Iron-mediated oxidative stress plays an essential role in ferritin-induced cell death

Free Radical Biology & Medicine 48 (2010) 1347–1357

Contents lists available at ScienceDirect

Free Radical Biology & Medicine j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / f r e e r a d b i o m e d

Original Contribution

Iron-mediated oxidative stress plays an essential role in ferritin-induced cell death Nikolaus Bresgen a,⁎, Heidi Jaksch a, Heide Lacher a, Ingo Ohlenschläger a, Koji Uchida b, Peter M. Eckl a a b

Department of Cell Biology, University of Salzburg, A-5020 Salzburg, Austria Graduate School of Bioagricultural Sciences, Nagoya University, Nagoya 464-8601, Japan

a r t i c l e

i n f o

Article history: Received 23 December 2009 Revised 9 February 2010 Accepted 12 February 2010 Available online 19 February 2010 Keywords: Ferritin Apoptosis Oxidative stress Iron Lipid peroxidation Hydroxynonenal Free radicals

a b s t r a c t Previously, we have demonstrated an apoptosis-inducing activity of an acidic, H-chain-rich isoferritin secreted from primary rat hepatocytes in vitro. Because this proapoptotic property may be responsible for the growth-inhibitory and immunosuppressive effects described for certain ferritin species, we aimed to address the mechanism by which ferritin can trigger cell death. Suggesting a pivotal role for iron, iron chelation by desferrioxamine significantly abrogates ferritin-mediated apoptosis and necrosis in primary rat hepatocytes and substantially lowers the extent of protein modification by 4-hydroxynonenal (HNE)—a major lipid peroxidation (LPO) product. Furthermore, supplementing the cultures with the radicalscavenging compound trolox also provided significant protection from ferritin-mediated apoptosis. Moreover, a significant increase in micronucleated cells upon exposure to ferritin indicates that ferritin also introduces damage to DNA. Based on these observations we therefore propose that endocytosis of extracellular ferritin increases the level of free ferrous iron in the lysosomal compartment, promoting Fenton chemistry-based oxidative stress involving LPO and increased lysosomal membrane permeability. Subsequently, the release of reactive lysosomal content leads to cellular damage, in particular modification of protein and DNA induced by HNE and other reactive aldehydic LPO products. Together, these effects will trigger apoptosis and necrosis based on the upregulation of p53, increased mitochondrial membrane permeability, and proapoptotic Fas signaling as described recently. In conclusion, based on their iron-storing ability, secreted acidic isoferritins may act as soluble mediators of oxidative stress under certain physiological and pathophysiological conditions. © 2010 Elsevier Inc. All rights reserved.

Introduction Although iron is essential to proper cell function, an excess of free Fe2+ can cause the generation of reactive oxygen species (ROS) based on the Fenton reaction, which will provoke oxidative cell damage. Hence, the cellular iron pool requires stringent control accomplished by the tight regulation of iron uptake via the transferrin/transferrinreceptor system and buffering of the intracellular iron pool by ferritin. Ferritin, an approximately 450-kDa protein composed of 24 heavy (Hchain) and light (L-chain) subunits, shows a remarkably high iron storage capacity of up to 4500 Fe3+ ions/molecule (on average 2000 Fe3+ ions/molecule can be found in biological samples). The ratio of

Abbreviations: CM, conditioned medium; DAPI, 4′,6-diamidino-2-phenylindole; DFO, desferrioxamine mesylate; EGF, epidermal growth factor; FER-CM, ferritin purified from CM collected after 3 h of primary rat hepatocyte culture; FER-T, commercially available ferritin preparation from rat liver; HNE-HisP, protein modification by HNE adducts to histidine residues; HNE, 4-hydroxy-2-nonenal; I, insulin; LMP, lysosomal membrane permeabilization; LPO, lipid peroxidation; MAPK, mitogen-activated protein kinase; MEM, minimum essential medium; MOMP, mitochondrial outer membrane permeabilization; PLF, placental isoferritin; ROS, reactive oxygen species. ⁎ Corresponding author. Fax: +011 43 662 8044 144. E-mail address: [email protected] (N. Bresgen). 0891-5849/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.freeradbiomed.2010.02.019

heavy- to light-chain subunits varies among different tissues, and with respect to their pI, either basic (L-rich) or acidic (H-rich) isoforms can be distinguished [1]. The ferritin H-chain harbors a ferroxidase center, which oxidizes Fe2+ upon entrance into the molecule, the Fe3+ being subsequently incorporated into a ferrihydrite (5Fe2O3·9H2O) mineral core, a process to which the L-chain activity is critical, especially at increased cellular iron levels [1,2]. Ferritin synthesis is regulated by cellular iron levels, independent of both subunits, mainly at the translational level and is modulated by oxygen supply [3]. Notably, ferritin has been shown to serve as a cellular antioxidant, the H-chain being central to protection from iron-mediated lipid peroxidation and oxidative stress as well as inhibiting apoptosis [4–6]. Importantly, ferritin can also be released from the cell based on the ER/Golgi-dependent secretory pathway, a rapid and highly specific process involving N-glycosylation [7]. Under normal conditions, only very small amounts of ferritin are released; however, ferritin levels in the blood are increased upon dietary iron supplementation and during pregnancy, in which an acidic placental isoferritin (PLF) participates in the downregulation of the maternal immune response to fetal antigens [8]. Furthermore, the content of ferritin in serum is also elevated in a number of pathological conditions, including acute and chronic inflammation as well as autoimmune diseases [9–12].

1348

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357

Isoferritins have also been recognized as acute-phase reactants and may also play a yet to be defined role in tissue repair and regeneration [13,14]. Of particular relevance is the finding that various types of cancer are accompanied by hyperferritinemia [15–17], and the release of H-chain-rich, acidic ferritin isoforms from cultured tumor cell has also been shown [18,19]. Although secretion and uptake of H-chain-rich isoferritins may serve to supply iron under certain circumstances [20], this “soluble” ferritin species also exhibits growth-adverse, immunomodulatory effects [8,19,21,22], which contrasts with the role of intracellular ferritin as guardian of the cellular iron pool. Furthermore, we have reported previously that primary hepatocytes secrete H-chain-rich acidic isoferritins with homology to PLF and melanoma-derived Hchain-rich isoferritins, which stimulate apoptosis by addressing proapoptotic Fas (CD95) and intrinsic mitochondrial signaling [14,23]. The mechanisms underlying the growth-adverse effects of secreted ferritins are still poorly defined, in particular the involvement of ferritin-stored iron is not elaborated. Interestingly, secreted ferritins can stimulate cytokine–chemokine interactions or act as proinflammatory cytokines by an iron-independent, receptorbased interference with MAPK signaling, for which uptake of the protein apparently is not essential [19,24–26]. On the other hand, the receptor-mediated endocytosis of recombinant ferritins and translocation to endo-/lysosomes have been demonstrated, which also interfere with cellular growth control [27–30]. In the latter respect translocation of extracellular iron-loaded ferritin to the lysosomal compartment could resemble a critical event, because an increase in the lysosomal pool of redox-active iron can trigger lysosomal membrane permeabilization (LMP), especially under prooxidant conditions [31–35]. It is well documented that LMP or lysosomal stress may become detrimental to the cell because of the release of the lysosomal content (e.g., Fe2+, ROS, cathepsins) leading to protein modification, DNA damage, and also the direct activation of mitochondrial outer membrane permeabilization (MOMP), which subsequently will trigger the onset of cell death [31,36,37]. The question therefore arises as to whether and to what extent iron, ROS, and lipid peroxidation are involved in cell death mediated by secreted acidic isoferritin. To address these questions, the effects of the well-characterized lysosomotropic iron chelator desferrioxamine mesylate (DFO) as well as the radical scavenger trolox (6-hydroxy-2,5,7,8-tetramethylchromane-2-carboxylic acid) on ferritin-mediated apoptosis and necrosis were investigated in primary hepatocytes. Furthermore, the role of lipid peroxidation (LPO) was determined by immunocytochemical in situ analysis using a monoclonal antibody that specifically recognizes 4hydroxy-2-nonenal (HNE)–histidine protein adduct formation [38]. In addition, the response of hepatocytes to secreted acidic isoferritins as well as iron-loaded ferritin prepared from rat liver was also investigated upon proliferative stimulation of parenchymal hepatocytes by the combination of epidermal growth factor (EGF) and insulin, addressing ferritin-mediated cytotoxicity and potential genotoxic effects. Materials and methods Materials Minimum essential medium (MEM) with Earle's salts and nonessential amino acids was obtained from Invitrogen (Vienna, Austria). Penicillin and streptomycin were obtained from MedPro (Vienna, Austria). Collagenase and other cell culture chemicals— unless otherwise specified—were obtained from Sigma Chemical Co. through Biotrade (Vienna, Austria). Percoll was obtained from GE Healthcare (Vienna, Austria). Plasticware was obtained from Greiner, Sarstedt, and Falcon/Becton–Dickinson (Vienna, Austria).

