Ceruloplasmin Ferroxidase Activity Stimulates Cellular Iron Uptake by a Trivalent Cation-specific Transport Mechanism*

The balance required to maintain appropriate cellular and tissue iron levels has led to the evolution of multiple mechanisms to precisely regulate iron uptake from transferrin and low molecular weight iron chelates. A role for ceruloplasmin (Cp) in vertebrate iron metabolism is suggested by its potent ferroxidase activity catalyzing conversion of Fe2+ to Fe3+, by identification of yeast copper oxidases homologous to Cp that facilitate high affinity iron uptake, and by studies of “aceruloplasminemic” patients who have extensive iron deposits in multiple tissues. We have recently shown that Cp increases iron uptake by cultured HepG2 cells. In this report, we investigated the mechanism by which Cp stimulates cellular iron uptake. Cp stimulated the rate of non-transferrin 55Fe uptake by iron-deficient K562 cells by 2–3-fold, using a transferrin receptor-independent pathway. Induction of Cp-stimulated iron uptake by iron deficiency was blocked by actinomycin D and cycloheximide, consistent with a transcriptionally induced or regulated transporter. Cp-stimulated iron uptake was completely blocked by unlabeled Fe3+ and by other trivalent cations including Al3+, Ga3+, and Cr3+, but not by divalent cations. These results indicate that Cp utilizes a trivalent cation-specific transporter. Cp ferroxidase activity was required for iron uptake as shown by the ineffectiveness of two ferroxidase-deficient Cp preparations, copper-deficient Cp and thiomolybdate-treated Cp. We propose a model in which iron reduction and subsequent re-oxidation by Cp are essential for an iron uptake pathway with high ion specificity.

The balance required to maintain appropriate cellular and tissue iron levels has led to the evolution of multiple mechanisms to precisely regulate iron uptake from transferrin and low molecular weight iron chelates. A role for ceruloplasmin (Cp) in vertebrate iron metabolism is suggested by its potent ferroxidase activity catalyzing conversion of Fe 2؉ to Fe 3؉ , by identification of yeast copper oxidases homologous to Cp that facilitate high affinity iron uptake, and by studies of "aceruloplasminemic" patients who have extensive iron deposits in multiple tissues. We have recently shown that Cp increases iron uptake by cultured HepG2 cells. In this report, we investigated the mechanism by which Cp stimulates cellular iron uptake. Cp stimulated the rate of non-transferrin 55 Fe uptake by iron-deficient K562 cells by 2-3-fold, using a transferrin receptor-independent pathway. Induction of Cp-stimulated iron uptake by iron deficiency was blocked by actinomycin D and cycloheximide, consistent with a transcriptionally induced or regulated transporter. Cp-stimulated iron uptake was completely blocked by unlabeled Fe 3؉ and by other trivalent cations including Al 3؉ , Ga 3؉ , and Cr 3؉ , but not by divalent cations. These results indicate that Cp utilizes a trivalent cation-specific transporter. Cp ferroxidase activity was required for iron uptake as shown by the ineffectiveness of two ferroxidase-deficient Cp preparations, copper-deficient Cp and thiomolybdate-treated Cp. We propose a model in which iron reduction and subsequent re-oxidation by Cp are essential for an iron uptake pathway with high ion specificity.
Iron is important in many biological processes because it is an ideal O 2 carrier and because it can function as a proteinbound redox element. Iron deficiency is common worldwide and in infants can cause severe neurological deficit (1). In contrast, iron in excess of cellular needs is toxic and dietary overload or hereditary hemochromatosis (HH) 1 leads to tissue iron deposi-tion and injury (2,3). Alterations of iron pools have been implicated in diverse human diseases, including neurodegenerative disease, aging, atherosclerosis, cancer, and microbial infection (4). The precise balance required to maintain appropriate intracellular and tissue iron levels has led to the evolution of several independent cellular iron transport pathways.
