Modulation of Cellular Iron Metabolism by Hydrogen Peroxide

Cellular iron uptake and storage are coordinately controlled by binding of iron-regulatory proteins (IRP), IRP1 and IRP2, to iron-responsive elements (IREs) within the mRNAs encoding transferrin receptor (TfR) and ferritin. Under conditions of iron starvation, both IRP1 and IRP2 bind with high affinity to cognate IREs, thus stabilizing TfR and inhibiting translation of ferritin mRNAs. The IRE/IRP regulatory system receives additional input by oxidative stress in the form of H2O2 that leads to rapid activation of IRP1. Here we show that treating murine B6 fibroblasts with a pulse of 100 μmH2O2 for 1 h is sufficient to alter critical parameters of iron homeostasis in a time-dependent manner. First, this stimulus inhibits ferritin synthesis for at least 8 h, leading to a significant (50%) reduction of cellular ferritin content. Second, treatment with H2O2induces a ∼4-fold increase in TfR mRNA levels within 2–6 h, and subsequent accumulation of newly synthesized protein after 4 h. This is associated with a profound increase in the cell surface expression of TfR, enhanced binding to fluorescein-tagged transferrin, and stimulation of transferrin-mediated iron uptake into cells. Under these conditions, no significant alterations are observed in the levels of mitochondrial aconitase and the DivalentMetal Transporter DMT1, although both are encoded by two as yet lesser characterized IRE-containing mRNAs. Finally, H2O2-treated cells display an increased capacity to sequester 59Fe in ferritin, despite a reduction in the ferritin pool, which results in a rearrangement of59Fe intracellular distribution. Our data suggest that H2O2 regulates cellular iron acquisition and intracellular iron distribution by both IRP1-dependent and -independent mechanisms.

To satisfy metabolic needs for iron, mammalian cells utilize transferrin (Tf), 1 the iron carrier in plasma. Cellular iron up-take involves binding of Tf to the cell-surface Tf receptor (TfR), followed by endocytosis. Within the acidified endosome, iron is released from the Tf-TfR complex and transported, most likely by the Divalent Metal Transporter DMT1, across the endosomal membrane to the cytosol, where it becomes bioavailable for the synthesis of iron proteins. Excess iron is stored in ferritin, a multisubunit protein consisting of H-and L-chains, that serves as the major intracellular iron storage device (reviewed in Refs. [1][2][3]. Sequestration of iron in ferritin is viewed as a detoxification step to reduce the risk of iron-mediated cell damage, which is based on the capacity of iron to catalyze the generation of toxic oxygen radicals (4). Balanced iron homeostasis is critical for health, and both iron deficiency as well as iron overload are associated with severe disorders (5).
At the cellular level, iron homeostasis is accomplished by the coordinate regulation of iron uptake and storage. The expression of TfR and ferritin is mainly controlled post-transcriptionally by iron regulatory proteins, IRP1 and IRP2. Under conditions of iron starvation, IRP1 and IRP2 are activated for high affinity binding to multiple "iron-responsive elements" (IREs) in the 3Ј-untranslated region (UTR) of TfR mRNA and to a single IRE in the 5Ј-UTR of the mRNAs encoding both H-and L-ferritin chains. This stabilizes TfR mRNA (6) and inhibits ferritin mRNA translation (7). Conversely, failure of IRPs to bind to cognate IREs in iron-replete cells leads to degradation of TfR mRNA and synthesis of ferritin (reviewed in Refs. 8 -10). The identification of additional IRE-containing mRNAs suggests that the functional significance of the IRE/IRP system stretches out beyond the control of cellular iron uptake and storage. The mRNAs encoding the enzymes 5-aminolevulinate synthase-2 (involved in erythroid heme synthesis), mammalian mitochondrial aconitase (m-aconitase), and the insect Ip subunit of succinate dehydrogenase (both catalyzing reactions in citric acid cycle) contain a "translation-type" IRE in their 5Ј-UTRs (11)(12)(13)(14)(15)(16). The mRNAs encoding the more recently discovered iron transporters DMT1 (17,18) and ferroportin/IREG1 (19 -21) contain a single and, in terms of function, incompletely characterized IRE in their 3Ј-or 5Ј-UTR, respectively. IRP1 and IRP2 share extensive homology and belong to the family of iron-sulfur cluster isomerases that also includes maconitase. However, their activities are controlled by distinct mechanisms. In iron-loaded cells, IRP1 assembles a cubane 4Fe-4S cluster that converts it to a cytosolic aconitase (c-aconitase) and prevents IRE-binding, whereas IRP2 is oxidized and degraded by the proteasome. Iron starvation increases IREbinding activity by disassembly of the 4Fe-4S cluster in IRP1 and stabilization/de novo synthesis of IRP2 (reviewed in Refs. 8 -10, 22). Iron regulatory proteins are subjected to regulation by additional iron-independent signals, including nitric oxide, hypoxia, and oxidative stress (reviewed in Refs. [23][24][25]. Of particular interest is the rapid induction of IRE binding activity of IRP1 in response to hydrogen peroxide (H 2 O 2 ) (26,27), because this "reactive oxygen intermediate" is implicated in iron toxicity. In the presence of catalytic amounts of ferrous iron, H 2 O 2 yields highly aggressive hydroxyl radicals (Fenton reaction) that readily attack membranes, proteins, and nucleic acids (4). Exposure of different cell types to micromolar concentrations of H 2 O 2 is sufficient to induce a rapid conversion of IRP1 from c-aconitase to the IRE-binding protein within 30 -60 min (26,27) by an incompletely characterized mechanism that involves signaling (28,29). In contrast to this, H 2 O 2 does not affect the activity of IRP2 (30). It should be noted that reactive oxygen species, including H 2 O 2 , are widely viewed as participants in a multitude of signaling pathways. These involve calcium signaling, mitogen-activated protein kinase cascades, tyrosine phosphorylation, regulation of phosphatases and phospholipases, or activation of transcription factors (reviewed in Refs. 31 and 32).
The effects of H 2 O 2 on cellular iron metabolism have been as yet only partially studied. We have previously utilized mouse B6 fibroblasts, a cell line predominantly expressing IRP1 and negligible levels of IRP2, to characterize the mechanism of IRP1 induction by H 2 O 2 (26,28,30,33). We also showed that a treatment of these cells with 100 M H 2 O 2 for 1 h inhibits ferritin synthesis, whereas longer treatments (4 -6 h) increase TfR mRNA levels, as a result of IRP1 activation (26). However, these responses have not been correlated with the biological activity of TfR and ferritin, in terms of iron uptake and sequestration. Here we extend the previous studies and investigate the effects of H 2 O 2 in the expression and function of several IRE-containing mRNAs, as reflected in the uptake of 59 Fetransferrin and intracellular management of 59 Fe.

EXPERIMENTAL PROCEDURES
Materials and Cell Culture-Desferrioxamine (DFO) was purchased from Novartis (Dorval, Canada), and H 2 O 2 was from Merck. Hemin, human apo-and holo-Tf, fluorescein isothiocyanate (FITC)-conjugated holo-Tf, and lactoferrin were from Sigma. B6 fibroblasts were grown and treated with H 2 O 2 as described (26).
Western Blotting-Total cell extracts (in RIPA lysis buffer) were analyzed by Western blotting (34) with antibodies against TfR (Zymed Laboratories Inc.), m-aconitase (a generous gift of Dr. Rick Eisenstein), actin (Sigma), or DMT1 (raised in rabbits against the peptide VFAE-AFFGKTNEQVVE, which corresponds to amino acids 260 -275 in human DMT1). Dilutions for antibodies are indicated in the respective figure legends.
Fluorescence-activated Cell Sorting (FACS)-To determine cell surface expression or the Tf-binding capacity of TfR, cells were scraped, suspended in medium, and tumbled with either 5 l/ml FITC-conjugated mouse TfR antibody (PharMingen) or with 50 g/ml FITC-conjugated human Tf (Sigma), respectively. Where indicated, a 50-fold molar excess human holo-Tf or lactoferrin was added prior to FITC-Tf. Excess FITC label was removed by washing twice with phosphate-buffered saline containing 0.1% bovine serum albumin. Cells were fixed with 3.7% formaldehyde and analyzed for fluorescence on a cell sorter (Beckman Coulter).