Animals Adult female Fischer 344 rats weighing approximately 100 g were obtained from Harlan (Germany). The animals were kept in a temperature- and humidity-controlled room with a 12-h light/dark cycle. Water and food were provided ad libitum. The animals were allowed to acclimate for at least 2 weeks before hepatocyte isolation. Culture medium (MEM) Serum-free MEM (Cat. No. 41500-018), supplemented with nonessential amino acids, pyruvate (10 mM), aspartate (0.2 mM), serine (0.2 mM), penicillin (100 U), and streptomycin (100 μg/ml) was used as the culture medium. Cell isolation and primary culture Parenchymal hepatocytes were isolated by the in situ two-step collagenase liver perfusion technique as described by Michalopoulos et al. [39]. After the freshly isolated cells were washed twice in calcium-free buffer (142 mM NaCl, 6.7 mM KCl, 10 mM Hepes, pH 7.4) cell numbers and viability (trypan blue exclusion assay) were determined, yielding ≥90% viable cells. From the crude cell pellet, dead cells and remaining nonparenchymal cells were removed by Percoll density gradient centrifugation as described previously [40] and the purified parenchymal cells were plated according to the protocols given below. Purification of ferritin from the conditioned medium (CM) Viable cells (5 × 106) were plated on 90-mm-diameter collagencoated petri dishes containing 10 ml of serum-free MEM and were incubated for 3 h under standard culture conditions (37°C, 5% CO2, 95% relative humidity). Thereafter, the culture supernatant (CM) was collected, detached cells and cell debris were removed by centrifugation (250 g for 10 min), and the conditioned medium (native CM3) was stored at 4°C. Purification of ferritin from the native CM3 was performed as described previously [23,40], yielding a fraction enriched in H-chain-rich, acidic isoferritins termed FER-CM (see also Fig. 1A). In addition, a commercially available preparation of ironloaded ferritin purified from rat liver (Cat. No. 4420-5539; Biotrend, Cologne, Germany) was also investigated, termed FER-T (tissue ferritin). For electrophoretic analysis, concentrated samples were separated by one-dimensional SDS–PAGE under reducing conditions on a 12% gel (Mini-Protean III; Bio-Rad, Hercules, CA, USA), and protein bands were visualized by silver staining. Molecular weights were determined by comparison with molecular weight markers (Bio-Rad). Protein concentrations were determined using the Bio-Rad protein assay. Protein identification Characterization of proteins separated by SDS–PAGE was performed by peptide mass fingerprinting and tandem mass spectrometry as described previously [23]. The peptide mass fingerprint and the tandem mass spectra were submitted to a search of the NCBI protein NR database using MASCOT (www. matrixscience.com). Treatment protocols After Percoll enrichment, 1 × 105 viable cells were plated on 13mm-diameter collagen-coated plastic coverslips in 24-well plates containing 1 ml MEM per well and were cultured for 1 h under standard conditions to allow cell attachment. Thereafter, the

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357

1349

Fig. 1. Influence of radical scavenging on ferritin-mediated cytotoxicity. (A) Primary hepatocytes were treated with 100 ng/ml ferritin (FER-CM, FER-T) alone (w/o Trolox) or in the presence of 100 mM trolox. Bars represent the means ± SD of the percentages of apoptotic cells determined after 24 (open bars) and 48 h (black bars) of treatment. (B) Cell densities determined after 48 h in control cultures (open bars) and cultures treated with 100 ng/ml FER-T (gray bars) or 100 ng/ml FER-CM (black bars) supplemented with 100 mM trolox or left unsupplemented. ⁎P b 0.05, ⁎⁎P b 0.005 compared to the control; +P b 0.05, ++P b 0.005 compared to treatment w/o trolox; N ≥ 4. The inset in (A) shows a representative sample of the applied ferritin preparations analyzed by SDS–PAGE and silver staining. Arrows point at ferritin subunits with a molecular mass of 22 kDa (H-chain monomer) and 43 kDa (assumed to represent an H–L chain heterodimer—see text). The three main contaminants in FER-T (arrowheads) have been identified as δ-aminoleavulinic acid dehydratase (NCBI ID P06214), glutathione reductase (NCBI ID P70619), and transferrin (NCBI ID P12346) (from below).

medium was exchanged for (i) fresh MEM, (ii) MEM + supplements (see below), or (iii) MEM containing 100 ng/ml FER-CM or 100 ng/ml FER-T with or without supplementation (see below), and the plates were returned to the incubator. Unless otherwise specified, no further medium change was carried out until fixation. Cells were fixed after 24, 48, and 72 (only in the case of proliferative stimulation) hours of culture. The following treatment protocols were performed: (i) 100 μM trolox added 1 h after plating in combination with FER-CM or FER-T; (ii) 100 μg/ml DFO, either applied for 1 h before exposure of the cultures to FER-CM without DFO supplementation (DFO-pre) or added in combination with FERCM (DFO-combined) 1 h after plating in nonsupplemented MEM; and (iii) proliferative stimulation by EGF (40 ng/ml) and insulin (0.1 μM) in combination with FER-CM or FER-T starting either 1 h after plating or according to the experimental protocols outlined in Fig. 5. Fixation, staining, and microscopical analysis Cyto- and genotoxic effects were determined by fluorescence microscopy, allowing a single-cell-based analysis of all relevant endpoints in parallel in the same culture (i.e., apoptosis, necrosis, mitosis, and micronucleus formation as well as formation of HNEmodified proteins). The morphology of DAPI-stained nuclei was used to identify (i) viable cells (regularly sized nuclei showing neither chromatin condensation nor fragmentation); (ii) apoptotic cells (the DAPI stain representing a valid method for the detection of advanced apoptotic stages [41] marked by crescent-like, condensed, or fragmented chromatin [42]); (iii) cells that have undergone necrosis as indicated by small, pyknotic nuclei revealing highly condensed chromatin [42]; (iv) mitotic cells, with chromosomes arranged in mitotic figures; and (v) cells showing one or more micronuclei in the vicinity of the nucleus.

Fixation and DAPI staining were performed as described previously [40]. For microscopical analysis the coverslips were mounted in glycerol and analyzed under the fluorescence microscope (Leitz Aristoplan) using epifluorescence combined with phase-contrast optics. Cultures were analyzed by scoring the coverslip for the numbers of adherent viable, apoptotic, necrotic, mitotic, and micronucleated cells. At least 1000 cells were analyzed for each time point and treatment group per experiment. Mean cell densities were determined by dividing the total cell number by the number of optical fields. Immunocytochemistry After being washed with calcium-free buffer (142 mM NaCl, 6.7 mM KCl, 10 mM Hepes, pH 7.4), the cells were fixed with methanol. Immunocytochemical detection of HNE-modified proteins was performed according to standard methods. The monoclonal mouse antibody mAbRS17 specific for the 4-hydroxy-2-nonenal– histidine adduct [43] was used as primary antibody, and binding of the primary antibody was visualized by an Alexa 555-conjugated secondary antibody (Sigma). Cell nuclei were counterstained by DAPI. Analysis was performed using a Leitz Aristoplan fluorescence microscope, and densitometric analysis of the micrographs was performed using the Zeiss LSM510 META confocal laser microscope software package. Statistical analysis Statistical analysis was performed by applying the SPSS statistical software package. Normal distribution of the results was checked by use of the one-sample Kolmogorov–Smirnov test. Student's twotailed t test for independent samples was applied to calculate the levels of significance for changes in the percentages of adherent

1350

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357

Table 1 Effect of iron chelation by desferrioxamine (DFO) on ferritin-mediated cytotoxicity in primary rat hepatocytes Apoptotic cells (%)

MEM MEM + DFO FER-CM FER-CM + DFO

Cell densitya

Necrotic cells (%)