The transferrin-mediated iron uptake pathway has been extensively studied and is an important model for understanding the recycling, receptor-mediated endocytosis pathway. In this pathway, diferric transferrin is endocytosed by a cell surface transferrin receptor and iron is delivered to the cell by an endosomal pathway accompanied by recycling of transferrin and its receptor (5). The physiological significance of the transferrin receptor is suggested by its presence in virtually all cultured cells and by the fact that hypotransferrinemic animals develop severe hemochromatosis (6). The central importance of the transferrin pathway is shown by recent studies of HH, the most common inherited disease in people of northern European descent. Positional cloning studies of Feder et al. (7) identified HFE as a candidate gene for HH, and two HFE mutations account for about 80 -85% of all HH. The normal HFE gene product, by binding to the transferrin receptor and lowering its affinity for transferrin, down-regulates intestinal iron absorption (8). Mutant HFE does not bind the transferrin receptor and lacks this suppressor activity.
Transferrin-independent cellular iron uptake has been described in many cultured cells including HepG2 (9) and K562 (10) cells. The function of the transferrin-independent system is unknown, but Kaplan et al. (11) have proposed that it clears potentially toxic, low molecular weight iron chelates (e.g. nontransferrin bound iron). Such chelates accumulate to a very large extent in HH patient plasma (12). Transferrin-independent uptake mechanisms are not well understood; however, two putative transporters, DCT1 and SFT, have been recently identified in mammalian cells (13,14). A cell surface ferrireductase is also active and may be required for transferrinindependent iron uptake (15,16).
Important insights into eukaryotic non-transferrin iron metabolism have come from recent yeast genetic studies. Two iron transporters (or "permeases"), Ftr1p from Saccharomyces cerevisiae (17) and the homologous Fip1p from Schizosaccharomyces pombe (18), facilitate high affinity iron uptake by irondeficient yeast. A surface ferrireductase, Fre1p (or its homologue Fre2p), is also necessary for yeast iron uptake (19). A requirement for copper in iron uptake has led to the cloning of genes encoding copper-containing proteins, Fet3p in S. cerevisiae (20 -22) and its homologue Fio1p in S. pombe (18), that are required for iron uptake. These proteins are membranespanning, multicopper oxidases which interact at the cell membrane with the transporters (17). Fet3p (and Fio1p) has limited sequence homology to human ceruloplasmin (Cp), and like Cp is a copper protein with a ferroxidase activity that catalyzes conversion of ferrous to ferric ion.
The similarities between Fet3p and Cp suggest a possible role for the latter in vertebrate cell iron uptake. We have shown recently that Cp increases 55 Fe uptake by iron-deficient HepG2 cells by a transferrin receptor-independent pathway (23). Furthermore, iron deprivation transcriptionally induces Cp synthesis by HepG2 cells (23), and Fet3p synthesis in yeast (20,21). Thus maintenance of iron homeostasis in liver (and possibly other tissues) during periods of iron deficiency may be an important physiological function of Cp. The recent observation that "aceruloplasminemia" patients with Cp gene defects have massive iron deposits in liver, pancreas, brain, retina, and other tissues indicates an important role for Cp in iron homeostasis in vivo (22,24,25).
In this report we investigate the molecular mechanism by which Cp increases iron uptake by erythroleukemic K562 cells. Our results show that Cp-stimulated iron uptake requires Cp ferroxidase activity and utilizes a novel, trivalent cation-specific transport pathway.
Measurement of 55 Fe Uptake by K562 Cells-Human erythroleukemic K562 cells (ATCC, Rockville, MA) were grown in iron-free RPMI 1640 medium supplemented with 10% fetal bovine serum. The cells were made iron-deficient by resuspension in serum-free RPMI 1640 medium containing bathophenanthroline disulfonate (100 M) for 4 -8 h. The cells were washed twice with phosphate-buffered saline and once with RPMI 1640 medium, and then resuspended in the same medium. A solution of 55 Fe-NTA was prepared by mixing 55 FeCl 3 (15-20 M) with a 4-fold molar excess of NTA in colorless RPMI 1640 medium containing 1 mM ascorbate (26). To measure iron uptake, 5 ϫ 10 6 K562 cells were incubated with 55 Fe-NTA (1.4 M) in 0.5 ml of RPMI 1640 medium for 20 -30 min at 25°C. An aliquot (0.1 ml) was removed and applied to a nitrocellulose filter subjected to vacuum. The filter was washed twice with 3 ml of 150 mM NaCl containing 10 M EDTA (to remove surfacebound iron) and cell-associated radioactivity was measured by liquid scintillation counting. The amount of 55 Fe bound after a "zero time" incubation (less than 15 s) was subtracted from experimental measurements to correct for nonspecific binding.
