Cellular Iron Depletion and the Mechanisms Involved in the Iron-dependent Regulation of the Growth Arrest and DNA Damage Family of Genes*

Iron plays a crucial part in proliferation while iron deficiency results in G1/S arrest, DNA damage, and apoptosis. However, the precise role of iron in cell cycle control remains unclear. We showed that iron depletion using the iron chelators, desferrioxamine (DFO), or 2-hydroxy-1-napthylaldehyde isonicotinoyl hydrazone (311), increased the mRNA levels of the growth arrest and DNA damage 45α gene, GADD45α (Darnell, G. and Richardson, D. R. (1999) Blood 94, 781–792). In this study, we examined the effect of iron depletion on up-regulating GADD family members involved in growth control, including cell cycle arrest, apoptosis, and DNA repair, making them therapeutic targets for tumor suppression. We showed the GADD family members were up-regulated by cellular iron depletion. Further, up-regulation of GADD45α after iron deprivation was independent of hypoxia-inducible factor-1α (HIF-1α), octamer-1 (Oct-1), p53 and early growth response 1 (Egr1). We then analyzed the regulatory elements responsible for iron depletion-mediated regulation of GADD45α and identified the specific transcription factor/s involved. This region was within −117 bp and −81 bp relative to the start codon where the consensus sequences of three transcription factors are located: the CCAAT-binding factor/nuclear factor-Y (NF-Y), the stabilizing molecule v-MYB and the enhancer, CCAAT enhancer-binding protein (CEBPα). Mutation analysis, shRNA studies, Western blotting, and electrophoretic mobility shift assays led to the identification of NF-Y in the transcriptional up-regulation of GADD45α after iron depletion. Furthermore, like GADD45α, NF-YA was up-regulated after iron chelation and down-regulated by iron supplementation. These results are important for understanding the mechanisms of iron depletion-mediated cell cycle arrest, DNA damage repair, and apoptosis.

Iron is vital for proliferation and multiple mechanisms are involved in the anti-tumor activity of iron chelators whose effect induces cell cycle arrest and apoptosis (1,2). Iron depletion results in: 1) inhibition of the iron-containing enzyme, ribonucleotide reductase, which is critical for DNA synthesis (1); 2) increased expression of the growth and metastasis suppressor, N-myc downstream-regulated gene 1 (3); 3) up-regulation of p53 (4); 4) decreased cyclin D1 expression (5) that is required for G 1 /S progression; 5) down-regulation of p21 CIP1/WAF1 protein in some cell types (5) that may induce apoptosis; 6) down-regulation of Bcl-2 and up-regulation of pro-apoptotic Bax (6,7), and 7) up-regulation of cell division cycle 14A, which acts on substrates important for cell cycle progression (e.g. p27 kip1 and cyclin E) (8).
The current investigation focuses on five members of the growth arrest and DNA damage (GADD) 2 family of genes, namely: GADD34, GADD45␣, GADD45␤, GADD45␥, and GADD153. The expression of these genes is often increased when cells are subjected to stress such as nutrient deprivation (9 -11) or exposure to DNA-damaging agents (12), which may cause cell cycle arrest and/or apoptosis. Our laboratory previously showed that iron depletion mediated by the iron chelators, DFO or 311, caused a pronounced concentration-and time-dependent increase in the expression of GADD45␣ mRNA in three cell types (13). Similar to the effect observed with iron depletion, GADD45␣ and GADD153 mRNA are upregulated during hypoxia (14), suggesting a possible role for hypoxia-inducible factor-1␣ (HIF-1␣), that can be activated by both hypoxia and iron depletion via prolyl-hydroxylases (1).