24 h

48 h

24 h

48 h

48 h

0.70 ± 0.55 0.14 ± 0.06‡ 2.58 ± 1.29⁎ 0.46 ± 0.35†

1.75 ± 0.73 1.04 ± 0.80 9.52 ± 2.34⁎⁎ 2.31 ± 0.44††

3.63 ± 1.59 1.30 ± 1.55‡ 10.55 ± 2.38⁎⁎ 3.49 ± 2.22†

3.71 ± 2.56 0.99 ± 0.64‡ 19.50 ± 11.21⁎ 3.69 ± 2.41†

1.28 ± 0.39 1.49 ± 0.63 0.74 ± 0.32⁎ 1.49 ± 0.38†

a Cell density is expressed as cells × 104/cm2. Ferritin (FER-CM) was applied at a concentration of 100 ng/ml. DFO was used at a concentration of 100 μg/ml. Data represent the means ± SD determined after 24 and 48 h of treatment. Student's two-tailed t test for independent samples; N ≥ 3. ‡ P ≤ 0.08 compared to MEM. ⁎ P ≤ 0.05. ⁎⁎ P ≤ 0.005 compared to MEM (control). † P ≤ 0.05. †† P ≤ 0.005 compared to cultures treated with FER-CM without DFO supplementation.

viable, mitotic, apoptotic, necrotic, and micronucleated cells as well as changes in cell density. The one-sample t test was used to examine the significance of effects expressed as fold increase/decrease compared to the corresponding controls. Results Oxidative stress in ferritin-mediated cell death To examine the role of oxidative stress in ferritin-mediated cell death, experiments employing the radical-scavenging vitamin E derivative trolox were performed. As demonstrated in Fig. 1A, trolox applied at a concentration of 100 μmol/L significantly (P ≤ 0.05; P ≤ 0.005) protected against ferritin-mediated apoptosis stimulated by 100 ng/ml FER-T or FER-CM. Although Fig. 1A suggests a moderately stronger effect for FER-CM, both ferritin species significantly (P ≤ 0.005) elevated the percentage of apoptotic cells in a time-dependent manner. Trolox inhibited this effect by approximately 50%, the apoptotic events

becoming significantly (P ≤ 0.05) reduced to 0.52 ± 0.24-fold for FER-T and to 0.52± 0.28-fold for FER-CM after 48 h in the presence of the radical scavenger (N = 5). Both ferritin species also stimulated necrotic cell death, the observed differences among FER-T and FER-CM not being significant: 18.87 ± 12.29% of necrotic cells after 48 h of exposure to FER-CM compared to 12.24 ± 2.93% in FER-T-treated cultures. At the same time point 7.49 ± 5.24% necrotic cells were observed in control cultures (N ≥ 4). Similar to its effect on apoptosis induction, 100 μM trolox significantly (P ≤ 0.05) inhibited necrosis in cultures treated with FERT, decreasing the percentage of necrotic cells to 0.54 ± 0.24-fold after 48 h of incubation (N = 5). In contrast, FER-CM-stimulated necrosis was less affected by trolox, the levels observed after 48 h of treatment moderately decreasing to 0.86 ± 0.32-fold (N = 5). In addition, supplementation with 100 μM trolox also restored increased cell densities in ferritin-treated cultures, an effect that was moderate in FER-T-treated cultures, but became significant (P ≤ 0.005) when FERCM was applied (Fig. 1B).

Fig. 2. Effects of desferrioxamine (DFO) on ferritin-mediated apoptosis. (A) Primary hepatocytes were (i) treated with 100 ng/ml ferritin (FER-CM) alone (w/o DFO), (ii) preincubated for 1 h with 100 μg/ml DFO (DFO pre) followed by ferritin treatment in the absence of DFO, or (iii) treated with a combination of ferritin and DFO (DFO comb.). Bars represent the means ± SD of the percentages of apoptotic cells determined after 48 h of treatment. (B, C) Differences between DFO preincubation and combined FER-CM/DFO treatment expressed as fold decrease (compared to ferritin treatment without DFO) in the percentage of apoptotic cells and cell density determined after 48 h of culture. ⁎P b 0.05, ⁎⁎P b 0.005 compared to the control and DFO pretreatment; +P b 0.05, ++P b 0.005 compared to treatment with ferritin alone (w/o DFO) (N = 4).

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357

With respect to purity, electrophoretic analysis of FER-T and FER-CM revealed marked differences among both preparations (Fig. 1A inset). Because the protein concentrations applied to the cultures refer to the total protein content of the preparation (i.e., 100 ng of total protein/ml), the cells will have been exposed to higher amounts of FER-CM compared to the less pure FER-T. In addition, both ferritin preparations share a ferritin band at 22 kDa (identified as the H- and L-chain monomer); however, a further prominent band at 43 kDa, most likely representing a ferritin H/L-chain dimer [23], is detectable only in FER-CM (Fig. 1A). Notably, a 43-kDa “superheavy” subunit is also found in H-chain-rich PLF and melanoma-derived H-chain-rich ferritin [19,22,44,45], which also share homology with the hepatocyte-derived isoferritin [23]. Hence, FER-CM eventually represents a ferritin with a higher content of the H-chain compared to FER-T, which may also contribute to the observed differences between both ferritin preparations. Effect of iron chelation on ferritin-mediated cytotoxicity In a further approach we investigated the role of iron in ferritinmediated cell death. Confirming a profound impact of ferritin-derived iron on cell integrity, iron chelation by DFO strongly inhibited the cytotoxic effects of FER-CM as demonstrated in Table 1 and Figs. 2 and 3. Apoptosis as well as the onset of necrotic cell death were significantly (P ≤ 0.05; P ≤ 0.005) inhibited upon DFO supplementation of FER-CM-treated cultures, lowering the percentages of apoptotic and necrotic cells to the control level (Table 1). In turn,

1351

cell density significantly (P ≤ 0.05) increased upon DFO supplementation in cultures exposed to FER-CM (Table 1). Moreover, DFO also reduced the low basal levels of apoptotic and necrotic cells found in control cultures (Table 1), although this was not significant (P ≤ 0.08). In contrast, “preloading” of the cells with the iron chelator by preincubation with DFO restricted to 1 h before treatment with 100 ng/ml FER-CM in the absence of DFO failed to protect the cells from ferritin-mediated cytotoxicity, the percentages of apoptotic and necrotic cells found after 48 h of culture remaining significantly (P ≤ 0.05; P ≤ 0.005) elevated above the control and cultures treated with a combination of FER-CM and DFO (Fig. 2). Ferritin-associated lipid peroxidation HNE is one of the major products of lipid peroxidation and causes the formation of protein as well as DNA adducts [46–48]. To examine the role of ferritin-associated LPO, the extent of HNE– histidine adduct formation was analyzed immunocytochemically. The results summarized in Fig. 3 indicate a remarkable increase in HNE-dependent protein modification (HNE–HisP) upon exposure to FER-CM (Fig. 3). After 24 h of exposure to FER-CM, a majority of the cells showed a remarkable accumulation of HNE–HisP, which was strongest in the perinuclear area (Figs. 3B and E). As shown in Fig. 3H, the specific staining pattern for HNE–HisP adjacent to the nuclei suggests the formation of aggregates of HNE-modified proteins in the vicinity of the nuclear envelope, resulting in a granular, “ringlike” appearance of HNE–HisP. Because immunoreactivity

Fig. 3. Ferritin-associated lipid peroxidation. Primary hepatocytes were treated for 24 h with (B, E, I) 100 ng/ml FER-CM, (C, F) a combination of 100 ng/ml FER-CM and 100 μg/ml DFO, (G, H) a combination of 100 ng/ml FER-CM with EGF (40 ng/ml) and insulin (100 nM), or (A, D—control) left untreated. (A–C and G–I) Immunoreactivity to a monoclonal antibody recognizing proteins modified by HNE (HNE–histidine adducts) employing Alexa 555 as secondary antibody. Note the enhanced immunoreactivity in the nuclear periphery (arrows). (D–F) The corresponding densitograms and the color coding relative to grayscale intensity. (G) Ferritin-mediated HNE protein modification is detectable only in the vicinity of the nucleus upon proliferative stimulation by EGF + insulin. (H) The perinuclear immunostaining pattern in a binucleated cell in detail, revealing inhomogeneous distribution suggesting an accumulation of HNE-modified proteins (left, optical nuclear cross section; right, optical focus on nuclear surface). (I) Different immunoreactivities for the anti-HNE–histidine antibody in a culture treated with FER-CM alone. Note that the cells on the left show reduced staining for HNE–histidine and improved cell spreading, which also holds true for (C) and (G).

