This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/authorsrights

Author's personal copy

Identification of H ferritin-dependent and independent genes in K562differentiating cells by targeted gene silencing and expression profiling

Roberta Misaggi a,1, Maddalena Di Sanzo a,1, Carlo Cosentino a, Heather M. Bond a, Domenica Scumaci a,Francesco Romeo a, Claudia Stellato b, Giorgio Giurato b, AlessandroWeisz b, Barbara Quaresima a, Tullio Barni a,Francesco Amato a, Giuseppe Viglietto a, Giovanni Morrone a, Giovanni Cuda a,Maria Concetta Faniello a,⁎, Francesco Costanzo a

a Department of Experimental and Clinical Medicine, Magna Græcia University of Catanzaro, Salvatore Venuta Campus, Viale Europa, 88100 Catanzaro, Italyb Laboratory of Molecular Medicine and Genomics, Department of Medicine and Surgery, University of Salerno, via Allende, 84081 Baronissi, Salerno, Italy

a b s t r a c ta r t i c l e i n f o

Article history:Accepted 30 October 2013Available online 14 November 2013

Keywords:DifferentiationMicroarrayGene expression

Ferritin is best known as the key molecule in intracellular iron storage, and is involved in several metabolic pro-cesses such as cell proliferation, differentiation and neoplastic transformation. We have recently demonstratedthat the shRNA silencing of the ferritin heavy subunit (FHC) in a melanoma cell line is accompanied by a consis-tentmodification of gene expression pattern leading to a reduced potential in termsof proliferation, invasiveness,and adhesion ability of the silenced cells.In this study we sought to define the repertoire of genes whose expression might be affected by FHC during thehemin-induced differentiation of the erythromyeloid cell line K562. To this aim, gene expression profiling wasperformed in four different sets of cells: i) wild type K562; ii) sh-RNA FHC-silenced K562; iii) hemin-treatedwild-type K562; and iv) hemin-treated FHC-silenced K562. Statistical analysis of the gene expression data, per-formed by two-factor ANOVA, identified three distinct classes of transcripts: a) Class 1, including 657 mRNAswhose expression is modified exclusively during hemin-induced differentiation of K562 cells, independentlyfrom the FHC relative amounts; b) Class 2, containing a set of 70 mRNAs which are consistently modified byhemin and FHC-silencing; and c) Class 3, including 128 transcripts modified by FHC-silencing but not by hemin.Our data indicate that FHC may function as a modulator of gene expression during erythroid differentiation andadd new findings to the knowledge of the complex gene network modulated during erythroid differentiation.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Iron is an essential macronutrient for living cells, involved in keyprocesses such as DNA synthesis (Beard et al., 1993). At the sametime, the presence of free iron in the cell induces the generation of reac-tive oxidative species (ROS) thatmay cause irreversible cellular damagethrough lipid peroxidation and DNA breakage (Ray et al., 2012). Thetask of maintaining intracellular iron in a soluble, nontoxic form is per-formed by ferritin, whose 24-mer shell is composed by a variable arrayof heavy (FHC) and light (FLC) subunits, encoded by two distinct genes(Costanzo et al., 1986). Despite a high degree of aminoacid homology,FHC and FLC subunits play different functions in the ferritin shell: FHC

possesses ferroxidase activity and is involved in rapid iron uptake andrelease, while FLC is devoid of enzymatic activity and contributes tolong-term iron storage (Levi et al., 1988). Tissue distribution of thetwo subunits is also unequal: FLC is particularly expressed in the liverand spleen and therefore dominant in the structure of the shell inthese tissues, whereas FHC is mostly present in ferritins from theheart and kidney as well as in several types of neoplastic cells(Bomford et al., 1981).

The structural and functional differences between the two subunitsand the diversity in tissue-specific distribution are also reflected at thetranscriptional level. Besides thepresence of distinct trans-acting factorsacting on the two ferritin promoters, a robust body of evidence suggeststhat the FHC promoter activity is differentially regulated in different celltypes as well as in different stages of cell life (Bevilacqua et al., 1992,1995, 1997a, 1998; D'Agostino et al., 1995; Faniello et al., 1999). More-over, the FHC promoter is positively regulated by cAMP (Bevilacquaet al., 1997b) and by cJun (Faniello et al., 2002) whereas E1A(Bevilacqua et al., 1997b) and p53 (Faniello et al., 2008) act as transcrip-tional repressors.

Gene 535 (2014) 327–335

Abbreviations: FHC, Ferritin Heavy Chain; FLC, Ferritin Light Chain; shRNA, shorthairpin RNA; qPCR, quantitative Real Time Polymerase Chain Reaction.⁎ Corresponding author. Tel.: +39 0961 3694354; fax: +39 0961 4090.

E-mail address: faniello@unicz.it (M.C. Faniello).1 These authors contributed equally to this work.

0378-1119/$ – see front matter © 2013 Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.gene.2013.10.067

Contents lists available at ScienceDirect

Gene

j ourna l homepage: www.e lsev ie r .com/ locate /gene

Author's personal copy

Recent data indicate that the FHC subunit, which is localized not onlyin the cytoplasm but also in mitochondria and nucleus, plays additionalroles in several cell regulatory pathways in addition to ironmetabolism(Alkhateeb and Connor, 2010). For instance, it has been recently discov-ered that FHC regulates angiogenesis in response to inflammation andtumor progression (Coffman et al., 2009) and that it interfereswith che-mokine receptor signaling. In particular, the FHC subunit is able to phys-ically interact with the CXCR4 receptor and this interaction determinesan inhibitory effect on the downstream signaling pathway,which is cru-cial in the regulation of neuronal migration, differentiation, and survival(Li et al., 2006). The interaction of FHC with CXCR4 is also important indetermining the extent of the nuclear translocation of FHC (Li et al.,2006). The role of nuclear H ferritin has not been fully elucidated yet:it has been proposed that, besides protecting the cell from oxidativeand UV-induced DNA damage, FHC might bind DNA and contributeto the transcriptional control of certain genes, including β-globin(Pountney et al., 1999).

