Hypoxia alters iron-regulatory protein-1 binding capacity and modulates cellular iron homeostasis in human hepatoma and erythroleukemia cells.

Ferritin and transferrin receptor expression is post-transcriptionally regulated by a conserved mRNA sequence termed the iron-responsive element (IRE), to which a transacting protein called the iron-regulatory protein (IRP) is bound. Our data demonstrate that hypoxia powerfully enhances IRE/IRP-1 binding in human cell lines. Using the human hepatoma cell line Hep3B as a model, we found that 16 h in a 1% oxygen atmosphere markedly increases IRE/IRP-1 binding as assessed by electromobility shift assay. Hypoxia also decreased cytosolic aconitase activity. The hypoxia-enhanced IRE/IRP-1 binding stabilized the transferrin receptor message, increased the cellular mRNA content by over 10-fold, and doubled surface receptor expression. Simultaneously, hypoxia suppressed ferritin message translation. Hypoxia's effect was most strikingly depicted by the absence of ferritin synthesis in cells challenged with inorganic iron. Our results contrast with previously reported data (Hanson, E. S., and Leibold, E. A. (1998) J. Biol. Chem. 273, 7588-7593) in which a 3% oxygen atmosphere reduced IRE/IRP-1 binding in rat hepatoma cells. We discuss some possible reasons for the differences. In aggregate with other investigations involving responses to hypoxia, iron, or nitric oxide, our data indicate that cellular iron metabolic responses are complex and that IRE/IRP-1 interactions vary between cell lines and perhaps between species.

Iron is a key element in cellular growth and metabolism. The element is part of the active site of many enzymes, often as a component of heme or as part of an iron-sulfur complex (1). As an enzyme prosthetic group, iron catalyzes redox reactions involving proteins, lipids, carbohydrates, and nucleic acids. The ability of the element to exist in either of two stable oxidation states (ferric, Fe 3ϩ ; ferrous, Fe 2ϩ ) is the key to its enzymatic activity. Unrestrained, however, iron can wreak havoc on cells through the production of reactive intermediates (2), including the deadly hydroxyl radical (OH . ).
Cellular iron is closely regulated, in part through the actions of the transferrin receptor and ferritin, the proteins of cellular iron uptake and storage, respectively (3). Expression of these two proteins is largely regulated at the post-transcriptional level (4). A conserved 28-base sequence termed the iron-responsive element (IRE) 1 exists in the ferritin 5Ј-UTR, whereas five such elements are located in the transferrin receptor 3Ј-UTR (5). The two cytoplasmic iron-regulatory proteins 1 and 2 (IRP-1 and IRP-2) recognize and bind to the IRE (6,7). IRE/ IRP-1 binding in the 5Ј-UTR blocks message translation. Such binding in the 3Ј-UTR stabilizes the message against enzymatic degradation (8).
Regulation of IRE/IRP-1 binding is a complex affair (9). Iron modulates IRP-1 binding to IRE elements by forming a 4Fe-4SH cluster within the protein. When the cluster is intact, IRP-1 cannot bind to the IRE and exists free in the cytosol (10). In this state, IRP-1 has aconitase activity. In the absence of iron, the 4Fe-4SH cluster collapses, aconitase activity is lost, and IRP-1 acquires IRE binding capacity. Therefore, IRP-1 is a dual function protein, serving alternatively as a cytoplasmic aconitase or as a transacting RNA-binding protein. In contrast to IRP-1, the IRP-2 molecule lacks aconitase activity and is susceptible to proteolytic degradation when it is free in the cytosol (11,12).
Investigators have described a number of other factors that modify IRE/IRP-1 binding and ferritin synthesis such as H 2 O 2 and nitric oxide (13,14). Cytoplasmic IRP-1 appears to exist in equilibrium between forms that either possess or lack aconitase activity. Ascorbic acid shifts the equilibrium toward aconitase (ϩ) IRP-1 (15). This creates a thermodynamic "sink" that favors greater IRP-1 release when iron interacts with the IRE/ IRP-1 complex.