1352

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357

Table 2 Influence of proliferative stimulation on ferritin-mediated cytotoxicity in primary hepatocytes Apoptotic cells (%)

Necrotic cells (%)

MI (%)a

AI (%)/MI (%)

Without EGF + insulin Control FER-T FER-CM

1.82 ± 0.49 (5) 5.53 ± 1.13 (5)⁎⁎ 8.08 ± 1.47 (5)⁎⁎

4.97 ± 3.09 (4) 12.24 ± 2.93 (4)⁎ 12.69 ± 4.62 (4)⁎

n.d. n.d. n.d.

n.a. n.a. n.a.

With EGF + insulin Control FER-T FER-CM

0.27 ± 0.22 (5)†† 0.34 ± 0.25 (4)†† 0.55 ± 0.13 (3)††

4.77 ± 3.70 (4) 7.13 ± 0.98 (3)† 5.54 ± 1.73 (3)†

1.21 ± 0.35 (5) 1.09 ± 0.59 (4) 0.65 ± 0.15 (3)⁎

0.22 ± 0.15 (5) 0.33 ± 0.25 (3) 0.91 ± 0.47 (3)

a MI, mitotic index (i.e., percentage of mitotic cells); n.d., not detectable (no proliferative stimulation); n.a., not applicable (absence of mitotic cells). Data represent the means ± SD determined after 48 h of incubation in MEM (control) or MEM supplemented with 100 ng/ml ferritin in the presence or absence of EGF (40 ng/ml) and insulin (100 nM). Student's two-tailed t test for independent samples. ⁎ P ≤ 0.05. ⁎⁎ P ≤ 0.005 compared to the corresponding controls. † P ≤ 0.05. †† P ≤ 0.005 compared to the corresponding cultures without proliferative stimulation.

inside these “circles” is weak, modification of nuclear proteins such as histones by HNE is unlikely or may occur at a very low level. Iron chelation by DFO substantially reduced HNE–HisP accumulation in FER-CM-treated cultures (Figs. 3C and F), which points to a causal role for ferritin-derived iron in LPO and HNE–HisP formation. Interestingly, cytosolic immunoreactivity to the anti-HNE–His antibody was reduced by DFO, whereas perinuclear staining for HNE–HisP remained detectable (Fig. 3F). The same holds true for proliferating hepatocytes in cultures treated with EGF and insulin, in which ferritin-dependent HNE–HisP accumulation takes place only in the perinuclear zone (Figs. 3G and H). Furthermore, HNE–HisP immunoreactivity may markedly vary between cells in FER-CM-treated cultures as demonstrated in Fig. 3I. In this representative micrograph cells with strong cytosolic staining for HNE–HisP are neighboring others without detectable cytosolic HNE–HisP immunoreactivity. Comparing Figs. 3C and I it also becomes evident that the variability of HNE–HisP accumulation seen in FER-CM cultures is remarkably reduced by DFO. Moreover, Fig. 3I also shows that reduced HNE–HisP accumulation is correlated with improved cell

spreading in FER-CM cultures, which also holds true for the effect of DFO (Figs. 3B and C). Ferritin and hepatocyte proliferation In agreement with our previous findings [14,23], apoptosis mediated by FER-CM and FER-T was significantly (P ≤ 0.005) inhibited upon proliferative stimulation by epidermal growth factor and insulin (EGF + I), culture conditions that also significantly (P ≤ 0.05) counteracted ferritin-mediated necrotic cell death (Table 2). Remarkably, inhibition of FER-CM-mediated apoptosis by EGF + I was significantly (P ≤ 0.005) more powerful than treatment with DFO or trolox (Fig. 4). In addition, DFO and EGF + I also antagonized FER-CMinduced necrotic cell death to similar degrees, which were significantly (P ≤ 0.05) higher compared to trolox treatment. As already mentioned above, EGF + I also strongly reduced ferritin (FER-CM)associated LPO, as evidenced by the absence of cytosolic immunoreactivity to the anti-HNE–His antibody, and also preserved a wellspread hepatocyte morphology in the presence of FER-CM (Fig. 3G).

Fig. 4. Comparison of different culture supplements with respect to their ability to inhibit ferritin-mediated cell death. Primary hepatocytes were treated for 48 h with 100 ng/ml FER-CM in the presence of either 100 μM trolox or 100 μg/ml DFO or a combination of 40 ng/ml EGF and 100 nM insulin. Bars represent the means ± SD of the fold decrease in the percentages of apoptotic and necrotic cells compared to cultures treated with FER-CM alone (positive control). ⁎P b 0.05, ⁎⁎P b 0.005 compared to FER-CM; +P b 0.05, ++P b 0.005 compared to supplementation with trolox (N ≥ 4).

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357 Fig. 5. Genotoxicity of ferritin. Primary hepatocytes were either (i) pretreated with 100 ng/ml ferritin (FER-CM, FER-T) for 24 h before proliferative stimulation by a combination of 40 ng/ml EGF and 100 nM insulin under normal culture conditions (i.e., without further ferritin administration) or (ii) exposed to a combination of 100 ng/ml ferritin with EGF and insulin after 24 h of regular primary culture (i.e., no ferritin supplementation). A schematic representation of the protocols is given below the diagrams (0, plating; MC, medium change). Controls were treated by the same protocols, but ferritin was omitted. Shown are the percentages of (A) mitotic and (B) micronucleated cells determined after 48 h of proliferative stimulation (summing up to a total incubation time of 72 h). (C) The standardized values (micronucleated cells (%)/MI (%)), accounting for the dependence of micronucleus formation on mitosis. ⁎P b 0.05, ⁎⁎P b 0.005 compared to the control; +P b 0.05 compared to pretreatment with ferritin (N ≥ 4).

1353

1354

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357

Although EGF + I protected from ferritin-mediated cell death, its efficiency in stimulating proliferation was reduced in the presence of FER-CM and FER-T. This effect was most prominent when EGF and insulin were added 1 h after plating together with FER-CM or FER-T. After this treatment, the percentages of mitotic cells detectable after 48 h of proliferative stimulation were significantly (P ≤ 0.05) lowered by FER-CM, which also markedly shifted the apoptosis-to-mitosis ratio (Table 2). A comparable effect was also obtained in cultures pretreated with FER-CM for the first 24 h of primary culture, followed by proliferative stimulation for 48 h in MEM (Fig. 5A—Pretreatment). Under both treatment protocols, FER-T also reduced the mitotic indices (Table 2 and Fig. 5A); however, this reduction was not significant. Notably, inhibition of EGF + I-stimulated cell proliferation was less pronounced when EGF + I and ferritin were added to the cultures after an initial cultivation period of 24 h in regular culture medium (Fig. 5A—Combined). Genotoxic potential of ferritin To assess potential DNA damage by ferritin treatment, micronucleus induction in proliferating hepatocytes was analyzed as a measure of genotoxicity [49]. Fig. 5B shows that exposure to ferritin upon proliferative stimulation (combined treatment) causes a significant (P ≤ 0.05) increase in the percentage of micronucleated cells. Although Fig. 5B suggests that this does not hold true for pretreatment with ferritin, it should not be ignored that micronuclei are formed only in the course of mitosis. Therefore, a reduced percentage of mitotic cells will yield a lower number of micronucleated cells, even if DNA damage has occurred. Therefore, the apparently unchanged numbers of micronucleated cells, which are accompanied by a significant (P ≤ 0.05; FER-CM) reduction in the percentage of mitotic cells in cultures pretreated with FER-CM or FERT, actually indicate an increase in DNA damage induced by ferritin. Indeed, by normalizing the percentage of micronucleated cells for the mitotic index (Fig. 5C), a substantial and significant (P ≤ 0.05; P ≤ 0.005) effect of ferritin on micronucleus formation can be envisaged in cultures pretreated with FER-T or FER-CM.

Fig. 6. A proposed model for ferritin-mediated cytotoxicity. For details see the text. Highlighted are the central events involved in cell death, triggered by ferritin uptake (black boxes), and the steps that are interfered with by iron chelation (DFO), radical scavenging (trolox), and EGF supplementation. Further essential intracellular regulatory elements (oxidant state and availability of ferritin for cytosolic iron sequestration), which may also interfere with the illustrated pathway but are not discussed in detail, are highlighted in gray letters.