Products of the GADD45 subgroup of genes encode three structurally related proteins, namely GADD45␣, GADD45␤, and GADD45␥, that play important roles in the G 2 /M checkpoint and apoptosis (15). However, only GADD45␣ has been demonstrated to activate p53-dependent G 2 /M arrest and inhibit cdc2 kinase (15). Neither GADD45␤ nor GADD45␥ are downstream targets of p53 (15). Interestingly, GADD45␣ is also known to interact with key cell cycle regulatory molecules, such as p21 CIP1/WAF1 (16), cdc2/cyclin B1 (17) and p38 mitogenactivated protein kinase (MAPK) (18). Studies using trans-* This work was supported in part by a Cancer Institute NSW (CINSW) fected cells suggest that GADD34 and GADD153 appear to have a direct role in initiating apoptosis rather than inducing cell cycle arrest (19,20). Overexpression of each GADD gene causes growth inhibition and/or apoptosis, while combined overexpression leads to synergistic or cooperative effects on anti-proliferative activity (21).
The current investigation has demonstrated that members of the GADD gene family are up-regulated in response to iron depletion. Our studies identify the constitutive transcriptional activator, nuclear factor Y (NF-Y), in the transcriptional upregulation of GADD45␣ after iron depletion.

EXPERIMENTAL PROCEDURES
Reagents-DFO was obtained from Novartis (Basel, Switzerland) and 311 was synthesized and characterized as described (22). Actinomycin D and ferric ammonium citrate were from Sigma-Aldrich.
Cell Culture-The human MCF7 breast cancer, SK-N-MC neuroepithelioma and SK-Mel-28 melanoma cell-types were obtained from the American Type Culture Collection (Manassas, VA). Murine embryonic fibroblasts (MEFs) from control and homozygous Oct-1, HIF-1␣. and Egr1 knock-out mice were obtained from Dr. D. Tantin (University of Utah, Salt Lake City, UT), Dr. R. Johnson (University of California, San Diego, CA) and Dr. J. Milbrandt (Washington University, St. Louis, MO), respectively. H1299 cells (p53-null) stably transfected with tetracycline (Tet)-regulated p53 were obtained from GenHunter (Nashville, TN). Cells were cultured at 37°C in the presence of medium containing 10% FCS (3). Unless otherwise specified, cells were incubated under standard conditions in an atmosphere of 95% air/5% CO 2 . In some experiments, cells were incubated under hypoxic conditions in an atmosphere of 1% O 2 , 94% N 2 , and 5% CO 2 (BOC Gases Ltd, Sydney, Australia).
RNA Isolation and RT-PCR-RNA was isolated using TRIzol (Invitrogen, Melbourne, Australia). RT-PCR was performed via standard procedures (23) using the primers in Table  1. RT-PCR was shown to be semi-quantitative by a protocol which demonstrated it was in the log-phase of amplification. Expression of ␤-actin was used as an RNA-loading control.
Luciferase Assays-Transfected MCF7 cells were incubated for 24 h/37°C with control medium or medium containing 25 M 311 or 250 M DFO. Protein extraction was performed using Passive Lysis Buffer (Promega, Madison, WI). Luciferase Assays were carried out using the Promega Luciferase Assay System as per the manufacturer's protocol. Luminescence was measured using a FLUOstar Omega Luminometer (BMG Labtech, Mornington, Australia). Mutagenesis Analysis-The introduction of specific mutations into the promoter deletion constructs was carried out using the GeneTailor kit (Invitrogen). Briefly, the modified target sequences were based on the consensus sequences of CCAAT enhancer-binding protein (CEBP), CEBP␣, GATA, v-MYB, and NF-Y and then confirmed by sequencing (AGRF). The plasmid carrying the Ϫ117 bp fragment of the GADD45␣ promoter region was methylated with a DNA methylase enzyme (4 units/l; Invitrogen) for 1 h/37°C. Then, a PCR reaction was carried out using a mutagenic and a corresponding complementary primer and the resulting amplified plasmid transformed in One-Shot MAX Efficiency DH5␣ TM -T1R competent cells (Invitrogen). Following transformation, the methylated (not mutated) DNA was digested by McrBC and Mrr endonucleases in Escherichia coli (25), leaving only unmethylated mutated products.
shRNA-SureSilencing shRNA plasmids were purchased from SABioscience (Frederick, MD) and transiently transfected using Lipofectamine TM LTX with PLUS TM Reagent (Invitrogen). RNA and protein were harvested following transfection and incubation of MCF7 cells with either control medium or this medium containing DFO (250 M) or 311 (25 M) for 24 h/37°C.