Ferritin expression is modulated during cell differentiation, mostlyat the transcriptional level, since significant modifications in theamount of FHC and FLC transcripts have been reported in rat thyroidcells (Chazenbalk et al., 1990) in rabbit vascular smooth-muscle cells(Liau et al., 1991) in Caco-2 cells (Bevilacqua et al., 1995) and in avariety of hematopoietic cell lines such as HL60, K562 and Frienderythroleukaemic cells (Iwasaki et al., 2006). Interestingly, the molecu-lar dissection of the FHC promoter revealed that the cis-elementinvolved in the differentiation-modulated transcription in Caco-2 cellsis represented by a proximal CAAT element, recognized by NF-Y(Bevilacqua et al., 2002) while in K562 cells the most significant ele-ment is an ARE sequence, where the differentiation-inducing agenthemin promotes the binding of several b-zip transcription factors(Iwasaki et al., 2006). During hemin-induced erythroid differentiationof K562 cells, the increase in ferritin synthesis is also due to an improvedtranslational efficiency (Iwasaki et al., 2006). Both transcriptional andtranslational upregulation of FHC fulfill the metabolic requirements ofthe differentiating cells for hemoglobinization and cytoprotectionagainst oxidative-stress mediated cytotoxicity (Wandzik et al., 2009).

The dramatic changes in FHC gene expression observed during celldifferentiation, particularly in erythroid differentiation, prompted usto investigate the role of FHC in K562 cells differentiating in responseto hemin. To this aimwe selectively silenced FHCmRNA and performedmicroarray analysis on the following four different sets of cells: i) un-treated wild type K562 (K562WT), ii) FHC-silenced K562 (K562shFHC),iii) hemin-treatedK562WT, and iv) hemin-treated K562shFHC. The resultsillustrated in this paper: i) indicate that FHC-silenced K562 cells are stillable to differentiate in response to hemin; ii) delineate a role for FHC asa relevant modifier of gene expression during erythroid differentiation;and iii) provide a list of potential FHC-modulated transcripts. All togeth-er our data confirm that the H ferritin subunit plays additional roles inthe cell aside from controlling iron metabolism, and provide new in-sights into its potential regulatory role in critical biological processes.

2. Materials and methods

2.1. Cell culture

K562 human erythroleukemia cells (ATCC number: CCL 243) werecultured in RPMI 1640 medium supplemented with 0.3 g/l glutamine,25 mMHEPES, and 10% fetal bovine serum (Sigma) at 37 °C in a humid-ified 5% CO2 atmosphere. Hemin was added at 50 μM as described(Pierantoni et al., 2003) for 24, 36,72 h.

2.2. Preparation of lentiviral supernatants and transduction of K562 cells

Lentiviral preparations and transductions were performed as previ-ously described (Mega et al., 2011). Briefly, 5 × 106 HEK-293T cellswere grown on 10-cm plates to 70–80% confluence and co-transfected

with 10 μg shRNA lentiviral DNA, 2 μg pCMV-VSV-G expressing enve-lope plasmid, and 18 μg packaging viral CMV delta 8.9 plasmid, usingthe calcium phosphate precipitation method. Eight hours later, freshmedium was added, and cells were cultured for an additional 2 days.Themediumwas harvested 48 h post-transfection, and filtered through0.45 μm filters. The supernatants were used to cross-transduce K562cells in the presence of 8 μg/ml polybrene (Sigma-Aldrich) and positivecloneswere isolated by puromycin selection (1 μg/ml). Cellswere stablytransduced with a lentiviral DNA containing either an shRNA that tar-gets the 196–210 region of the FHC mRNA (sh29432) (K562shFHC), ora control shRNA without significant homology to known humanmRNAs (K562shRNA).

2.3. RNA extraction and semi-quantitative reverse transcriptasepolymerase chain reaction (RT-PCR)

Total RNA was extracted from three distinct K562 cell cultures,namely: i) wild-type K562WT, ii) FHC-silenced K562 (K562shFHC), andiii) scrambled-shRNA K562 (K562shRNA). For each culture type two bio-logical replicates were used. The TRIzol method (Invitrogen Life Tech-nologies) was used according to the manufacturer's instructions andreverse-transcription was subsequently performed with a reverse tran-scription kit (Invitrogen) (Quaresima et al., 2008). FHC cDNA (2 μl) wasamplified with the following primers: forward 5′-cat caa ccg cca gat caac-3′; reverse, 5′-gat ggc ttt cac ctg ctc at-3′. A human GAPDH cDNA frag-ment was amplified as the internal control for the amount of cDNA inthe PCR (Faniello et al., 2006). The amplified products were electropho-resed on 1.5% agarose gels.

2.4. Western blotting analysis

K562 cells (K562WT; K562shFHC; K562shRNA) were lysed in the follow-ing buffer [20 mM Hepes pH 7.9, 420 mM NaCl, 1% Triton X-100, 1 mMEDTA, 25% glycerol, 1 mM PMSF, 1 mM Na3VO4, 1 mM DTT, 1 μg/mlaprotinin, 1 μg/ml leupeptin] for 30 min on ice. After removal of thecell debris by centrifugation (12,000 ×g, 30 min), the concentration ofproteins in the supernatant was measured by the Bio-Rad protein assayaccording to the manufacturer's instructions (BioRad Laboratories).

