Overexpression of Wild Type and Mutated Human Ferritin H-chain in HeLa Cells

Transfectant HeLa cells were generated that expressed human ferritin H-chain wild type and an H-chain mutant with inactivated ferroxidase activity under the control of the tetracycline-responsive promoter (Tet-off). The clones accumulated exogenous ferritins up to levels 14–16-fold over background, half of which were as H-chain homopolymers. This had no evident effect in the mutant ferritin clone, whereas it induced an iron-deficient phenotype in the H-ferritin wild type clone, manifested by ∼5-fold increase of IRPs activity, ∼2.5-fold increase of transferrin receptor, ∼1.8-fold increase in iron-transferrin iron uptake, and ∼50% reduction of labile iron pool. Overexpression of the H-ferritin, but not of the mutant ferritin, strongly reduced cell growth and increased resistance to H2O2 toxicity, effects that were reverted by prolonged incubation in iron-supplemented medium. The results show that in HeLa cells H-ferritin regulates the metabolic iron pool with a mechanism dependent on the functionality of the ferroxidase centers, and this affects, in opposite directions, cellular growth and resistance to oxidative damage. This, and the finding that alsoin vivo H-chain homopolymers are much less efficient than the H/L heteropolymers in taking up iron, indicate that functional activity of H-ferritin in HeLa cells is that predicted from thein vitro data.

Ferritins are the major iron storage proteins, ubiquitous in mammalian cells and tightly regulated by iron. They are made of 24 subunits that assemble in an almost spherical shell to delimit a cavity where iron is accommodated and concentrated in a mineral and compact form. Mammalian ferritin are composed of two subunit types, the H-and L-chains, with ϳ50% sequence identity and very similar three-dimensional structures made of a four-helix bundle (1). The mRNAs for the two chains have nearly identical iron-responsive elements (IREs) 1 close to the 5Ј termini that bind to the iron sensors ironregulatory proteins (IRPs) with a mechanism that determines a tight translational iron-dependent regulation of protein expression (2,3). Iron supplementation to cells in culture determines a strong up-regulation of both ferritin chains, whereas treatment with chelating agents such as desferrioxamine determines an almost total suppression of ferritin accumulation in about 24 h (4). Ferritin function has been studied mainly in vitro, using horse spleen ferritin first and recombinant proteins later. The recombinant ferritins have the property of being composed by a single subunit type, the H-chain, L-chain, or artificial variants (5). Although these homopolymers are virtually nonexistent in mammalian cells, the approach proved instructive and showed important functional differences between the two gene products. The recombinant L-chain homopolymers purify from Escherichia coli as iron poor proteins, induce a slow uptake of iron from Fe(II) salts, and are more stable to denaturants than the H-homopolymers (6,7). The L-subunit has no catalytic activity on its own, but it facilitates the activity of the H-subunits by offering sites for iron nucleation and mineralization and increasing the turnover at the ferroxidase centers (8 -10). Recombinant H-chain homopolymers purify from E. coli as iron-containing molecules and induce fast iron oxidation (5). This is due to the presence of ferroxidase centers buried inside the protein fold and consisting in di-iron binding sites coordinated by atoms of seven residues that are conserved in most ferritins from animals, plants, and bacteria (1,11). These catalytic sites accelerate Fe(II) oxidation, which is a rate-limiting step in the mechanism of ferritin iron incorporation, in a reaction that consumes one dioxygen molecule/two Fe(II) ions with the production of hydrogen peroxide (10). Present understanding of the mechanism of in vitro ferritin iron uptake indicates that Fe(II) ions move inside the cavity through hydrophilic channels on the 3-fold symmetry axes and the transfer is facilitated by the presence of local metal binding sites (12). The iron then localizes in the ferroxidase centers where a fast interaction with O 2 occurs with the formation of a transient peroxodiferric complexes that readily decay (13), the resulting -oxo bridged di-iron intermediates then split, and the mononuclear Fe(III) complexes move to the cavity where they hydrolyze to build iron cores of up to 4000 iron atoms (1,14). The vacated ferroxidase center is then available for other cycles of iron oxidation (10). In vitro, in parallel to the catalyzed reaction, spontaneous iron oxidations occur with the stoichiometry of four Fe(II) ions per O 2 molecule and the formation of water. These reactions predominate at high iron concentrations and high pH values and occur also inside the cavity on iron core surface after it has reached a sufficiently large size (15). Thus, in the ferritin iron is oxidized on the catalytic center and on the surface of the mineral core, but the physiological relevance of the latter is dubious. In fact, in conditions in which spontaneous iron hydrolysis does not occur, ferritin can still incorporate iron, although with lower efficiency, as in the case Fe(III) chelated to citrate or nitrilotriacetate (NTA) in the presence of ascorbate. This reaction, which is possibly closer to that occurring in the cell, necessitates of the ferroxidase activity of H-chain and the integrity of hydrophilic channels (16).