Chemiluminescent EMSA-Studies examining DNA-protein interactions were carried out using the LightShift Chemiluminescent EMSA kit (Thermo Scientific, Rockford, IL) as per the manufacturer's protocol. The double-stranded biotinylated oligonucleotides were based on the NF-Y consensus sequence in the GADD45 promoter and a mutated sequence, namely 5Ј-atagcccaatggccaagctg-3Ј and 5Ј-atagctagctggccaagctg-3Ј, respectively. Briefly, nuclear proteins were extracted using NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Scientific, Scoresby, Australia). The DNA/protein-binding reaction was carried out for 20 min at room temperature. Then equal amounts of protein (2 g), as determined by the Bio-Rad protein assay reagent, were loaded onto a native 6% polyacrylamide gel in 0.5ϫ TBE and run at 100 V for 1 h at room temperature. The binding reaction was subsequently transferred onto a nylon membrane (GE Healthcare, Australia) at 100 V/30 min at room temperature. To cross-link transferred DNA to the membrane, a 60 s exposure at 120 mJ/cm 2 at room temperature was performed using the UV Stratalinker 1800 (Stratagene, Santa Clara, CA). The biotin-labeled DNA was then detected by chemiluminescence.
Statistics-Results presented are typical of three independent experiments and given as mean Ϯ S.D. Experimental data were compared using Student's t test. Results were considered statistically significant when p Ͻ 0.05.

GADD Family mRNA Levels Are Increased after Iron
Depletion-Incubation of cells with DFO or 311 resulted in a marked and significant increase (p Ͻ 0.05-0.001) in the levels of GADD34, GADD45␣, -␤, -␥, and GADD153 mRNA compared with cells grown in control medium (Fig. 1A). As positive controls, we also examined the effect of chelators on the expression of p21 WAF1/CIP1 and TfR1 mRNA which are known to be up-regulated after iron chelation (5, 13) and were significantly (p Ͻ 0.05-0.01) increased after incubation with the ligands relative to the control. To assess the role of cellular iron depletion on the up-regulation of the GADD gene family, cells were incubated with 311 that had been pre-saturated with iron(III) to form its iron complex (i.e. 2:1 ligand-metal ion complex, 311-Fe) which cannot bind cellular iron pools (22). In contrast to the ability of 311 (25 M) to significantly (p Ͻ 0.01-0.001) increase the expression of GADD family members, p21 WAF1/CIP1 and TfR1 mRNA (Fig. 1A), its respective iron complex (25 M) had no significant effect on up-regulating these genes relative to the control (supplemental Fig. S1). Similar results were also found comparing DFO (250 M) to the DFO-Fe(III) complex (1:1 ligand-metal ion complex (250 M); data not shown). Collectively, these data demonstrate that it is the ability of chelators to bind iron that results in up-regulation of the GADD gene family. In these experiments, 311 was used in preference to DFO due to its rapid permeation of cells to deplete iron pools (13,26). The effect of hypoxia (1% O 2 , 94% N 2 , 5% CO 2 ) was also assessed as the effect of iron chelation on increasing GADD family expression could be mediated through inhibition of iron-requiring prolyl hydroxylases that would increase HIF-1␣ expression (27). As a positive control, to induce the expression of the GADD family, cells were incubated with a low concentration (5 nM) of the DNA-damaging agent, actinomycin D (Act D) (3,28,29). At this concentration, Act D does not inhibit transcription, but leads to DNA damage, which is known to up-regulate GADD genes (3,28).

Alterations of GADD Family mRNA and Protein Expression after Incubation with Chelators Is Time-dependent-To
Incubation with control medium alone led to either only a slight or no increase in the expression of the GADD family genes. In contrast, 311 caused significant (p Ͻ 0.001) up-regulation in the expression of all these genes after 24 h, with the initial increase in their mRNA levels being evident after 4 -16 h Western blot analysis was then performed to assess protein expression of these members of the GADD gene family. The expression of these proteins remained relatively constant (GADD34) or was increased (GADD45␣, -␤, -␥, and GADD153) over a 24-h incubation when cells were maintained in control medium alone (supplemental Fig. S3A). It was notable that a double band was sometimes visible for the GADD45␣ blots that has also been described previously by others and may be due to post-transcriptional processing of the protein (30).