The current work uses the human hepatoma cell line Hep3B to examine the effect of hypoxia on cellular iron metabolism. These cells mimic fetal hepatocytes in many respects and were the first cells shown to produce human erythropoietin in response to hypoxia (16). Their response to hypoxia has been studied extensively, making them an excellent model system. Hypoxia induces the expression of HIF-1 in Hep3B cells, a universal transcription factor that activates genes needed to adapt to low oxygen tensions (17,18). The expression of HIF-1 (and its DNA binding capacity) increases as the oxygen tension falls, peaking at a 0.5-1% oxygen concentration (19). We now demonstrate that hypoxia promotes powerful IRE/IRP-1 binding in these cells that resists even iron-mediated dissociation. Hypoxia markedly alters the transcription of several genes (17), including those encoding erythropoietin (epo) and vascular endothelial growth factor (18 -22). The promoters of these genes contain binding sites for HIF-1. Hypoxia changes gene transcription by altering HIF-1 binding (23). In distinction to these well-established effects of hypoxia, the effect on IRE/ IRP-1 interaction is a post-transcriptional phenomenon. Hanson and Leibold (24) reported previously that exposing rat hepatoma cells to a 3% atmosphere decreased IRP-1 binding to the IRE but did not alter IRP-2 binding. We discuss possible reasons for the differences.
K562 human erythroleukemia cells were maintained in RPM 1640 medium supplemented with 10% fetal bovine serum (Biowhittaker) and penicillin/streptomycin at a density of 5 ϫ 10 5 cells/ml. Hypoxia treatment was as outlined above.
Electromobility Shift Assay-IRP binding to RNA was assessed by electromobility shift assay as described previously (4,15). Briefly, cytoplasmic cell extracts were prepared from Hep3B or K562 cells grown under conditions of hypoxia or normoxia in the presence or absence of 100 M desferrioxamine or 10 g/ml ferric ammonium citrate. An excess quantity of [ 32 P]UTP-labeled RNA transcript (from pSPT-fer, a 28nucleotide fragment encoding the human H-ferritin IRE; a generous gift of Dr. Kü hn (25)) was incubated at room temperature for 30 min with 3 g of protein of fresh cytoplasmic cell lysate. RNase T1 (1 unit/reaction) and heparin (5 mg/ml) were added sequentially for 10 min each. IRE-IRP-1 complex was analyzed on a 6% nondenaturing polyacrylamide gel, as detailed previously (15).
Northern Blot Analysis-RNA was isolated from control or experimentally manipulated cells using a STAT-30 kit from TEL-TES B, Inc. (Friendswood, TX), following the manufacturer's instructions. Twenty g of RNA were separated on an agarose/formaldehyde gel and immobilized to Hybond-N nylon membrane (Amersham, Arlington Height, IL) using a 0.05 N fixation (26). The RNA was hybridized with a 600-bp pStI fragment of TfR cDNA labeled by the Random primer method according to the supplier's protocol. After hybridization and washing by the method of Church and Gilbert (27), the membrane was exposed to x-ray film for 2 days. The TfR signal was stripped off by boiling the membrane in distilled water for 1 min. The membrane then was reprobed with human L-ferritin cDNA probe (28).
Nuclear Run On Assay-Nuclei were isolated from Hep3B cells grown under normoxic or hypoxic conditions using a modified version of the protocol of Greenberg and Ziff (29). Briefly, 5 ϫ 10 7 cells were harvested, washed with phosphate-buffered saline, and treated with 4 ml of hypotonic solution (5 M NaCl, 1 M Tris-HCl, pH 7.4, 1 M MgCl 2 , and 1 l/ml ␤-mercapthoethanol) for 15 min on ice, followed by 0.1% Triton and 140 mM KCl. The cells were homogenized, and the nuclei were collected after repeated centrifugation. The newly synthesized RNA was labeled with [ 32 P]UTP and hybridized to a nitrocellulose membrane containing the coding region of ferritin L chains (28), TfR (25), or epo (22). After a 72-h hybridization at room temperature, the membrane was washed twice with 2ϫ SSC/0.2% SDS for 20 min and 0.2ϫ SSC/ 0.2% SDS for 40 min at 68°C and exposed to x-ray film for 3 days.