A total of 70 μg protein extract was boiled for 10 min in SDS samplebuffer, separated by 12% SDS-PAGE and the proteinswere transferred toa nitrocellulose membrane by electroblotting. Non-specific reactivitywas blocked in nonfat dry milk in TPBS [5% (w/v) milk in PBS (pH 7.4)and 0.005% Tween 20] for 2 h at room temperature. The membranewas incubated with rabbit anti-H ferritin antibody (1:200; Santa CruzBiotechnology) for 2 h at room temperature, followed by incubationwith goat anti-rabbit horseradish peroxidase-conjugated secondary an-tibody (1:5000; Santa Cruz Biotechnology). The membrane was devel-oped by ECL-Western blot detection reagents according to themanufacturer's instructions (Santa Cruz Biotechnology). γ-Tubulinwas used as a loading control.

2.5. Microarray analysis

Formicroarray analysis RNA extracted from two biological replicateswere pooled. RNA concentration was determined with a Nanodrop(NanoDrop, Wilmington, Delaware, USA) spectrophotometer and itsquality was assessed with an Agilent 2100 Bioanalyzer (Agilent Tech-nologies, Milano, Italy). For each sample, 500 ng total RNAwere synthe-sized to biotinylated cRNA using the Illumina RNA Amplification Kit(Ambion, Inc., Austin, TX). Synthesis was carried out according to themanufacturers' instructions. cRNA concentration and the quality wereassessed as described above (Quaresima et al., 2008).

Each RNA sample, obtained by combining equal amounts of RNAfrom two independent biological replicates, was analyzed inHumanHT12 v.4.0 Expression BeadChips arrays (Illumina Inc., SanDiego, CA, USA) by hybridizing in each case 750 ng cRNA to the array

328 R. Misaggi et al. / Gene 535 (2014) 327–335

Author's personal copy

for 18 h, to according to the protocol provided by the manufacturer.Arrays were washed and stained with streptavidin-conjugated Cy3(GE Healthcare Milano, Italy), before washing drying and scanningwith the Illumina BeadArray Reader (Illumina Inc., San Diego, CA,USA) (Cicatiello et al., 2010). For data analysis, the intensity files wereloaded into the Illumina GenomeStudio v.2010.3 (Gene ExpressionModule: 1.8.0) software for quality control and gene expression analy-sis. First, the quantile normalization algorithm was applied on thedataset to correct systematic errors. For expression analysis, technicalreplicates of each sample were grouped together and genes with a de-tection p-value below 0.05, corresponding to a false positive rate of5%, were considered as detected (Mutarelli et al., 2008).

2.6. Database submission of microarray data

Themicroarray data were prepared according tominimum informa-tion about a microarray experiment (MIAME) recommendations(Brazma et al., 2001) and deposited in the GEO database (http://www.ncbi.nlm.nih.gov/geo/). The GEO accession number for the platform isE-MTAB-1336 and the study may be viewed at: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=E-MTAB-1336.

2.7. Statistical analysis

A two-factor ANOVA test by software Matlab Statistics Toolbox hasbeen run to statistically identify the effects of the two factors thatchange among the experiments: FHC silencing and hemin stimulus.Therefore, the two-way ANOVA test provides a means to simul-taneously evaluate the differential expression among four cellpopulations: wild-type, FHC-silenced, wild-type stimulated by hemin,and FHC-silenced stimulated by hemin.

The two-way ANOVA test yields, for each gene, three p-values:i) pFHC is associated to the statistical significance of the effect of FHCon the differential expression; ii) pHem is associated to the statisticalsignificance of the effect of hemin on the differential expression; andiii) pFHC+Hem is associated to the statistical significance of the interac-tion effect between FHC and hemin; a significant value indicates thatthere is a combined effect of FHC and hemin, which is not explainablejust by the superposition of the effects of the single factors.

On the basis of the above values, the differentially expressed geneshave been divided into three classes: Class 1—genes influenced byhemin changes only: significant values only for pHem; Class 2—genesinfluenced by both FHC and hemin changes: significant values for both

Fig. 1. Heavy ferritin subunit mRNA and protein expression. (A) RT-PCR analysis from K562 (K562WT) cells, K562 transduced with shFHC (K562shFHC), and K562 transduced with shRNAcontrol (K562shRNA). (B) Western blot analysis from K562WT, K562shFHC and K562shRNA.

Fig. 2. Erythroid differentiation of K562 cells after induction by hemin. (A)Western blot analysis of K562 cells for anti FHC, γ-globin, and γ-tubulin antibodies after treatment with heminfor indicated times. (B)Western blot analysis of K562WT and K562shFHC cells treated (H) or untreated (CTR) with hemin for 72 h for anti FHC, γ-globin, and γ-tubulin antibodies. (C) Real-time PCR analysis performed on the K562WT and K562shFHC cells treated (H) or untreated (CTR) with hemin for 72 h to evaluate mRNA expression levels of FHC. The real time data isexpressed as themean ± SD of three independent experiments performed in triplicates. (D) Real-time PCR analysis performedon theK562WT andK562shFHC cells treated (H) or untreated(CTR) with hemin for 72 h to evaluate mRNA expression levels of FLC. The real time data is expressed as the mean ± SD of three independent experiments performed in triplicates.

329R. Misaggi et al. / Gene 535 (2014) 327–335

Author's personal copy

pFHC and pHem; and Class 3—genes influenced by FHC changes only:significant values only for pFHC.

