Ferritin Contributes toMelanoma Progression byModulating CellGrowth and Sensitivity to Oxidative StressAlfonso Baldi,1,5 Daniela Lombardi,6 Patrizia Russo,1 Emanuele Palescandolo,1 Antonio De Luca,1

DanieleSantini,3 FelicianoBaldi,5 LuigiRossiello,5 MariaLuciaDell’Anna,4 AriannaMastrofrancesco,4

VittoriaMaresca,4 EnricaFlori,4 PierGiorgioNatali,2 MauroPicardo,4 andMarcoG.Paggi1

Abstract Purpose: Employing an in vitro model system of human melanoma progression, we previouslyreported ferritin light chain (L-ferritin) gene overexpression in the metastatic phenotype.Here, we attempted to characterize the role of ferritin in the biology of human melanoma and inthe progression of this disease.Experimental Design: Starting from the LM human metastatic melanoma cell line, we engi-neeredcellclones inwhichL-ferritin geneexpressionwasdown-regulatedby the stableexpressionof a specific antisense construct. These cells were then assayed for their growth capabilities,chemoinvasive properties, and sensitivity to oxidative stress. Additionally, ferritin protein contentin primary and metastatic human melanomas was determined by immunohistochemistry.Results: Artificial L-ferritin down-regulation in the LM cells strongly inhibited proliferationand chemoinvasion in vitro and cell growth in vivo. In addition, L-ferritin down-regulated cellsdisplayed enhanced sensitivity to oxidative stress and to apoptosis. Concurrently, immunohisto-chemical analysis of a human melanoma tissue array revealed that ferritin expression level inmetastatic lesions was significantly higher (P < 0.0001) than in primary melanomas. Further-more, ferritin expression was constantly up-regulated in autologous lymph node melanomametastases when compared with the respective primary tumors in a cohort of 11patients.Conclusions: These data suggest that high ferritin expression can enhance cell growth andimprove resistance to oxidative stress in metastatic melanoma cells by interfering with theircellular antioxidant system. The potential significance of these findings deserves to be validatedin a clinical setting.

The hierarchy of the molecular events underlying humanmelanoma progression is still largely unknown. For this reason,more information is clearly needed regarding genetic changestriggering melanoma onset and progression (1–3). In the pastyears, several genes have been identified as differentiallyexpressed during melanocytic tumor progression (4–10), and

the technology of the cDNA arrays has been recently successfulin pointing out genes whose expression is associated withmetastatic potential in melanoma cells (11, 12). Indeed, theidentification of genes that are differentially regulated whenmelanoma cells attain metastatic potential may lead to thecharacterization of patients with high risk of developingmetastasis and possibly to the recognition of more specifictherapeutic targets.

Recently, we took advantage of an in vitro model ofmelanoma progression consisting of two cell lines: LP, whichis derived from a primary human melanoma, and LM, which isderived from a supraclavicular lymph node metastasis of thesame patient, the latter showing enhanced cell proliferationand clonogenic capacity (13). By means of cDNA arrays, weidentified several genes whose expression was modulated inour model system and, among these, the gene encoding forferritin light chain (L-ferritin) whose expression was found up-regulated in the LM cell line (14).

Ferritins are important regulators of the intracellular ironcontent. Pharmacologically obtained iron depletion inducesmajor cellular alterations, including cell cycle arrest andapoptosis (15), but at elevated tissue concentrations, ironresults as a potential toxicant (16). Storage of intracellular ironin the ferritin molecules together with a down-modulation oftransferrin receptor are the two key mechanisms through whichthe human tissues are shielded from the toxic effects of excess

www.aacrjournals.org Clin Cancer Res 2005;11(9) May1, 20053175

Authors’ Affiliations: 1Laboratory ‘‘C,’’ Department for the Development ofTherapeutic Programs and 2Laboratory of Immunology, Center for ExperimentalResearch, Regina Elena Cancer Institute ; 3Service of Oncology, CampusBioMedico University; 4San Gallicano Dermatological Institute, Rome, Italy;5Deparment of Biochemistry and Biophysics ‘‘F. Cedrangolo,’’ Section of AnatomicPathology, Second University of Naples, Naples, Italy; and 6Department ofExperimental Medicine, University of L’Aquila, L’Aquila, ItalyReceived 4/1/04; revised1/11/05; accepted 2/3/05.Grant support:Associazione Italiana per la Ricerca sul Cancro grant (M.G. Paggi),Ministero della Salute grants (M.G. Paggi and M. Picardo), General Broker Service,International Society for the Study of Comparative Oncology, and Istituto perl’Ambiente e l’Educazione Schole¤ Futuro grants (ISSCO and Futura-ONLUS)(A. Baldi, F. Baldi, and D. Santini), and Ateneo ex 60% grant (D. Lombardi).The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordancewith18 U.S.C. Section1734 solely to indicate this fact.Requests for reprints: Marco G. Paggi, Laboratory ‘‘C,’’ Department for theDevelopment of Therapeutic Programs, Regina Elena Cancer Institute, Centerfor Experimental Research,Via E. Chianesi 53, 00128 Rome, Italy. Phone: 39-06-5266-2550; Fax: 39-06-5266-2572; E-mail: paggi@ccd.inemm.cnr.it.

F2005 American Association for Cancer Research.

Human Cancer Biology

iron ions (17). Ferritins are made of 24 subunits assembled toform the apoferritin shell. The protein subunits are of two types,light chain and heavy chain, which share extensive homology.The heavy chain is predominant in the so-called acidicisoferritins, prevalently detectable in heart, kidney, and pla-centa. The light chain is usually found in the so-called basicisoferritins, detected in liver and spleen. Light and heavy chainferritin subunits are encoded by two specific genes that areunder separate transcriptional control. To date, the independentfunction and regulation of light and heavy chain ferritins stillremains an open issue (16, 18, 19). Translation of both ferritinsubunits is thought to be activated by iron through interactionswith the iron regulatory element motifs present in the promoterregion of the two genes and through cytoplasmic iron sensorscalled iron regulatory proteins (20). Tissue ferritin expressionchanges during development as well as in various pathologicstates. In the past years, there has been considerable interest inferritin as an oncofetal protein and as a marker associated withcellular proliferation (19). Findings regarding the relationshipbetween ferritin and cancer seem, at least in part, conflicting,which reflects the relative paucity of experiments in this area.Nevertheless, increased ferritin concentration in tumor versusnormal tissue has been reported in several malignancies, such ascolon and breast cancer (21–23) and seminoma (24).

Melanoma progression is associated with aberrant redoxregulation leading to a continuous oxidative stress andproduction of reactive oxygen species (ROS), chemical speciescontaining one or more unpaired electrons (25).

Due to the considerable novel interest raising on themechanisms linking iron metabolism and cancer cells (15, 26),we have sought to assess the functional significance of ferritin inthe biology of human melanoma, a tumor type characterized bya peculiar iron metabolism (27, 28). By means of artificial down-regulation of L-ferritin in metastatic human melanoma cells, weinvestigated the possible role of ferritin in the progression of thistumor and in its peculiar response to oxidative stress. Addition-ally, we determined immunohistochemically ferritin expressionlevel in primary and metastatic melanomas.