The in vivo functionality of ferritin is more complex than that predicted by the in vitro data, because of interactions with other molecules and the poor characterization of the iron forms accessible to the ferritin inside the cell. The analysis of the distinct roles of H-and L-subunits is facilitated by the bypassing of the IRE-IRPs iron-mediated regulation that equally affects the two chains. This occurs in the hereditary hyperferritinemia cataract syndrome where heterogeneous point mutations in the IRE sequence of L-subunit determine a constitutive up-regulation of the L-ferritin expression with protein accumulation in serum and tissues to levels 10 -20-fold higher than normal, without evident effects on body iron metabolism (17,18). Analysis of immortalized cells from these subjects showed that the accumulated L-subunit homopolymers were incompetent in iron incorporation within the cells (19). To study the role of H-subunit, its cDNA was inserted under strong promoters and used to obtain transient transfected COS cells. However the system proved not very informative, because the exogenous ferritin did not coassemble with the endogenous ones and the resulting transfectant ferritins were apparently nonfunctional in iron incorporation or in modifying cellular iron metabolism (20). More instructive was the production of mouse erythroleukemic (MEL) cells lines stably transfected with mouse ferritin H-chain cDNA in which the IRE was inactivated and H-ferritin was constitutively up-regulated (21). The cells accumulated H-ferritin up to 5-fold over background and were found to be iron starving and to have significant reduction of labile iron pool; in addition they showed reduced heme and hemoglobin synthesis, were more resistant to oxidative damage, and had multidrug resistance properties (21)(22)(23). The results indicate that the H-subunit has active roles in regulating iron-related metabolic processes, whereas the L-chain seem inactive. This opens the question of whether the H-chain activity is dependent only on its ferroxidase activity. With the aim of clarifying this point, we produced HeLa cell lines transfected with cDNA for the full human H-chain and a variant with inactivated ferroxidase activity. The cDNAs were under the tetracycline-responsive promoter, and protein expression was induced by withdrawal of doxycycline. It was found that in HeLa cells H-chain overexpression induced iron deprivation, which reduced cell growth and increased resistance to oxidative damage by H 2 O 2 . These effects were reverted by prolonged cell supplementation with ferric salts and were absent in the mutant ferritin transfectants. The results show that H-ferritin functional activity in HeLa cells is mediated by the ferroxidase activity and indicate that most of the features of the in vitro mechanism of ferritin iron incorporation apply to the in vivo conditions.

EXPERIMENTAL PROCEDURES
Plasmid Construction and Cells Culture-The cDNAs for the human entire coding sequences of human H-chain wild type and H-222 mutant (E62K, H65G) were generated by polymerase chain reaction and subcloned into pUDH10 -3 vector (24) under the control of tTA promoter to obtain pUD-HFt and pUD-222Ft plasmids that encode for the entire ferritin H-chain and its mutant 222, respectively. HeLa-tet Off cells (CLONTECH) were co-transfected with 3.5 g of pUD-HFt or pUD-222Ft plasmids and with 1 g of pTK-Hyg plasmid (5:1 molar ratio) (CLONTECH) using calcium phosphate method (25). The colonies were grown in DMEM (Life Technologies, Inc.) with hygromycin D (150 g/ml) and doxycycline (2 g/ml), and the surviving ones were screened for the integration of ferritin genes by polymerase chain reaction. Clones were further tested for protein expression by growing them up to 30 days in the absence of doxycycline and analyzing ferritin content in the cell extracts. The selected cells were maintained in DMEM (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (CLON-TECH), 100 g/ml G418 (Geneticin, Sigma), 150 g/ml hygromycin D (CLONTECH), 100 units/ml penicillin, 100 g/ml streptomycin, 1 mM L-glutamine, and with (Doxϩ) or without (DoxϪ) 2 ng/ml doxycycline (Sigma).
Ferritin Quantification and Immunoblotting-Extracts of 10 6 cells were analyzed for ferritin content by using ELISA assays based on monoclonal antibodies specific for the H-ferritin (rH02) and the Lferritin (LF03) calibrated on the corresponding recombinant homopolymers expressed in E. coli (26). Protein content was evaluated by BCA method (Pierce) calibrated on bovine serum albumin. In immunoblot experiments 3 g of soluble proteins were loaded on 12% SDS-PAGE, and the nitrocellulose filters were incubated with anti-TfR antibody (dilution 1:1000) (Zymed Laboratories Inc.) followed by secondary, peroxidase-labeled antibody (Envision, DAKO). Bound activity was revealed by ECL (Amersham Pharmacia Biotech).