Generation of 59 Fe-Tf-59 FeCl 3 (PerkinElmer Life Sciences) was mixed with sodium citrate (1:50 molar ratio in a total volume of 1 ml) and incubated for 1 h at room temperature. The resulting 59 Fe-citrate was mixed with apo-Tf (2:1 molar ratio); the volume was brought up to 4 ml in 0.6 M NaHCO 3 , and incubation was continued overnight. 59 Fe-Tf was separated from 59 Fe-citrate on a Centricon Plus-20 filter (Amicon), and its concentration was calculated spectrophotometrically at 465 nm (⑀ ϭ 4620 M Ϫ1 dm Ϫ1 ).
Cellular Uptake of 59 Fe-Tf and Immunoprecipitation of 59 Fe-Ferritin-Cells were labeled with 59 Fe-Tf in minimal essential medium containing 25 mM Hepes, pH 7.4, 10 mM NaHCO 3 , and 1% bovine serum albumin. Labeling was terminated by washing with ice-cold phosphatebuffered saline, and cells were monitored for radioactivity on a ␥-counter. For immunoprecipitation of 59 Fe-ferritin, cytoplasmic lysates were prepared in the same way as lysates of 35 S-labeled cells (see above), and 1 mg was tumbled at 4°C with 5 l of rabbit polyclonal ferritin antibodies (Roche Molecular Biochemicals). Following addition of protein A-coupled Sepharose CL-4B beads (Amersham Pharmacia Biotech), immunoprecipitated material was washed twice in lysis buffer, and radioactivity was monitored on a ␥-counter.

H 2 O 2 Elicits a Time-dependent Stimulation of TfR and Inhibition of Ferritin Synthesis-
We have shown previously that treatment of cells with micromolar concentrations of H 2 O 2 results in rapid induction of IRP1 to bind to IREs and that IRE binding activity remains elevated for at least 4 h following removal of the inducer (30,33). This observation prompted us to study the effects of H 2 O 2 on the expression of TfR and ferritin, two crucial proteins of iron metabolism under the control of the IRE/IRP system. Our analysis covers intervals of up to 8 h following exposure of cells to a bolus of 100 M H 2 O 2 , allowing IRP1 activity to peak and decrease to basal levels (30). No apparent toxicity was observed by the trypan blue exclusion assay, under all experimental conditions employed in this study, in line with earlier observations that exogenous H 2 O 2 is very rapidly degraded by these cells (33). Nevertheless, a single bolus of 100 M H 2 O 2 is sufficient to sustain a threshold of ϳ10 M H 2 O 2 for about 15 min, which is the minimum concentration required to elicit IRP1 activation (33). Thus, we established experimental conditions to activate IRP1 and study the effects of H 2 O 2 on cellular iron metabolism in the absence of potential toxic side effects of H 2 O 2 . B6 fibroblasts were first treated with 100 M H 2 O 2 for 1 h and metabolically labeled with [ 35 S]methionine/cysteine for 2 h either immediately or at different time points after treatment, and TfR and ferritin synthesis were assessed by immunoprecipitation ( Fig. 1, top  panel). In cells previously treated with the iron chelator DFO (100 M), TfR synthesis is stimulated 3.3-fold compared with untreated control cells, whereas synthesis of ferritin H-and L-chains is strongly inhibited (11 and 12% of control, respectively, lanes 1 and 2). Treatment with H 2 O 2 initially does not affect TfR expression (lanes 2 and 3) but clearly stimulates TfR synthesis by 2-and 2.1-fold, within 4 and 6 h after its withdrawal, respectively (lanes 4-7). Soon afterward, TfR synthesis declines to almost control (1.1-fold) levels (lanes 8 and 9). In contrast to TfR, ferritin expression is affected immediately after H 2 O 2 treatment; synthesis of ferritin H-and L-chains is reduced to 29 and 22% of control (lanes 2 and 3), in agreement with earlier observations (26). Ferritin synthesis remains at low levels even after 4 (28% for H-and 26% for L-chain) and 6 h (35% for H-and 31% for L-chain) following H 2 O 2 withdrawal (lanes 4 -7). After 8 h, ferritin synthesis only partially (60%) recovers, even though TfR synthesis has essentially returned to basal levels (lanes 8 and 9). As a control, the non-iron-regulated protein Sam68 (68-kDa Src substrate associated during mitosis) was immunoprecipitated from TfR/ferritin-immunodepleted supernatants. Synthesis of Sam68 essentially remains unchanged during the course of the treatment (Fig. 1 whereas iron chelation dramatically reduces ferritin to 6% of control levels (lanes 1-3). Treatment with 100 M H 2 O 2 for 1 h initially decreases the ferritin content to 69% (lanes 3 and 4). Further reductions to 55 and 42% are evident 2 and 4 h after H 2 O 2 withdrawal, respectively (lanes 5 and 6). Ferritin concentration tends to increase very slightly to 49 and 47% after 6 and 8 h (lanes 7 and 8), in line with the partial recovery in de novo ferritin synthesis at these time points (Fig. 1). We conclude that H 2 O 2 leads to a marked reduction in the ferritin pool for at least 8 h after the treatment.