Materials andMethods

Cell lines. The in vitro model system used consisted of two celllines derived from a primary cutaneous melanoma (Clark’s level V;Breslow 12 mm; LP) or from its supraclavicular lymph node metastasis(LM). LP and LM cells were cultured in RPMI 1640 plus 10% FCS in a5% CO2 atmosphere. LM cells have enhanced proliferation rate andclonogenic capacity (13). After f10 in vitro passages, LP cells showed adifferent, more malignant, phenotype.7

To maintain cells in fatty acid precursor–supplemented cultureconditions, linoleic acid was added to the medium at the finalconcentration of 7 � 10�5 mol/L from a stock solution in absoluteethanol. An equal volume of absolute ethanol was added to themedium of control cells. Bovine serum albumin at the finalconcentration of 0.02% was added as a fatty acid carrier to bothcontrol and treated cells. Cells were exposed to fatty acids precursorsfor 6 days before analysis.

cDNA constructs. The human L-ferritin (Genbank accession no.M11147) cDNA was cloned in sense or antisense orientation in thepcDNA3-T7-Tag vector (Invitrogen, Carlsbad, CA) between BamHI andEcoRI sites. The primers used for the L-ferritin construct were (forward)

AGCGGATCCATGAGCTCCCAGATTCGTCAG and (reverse) AGC-GAATTCTTAGTCGTGCTTGAGAGTGAG. The primers used for theantisense-L-ferritin (AS-L-ferritin) construct were (forward) AGC-GAATTCATGAGCTCCCAGATTCGTCAG and (reverse) AGCGGATCCT-TAGTCGTGCTTGAGAGTGAG.

Northern blot analysis. Northern blot analysis was done asdescribed (14) using the human L-ferritin cDNA as a probe. Asubsequent hybridization with h-actin was used for normalization.Quantification of the bands was obtained by the ImageQuant 5.0software (Molecular Dynamics, Sunnyvale, CA).

Western blot analysis. Western blotting on cell lysates was done asdescribed (29) using a rabbit anti-human ferritin antibody (DAKO,Carpinteria, CA) at a working dilution of 1:500. Normalization wasdone using an anti-HSP70 mouse monoclonal antibody (OncogeneScience, Manhasset, NY). Quantification of the bands was obtained bythe ImageQuant 5.0 software.

Transfections. LM cells (7 � 105) were transfected with 30 AgpcDNA3-T7-Tag plasmid as control, with the pcDNA3-T7-L-ferritin-Tag

construct (sense), or with the pcDNA3-T7-AS-L-ferritin-Tag construct,containing the ferritin light chain cDNA sequence in antisense

orientation, by electroporation as described (30). Under these con-ditions, 55% transfection efficiency was achieved. After growing electro-

porated cells in G418-containing medium (250 Ag/mL) for 14 days,single clones were isolated and grown. Control clones were propagated at60% to 70% of confluence. Because L-ferritin down-regulated LM clones

underwent apoptosis when kept >3 days in culture, these cells werepropagated every other day at 40% to 50% of confluence.

Proliferation assay. Cells were seeded in a 96-well plate (5 � 103

per well) 12 hours before the experiment. Cell growth was evaluated

using the Cell Proliferation Kit II (Roche Molecular Biochemicals,

Indianapolis, IN) following manufacturer’s instructions. This methodemploys a colorimetric procedure based on the tetrazolium salt

sodium, 2,3-bis[2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-

carboxanilide inner salt, which is transformed in a formazan dye onlyin metabolically active cells. The formazan dye was measured after

4 hours at a 492 nm wavelength. Values F SD are the average of threeexperiments done in decuplicate.

In vitro invasion assay. The chemoinvasion assay was done as

described (31) with some minor modifications. Briefly, cell cultureinserts incorporating 6.4 mm diameter membranes (8 Am pore size,

Becton Dickinson, Franklin Lakes, NJ) were used. Membranes werecoated with 10 Ag each of Matrigel basement membrane matrix (Becton

Dickinson). Results were expressed as the number of invaded cells on

the lower surface of the membrane. Ten microscope fields per eachmembrane were counted. Each experiment was done at least five times.

Terminal deoxynucleotidyl transferase–mediated dUTP nick end

labeling assay. Cells were fixed in paraformaldehyde, washed in

distilled water, and exposed briefly to 3% H2O2 to inactivate

endogenous peroxidase. The terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) reaction was done using

the peroxidase-based Apoptag kit (Intergene, Purchase, NY). TheTUNEL-positive cells were revealed by diaminobenzidine and H2O2

according to the manufacturer’s instructions. Finally, stained cells were

slightly counterstained with hematoxylin. TUNEL-positive cells wereconsidered as apoptotic.

Induction and evaluation of apoptosis. Cells were grown up to 70%confluence; then, apoptosis was induced by the addition of H2O2 (at thefinal concentrations of 25, 50, 100, and 200 Amol/L) or the anti-Fasantibody CH11 (Upstate Biotechnology, Lake Placid, NY; at the finalconcentrations of 15, 30, 60, and 120 ng/mL). Cells were assayed after60 minutes. Cell cycle analysis was done by flow cytometry using theCycleTest Plus DNA Reagent kit (Becton Dickinson) followingmanufacturer’s instructions. Cells were analyzed by a FACSCaliburflow cytometer (Becton Dickinson; 1 � 104 events per sample). Toestimate apoptosis, the acquired data were elaborated with theCellQuest version 3.3 software and analyzed with the ModFit LTversion 3.0 software.

www.aacrjournals.orgClin Cancer Res 2005;11(9) May1, 2005 3176

7 C. Leonetti, personal communication.

Human Cancer Biology

In vivo analysis of tumor growth inhibition. LM cells (1 � 106)

stably transfected with pcDNA3-T7-Tag plasmid or with the pcDNA3-T7-AS-L-ferritin-Tag construct were injected s.c. into the flanks of male

CD1 nu/nu mice (Charles River, Calco, Italy) in a final volume of

350 AL. Eight mice for each experimental point were employed. Tumorvolume was monitored by standard procedures. After 39 days, animals

were sacrificed and tumors were removed and processed for RNAextraction and for histology.

Descriptive statistical analysis was used to monitor tumor volume

modifications (expressed as median value and 95% confidential

interval of tumor volumes). Sheffe test and two-way ANOVA analysis

were used to compare tumor growth (tumor volume) of different

cancer clones (multiple comparisons). Ps V 0.05 were regarded as

statistically significant in two-tailed tests. SPSS software version 10.00

(SPSS, Chicago, IL) was used for statistical analysis.All procedures involving animals and their care were conducted

in conformity with the institutional guidelines in compliance with

national (D.L. No. 116, G.U., Suppl. 40, Feb. 18, 1992; Circolare No. 8,G.U., July 1994) and international laws (EEC Council Directive 86/

609, OJ L 358. 1, Dec 12, 1987; Guide for the Care and Use of

Laboratory Animals, U.S. National Research Council, 1996).Antioxidant enzymatic activities assay. Superoxide dismutase

(SOD) activity was evaluated as described (32). The results werereported as units/mg of protein.