Incubation with 311 led to an increase in GADD34, GADD45␤, GADD45␥, and GADD153 protein expression after only 2 h relative to the 0 h time point, while for GADD45␣, an increase in expression only became evident after a 8 h incuba-tion (supplemental Fig. S3B). Hypoxia increased GADD34, GADD45␤, GADD45␥, and GADD153 protein expression after 4 h, while an increase in GADD45␣ was only found after an 8-h incubation (supplemental Fig. S3C).
Iron Supplementation Decreases GADD mRNA Levels following Iron Chelation-To determine if the observed iron chelator-mediated up-regulation of GADD mRNA was due to cellular iron depletion, we examined the effect of iron repletion by incubating cells with the iron donor, ferric ammonium citrate (FAC) (3,31). In these studies, after a primary incubation of 24 h These results showed that the mRNA levels of the GADD family members were slightly reduced or significantly (p Ͻ 0.05) decreased when cells were incubated with control medium for 24 h and then reincubated with FAC for 24 h when compared with cells incubated with control media for 48 h (Fig. , 94% N 2 , 5% CO 2 ) with control medium only. C, up-regulation of GADD45␣ and TfR1 (as control) mRNA levels after incubation with DFO or 311 is markedly decreased upon re-incubation with the iron donor, ferric ammonium citrate (FAC, 100 g/ml). Results shown are typical gels from three experiments, while the densitometry is mean Ϯ S.D. *, p Ͻ 0.05; **, p Ͻ 0.01; ***, p Ͻ 0.001, relative to the control. OCTOBER Fig. S4, lanes 1 and 2). In contrast, when cells were incubated with either DFO or 311 for 24 h to induce iron depletion and then reincubated with either control medium or the chelator, there was a marked increase in the mRNA levels of the GADD family members, or the positive controls for iron depletion, namely p21 or TfR1. However, when cells were first incubated with DFO or 311 and then subsequently reincubated with FAC, there was a marked decrease in the mRNA levels of the GADD members, p21 and TfR1, indicating cellular Fe-repletion. These results indicate that like the TfR1, mRNA expression of the GADD family members was regulated by cellular iron levels.

Iron Depletion and GADD Gene Family Expression
Deletion Analysis of the GADD45␣ Promoter Region-Considering the regulation of the GADD gene family by cellular iron levels shown above, further work examined the mechanism involved. In these studies, we focused on GADD45␣ as it is the best characterized of this group of genes and shows important functions relevant to the response to iron depletion (13). Further, GADD45␣ activates p53-dependent G 2 /M arrest and inhibits cdc2 kinase (15). To identify the regulatory element responsible for iron chelator-mediated up-regulation of GADD45␣ expression, promoter deletion constructs (from the 5Ј-end of the promoter) of various sizes were joined to a luciferase reporter vector ( Fig. 2A). The plasmids were transiently transfected into MCF7 cells, which were then incubated with either control medium or this medium containing 311 (25 M) for 24 h/37°C. As shown in Fig. 2A, after incubation with 311 relative to the control, the reporter plasmid containing Ϫ234 bp of the GADD45␣ promoter relative to the transcription start site showed the greatest increase (p Ͻ 0.001) in luciferase activity as compared with the empty vector (pGVB2) or to any longer constructs (Ϫ817, Ϫ1326, and Ϫ2256 bp). On the other hand, transfection of smaller fragments of the promoter (Ϫ10 and Ϫ81 bp) showed almost no luciferase activity ( Fig. 2A).