, bottom panel). Effects of H 2 O 2 on the Steady-state Levels of TfR and
To examine whether stimulation of TfR synthesis by H 2 O 2 is associated with an increase in TfR concentration, we analyzed steady-state levels of TfR by Western blotting (Fig. 2C). Treatment of cells with 100 M H 2 O 2 for 1 h leads to gradual accumulation of TfR after 2-8 h (lanes 3-8). H 2 O 2 -mediated induction of TfR reaches a maximum 6 and 8 h after the treatment (1.9-and 1.8-fold, respectively). As expected, treatments with DFO or hemin result in 2.2-fold increase and 0.6-fold decrease of TfR, respectively (lanes 1-3). In this experiment, cells were solubilized in RIPA lysis buffer, to extract membrane-bound TfR efficiently, but similar results were obtained with cytoplasmic extracts (not shown). Fig. 2C suggest that H 2 O 2 stimulates TfR expression. We next designed experiments to address whether this is accompanied by increased Tf binding activity. The fraction of TfR expressed on the cell surface is crucial for Tf binding. In a previous report it was shown that H 2 O 2 negatively affects the size of this fraction, at least in human hematopoietic K562 and HL-60 cells (35). In light of these findings, we analyzed relative changes in cell surface expression of TfR in mouse B6 fibroblasts by means of FACS, using FITC-conjugated TfR antibodies (Fig. 3A). The levels of TfR on the cell surface essentially remain unaltered within 2 h after exposure of cells to H 2 O 2 (100 M H 2 O 2 for 1 h) (lanes 3-5), but increase by 1.4-, 1.5-and 1.9-fold within 4, 6, and 8 h, respectively (lanes 6 -8). A profound cell surface expression of TfR is achieved by treatment with DFO, whereas administration of hemin does not appear to cause any notable alterations (lanes 1-3).

H 2 O 2 Leads to Increased Expression of Functional TfR on the Cell Surface-The data shown in
By having established that exposure of cells to H 2 O 2 is associated with increased expression of TfR, including its cell surface fraction, we then employed a functional assay to evaluate the effects of H 2 O 2 on Tf binding activity. Cells were incubated with FITC-conjugated Tf under conditions allowing its binding to TfR. Changes in relative fluorescence were then monitored by FACS (Fig. 3B (Fig. 3B, bars 3-8). The increase was slightly elevated when incubations were performed at 37°C (compare 1.2-, 1.4-, and 1.7-fold at 4°C with 1.4-, 1.7-, and 1.8-fold increase at 37°C, 4, 6, and 8 h after treatment, respectively). Consistent with the data described above, iron chelation with DFO elicits stronger effects on FITC-Tf binding to TfR than H 2 O 2 (up to 3.8-fold induction, Fig. 3B, bar 2). As expected, the effects of hemin are inhibitory (bar 1).
