Role of Ceruloplasmin in Macrophage Iron Efflux during Hypoxia*

The reticuloendothelial system has a central role in erythropoiesis and iron homeostasis. An important function of reticuloendothelial macrophages is phagocytosis of senescent red blood cells. The iron liberated from heme is recycled for delivery to erythrocyte precursors for a new round of hemoglobin synthesis. The molecular mechanism by which recycled iron is released from macrophages remains unresolved. We have investigated the mechanism of macrophage iron efflux, focusing on the role of ceruloplasmin (Cp), a copper protein with a potent ferroxidase activity that converts Fe2+ to Fe3+ in the presence of molecular oxygen. As shown by others, Cp markedly increased iron binding to apotransferrin at acidic pH; however, the physiological significance of this finding is uncertain because little stimulation was observed at neutral pH. Introduction of a hypoxic atmosphere resulted in marked Cp-stimulated binding of iron to apotransferrin at physiological pH. The role of Cp in cellular iron release was examined in U937 monocytic cells induced to differentiate to the macrophage lineage. Cp added at its normal plasma concentration increased the rate of 55Fe release from U937 cells by about 250%. The stimulation was absolutely dependent on the presence of apotransferrin and hypoxia. Cp-stimulated iron release was confirmed in mouse peritoneal macrophages. Stimulation of iron release required an intracellular “labile iron pool” that was rapidly depleted in the presence of Cp and apotransferrin. Ferroxidase-mediated loading of iron into apotransferrin was critical for iron release because ferroxidase-deficient Cp was inactive and because holotransferrin could not substitute for apotransferrin. The extracellular iron concentration was critical as shown by inhibition of iron release by exogenous free iron, and marked enhancement of release by an iron chelator. Together these data show that Cp stimulates iron release from macrophages under hypoxic conditions by a ferroxidase-dependent mechanism, possibly involving generation of a negative iron gradient.

The reticuloendothelial system has a central role in erythropoiesis and iron homeostasis. An important function of reticuloendothelial macrophages is phagocytosis of senescent red blood cells. The iron liberated from heme is recycled for delivery to erythrocyte precursors for a new round of hemoglobin synthesis. The molecular mechanism by which recycled iron is released from macrophages remains unresolved. We have investigated the mechanism of macrophage iron efflux, focusing on the role of ceruloplasmin (Cp), a copper protein with a potent ferroxidase activity that converts Fe 2؉ to Fe 3؉ in the presence of molecular oxygen. As shown by others, Cp markedly increased iron binding to apotransferrin at acidic pH; however, the physiological significance of this finding is uncertain because little stimulation was observed at neutral pH. Introduction of a hypoxic atmosphere resulted in marked Cp-stimulated binding of iron to apotransferrin at physiological pH. The role of Cp in cellular iron release was examined in U937 monocytic cells induced to differentiate to the macrophage lineage. Cp added at its normal plasma concentration increased the rate of 55 Fe release from U937 cells by about 250%. The stimulation was absolutely dependent on the presence of apotransferrin and hypoxia. Cp-stimulated iron release was confirmed in mouse peritoneal macrophages. Stimulation of iron release required an intracellular "labile iron pool" that was rapidly depleted in the presence of Cp and apotransferrin. Ferroxidase-mediated loading of iron into apotransferrin was critical for iron release because ferroxidasedeficient Cp was inactive and because holotransferrin could not substitute for apotransferrin. The extracellular iron concentration was critical as shown by inhibition of iron release by exogenous free iron, and marked enhancement of release by an iron chelator. Together these data show that Cp stimulates iron release from macrophages under hypoxic conditions by a ferroxidasedependent mechanism, possibly involving generation of a negative iron gradient.
Reticuloendothelial macrophages have a central role in regulating normal iron homeostasis (1). Macrophages in bone marrow, liver, and spleen recognize and phagocytose senescent or damaged erythrocytes, and the heme iron is processed and returned to circulation for reutilization by red cell precursors during erythropoiesis (2). Macrophages may also utilize iron to generate toxic reactive oxygen species as a defense mechanism against pathogens (3). Macrophage iron release appears to be tightly regulated. The relative amount of iron released, compared with the amount shunted for storage as ferritin, depends on iron status and is increased during iron deficiency. The rate of macrophage iron release is also regulated by, or at least coupled to, the rate of erythropoiesis. The reticuloendothelial system accumulates iron during several pathological states including chronic inflammation and renal failure, indicating an inability of the macrophage to release iron normally.