To control the family-wise error rate (FWER) to about 0.05, theBonferroni correction method, which yields a threshold for significanceof p b 1e − 6 was applied. Furthermore, to ensure that the identifiedgene expression changes are also biologically relevant, the four geneclasses have also been filtered, to include only those that have a foldchange (FC) b −1.5 or N1.5. For each class, the filter on the fold changeis applied only with respect to the comparisons that are pertinent tothat class, i.e. considering which factor is significant for that class: inClass 1 we took into account the FC only in the comparisonsK562WT+Hem_vs_K562WT and K562shFHC+Hem_vs_K562shFHC,since the effect of FHC silencing is not statistically significant for thegenes of this class, in Class 2 we considered the FC of all the four com-parisons considered for classes A and B, while in Class 3 onlyK562shFHC_vs_K562WT and K562shFHC+Hem_vs_K562WT+Hemcomparisons were considered.

2.8. Quantitative real time PCR (qRT-PCR)

Total RNA extraction for semi-quantitative reverse transcriptase po-lymerase chain reaction was performed on the following cell lines:K562WT treated and untreated with hemin; K562shFHC treated and un-treated with hemin. Real time PCR was performed using as a sourcefor the cDNA synthesis either the same RNA samples subjected to mi-croarray or RNA “ad hoc” prepared from biological replicates. Each frag-ment was amplified in a 20 μl reaction containing 1× SYBR Green I PCRMaster mix (Bio-Rad Laboratories), 20 pmol of each primer pair, 20 ngof cDNA (total RNA equivalent) and nuclease-free water. The thermalprofile consisted of 1 cycle at 95 °C for 3 min followed by 45 cycles at95 °C for 10 s, 60 °C for 10 s, 72 °C for 20 s. The PCR reaction wasfollowed by a melting profile analysis. A human glyceraldehyde3-phosphate dehydrogenase (GAPDH) cDNA fragment was amplifiedas the internal control for the amount of cDNA in the qPCR. Each exper-iment was performed in triplicate.

Fig. 3. Effects of hemin treatment and FHC silencing on mRNA expression in K562 cells. Heatmaps summarizing mRNA expression profiling results obtained in K562 cells upon hemintreatment and/or FHC silencing. The following gene expression profiling, performed in triplicate fold-changes calculated in the comparisons pertinent to each class are reported.(A) 657 mRNAs influenced by hemin treatment in both wt and shFHC cells (Class 1). (B) 128 mRNAs influenced by FHC silencing in both untreated and hemin-treated cells (Class 3).(C) 70 mRNAs influenced by both hemin and FHC silencing (Class 2).

330 R. Misaggi et al. / Gene 535 (2014) 327–335

Author's personal copy

The primer sequences were as follows:

MAGEA4 forTGC TCT GAC CAG CCG CTT GG

MAGEA4 revAGG GCA GCA GGC AAG AGT GC

SPARC forGTG CAG AGG AAA CCG AAG AG

SPARC revTCA TTG CTG CAC ACC TTC TC

FHC forCATCAACCGCCAGATCAAC

FHC revGATGGCTTTCACCTGCTCAT

FLC forCGTCAGAATTATTCCACCGACG

FLC revGAGTGAGCCTTTCGAAGAG

3. Results

3.1. Effects of FHC silencing and hemin treatment on K562 cells

The primary goal of this study was the identification of genes whoseexpression is directly or indirectly regulated by the FHC during ery-throid differentiation. As a model system we used the cell line K562,that is known to differentiate along the erythroid lineage upon treat-ment with hemin. Firstly, we applied the shRNA silencing technique toknock-down FHC expression in K562 cells. Fig. 1 shows a representativeanalysis of the heavy ferritin subunit mRNA (A) and protein (B) expres-sion in wild-type and FHC-silenced K562 cells; both transcript and pro-tein levels are consistently reduced upon FHC silencing.

The effects of erythroid differentiation on FHC expression in K562cells were evaluated by a time-course exposure of the cells to 50 μMhemin. As a control, γ-globin levels were also monitored in the assay,since it is known that synthesis of this protein is heavily potentiatedunder these conditions (Iwasaki et al., 2006). Fig. 2A shows howhemin increases FHC and γ-globin protein levels at 24, 36 and 72 h;this latter time point was used for further experiments. Fig. 2B showsthe western blot analysis of FHC and γ-globin in hemin-treated wild-type and silenced K562 cells, compared with untreated controls.Hemin produced a significant increase in FHC and γ-globin proteinlevels both in wild-type and FHC-silenced cells. qRT-PCR analysis ofFHC and FLC mRNAs, shown in Figs. 2C and D, respectively, indicatesthat FHC mRNA levels remain almost undetectable in the silencedcells,while FLC transcription is increased approximately to the same de-gree in both wild type and FHC-silenced K562.

Fig. 4. Ingenuity Pathways Analysis (IPA). In order to investigate potential interactions ofdifferentially regulated genes, datasets representing almost 800 genes with an altered ex-pression profile obtained from the microarray analysis and belonging to the three classesdescribed in the text were imported in the Ingenuity Pathway Analysis tool and the fol-lowing data is illustrated: (A) Class 1: the list of the top five networkswith their respectivescores (top) and the network representation of themost highly rated network (cell deathand survival, cardiac damage, cardiovascular disease) (bottom); (B) Class 2: the list of thetopfive networkswith their respective scores (top) and the network representation of themost highly rated network (cell-to-cell signaling and interaction, hematological systemdevelopment and function, immune cell trafficking) (bottom); (C) Class 3: the list of thetop five networks with their respective scores (top) and the network representation ofthe most highly rated network (cell death and survival, cellular growth and proliferation,cellular development) (bottom). Shaded genes in the networkwere determined to be sig-nificant from the statistical analysis. Red shaded are the upregulated and green shaded arethe downregulated genes, respectively. Intensity of shading correlates with the degree ofup- or downregulation. A solid line represents a direct interaction between two genes,while a dotted line indicates an indirect interaction.