The specificity of the FITC-Tf binding assay is illustrated in Fig. 3C. Co-incubation of FITC-Tf with 50-fold excess non- labeled Tf competitor strongly reduces fluorescence intensity to 24.4% in untreated and to 10% in DFO pretreated cells. In contrast, addition of 50-fold excess lactoferrin as a nonspecific competitor only slightly interferes with FITC-Tf binding (ϳ15% reduction). Incubations with these competitors were performed at 37°C, and similar results were obtained at 4°C (not shown). Taken together, our findings suggest that exposure of B6 cells to H 2 O 2 leads not only to an increase in TfR steady-state levels but also stimulates its cell surface expression and the Tf-binding capacity. These conditions are predicted to favor enhanced cellular iron uptake from Tf.  (Fig. 3), the differences in 59 Fe uptake in response to these stimuli are not particularly strong, suggesting that the Tf-TfR cycle may be subjected to additional controls. Nevertheless, these data show that H 2 O 2 -treated cells have an increased capacity to take up iron.
Under the conditions of the iron uptake experiment (e.g. 6 -8 h following H 2 O 2 treatment), ferritin synthesis is still partially repressed (Fig. 1), whereas cellular ferritin content has dropped to Ͻ50% of control levels (Fig. 2B). Since ferritin plays a major role in iron detoxification as an iron-storage sink, we wondered how cells respond to increased iron uptake when ferritin levels are reduced. To address this question, B6 fibroblasts were labeled with 5 M 59 Fe-Tf (as in Fig. 4A) for 15 and 30 min and 1 and 2 h. Cytoplasmic extracts were analyzed by quantitative immunoprecipitation with ferritin antibodies, and ferritin-associated 59 Fe was plotted against the time of labeling (Fig. 4B). Ferritin immunoprecipitates from H 2 O 2 -treated cells display a marked increase in 59 Fe content compared with untreated control cells. After 2 h of labeling, ϳ8.16 pmol of 59 Fe/mg protein in extracts of control cells are associated with ferritin, whereas this value increases to ϳ14.85 pmol of 59 Fe/mg protein (181%) in extracts from H 2 O 2 -treated cells. Ferritin-associated 59 Fe in extracts of cells pretreated with DFO is very low (ϳ0.9 pmol of 59 Fe/mg of protein after 2 h of labeling, representing 11% of control), most likely due to sequestration of iron by the chelator. When cells were prelabeled with 59 Fe-Tf for 2 h and then left untreated or treated with H 2 O 2 , no differences in the amount of ferritin-associated 59 Fe were observed (not shown).
These data suggest that H 2 O 2 -treated cells have an increased capacity to store newly internalized iron in ferritin, despite the reduction in the translation and in the intracellular pool of ferritin. They also imply that their fraction of ferritinassociated 59 Fe is significantly enriched. To calculate the distribution of 59 Fe in control, H 2 O 2 -, and DFO-treated cells, we also measured radioactivity in the ferritin-immunodepleted extracts and in the insoluble cell fraction (similar methodology has been employed by others (36)) and depicted the results in form of pie charts (Fig. 4C) (Fig. 4A), the former store ϳ1.66 and the latter ϳ3.05 pmol of 59 Fe in ferritin. This represents an almost 2-fold increase under conditions where only half the amount of ferritin is available (Fig. 2B).
The Steady-state Levels of m-Aconitase and DMT1 Are Not Affected by H 2 O 2 -By having established that H 2 O 2 modulates the expression (and the function) of TfR and ferritin, we asked whether H 2 O 2 also affects the abundance of m-aconitase and DMT1, both encoded by IRE-containing mRNAs. Western blotting analysis at different time points after treatment of B6 cells with H 2 O 2 does not show any significant alterations in steadystate levels of m-aconitase (Fig. 5A, top panel, lanes 3-8).
Overnight iron perturbations with DFO or hemin yield a similar outcome (lanes 1 and 2). Probing with an antibody against ␤-actin (bottom panel) suggests that the slight reduction in the intensity of the m-aconitase band on lane 7 is of no functional importance and rather reflects unequal loading.