Understanding the regulation of macrophage iron release has been an area of intense investigation during the past three decades. Several fundamental issues remain unresolved including the intracellular form of released iron, its mechanism of transport through the plasma membrane, the form of the emerging iron, and the mechanism of iron delivery to apotransferrin (4). A mechanism for the terminal step in the export process was suggested by Osaki et al. (5). They showed that ceruloplasmin (Cp, 1 EC 1.16.3.1) has a potent ferroxidase activity that catalyzed the oxidation of Fe 2ϩ to Fe 3ϩ at the expense of O 2 , and accelerated the binding of iron by apotransferrin. They proposed that this process generated a steep, negative free-iron concentration gradient that increased iron efflux from the cell (5,6). The finding of massive tissue iron deposits in hereditary Cp deficiency patients (7,8), and ferrokinetic studies in mice with a targeted disruption in the Cp gene, support an important role of Cp in tissue iron release (9).
Arguments have been raised against this proposed mechanism of iron release. The rapid autoxidation of iron under physiological conditions, particularly in the presence of apotransferrin (which itself may have ferroxidase activity), may eliminate the need for Cp-dependent catalysis to oxidize and load iron into apotransferrin (10 -12). In fact, the rate of iron binding by apotransferrin is too fast to measure by usual kinetic techniques at pH 7, thus in vitro studies of Cp-mediated iron loading into apotransferrin are done under acidic conditions that slow the reaction, generally at pH 5.5 to 6.0 (13). It is also difficult to reconcile the gradient mechanism with the inverse relationship between plasma Cp concentration and transferrin iron saturation, for example, newborns have very high transferrin iron saturation and low Cp, whereas their mothers have the opposite condition (14). Likewise, the marked elevation of plasma Cp during the anemia of inflammation, when reticuloendothelial cell iron stores are elevated, remains paradoxical (15). The requirement for apotransferrin in macrophage iron release in vitro is also controversial; several investigators have shown a lack of stimulation by apotransferrin (16,17), whereas others have found a stimulatory effect (18). In fact, under some circumstances, Cp may enhance iron uptake by cells, possibly by a trivalent cation transporter on the cell surface (19 -21).
The role of plasma Cp in iron release by macrophages has not been investigated experimentally, and studies from non-macrophage cells are inconsistent. Young et al. (22) reported a 40% stimulation of iron release from human hepatocarcinoma HepG2 cells by Cp in the presence of apotransferrin. The mechanism of stimulation was not specifically investigated in detail, and in view of the comparable stimulation of iron release by either apo-or holotransferrin, an important role of iron loading to transferrin is unlikely. In other studies, Cp increased iron release from iron-loaded HepG2 cells (23) and from central nervous system astrocytes (24), but the mechanism underlying the stimulation was unclear because it occurred even in the absence of apotransferrin. Interestingly, astrocyte Cp is glycosylphosphatidylinositol-linked suggesting that both free and membrane-bound forms of Cp are involved in iron transport processes. Three other laboratories reported that Cp did not increase iron release (even in the presence of apotransferrin) from HepG2 or erythroleukemic K562 cells (19,20), placental BeWo cells (25), or glioblastoma BT325 cells (21).
In this report we rigorously investigate the requirements for Cp stimulation of iron binding to apotransferrin and iron release from myeloid cells. We show that Cp markedly enhances iron release from macrophages only in the presence of apotransferrin and only under hypoxic conditions. Regarding the mechanism of Cp-stimulated macrophage iron release, we show that the iron source is the intracellular labile iron pool (LIP), that Cp ferroxidase activity is required, and that development of a negative iron gradient is an essential feature of iron release from these cells.

