iron by the duodenum as well as in K562 and other tissue culture cells. It contained both mobilferrin (which is known to bind ferric iron) and a flavin monooxygenase. It reduced ferric iron to ferrous iron by donating one electron from NADPH to the iron and one to oxygen. Other peptides were found in the protein complex, and these multiple peptides were shown by cross-linkage experiments to belong to one entity. Included were the two peptides of integrin and FMO, both of which were previously thought to be exclusively membrane proteins. Mobilferrin is identical to calreticulin for the first 20 NH2-terminal amino acids, and antibodies to both the NH2- and COOH-terminal sequences of calreticulin react with mobilferrin, so that there appears to be a high degree of homology, and it is possible that the two proteins are identical. Calreticulin has been frequently demonstrated to have a chaperone function and could solubilize FMO, and presumably the histocompatibility locus A complex, as part of its chaperone functions (22). Although it may be surprising to find so many “insoluble” proteins in a cytoplasmic complex, it is not, therefore, without precedent (13).

Paraferritin is one of a large group of proteins capable of reducing ferric iron. Among those recently described are the NADH-ferric reductase activity associated with dihydropteridine reductase (16), plasma membrane-associated reductase (27), and duodenal cytochrome b(Dcytb) (18). Strong arguments have been presented that the ascorbate-dependent trans-plasma membrane oxidoreductase does not function in iron transport (17). It was postulated that Dcytb reduces dietary ferric iron to ferrous iron at the exterior luminal surface of the duodenal mucosa to allow iron to be transported by DMT-1. In contrast, iron uptake studies (4) showed that ferrous iron was not an essential intermediate in ferric iron uptake. No biochemical association between DMT-1 and Dcytb has been demonstrated yet. Whereas Dcytb may or may not be complexed to paraferritin in some fashion, the purified paraferritin did not have a spectra consistent with the presence of cytochromes (25).

This study shows that DMT-1 is part of the paraferritin complex. Both immunoprecipitates of paraferritin obtained by using antibody to mobilferrin/calreticulin of tissue culture cells and immunoblots of purified protein demonstrated the presence of DMT-1. Although the immunoprecipitates might possibly contain “trapped” DMT-1, or since mobilferrin/calreticulin has chaperone activity that the DMT-1 may have complexed with these proteins specifically or nonspecifically, these should not have contaminated the purified paraferritin preparations because of the unusual and large molecular weight of paraferritin.

The observations in this study demonstrated an association between DMT-1 and paraferritin, but they did not provide an understanding of the role this would play in the relationship between the two pathways. The entry of ferric and ferrous iron into cells involves clearly distinct mechanisms, which now seemingly converge on paraferritin. The product of the ferrireductase activity of paraferritin, ferrous iron, is the substrate for DMT-1, which might lead to the speculation that DMT-1 is somehow involved in the next step: transport to or into the mitochondrion. The means by which iron crosses into the mitochondrion is not known, but it does appear to require cotransport with a proton, analogous to that used by DMT-1. However, there is equally clearly a direct ferrous pathway from the plasma membrane of the cell as well. Whether the phenotype of the Belgrade rat and mk mouse can best be explained by a defect in the plasma membrane pathway (and, therefore, involving dietary derived ferrous iron), perhaps by an inability to relocate DMT-1 to the membrane, or by a defect in the functional parameters of paraferritin (involving dietary-derived ferric iron) has not been clarified.

Both mobilferrin and DMT-1 have been found in the membranes and in paraferritin in the cytoplasm. It is not known if the proteins with their bound iron are carried between the different locations or if only the iron is shuffled between the various sites. Since ferric iron is insoluble and ferrous iron is extremely toxic, it is tempting to speculate that the proteins themselves are moved to prevent the release of free iron.


This study was supported by the National Institute of Diabetes and Digestive and Kidney Diseases Merit Award 2-R37-DK-36112.


  • Address for reprint requests and other correspondence: J. N. Umbreit, USA Cancer Center, 307 Univ. Blvd., Rm. 414, Mobile, AL 36688 (E-mail: jumbreit{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • 10.1152/ajpgi.00199.2001


View Abstract