Cytosolic Aconitase Assay-Cytosolic aconitase activity was assessed by the consumption of cis-aconitate as measured by the spectrophotometric absorbance at 240 nm (as detailed previously in Ref. 15). Briefly, mitochondrial-free lysates from Hep3B cells grown under conditions of hypoxia or normoxia (50 g) were incubated with 200 M cis-aconitate in 50 mM Tris-HCl buffer, pH 7.2, 100 mM NaCl, and 0.02% bovine serum albumin at room temperature in a volume of 1 ml. Specific activity (M substrate converted/mg protein/min) was calculated as described by Emery-Goodman et al. (30).
Cytosolic Ferritin Content Determination-The steady-state level of cytosolic ferritin was determined using a Coat-A-Count Ferritin IRMA assay kit whose limit of ferritin detectability is 0.1 ng/ml (Diagnostic Products Corp., Los Angeles, CA). Briefly, 10 6 cells/assay were harvested, washed, and solubilized in a detergent buffer (1% Triton X-100, 150 mM NaCl, 20 mM Tris-HCl, pH 7.5, and 0.2 mM phenylmethylsulfonyl fluoride). Ferritin was immunoprecipitated with monoclonal antibody to the L subunit and detected with a second antibody labeled with 125 I.
Ferritin Biosynthesis-To assess ferritin biosynthesis, 10 6 cells were incubated for 1 h at 37°C in methionine-free medium (RPMI 1640 or ␣-minimum essential medium, without serum) supplemented with 250 Ci/ml [ 35 S]methionine. The cells were washed and solubilized in the lysis buffer (above), and the newly synthesized ferritin subunits were immunoprecipitated with the polyclonal antibody against ferritin. The ferritin subunits were separated on SDS-phosphate-urea gel and visualized by autoradiography (26).
Transferrin Receptor Expression-After treatment, cells were chilled to 4°C and incubated for 1 h with 125 I-labeled transferrin at concentrations ranging from 0.5 to 500 nM (30). Plates were washed with cold phosphate-buffered saline and lysed in situ in a lysis buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 0.1% Triton X-100), and the cell-associated radioactivity was measured. A parallel set of cells was incubated with 125 I-labeled transferrin along with a 100-fold excess of unlabeled transferrin to assess nonspecific binding. Scatchard analysis provided the number of receptor sites/cell as well as the binding affinity. The total transferrin receptor pool was determined by solubilizing the cells in detergent buffer and removing the insoluble debris by centrifugation. A 150-l aliquot of cell extract was incubated with 50 l of 125 I-labeled transferrin (0.5-500 nM) for 30 min with or without a 100-fold excess of unlabeled ligand at room temperature. The reaction was stopped by adding 250 l of ice-cold 60% ammonium sulfate. The precipitate containing the transferrin/transferrin receptor complex was pelleted by centrifugation, washed, and assessed for radioactivity using a gamma counter (31).

RESULTS
Hypoxia Enhances IRE/IRP-1 Binding-We used the electromobility shift assay to assess IRE/IRP-1 interaction in control Hep3B cells (21% O 2 ) and cells exposed to 1% O 2 for 16 h. Fig. 1 shows that the baseline IRE/IRP-1 binding (C) in the

FIG. 1. Hypoxia promotes IRE/IRP-1 binding in Hep3B cells.
Hep3B cells (10 6 cells at 75% confluence) were grown for 16 h under normoxic (21% O 2 ) or hypoxic conditions (1% O 2 ). In addition to untreated cells for each condition (C), cells were incubated with 100 M desferrioxamine (D) or 10 g/ml ferric ammonium citrate (Fe). Three mg of fresh cell extract were incubated with a 32 P-labeled IRE probe. The IRE/IRP-1 complex was resolved on a 6% polyacrylamide gel that was subsequently fixed, dried, and used to expose x-ray film. Parallel cell extracts were used for IRE/IRP-1 binding in an incubation mixture containing 2% ␤-mercaptoethanol, which induces maximal IRE/IRP-1 binding. The experiment was repeated several times, and one representative result is shown.
hypoxic cells exceeds that in the normoxic cells by at least 10-fold.
Consistent with a substantial body of previous information, IRE/IRP-1 binding rose dramatically when cells in a 21% O 2 atmosphere were treated for 16 h with 100 M desferrioxamine (D). By contrast, desferrioxamine treatment did not increase IRE/IRP-1 binding in cells in a 1% O 2 atmosphere. Therefore, chelation of intracellular iron does not further enhance hypoxia-stimulated IRE/IRP-1 binding. This suggests that hypoxia promoted IRE/IRP-1 association to the maximum possible extent.