EXPERIMENTAL PROCEDURES
Materials-Purified human Cp was obtained from Vital Products (Boynton Beach, FL) and Calbiochem (La Jolla, CA) and homogeneity was verified by an absorbance ratio (610 nm/280 nm) greater than 0.04. Integrity of the protein was established by SDS-PAGE and Coomassie stain; intact, 132-kDa protein was the predominant form, and the primary degradation product was the 115-kDa protein present in human serum (26). Cp mass was determined by nephelometry by the Reference Laboratory at the Cleveland Clinic Foundation, and metal content was determined by inductively coupled plasma mass spectroscopy. Cp contained 6.3 Ϯ 0.3 copper atoms per molecule, and the iron content was below the instrument detection level (i.e. Ͻ0.1 atoms per molecule). Human apotransferrin and human holotransferrin were obtained from Sigma, and the mass determined nephelometrically. The iron content of apo-and holotransferrin was 0.060 Ϯ 0.002 and 2.03 Ϯ 0.02 atoms per molecule, respectively; the copper content of the transferrin preparations was negligible (i.e. Ͻ0.001 atoms per molecule). RPMI 1640 medium (Invitrogen, Carlsbad, CA) was selected for cell experiments because it does not contain iron, copper, or ascorbic acid, a metal ion reducing agent (27). Reduced glutathione, nitrilotriacetic acid (NTA), ferrous ammonium sulfate, ferric chloride, bathophenanthroline disulfonic acid, phorbol myristoyl acetate (PMA), desferrioxamine, and all other reagents were from Sigma. 55 FeCl 3 (9.1 mM, 20 mCi/mg) was from PerkinElmer Life Sciences. A 55 Fe-NTA solution was prepared by incubating 55 FeCl 3 (9.1 mM) and NTA (45.5 mM) for 1 h. The resulting 55 Fe-NTA solution was mixed at a 1:9 molar ratio with unlabeled iron-NTA (prepared similarly) in serum-free medium containing ascorbate (100 M) so that the final concentration of 55 Fe-NTA was 10 M. Iron solutions were freshly prepared before each experiment.
Determination of Iron Binding by Apotransferrin-Iron binding to apotransferrin was measured by a modification of the method of Osaki et al. (5). Ferrous ammonium sulfate was dissolved in glycine (0.1 mM) buffer at pH 3.0. Apotransferrin and Cp were dissolved in RPMI 1640 medium (without phenol red dye) containing 20 mM Hepes at pH 7.4. Apotransferrin was used at 55 M, the upper limit of the normal range in adult serum, as described previously (5,13). For most experiments, the solutions were brought to 1% O 2 , 5% CO 2 , and 94% N 2 by bubbling with a mixture of N 2 , CO 2 , and 21% O 2 ϩ 79% N 2 at the appropriate ratio using a 3-channel gas controller and mixer. In some experiments the O 2 or CO 2 concentration was varied independently. The solutions were added to a plastic cuvette in a final volume of 1 ml. The cuvette was sealed with parafilm and transferrin-bound iron was measured as A 460 nm at 2-s intervals. The data were fitted to a hyperbola by nonlinear least square optimization, and the initial rate was determined from the fitted parameters.
Determination of 55 Fe Release from U937 Cells-U937 cells were grown to a density of about 1 ϫ 10 6 cells/ml in RPMI 1640 medium containing 10% fetal bovine serum. The cells were centrifuged at 1,000 ϫ g for 5 min, the supernatant was aspirated, and the pellet washed with phosphate-buffered saline (PBS). The cells (7 ϫ 10 5 cells/ well in a 12-well plate) were differentiated toward the macrophage lineage by incubation for 16 h with PMA (15 ng/ml). To load the cells with iron, the media and non-adherent cells were aspirated, and the adherent cells were washed and incubated in serum-free RPMI 1640 medium for 24 h. The cells were incubated with 55 Fe-NTA (10 M) in the same medium containing ascorbate (100 M) for 3 h (or in some experiments, 24 h) in a hypoxia chamber (Pro-Ox, Reming, Redfield, NY). The chamber was maintained at 37°C with an atmosphere regulated to contain 1% O 2 using an oxygen controller (Pro-Ox model 110, Reming), and with the remainder occupied by a mixture of 5% CO 2 and 95% N 2 . Cell viability after 3 and 24 h of hypoxia and normoxia was determined by trypan blue dye exclusion; no significant differences in viability were observed and cell viability was greater than 95% under all conditions. The medium was aspirated and the cells were washed with ice-cold PBS containing 100 M EDTA to remove iron non-specifically bound to the cell surface, and twice with ice-cold PBS. 55 Fe-NTA uptake was measured in triplicate wells by lysis in 20% Nonidet P-40 followed by liquid scintillation counting. To measure 55 Fe release, test reagents (apotransferrin, Cp) were dissolved in serum-free RPMI 1640 medium containing 20 mM Hepes at pH 7.4, and the solutions exhaustively bubbled with gases at the appropriate O 2 and CO 2 concentrations before addition (0.5 ml) to cells. The cells were placed in the hypoxia chamber for a 15-min iron release interval (unless otherwise noted). The medium was collected, centrifuged to remove suspended cells and cell debris, and the supernatant was collected and counted by liquid scintillation.