Measurement of 59 Fe Uptake and Incorporation into K562 Cell Protein-K562 cells, made iron-sufficient and iron-deficient as above, were incubated with 59 Fe-NTA (1.4 M) in the presence or absence of Cp (30 g/ml) for 20 min at 25°C. The cells were washed two times by centrifugation at 16,000 ϫ g, and then lysed in 20 mM HEPES, pH 8.0, 2 mM MgCl 2 , 1% aprotinin, and 200 g/ml leupeptin by repeated passage through a 27-gauge needle at 0°C. Unbroken cells, nuclei, and debris were removed by centrifugation at 2000 ϫ g for 5 min. Membrane fragments were separated from the cytosol-containing fraction by centrifugation at 16,000 ϫ g for 10 min at 4°C. Radioactivity in the intact cells, and membrane and cytosol fractions was measured by ␥-counting.
To determine incorporation of 59 Fe into individual proteins, aliquots (20 g of protein) of membrane and cytosolic fractions were subjected to SDS-PAGE (6.5%) under nonreducing conditions and autoradiography (Hyperfilm, Amersham, Buckinghamshire, United Kingdom). To evaluate total protein in these fractions, the gel was stained with Coomassie Blue. In a control experiment, ferritin in the cytosolic fraction was immunodepleted by incubation for 18 h with rabbit anti-human ferritin IgG (1:1000) or with a nonspecific antibody (rabbit anti-human G q␣ IgG; Santa Cruz Biotechnology, Santa Cruz, CA), then with protein A-Sepharose, and centrifuged at 16,000 ϫ g.
Preparation of Apoceruloplasmin and Reconstitution of Holoceruloplasmin-Apoceruloplasmin (apoCp) was prepared by removal of Cp copper by complexing with cyanide under reducing conditions as described by Musci et al. (27). Cp was dialyzed at 4°C against 100 mM sodium acetate buffer (pH 5.9) containing 10 mM ascorbate under anaerobic conditions achieved by continuous bubbling with N 2 gas. Dialysis was continued until the "blue" coppers were reduced as shown by complete decolorization of the sample. Potassium cyanide (50 mM) was added and the dialysis continued for 5 h. The copper-cyanide complex and excess cyanide were removed by dialysis against 100 mM sodium acetate containing 1 mM cysteine, and then by an overnight dialysis against buffer containing 50 mM MOPS, 150 mM potassium chloride (pH 7.0). Inactivation was verified by an "in-gel" ferroxidase assay (see below).
Holoceruloplasmin was reconstituted from apoCp as described by Musci et al. (27). A Cu 1ϩ -glutathione complex was prepared immediately before use (28) by addition of copper sulfate (0.1 mM) to GSH (0.3 mM) in 0.1 M phosphate buffer (pH 7.0). A 7-fold molar excess of the Cu 1ϩ -GSH complex was added to apoCp in 50 mM MOPS, 150 mM potassium chloride, 5 mM magnesium chloride (pH 7.0) for 4 h. Unbound copper was removed by overnight dialysis against phosphatebuffered saline.
Inhibition of Cp Ferroxidase Activity by Thiomolybdate-Cp (7.61 M) was incubated for 20 min with ammonium tetrathiomolybdate (1.52 mM) in acetate buffer (320 mM, pH 6.0) as described by Chidambaram et al. (29). Excess thiomolybdate was removed by dialysis overnight against the same buffer, then against phosphate-buffered saline. Removal of unreacted thiomolybdate was monitored by monitoring optical density at 470 nm.
Measurement of Cp Ferroxidase Activity-Cp ferroxidase activity was measured by a modification of the method of Schen and Rabinovitz (30). In brief, Cp or modified Cp was subjected to electrophoresis on a 1% agarose gel (Seakem Gold, FMC, Rockland, ME) under nondenaturing conditions. After transfer to an Immobilon-P membrane, the blot was incubated with 0.08% ferrous sulfate in acetate buffer (pH 5.7), for 30 min at 37°C, followed by incubation with 1% potassium ferrocyanide in 0.1 N HCl. Ferric ion (formed by Cp-catalyzed oxidation of ferrous ion) was detected by the rapid formation of a "Prussian blue" band which was stable for at least 24 h.