Previous studies using these constructs to demonstrate the transcription factor responsible for up-regulation of GADD45␣ in response to UV-irradiation identified Oct-1 in the Ϫ234 bp region as being responsible (32). However, studies using Oct-1 ϩ/ϩ and Oct-1 Ϫ/Ϫ MEFs showed that incubation with 311 (25 M) could induce GADD45␣ expression to a similar degree in both cell types (Fig. 2B), demonstrating that Oct-1 was not essential for the up-regulation of GADD45␣ in response to iron depletion. Similar analysis, using appropriate cell types with or without the molecules of interest, was also extended to assess the role of p53, Egr1, and Hif-1␣ in the up-regulation of GADD45␣ after iron depletion (Fig. 2B). Indeed, these transcription factors were considered as potential candidates for transcriptional up-regulation of GADD45␣ as they are also regulated by intracellular iron levels (33). In addition, the role of p53 and Egr-1 in regulating GADD45␣ expression has been described (34), whereas the up-regulation of GADD45␣ mRNA during hypoxia was previously demonstrated (14) and confirmed in the current study (Fig. 1B). In each case, the expression of GADD45␣ was up-regulated after incubation with 311 in cells with or without p53, Egr1, or Hif-1␣, indicating that they were not essential for this effect.
To identify a specific region within the promoter responding to iron depletion, we performed additional promoter truncations (Fig. 2C). The resulting fragments were then cloned into a pGL4 luciferase reporter plasmid which was chosen because the pGVB2 vector used in Fig. 2A was no longer commercially available. Constructs lacking some of the potential transcription factor consensus sequences were generated by sequential deletion of the GADD45␣ promoter. The truncated constructs were then tested using the luciferase reporter assay in transfected MCF7 cells.
As shown in Fig. 2D, the deletion of the promoter region at base number Ϫ234, Ϫ143, Ϫ124, and Ϫ117 bp resulted in a marked and significant (p Ͻ 0.001) increase of the luminescence after incubation with 311 or DFO relative to the control. This is possibly due to the removal of extraneous sequence not necessary for transcription, leading to a more compact and efficient transcriptional unit responsible for GADD45␣ induction after iron deprivation. Further truncation of the promoter region to the Ϫ89 bp construct resulted in a 47-48% reduction of luminescence compared with the Ϫ117 bp construct, while again the Ϫ81 bp construct showed virtually no activity (Fig.  2D). The level of luminescence observed for the Ϫ89 bp construct was comparable to that found in the original Ϫ234 bp fragment, indicating that the core promoter could be present in the Ϫ81 to Ϫ89 bp region (Fig. 2D). Taken together, these results indicate that a transcription element is located between Ϫ89 and Ϫ81 bp, but this element is not sufficient for the highest induction of GADD45␣ expression. In fact, this can only be achieved by the involvement of a possible enhancer element located between the Ϫ89 and Ϫ117 bp region.
Mutagenesis Analysis of the GADD45␣ Promoter Region-The presence of transcription factor binding sites in the promoter region of GADD45␣ was analyzed using TFBIND and TESS software. Both programs score the highest probability of each potential transcription factor to bind to an appropriate binding site. According to this analysis, the potential consensus sequences with the greatest scores in the region between Ϫ89 and Ϫ117 bp were: CEBP␣ (90%), GATA (89%), OCT-1 (85%), and CEBP (81%) (Fig. 3A). Further, in the region between Ϫ81 and Ϫ89 bp, the potential consensus sequences with the highest scores were: (i) the CCAAT-binding protein (CCAAT-BP), NF-Y (85%); (ii) v-MYB (85%); and (iii) a second NF-Y site (86%; Fig. 3A). Because of the lack of specific restriction enzyme sites in the region between Ϫ117 bp and Ϫ89 bp, a deletion strategy was not possible. Hence, the site-directed mutagenesis technique was employed to introduce a four base specific substitution into each consensus sequence (Fig. 3B). This technique has been previously shown to prevent transcription factor binding (35).
The introduction of mutations in CEBP, GATA, and the second NF-Y region did not induce a statistically significant reduction of luminescence compared with the non-mutated control plasmid pGL4 -117 (Fig. 3C). In contrast, mutation of the CEBP␣ sequence resulted in a reduction of luciferase gene expression by 46 -47% relative to pGL4 -117 after treatment of transfected cells with 311 or DFO. Interestingly, the mutations directed against the CCAAT-BP region (i.e. CAAT/NF-Y and NF-Y/v-MYB) almost completely abrogated luminescence rel-ative to pGL4 -117. This mutation not only affects the core of the CCAAT sequence, but also disrupts the binding sequence of two factors potentially involved in the regulation of GADD45␣ expression, namely NF-Y and v-MYB (Fig.  3C). In particular, NF-Y has previously been described as a CCAAT-BP (36), while v-MYB may be necessary for stabilization of NF-Y binding to its site (37). Hence, GADD45␣ expression can be regulated by the binding of NF-Y and this interaction appeared to receive stabilization by v-MYB and was enhanced by CEBP␣.