Discussion Here we demonstrate for the first time that physiological (not recombinant) H-chain-rich isoferritins released from primary hepatocytes can exert detrimental effects on cell survival by promoting iron-dependent oxidative stress and lipid peroxidation under normal culture conditions. As a consequence, the accumulation of cellular lesions, in particular modifications of proteins and DNA damage, renders cells susceptible to apoptosis and necrosis. These results provide strong evidence for the existence of iron-dependent pathways regulating cell survival (Fig. 6). Ferritin, redox-active iron, and lysosomal permeability Distinct “labile” pools of redox-active iron are critical in the maintenance of intracellular iron homeostasis, in particular defining the susceptibility to cell death under conditions of oxidative stress—a context that has been studied extensively for the lysosomal pool of “free” iron [32,34,35,50–52]. Notably, the release of reactive compounds from oxidatively damaged lysosomes caused by iron can lead to lethal cell injury, which also includes DNA strand breaks [35,36,52]. Hence it seems reasonable that ferritin uptake will also stimulate LMP by increasing the lysosomal pool of free iron and as a consequence promote the onset of apoptosis and necrosis. Importantly, cellular uptake of ferritin is based on endocytosis mediated by H- and L-chainspecific ferritin binding sites followed by translocation of the endocytosed protein to the lysosomal compartment, where the iron is released by proteolytic ferritin degradation [20,27,28,53–56]. Based

on this mechanism, serum ferritin may serve as an iron supply especially in the brain, H-chain ferritins delivering more iron to the cell than L-chain-rich isoforms [20]. With respect to our findings, however, this kind of iron delivery will establish an ambiguous situation for the cell: increasing the iron pool at an enhanced risk of lysosomal instability. Hence, ferritin import may represent a potential cytotoxic challenge, taking into consideration that 2000–4500 iron atoms per ferritin molecule are incorporated. This assumption is supported by the concentration dependence of ferritin-mediated cytotoxicity [40] and may also be reflected by the enhanced cytotoxicity observed for FER-CM compared to the less pure FER-T. Interestingly, a 43-kDa ferritin subunit that is suggested to represent an H/L heterodimer [23], can be resolved upon electrophoresis in FER-CM, but is not detectable in FER-T (Fig. 1A). Dimeric (43 kDa) and trimeric (64 kDa) ferritin subunits have also been reported for other secreted H-chain-rich isoferritins [19,22,44], the observed electrophoretic patterns most likely pointing to isoformspecific conformational properties. However, the question of to what degree this could contribute to the stronger effect observed for FERCM remains the subject of further investigation. Nevertheless, it is conceivable that the H/L dimer renders the FER-CM richer in the Hchain, which results in an enhanced iron uptake due to the increased ferroxidase activity compared to basic, L-chain-rich isoferritins (such as that found in liver tissue). As a consequence, FER-CM as well as other secreted acidic isoforms may be able to store more iron and thus have a higher capacity to promote its pro-oxidant and cytotoxic potential.

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357

Indeed, experimental evidence suggests that the growthinhibiting and cytotoxic effects reported for several ferritin isoforms are dependent on the H-chain [14,19,21–23,28,53,57,58]. Although this subunit may mediate iron-independent responses in certain physiological contexts [25,26], the pronounced protective effect of iron chelation on FER-CM-induced cell death supports an iron-dependent pathway for ferritin-mediated apoptosis and necrosis. In this context it has to be mentioned that iron-chelating DFO is a lysosomotropic compound, which, similar to ferritin, is taken up by endocytosis [59]. Together with the finding that direct chelation of iron stored in ferritin is not likely but requires proteolytic digestion [54], this finding emphasizes a dependence of ferritin-mediated cytotoxicity on lysosomal iron liberation. Surprisingly, DFO inhibited cell death only when it was applied in combination with ferritin, whereas preincubation with the chelator failed to protect the cells from subsequent ferritin exposure. Whether this may indicate the possible coexistence of lysosomes “preloaded” with DFO and “ferritin-loaded” lysosomes free of DFO in the same cell remains the subject of further investigation. Interestingly, DFO also reduced the incidence of cell death in untreated control cultures, indirectly supporting the role of lysosomes in cellular ferritin turnover [35,60]. On the other hand, it has to be assumed that DFO also will counteract the autocrine, density-dependent stimulation of cell death arising from ferritin release to the culture medium itself [14]. The role of oxidative stress in ferritin-mediated cytotoxicity Once liberated from ferritin, free redox-active Fe2+ will stimulate the formation of ROS inside the lysosome via the Fenton reaction. As a consequence there is lipid peroxidation promoting LMP, a cascade of events that has been shown to be deleterious to cell integrity, in particular under conditions of increased oxidative stress [35,36,50,59]. Obviously, this mechanism is also central to apoptosis and necrosis induction by ferritin, as indicated by the inhibition of ferritin-induced cytotoxicity and HNE–HisP accumulation by radical scavengers (trolox) and iron chelators (DFO). Interestingly, iron chelation inhibited ferritin-mediated cell death at higher efficiency compared to radical scavenging by trolox. This accounts for a very strong “upstream” protection by DFO, removing the radical-forming “catalyst” Fe2+ itself. However, the highly lysosomotropic nature of DFO contrasts with the hydrophilic properties of trolox, which may hinder lysosomal uptake. As a consequence, trolox may act mainly as a ROS scavenger in the cytosol [61]. Therefore, the moderate effect of trolox on ferritin-mediated necrosis induction could be explained by a reduced potential to abrogate severe lysosomal ROS formation. However, scavenging of cytosolic ROS, which will be released upon LMP or may also be generated de novo in the cytosol by Fe2+ ions leaking from ruptured lysosomes, is sufficient to inhibit the onset of ferritin-mediated apoptosis. Hence, LMP resulting from iron-induced, lysosomal LPO will contribute to cytosolic LPO, which “amplifies” the primary lysosomal insult. The remarkable accumulation of HNE–HisP in the cytosol of ferritin-treated cells and its almost complete absence in the presence of DFO support this assumption. However, perinuclear accumulation of HNE–HisP persists in the presence of DFO, although to a lesser extent. Moreover, this is the only site of HNE–HisP accumulation in cells exposed to FER-CM in the presence of EGF + I. Furthermore, reduced cytosolic HNE–HisP accumulation is accompanied by improved cell spreading upon supplementation with DFO or EGF + I. Similarly, in FER-CM culture itself, individual responses with respect to cytosolic HNE–HisP and cell spreading are evident (Fig. 3I). Based on this observation it cannot be excluded that cell dynamics, in particular cytoskeletal rearrangements, receptor and endosomal sorting, and intracellular trafficking, may participate in the positioning of ferritin-loaded lysosomes and thus determine “compartmentalization” of LPO.