Catalase activity was determined by the disappearance of H2O2

(10 mmol/L) measured at 240 nm in PBS (pH 7.4; ref. 33). The resultswere reported as units/mg of protein.

Reactive oxygen species detection. ROS production was detected by

2V,7V-dichlorofluorescein diacetate (Fluka AG, Buchs, Switzerland), asubstance oxidizable to fluorescent dichlorofluorescein by several

different ROS and peroxides. Cells were gently detached, washed, andincubated (30 minutes at 37jC, 5% CO2) in the presence of 2.5 Amol/L

2V,7V-dichlorofluorescein diacetate in PBS plus 5 mmol/L glucose. Flow

cytometric analysis was immediately done by a Cytoron Absolute(Ortho Diagnostic System, Raritan, NJ) with excitation and emission

setting at 488 and 530 nm, respectively. The intracellular ROS werequantified using the median of FL1 channel of fluorescence, which

matches with the maximal number of cells with the highestfluorescence.

Vitamin E analysis. Vitamin E was extracted (34) and measured(35) as described. Results are reported as the mean of two determi-nations from two different experiments and expressed as ng/mg protein.

Thiobarbituric acid–reactive substances analysis. The evaluation ofthiobarbituric acid–reactive substances (TBARS) was done as described(36). The assay reveals mainly malondialdehyde, which forms aspectrophotometrically detectable adduct with thiobarbituric acid. ForTBARS quantitative determination, malondialdehyde standard solu-tions of known concentration were used instead of the cell lysate. Wedid three different determinations in duplicate. TBARS, detected aspmol/mg of protein, were expressed as percentage variations due tointrinsic variability of the detection system itself.

Polyunsaturated fatty acid analysis. Lysed cells were extracted twicein chloroform/methanol (2:1) in the presence of butylated hydrox-ytoluene (100 Ag) as antioxidant. Phospholipid fraction was purifiedby TLC, and tricosanoic acid ethyl ester (100 Ag) was added as internalstandard. The fatty acids of phospholipid fraction were transmethylatedwith sodium methoxide in methanol and analyzed by a combined gaschromatography-mass spectrometry system (Hewlett-Packard, PaloAlto, CA, 5890 II gas chromatography combined with 5989 massspectrometry) on capillary column (FFA-P, 60 m � 0.32 Am �0.25 mm, Hewlett-Packard). Helium was used as carrier gas. Oventemperature gradient from 80jC to 220jC at 10jC/min was used (37).The results were obtained after time integration of the chromatogramand final processing of the peak areas and reported as polyunsaturatedfatty acid (PUFA) Ag/mg of protein.

Protein determination. Protein determination was done accordingto Bradford (38).

Immunohistochemistry. Immunohistochemistry on the humanmalignant melanoma tissue array CK1 slide (SuperBioChips Labora-tory, Seoul, South Korea) as well as on 11 primary melanomas andmatched lymph node metastases was done as described previously(14) using the rabbit anti-human ferritin antibody (A0133) at 1:250dilution or the monoclonal mouse antibody to human melanoma(clone HMB45, DBS, Pleasanton, CA) at 1:50 dilution after a10-minute Pronase treatment (DAKO). 3-Amino-9-ethylcarbazide wasthe final chromogen and hematoxylin is the nuclear counterstain. Thepercentage of ferritin-stained cells per field (�250) at light microscopywas calculated and compared in different specimens by three separateobservers (A.B., L.R., and F.B.) in a double-blind fashion, choosingareas almost completely composed by HMB45-positive cells. The scorewas as follows: +, up to 5% positive cells per field; ++, 6% to 15%positive cells per field; +++, 16% to 25% positive cells per field; and++++, >25% positive cells per field. The level of concordance,expressed as the percentage of agreement between the observers, was92%. In the remaining specimens, the score was obtained aftercollegial revision and agreement. Spearman’s rank correlation orFisher’s exact test was used to assess relationship between ordinal data.Ferritin values in the different subset of tumor (primary tumors,metastases) and in different pathologic stages of disease (tumor-node-metastasis) were compared according to Mann-Whitney U test fornonparametric independent variables. Ps V 0.05 were regarded asstatistically significant in two-tailed tests. SPSS software was used forstatistical analysis.

The LF03 anti-human ferritin light chain monoclonal antibody waselicited by human liver ferritin and selected for its specificity to ferritinlight chain (39). The suitability of this antibody for immunohisto-chemical analysis has been described elsewhere (40). The antibody wasused at a 1:500 dilution and incubated for 1 hour at room temperature.

The immunohistochemical results obtained with the polyclonalrabbit anti-human ferritin antibody raised against liver ferritin(prevalently L-ferritin) were confirmed in a selected number ofspecimens using the LF03 monoclonal antibody.

Results

Ferritin light chain gene expression in LP and LM cells.Northern and Western blot analyses (Fig. 1A and B, respec-tively) revealed higher L-ferritin gene and ferritin proteinexpression in the metastatic LM cells when compared withthe LP cells, the ones derived from the primary tumor.Quantitative results are reported in the legend of Fig. 1. Thisresult confirmed our previous findings in which, by cDNAarrays, we identified the L-ferritin gene as up-regulated in theLM cell line (14). In addition, immunohistochemical ferritinexpression was found from very low to undetectable in10 benign nevi. Figure 1C shows a representative staining.

Effect of L-ferritin down-regulation on proliferative andchemoinvasive properties of LM cells in vitro. These findingsprompted us to investigate the possible role of L-ferritin down-modulation in the phenotypic differences observed betweenLP and LM melanoma cell lines. To this end, we subcloned theL-ferritin cDNA in antisense orientation (AS-L-ferritin) into aconstitutive expression vector and transfected it into the LMcell line. Isolated individual clones of stable transfectants weregrown up to 40% to 50% confluence for 48 hours. Cell lysateswere analyzed for ferritin protein expression by Western blotusing a commercial polyclonal rabbit antibody raised againstferritin isolated from human liver, a tissue in which the lightchain of ferritin is prevalent over the heavy chain molecule(19). Six of nine selected transfectant clones displayed lowerlevels of ferritin when compared with the LM cells transfected

www.aacrjournals.org Clin Cancer Res 2005;11(9) May1, 20053177

Ferritin in HumanMelanoma Progression

with the control vector (Fig. 2A). All these clones actuallydisplayed a ferritin amount lower than the amount detectablein the LP cell line itself. Ferritin, when in lower amount, wasdetected as a doublet in 12% acrylamide gels possibly due to apartial cross-reaction with H-ferritin. Quantitative results arereported in the legend of Fig. 2. Among the selected clones, wechose those with the lowest ferritin protein expression(i.e., clones 2, 4, and 6). These clones were tested for theirproliferative and chemoinvasive properties. L-ferritin down-regulation resulted in a major decrease in the proliferativeactivity of the LM cells (Fig. 2B). Sense L-ferritin-transfectedLM cells did not display significant changes neither in ferritinamount expression by Western blot nor in proliferation rate(data not shown). Furthermore, L-ferritin down-regulationstrongly inhibited the ability of LM cells to invade an in vitro –constituted extracellular matrix (Fig. 2C).