Metabolic Labeling and Immunoprecipitation-Cells (5 ϫ 10 5 ) were grown for 1 h in DMEM w/o L-glutamine, methionine, and cysteine (ICN), 0.5% fetal calf serum, 0.5% bovine serum albumin and then were labeled for 18 h with 50 Ci/ml [ 35 S]methionine (ICN) in the same medium (19). The cells were washed with phosphate-buffered saline and then lysed with 500 l of lysis buffer (20 mM Tris-HCl, pH 8.0, 200 mM LiCl, 1 mM EDTA, 0.5% Nonidet P-40). Total radioactivity associated to the soluble proteins were determined by trichloroacetic acid precipitation. For immunoprecipitation studies, 4 ϫ 10 6 cpm of cytosolic lysates were precleared by incubation with 30 l of protein A-Sepharose 50% v/v (Sigma) for 1 h at 4°C with gentle shaking and centrifuged for 1 min at 14,000 rpm. Then anti-ferritin L-chain antibody LFO3 (30 g) and protein A-Sepharose (30 l) were added, the samples were incubated for 1 h at 4°C, and the precipitates were collected. The soluble fractions were further incubated for 1 h at 4°C with 30 g of antiferritin H-chain antibody rH02 and protein A-Sepharose (30 l) and precipitated (19). The immunobeads were washed resuspended in SDS buffer, boiled for 10 min, and loaded on 12% polyacrylamide SDS-PAGE. The gels were treated with autoradiography image enhancer (Amplify; Amersham Pharmacia Biotech), dried, and exposed. The intensity of ferritin subunit bands was quantified by densitometry. In other experiments, after 18 h labeling with [ 35 S]methionine, the cells were washed twice with Dulbecco's phosphate-buffered saline and then grown in the complete medium. At the indicated times the cells were harvested and lysed, and 10 g of soluble extract protein were subjected to immunoprecipitation with anti-H-ferritin antibody as above.
Iron Regulatory Protein Activity-Electromobility shifts assays for IRP activity were performed as in Ref. 27. Cells (2 ϫ 10 5 ) were lysed, and 2-g samples of soluble proteins were incubated with a molar excess of 32 P-labeled H-ferritin IRE probe, RNase T1, and heparin in the presence or the absence of 2% 2-mercaptoethanol. RNA protein complexes were separated on 6% nondenaturing PAGE and exposed to autoradiography.
LIP Assays-The cellular LIP was measured essentially as described in Ref. 23. Cells (2 ϫ 10 6 ) were incubated with 0.250 M calcein-AM (Molecular Probes) for 15 min at 37°C in bicarbonate free medium containing 1 mg/ml bovine serum albumin and 20 mM HEPES, pH 7.3. After washing the cells were resuspended in 2 ml of HBS (145 mM NaCl, pH 7.2, 20 mM HEPES), and 1 ml of cell suspension was placed in a stirred, thermostated (37°C) cuvette; the fluorescence was monitored at an excitation of 488 nm and an emission of 517 nm in an instruments LS50B (Perkin-Elmer). The quenching of calcein by intracellular iron was revealed by addition of 100 M isonicotionoyl salicylaldehyde hydrazone (kindly donated by Dr. P. Ponka).
Cellular 55 Fe Incorporation-Cells (2 ϫ 10 5 ) were grown for 18 h in the presence of 1 M 55 Fe-transferrin in DMEM, 0.5% fetal calf serum, 0.5% bovine serum albumin. The cells were washed and lysed in 0.3 ml of lysis buffer, and after centrifugation 10 l of the soluble fraction was mixed with 0.3 ml of Ultima Gold (Packard) and counted for 1 min in a scintillator counter (Packard). The soluble proteins were analyzed also on nondenaturing PAGE (7% polyacrylamide) directly or after immunoprecipitation (19). Gels were dried and exposed to autoradiography.
In the experiments to evaluate the effect of cell integrity on ferritin iron uptake, the cells in DMEM, 0.5% fetal calf serum, 0.5% bovine serum albumin or in lysis buffer were added to 2 Ci/ml 55 Fe-NTA (1:10), 200 M ascorbic acid and incubated for 3 h at 37°C, and the protein was separated in nondenaturing PAGE.
Analysis of Cell Growth-To evaluate the rate of cell growth, the cells were counted, and 10 4 of them were plated in 96 wells culture plates (Greiner) in octaplicate in normal medium. They were grown for 18 h at 37°C in DMEM, then MTT was added, and after 3 h the developed color was read at 570 nm (28), following manufacturer's instructions (MTT assay, Sigma). In the experiments in which the cytotoxic effect of H 2 O 2 was analyzed, the cells were plated at different concentrations (1.5 ϫ 10 4 of the induced Hwt-tTA cells and 10 4 of the other cells) to obtain analogous reading at 570 nm after 18 h growth at 37°C. Then cells were washed with phosphate-buffered saline, incubated for 1 h in serum free medium, added of different concentrations of H 2 O 2 , and further incubated for 2 h. Finally, the plates were washed twice, and cellular vitality was measured by incubating the cells in MTT for 3 h as above.