Preparation of Mouse Peritoneal Macrophages-C57BL/6J mice were injected with 0.5 ml of thioglycollate medium into the peritoneum. After 72 h, the mice were sacrificed and the peritoneum was injected with 10 ml of ice-cold PBS. The peritoneal lavage was removed and the cells were centrifuged (1,000 rpm, 10 min, 4°C) and resuspended (50,000 cells per well in a 12-well plate) in RPMI 1640 with 10% fetal bovine serum supplemented with L-glutamine and penicillin/streptomycin. Non-adherent cells were removed by rinsing with PBS.
Determination of the Intracellular Labile Iron Pool by Quenching of Calcein Fluorescence-The intracellular LIP was determined by an assay based on quenching of calcein fluorescence by bound iron (28). PMA-treated U937 cells were iron-loaded by incubating for 3 or 24 h with unlabeled Fe-NTA (10 M) and ascorbate (100 M) in serum-free RPMI 1640 medium. The entire experiment was done in a hypoxic atmosphere consisting of 1% O 2 , 5% CO 2 , and 94% N 2 . The cells were washed with ice-cold PBS containing EDTA (100 M) and then twice with ice-cold PBS. Fresh RPMI 1640 medium was replaced and the iron-loaded cells were permitted to release iron for 15 min in the presence of various test agents. The cultures were then washed and incubated with calcein (0.5 M) in serum-free RPMI 1640 medium (without phenol red) for 10 min. The cells were washed and observed in an inverted fluorescent microscope, and the mean cellular fluorescence intensity (488 nm excitation/517 nm emission) in 15 or more cells was quantitated (Image Pro Plus). To determine the LIP size before the iron release period, one set of wells was examined immediately after calcein treatment. To provide a no-iron control, one set of wells for each treatment condition was incubated for 30 min with desferrioxamine (100 M) to chelate all free and calcein-bound iron. The difference in intensity between a given cell treatment and cells treated with desferrioxamine gives a relative measure of the LIP.
Preparation of Apo-Cp by Reductive Copper Chelation and Reconstitution of Holo-Cp-Apo-Cp was prepared by removal of copper by complexing with cyanide under reducing conditions as described by Musci et al. (29). Cp was dialyzed at 4°C against sodium acetate buffer (100 mM, pH 5.9) containing ascorbate (10 mM) under anaerobic conditions achieved by continuous bubbling with N 2 gas. Dialysis was continued until the "blue" coppers were reduced as shown by complete sample decolorization. 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 sodium acetate (100 mM) containing cysteine (1 mM), and then by an overnight dialysis against buffer containing KCl (150 mM) and MOPS (50 mM, pH 7.0). Holo-Cp was reconstituted from apo-Cp as described (29). A Cu 2ϩ -reduced glutathione complex was prepared immediately before use by addition of copper sulfate (0.1 mM) to reduced glutathione (0.3 mM) in phosphate buffer (0.1 M, pH 7.0) containing ascorbate (100 M). A 7-fold molar excess of the Cu 2ϩ -reduced glutathione complex was added to apo-Cp in KCl (150 mM), MgCl 2 (5 mM), and MOPS (50 mM, pH 7.0) for 4 h. Unbound copper was removed by overnight dialysis against PBS.

Cp Increases Iron Binding to Apotransferrin under Hypoxic
Conditions-Previous studies have shown that Cp stimulated iron binding to apotransferrin under non-physiological, acidic conditions (pH 5.5 to 6.0) (5, 13). We verified this observation and found that Cp increased the initial rate of binding of ferrous iron to apotransferrin in acetate-buffered solution at pH 6.0 by about 10-fold (Fig. 1A). A similar experiment over a range of pH showed that the stimulation of iron binding to apotransferrin by Cp decreased as the pH increased, and the stimulation at pH 7.4 was less than 20% (Fig. 1B). Essentially identical results were observed when acetate buffer was replaced with RPMI 1640 culture medium containing 20 mM Hepes (not shown). The decrease in the apparent stimulatory activity of Cp was most likely because of the high rate of iron autoxidation at physiological pH.