With each electromobility shift assay, a parallel incubation was performed that included 2-mercaptoethanol. This agent promotes maximal IRE/IRP-1 binding when added to the in vivo incubation mixture. As shown in Fig. 1, maximal IRE/IRP binding was equivalent for cells in a 21% O 2 or 1% O 2 atmosphere. Therefore, hypoxia increases the extent of IRE/IRP-1 binding but does not change the overall IRE/IRP-1 binding capacity of the cells.
The expected decrease in IRE/IRP-1 binding that occurred with the addition of iron in the form of ferric ammonium citrate to the normoxic cells was paralleled in the hypoxic cells. Therefore, hypoxia enhances IRE/IRP-1 binding without altering its physiological response to iron.
The 16-h period of hypoxia did not affect the viability of Hep3B cells as assessed by trypan blue staining. Hypoxia enhanced IRE/IRP-1 binding in cells exposed to 1% O 2 for only 4 h (Fig. 2). The increase was only 2-fold over the baseline, suggesting that the metabolic changes that augment IRE/IRP-1 binding develop over time.
We also examined the effect of hypoxia on IRE/IRP-1 binding in K562 cells. Fig. 3 shows that 16 h of hypoxia enhanced IRE/IRP-1 binding by only 2-fold (as assessed by densitometric scanning). In contrast, incubating the cells with erythropoietin at a concentration of 50 units/ml boosted IRE/IRP-1 binding by 5-fold, consistent with the observations by other investigators (32). Therefore, K562 cells can shift IRE/IRP-1 binding characteristics in response to environmental stimuli. The response of Hep3B cells and K562 cells to hypoxia was qualitatively similar but differed in magnitude. The standard electromobility shift assay separates RNA/protein complexes by charge. The technique does not separate IRE/IRP-1 and IRE/IRP-2 in extracts derived from human hepatoma or erythroleukemia cells (in contrast to murine cells) (32). Western blot analysis indicates that the level of IRP-2 protein is very low in these cell lines (data not shown). Supershift experiments were inconclusive.
Hypoxia Decreases Cytosolic Aconitase Activity-Cytosolic aconitase activity and RNA binding capacity are mutually exclusive properties of IRP-1. The hypoxia-induced increase in IRE/IRP-1 binding should lessen the amount of free cytosolic IRP-1 protein. IRP-1 is an aconitase enzyme when it is free in the cytosol, but it lacks this activity when it is part of an IRE/IRP-1 complex. On this basis, we predicted that hypoxia would lower cytosolic aconitase activity.
K562 cells and Hep3B cells exposed to 1% O 2 for 16 h were lysed, and the 100,000 ϫ g supernatant was collected. Cytochrome c oxidase activity was 1.5 unit in the pellet and was undetectable in the supernatant, indicating no contamination by this mitochondrial enzyme. The same should hold for mitochondrial aconitase (15). As shown in Table I, hypoxia markedly decreased cytosolic aconitase activity in both cell lines. The magnitude of the decline produced by hypoxia was similar to that seen with desferrioxamine treatment. The significant decrease in enzyme activity correlates well with the marked increase in the binding of IRP-1 to IREs. Although hypoxia shifts IRP-1 onto IRE-containing transcript, the aconitase that remains in the cytosol still shows iron-dependent regulation in a 1% oxygen atmosphere (  and eukaryotic organisms, hypoxia can induce transcriptional activity of physiologically important genes (for a review, see Ref. 17). Northern blot analysis (Fig. 4) showed an increase of greater than 10-fold in the steady-state level of TfR mRNA in Hep3B cells exposed to hypoxia. Nuclear run on assay was performed to determine whether the observed increase was due to enhanced transcription. We predicted that the hypoxia-induced increase in transferrin receptor message represented stabilization of the message due to binding of the IRP-1 to the IREs in the 3Ј-UTR of the transcripts.