RESULTS
Ferritin Expression-The cDNAs for human H-ferritin wild type and for its mutant named 222 in which ferroxidase activity was inactivated by the substitutions E62K ϩ H65G (29) were subcloned into the pUHD10-3 vector under the control of inducible tetracycline responsive promoter (24). The constructs were used to transfect HeLa cells, and among the stable transfectant clones obtained we selected the ones that expressed the highest levels of ferritin after induction. The clones were named Hwt-tTA and H222-tTA, respectively. Ferritin levels in the cell homogenates were measured by specific ELISA assays, and we found that the ferritin expression was maximally repressed at doxycycline (Dox) concentrations Ͼ2 ng/ml and fully derepressed with Dox concentrations Ͻ0.015 ng/ml (not shown). Upon withdrawal of the drug (from 2 ng/ml Dox), the H-ferritin concentration increased steadily in the H222-tTA clone up to day 7 and then leveled off, whereas H-ferritin accumulation in Hwt-tTA clone showed a biphasic pattern with a maximum around day 4 followed by a decrease (Fig. 1). The maximum H-ferritin concentrations reached in the two clones were similar and 14 -16-fold higher than that of untransfected parent cells (Table I). In the repressed Doxϩ H222-tTA clone H-ferritin level was analogous to that of untransfected HeLa cell, whereas in the Hwt-tTA clone it was slightly higher possibly for some leakiness. The concentration of the partner L-ferritin was consistently higher in the repressed Doxϩ than in the derepressed DoxϪ clones (Table I). For a qualitative analysis of the isoferritins the cells were metabolically labeled with [ 35 S]Met, and the homogenates were immunoprecipitated first with anti-ferritin L-chain and then with anti-H-chain antibodies, a system that separates ferritin heteropolymers from the H-homopolymers (20). The H-chain homopolymers were undetectable in the untransfected and repressed cells, as expected, whereas they were highly represented in both derepressed clones (Fig. 2, lanes 5 and 9). Densitometry showed that the ferritin heteropolymers in derepressed DoxϪ cells (Fig. 2, lanes  2 and 7) had a H-to L-chain ratio of about 20 to 1, whereas in the repressed or untransfected cells, the ratio was in the range of 4 -5 to 1 (Fig. 2, lanes 1, 3, and 8).
Cells were metabolically labeled with [ 35 S]Met for 18 h, and then the ferritins immunoprecipitated at various times with anti-H-chain antibody. Ferritin remained detectable up to 24 h after labeling and disappeared at similar rates in the untransfected parent cells and in the transfected cells, repressed or not (Fig. 3). The H to L ratio did not change throughout the experiments, indicating that the two chains degrade at similar rates and that the homo-and hetero-polymers of Hwt or H222 mutant have similar stability. The calculated half time of about 18 -24 h is in agreement with that found in human fibroblasts (30).
Ferritin Overexpression and Cellular Iron Metabolism-Electromobility shift assays for IRE binding activity of the IRPs were carried out on the cell extracts. The activity, indicated by the intensity of the IRE-IRP complex, was analogous in the repressed and derepressed H222-tTA clone, whereas it was about 5-fold higher in the derepressed than the repressed Hwt-tTA clone (Fig. 4A). In derepressed Hwt-tTA clone, the IRP activity was down-regulated by iron supplementation, although less so than in the corresponding repressed cells, and it was only marginally up-regulated by desferrioxamine treatment, likely because already highly activated (Fig. 4B). The effect of IRP up-regulation in the derepressed Hwt-tTA clone was manifested by a 2.5-fold higher accumulation of transferrin receptor, detected by Western blotting, compared with the other cells (Fig. 4C).