Effect of L-ferritin down-regulation on apoptosis of LM cellsin vitro. TUNEL analysis, done on cells cultured for >48hours, provided the first indication that apoptosis was moreevident in L-ferritin down-regulated LM cells stably expressingAS-L-ferritin rather than in LM control transfectants (56.0 F13% versus 11.9 F 4%; Fig. 3A). With the aim of further

investigating the higher apoptotic rate of L-ferritin down-regulated cells, we did cell cycle analysis by flow cytometry ofeither control or AS-L-ferritin stable LM transfectants afterincubation in the presence of either H2O2 (25) or the anti-Fasmonoclonal antibody CH11 (41–44). Both these stimuliproduced a remarkable increase in apoptotic events in AS-L-ferritin stable LM transfectants rather than in control LM cells(Fig. 3B and C), demonstrating that ferritin down-regulationsignificantly lowered the threshold of apoptosis in LM humanmelanoma cells.

www.aacrjournals.orgClin Cancer Res 2005;11(9) May1, 2005 3178

Fig.1. Up-regulationof L-ferritinmRNA in the LMcell line.A, Northernblot analysisrevealed L-ferritin up-regulation in the highly metastatic LM cell line.Total RNA(10 Ag) from LPand LM cell lines was probed with 32P-labeled full-length L-ferritincDNA. After normalization by h-actin detection, L-ferritin mRNA content resulted2.5-fold higher in LM cells.B,Western blot analysis after a10% SDS-PAGEshowed ferritin protein amounts in LP and LM cells. After normalization by HSP70detection, ferritin protein content resulted 3.3-fold higher in LM cells.C, immunohistochemical detection in a nevus showed a barely detectableamount of ferritin.

Fig. 2. L-ferritin down-regulated LM clones and their in vitro growth and invasiveproperties.A,Western blot analysis after a12% SDS-PAGE. LM cells transfectedwith the empty vector (control) or with theAS-L-ferritin cDNA (clones1-6) wereassayed for ferritin expression. HSP70 detectionwas used for normalization. Eachlanewas loadedwith 30 Agprotein. After normalization by HSP70 detection, ferritinprotein content in control LM cells resulted10.1-,13.8-, 4.8-,10.1-, 3.7-, and8. 8-fold higher than in AS-L-ferritin LM clones1, 2, 3, 4, 5, and 6, respectively.B, proliferation assay: in vitro effects of L-ferritin down-modulation in LM cells(control, clones 2, 4, and 6). Proliferation rate was expressed as the amount offormazan detected spectrophotometrically after 4 hours. C, chemoinvasion assay:in vitro effects of L-ferritin down-modulation in LMcells (control, clones 2, 4, and 6).Invasiveness was expressed as the number of invaded cells on the lower surface ofthe coated membrane.

Human Cancer Biology

Effect of L-ferritin down-regulation on tumorigenicity of LMcells in vivo. L-ferritin down-regulated LM clones as well ascontrol cells were injected into the flanks of nude mice, andtumor volumes were measured thrice weekly. Tumors originat-ed from the L-ferritin down-regulated LM clones grewsignificantly slower than controls, with clones 2 and 4 showingthe strongest growth inhibition (Fig. 4A). Both Northernblot and immunohistochemistry confirmed the decrease inL-ferritin gene and protein expression in the xenograft explantsderived by L-ferritin down-regulated LM cells (Fig. 4B and C,respectively). The extent of tumor growth inhibition reachedstatistical significance at different times [i.e., control cells versus

L-ferritin down-regulated LM clones 2 and 4 at day 22 (P =0.039 and 0.045, respectively)], whereas control cells versusAS-L-ferritin LM clone 6 reached a borderline statisticalsignificance at day 34 (P = 0.08).

L-ferritin down-modulation is associated with the developmentof an oxidative stress. To investigate whether L-ferritin down-regulation could impair the antioxidant defense system, thuscausing lipoperoxidative effects and cell damage, we evaluatedlipophilic and enzymatic antioxidants PUFA and TBARScontent in AS-L-ferritin-transfected and control-transfected LMcells.

In L-ferritin down-regulated LM cells, SOD activity wassignificantly higher (12.71 F 2.46 versus 4.54 F 1.21 units/mgprotein), whereas catalase activity resulted significantly lower(6.61 F 1.52 versus 16.21 F 3.09 units/mg protein) whencompared with control LM cells (P < 0.005). Consequently, theSOD/catalase ratio was higher in L-ferritin down-regulated LMcells (1.92) with respect to control cells (0.28), thus facilitatingH2O2 accumulation in the former.

The amounts of total PUFAs were 0.11 F 0.032 and 0.39 F0.066 Ag/mg of protein in L-ferritin down-regulated LM cells andcontrol cells, respectively (P < 0.005). TBARS, expressedas percentage variations, were significantly higher (41 F 18%)in L-ferritin down-regulated LM cells than in control cells(P < 0.001). Moreover, vitamin E levels were found higher inL-ferritin down-regulated LM cells than in control cells(7.64 F 1.22 and 5.84 F 0.86 ng/mg of protein, respectively;P < 0.01) as a possible compensatory mechanism to counteractoxidative stress (35, 45). All these data are summarized in Table 1.

The intracellular ROS level, however, as evaluated by the2V,7V-dichlorofluorescein diacetate fluorescence, was not signi-ficantly increased in L-ferritin down-regulated LM cells withrespect to controls (2V,7V-dichlorofluorescein diacetate fluores-cence median of 134 and 130, respectively).

To counteract the effects of fatty acid precursor depletion,a phenomenon generally occurring in cultured cells (46), in aseparated set of experiments, we supplemented the culturemedium of both control LM and L-ferritin down-regulated LMcells with the fatty acid precursor linoleic acid (added asspecified in Materials and Methods). Cells were then assayedfor TBARS and PUFA content and for proliferative capabilities.The data obtained from two different determinations, eachdone in triplicate, indicated that linoleic acid–supplementedcells showed increased TBARS content, which was morepronounced in L-ferritin down-regulated LM than in controlLM cells (20.4% and 50.5%, respectively). As far as PUFAsare concerned, we evaluated linoleic (C18:2D9,12) andarachidonic acid (C20:4D5,8,11,14), which are the mostrelevant cell membrane unsaturated components. In cellssupplemented with linoleic acid, we found a more pro-nounced increase in control LM cells than in L-ferritin down-regulated LM cells (8.27- and 4.92-fold, respectively). In turn,arachidonic acid, a key target of lipoperoxidative damage, wasfound considerably more reduced in L-ferritin down-regulatedLM than in control LM cells (8.83- and 3.55-fold reduction,respectively).