1355

Genotoxicity of ferritin Ferritin exposure also induced micronucleus formation. This is in good agreement with the finding that LMP-mediated iron release can lead to DNA damage, especially under conditions of oxidative stress [52,59]. Notably, this can be catalyzed in the nucleus by both relocalized iron and HNE formed upon iron-mediated LPO [62]. Although the perinuclear accumulation of HNE–HisP in cells of EGF + I/FER-CM-treated cultures accounts for an increased HNE concentration close to the nucleus, the lack of HNE–His immunoreactivity in the nucleus makes it unlikely that HNE by itself enters the nucleus and leads to DNA adducts by direct Michael addition. However, in hepatocytes HNE may also be metabolized to epoxynonanal, which can lead to the formation of DNA adducts [48] and thus lead to micronucleus formation. Nevertheless, it should not be neglected that HNE may also form aldehydic adducts to proteins located in the vicinity of the nuclear lamina, including components of the mitotic spindle, and it is reasonable that both effects may contribute to chromosomal destabilization and micronucleus formation. Ferritin-mediated cell death—role of MOMP With respect to the endpoint of these events, in particular the onset of apoptosis, our previous findings have shown that ferritin triggers the upregulation of p53, involving proapoptotic Fas signaling via FasL, which is also upregulated upon ferritin exposure and stimulates intrinsic mitochondrial responses, including MOMP activation by Bid [14,23]. Among several downstream targets, MOMP seems of particular relevance in ferritin-mediated cell death, because it may be stimulated by several interrelated mechanisms: (i) stress/ damage-dependent p53 signaling including gene transactivation and direct mitochondrial accumulation of p53 [63], (ii) translocation of FasL/Fas–caspase 8-activated Bid to the outer mitochondrial membrane, and (iii) the direct coupling of MOMP to LMP via release of ROS and Fe2+ including cytosolic LPO as well as leakage of cathepsins, which may proteolytically activate Bid [36,37]. Moreover, in addition to its adverse effects on protein and DNA function, HNE by itself may also directly trigger p53 as well as Fas-dependent proapoptotic pathways [64] and thus act as an important messenger in ferritinmediated cytotoxicity. Therefore, ferritin uptake will primarily promote LMP, but the multiple downstream effects will converge into a strong attack on mitochondrial integrity in hepatocytes, the degree of MOMP critically affecting the final decision between survival, apoptosis, or necrosis. The potential to suppress ferritinmediated cell death follows the order trolox bDFO bEGF + I, the EGF + I protecting at about the same efficiency obtained with neutralizing anti-H-ferritin antibodies or the corticosteroid dexamethasone (DEX) [23]. Whether antibody binding blocks receptor-mediated endocytosis of ferritin remains to be elaborated. However, EGF as well as DEX is able to inhibit MOMP in hepatocytes by upregulating bcl-XL and bxl-2 and downregulating proapoptotic Bid [65–67]. Importantly, in addition to antagonizing MOMP, bcl-XL also inhibits LMP in hepatocytes [68]. Thus it is conceivable that an EGF-mediated upregulation of bcl-XL will not only lead to MOMP inhibition but also counteract ferritin-mediated LMP, which may explain the strong protective effect of EGF on ferritin-mediated cell death, including the inhibition of LPO, as indicated by the absence of cytosolic HNE–HisP accumulation in EGF + I-treated cultures. In this context our findings further suggest that ferritin-mediated cytotoxicity switches from cell death induction in proliferatively quiescent cells to inhibition of mitosis in proliferating cells, when DNA damage as well as restricted perinuclear HNE–HisP accumulation is still detectable. Therefore, the signals transduced by p53 in response to ferritin exposure obviously lead to cell cycle arrest in proliferating hepatocytes. In addition, however, HNE generated upon ferritin uptake may also mediate cell cycle arrest in proliferating primary

1356

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357

hepatocytes. It is an accepted concept that HNE acts as a stress signaling molecule that can trigger growth arrest in proliferating cells by targeting cell cycle regulatory elements such as certain protooncogenes, including p53, and also interferes with growth-controlling pathways such as extracellular signal-regulated kinases, c-Jun aminoterminal kinases, phosphatidylinositol kinase/Akt (serine/threonine kinase), and protein kinase C signaling (for reviews see [69,70]). Furthermore, HNE can directly interfere with EGF receptor/MAPKmediated signaling, which leads to reduced cell proliferation [71], an effect that may explain the reduced efficiency of EGF in stimulating mitosis in ferritin-treated primary hepatocytes. Interestingly, NF-κB gene transactivation, which plays an important role in liver regeneration, is also downregulated by HNE [72,73]. Thus, the cross talk between HNE signaling and the EGF receptor–MAPK pathway, as well as the influence of HNE on further targets such as NF-κB, may establish a critical determinant for the proliferative/regenerative process, in particular under pro-oxidant conditions such as inflammation. Therefore, by promoting HNE production, ferritin will interfere with this cross talk and thus not only affect the in vitro growth of primary hepatocytes but also may play a yet to be defined role in the development of liver disease. To summarize, depending on specific growth conditions, p53- as well as HNE-dependent signaling will markedly affect the response to ferritin, the outcome differing between proliferating and resting cells. This could also explain why growth-inhibiting effects, but no distinct proapoptotic- or necrosis-inducing properties, have been reported for different ferritin species studied in proliferating cell lines [8,19,21,22,30,57]. In conclusion, ferritin uptake seriously affects cell survival depending on specific culture conditions in vitro. Obviously, individual cell properties define the resistance to ferritin “attack,” which is exemplified by the variability of HNE–HisP accumulation seen in FERCM-treated cultures reflecting cell-specific differences in the degree of LMP. It has been shown that lysosomal responses become heterogeneous under conditions of oxidative stress, and cellular antioxidant defenses, which stabilize lysosomal integrity, become critical for cell survival [32,35]. Hence, in ferritin-mediated oxidative stress the individual antioxidant capacity of a cell will determine the degree of LMP and affect the severity of the downstream effects (e.g., LPO, MOMP), thus defining the outcome: maintained proliferative capacity, cell cycle arrest, apoptosis, or—in the worst case—rapid necrotic cell death. Therefore, according to recent concepts [37], restricted LMP may either allow survival or stimulate apoptosis, whereas massive LMP will initiate a rapid onset of necrotic cell death. Intriguingly, the cytotoxicity of hepatocyte-derived acidic isoferritins follows a dose response that is marked by the incremental transition from apoptosis to necrosis at concentrations above 100 ng/ml [40], indicating an escalation of LMP at higher ferritin concentrations, at which the antioxidant capacity will rapidly become depleted in an increasing number of the cells. The motive of a pro-oxidant “challenge” exerted by secreted isoferritins, addressing the cell-specific tolerance to oxidative stress, is intriguing when these in vitro findings are transferred to the in vivo situation. As previously reported [14] density-dependent accumulation of ferritin in the culture medium will establish an in vitro environment, which becomes increasingly restrictive to cell survival and gives rise to selective cell elimination by promoting “intracellular oxidative stress.” Further investigations have to show whether these in vitro findings account for a putative in vivo role of secreted acidic isoferritins, acting as soluble mediators of cellular oxidative stress in tissue homeostasis and immune surveillance. Acknowledgments The authors gratefully acknowledge K.F. Lottspeich, E. Kuffner, and M. Zobawa (Max Planck Institute of Biochemistry, Martinsried,