To minimize iron autoxidation, and to replicate tissue conditions, iron binding to apotransferrin was measured at pH 7.4 under conditions of varying degrees of hypoxia. The gas phase included 5% CO 2 because maximal iron binding to apotransferrin requires HCO 3 Ϫ anion, which serves as a bridging ligand between the protein and the iron, and completes the coordination requirements of the metal ion (30). The rate of iron binding to apotransferrin was essentially linear with respect to O 2 concentration in the absence of Cp ( Fig. 2A). In contrast, the rate of iron binding to apotransferrin in the presence of Cp increased dramatically as the O 2 level was increased from 0 to 1% of the total applied gas. The rate of iron binding did not increase further even at an atmospheric O 2 level of 21%. The net stimulation of iron binding to apotransferrin by Cp was about 10 -15-fold at low O 2 (1% or less), whereas the stimulation was 10 -20% under normoxic conditions. To determine whether HCO 3 Ϫ anion influenced the stimulatory activity of Cp, an experiment was done in which the CO 2 level was varied from 0 to 5% while the O 2 level was maintained at 1%. Inclusion of 20 mM Hepes in the medium maintained the pH to within 0.05 units at all CO 2 levels. Cp-independent and -dependent iron binding to apotransferrin both had an absolute requirement for CO 2 . Cp-dependent iron binding was halfmaximal at about 2% CO 2 and near-maximal at about 5% CO 2 (Fig. 2B). The dependence of iron binding on Cp concentration was determined under optimal conditions, i.e. 1% O 2 , 5% CO 2 , and pH 7.4. The initial rate of iron binding to apotransferrin increased almost linearly as Cp was increased up to its physiological plasma concentration of 300 g/ml (Fig. 2C).  (55 M) in the presence (q) or absence (E) of Cp (120 g/ml) in phenol red-free, RPMI 1640 medium at pH 7.4. The O 2 level in the applied gas was varied from 0 to 21% while maintaining CO 2 at a constant 5% level, the remainder was made up with N 2 . The initial rate of iron binding to apotransferrin was measured as described in the legend to Fig. 1. B, iron binding to apotransferrin was measured as in A except that CO 2 levels were varied while maintaining a constant 1% level of O 2 , the remainder was N 2 . C, iron loading into apotransferrin was measured as a function of Cp concentration in 1% O 2 , 5% CO 2 , and 94% N 2 as described in A.
Cp Enhances Iron Release from Monocytic Cells-We investigated the influence of Cp on iron release from monocytic cells using the human promonocytic U937 cell line pretreated with PMA to induce differentiation toward a macrophage phenotype. Under hypoxic conditions, and in the presence of apotransferrin, a linear increase in cellular iron release was observed up to 300 g/ml Cp, the mean plasma concentration in healthy adults (Fig. 3A). At its physiological plasma concentration, Cp stimulated iron release by almost 2.5-fold, and by about 3.5-fold at a concentration of 600 g/ml, the extreme pathophysiological plasma concentration during a severe acute phase response. The small deviation from linearity when the concentration was increased to 600 g/ml was probably not because of saturation of any Cp-mediated process (for example, binding to cell surface receptors or apotransferrin), but rather because of the fact that even in the short, 5-min release period used here, about 85% of the pre-loaded 55 Fe-NTA was released from the cells (Fig. 3B). A control experiment showed that apotransferrin was required for the stimulation of iron release because Cp by itself had no significant influence on release from cells (Fig. 3B).
We investigated the effect of O 2 on basal and Cp-stimulated iron release from U937 cells. In the absence of Cp, iron release was slow, about 20 pmol/well, increasing gradually to more than 50 pmol/well at atmospheric O 2 (Fig. 3C). Cp markedly stimulated iron release, particularly at low O 2 levels. At 0.1 or 1% O 2 , Cp (at 120 g/ml) increased cellular iron release by about 50 to 75% (Fig. 3C). The rate of iron release in the presence of Cp was maximal at 1% O 2 , and did not change as O 2 was increased to 21%. These results were very nearly parallel to the Cp-mediated stimulation of iron binding to apotransferrin ( Fig. 2A). The biphasic nature of the release curve in the absence of Cp suggests that there may be two release mechanisms, possibly O 2 -dependent and -independent mechanisms.
The effect of Cp on iron release from authentic macrophages was also investigated. Mouse macrophages were obtained after peritoneal injection of thioglycollate into C57BL/6J mice. Iron release was measured from macrophages pre-loaded with 55 Fe-NTA. Cp at 300 g/ml nearly doubled iron release from mouse macrophages in the presence of apotransferrin under hypoxic conditions (Fig. 4A). A small (10 to 20%) stimulation of iron release by apotransferrin alone was reproducibly observed. Under normoxic conditions, the combination of Cp and apotransferrin did not increase iron release more effectively than apotransferrin alone (Fig. 4B).