Furthermore, linoleic acid supplementation produced a48.26% decrease in control LM cells and a 38.17% decreasein L-ferritin down-regulated LM cells, which were alreadygrowth inhibited by the expression of the AS-L-ferritinconstruct (Fig. 5).

www.aacrjournals.org Clin Cancer Res 2005;11(9) May1, 20053179

Fig. 3. Susceptibility to apoptosis of L-ferritin down-regulated LM cells. A,TUNELanalysis in control and in L-ferritin down-regulated LM cells; cells with darknuclei were considered as apoptotic.B, induction of apoptosis in control and inL-ferritin down-regulated LM cells (AS-L-ferr.) by H2O2 at the concentrationsindicated. C, induction of apoptosis in control and in L-ferritin down-regulated LMcells by addition of the CH11anti-Fas monoclonal antibody at the concentrationsindicated.Columns , mean from four to six independent assays; bars, SD. *,P < 0.05,significant; **, P < 0.001, highly significant, calculated by the Student’s t test.These results apply to theAS-L-ferritin stable transfectant LM clone 2; the behaviorof clones 4 and 6 was essentially overlapping.

Ferritin in HumanMelanoma Progression

Immunohistochemical analysis of ferritin expression in humanmelanomas. The expression patterns of ferritin were assessed byimmunohistochemistry done on a tissue array consisting of 50samples of human melanomas ranging from in situ to metastatic.Ferritin was found always expressed in the cytoplasm. Ofinterest, strong ferritin staining was also detected in spindle-shaped cells or histiocytes of the stroma surrounding theneoplastic tissue. A broad range of ferritin expression rangingfrom no staining to a high number of positive cells per field

(>25%) was observed. Interestingly, we found higher ferritinexpression level in metastatic lesions, which resulted statisticallysignificant when compared with primary melanomas (P <0.0001). Moreover, a significant direct correlation was alsofound between ferritin expression and T or M status of theprimary melanomas (P < 0.0001 and P = 0.003, respectively).These data are summarized in Table 2. In addition, ferritinexpression in 11 human primary melanomas and in therespective autologous lymph node metastases showed, in allthese cases, a more abundant ferritin expression in the metastaticlesions (Fig. 6; Table 3). Interestingly, ferritin expression wasdramatically up-regulated also in two cases of metastatic lesionto the skin when compared with the respective primarymelanoma, as shown in Fig. 5, where the insets (top right) showimmunostaining for HMB45 expression in the consecutivetissue section.

www.aacrjournals.orgClin Cancer Res 2005;11(9) May1, 2005 3180

Fig. 4. LM clones stably expressingAS-L-ferritin and their in vivo growthproperties. A, in vivo growth rates of thecontrol and L-ferritin down-regulatedLM cells (clones 2, 4, and 6). Bars, 95%confidential interval.Tumor volumes(median values) were expressed inmm3.B, Northern blot analysis for L-ferritin inxenografts derived from control or L-ferritindown-regulated LM cells (clone 2). Afternormalization by h-actin detection, L-ferritinmRNA content resulted1.5-fold lower inL-ferritin down-regulated LM cells.C, immunohistochemical determination offerritin in xenografts derived from controlor L-ferritin down-regulated LM cells(clone 2, top and bottom, respectively).

Table 1. Pattern of the antioxidants and oxidativestress variables in control LM cells and L-ferritindown-regulated LM cells

ControlLM cells

L-ferritindown-regulatedLM cells P

SOD (units/mgof protein)

4.54F1.21 12.71F2.46 <0.005

Catalase (units/mgof protein)

16.27F 3.09 6.62F 1.50 <0.005

SOD/catalase ratio 0.28 1.92 N/AVitamin E(ng/mgof protein)

5.84F 0.86 7.64F1.22 <0.005

PUFAs(Ag/mgof protein)

0.39F 0.066 0.1F0.03 <0.005

TBARS(percentage value)

100 141F18 <0.001

NOTE:Values are the average F SD from three different experiments. Signifi-cance testing (P) was done using two-tailed t test.

Table 2. Immunohistochemical detection of ferritin inheterologous primary andmetastatic melanomas

Variable

Spearman’srank correlationcoefficient P

Primary vs metastatic 0.652 <0.0001T 0.634 <0.0001N 0.170 NSM 0.548 0.003

NOTE: Spearman’s rank correlation between ferritin expression and primaryversus nonrelated metastatic melanomas,T, N, or M clinical stage of disease,respectively.

Human Cancer Biology

Discussion

In this study, we show that high levels of L-ferritin arecorrelated with the metastatic phenotype in an in vitro model ofmelanoma progression. Artificial down-regulation of L-ferritinin LM metastatic melanoma cells is capable in fact of reducingtheir proliferation rate in vitro and in vivo as well as their invasivepotential. Furthermore, in these cells, biochemical changes dueto an oxidative stress and increased apoptotic rate wereobserved. Finally, in vivo analysis of human melanomas showedthat ferritin expression in primary tumors was significantlycorrelated with T and M status, being also significantly higher inmetastatic versus primary melanomas.

Although the biological role of ferritin in tumor cells is stillnot completely understood, various studies suggest thatintracellular ferritin level is a reliable marker of cell prolifera-tion (19). Iron deprivation has been shown to significantlyinhibit cell cycle and proliferation in cancer cells (15). In fact, iniron-deprived human breast cancer cells, down-regulation ofcyclin-dependent kinase activities, and decrease in cyclin D/cyclin-dependent kinase 4 protein levels is found, thus reducingtheir proliferative potential (47). In addition, it has beenreported that iron chelators induce apoptosis in tumor cells,indicating a potential role of intracellular iron in the regulationof programmed cell death (15, 48–51). Several studies haveconfirmed the expression of ferritin in tumor cells, where highconcentrations often correlate with higher degrees of cellproliferation (16, 21–24, 52–56). It is well known that cancerpatients often exhibit an elevated level of serum ferritin usuallycorrelated with a poor prognosis (16, 19, 57). This is also true inthe case of human melanoma, where increased serum ferritinconcentrations are associated with progressive metastaticdisease (58). In addition, ferritin is produced and secreted as

a cell growth factor by several cancer cells in vitro (59) and isconsidered responsible for some of the melanoma-inducedimmunosuppressive effects (60).

Melanoma cells often display an increased state of chronicoxidative stress when compared with melanocytes (35, 61).High levels of H2O2 stimulate the activity of several transcrip-tional factors, including nuclear factor-nB (25, 62), known toenhance survival pathways by inducing gene expression fordifferent growth factors; therefore, melanoma cells, withendogenous high ferritin levels, can be endowed with highersurvival and biological aggressiveness (62). Recently, ferritinheavy chain (H-ferritin) has been shown to play a key role inregulating apoptosis during inflammation. In fact, H-ferritin,induced by nuclear factor-nB, is required to prevent sustainedc-Jun NH2-terminal kinase cascade activation, thus inhibitingROS-mediated tumor necrosis factor-a-induction and theconsequent apoptosis. In essence, H-ferritin-driven inhibitionof c-Jun NH2-terminal kinase signaling depends on suppressingROS accumulation and is achieved through its ability tosequester iron ions (63).