331R. Misaggi et al. / Gene 535 (2014) 327–335

Author's personal copy

3.2. Gene expression profiling

In order to perform a comparative analysis, total RNAwas extractedfrom two biological replicates of K562WT and K562shFHC, as well as fromtwo biological replicates of hemin-treated K562WT and K562shFHC cells.Each pair of replicates was then pooled and cRNAs were producedfrom each cell line and used to hybridize three microarrays (technicalreplicates). Analysis of microarrays was performed by IlluminaGenomeStudio v.2010.3 software (see Materials and methods). Thedata obtained were subjected to a standard two-factor ANOVA withthe aim of highlighting significant differences in mRNA expressionunder the different experimental conditions. The ANOVA test wasspecifically chosen to ascertain whether FHC silencing alone, heminalone, or both might have contributed to the differential expression ofa specific gene in each cell population.

The results of this analysis have been grouped in three classes(Supplementary Table S1): i) Class 1, comprising 657 RNAs whose ex-pression appears to be uniquelymodified by hemin treatment, indepen-dently from the presence or absence of FHC; ii) Class 2, including 70RNAs that are modulated by hemin-triggered differentiation and FHCsilencing; and iii) Class 3, which entails 128 RNAs whose expression isexclusively controlled by FHC. Heatmap of differentially expressedgenes in the three distinct classes is provided as Fig. 3.

3.3. Classification of pathways among differentially expressed genes

Pathways potentially dysregulated following FHC-silencing andhemin treatment of K562 cells were investigated by importing intothe Ingenuity Pathway Analysis tool (IPA) the list of the differentiallyexpressed genes included in the three classes described above.

IPA analysis highlighted several interesting networks in each of thethree classes (Supplementary Table S2). Fig. 4 represents the list of thetop 5 networks in Class 1 (A), Class 2 (B) and Class 3 (C). Of these net-works: i) cell death and survival, cardiac damage, and cardiovasculardisease resulted in the highest rated network in class 1 (with 28 focusmolecules and a significance score of 35); ii) cell-to-cell signaling and in-teraction, hematological system development and function, and immunecell trafficking was the highest rated network in class 2 (with 11 focusmolecules and a significance score of 20); and, iii) cell death and survival,cellular growth and proliferation, and cellular development resulted in thehighest rated network of class 3 (with 15 focus molecules and a signifi-cance score of 23). It might be useful to mention that the score is theprobability that a collection of genes equal to or greater than the num-ber in a network can be achieved by chance alone. A score of 3 indicatesa 1/1000 chance that the focus genes are in a network not due to ran-dom chance (Dubey et al., 2011). The list of genes in each networkwith their respective fold changes in the array data is shown in Supple-mentary Table S3.

To further characterize the effects of FHC silencing and hemin treat-ment on K562 cells, the differentially expressed genes identified by mi-croarray analysis within the three distinct classes were fed into theProtein ANalysis THrough Evolutionary Relationship (PANTHER) webresource. Table 1 shows a list of the enriched and depleted biologicalpathways. Hemin-induced differentiation determined several changesin critical metabolic pathways: among them, a significant overexpres-sion of genes involved in general transcription, translation and nucleo-tide processing was observed, together with a marked downregulationof transcripts related to cell–cell adhesion, signal transduction and cellcommunication. On the other hand, when K562 cells were simulta-neously subjected to FHC-knock down and hemin treatment, a signifi-cant upregulation of pathways related to RNA metabolic processes,cholesterol metabolic processes and blood circulation, and a downregu-lation of general cellular processes emerged from the PANTHER analy-sis. Finally, FHC silencing alone produced an upregulation of genesrelated to response to toxin, macrophage activation and RNA localiza-tion, while the main downregulated pathways under these conditionsresulted to be transcription, primary metabolic processes and cell cycle.

4. Discussion

Changes in gene expression in K562 cells undergoing differentiationtoward the erythroid or megakaryocytic lineages have been extensivelycharacterized, more recently bymeans of high-throughput genomic andproteomic techniques (Addya et al., 2004; Tsolmon et al., 2011), leadingto the identification of a number of differentially regulated products in-cluding transcription factors, signalingmolecules and enzymes critical incell metabolism (Addya et al., 2004; Tsolmon et al., 2011). One of theproteinswhose expression is positively modulated during erythroid dif-ferentiation is the H subunit of the ferritin molecule (Iwasaki et al.,2006). Ferritin is the major iron storage protein in the cell, where it iscompartmentalized predominantly in the cytoplasm and, in variable

Table 1Enriched/depleted pathways from PANTHER analysis.

Class Pathways Homo sapiensgenes (ref)

Differentiallyexpressed gene list

# # Expected +/−

1 Metabolic process 8351 251 200.84 +tRNA aminoacylation for proteintranslation

48 8 1.15 +

Primary metabolic process 8037 236 193.29 +General transcription from RNApolymerase II promoter

76 8 1.83 +

Porphyrin metabolic process 34 5 0.82 +Translation 453 22 10.89 +Nucleobase, nucleoside, nucleotideand nucleic acid metabolic process

3987 122 95.89 +

Transcription from RNApolymerase II promoter

2343 77 56.35 +

Transcription 2353 77 56.59 +DNA metabolic process 520 23 12.51 +Sensory perception 864 5 20.78 −Cell–cell adhesion 843 5 20.27 −Cell surface receptor linkedsignal transduction

2236 29 53.78 −

Signal transduction 4293 73 103.25 −Nervous system development 1497 18 36 −Reproduction 1143 12 27.49 –

Ectoderm development 1698 22 40.84 −Cell adhesion 1411 17 33.93 −System development 2277 34 54.76 −System process 2466 38 59.31 −Visual perception 527 3 12.67 −Cell communication 4504 81 108.32 −G-protein coupled receptorprotein signaling pathway