The effects of H 2 O 2 on DMT1 mRNA were assessed by Northern blotting. Probing with a mouse DMT1 cDNA reveals two hybridizing bands of 3.1 and 2.3 kilobases (Fig. 5B, top panel) that possibly correspond to the non-IRE and IRE-containing isoforms of DMT1 mRNAs, respectively (37,38). By normalizing to the ␤-actin signal (bottom panel), no obvious differences in the intensity of both bands are observed between samples from untreated control (lane 5), H 2 O 2 -treated (lanes 1-4) or iron-perturbed cells (lanes 5-7). This finding is also mirrored at the protein level; Western blotting with antibodies against DMT1 (Fig. 5C) shows a faint band with an apparent molecular mass of ϳ65 kDa that has the same intensity in samples from untreated control, H 2 O 2 -treated, or iron-perturbed cells (lanes [1][2][3][4][5]. Since the sizes of polypeptides encoded by the IRE-and non-IRE-DMT1 mRNAs differ by only 7 amino acids, the ϳ65-kDa band most likely corresponds to a mixture of both isoforms. The specificity of the interaction is demonstrated on lane 6. Probing the filter in the presence of excess DMT1 antigenic peptide does not produce the 65-kDa signal. Thus, in B6 cells, the expression of both non-IRE-and IRE-containing isoforms of DMT1 mRNAs does not appear to respond to iron or H 2 (Fig. 1) as a result of the accumulation of TfR mRNA ( Fig.  2A). In contrast, ferritin (at least H-chain) mRNA levels remain unaltered for up to 8 h after H 2 O 2 challenge ( Fig. 2A), but ferritin (H-and L-chains) synthesis is strongly inhibited immediately after H 2 O 2 withdrawal and slowly recovers afterwards (Fig. 1). As IRP1 is known to stabilize TfR mRNA and inhibit ferritin mRNA translation by binding to their respective IREs, these responses underlie the causal relationship of IRP1 induction by H 2 O 2 . In kinetic terms, the activation of IRP1 is in perfect agreement with the regulatory effects on its downstream targets. The translational inhibition of ferritin is rapid and temporally coincides with the increase in IRE binding activity (30), whereas accumulation of TfR mRNA is delayed and follows its stabilization by binding of IRP1. The decline of IRE binding activity to basal levels Ͼ4 h after H 2 O 2 treatment (30) is associated with a decrease in TfR synthesis, as a result of TfR mRNA destabilization, and gradual recovery of ferritin mRNA translation (Figs. 1 and 2). We conclude that H 2 O 2 modulates the expression of ferritin and TfR via activation of IRP1.
We have also studied the effects of H 2 O 2 (and iron donors/ chelators) on the abundance of m-aconitase and DMT1. Both  4 -6). Aliquots of 6.5 ϫ 10 5 cells (duplicates) were suspended in medium, mixed first with the indicated competitors and then with 50 g/ml FITC-conjugated Tf (Sigma), and tumbled at 37°C for 40 min. Following wash (to remove excess FITC label) and fixation, fluorescent cells were analyzed by FACS. Columns 1 and 4 (gray bars), no competitor; columns 2 and 5 (white bars), 50-fold molar excess of holo-Tf; columns 3 and 6 (black bars), 50-fold molar excess of lactoferrin. Remaining FITC-Tf binding in the presence of competitors is calculated compared with respective controls (lanes 1-3 and 4 -6).
proteins are encoded by IRE-containing mRNAs. However, under our experimental conditions, we did not observe any ironor H 2 O 2 -dependent alterations in their steady-state levels (Fig.  5). The IRE in m-aconitase mRNA is located in the 5Ј-UTR and is functional as a translational regulator in vitro (13,14). However, the range of iron-dependent regulation of m-aconitase translation in vivo lags orders of magnitude behind the respective range of ferritin regulation (14,15,39). A potential explanation for this is offered by the structural differences between m-aconitase and ferritin IREs. The former contains a C-bulge and the latter an internal loop/bulge that confer to them differential binding specificity toward IRP1 and IRP2 in vitro (40). The functionality of m-aconitase IRE in de novo synthesis of m-aconitase has been demonstrated by sensitive immunoprecipitation assays following iron perturbations and metabolic labeling of several cell lines with [ 35 S]methionine (15). Relatively small but significant effects of iron on the steady-state levels of m-aconitase have been documented by Western blotting analysis of mouse (14) and rat (39) tissues following long term (over several weeks) modulation of dietary iron intake. In light of these data, we conclude that short term (Ͻ12 h) iron perturbations or treatments with H 2 O 2 are not sufficient to lead to any detectable alterations in m-aconitase steady-state levels (Fig. 5A).