Role of Labile Iron Pool in Cp-stimulated Iron Release from U937 Cells-To investigate whether Cp can accelerate the release of iron in a storage pool, or whether the released iron is restricted to the intracellular LIP, experiments were done under iron uptake conditions likely to influence the amount of iron in these pools. Iron was loaded under four conditions. U937 cells loaded with a low level of 55 55 Fe-NTA (10 nmol) for 3 h in a hypoxic atmosphere maintained at 1% O 2 , 5% CO 2 , and 94% N 2 . Cellular iron release was measured as a function of Cp concentration (E) in the presence of apotransferrin (55 M) and in the same hypoxic atmosphere. After a 5-min release interval, the medium was collected (cells centrifuged for 5 min at 2,000 rpm) and 55 Fe determined by liquid scintillation counting and expressed as picomole/well. B, total 55 Fe taken up by the cells during the uptake phase is indicated (open bar). As controls, iron release was measured in the presence of medium alone or in the presence of Cp (300 g/ml) in the absence of apotransferrin (striped bars). C, iron release from U937 cells was measured after a 15-min release interval as a function of O 2 level balanced by a gas mixture containing 5% CO 2 and 95% N 2 in the presence of apotransferrin (55 M), and in the presence (q) or absence (E) of Cp (120 g/ml). Iron release was measured as in A. signed to minimize the LIP there was no detectable stimulation of iron release by either apotransferrin or by the combination of apotransferrin and Cp (Fig. 5A). In contrast, in cells loaded using a strategy to maximize the LIP there was about a 2-fold stimulation of iron release by apotransferrin and a 3.5-fold stimulation by apotransferrin and Cp added together (Fig. 5B). Cells loaded with iron under conditions expected to give intermediate LIP levels showed only minor enhancements of iron release by apotransferrin and Cp (Fig. 5, C and D). It is noteworthy that the initial amount of iron taken up by cells incubated with 10 nmol of iron for 24 h (Fig. 5D) is almost twice that in cells loaded with 10 nmol of iron for 3 h (Fig. 5B), yet the former exhibits a substantially lower basal rate of iron release, and is not susceptible to stimulation by apotransferrin and Cp. This result indicates that basal and Cp-stimulated iron release does not depend on the absolute amount of iron taken up by the cell, but more likely depends on the size of the LIP.
To confirm the significance of free, intracellular iron in Cpstimulated iron release, the LIP was measured before and after iron release. We took advantage of the highly efficient fluorescence quenching of calcein by iron (31). Cells were iron-loaded by incubation with 10 nmol of iron for 3 h, the condition designed to generate a high level of free intracellular iron. After brief incubation with calcein, the LIP was determined by cellular fluorescence intensity. The fluorescence intensity of cells incubated with medium alone was low indicating that the LIP remained large (Fig. 6A). In contrast, fluorescence intensity of cells incubated with apotransferrin and Cp was substantially higher, indicating a decrease in LIP size. The relative LIP size was quantitated as the difference in fluorescence intensity between treated cells and cells incubated with desferrioxamine (to chelate calcein-bound iron). Apotransferrin decreased the LIP by about 30% compared with medium alone, and Cp in combination with apotransferrin decreased the LIP by almost 60% (Fig. 6B). When cells were incubated with 10 nmol of iron for 24 h the LIP size was nearly an order of magnitude smaller, and was not further reduced by treatment with Cp and apo-transferrin (Fig. 6C). Other experiments confirmed that the other incubation conditions described in Fig. 5, i.e. 1 nmol of iron for 3 or 24 h, also produced cells with a small LIP that was not substantially reduced by Cp and apotransferrin (not shown). Together these studies suggest that the rate of Cpstimulated iron release depends on the LIP size, that the release decreases the LIP, and that the stored iron pool is not susceptible to Cp-stimulated release under the conditions of these experiments.