The influence of Hwt or H222-ferritin overexpression on cellular LIP was measured by a fluorescent permeable metal sensor calcein-AM (23). Cells were loaded by 15 min of incubation with 0.250 M calcein-AM at 37°C, and after extensive washing the highly permeable iron chelator isonicotionoyl salicylaldehyde hydrazone was added. The increase in fluorescence secondary to isonicotionoyl salicylaldehyde hydrazone sequestration of calcein bound iron, which cause dequenching, provides a reliable index of labile iron concentration. The variation of fluorescence was analogous in the derepressed and repressed H222-tTA clone (Fig. 5B), whereas in the repressed Hwt-tTA clone it was about 2-fold higher than in the dere-  pressed conditions, implying that H-ferritin overexpression reduced iron available to calcein binding (Fig. 5A). Cells were incubated for 18 h with 55 Fe-Tf, and, after washing, the radioactivity associated to the soluble fraction of cell homogenates or to ferritins was determined. Total cellular iron uptake was analogous in the parent HeLa cells and in the transfected and repressed clones, whereas it increased of about 1.8-fold in the derepressed Hwt-tTA, but not in the derepressed H222-tTA clone (Fig. 6A). Autoradiography of 55 Fe labeled cellular homogenates separated on nondenaturing PAGE showed that essentially all protein-bound radioactivity was associated to ferritin, and that band intensities were stronger in the derepressed Hwt-tTA clone than in control cells (Fig.  6B). To further analyze in which ferritin compartment the iron was allocated, the 55 Fe-labeled homogenates were immunoprecipitated with saturating amount of anti-L-ferritin antibody to remove ferritin heteropolymers and then analyzed on nondenaturing PAGE in parallel with the untreated homogenates. Densitometry of the autoradiographies showed that the fraction of ferritin-iron associated to H-homopolymers accounted for about 18% of the total in the derepressed Hwt-tTA clone (Fig. 6C, lane 8) and Ͻ5% in the derepressed 222-tTA clone (Fig. 6C, lane 4). In the repressed clones no ferritin iron remained in solution after immunoprecipitation, because of the absence of H-homopolymers.
Cells can be iron loaded by incubation with 55 Fe-NTA (2 Ci/ml) in the presence of 200 M ascorbate, conditions that can be used to load in vitro ferritins with ferroxidase activity (16). Thus, we performed experiments in which equal amounts of cells of the Hwt-tTA clone were incubated for 3 h with the same solution of 55 Fe-NTA ascorbate either before and after cell lysis with Nonidet P-40. Then the homogenates were run on nondenaturing gel electrophoresis, and the ferritin-bound iron was detected by autoradiography. In the Doxϩ and DoxϪ intact cells ferritin iron incorporation was very similar, whereas in the lysed cells ferritin iron incorporation was about 4-fold higher in the derepressed than in the repressed Hwt-tTA clones (Fig. 6D). Controls of untransfected parent HeLa cells gave results analogous to the repressed cell (not shown).
Ferritin Overexpression and Cellular Growth-The Hwt-tTA and H222-tTA clones could be maintained in culture in the g of extract soluble protein were immunoprecipitated with saturating amounts of anti-H-ferritin antibody, and the precipitated proteins were loaded on 12% polyacrylamide SDS-PAGE under denaturing conditions. The gels were dried and exposed to autoradiography. A, experiments on the Hwt-tTA clone in the presence or absence of doxycycline, representative of three independent experiments. B, densitometry quantification of ferritin in the indicated HeLa cells clones expressed as percentage of the initial signal.

FIG. 4. IRPs activity in the induced and noninduced cells.
Transfected H222-tTA and Hwt-tTA cells were grown for 7 and 4 days, respectively, in the presence (Doxϩ) or absence (DoxϪ) of doxycycline and then analyzed. A, samples of 2 g of total soluble protein extracts were incubated with a 32 P-labeled IRE H-ferritin probe in the absence or presence of 2% 2-mercaptoethanol (2-␤-SH), and the RNA-protein complexes were separated on nondenaturing gel electrophoresis and exposed to autoradiography. Representative data of five independent experiments are shown. B, Hwt-tTA cells growth in the presence or absence of doxycycline were grown for 18 h in medium supplemented with 100 M ferric ammonium citrate, 100 M desferrioxamine or in control medium (C), and the cell homogenates were subjected to band shift assay as in A. C, immunoblotting analysis with anti-human transferrin receptor antibody. The samples of 3 g total soluble protein extracts as in A were loaded on 12% polyacrylamide SDS-PAGE, transferred to nitrocellulose, overlaid with mouse anti-human transferrin receptor antibody (␣-TfR) and secondary horseradish peroxidase-labeled secondary, and finally developed by ECL. absence of Dox for up to 6 weeks without evident signs of toxicity. However, we observed that Hwt-tTA reached confluence more readily in the presence of Dox than in its absence, whereas the drug had no evident effects on cellular growth of the H222-tTA clone. More quantitative data were collected by seeding an equal number of cells of the different clones and analyzing them after 18 h growth under the same conditions. Cell protein content and cell viability monitored by MTT assay was not affected by the presence or the absence of Dox in the H222-tTA clone, although the values were reduced compared with the untransfected parent HeLa cells (Table II). In contrast, both protein content and total cell viability was more than 2-fold higher in the Hwt-tTA clone grown in the presence than in the absence of Dox (Table II). Analysis of cells from different induction experiments and collected at different times of induction showed an almost linear and negative correlation between the growth rate monitored by MTT assay and Hferritin content when above 2,000 ng/mg protein (Fig. 7). The observed differences in cell growth were evidently not caused by a direct effect of doxycycline, and in fact its presence or absence did not affect cell mortality detected by trypan blue exclusion (Table II). To asses whether the H-ferritin effect was related to iron availability, the cells were grown for 4 days in the presence of 100 M ferric ammonium citrate and then splitted, and cell growth was analyzed with the MTT assay. The treatment had no detectable effect on parent untransfected cells and on the repressed Hwt-tTA cells, while restoring the derepressed Hwt-tTA cell growth to the level on the controls (Table II).