Germany) for performing peptide mass fingerprint analysis and Brigitte Fiedler and Verena Fagerer (University of Salzburg) for technical support. References [1] Harrison, P. M.; Arosio, P. The ferritins: molecular properties, iron storage function and cellular regulation. Biochim. Biophys. Acta 1275:161–203; 1996. [2] Levi, S.; Santambrogio, P.; Cozzi, A.; Rovida, E.; Corsi, B.; Tamborini, E.; Spada, S.; Albertini, A.; Arosio, P. The role of the L-chain in ferritin iron incorporation: studies of homo and heteropolymers. J. Mol. Biol. 238:649–654; 1994. [3] Sammarco, M. C.; Ditch, S.; Banerjee, A.; Grabczyk, E. Ferritin L and H subunits are differentially regulated on a post-transcriptional level. J. Biol. Chem. 283: 4578–4587; 2008. [4] Cozzi, A.; Santambrogio, P.; Levi, S.; Arosio, P. Iron detoxifying activity of ferritin: effects of H and L human apoferritins on lipid peroxidation in vitro. FEBS Lett. 277: 119–122; 1990. [5] Zhao, G.; Arosio, P.; Chasteen, N. D. Iron(II) and hydrogen peroxide detoxification by human H-chain ferritin: an EPR spin-trapping study. Biochemistry 45: 3429–3436; 2006. [6] Berberat, P. O.; Katori, M.; Kaczmarek, E.; Anselmo, D.; Lassman, C.; Ke, B.; Shen, X.; Busuttil, R. W.; Yamashita, K.; Csizmadia, E.; Tyagi, S.; Otterbein, L. E.; Brouard, S.; Tobiasch, E.; Bach, F. H.; Kupiec-Weglinski, J. W.; Soares, M. P. Heavy chain ferritin acts as an antiapoptotic gene that protects livers from ischemia reperfusion injury. FASEB J. 17:1724–1726; 2003. [7] Ghosh, S.; Hevi, S.; Chuck, S. L. Regulated secretion of glycosylated human ferritin from hepatocytes. Blood 103:2369–2376; 2004. [8] Sirota, L.; Kupfer, B.; Moroz, C. Placental isoferritin as a physiological downregulator of cellular immunoreactivity during pregnancy. Clin. Exp. Immunol. 77: 257–262; 1989. [9] Hann, H. W.; Kim, C. Y.; London, W. T.; Blumberg, B. S. Increased serum ferritin in chronic liver disease: a risk factor for primary hepatocellular carcinoma. Int. J. Cancer 43:376–379; 1989. [10] You, S. A.; Wang, Q. Ferritin in atherosclerosis. Clin. Chim. Acta 357:1–16; 2005. [11] Recalcati, S.; Invernizzi, P.; Arosio, P.; Cairo, G. New functions for an iron storage protein: the role of ferritin in immunity and autoimmunity. J. Autoimmun. 30: 84–89; 2008. [12] Zandman-Goddard, G.; Shoenfeld, Y. Ferritin in autoimmune diseases. Autoimmun. Rev. 6:457–463; 2007. [13] Koc, M.; Taysi, S.; Sezen, O.; Bakan, N. Levels of some acute-phase proteins in the serum of patients with cancer during radiotherapy. Biol. Pharm. Bull. 26: 1494–1497; 2003. [14] Bresgen, N.; Ohlenschlager, I.; Wacht, N.; Afazel, S.; Ladurner, G.; Eckl, P. M. Ferritin and FasL (CD95L) mediate density dependent apoptosis in primary rat hepatocytes. J. Cell. Physiol 217:800–808; 2008. [15] Gray, C. P.; Arosio, P.; Hersey, P. Association of increased levels of heavy-chain ferritin with increased CD4+ CD25+ regulatory T-cell levels in patients with melanoma. Clin. Cancer Res. 9:2551–2559; 2003. [16] Miyata, Y.; Koga, S.; Nishikido, M.; Hayashi, T.; Kanetake, H. Relationship between serum ferritin levels and tumour status in patients with renal cell carcinoma. BJU Int. 88:974–977; 2001. [17] Riley, R. D.; Heney, D.; Jones, D. R.; Sutton, A. J.; Lambert, P. C.; Abrams, K. R.; Young, B.; Wailoo, A. J.; Burchill, S. A. A systematic review of molecular and biological tumor markers in neuroblastoma. Clin. Cancer Res. 10:4–12; 2004. [18] Tran, T. N.; Eubanks, S. K.; Schaffer, K. J.; Zhou, C. Y.; Linder, M. C. Secretion of ferritin by rat hepatoma cells and its regulation by inflammatory cytokines and iron. Blood 90:4979–4986; 1997. [19] Gray, C. P.; Franco, A. V.; Arosio, P.; Hersey, P. Immunosuppressive effects of melanoma-derived heavy-chain ferritin are dependent on stimulation of IL-10 production. Int. J. Cancer 92:843–850; 2001. [20] Fisher, J.; Devraj, K.; Ingram, J.; Slagle-Webb, B.; Madhankumar, A. B.; Liu, X.; Klinger, M.; Simpson, I. A.; Connor, J. R. Ferritin: a novel mechanism for delivery of iron to the brain and other organs. Am. J. Physiol., Cell Physiol. 293:C641–C649; 2007. [21] Broxmeyer, H.; Bognacki, J.; Dorner, M.; de Sousa, M. Identification of leukemiaassociated inhibitory activity as acidic isoferritins: a regulatory role for acidic isoferritins in the production of granulocytes and macrophages. J. Exp. Med. 153: 1426–1444; 1981. [22] Moroz, C.; Traub, L.; Maymon, R.; Zahalka, M. A. PLIF, a novel human ferritin subunit from placenta with immunosuppressive activity. J. Biol. Chem. 277: 12901–12905; 2002. [23] Bresgen, N.; Ohlenschlager, I.; Fiedler, B.; Wacht, N.; Zach, S.; Dunkelmann, B.; Arosio, P.; Kuffner, E.; Lottspeich, F.; Eckl, P. M. Ferritin—a mediator of apoptosis? J. Cell. Physiol. 212:157–164; 2007. [24] Moroz, C.; Grunspan, A.; Zahalka, M. A.; Traub, L.; Kodman, Y.; Yaniv, I. Treatment of human bone marrow with recombinant placenta immunoregulator ferritin results in myelopoiesis and T-cell suppression through modulation of the cytokine-chemokine networks. Exp. Hematol. 34:159–166; 2006. [25] Zahalka, M. A.; Barak, V.; Traub, L.; Moroz, C. PLIF induces IL-10 production in monocytes: a calmodulin–p38 mitogen-activated protein kinase-dependent pathway. FASEB J. 17:955–957; 2003. [26] Ruddell, R. G.; Hoang-Le, D.; Barwood, J. M.; Rutherford, P. S.; Piva, T. J.; Watters, D. J.; Santambrogio, P.; Arosio, P.; Ramm, G. A. Ferritin functions as a proinflammatory cytokine via iron-independent protein kinase C zeta/nuclear factor kappaBregulated signaling in rat hepatic stellate cells. Hepatology 49:887–900; 2009.

N. Bresgen et al. / Free Radical Biology & Medicine 48 (2010) 1347–1357 [27] Chen, T. T.; Li, L.; Chung, D. H.; Allen, C. D. C.; Torti, S. V.; Torti, F. M.; Cyster, J. G.; Chen, C. Y.; Brodsky, F. M.; Niemi, E. C.; Nakamura, M. C.; Seaman, W. E.; Daws, M. R. TIM-2 is expressed on B cells and in liver and kidney and is a receptor for H-ferritin endocytosis. J. Exp. Med. 202:955–965; 2005. [28] Moss, D.; Hibbs, A. R.; Stenzel, D.; Powell, L. W.; Halliday, J. W. The endocytic pathway for H-ferritin established in live MOLT-4 cells by laser scanning confocal microscopy. Br. J. Haematol. 88:746–753; 1994. [29] Moss, D.; Powell, L. W.; Arosio, P.; Halliday, J. W. Effect of cell proliferation on Hferritin receptor expression in human T lymphoid (MOLT-4) cells. J. Lab. Clin. Med. 120:239–243; 1992. [30] Fargion, S.; Cappellini, M. D.; Fracanzani, A. L.; De Feo, T. M.; Levi, S.; Arosio, P.; Fiorelli, G. Binding and suppressive activity of human recombinant ferritins on erythroid cells. Am. J. Hematol. 39:264–268; 1992. [31] Kroemer, G.; Jaattela, M. Lysosomes and autophagy in cell death control. Nat. Rev. Cancer 5:886–897; 2005. [32] Nilsson, E.; Ghassemifar, R.; Brunk, U. T. Lysosomal heterogeneity between and within cells with respect to resistance against oxidative stress. Histochem. J. 29: 857–865; 1997. [33] Persson, H. L.; Yu, Z.; Tirosh, O.; Eaton, J. W.; Brunk, U. T. Prevention of oxidantinduced cell death by lysosomotropic iron chelators. Free Radic. Biol. Med. 34: 1295–1305; 2003. [34] Yu, Z.; Persson, H. L.; Eaton, J. W.; Brunk, U. T. Intralysosomal iron: a major determinant of oxidant-induced cell death. Free Radic. Biol. Med. 34:1243–1252; 2003. [35] Kurz, T.; Terman, A.; Gustafsson, B.; Brunk, U. Lysosomes in iron metabolism, ageing and apoptosis. Histochem. Cell Biol. 129:389–406; 2008. [36] Boya, P.; Kroemer, G. Lysosomal membrane permeabilization in cell death. Oncogene 27:6434–6451; 2008. [37] Guicciardi, M. E.; Leist, M.; Gores, G. J. Lysosomes in cell death. Oncogene 23: 2881–2890; 2004. [38] Toyokuni, S.; Miyake, N.; Hiai, H.; Hagiwara, M.; Kawakishi, S.; Osawa, T.; Uchida, K. The monoclonal antibody specific for the 4-hydroxy-2-nonenal histidine adduct. FEBS Lett. 359:189–191; 1995. [39] Michalopoulos, G.; Cianciulli, H. D.; Novotny, A. R.; Kligerman, A. D.; Strom, S. C.; Jirtle, R. L. Liver regeneration studies with rat hepatocytes in primary culture. Cancer Res. 42:4673–4682; 1982. [40] Bresgen, N.; Rolinek, R.; Hochleitner, E.; Lottspeich, F.; Eckl, P. M. Induction of apoptosis by a hepatocyte conditioned medium. J. Cell Physiol. 198:452–460; 2004. [41] Huerta, S.; Goulet, E. J.; Huerta-Yepez, S.; Livingston, E. H. Screening and detection of apoptosis. J. Surg. Res. 139:143–156; 2007. [42] Wyllie, A. H.; Kerr, J. F.; Currie, A. R. Cell death: the significance of apoptosis. Int. Rev. Cytol. 68:251–306; 1980. [43] Hashimoto, M.; Sibata, T.; Wasada, H.; Toyokuni, S.; Uchida, K. Structural basis of protein-bound endogenous aldehydes: chemical and immunochemical characterizations of configurational isomers of a 4-hydroxy-2-nonenal–histidine adduct. J. Biol. Chem. 278:5044–5051; 2003. [44] Garty, B.; Kaminsky, E.; Moroz, C. The immunosuppressive human placental ferritin subunit p43 is produced by activated CD4+ lymphocytes. Clin. Diagn. Lab. Immunol. 2:225–226; 1995. [45] Moroz, C.; Kupfer, B.; Twig, S.; Parhami-Seren, B. Preparation and characterization of monoclonal antibodies specific to placenta ferritin. Clin. Chim. Acta 148: 111–118; 1985. [46] Esterbauer, H.; Schaur, R. J.; Zollner, H. Chemistry and biochemistry of 4hydroxynonenal, malonaldehyde and related aldehydes. Free Radic. Biol. Med. 11: 81–128; 1991. [47] Uchida, K.; Stadtman, E. R. Modification of histidine residues in proteins by reaction with 4-hydroxynonenal. Proc. Natl. Acad. Sci. USA 89:4544–4548; 1992. [48] Sodum, R. S.; Chung, F. L. 1,N2-Ethenodeoxyguanosine as a potential marker for DNA adduct formation by trans-4-hydroxy-2-nonenal. Cancer Res. 48:320–323; 1988. [49] Eckl, P. M.; Raffelsberger, I. The primary rat hepatocyte micronucleus assay: general features. Mutat. Res. 392:117–124; 1997. [50] Lu, Z.; Nie, G.; Li, Y.; Soe-Lin, S.; Tao, Y.; Cao, Y.; Zhang, Z.; Liu, N.; Ponka, P.; Zhao, B. Overexpression of mitochondrial ferritin sensitizes cells to oxidative stress via an iron-mediated mechanism. Antioxid. Redox Signal. 11:1791–1803; 2009. [51] Tenopoulou, M.; Kurz, T.; Doulias, P. T.; Galaris, D.; Brunk, U. T. Does the calceinAM method assay the total cellular 'labile iron pool' or only a fraction of it? Biochem. J. 403:261–266; 2007.