1042 12 25.06 −

Gamete generation 1023 12 24.6 −Neurological system process 2172 35 52.24 −

2 RNA metabolic process 668 2 0.27 +RNA catabolic process 70 1 0.03 +Cholesterol metabolic process 117 1 0.05 +Blood circulation 246 1 0.1 +

3 Response to toxin 99 3 0.47 +Macrophage activation 352 5 1.67 +Localization 160 3 0.76 +Neurological system process 2172 3 10.32 −Synaptic transmission 756 0 3.59 −Transcription 2353 5 11.18 −Metabolic process 8351 30 39.67 −Transcription from RNApolymerase II promoter

2343 5 11.13 −

Primary metabolic process 8037 29 38.18 −G-protein coupled receptorprotein signaling pathway

1042 1 4.95 −

System process 2466 6 11.71 −Carbohydrate metabolic process 984 1 4.67 −Mitosis 621 0 2.95 −Cell cycle 1867 4 8.87 −

#: number of genes expected in the list for this PANTHER category, based on the referencelist; +/−: over/under representation of a category.

332 R. Misaggi et al. / Gene 535 (2014) 327–335

Author's personal copy

amounts, in mitochondria and nuclei. The increase in FHC levels duringerythroid differentiation of K562 cells has been linked to its wellknown role in ironmetabolism (i.e.more ferritin is necessary to enhanceiron storage capability for the hemoglobinization process). However, thehemin-mediated increase in ferritinmRNA in K562 cellswas not blockedby deferoxamine (Mattia et al., 1989), suggesting that this induction is atleast in part independent from the amounts of intracellular chelatableiron. In addition, cytoplasmic FHC may contribute, in these cells, to thecontrol of the balance between cell death and survival, since this proteinhas been shown to modulate the ERK1/2, and the TNF/JNK pathways(Bubici et al., 2004; Li et al., 2006). To better understand the role ofFHC in differentiating K562 cells, we characterized the changes in geneexpression induced by FHC silencing upon hemin exposure.

A two-way ANOVA performed on the microarray dataset obtainedfrom wild type and FHC-silenced cells, in the presence or absence ofhemin, identified three distinct classes of modulated transcripts. Thefirst class defines a subset of RNAs whose expression is modified, interms of up- or down-regulation, during erythroid differentiation, inde-pendently from the intracellular levels of FHC. Differentiation is a highlyregulatedmulti-stage process based on the specific activation or repres-sion of numerous genes; as expected, this class contains the largest

number of transcripts (more than 650), some of which have been al-ready identified in previous studies (Jacquel et al., 2006). The secondclass comprises approximately 70 RNAs that are differentially expressedin response to both hemin stimulation and FHC silencing. This class in-cludes genes whose expression profiles highlight synergistic actionsexerted by hemin and FHC. The third class of transcripts defines agroup of 128 genes whose expression is influenced exclusively bythe intracellular FHC content and remains unmodified during K562erythroid differentiation.

Network analysis performed by IPA and PANTHER revealed that celldeath and survival are among the most favored pathways in all classes,whereas other networks appear to correlate specifically with heminplus FHC silencing or with FHC silencing alone. Considering these re-sults, it seems appropriate to conclude that FHC, alone or in cooperationwith other stimuli, plays a role in the modulation of gene expressionthat is relevant to erythroid differentiation. Consistent with this notionis the observation that ID2, a known inducer of erythropoiesis (Ji et al.,2008) is downregulated by FHC silencing in hemin-stimulated K562.Of particular interest is the presence in class 2 of the mRNAs encodingthe transcription factors SOX18, here identified for the first time as adifferentiation-regulated gene in K562, C/EBP gamma, and ATF5. All

Fig. 5. Validation of the microarray data by quantitative RT-PCR (qRT-PCR). qRT-PCR was done in K562WT and K562shFHC, with and without hemin, in order to validate two genes whoseexpression is regulated by FHC silencing alone. The real time data is expressed as the mean ± SD of three independent experiments performed in triplicates.

333R. Misaggi et al. / Gene 535 (2014) 327–335

Author's personal copy

these factors are implicated in development and differentiation(Downes and Koopman, 2001; Greene et al., 2009; Yan et al., 2012).Moreover, this class includes also NXF1, belonging to a group of evolu-tionarily conserved proteins classified as nuclear export factors(Erkmann and Kutay, 2004) and Zfp36, a zinc finger protein that hasbeen reported to regulate the half-life of the mRNAs encoding impor-tant immune mediators (Brahma et al., 2012). These data furtherstrengthen the role played by FHC in regulating gene expression andopen the way for additional studies in this direction.

Another interesting aspect of the role of FHC in K562 emerges fromthe analysis of class 3 genes that are regulated by FHC silencing alone,independently from hemin-induced differentiation. Two of thesegenes, namely MAGE-A4 and SPARC, reported to act as tumor suppres-sor genes (Chlenski and Cohn, 2010; Goodyear et al., 2010), are up-regulated under these conditions (see Fig. 5). In particular, MAGE-A4is a member of the melanoma-associated antigen family and acts as apro-apoptotic factor. Its expression has been documented to be alteredin several tumor types,with a very poor expression in human leukemias(Goodyear et al., 2010). SPARC is an acidic, cysteine-rich, secreted pro-tein involved in apoptosis, angiogenesis, cell cycle and proliferation(Chlenski and Cohn, 2010). SPARC has been found overexpressed inhuman colo-rectal cancers sensitive to chemotherapy where it acts asa chemotherapy sensitizer. In an independent study we found thatSPARC expression is modulated by FHC silencing also in human meta-static melanoma cells (Di Sanzo et al., 2011). These data not only vali-date the finding in K562 cells, but also delineate a potential role forFHC in the modulation of the expression of tumorigenesis-associatedgenes in diverse cell types.