Abrogation of NF-YA Expression by shRNA Prevents the Iron Chelator-mediated Increase in GADD45␣ mRNA and Protein
Expression-To further examine the involvement of NF-Y in the regulation of GADD45␣ expression, we transfected MCF7 cells with two plasmids containing sequences for two different shRNAs (i.e. shRNA#1 and shRNA#2) to silence the critical A C, schematic diagram showing the restriction enzyme sites used to prepare GADD45␣ promoter constructs that were then used to generate the constructs implemented in D. D, luciferase assay was performed using MCF7 cells transiently transfected with promoter deletion constructs to identify the specific consensus sequences responsible for the up-regulation of GADD45␣ mRNA after iron depletion in the Ϫ234 and Ϫ81 bp region of the GADD45␣ promoter. Transfected cells were incubated for 24 h/37°C with control medium or this medium containing DFO (250 M) or 311 (25 M). Results shown are typical gels from three experiments, while the densitometry is mean Ϯ S.D. ***, p Ͻ 0.001, relative to the control. OCTOBER (Fig. 4A). Relative to the scrambled shRNA control vector, the transfected NF-YA-shRNA plasmids markedly and significantly (p Ͻ 0.001) inhibited NF-YA mRNA in control cells, but also those incubated with DFO or 311 (Fig. 4A). As a consequence of the inhibition of NF-YA expression, GADD45␣, GADD45␤, and GADD153 ( Fig. 4A and supplemental Fig. S5, B and D) mRNA levels were markedly and significantly (p Ͻ 0.01-0.001) lower when compared with the scrambled control when cells were incubated with DFO or 311. Hence, abrogation of NF-YA not only affects GADD45␣ mRNA levels, but also GADD45␤ and GADD153 mRNA, demonstrating a common mechanism of regulation. In contrast, there was no difference in the expression of GADD34, GADD45␥, and TfR1 between the scrambled shRNA control and the NF-YA-shRNA (supplemental Fig. S5, A, C, E). The lack of effect of NF-YA-shRNA on the expression of these latter genes provides appropriate controls for potential off-target effects which could lead to misinterpretation. Expression of TfR1 mRNA which is iron regulated (39), was assessed as a relevant internal control for iron depletion. Again, these results confirm an important role of NF-YA in the chelator-mediated expression of GADD45␣ and indicate a potential role of this transcription factor in regulating GADD45␤ and GADD153.

Iron Depletion and GADD Gene Family Expression
To further examine the role of NF-Y in GADD45␣ expression, we also performed Western blotting on lysates from NF-YA-shRNA transfected cells (Fig. 4B). These results confirmed that the protein expression of both NF-YA and GADD45␣ were significantly (p Ͻ 0.01-0.001) lower in NF-YA-shRNA-transfected cells relative to the scrambled shRNA-treated cells (Fig.  4B). Taken together, these results demonstrate that NF-YA is necessary for the up-regulation of GADD45␣ under iron deprivation, as in NF-YA-deficient cells the marked up-regulation of GADD45␣ expression by iron depletion by DFO and 311 was no longer observed.