Characterization of Cp-stimulated Iron Uptake by K562
Cells-We have previously shown that Cp stimulates high affinity iron uptake by iron-deficient HepG2 cells (23). For a more detailed analysis of the mechanism of Cp-stimulated iron uptake, we used human erythroleukemic K562 cells since the transferrin-dependent and -independent uptake pathways are well characterized (10). Unlike HepG2 cells which secrete Cp in an amount sufficient to promote iron uptake (23), K562 cells do not secrete Cp (or transferrin, Ref. 31) thereby permitting more precise control of conditions. Cp stimulated 55 Fe-NTA uptake by iron-deficient K562 cells by about 2-3-fold (Fig. 1A). Halfmaximal stimulation was at about 10 g/ml Cp and maximal uptake was at about 30 g/ml, a level much lower than the unevoked human plasma concentration of 300 g/ml. The stimulation of iron uptake by Cp was observed in iron-deficient cells but not in iron-sufficient cells (Fig. 1B), a finding consistent with studies of HepG2 cells (23). Cp stimulated iron uptake by decreasing the apparent K m of 55 Fe-NTA uptake from about 2 M to about 0.5 M iron, without changing the maximum uptake rate (Fig. 1B). This result indicated that the number of active transporters was not altered by Cp, but rather that the affinity of the transporter for iron was increased or that the amount of substrate was increased by Cp.
Several control experiments were done to show that Cp stimulated iron entry into the cell, and thus exclude the possibility that Cp increased extracellular binding of iron to the cell surface (e.g. by ferroxidase-dependent conversion of soluble Fe 2ϩ to insoluble Fe 3ϩ ). First, we measured iron uptake using a procedure specifically designed to remove complexes of surfacebound iron (32). According to this method, cells are rinsed with a solution containing 5 M sodium dithionite to reduce Fe 3ϩ to Fe 2ϩ and with 5 M bathophenanthroline disulfonate to specifically chelate Fe 2ϩ . This procedure gave results essentially identical to those obtained with the normal rinse solution containing 10 M EDTA; Cp increased iron uptake by 2-3-fold in iron-deficient K562 cells when measured using either the "reducing" wash or the normal wash (data not shown). To verify iron internalization, intracellular and membrane-bound iron was measured. K562 cells were incubated with 59 Fe-NTA, the cells were lysed, and membrane and cytosol fractions isolated by centrifugation. In this experiment Cp stimulated about a 2-fold increase in radioactivity measured in intact (and iron-deficient) cells, and nearly all of this increase was accounted for by 59 Fe in the cytosolic fraction (Fig. 1C); 10% or less of the total cell radioactivity was bound to membranes under any condition. To verify the intracellular localization of 59 Fe, incorporation into protein was determined by SDS-PAGE and autoradiography. Radiolabeled protein was not detected in the membrane fraction of any samples (Fig. 1, D and E). However, the cytosolic fraction exhibited a single major 59 Fe-labeled band that co-migrated with ferritin (Fig. 1F). Identification of the 59 Fe-labeled band as ferritin was confirmed by immunodepletion with anti-ferritin IgG, but not by a nonspecific antibody (Fig. 1F). Cp increased incorporation of 59 Fe into ferritin in lysates made from iron-deficient cells by about 90%; equal protein loading of the gel lanes was shown by Coomassie Blue stain (Fig. 1G). These results demonstrate that the increase in iron associated with cells in the presence of Cp is due to cellular uptake and entry into a metabolically active intracellular iron pool.
The finding that Cp did not stimulate iron uptake by ironsufficient cells (Fig. 1B) suggested that a factor required for uptake, e.g. the transporter itself, was induced or activated by iron deficiency. To test whether this factor must be synthesized de novo, K562 cells were treated with actinomycin D or cycloheximide prior to iron depletion. Cp-stimulated iron uptake was completely blocked by both inhibitors suggesting transcriptional up-regulation of an iron transporter (or a regulatory protein) by iron deficiency (Fig. 1H). To determine whether Cp stimulates iron efflux as well as influx, K562 cells were preloaded with 55 Fe-NTA, and Cp-stimulated iron release was measured. Under these conditions, Cp did not enhance iron release (Fig. 1I). The addition of apo-transferrin or NTA to the medium as Fe 3ϩ acceptors did not alter this result. This result was consistent with our previous findings in HepG2 cells (23), but not with results of Young et al. (33) who reported a small Cp-stimulated increase in iron efflux from HepG2 cells. The difference may be due to experimental conditions, e.g. our methods measure the effects of Cp on initial rates whereas the 18-h efflux intervals of Young et al. (33) may reflect equilibrium processes shifted by Cp.