The observation that FHC silencing did not significantly affect γ-globin synthesis in response to hemin might appear surprising. Howev-er, the increased iron demand for the hemoglobinization process maybemainly fulfilled by the L ferritin subunit—whosemRNA accumulationin the FHC-silenced cells is stimulated by hemin (Fig. 2D)—and in partby the increased translation efficiency of the residual FHC mRNA inhemin stimulated K562 (compare lanes 5, 6 and 8 of Fig. 2). On theother hand, a substantial alteration of the intracellular iron metabolismwould be expected tomodify significantly the expression of other genesinvolved in this process (i. e. transferrin receptor, HO1, ferroportin); inourmicroarray analyses, none of these genes was found consistently af-fected following FHC-silencing.

Taken together, the results presented in this paper confirm the func-tional role of the ferritin FHC subunit in cell homeostasis, underscore itsfunction as a regulator of gene expression andprovide novel insights intothe complex gene network modulated during erythroid differentiation.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.gene.2013.10.067.

Conflict of interest

The authors declare that they have no competing interests.

Acknowledgments

This researchwas supported by:Ministero dell'Istruzione, Universitàe Ricerca (Grants PRIN-2008YFRLC8 and 2010LC747T 002, PONa300138), Associazione Italiana per la Ricerca sul Cancro (GrantsIG-9204 and IG-13176) and Universita' di Salerno (Grants FARB 2011and 2012). CS is a PhD student of the Research Doctorate “MolecularOncology, Experimental Immunology and Innovative Therapy Develop-ment”, University of Catanzaro “Magna Græcia”.

References

Addya, S., et al., 2004. Erythroid-induced commitment of K562 cells results in clusters ofdifferentially expressed genes enriched for specific transcription regulatory elements.Physiol. Genomics 19, 117–130.

Alkhateeb, A.A., Connor, J.R., 2010. Nuclear ferritin: a new role for ferritin in cell biology.Biochim. Biophys. Acta 1800, 793–797.

Beard, J.L., Connor, J.R., Jones, B.C., 1993. Iron in the brain. Nutr. Rev. 51, 157–170.Bevilacqua, M.A., Giordano, M., D'Agostino, P., Santoro, C., Cimino, F., Costanzo, F., 1992.

Promoter for the human ferritin heavy chain-encoding gene (FERH): structural andfunctional characterization. Gene 111, 255–260.

Bevilacqua, M.A., et al., 1995. Transcriptional activation of the H-ferritin gene in differen-tiated Caco-2 cells parallels a change in the activity of the nuclear factor Bbf. Biochem.J. 311, 769–773.

Bevilacqua, M.A., Faniello, M.C., Cimino, F., Costanzo, F., 1997a. Okadaic acid stimulates Hferritin transcription in HeLa cells by increasing the interaction between the p300CO-activator molecule and the transcription factor Bbf. Biochem. Biophys. Res.Commun. 240, 179–182.

Bevilacqua, M.A., et al., 1997b. A commonmechanism underlying the E1A repression andthe cAMP stimulation of the H ferritin transcription. J. Biol. Chem. 272, 20736–20741.

Bevilacqua, M.A., Faniello, M.C., Russo, T., Cimino, F., Costanzo, F., 1998. P/CAF/p300complex binds the promoter for the heavy subunit of ferritin and contributes to itstissue-specific expression. Biochem. J. 335, 521–525.

Bevilacqua, M.A., Faniello, M.C., Iovine, B., Russo, T., Cimino, F., Costanzo, F., 2002.Transcription factor NF-Y regulates differentiation of CaCo-2 cells. Arch. Biochem.Biophys. 407, 39–44.

Bomford, A., Conlon-Hollingshead, C., Munro, H.N., 1981. Adaptive responses of rat tissueisoferritins to iron administration. Changes in subunit synthesis, isoferritinabundance, and capacity for iron storage. J. Biol. Chem. 256, 948–955.

Brahma, P.K., et al., 2012. The mRNA-binding protein Zfp36 is upregulated byβ-adrenergic stimulation and represses IL-6 production in 3T3-L1 adipocytes. Obesity20, 40–47.

Brazma, A., et al., 2001. Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nat. Genet. 29, 365–371.

Bubici, C., Papa, S., Pham, C.G., Zazzeroni, F., Franzoso, G., 2004. NF-kappaB and JNK: an in-tricate affair. Cell Cycle 3, 1524–1529.

Chazenbalk, G.D., Wadsworth, H.L., Rapoport, B., 1990. Transcriptional regulation of ferri-tin H messenger RNA levels in FRTL5 rat thyroid cells by thyrotropin. J. Biol. Chem.265, 666–670.

Chlenski, A., Cohn, S.L., 2010. Modulation of matrix remodeling by SPARC in neoplasticprogression. Semin. Cell Dev. Biol. 21, 55–65.

Cicatiello, L., et al., 2010. Estrogen receptor alpha controls a gene network in luminal-likebreast cancer cells comprising multiple transcription factors and microRNAs. Am.J. Pathol. 176, 2113–2130.

Coffman, L.G., Parsonage, D., D'Agostino Jr., R., Torti, F.M., Torti, S.V., 2009. Regulatoryeffects of ferritin on angiogenesis. Proc. Natl. Acad. Sci. U. S. A. 106, 570–575.

Costanzo, F., et al., 1986. Structure of gene and pseudogenes of human apoferritin H.Nucleic Acids Res. 14, 721–736.

D'Agostino, P., et al., 1995. Negative and positive elements in the promoter region of thehuman apoferritin L gene. Biochem. Biophys. Res. Commun. 215, 329–337.

Di Sanzo, M., et al., 2011. H ferritin gene silencing in a human metastatic melanoma cellline: a proteomic analysis. J. Proteome Res. 10, 5444–5453.