Cellular Mechanism of Cp-stimulated Iron Release from U937 Cells-The inability of Cp to stimulate cellular iron release in the absence of apotransferrin suggests that the primary function of Cp is the conversion of ferrous ion to ferric ion for efficient binding to apotransferrin. The requirement for Cp ferroxidase activity was tested using a ferroxidase-defective Cp preparation. Copper was removed by chelation under reducing conditions (20,29). Apo-Cp was almost completely ferroxidaseinactive as measured by the ability to stimulate iron binding to apotransferrin (Fig. 7, inset). As a control to show that the chelation procedure did not globally disrupt the protein, copper-depleted Cp was reconstituted to holo-Cp by addition of excess Cu 2ϩ in the presence of glutathione (20,29), and was shown to regain essentially all ferroxidase activity. We tested the ability of these Cp preparations to stimulate iron release from U937 cells. Native Cp, added at the physiological serum concentration of 300 g/ml and in the presence of apotransferrin, stimulated iron release by about 3-fold compared with control, untreated cells (Fig. 7). Ferroxidase-defective apo-Cp had no stimulatory activity, and iron release was nearly the same as in wells containing only apotransferrin. Reconstituted, ferroxidase-active holo-Cp was as active in releasing iron from cells as native Cp. These results show that ferroxidase activity is required for Cp-stimulated iron release from cells.
The requirement for apotransferrin and ferroxidase-active Cp for stimulation of cellular iron release is consistent with the proposed gradient mechanism (5). According to this mechanism, iron is released from cells as Fe 2ϩ , Cp ferroxidase activity converts the ion to Fe 3ϩ that increases its binding to apotransferrin, which in turn creates a negative concentration gradient with respect to ferrous ion, thereby stimulating iron release. We have done several additional experiments to test the validity of the hypothesis in the context of U937 cell iron release. The effect of holotransferrin on iron release was measured because it should be unable to act as an iron ''sink,'' and thus should not facilitate formation of a negative iron gradient. Cp did not stimulate iron release in the presence of holotransferrin (Fig. 8). Surprisingly, holotransferrin reproducibly stimulated iron release by about 20%, i.e. about the same stimulation observed with apotransferrin. However, the inability of Cp to synergize with holotransferrin suggests that the apo-and holoforms of transferrin may use distinct mechanisms of iron release. Furthermore, if the synergistic stimulation of iron release by Cp and apotransferrin depends on formation of an iron gradient, then the addition of excess extracellular iron should inhibit gradient formation and iron release. Indeed, iron-NTA (10 M) completely blocked the stimulation by Cp and apotransferrin (Fig. 8). Iron-NTA added by itself almost completely blocked iron release showing that even basal release requires an iron gradient. Finally, we examined the effect of an artificial gradient on iron release using bathophenanthroline sulfonate, a cell-impermeant iron chelator. The chelator markedly stimulated iron release from the cells, in fact, essentially all of the loaded iron was removed within 15 min. These results are consistent with a mechanism in which Cp, in collaboration with apotransferrin, enhances monocytic cell iron release by creation of a negative iron gradient with respect to the cell surface. DISCUSSION Our results indicate that the gradient hypothesis (5), proposed to explain Cp-stimulated iron release by the gut, also applies to iron release by macrophages. According to the original proposal, iron oxidation by plasma Cp, and subsequent binding to apotransferrin, creates a negative iron gradient with respect to the mucosal cell surface that increases gut iron release. Experiments in vivo and in perfused organs have confirmed the role of Cp in iron release from gut and liver (5,(32)(33)(34). However, recent studies suggest that this release mechanism may not apply to the gut, because iron absorption is normal in Cp-null mice (9). The ferroxidase activity of hephaestin, a Cp homologue in the gut, may substitute for Cp (35,36). However, the perinuclear localization of hephaestin suggests that it is more likely involved in intracellular iron transport than in release. Compelling evidence for a role of Cp in release of iron from reticuloendothelial cells and hepatocytes is provided by studies in Cp-null mice (9). Although the case for Cp mediating inter-tissue iron transport is strong, the cellular and molecular mechanisms remain elusive.