Nuclei were isolated from control Hep3B cells or from cells exposed to 1% O 2 for 16 h. Hypoxia affected neither the cell viability nor yield of isolated nuclei. Fig. 5 shows no effect by hypoxia on the transcription of the transferrin receptor message. In contrast, hypoxia increased the synthesis of erythropoietin message, as expected (16). These data confirm that hypoxia increases TfR mRNA levels in Hep3B cells by stabilizing the message.
Hypoxia Increases Transferrin Receptor Surface Expression-An increase in IRP binding to the IREs in the 3Ј-UTR of the transferrin receptor message raises the level of transferrin receptor mRNA and secondarily increases transferrin receptor expression. This phenomenon is seen most strikingly in cells treated with desferrioxamine. Because hypoxia mimics the effect of desferrioxamine on IRE/IRP-1 binding and transferrin receptor mRNA expression in these cells, we predicted that hypoxia would also increase transferrin receptor protein expression. Fig. 6 shows that surface expression of transferrin receptors doubles in Hep3B cells exposed to hypoxia for 16 h (2.5 ϫ 10 4 versus 4.9 ϫ 10 4 receptors/cell). Scatchard analysis of the data shows no change in ligand/receptor affinity (K d ϭ 8.6 nM). Binding studies also show that the total transferrin binding capacity of the cells increased by more than 2-fold (5 versus 13.7 ng/10 6 cells). The increase in transferrin receptor message produced by 16 h of hypoxia is functionally significant in that it substantially raises transferrin receptor surface expression. In contrast, a 4-h period of hypoxia had no effect on transferrin receptor expression (data not shown).
Hypoxia Decreases Cellular Ferritin Synthesis and Content-A second functional consequence of IRE/IRP-1 binding is impaired translation of messages containing an IRE element in the 5Ј-UTR. For example, desferrioxamine chelates intracellular iron, increases IRE/IRP-1 binding, and blocks translation of the ferritin message. We predicted that the increase in IRE/IRP binding produced by hypoxia would also block translation of the ferritin message.
We therefore assessed ferritin synthesis in control cells and in cells exposed to hypoxia for 16 h. Fig. 7 is a polyacrylamide gel electrophoresis of [ 35 S]methionine-labeled ferritin immunoprecipitated with polyclonal anti-ferritin antibody. Ferritin synthesis is diminished modestly in the hypoxic cells relative to controls. Iron normally increases ferritin synthesis by dissociating the IRP from the IRE. Fig. 7 shows that this occurred in control cells. However, hypoxic cells did not increase ferritin synthesis in response to iron. This indicates that the hypoxiainduced IRE/IRP-1 binding is very tight and resists dissociation in response to iron.
Reduced ferritin synthesis in hypoxic cells should reduce the cellular ferritin content. We assessed the ferritin content of control cells and hypoxic cells using a commercially available radioimmunoassay kit. Table II shows that the ferritin content of hypoxic cells is about one-third that of the control. Also consistent with the previous ferritin biosynthesis studies, the  4. Hypoxia substantially increases the steady-state level of the transferrin receptor message. RNA was isolated from untreated control (C) Hep3B cells, cells treated with 100 M desferrioxamine (D), or 10 g/ml ferric ammonium citrate (Fe) for 16 h under normoxia (21% O 2 ). Parallel isolations were performed for hepatoma cells exposed to hypoxia (1% oxygen for 16 h). 20 g of RNA were separated on a 1% agarose formaldehyde gel and immobilized to a Hybond (Amersham) nylon membrane. The blot was consecutively hybridized with a 600-bp PstI fragment of TfR and a 670-bp PstI fragment of L-ferritin. The fragments were radiolabeled using the random primer method, with the hybridization and washing done according to Church and Gilbert (27). The figure shows a representative result selected from several experiments. Viability was analyzed with trypan blue staining, and 10 7 nuclei were used from each cell population. cDNA (3 g/lane) of the corresponding genes was immobilized to the nitrocellulose membrane. The newly synthesized mRNAs were labeled with [ 32 P]UTP, and 10 7 cpm/ml probe were used for hybridization. After 72 h, the membranes were washed as described by Greenberg and Ziff (29) and exposed to x-ray film for 4 days. The experiment was repeated twice with similar results. ferritin content of control cells treated with 10 mg/ml ferric ammonium citrate was 10-fold greater than the baseline. In striking contrast, iron salt increased the ferritin content of hypoxic cells by only about 20%. DISCUSSION The interplay of IRE/IRP-1 binding in the post-transcriptional regulation of ferritin and transferrin receptor synthesis was first described in the context of iron-mediated changes (1,5). In the presence of iron, the reactive center of the IRP-1 protein forms a 4Fe-4SH structure that confers aconitase activity to IRP-1, the first of the two presently known IRP proteins to be described. In the absence of iron, the active site collapses, obliterating aconitase activity. This change is counterbalanced by the acquisition of RNA binding capacity with the IRE sequence as the binding target.