Downes, M., Koopman, P., 2001. SOX18 and the transcriptional regulation of blood vesseldevelopment. Trends Cardiovasc Med 11, 318–324.

Dubey, R., Chhabra, R., Saini, N., 2011. Small interfering RNA against transcription factor STAT6leads to increased cholesterol synthesis in lung cancer cell lines. PLoS One 6 (12), e28509.

Erkmann, J.A., Kutay, U., 2004. Nuclear export of mRNA: from the site of transcription tothe cytoplasm. Exp. Cell Res. 296, 12–20.

Faniello, M.C., et al., 1999. The B subunit of the CAAT-binding factor NFY binds the centralsegment of the co-activator p300. J. Biol. Chem. 274, 7623–7626.

Faniello, M.C., et al., 2002. An alternative model of H ferritin promoter transactivation byc-Jun. Biochem. J. 363, 53–58.

Faniello, M.C., et al., 2006. Detection and functional analysis of an SNP in the promoter ofthe human ferritin H gene that modulates the gene expression. Gene 377, 1–5.

Faniello, M.C., et al., 2008. p53-Mediated downregulation of H ferritin promoter transcrip-tional efficiency via NF-Y. Int. J. Biochem. Cell Biol. 10, 2110–2119.

Goodyear, O., et al., 2010. Induction of a CD8+ T-cell response to the MAGE cancer testisantigen by combined treatment with azacitidine and sodium valproate in patientswith acute myeloid leukemia and myelodysplasia. Blood 116, 1908–1918.

Greene, L.A., Lee, H.Y., Angelastro, J.M., 2009. The transcription factor ATF5: role inneurodevelopment and neural tumors. J. Neurochem. 108, 11–22.

Iwasaki, K.,Mackenzie, E.L., Hailemariam, K., Sakamoto, K., Tsuji, Y., 2006. Hemin-mediatedregulation of an antioxidant-responsive element of the human ferritin H gene and roleof Ref-1 during erythroid differentiation of K562 cells. Mol. Cell. Biol. 26, 2845–2856.

Jacquel, A., et al., 2006. A survey of the signaling pathways involved in megakaryocyticdifferentiation of the human K562 leukemia cell line by molecular and c-DNA arrayanalysis. Oncogene 25, 781–794.

Ji, M., Li, H., Suh, H.C., Klarmann, K.D., Yokota, Y., Keller, J.R., 2008. Id2 intrinsically regu-lates lymphoid and erythroid development via interaction with different target pro-teins. Blood 112, 1068–1077.

Levi, S., et al., 1988. Mechanism of ferritin iron uptake: activity of the H-chain and deletionmapping of the ferro-oxidase site. A study of iron uptake and ferro-oxidase activity ofhuman liver, recombinant H-chain ferritins, and of two H-chain deletion mutants.J. Biol. Chem. 263, 18086–18092.

Li, R., Luo, C., Mines, M., Zhang, J., Fan, G.H., 2006. Chemokine CXCL12 induces binding offerritin heavy chain to the chemokine receptor CXCR4, alters CXCR4 signaling, and in-duces phosphorylation and nuclear translocation of ferritin heavy chain. J. Biol. Chem.281, 37616–37627.

Liau, G., Chan, L.M., Feng, P., 1991. Increased ferritin gene expression is both promoted bycAMP and a marker of growth arrest in rabbit vascular smooth muscle cells. J. Biol.Chem. 266, 18819–18826.

334 R. Misaggi et al. / Gene 535 (2014) 327–335

Author's personal copy

Mattia, E., den Blaauwen, J., Ashwell, G., van Renswoude, J., 1989. Multiple post-transcriptional regulatory mechanisms in ferritin gene expression. Proc. Natl. Acad.Sci. U. S. A. 86, 1801–1805.

Mega, T., et al., 2011. Zinc finger protein 521 antagonizes early B-cell factor 1 and modu-lates the B-lymphoid differentiation of primary hematopoietic progenitors. Cell Cycle10, 2129–2139.

Mutarelli, M., et al., 2008. Time-course analysis of genome-wide gene expression data fromhormone-responsive human breast cancer cells. BMC Bioinforma. 26 (9 Suppl. 2), S12.

Pierantoni, G.M., et al., 2003. High-mobility group A1 proteins are overexpressed inhuman leukaemias. Biochem. J. 372, 145–150.

Pountney, D., Trugnan, G., Bourgeois, M., Beaumont, C., 1999. The identification of ferritin inthe nucleus of K562 cells, and investigation of a possible role in the transcriptional reg-ulation of adult β-globin gene expression. J. Cell Sci. 112, 825–831.

Quaresima, B., et al., 2008. BRCA1 5083del19 mutant allele selectively up-regulatesperiostin expression in vitro and in vivo. Clin. Cancer Res. 14, 6797–6803.

Ray, P.D., Huang, B.W., Tsuji, Y., 2012. Reactive oxygen species (ROS) homeostasis andredox regulation in cellular signaling. Cell. Signal. 24, 981–990.

Tsolmon, S., Nakazaki, E., Han, J., Isoda, H., 2011. Apigetrin induces erythroid differentia-tion of human leukemia cells K562: proteomics approach. Mol. Nutr. Food Res. 55(Suppl. 1), S93–S102.

Wandzik, K., Zahn, C., Dassler, K., Fuchs, H., 2009. Substantial changes of cellular iron ho-meostasis during megakaryocytic differentiation of K562 cells. Dev Growth Differ 51,555–565.

Yan, C., Wang, X., Cao, J., Wu, M., Gao, H., 2012. CCAAT/enhancer-binding protein γ is acritical regulator of IL-1β-induced IL-6 production in alveolar epithelial cells. PLoSOne 7 (4), e35492.

335R. Misaggi et al. / Gene 535 (2014) 327–335