Our results suggest that Cp has an important role in macrophage iron flux during hypoxia. The stimulation of cellular iron release is about 10% at atmospheric O 2 and increases almost linearly to its maximal, 2-3-fold stimulation at 1% O 2 . Thus, Cp is likely to influence cellular iron release under physiological conditions in which the O 2 level is about 5% in well perfused interstitium, decreasing to about 1% within the tissue (37). Early observations that high altitude induces a dramatic erythropoietic response (38) suggests an important role of hypoxia in iron metabolism. Cp may participate in hypoxia-stimulated erythropoiesis via conversion of iron to the Fe 3ϩ form for binding to apotransferrin, or possibly to ferric iron transport systems on cell surfaces (19,20). Hypoxia also induces the expression of important iron regulatory proteins, for example, erythropoietin, the major hormonal regulator of erythroid cell development and erythropoiesis. Hypoxia induces hypoxia-inducible factor-1␣, a transcription factor that binds the erythropoietin gene hypoxia-responsive elements (39). Hypoxia-inducible factor-1 also transactivates other key genes involved in iron metabolism, namely transferrin (40), transferrin receptor (41,42), heme oxygenase (43), and Cp (44,45). Patients with chronic obstructive pulmonary diseases that induce hypoxia have a 35-100% increase in serum Cp (46 -48). In addition, exposure of healthy adults to high altitude hypoxia signifi- FIG. 7. Cp ferroxidase activity is required for enhanced iron release from U937 cells. PMA-treated U937 cells were loaded with 55 Fe-NTA as described in the legend to Fig. 3; total cellular iron uptake was 157 Ϯ 7 pmol/well. The cells were incubated for 15 min with apotransferrin (apo-Tf, 55 M) in the presence of Cp, apo-Cp, or reconstituted (reconst.) Cp (all Cp preparations were added at 300 g/ml) in an atmosphere of 1% O 2 and 5% CO 2 , and 94% N 2 . Iron release was measured as described in the legend to Fig. 3. Inset, ferroxidase (Fxase) activity of Cp, apo-Cp, and reconstituted Cp was quantitated as the rate of iron binding to apotransferrin (apo-Tf) in an atmosphere containing 1% O 2 and 5% CO 2 , and 94% N 2 as described in the legend to Fig. 2.   FIG. 8. Requirement of a negative iron gradient for Cp-stimulated iron release from U937 cells. PMA-treated U937 cells were loaded with 55 Fe-NTA as described in the legend to Fig. 3 (open bar). The cells were incubated with Cp (300 g/ml), apotransferrin (55 M), holotransferrin (55 M), iron-NTA (10 M), and bathophenanthroline sulfonate (BPS, 100 M) as indicated and iron release was measured after 15 min. cantly increases serum Cp (49). The finding that Cp expression is induced by hypoxia provides evidence for an important role of Cp in erythropoiesis. At the post-transcriptional level, hypoxic treatment of hepatoma and erythroleukemic cells increases the interaction of iron regulatory protein-1 with the iron-responsive element in transferrin receptor mRNA, stabilizing the transcript and increasing receptor expression (50). Hypoxia inhibits ferritin translation in these cells by a similar molecular mechanism (50). Given their central role in erythropoiesis and iron homeostasis, surprisingly little is known about the effects of hypoxia on iron metabolism in macrophages. Hypoxia activates hypoxia-inducible factor-1 in macrophages (51,52), but the role of hypoxia-inducible factor-1 in expression of ironrelated genes has not been explored. Interestingly the effect of hypoxia on the iron regulatory protein-1/iron-responsive element interaction in mouse peritoneal macrophages is different from that described in other cells; hypoxia inhibits the binding and enhances ferritin expression (53). Much remains to be learned about the effects of hypoxia on macrophage expression of Cp and other iron-related genes.
An outstanding issue in macrophage iron release is the nature and identity of the putative plasma membrane transporter. One candidate is ferroportin-1 (FPN1), a ferrous ion transporter implicated in basolateral iron release from duodenal epithelial cells (54 -56). Expression of FPN1 in Xenopus oocytes increases iron efflux but the requirement for Cp has not been resolved (55,56). FPN1 is highly expressed in reticuloendothelial cells of the spleen, liver, and marrow, but its intracellular localization in macrophages suggests that it may not be involved in iron export in these cells (54,57). A recent report suggests that coupling of the glycosylphosphatidylinositol-anchored form of Cp to FPN1 facilitates iron efflux from brain astrocytes (24). Likewise, soluble Cp may interact with FPN1 or another cell surface iron transporter, and a saturable Cp receptor or binding site has been described on the surface of monocytes and macrophages (58,59). However, the linear stimulation of cell iron release by up to 600 g/ml Cp suggests a non-saturable process not consistent with a receptor-mediated mechanism.
Fundamental questions on the function of Cp during macrophage iron release remain to be answered, including the chemical form of iron and requirements for ion co-and countertransport. Our results showing that iron transport rates increase linearly up to about 600 g/ml provide a mechanistic basis for the increase of Cp to this level during the acute phase reaction. How people with plasma Cp of one-tenth normal, i.e. about 30 g/ml, maintain normal iron homeostasis remains unclear (10). Finally, it is paradoxical that the increase in plasma Cp during inflammation does not decrease iron storage pools, instead, reticuloendothelial iron stores markedly increase during inflammation (60).