The IRE in DMT1 mRNA is located in the 3Ј-UTR and has as yet only partially been characterized. The levels of DMT1 mRNA (IRE-containing isoform) are increased in iron-deficient enterocytes from duodenal samples of hemochromatosis patients (41) or HFE Ϫ / Ϫ mice (42). In addition, a radiolabeled DMT1 IRE probe is functional in gel retardation assays with cell extracts (38). 2 Taken together, these results would argue for a role of the IRE in controlling the stability of DMT1 mRNA. However, whereas a single IRE is sufficient to function as a translational control element, earlier experiments showed that the minimum requirement for regulating the stability of TfR mRNA is defined by a combination of more than one IRE together with additional non-IRE sequences (43). According to the findings in Ref. 43, the single IRE in the 3Ј-UTR of DMT1 mRNA would not qualify in its own right as a regulator of its stability via IRE/IRP interactions. This view is supported by the data presented in Fig. 5B. The Northern analysis does not reveal any significant differences in the abundance of the 2.3and 3.1-kilobase transcripts, following treatments with H 2 O 2 (over 8 h) or iron donors/chelators (overnight). The lack of iron responsiveness in the abundance of DMT1 mRNA is in agreement with recent data (38,44). We speculate that the iron-dependent regulation in the expression of the IRE-containing isoform of DMT1 mRNA in enterocytes may involve additional factors that are not present in B6 cells. However, it should be noted that we do not have sufficient information on the relative distribution of DMT1 encoded by the IRE-or non-IRE forms of DMT1 mRNA, which may be an important factor for the overall interpretation of the data. Along these lines, it is not unexpected that under our experimental conditions iron or H 2 O 2 essentially has no effect on DMT1 steady-state levels (Fig. 5C). 2 A. Caltagirone, G. Weiss, and K. Pantopoulos, unpublished data.   (31,32). We thus wondered whether exposure of cells to H 2 O 2 affects the expression and function of genes of iron metabolism at different levels, either upstream or downstream of the IRE/IRP regulatory system (for example transcriptionally or post-translationally). There is evidence that ferritin synthesis is transcriptionally activated in response to various forms of oxidative stress as part of a homeostatic antioxidant defense mechanism (45-47). More recently, a functional "antioxidant response element" has been identified in the promoters of L- (48) and H-ferritin (49).
This element is shared in promoter regions of several phase II detoxification genes and functions as a transcriptional enhancer in response to pro-oxidant stimuli. Treatment of mouse BNL CL.2 normal liver cells or Hepa1-6 hepatoma cells with Ͼ250 M H 2 O 2 stimulated a delayed (after 8 h) transcriptional activation of H-and L-ferritin mRNAs via the antioxidant response element, which gradually overcame the initial IRP1mediated translational inhibition of ferritin synthesis (49). The data presented in Fig. 2A suggest that in B6 cells H 2 O 2 fails to increase ferritin mRNA levels over 8 h (at least that of ferritin H-chain). Although it is conceivable that application of more stringent conditions of oxidative stress may stimulate ferritin mRNA transcription, it is apparent that low micromolar concentrations of H 2 O 2 exhibit solely inhibitory effects on ferritin expression. These are reflected in the decrease in ferritin synthesis (Fig. 1) and the reduction of ferritin pool (Fig. 2B). In preliminary pulse-chase experiments, H 2 O 2 did not appear to affect ferritin half-life (not shown), suggesting that H 2 O 2 -mediated translational inhibition of ferritin synthesis suffices to reduce dramatically (Ͻ50% of control levels) the intracellular ferritin pool for at least 8 h.