Ferritin Overexpression and Oxidative Stress-To asses the resistance to oxidative stress, cells were incubated with various concentrations of H 2 O 2 , and then their viability was monitored by the MTT assay. Fig. 8A shows that H 2 O 2 toxicity to HeLa cells becomes evident at concentrations above 300 M to reach a plateau at 600 M, where cell viability decreases to 30 -40% of the untreated ones. The viability plots were analogous for the HeLa parent cells and repressed or derepressed H222-tTA clone, indicating that doxycycline had no major effect on the system. However, Hwt-tTA clone, which behaved as the other ones when repressed showed to be completely viable when exposed to the H 2 O 2 concentration of 600 M, which was highly toxic to the other cells. To asses whether the protective effect in the derepressed Hwt-tTA clone was related to iron availability, the cells were incubated for 4 days with 100 M ferric ammonium citrate and reanalyzed. The treatment did not modify the resistance to H 2 O 2 of the other clones, although it fully reverted the protective effect observed in the derepressed Hwt-tTA clone, which became as sensitive as the repressed one to 600 M H 2 O 2 (Fig. 8B). DISCUSSION We generated two transfected HeLa clones that express exogenous ferritin under the inducible tetracycline responsive promoter and accumulate analogous levels of ferritin, up to 14 -16-fold over background, under the same conditions. The two ferritins differ for two amino acid substitutions which are essential for ferroxidase activity, the H-chain wild type induces a fast iron oxidation in vitro, whereas the mutant 222 does not (11,16). The proteins are antigenically identical, and in vitro they equally co assemble with the L-chain (9). The two clones seemed adequate to study the cellular effects of high levels H-ferritin accumulation (Ͼ5-fold higher than the ones obtained by iron induction) and to analyze how these effects are related to ferroxidase activity.
The two clones showed a number of similarities; in the presence of doxycycline the exogenous ferritin expression was low or undetectable, and the cells behaved similarly to the untransfected parent HeLa cells in all the analyses we performed. After withdrawal of the drug the ferritin synthesis in both clones was accompanied by a similar shift of the H:L ratio in the heteropolymeric fraction of the isoferritins and by the accumulation of an amount of H-homopolymers comparable with that of the Fe-transferrin, and then cells were washed, lysed on the plates, and analyzed. A, radioactivity of total soluble homogenates expressed as pmol of radioactive iron/mg of soluble protein, an index of cellular iron uptake. Means and S.D. of three independent experiments. B, samples of soluble cell homogenates containing 3 g of protein were loaded on nondenaturing PAGE and exposed to autoradiography. Ferritin mobility is indicated by the arrow. C, samples of 250 g of the soluble protein homogenates were immunoprecipitated with saturating amount of anti-L-ferritin antibody to remove the ferritin heteropolymer isoferritin population. The resulting soluble fractions (10 g/well, lanes 2, 4, 6, and 8) were loaded on nondenaturing PAGE and run in parallel with the corresponding nonimmunoprecipitated samples (10 g/well, lanes 1, 3,  5, and 7) that contain all isoferritin populations. The gels were exposed to autoradiography and evaluated by densitometry. D, cells (2 ϫ 10 5 ) of the repressed or derepressed Hwt-tTA clone were incubated with 2 Ci/ml of 55 Fe-NTA, 200 M ascorbate for 3 h in serum free DMEM (Ϫ lysis) or 0.5% Nonidet P-40, 20 mM Tris-HCl, pH 8.0, 200 mM LiCl, 1 mM EDTA (ϩ lysis). Equal amounts of soluble homogenates (3 g) were loaded on nondenaturing PAGE and then exposed to autoradiography. heteropolymers. In various experiments with both clones in which total H-ferritin concentration ranged between 2,000 and 5,000 ng/mg, we found that the amount of H-homopolymers varied, whereas the L-to H-ratio in the heteropolymers was consistently ϳ20 -22:1, i.e. about one L-chain/molecule. The shift in the isoferritin compositions caused by H-chain overexpression was indirectly confirmed by the lower apparent accumulation of the L-ferritin type detected by the ELISA assay, which is sensitive to the quaternary structure of the protein (26). The data demonstrate that in the transfected HeLa clones the exogenous ferritin subunits co-assemble with the endogenous ones, at variance with the transient expression system in COS cells where the exogenous and endogenous ferritins did not coassemble and provided little indication on the effects of ferritin overexpression (20). The slower accumulation of the exogenous ferritins in the HeLa clones (Fig. 1) compared with the faster one in the transient transfection system (up to 1000fold over background in 48 h) may allow the pools of endogenous and exogenous ferritin subunits to equilibrate. This result and the finding that exogenous and endogenous ferritins degrade with the same kinetics indicated that HeLa cells do not distinguish the exogenous subunits from the endogenous ones and that the system is adequate to analyze ferritin functionality in vivo.