Transport Mechanism of Cp-stimulated Iron Uptake-The conversion of Fe 2ϩ to Fe 3ϩ by Cp ferroxidase activity is thought to facilitate the binding of iron to apotransferrin (34). It is thus possible that Cp stimulates cellular iron uptake by stimulating the formation of Fe 2 -transferrin which delivers Fe 3ϩ by transferrin receptor-mediated endocytosis. Although K562 cells are not known to secrete transferrin (31), this pathway was tested by addition of a rabbit polyclonal anti-human transferrin antibody to the incubation medium. The antibody only slightly inhibited Cp-stimulated iron uptake but completely blocked 55 Fe uptake mediated by transferrin ( Fig. 2A). Essentially identical results were obtained using a monoclonal anti-human transferrin receptor antibody (not shown). These results show that Cp-stimulated iron uptake was transferrin receptor-independent, but did not exclude the possibility that other receptormediated endocytic processes were involved, e.g. endocytosis of the putative Cp receptors (35,36) or other iron-protein receptors. The lysosomatropic inhibitor, amantadine, was used to inhibit receptor-mediated endocytosis (37). Amantadine had essentially no effect on Cp-stimulated 55 Fe-NTA uptake, but was a very effective inhibitor of transferrin-mediated iron uptake (Fig. 2B). Comparable results were obtained using ammonium chloride (38) to inhibit receptor-mediated endocytosis (data not shown).
Cation Specificity and Role of Ferroxidase Activity in Cpstimulated Iron Uptake-The cation specificity of Cp-stimulated cell transport was investigated. Multiple divalent cations, including Ca 2ϩ , Mg 2ϩ , Cu 2ϩ , or Zn 2ϩ (and Mn 2ϩ and Co 2ϩ , not shown), added at a 70-fold molar excess did not substantially compete for Cp-stimulated 55 Fe uptake (Fig. 3). In the same experiment Fe 2ϩ (FeCl 3 plus ascorbic acid) in the presence of Cp completely suppressed uptake. This result suggests that Cp stimulates a transporter with absolute specificity for Fe 2ϩ or a trivalent cation transporter; the latter is consistent with a requirement for ferroxidase activity of Cp.
The role of Cp ferroxidase activity was examined using ferroxidase-defective Cp. We took advantage of the ability of thiomolybdate to bind copper and irreversibly inhibit Cp ferroxidase activity (29). Treatment of Cp with ammonium tetrathiomolybdate completely blocked its iron uptake activity (Fig. 4A). Thiomolybdate did not inhibit iron uptake by inacti-vation of the cellular transport pathway since the addition of untreated Cp to cells incubated with thiomolybdate-treated Cp fully restored iron uptake. The inactivation by thiomolybdate treatment was confirmed by an in-gel assay for ferroxidase activity (Fig. 4B). The integrity of apoCp was tested by SDS-PAGE analysis and Coomassie Blue staining which revealed a 132-kDa undegraded protein (Fig. 4C). To verify the role of Cp ferroxidase activity in iron transport, ferroxidase-defective apoCp was prepared by removal of copper by complexing with cyanide under reducing conditions (27). ApoCp did not stimulate 55 Fe influx into iron-deficient K562 cells (Fig. 4D). The absence of apoCp ferroxidase activity was confirmed by an in-gel assay (Fig. 4E), and the structural integrity of the protein was shown by SDS-PAGE and Coomassie Blue staining (Fig. 4F). To further demonstrate the integrity of apoCp, it was reconstituted to holoCp by incubation with copper in the presence of GSH as a reducing agent. The reconstituted Cp re-gained about 80% of its ferroxidase activity and about 70% of its iron uptake enhancing activity (Fig. 4, D and E). The complete inactivity of two ferroxidase-defective forms of Cp is persuasive evidence that ferroxidase activity is necessary for Cpstimulated iron uptake, and suggests that the Fe 3ϩ product is transported into the cell by a trivalent cation transporter. The results also suggest that Cp increases iron uptake by increasing available substrate (by formation of Fe 3ϩ ), not by activation of the transporter itself.