Factors other than iron modulate IRE/IRP-1 interaction. Nitric oxide promotes IRP-1 binding to the IRE (33, 34). The likely mechanism is an attack by the free radical on the iron in the 4Fe-4SH cluster. The result is a marked reduction in ferritin synthesis, along with a rise in transferrin receptor expression. Ascorbic acid also strikingly alters the synthesis profile of ferritin and the transferrin receptor. Ascorbate alone does not change ferritin synthesis (15,26). However, the vitamin strikingly increases ferritin synthesis in response to iron. Ascorbate does not directly alter the IRE/IRP-1 interaction but rather shifts free IRP-1 between states with or without aconitase activity (15). The result is that a larger fraction of IRP-1 dissociates from the ferritin message in response to iron.
The present report details our experience with the effects of hypoxia on cellular iron metabolism using two human cell lines, K562 and Hep3B. We focused on the latter because of the extensive studies of hypoxia in the expression of epo and HIF-1-mediated gene responses (20 -23). Electromobility shift assay shows that IRE/IRP-1 binding increases substantially in cells maintained for 16 h in a 1% O 2 atmosphere. The phenomenon occurred in both Hep3B and K562 cells, although the magnitude was much greater in the former. The electromobility shift assay involves the addition of radiolabeled RNA probe (in this case, IRE) to cytosol isolated from control and hypoxia-treated cells. The increase in IRE/IRP-1 binding in this assay means that the change in the IRP produced by the in vivo manipulation of cellular oxygen status persists after the cells have been lysed.
Several readouts of cell activity indicate that hypoxia enhances IRE/IRP-1 binding in living cells as well. Messages with IRE elements in the 3Ј-UTR resist enzymatic digestion once IRPs bind to the IREs (5). The striking increase in transferrin receptor message levels in cells exposed to hypoxia attests to IRE/IRP-1 binding in vivo.
The decrease in ferritin biosynthesis in hypoxic cells is further indication of significantly enhanced in vivo IRE/IRP-1 binding. Attachment of IRP-1 to IRE elements in the 5Ј-UTR blocks the translation of the transcript. Hypoxia decreases ferritin translation without altering message levels. The effect is most prominent in hypoxic cells that are simultaneously exposed to iron. Hypoxia completely abrogates the iron-mediated enhancement of ferritin synthesis.
IRE/IRP-1 binding as a biochemical event alters the biology of the cell. Transferrin receptor expression doubles in Hep3B cells exposed to hypoxia for 16 h. Scatchard analysis shows an identical transferrin binding affinity for control and hypoxic FIG. 6. Hypoxia increases cell surface expression of the transferrin receptor. Hep3B cells (5 ϫ 10 5 ) were exposed to hypoxia for 16 h. Cell surface 125 I-labeled transferrin binding was measured as described under "Materials and Methods." Briefly, cells were incubated in 25 mM HEPES, pH 7.4, 150 mM NaCl, and 1 mg/ml bovine serum albumin with the indicated concentrations of 125 Ilabeled transferrin (0.5-50 nM) with or without 100-fold excess cold transferrin for 1 h at 4°C. After the binding, the buffer was removed, the cells were solubilized in situ, and the radioactivity was measured by gamma counting. The number of binding sites/cell (after correcting for nonspecific binding) are as follows: control, 2.5 ϫ 10 4 ; and hypoxia-treated cells, 4.9 ϫ 10 4 . The dissociation constant (K d ) was 8.6 nM for both cell populations. The binding studies were repeated several times, and the result of one representative experiment is shown.