1357

[52] Tenopoulou, M.; Doulias, P. T.; Barbouti, A.; Brunk, U.; Galaris, D. Role of compartmentalized redox-active iron in hydrogen peroxide-induced DNA damage and apoptosis. Biochem. J. 387:703–710; 2005. [53] Moss, D.; Fargion, S.; Fracanzani, A. L.; Levi, S.; Cappellini, M. D.; Arosio, P.; Powell, L. W.; Halliday, J. W. Functional roles of the ferritin receptors of human liver, hepatoma, lymphoid and erythroid cells. J. Inorg. Biochem. 47:219–227; 1992. [54] Kidane, T. Z.; Sauble, E.; Linder, M. C. Release of iron from ferritin requires lysosomal activity. Am. J. Physiol., Cell Physiol. 291:C445–C455; 2006. [55] Kalgaonkar, S.; Lonnerdal, B. Receptor-mediated uptake of ferritin-bound iron by human intestinal Caco-2 cells. J. Nutr. Biochem. 20:304–311; 2009. [56] Meyron-Holtz, E. G.; Fibach, E.; Gelvan, D.; Konijn, A. M. Binding and uptake of exogenous isoferritins by cultured human erythroid precursor cells. Br. J. Haematol. 86:635–641; 1994. [57] Broxmeyer, H.; Williams, D.; Geissler, K.; Hangoc, G.; Cooper, S.; Bicknell, D.; Levi, S.; Arosio, P. Suppressive effects in vivo of purified recombinant human H-subunit (acidic) ferritin on murine myelopoiesis. Blood 73:74–79; 1989. [58] Fargion, S.; Fracanzani, A.; Brando, B.; Arosio, P.; Levi, S.; Fiorelli, G. Specific binding sites for H-ferritin on human lymphocytes: modulation during cellular proliferation and potential implication in cell growth control. Blood 78: 1056–1061; 1991. [59] Doulias, P. T.; Christoforidis, S.; Brunk, U. T.; Galaris, D. Endosomal and lysosomal effects of desferrioxamine: protection of HeLa cells from hydrogen peroxideinduced DNA damage and induction of cell-cycle arrest. Free Radic. Biol. Med. 35: 719–728; 2003. [60] Radisky, D. C.; Kaplan, J. Iron in cytosolic ferritin can be recycled through lysosomal degradation in human fibroblasts. Biochem. J. 336 (Pt. 1):201–205; 1998. [61] Wen, Y.; Leake, D. S. Low density lipoprotein undergoes oxidation within lysosomes in cells. Circ. Res. 100:1337–1343; 2007. [62] Eckl, P. M.; Ortner, A.; Esterbauer, H. Genotoxic properties of 4-hydroxyalkenals and analogous aldehydes. Mutat. Res./Fundam. Mol. Mech. Mutagen. 290:183–192; 1993. [63] Erster, S.; Mihara, M.; Kim, R. H.; Petrenko, O.; Moll, U. M. In vivo mitochondrial p53 translocation triggers a rapid first wave of cell death in response to DNA damage that can precede p53 target gene activation. Mol. Cell. Biol. 24:6728–6741; 2004. [64] Awasthi, Y. C.; Sharma, R.; Sharma, A.; Yadav, S.; Singhal, S. S.; Chaudhary, P.; Awasthi, S. Self-regulatory role of 4-hydroxynonenal in signaling for stressinduced programmed cell death. Free Radic. Biol. Med. 45:111–118; 2008. [65] Ethier, C.; Raymond, V. A.; Musallam, L.; Houle, R.; Bilodeau, M. Antiapoptotic effect of EGF on mouse hepatocytes associated with downregulation of proapoptotic Bid protein. Am. J. Physiol. Gastrointest. Liver Physiol. 285: G298–G308; 2003. [66] Bailly-Maitre, B.; de Sousa, G.; Boulukos, K.; Gugenheim, J.; Rahmani, R. Dexamethasone inhibits spontaneous apoptosis in primary cultures of human and rat hepatocytes via Bcl-2 and Bcl-xL induction. Cell Death Differ. 8:279–288; 2001. [67] Musallam, L.; Ethier, C.; Haddad, P. S.; Bilodeau, M. Role of EGF receptor tyrosine kinase activity in antiapoptotic effect of EGF on mouse hepatocytes. Am. J. Physiol. Gastrointest. Liver Physiol. 280:G1360–G1369; 2001. [68] Feldstein, A. E.; Werneburg, N. W.; Li, Z.; Bronk, S. F.; Gores, G. J. Bax inhibition protects against free fatty acid-induced lysosomal permeabilization. Am. J. Physiol. Gastrointest. Liver Physiol. 290:G1339–G1346; 2006. [69] Poli, G.; Schaur, R. J.; Siems, W. G.; Leonarduzzi, G. 4-Hydroxynonenal: a membrane lipid oxidation product of medicinal interest. Med. Res. Rev. 28: 569–631; 2008. [70] Barrera, G.; Pizzimenti, S.; Dianzani, M. U. Lipid peroxidation: control of cell proliferation, cell differentiation and cell death. Mol. Aspects Med. 29:1–8; 2008. [71] Liu, W.; Akhand, A. A.; Kato, M.; Yokoyama, I.; Miyata, T.; Kurokawa, K.; Uchida, K.; Nakashima, I. 4-Hydroxynonenal triggers an epidermal growth factor receptorlinked signal pathway for growth inhibition. J. Cell Sci. 112 (Pt. 14):2409–2417; 1999. [72] Ji, C.; Kozak, K. R.; Marnett, L. J. IkappaB kinase, a molecular target for inhibition by 4-hydroxy-2-nonenal. J. Biol. Chem. 276:18223–18228; 2001. [73] Page, S.; Fischer, C.; Baumgartner, B.; Haas, M.; Kreusel, U.; Loidl, G.; Hayn, M.; Ziegler-Heitbrock, H. W.; Neumeier, D.; Brand, K. 4-Hydroxynonenal prevents NFkappaB activation and tumor necrosis factor expression by inhibiting IkappaB phosphorylation and subsequent proteolysis. J. Biol. Chem. 274:11611–11618; 1999.