To test whether the transporter is Fe 3ϩ -specific or if it transports multiple trivalent cations, we examined the ability of Al 3ϩ , Cr 3ϩ , and Ga 3ϩ to compete for iron uptake. All three trivalent cations inhibited Cp-stimulated iron cellular uptake in a concentration-dependent manner, reaching greater than 90% inhibition at about 10 M (Fig. 5A). Since trivalent cations may inhibit Cp ferroxidase activity (39), we investigated whether the effects of Al 3ϩ , Cr 3ϩ , and Ga 3ϩ on iron uptake were due to inhibition of ferroxidase activity rather than by competition for a transporter. Cr 3ϩ and Al 3ϩ , at a concentration that blocks iron uptake (10 M), did not inhibit Cp ferroxidase activity, although some inhibition was found at higher ion concentrations (Fig. 5, B and C). Ga 3ϩ , even at 100 M, did not inhibit ferroxidase activity (Fig. 5D). Together these data show that trivalent cations do not inhibit Cp-stimulated iron uptake by inhibition of Cp ferroxidase activity. More likely, these ions may block iron uptake by competing with iron for a binding site on a trivalent cation-specific transporter.
The inhibition of iron transport by multiple trivalent cations indicates a nonspecific trivalent cation transporter. However, the competition results do not prove that these competing cations are actually transported; they may compete with Fe 3ϩ for the binding site on the transporter without subsequent entry into the cell. To test this possibility, 51 CrCl 3 uptake by K562 cells was measured. 51 Cr 3ϩ uptake was high and exhibited two properties consistent with uptake by the transporter used for Cp-stimulated iron uptake (Fig. 6). First, 51 Cr uptake was much higher in iron-deficient K562 cells than in iron-sufficient FIG. 4. Requirement for ferroxidase activity in Cp-induced iron uptake. A, Cp was reacted with ammonium tetrathiomolybdate (Cp TM ) as described under "Experimental Procedures." Iron-deficient K562 cells were incubated with 55 Fe-NTA (1.4 M) in the presence of Cp (30 g/ml), Cp TM (30 g/ml), or Cp TM (30 g/ml) plus Cp (30 g/ml; 1 ϫ Cp) or plus Cp (150 g/ml; 5 ϫ Cp), or with thiomolybdate alone (2.35 M). Iron uptake was measured after 20 min. B, ferroxidase activity of 5 g of Cp and Cp TM were measured by an in-gel assay. C, Cp and Cp TM (5 g) were subjected to SDS-PAGE and Coomassie Blue staining. D, apoCp was prepared by complexing copper with cyanide as described under "Experimental Procedures." Cp was reconstituted with copper (Cp reconst. ) by incubation with Cu 1ϩ -GSH. Iron-deficient K562 were incubated with 55 Fe-NTA (1.4 M) in the presence of medium only, Cp, apoCp, or Cp reconst. for 20 min, and then 55 Fe uptake was measured. E, ferroxidase activity of 5 g of Cp, apoCp, and Cp reconst. was measured by an in-gel assay. F, integrity of Cp, apoCp, and Cp reconst. was examined by SDS-PAGE (4 -12% gel) and Coomassie Blue staining. cells. Second, 51 Cr uptake was inhibited by Fe 3ϩ (and by Cr 3ϩ and Al 3ϩ ), and by Fe 2ϩ plus Cp, but not by Fe 2ϩ alone (Fig. 6). Cp did not increase 51 Cr uptake, thus providing additional evidence that Cp does not directly activate the transporter. Together these results show that the transporter mediating Cp-stimulated Fe 3ϩ uptake is a high affinity trivalent cation transporter. DISCUSSION Our results show that Cp stimulates cellular iron uptake by ferroxidase-dependent conversion of Fe 2ϩ to Fe 3ϩ and transport by a trivalent cation-specific transporter. This uptake mechanism recalls the newly elucidated iron uptake pathway in S. cerevisiae and S. pombe shown in Fig. 7A (24). According to this model the yeast ferrireductase Fre1p/Fre2p initiates the uptake pathway by one-electron reduction of Fe 3ϩ to Fe 2ϩ . The reduced metal is reoxidized to Fe 3ϩ by Fet3p (or Fio1p), a Cp homologue with ferroxidase activity. Finally, Fe 3ϩ is transported into the cell by Ftr1p (or Fip1p), a six-spanning membrane transporter that physically interacts with Fet3p (Fio1p). These yeast proteins are all transcriptionally up-regulated during iron deficiency by activation of the Aft1p transcription factor (40). By analogy we propose a model in which Cp is positioned in the center of an analogous iron transport pathway for mammalian cells (Fig. 7B). In this model extracellular iron is first reduced by a surface ferrireductase, re-oxidized by Cp, and finally transported into the cell by a trivalent cationspecific transporter. The evidence for the mammalian pathway elements, and the similarities and differences with their yeast counterparts, are addressed below.