Ferritin and Cellular Iron Metabolism-The derepressed Hwt-tTA clone showed increased IRPs activity (about 5-fold), increased transferrin receptor (ϳ2.5-fold), and increased uptake of iron-transferrin (ϳ1.8-fold). This is fully consistent with the analysis of MEL transfected cells, where a lower up-regulation (3-5-fold over background) of ferritin H-chain induced a similar iron-deficient phenotype with increased IRP activity (21). The finding that in the transfected HeLa clone the IRPs activity was modulated by iron supplementation with ferric ammonium citrate and iron chelation with desferrioxamine indicated that the IRE-IRP system is not directly affected by the ferritin overexpression but rather responds to a deprivation of the cellular iron availability. This was confirmed by experiments to measure the LIP using the calcein method; in the H derepressed clone the LIP was about 50% smaller than in the repressed ones (Fig. 5) in agreement with results on transfected MEL cells (23).
The data confirm that the size of LIP is regulated by the amount of H-ferritin accessible to it probably for a chemical equilibrium between labile iron and the ferritin (22,23), as predicted by models of IRE-IRPs machinery where IRPs sense the iron in the LIP and direct ferritin synthesis to sequester the amount in excess (2,3). We add the demonstration that the functionality of the H-ferritin in the HeLa cells is totally dependent on its ferroxidase activity, because the overexpression of the mutant 222 chain in H222-tTA clone did not modify IRP activity, transferrin receptor accumulation, iron-transferrin uptake, or the size of the LIP. Its effect on cellular iron metabolism was negligible, as that found for the ferritin L-chains in cell lines forms hereditary hyperferritinemia cataract syndrome subjects (19).
A question we addressed was whether the human H-chain homopolymers are competent to incorporate iron when inside the cell. The recombinant H-ferritin can take up iron in prokaryotes, because it purifies from E. coli with 200 -300 iron atoms per molecule (8); however, cellular models to verify whether the same occur in mammalian cells, which have different regulations of iron, are missing, because H-chain homopolymers are not found in nature, and the transient transfected COS cells did not demonstrate to be informative (20). In 55 Fe-labeled derepressed Hwt-tTA cells ferritin heteropolymers were subtracted with anti-L-chain antibody precipitation and the resulting soluble H-homopolymers were found to contain only 18% of the total ferritin iron, although they accounted for about 50% of total ferritin protein (Fig. 6C). This indicates that the heteropolymers were about 5-fold more efficient than the H-homopolymers in incorporating iron. More intriguing were the results with the H222-tTA cells where the overex-TABLE II Ferritin overexpression and cell growth Cells of the indicated clones were plated at 2 ϫ 10 5 for trypan blue and protein assays or 10 4 for MTT assay. After 18 h of growth cell mortality was monitored by trypan blue exclusion method, the protein content were quantified by BCA assay, and total cell viability was determined by MTT assay also on cells grown for 4 -7 days in 100 M ferric ammonium citrate (ϩFAC). Means and S.D. of at least three independent experiments. ND, not determined.  pressed 222 subunit can associate with the endogenous H-or L-chains. The soluble ferritin fraction after immunoprecipitation was composed by heteropolymers of H-and 222 mutant and contained Ͻ5% of total ferritin iron (Fig. 6C). This indicates that the 222/H heteropolymers are much less efficient than H-homopolymers in iron incorporation. The data are in good agreement with the in vitro results showing that the L-chain contributes substantially to the efficiency of ferritin iron incorporation by promoting iron nucleation mineralization, whereas the 222 mutant subunit is the least efficient, because of the lack of both ferroxidase and iron mineralization capacity (16,31).