In our experimental model, LM metastatic cells clearly seemto depend on their high intracellular ferritin content asindicated by the reduction of their proliferative and invasivepotential when L-ferritin was experimentally down-regulated.Thus, we can hypothesize that enhanced proliferation of cancercells is supported by high ferritin level with resulting decrease ofthe overall oxidative stress and the associated apoptosis.

L-ferritin down-modulation in LM cells was also associatedwith increased SOD and decreased catalase activities. Chronicpro-oxidant stimuli can induce SOD activity both in vitro andin vivo (64–66), whereas catalase, a heme-containing enzyme, ismore susceptible to inactivation by exposure to high doses ofperoxidizing substances, including its own substrate (H2O2;ref. 67). The unbalance between SOD and catalase activitiespredisposes to intracellular H2O2 production because catalase,in association with glutathione peroxidase, is devoted to removeH2O2 (68). When the production of H2O2 overwhelms catalaseactivity, in the presence of free transitional metals, such as iron,

www.aacrjournals.org Clin Cancer Res 2005;11(9) May1, 20053181

Fig. 5. In vitro growth properties of LM cells after supplementation of theculture mediumwith linoleic acid. Control and L-ferritin down-regulated LM cells(clone 2) were assayed in standard growth conditions or after supplementation ofthe culture mediumwith linoleic acid as specified in Materials andMethods.Proliferation rate was expressed as the amount of formazan detectedspectrophotometrically after 4 hours.

Table 3. Immunohistochemical detection of ferritin inautologous primary andmetastatic melanomas

Caseno.

Primarymelanoma

Matchedmetastasis

1 + +++2 ++ ++++3 ++ ++++4 +++ ++++5 + +++6 ++ ++++7 ++ +++8 + ++9 ++ +++10 + ++11 + ++

NOTE: +, up to 5% positive cell per field; ++, 6% to 15% positive cells perfield; +++, 16% to 25% positive cells per field; ++++, >25% positive cellsper field.

Ferritin in HumanMelanoma Progression

Fenton reaction is promoted, generating extremely reactivehydroxyl radical (69, 70). In fact, alteration of SOD and catalaseactivities in LM ferritin down-modulated cells was associatedwith peroxidation of the cell membrane unsaturated lipid andapoptosis. In addition, after linoleic acid supplementation, theincrease of linoleic acid content in the cell membrane was moreevident in control than in L-ferritin down-regulated transfectedcells, suggesting that, in the latter, linoleate could be quicklyconsumed by the lipoperoxidative damage. The pronounceddecrease in arachidonic acid level and TBARS accumulationstrongly suggest a higher susceptibility to lipid peroxidation aswell.

The attempt we made to up-regulate L-ferritin expression inLM cells via transfection of the L-ferritin cDNA in ‘‘sense’’orientation resulted in the generation of clones unable to expressany detectable increase in ferritin content; moreover, these cellsdid not show any significantly altered proliferative activity. Thiscan be possibly ascribed to threshold or saturation effects.

In association with the lipid peroxidation process, weobserved an increase in vitamin E concentration as an effort tocounteract cell membrane alterations. Comparable alterationshave been described in cultured melanocytes from pheomelanicpatients, which are more susceptible to proapoptotic stimuli(71). Similar events can possibly occur in L-ferritin down-regulated LM cells as well, thus increasing their sensitivity to thecytotoxic effect of proapoptotic chemotherapeutic drugs. In fact,we observed a significantly higher rate of spontaneous orinduced apoptosis in L-ferritin down-regulated LM cells than incontrol. All these events are essentially in agreement with theincreased production of Fenton-generated hydroxyl radicals. Asan alternative explanation, ROS-induced genomic damagecould favor the selection of a phenotype particularly resistantto apoptosis. These hypotheses are not mutually exclusive.

A synergistic effect between ferritin down-modulation andthe effect of chemotherapeutic anticancer agents able to raisethe intracellular iron content (72) could be envisaged ascandidates for clinical testing.

To the best of our knowledge, this is the first study showingthat ferritin may be a growth regulatory element of the molecularpathway controlling proliferation and invasiveness of mela-noma cells. Furthermore, we show that metastatic melanomasexpress higher ferritin levels, which may account for the elevatedserum levels of ferritin in melanoma patients with progressivemetastatic disease. Recently, comparative analysis of humanSerial Analysis of Gene Expression libraries from primary andmetastatic melanomas and other malignances has identifiedL-ferritin as one of the genes whose expression best distin-guishes the melanoma libraries from the other ones (73).

It has been proposed that during melanomagenesis theintrinsic antioxidant control of melanocytes is lost, andinappropriate redox-sensitive transcription factors activationoccurs, contributing to the establishment of an antiapoptoticphenotype and possibly concurring in determining theresistance of melanoma cells to chemotherapeutic agents(50). As the transformed melanocyte population expands,restrictions due to the high redox cycling can generate aselective growth advantage for tumor cells expressing highlevels of ferritin, which contributes to lower the toxic effects ofexcess intracellular iron ions. In conclusion, this generates aphenotype more resistant to apoptosis.

In view that the in vivo analysis has shown a significantincrease of ferritin amount in melanoma metastases, furtherstudies are required to determine whether ferritin up-regulationis a common and biologically critical event during melanomaprogression. The present findings may be relevant in generatingtherapeutic protocols that include modulation of intracellulariron content in cancer cells.

Acknowledgments

We thank Dr. C. Leonetti for providing the LPand LM cell lines, Prof. P. Arosio forthe LF03 anti-human ferritin light chain monoclonal antibody, and Drs. B.Vincenziand A. Abbate for the statistical analysis. This work is dedicated to the memory ofTullio Battista who gave his invaluable contribution during the early phases of thisproject.

www.aacrjournals.orgClin Cancer Res 2005;11(9) May1, 2005 3182

Fig. 6. Expression pattern of ferritin inhumanmelanomas and autologous lymphnode and cutaneous metastases. Lowferritin expression in a primary humanmelanoma (A) andhigh ferritin expression inthe autologous lymph node metastasis(A1) and autologous cutaneous metastasis(A2).Very low ferritin expression in aprimary humanmelanoma (B) and highferritin expression in the autologous lymphnode metastasis (B1) and autologouscutaneous metastasis (B2). Originalmagnification for all the panels,�250.Insets , immunostaining for HMB45 antigenin the respective consecutive tissuesection.Themelanomas represented as‘‘A’’and ‘‘B’’correspond to cases1and 7,respectively, reported inTable 3.

References1. Newton Bishop JA. Molecular pathology of mela-noma. CancerMetastasis Rev1997;16:141^54.2.Welch DR, Goldberg SF. Molecular mechanisms con-trolling humanmelanoma progression and metastasis.Pathobiology1997;65:311^30.