The requirement for a surface ferrireductase activity for transferrin-independent iron uptake has been shown for human HeLa (16), HepG2 (9), and K562 cells (15). Our studies have not addressed the role of ferrireductase activity in Cpstimulated iron uptake; however, the requirement for the reductant ascorbate (not shown) is consistent with an important role of a ferrireductase under physiological conditions (i.e. in the absence of exogenous reductant). The identity of the mammalian ferrireductase is not known, but the phagocyte NADPH oxidase complex may be involved since Fre1p has partial sequence homology to its gp91-phox subunit (19).
Our results show that Cp stimulates 55 Fe uptake by irondeficient K562 cells by up to 2-3-fold. An absolute requirement for ferroxidase activity for Cp-stimulated iron uptake was shown using two ferroxidase-deficient preparations: thiomolybdate-treated Cp and copper-depleted apoCp. This part of the pathway is analogous to the yeast pathway since Fet3p (and Fio1p), a protein with some sequence homology to Cp, has been implicated in high affinity iron uptake (20). Fet3p, like Cp, is a multicopper oxidase with O 2 -dependent ferroxidase activity (21). A requirement for Fet3p ferroxidase activity in iron uptake has been suggested by the correlation between ferroxidase activity (and Fet3p expression) and iron uptake (20,22). A recent study has shown directly, by site-directed mutagenesis, that the ferroxidase activity of Fet3p is required for high affinity iron uptake by yeast (41). A similarity between the yeast and mammalian systems is found in the transcriptional upregulation by iron deficiency of Fet3p in S. cerevisiae (20) and Cp in HepG2 cells (23). K562 cells do not produce Cp, indicating tissue-specific expression (not shown). A species-specific difference is seen in the cellular localization of the ferroxidase. In yeast, Fet3p is membrane-bound and physically interacting with a membrane transporter (21). In contrast, mammalian Cp has a signal sequence and is efficiently secreted. This difference may reflect the paracrine nature of iron homeostasis in mammalian systems that is absent in yeast. The spatial relationship between Cp and the trivalent cation transporter is not known. One possibility is that Cp transiently associates with the transporter to increase the efficiency of Fe 3ϩ delivery. The identification of putative Cp-binding proteins on the surface of multiple cell types (36), which are up-regulated by iron deficiency (42), is consistent with this mechanism. Alternatively, Fe 3ϩ provided by Cp may be delivered to the transporter by diffusion.