Previous studies in human hematopoietic K562 and HL-60 cells showed that oxidative stress (either in the form of menadione or extracellular H 2 O 2 ) results in a rapid (within 30 min) redistribution of TfR in intracellular compartments without alterations in TfR levels (35,50). These results appear to be IRP1-independent and are in contrast to the stimulatory effects of H 2 O 2 on the synthesis, accumulation, cell surface expression, and Tf binding activity of TfR observed in B6 cells (Figs. 1-3). It is well established that the regulation of TfR expression is more complex in erythroid cells, the major iron consumers in the body, and involves transcriptional as well as post-transcriptional mechanisms (51)(52)(53)(54). Thus, it is conceivable that additional, IRP1-independent pathways regulate TfR in various cell types in response to H 2 O 2 . Nevertheless, it would be interesting to investigate the effects of H 2 O 2 on IRP1 activity in K562 and HL-60 cells.
In B6 fibroblasts, the H 2 O 2 -mediated increase in TfR expression correlates with a modest (ϳ11.5%) but significant increase of 59 Fe-Tf uptake (Fig. 4A), despite the fact that Tf binding activity is stimulated 1.7-fold (Fig. 3B). Similarly, iron-starved cells (treated with DFO) take up ϳ24.9% more 59 Fe-Tf than untreated controls, despite a 3.1-fold induction in Tf binding activity (Figs. 4A and 3B). These findings imply that intracellular iron release during the Tf-TfR cycle may be controlled at additional checkpoints but are also compatible with the idea that subtle perturbations in intracellular iron balance may be sufficient to elicit significant pathophysiological responses.
The experiments with 59 Fe-Tf yielded another unanticipated result. H 2 O 2 -treated cells have an increased capacity to store 59 Fe in ferritin, at a time point where the ferritin pool is dramatically reduced (Figs. 4B and 2B), leading to changes in intracellular 59 Fe distribution (Fig. 4C). The reason for this is not clear, but it is tempting to hypothesize that H 2 O 2 signaling interferes with the incompletely defined mechanism of iron sequestration in ferritin. It has been proposed that ferritin subunits may be arranged in a flexible and dynamic structure allowing iron entry/release by localized unfolding. In this sense, it is conceivable that changes in iron entry in response to extracellular stimuli may be associated with post-translational modification of ferritin that could affect such localized unfolding (55). In fact, there is evidence in older literature that ferritin can be phosphorylated in vitro (56). Along these lines, it will be interesting to examine the phosphorylation status of ferritin in cells, following a treatment with H 2 O 2 . From a physiological point of view, the increased capacity of ferritin to  Fig. 2C were analyzed by Western blotting. The membrane was probed with 1:500 diluted rabbit antibody raised against bovine heart m-aconitase (a generous gift of Dr. Rick Eisenstein) and reprobed with 1:200 diluted rabbit antibody against ␤-actin; m-aconitase and ␤-actin were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech). Lanes are as in Fig. 2C. B, analysis of DMT1 and ␤-actin mRNAs by Northern blotting. 15 g of total RNA from B6 cells, treated as indicated, was resolved on an agarose gel (1%), electrotransferred onto a nylon membrane, and hybridized to the respective 32 P-radiolabeled cDNA probes. Radioactive bands were visualized by autoradiography. Lanes 1-4, 100 M H 2 O 2 (1 h), following wash and further incubation for another 2, 4, 6 or 8 h; lane 5, untreated control; lane 6, 100 M hemin (overnight); and lane 7, 100 M DFO (overnight). C, analysis of DMT1 by Western blotting. 30 g total cell extracts from B6 cells, treated as indicated, were resolved by SDS-PAGE on a 10% gel and electrotransferred onto a nitrocellulose membrane. The membrane was probed with 1:200 diluted rabbit antibody against DMT1, and DMT1 was detected by a chromogenic (alkaline phosphatase) reaction. sequester iron may have evolved for protection against ironmediated injury and certainly adds to the complexity of cellular responses to oxidative stress. In summary, by utilizing B6 cells as a model system, we conclude that H 2 O 2 elicits complex effects on cellular iron metabolism. These are both dependent and independent from the IRE/IRP regulatory system and may be further complicated in various cell types.