NF-YA Protein Expression Is Regulated by Intracellular Iron
Levels-To further assess the role of NF-Y in GADD45␣ upregulation after iron depletion, we also assessed if NF-Y protein expression was regulated by cellular iron levels (Fig. 4C). In these studies, standard techniques (3,40) to modify iron levels implementing control medium, DFO (250 M) or 311 (25 M), or FAC (100 g/ml) were used as described for Fig. 1C and supplemental Fig. S4. NF-Y is a hetero-trimeric complex with NF-YA being the main subunit and the one responsible for CCAAT binding (38). Hence, it was chosen as the best repre-sentation of the protein complex to assess using Western blot analysis. As shown in Fig. 4C, GADD45␣ and NF-YA expression was significantly (p Ͻ 0.001) elevated in cells incubated with either DFO or 311 relative to cells incubated with control medium alone. A primary incubation with control medium, DFO or 311 followed by FAC resulted in a marked and significant (p Ͻ 0.001) decrease in both NF-YA and GADD45␣ expression (Fig. 4C). As a positive control, expression of the well known iron-regulated protein, TfR1 (39), was also assessed and demonstrated up-regulation after iron depletion by the  OCTOBER 14, 2011 • VOLUME 286 • NUMBER 41 chelators and then down-regulation after control or chelatortreated cells were incubated with FAC.

Iron Depletion and GADD Gene Family Expression
To assess whether the observed iron-dependent regulation of these genes was specific to MCF7 cells, we assessed NF-YA and GADD45␣ expression in multiple cell-types. Both human SK-N-MC neuroepithelioma and SK-Mel-28 melanoma cell lines showed a very similar pattern of expression as found in MCF7 cells with NF-YA and GADD45␣ being significantly (p Ͻ 0.001) up-regulated by iron chelators and down-regulated by FAC (supplemental Fig. S6). Taken together, these findings indicate that NF-Y is regulated by intracellular iron levels and corroborates the findings above concerning its role in regulating GADD45␣ expression.
Analysis of the Interaction between NF-Y and the Promoter of GADD45␣-To directly demonstrate the ability of NF-Y to bind the CCAAT binding region in the GADD45␣ promoter, EMSAs were carried out using nuclear protein extracts from cells incubated for 24 h with either control medium or medium containing DFO (250 M) or 311 (25 M). Double-strand bio-tin-labeled oligonucleotides were designed according to the wild-type and mutated GADD45␣ sequences reported in Fig. 3, A and B. As shown in Fig. 5A, when the binding reaction was carried out in the absence of a nuclear protein extract (lane 1), the only visible band was the excess free biotin-labeled DNA (indicated as free biotin labeled oligo in Fig. 5A). On the other hand, when the nuclear protein extract was added (lane 2), a band-shift occurred due to the binding of transcription factor(s) to this specific sequence. When the same unlabeled oligonucleotide was added in a 50-fold excess to the binding reaction as a specific competitor (lane 3), the shift was prevented. It is notable that the shifted band derived from the nuclear protein extract of cells treated with DFO or 311 was markedly and significantly (p Ͻ 0.001) greater than that present in control cells. This indicates that the up-regulation of NF-Y protein expression observed in cells incubated with 311 or DFO (Fig. 4, B and C) was also reflected by the greater binding of this transcription factor to its consequence sequence within the GADD45␣ promoter (Fig. 5A). To assess the specificity of the binding between NF-Y and the CCAAT binding region in the GADD45␣ promoter, a gel shift assay was also carried out in parallel to the studies in Fig. 5A, but using a mutated biotin-labeled target DNA (Fig. 5B). For consistency, the chosen sequence was the same used to introduce a four base mutation (from CCAA to TAGC) in the mutagenesis analysis, which practically abrogated reporter activity (Fig. 3, B and C). As shown in Fig. 5B, the EMSA analysis did not result in any band shift in the presence or absence of nuclear lysate with the only signal being represented by the excess labeled mutated target DNA (free biotin-labeled oligo). This confirms the results from the mutagenesis analysis indicating the high specificity of the binding of NF-Y to the CCAAT sequence and the importance of this binding for up-regulating GADD45␣ expression under iron deprivation conditions.

DISCUSSION
For the first time, we investigated regulation of members of the GADD gene family after iron depletion and have demonstrated that this group of genes are up-regulated at both the mRNA and protein level following iron depletion. Furthermore, we have shown that GADD45␣ responds to iron depletion via the binding of NF-Y to the CCAAT binding region.