The ion specificity of the mammalian non-transferrin iron uptake pathway(s) is uncertain, due in part to differences between the cells and uptake assays used in various laboratories. Our experiments show that the Cp-stimulated iron transporter in K562 cells is trivalent cation-specific based on the lack of inhibition by divalent cations, the inhibition by trivalent cations, the transport of 51 Cr, and the requirement for ferroxidase activity in iron transport. Others have suggested that Fe 2ϩ is FIG. 6. Chromium uptake by K562 cells. 51 Cr-NTA (0.14 M) was incubated with iron-sufficient (open hatched bars) or iron-deficient (closely hatched bars) K562 cells in the presence or absence of Cp (30 g/ml). In other wells, Fe-deficient cells were incubated with 100 M chloride salts of Cr 3ϩ , Al 3ϩ , Fe 3ϩ (added in the absence of 1 mM ascorbate), or Fe 2ϩ (FeCl 3 in the presence of 1 mM ascorbate) in the presence or absence of Cp (30 g/ml). 51 Cr uptake was measured after 20 min by the same method used for 55 Fe-NTA uptake except that 51 Cr was determined by ␥-ray scintillation counting. the principle transport form based on the requirement for reducing agents and inhibition by Fe 2ϩ -specific chelators (9). However, in light of recent studies showing the participation of both reduced and oxidized iron in the transport pathway, these approaches do not give specific information on the actual ion transported. Further evidence for Fe 2ϩ transport has been provided by studies showing that non-transferrin iron uptake is blocked by divalent cations (43). However, studies in K562 cells show an inability of divalent cations to compete, a finding in agreement with our own (10). Two previous reports suggest that Ga 3ϩ and other trivalent cations regulate iron uptake (44,45), but there is no direct evidence for iron uptake via a trivalent cation-specific pathway. To our knowledge the ion specificity of the yeast iron transporters Ftr1p and Fio1p have not been reported.
The identity of the transporter facilitating Cp-stimulated iron uptake is not known. Identification of the specific genes encoding the two yeast transporters has not been helpful since homologous mammalian cDNA's have not been identified. Several mammalian iron transporters have recently been discovered and are pathway candidates. One such iron transporter is DCT1, a member of the Nramp gene family (14). However, divalent cations are efficiently transported by DCT1 suggesting it is not linked to Cp-stimulated transport. A possible role for SFT, an iron transporter cloned from phorbol ester-treated K562 cells, is supported by the fact that the SFT-mediated iron transport is not inhibited by most divalent cations. However, its localization in recycling endosomes, and the stimulated uptake of both transferrin-bound iron and non-transferrin iron uptake, indicate a probable role in intracellular iron transport (13).
The presence of iron reduction and oxidation steps in a single pathway is surprising. In fact, early models of the yeast uptake pathway described an extracellular ferrireductase and an intracellular ferroxidase, thus avoiding a "futile" cycle (20,46). This model has since been modified by biochemical studies, and the ferroxidase domain of Fet3p, like Cp, is extracellular (21). The function of the ferrireductase is most likely to release soluble Fe 2ϩ from Fe 3ϩ complexes (Fe 3ϩ -NTA in our experiments, possibly non-transferrin bound iron in vivo). The function of the ferroxidase is clearly to convert the solubilized Fe 2ϩ to the Fe 3ϩ form that is recognized by the downstream trivalent cation transporter. The advantage of a two-step pathway of iron transport rather than a simpler one-step pathway of direct Fe 2ϩ transport via a divalent cation-specific transporter is uncertain. One possibility is that the two-step pathway gives higher cation specificity. For example, a trivalent cation-specific transporter could greatly improve specificity because Fe 3ϩ will compete successfully with the relatively rare trivalent cations found in tissues. In the case of a divalent cation transporter, Fe 2ϩ would have a much more difficult task competing with abundant ions such as Mg 2ϩ , Zn 2ϩ , etc. According to this proposed mechanism, the ferrireductase activity improves iron solubility and the ferroxidase activity improves transport specificity.
The precise role of Cp in iron homeostasis in vivo is poorly understood. The data presented here and elsewhere suggest that alterations in plasma Cp concentration or activity may alter iron homeostasis, and could contribute to human disease. The finding of hemochromatosis in aceruloplasminemia patients is generally considered as evidence that the normal function of Cp is to mediate iron efflux from cells and tissues (47,48). However, these studies may also be consistent with a role of Cp in cellular iron uptake. For example, iron taken into cells by the Cp-mediated pathway might reduce subsequent iron uptake more efficiently than iron taken in by other path-ways. In this case, Cp may act in effect as a negative regulator of iron uptake, and iron-loading would occur in its absence. Alternatively, the defect in Cp-mediated iron influx may lead to extracellular accumulation of an iron form to a concentration that permits nonspecific, low affinity uptake by cells, thus explaining the observed tissue overload. A more detailed understanding of the function of Cp in cellular iron metabolic pathways will put us in a better position to answer questions on the role of Cp in iron homeostasis in physiological and pathological states.