3. Nyormoi O, Bar-Eli M. Transcriptional regulation ofmetastasis-related genes in human melanoma. ClinExpMetastasis 2003;20:251^63.4. van Groningen JJ, Bloemers HP, Swart GW. Identi-fication of melanoma inhibitory activity and other

differentially expressed messenger RNAs in humanmelanoma cell lines with different metastatic capacityby messenger RNA differential display. Cancer Res1995;55:6237^43.5. HashimotoY, Shindo-Okada N,Tani M,Takeuchi K,

Human Cancer Biology

www.aacrjournals.org Clin Cancer Res 2005;11(9) May1, 20053183

Toma H,Yokota J. Identification of genes differentiallyexpressed in association with metastatic potential ofK-1735murinemelanomabymessengerRNAdifferen-tial display. Cancer Res1996;56:5266^71.6. Simon HG, Risse B, Jost M, Oppenheimer S, Kari C,Rodeck U. Identification of differentially expressedmessenger RNAs in humanmelanocytes andmelano-ma cells. Cancer Res1996;56:3112^7.7. IshiguroT, Nakajima M, Naito M, MutoT, TsuruoT.Identification of genes differentially expressed in B16murine melanoma sublines with different metastaticpotentials. Cancer Res1996;56:875^9.8. Duncan LM, Deeds J, Hunter J, et al. Down-regulation of the novel genemelastatin correlates withpotential for melanoma metastasis. Cancer Res 1998;58:1515^20.9. Zendman AJ, Cornelissen IM,Weidle UH, Ruiter DJ,van Muijen GN. TM7XN1, a novel human EGF-TM7-like cDNA, detected with mRNA differential displayusing human melanoma cell lines with different meta-static potential. FEBSLett1999;446:292^8.10. Huang F, AdelmanJ, Jiang HP, Goldstein NI, FisherPB. Identification and temporal expression pattern ofgenes modulated during irreversible growth arrestand terminal differentiation in human melanoma cells.Oncogene1999;18:3546^52.11. Clark EA, GolubTR, Lander ES, Hynes RO. Genomicanalysis of metastasis reveals an essential role forRhoC. Nature 2000;406:532^5.12. Bittner M, Meltzer P, ChenY, et al. Molecular classi-fication of cutaneous malignant melanoma by geneexpression profiling. Nature 2000;406:536^40.13. Leonetti C, BiroccioA, CandiloroA, et al. Increase ofcisplatin sensitivity by c-myc antisense oligodeoxynu-cleotides in a humanmetastatic melanoma inherentlyresistant to cisplatin.ClinCancerRes1999;5:2588^95.14. Baldi A, BattistaT, De Luca A, et al. Identification ofgenes down-regulated during melanoma progression:a cDNA array study. Exp Dermatol 2003;12:213^8.15. Le NT, Richardson DR. The role of iron in cell cycleprogression and the proliferation of neoplastic cells.Biochim Biophys Acta 2002;1603:31^46.16. Torti FM,Torti SV. Regulation of ferritin genes andprotein. Blood 2002;99:3505^16.17.Harrison PM, Arosio P. The ferritins: molecularprop-erties, iron storage function and cellular regulation.Biochim Biophys Acta1996;1275:161^203.18. Boyd D,Vecoli C, Belcher DM, Jain SK, DrysdaleJW. Structural and functional relationships of humanferritin H and L chains deduced from cDNA clones.JBiol Chem1985;260:11755^61.19. Bomford AB, Munro HN. Ferritin gene expression inhealth and malignancy. Pathobiology1992;60:10^8.20. Hentze MW, Kuhn LC. Molecular control of verte-brateironmetabolism:mRNA-basedregulatorycircuitsoperatedby iron, nitricoxide, andoxidative stress.ProcNatlAcadSciUSA 1996;93:8175^82.21.Vaughn CB,Weinstein R, Bond B, et al. Ferritin con-tent in human cancerous and noncancerous colonictissue. Cancer Invest1987;5:7^10.22. Guner G, Kirkali G,Yenisey C,Tore IR. Cytosol andserum ferritin in breast carcinoma. Cancer Lett 1992;67:103^12.23.Weinstein RE, Bond BH, Silberberg BK.Tissue ferri-tin concentration in carcinoma of the breast. Cancer1982;50:2406^9.24. Cohen C, Shulman G, Budgeon LR. Immunohisto-chemical ferritin in testicular seminoma. Cancer 1984;54:2190^4.25. Meyskens FL Jr, McNulty SE, Buckmeier JA, et al.Aberrant redox regulation in human metastatic mela-noma cells compared to normal melanocytes. FreeRadic Biol Med 2001;31:799^808.26. Kwok JC, Richardson DR. The iron metabolism ofneoplastic cells: alterations that facilitate proliferation?Crit Rev Oncol Hematol 2002;42:65^78.27. Richardson DR. The role of the membrane-boundtumour antigen, melanotransferrin (p97), in iron up-take by the human malignant melanoma cell. Eur JBiochem 2000;267:1290^8.28. Food MR, Des RR. Iron uptake by melanoma cells