FIG. 7. Hypoxia reduces baseline ferritin synthesis and blocks iron-stimulated ferritin synthesis. Hep3B cells (5 ϫ 10 6 ) at 75% confluence were treated with 10 g/ml ferric ammonium citrate (Fe) for 16 h under conditions of normoxia (21% oxygen) or hypoxia (1% oxygen). Control and iron-treated cells were washed with phosphate-buffered saline and metabolically labeled with [ 35 S]methionine in a methioninefree medium for 2 h. The cells were washed and solubilized in a buffer containing 1% Triton X-100, 0.15 M NaCl, 0.02 M Tris-HCl buffer, pH 7.5, and 0.2 mM phenylmethylsulfonyl fluoride. The newly synthesized, labeled ferritin molecules were immunoprecipitated with human antiferritin antibody, and the complex was separated on a 15% SDS-phosphate-urea gel. The experiments were repeated several times, and one representative result is shown. cells. The increase in transferrin surface binding therefore reflects an increase in surface expression of transferrin receptors rather than a change in the binding characteristics of a fixed number of receptors.
Hanson and Leibold (24) examined rat hepatoma cells in a 3% oxygen atmosphere and found a time-dependent decrease in IRE/IRP-1 binding. These cells also contain substantial quantities of IRP-2, whose binding to the IRE was not affected by hypoxia. Several possibilities could account for the divergent results between their experiments and our experiments.
The most apparent difference is that they used a rodent cell line, whereas our cell lines were derived from humans. One difference in this respect is that their cell line prominently expresses both IRP-1 and IRP-2, whereas our cell lines produce less IRP-2. The metabolic machinery may be set differently in cells from the two species with respect to the hypoxia response owing to this difference in IRP-1 and IRP-2 expression. Another example of a cell-specific response occurs in rat oligodendroglial cells that increase ferritin production under hypoxic conditions, whereas astrocytes and neurons do not (35). Interestingly, mouse peritoneal macrophages that are of similar derivation as oligodendroglial cells also show decreased IRE/ IRP binding and increased ferritin synthesis with hypoxia (36).
Pleiotrophic cell and tissue responses to a particular stimulus may be due in part to the fact that IRP-1 and IRP-2 have distinct binding characteristics for IRE structures with variations in the base sequences of the loop structure (37). Differences in loops and bulge/loops between IRE isoforms produce dramatic differences in the relative binding affinity of IRP-1 and IRP-2 (38). IRP-2 has the greatest variation in interactions with IRE isotypes, raising the possibility that IRP-2 contributes substantially to differences in IRE-dependent regulation in vivo. IRP-1 and IRP-2 function independently as translational repressors in vivo (39). This fact is driven home strikingly by a cell line that expresses no IRP-1 and yet responds appropriately to all iron-mediated stimuli solely through IRP-2 (40). The differences between IRP-1 and IRP-2 may allow fine tuning to a host of stimuli.
The relative expression of IRP-1 and IRP-2 varies greatly between species and between tissues in a single species (37,41). These differences have likely contributed to some of the conflicting results in the literature concerning changes in cellular iron metabolism in response to various stimuli. A case in point involves nitric oxide effects on cellular iron homeostasis. One group of investigators reported that the effect of nitric oxide is slow and analogous to desferrioxamine (42), whereas another group found the changes to be rapid (43). The explanation for the divergent reports may be the fact that in some cells, such as macrophages, differences in relative expression of IRP-1 and IRP-2 bring the differences in their binding specificities into greater relief (44).
Clearly, the cellular response to hypoxia is both important and complex. Genes are activated, enzyme levels change, and the local generation of free radicals is altered. This work raises the possibility that the relative expression of IRP-1 and IRP-2 also contributes to the adaptation of particular cells to hypoxia. More work is needed to tease apart the many variables in these systems and to understand this important biological response.  10 6 K562 and Hep3B cells were treated with iron as ferric ammonium citrate at a concentration of 10 g/ml or desferrioxamine at a concentration of 100 M for 16 h under conditions of hypoxia (1% O 2 ) or normoxia (21% O 2 ). Hypoxia did not affect cell viability as judged by trypan blue staining. The cells were harvested, washed with phosphate-buffered saline buffer three times and lysed with the same buffer used for electromobility shift experiments. The ferritin content was measured using a Coat-A Count Ferritin IRMA kit (Diagnostic Products Corp.) according to the supplier's manual. All treatments were performed in duplicate and repeated several times.