Previous studies investigating the effects of UV light on GADD45␣ expression have implicated Oct-1 (35). Interestingly, the combined action of Oct-1 and NF-YA has also been shown to be necessary for the up-regulation of GADD45␥ after incubation of cancer cells with the histone deacetylase inhibitor, trichostatin (41). However, we demonstrated that Oct-1 was not essential for iron depletion-mediated up-regulation of GADD45␣ as both DFO and 311 still markedly up-regulated this gene in Oct-1 Ϫ/Ϫ cells relative to their Oct-1 ϩ/ϩ counterparts (Fig. 2B). Considering the results of the present investigation and those assessing UV-irradiation (35), it is likely that different transcription factors will respond to different stimuli to up-regulate the GADD family. Certainly, we have demonstrated that after iron depletion, up-regulation of NF-Y is important for the increased expression of GADD45␣, GADD45␤, and GADD153, while other factor(s) appear necessary for the increased expression of GADD34 and GADD45␥ under these conditions (supplemental Fig. S5). Considering the function of the GADD family, it is conceivable that increase in the expression of these GADD family members could be mediated through DNA damage mediated by iron depletion and this deserves further investigation.
A previous study examining GADD153 (42) is of particular interest as the GADD family have been shown to be induced by a variety of agents involved in mediating endoplasmic reticulum (ER) stress (42)(43)(44). A short region within the GADD153 promoter has been identified to be required for its inducibility by ER stress and contains two overlapping domains, namely a CCAAT-binding site for the NF-Y transcription factor and a GCACG variant of the ER-stress element (ERSE) (42). This response element differs from other ERSEs, since only a single copy is present in the GADD153 promoter and it contains an AG-rich spacer between the two main domains instead of one which is GC-rich (42).
In the current investigation, a similar potential ERSE was identified within Ϫ90 and Ϫ70 bp of the GADD45␣ promoter. As identified in the GADD153 ERSE, the GADD45␣ response element is also present as a single copy, but maintains the typical GC-rich spacer. It is also notable that the sequence of the domain potentially responsible for the ER-stress inducibility has a substitution of cytosine with thymine (GCATG instead of GCACG) and that this sequence was identified as a potential NF-Y binding site (Fig. 3A). However, mutagenesis analysis of this particular sequence revealed that it was not necessary for GADD45␣ up-regulation after iron depletion (Fig. 3C), and thus, it may be responsible for the induction of this gene after ER stress only. Hence, the up-regulation of GADD45␣ and GADD153 observed in this study may be a response related to iron depletion rather than ER stress.
It is notable that like GADD45␣ and GADD153, a CCAAT binding sequence is present in the promoter region of GADD45␤. Hence, this could explain why the repression of NF-YA was able to markedly prevent GADD45␤ expression after incubation with DFO or 311 (supplemental Fig. S5B). Interestingly, the CCAAT binding sequence in the promoter region of GADD45␥, does not lead to NF-Y binding since the repression of NF-YA was not able to prevent GADD45␥ expression after incubation with DFO or 311 (supplemental Fig. S5C). In contrast, GADD34 does not contain a CCAAT binding sequence, and as such, its expression after iron depletion was not affected by the down-regulation of NF-YA (supplemental Fig. S5A). This indicates that other mechanisms were involved in the up-regulation of GADD34 and GADD45␥ under these conditions. In fact, previous studies have indicated that ER stress and hypoxia can activate phosphorylation of the eukaryotic initiation factor 2␣ to inhibit overall protein synthesis (45). Paradoxically, this also leads to translation of a subset of mRNAs, such as the transcription factor, ATF4, which induces transcription of downstream stress-induced genes such as GADD34 (46). Hence, this mechanism may be responsible for the up-regulation of the latter gene after iron depletion. Collectively, the analysis above indicates a complex response to iron depletion that involves NF-Y and potentially ATF4.
In summary, iron depletion significantly increased the expression of the GADD gene family. Furthermore, the induction of GADD45␣ expression by iron depletion could be attributed to the binding of NF-Y to the CCAAT-binding motif in the GADD45␣ promoter with the aid of v-MYB as a stabilizing molecule and with the enhancement role of CEBP␣. Together, these results are relevant to understanding the molecular alterations involved in cell cycle arrest during iron deficiency and the mechanism of chelators at inducing anti-proliferative activity and apoptosis.