from the soluble form of the transferrin homologue,melanotransferrin. Redox Rep 2002;7:279^82.29. Baldi A, De Luca A, Claudio PP, et al.The Rb2/p130geneproductisanuclearproteinwhosephosphorylationis cellcycle regulated. JCell Biochem1995;59:402^8.30. Lombardi D, Palescandolo E, Giordano A, PaggiMG. Interplay between the antimetastatic nm23 andthe retinoblastoma-related Rb2/p130 genes in pro-moting neuronal differentiation of PC12 cells. CellDeath Differ 2001;8:470^6.31. Albini A, Iwamoto Y, Kleinman HK, et al. A rapidin vitro assay for quantitating the invasive potential oftumor cells. Cancer Res1987;47:3239^45.32. Spitz DR, Oberley LW. An assay for superoxide dis-mutase activity in mammalian tissue homogenates.Anal Biochem1989;179:8^18.33. Clairborne A. Catalase activity. In: Greenwald RA,editor. Handbook of methods for oxygen radical re-search. Boca Raton (FL): CRC;1985. p. 283^4.34. Burton GW, Webb A, Ingold KU. A mild, rapid, andefficient method of lipid extraction for use in determin-ing vitamin E/lipid ratios. Lipids1985;20:29^39.35. Picardo M, Grammatico P, Roccella F, et al. Imbal-ance in the antioxidant pool in melanoma cells andnormal melanocytes from patients with melanoma.J Invest Dermatol1996;107:322^6.36. Jentzsch AM, Bachmann H, Furst P, Biesalski HK.Improved analysis of malondialdehyde in human bodyfluids. Free Radic Biol Med1996;20:251^6.37. Passi S, MorroneA, De Luca C, Picardo M, IppolitoF. Blood levels of vitamin E, polyunsaturated fattyacids of phospholipids, lipoperoxides and glutathioneperoxidase in patients affected with seborrheic der-matitis. JDermatol Sci1991;2:171^8.38. Bradford MM. A rapid and sensitive method for thequantitation of microgram quantities of protein utiliz-ing the principle of protein-dye binding. AnalBiochem1976;72:248^54.39. Luzzago A, Arosio P, Iacobello C, et al. Immuno-chemical characterization of human liver and heart fer-ritins with monoclonal antibodies. Biochim BiophysActa1986;872:61^71.40. Francanzani AL, Fargion S, Romano R, et al. Im-munohistochemical evidence for a lack of ferritin induodenal absorptive epithelial cells in idiopathic he-mochromatosis. Gastroenterology 1989;96:1071^8.41. Garrone P, Neidhardt EM, Garcia E, Galibert L, vanKooten C, BanchereauJ. Fas ligation induces apopto-sis of CD40-activated human B lymphocytes. J ExpMed1995;182:1265^73.42. Nagata S, Golstein P. The Fas death factor. Science1995;267:1449^56.43. Fulda S, Sieverts H, Friesen C, Herr I, Debatin KM.The CD95 (APO-1/Fas) system mediates drug-induced apoptosis in neuroblastoma cells. Cancer Res1997;57:3823^9.44. Ariza ME, Broome-Powell M, Lahti JM, Kidd VJ,Nelson MA. Fas-induced apoptosis in human malig-nant melanoma cell lines is associated with the acti-vation of the p34(cdc2)-related PITSLRE proteinkinases. J Biol Chem 1999;274:28505^13.45. ChowCK.Vitamin E andoxidative stress. Free RadicBiol Med1991;11:215^32.46. Sheridan AM, Fitzpatrick S,Wang C,Wheeler DC,LieberthalW. Lipid peroxidation contributes to hydro-gen peroxide induced cytotoxicity in renal epithelialcells. Kidney Int1996;49:88^93.47. Kulp KS, Green SL,Vulliet PR. Iron deprivation inhibitscyclin-dependentkinaseactivityanddecreasescyclinD/CDK4proteinlevels inasynchronousMDA-MB-453hu-manbreastcancercells. ExpCellRes 1996;229:60^8.48.Hileti D, Panayiotidis P, Hoffbrand AV. Iron chelatorsinduce apoptosis in proliferating cells. Br J Haematol1995;89:181^7.49. Leardi A, CaragliaM, Selleri C, et al. Desferioxamineincreases irondepletion and apoptosis inducedby ara-C of human myeloid leukaemic cells. Br J Haematol1998;102:746^52.50.Meyskens FL Jr, Farmer P, FruehaufJP. Redox reg-ulation inhumanmelanocytes andmelanoma.PigmentCell Res 2001;14:148^54.

51. Cen D, Gonzalez RI, Buckmeier JA, Kahlon RS,Tohidian NB, Meyskens FL Jr. Disulfiram induces ap-optosis in human melanoma cells: a redox-relatedprocess. Mol CancerTher 2002;1:197^204.52. Rossiello R, CarrieroMV, Giordano GG. Distributionof ferritin, transferrin and lactoferrin in breast carci-noma tissue. JClin Pathol1984;37:51^5.53. FlemingS. Immunocytochemical localizationof ferritininthekidneyandrenaltumours.EurUrol1987;13:407^11.54. Carter RL, Hall JM, Corbett RP. Immunohistochem-ical staining for ferritin inneuroblastomas.Histopathol-ogy1991;18:465^8.55. Yang HB, Hsu PI, Lee JC, Chan SH, Lin XZ, ChowNH. Adenoma-carcinoma sequence: a reappraisalwith immunohistochemical expression of ferritin.J Surg Oncol1995;60:35^40.56.WeberA, Engers R, Nockemann S, Gohr LL, Zur HA,Gabbert HE. Differentially expressed genes in associa-tion with in vitro invasiveness of human epithelioidsarcoma. Mol Pathol 2001;54:324^30.57. Hazard JT, Drysdale JW. Ferritinaemia in cancer.Nature1977;265:755^6.58. LugerTA, LinkeschW, Knobler R, Kokoschka EM.Serial determination of serum ferritin levels in patientswithmalignantmelanoma.Oncology1983;40:263^7.59. Kikyo N, Suda M, Kikyo N, et al. Purification andcharacterization of a cell growth factor from a humanleukemia cell line: immunological identity with ferritin.Cancer Res1994;54:268^71.60. Gray CP, Franco AV, Arosio P, Hersey P. Immuno-suppressive effects of melanoma-derived heavy-chainferritin are dependent on stimulation of IL-10 produc-tion. IntJCancer 2001;92:843^50.61. Chedekel MR. Photochemistry and photobiology ofepidermal melanins. Photochem Photobiol 1982;35:881^5.62. Meyskens FL Jr, Buckmeier JA, McNulty SE,Tohidian NB. Activation of nuclear factor-nB inhuman metastatic melanoma cells and the effect ofoxidative stress. Clin Cancer Res 1999;5:1197^202.63. PhamCG, Bubici C, Zazzeroni F, et al. Ferritinheavychain upregulation by NF-nB inhibitsTNFa-inducedapoptosis by suppressing reactive oxygen species.Cell 2004;119:529^42.64. Rohrdanz E, Schmuck G, Ohler S, Tran-Thi QH,Kahl R. Changes in antioxidant enzyme expression inresponse to hydrogen peroxide in rat astroglial cells.ArchToxicol 2001;75:150^8.65. Meewes C, Poswig A, Dissemond J, Brenneisen P,Scharffetter-Kochanek K. Adaptive antioxidantdefence in human dermal fibroblasts. Br J Dermatol2000;143:662^3.66. Meewes C, Brenneisen P, Wenk J, et al. Adaptiveantioxidant response protects dermal fibroblasts fromUVA-induced phototoxicity. Free Radic Biol Med2001;30:238^47.67. ArmeniT, BattinoM, Stronati A, et al.Total antioxidantcapacity andnuclearDNAdamage inkeratinocytes afterexposure toH2O2.BiolChem2001;382:1697^705.68. Yohn JJ, Norris DA,Yrastorza DG, et al. Disparateantioxidant enzyme activities in cultured human cuta-neous fibroblasts, keratinocytes, and melanocytes.J Invest Dermatol1991;97:405^9.69. Gutteridge JM. Biological origin of free radicals,and mechanisms of antioxidant protection. ChemBiolInteract 1994;91:133^40.70.Minotti G, Aust SD.The role of iron inoxygen radicalmediated lipid peroxidation. Chem Biol Interact 1989;71:1^19.71. MarescaV, Roccella M, Roccella F, et al. Increasedsensitivity to peroxidative agents as a possible patho-genic factor of melanocyte damage in vitiligo. J InvestDermatol1997;109:310^3.72. Kwok JC, Richardson DR. Anthracyclines induceaccumulation of iron in ferritin in myocardial and neo-plastic cells: inhibition of the ferritin iron mobilizationpathway. Mol Pharmacol 2003;63:849^61.73.Weeraratna AT, Becker D, Carr KM, et al. Generationand analysis of melanoma SAGE libraries: SAGE adviceon the melanoma transcriptome. Oncogene 2004;23:2264^74.

Ferritin in HumanMelanoma Progression