Another question we tried to address is whether in the cells the rate of ferritin iron incorporation is limited by the rate of cellular iron transfer through cell membrane under conditions of iron loading. Normally, Fe(II) salts are used to study ferritin iron uptake in vitro and the more stable Fe(III) chelates are used for cellular iron uptake; thus the rates of ferritin and cellular iron incorporations cannot be compared. We reasoned that 55 Fe-NTA, in presence of ascorbate, is taken up by cells and is also specifically incorporated into ferritins in vitro (16), and thus an analysis of ferritin iron incorporation in the absence of the presence of intact cell membranes may provide an indication on the limiting step of the transfer of iron from the medium to the ferritin. The finding that in the repressed Hwt-tTA cells the amount of iron associated to ferritin was the same in the lysed and intact cells suggested that the cell membrane transport was not the limiting step; however, in the derepressed cells, ferritin iron incorporation was 4-fold higher in the absence than in the presence of the membrane, indicating that the ferroxidase activity exceeded cell capacity to transfer iron. The finding may help to explain why the excess ferritin in the derepressed cells induces iron deficiency.
Ferritin, Cell Proliferation, and Oxidative Damage-The most evident phenotype associated to H-ferritin overexpression was a reduced growth rate; after withdrawal of the doxycycline the Hwt-tTA clone started growing slowly without evident modification of cell morphology or sign of toxicity. All the parameters we analyzed confirmed and extended the observation to the point that we noticed a correlation between cell Hferritin content and cell growth (Fig. 7). This has not been described before, likely because in transfected MEL cells the level of H-ferritin accumulation was lower than that obtained in HeLa cells (21), and the experimental limits of the transient transfections did not allow to study cell growth (20). The finding explains why in various experiments we could not obtain stable cell lines that constitutively expressed high level of H-ferritin (not shown). The derepression of H222-tTA clone did not modify cell growth, and the suppressive effect of H-ferritin in the derepressed Hwt-tTA clone was totally reverted by prolonged preincubation with iron (Table II) but not by short 2-h incubations with iron (not shown). Iron further up-regulates the expression of the endogenous ferritins and only when supplied long enough to replenish the labile iron pool it abolished H-ferritin suppressive effect. The data demonstrate that the negative effect of H-ferritin on cell growth in HeLa cells was mediated by the ferroxidase activity and is secondary to the iron deficiency induced by it. Thus, ferritin acts as an iron chelator, possibly by reducing iron available to ribonucleotide reductase (32).
We found also that overexpression of the H-chain significantly increased cell resistance to H 2 O 2 ; for instance only 30 -40% of the repressed Hwt-tTA cells remained viable after incubation with 600 M H 2 O 2 compared with 100% of the same derepressed cells (Fig. 8). Similar, but not as strong, antioxidant effect was observed also in the transfected MEL cells (33). We add that the effect was absent in the derepressed H222-tTA clone and that it was reverted by prolonged preincubation with iron. This seems to parallel the suppressive effect, being mediated by the ferroxidase activity and its iron withholding in the cell.
In conclusion the data demonstrate that the activity of the H-ferritin overexpressed in HeLa cells is similar to that observed in vitro: (i) it sequesters iron with a mechanism that necessitates the integrity of the ferroxidase center, (ii) one or few H-chains/molecule are sufficient to confer the capacity to sequester iron, (iii) the activity is strongly enhanced by the co-presence of the L-chain in the same molecule, and (iv) Hhomopolymers are functional, although less so than the H/L heteropolymers. In addition they show that all the phenotypes induced by H-ferritin overexpression are iron-mediated implying that in HeLa cells the only evident activity of the protein is to modulate iron availability. The results do not support the hypothesis that the association with ceruloplasmin is necessary for H-ferritin functionality (34). The data point also to an intriguing role of H-ferritin; if it leaves too little iron available for cellular needs, then proliferation is reduced and at the same the resistance to oxidative stress is increased. This fits well with present understanding of iron function and toxicity and with the hypothesis that iron homeostasis needs to be tightly regulated; it is also in agreement with the observation that proliferating cells have little ferritin and much transferrin receptor, whereas resting cells up-regulate ferritin to be more protected (35). The availability of two subunit types that cooperate in iron incorporation may be particularly useful to expand cell iron storage compartment (i.e. number of ferritin cavities) without affecting the capacity of iron sequestration (linked to the number of ferroxidase centers) (19). FIG. 8. Effect of H 2 O 2 . The cells were grown to obtain maximal ferritin accumulation in the absence (A) or in the presence (B) of 100 M ferric ammonium citrate. Clone Hwt-tTA was grown for 4 days, and clone H222-tTA was grown for 7 days. The cells were plated, incubated for 2 h with the indicated concentration of H 2 O 2 in serum-free medium, and washed, and then cellular vitality was measured by MTT assay. Values were plotted as percentages of the absorbance of the untreated control cells. Data are the means the Ϯ S.D. of three experiments in octaplicate.